" },
{ MU, A, 0, 0, "([a-z]{0,3}c;)+", "ccccc;c;cc;ccc;cccccccccccccccc;" },
/* Bracket repeats with limit. */
{ MU, A, 0, 0, "(?:(ab){2}){5}M", "abababababababababababM" },
{ MU, A, 0, 0, "(?:ab|abab){1,5}M", "abababababababababababM" },
{ MU, A, 0, 0, "(?>ab|abab){1,5}M", "abababababababababababM" },
{ MU, A, 0, 0, "(?:ab|abab){1,5}?M", "abababababababababababM" },
{ MU, A, 0, 0, "(?>ab|abab){1,5}?M", "abababababababababababM" },
{ MU, A, 0, 0, "(?:(ab){1,4}?){1,3}?M", "abababababababababababababM" },
{ MU, A, 0, 0, "(?:(ab){1,4}){1,3}abababababababababababM", "ababababababababababababM" },
{ MU, A, 0, 0 | F_NOMATCH, "(?:(ab){1,4}){1,3}abababababababababababM", "abababababababababababM" },
{ MU, A, 0, 0, "(ab){4,6}?M", "abababababababM" },
/* Basic character sets. */
{ MU, A, 0, 0, "(?:\\s)+(?:\\S)+", "ab \t\xc3\xa9\xe6\x92\xad " },
{ MU, A, 0, 0, "(\\w)*(k)(\\W)?\?", "abcdef abck11" },
{ MU, A, 0, 0, "\\((\\d)+\\)\\D", "a() (83 (8)2 (9)ab" },
{ MU, A, 0, 0, "\\w(\\s|(?:\\d)*,)+\\w\\wb", "a 5, 4,, bb 5, 4,, aab" },
{ MU, A, 0, 0, "(\\v+)(\\V+)", "\x0e\xc2\x85\xe2\x80\xa8\x0b\x09\xe2\x80\xa9" },
{ MU, A, 0, 0, "(\\h+)(\\H+)", "\xe2\x80\xa8\xe2\x80\x80\x20\xe2\x80\x8a\xe2\x81\x9f\xe3\x80\x80\x09\x20\xc2\xa0\x0a" },
{ MU, A, 0, 0, "x[bcef]+", "xaxdxecbfg" },
{ MU, A, 0, 0, "x[bcdghij]+", "xaxexfxdgbjk" },
{ MU, A, 0, 0, "x[^befg]+", "xbxexacdhg" },
{ MU, A, 0, 0, "x[^bcdl]+", "xlxbxaekmd" },
{ MU, A, 0, 0, "x[^bcdghi]+", "xbxdxgxaefji" },
{ MU, A, 0, 0, "x[B-Fb-f]+", "xaxAxgxbfBFG" },
{ CMU, A, 0, 0, "\\x{e9}+", "#\xf0\x90\x90\xa8\xc3\xa8\xc3\xa9\xc3\x89\xc3\x88" },
{ CMU, A, 0, 0, "[^\\x{e9}]+", "\xc3\xa9#\xf0\x90\x90\xa8\xc3\xa8\xc3\x88\xc3\x89" },
{ MU, A, 0, 0, "[\\x02\\x7e]+", "\xc3\x81\xe1\xbf\xb8\xf0\x90\x90\xa8\x01\x02\x7e\x7f" },
{ MU, A, 0, 0, "[^\\x02\\x7e]+", "\x02\xc3\x81\xe1\xbf\xb8\xf0\x90\x90\xa8\x01\x7f\x7e" },
{ MU, A, 0, 0, "[\\x{81}-\\x{7fe}]+", "#\xe1\xbf\xb8\xf0\x90\x90\xa8\xc2\x80\xc2\x81\xdf\xbe\xdf\xbf" },
{ MU, A, 0, 0, "[^\\x{81}-\\x{7fe}]+", "\xc2\x81#\xe1\xbf\xb8\xf0\x90\x90\xa8\xc2\x80\xdf\xbf\xdf\xbe" },
{ MU, A, 0, 0, "[\\x{801}-\\x{fffe}]+", "#\xc3\xa9\xf0\x90\x90\x80\xe0\xa0\x80\xe0\xa0\x81\xef\xbf\xbe\xef\xbf\xbf" },
{ MU, A, 0, 0, "[^\\x{801}-\\x{fffe}]+", "\xe0\xa0\x81#\xc3\xa9\xf0\x90\x90\x80\xe0\xa0\x80\xef\xbf\xbf\xef\xbf\xbe" },
{ MU, A, 0, 0, "[\\x{10001}-\\x{10fffe}]+", "#\xc3\xa9\xe2\xb1\xa5\xf0\x90\x80\x80\xf0\x90\x80\x81\xf4\x8f\xbf\xbe\xf4\x8f\xbf\xbf" },
{ MU, A, 0, 0, "[^\\x{10001}-\\x{10fffe}]+", "\xf0\x90\x80\x81#\xc3\xa9\xe2\xb1\xa5\xf0\x90\x80\x80\xf4\x8f\xbf\xbf\xf4\x8f\xbf\xbe" },
{ CMU, A, 0, 0 | F_NOMATCH | F_PROPERTY, "^[\\x{100}-\\x{17f}]", " " },
{ M, A, 0, 0 | F_NOMATCH, "[^\\S\\W]{6}", "abcdefghijk" },
/* Unicode properties. */
{ MUP, A, 0, 0, "[1-5\xc3\xa9\\w]", "\xc3\xa1_" },
{ MUP, A, 0, 0 | F_PROPERTY, "[\xc3\x81\\p{Ll}]", "A_\xc3\x89\xc3\xa1" },
{ MUP, A, 0, 0, "[\\Wd-h_x-z]+", "a\xc2\xa1#_yhzdxi" },
{ MUP, A, 0, 0 | F_NOMATCH | F_PROPERTY, "[\\P{Any}]", "abc" },
{ MUP, A, 0, 0 | F_NOMATCH | F_PROPERTY, "[^\\p{Any}]", "abc" },
{ MUP, A, 0, 0 | F_NOMATCH | F_PROPERTY, "[\\P{Any}\xc3\xa1-\xc3\xa8]", "abc" },
{ MUP, A, 0, 0 | F_NOMATCH | F_PROPERTY, "[^\\p{Any}\xc3\xa1-\xc3\xa8]", "abc" },
{ MUP, A, 0, 0 | F_NOMATCH | F_PROPERTY, "[\xc3\xa1-\xc3\xa8\\P{Any}]", "abc" },
{ MUP, A, 0, 0 | F_NOMATCH | F_PROPERTY, "[^\xc3\xa1-\xc3\xa8\\p{Any}]", "abc" },
{ MUP, A, 0, 0 | F_PROPERTY, "[\xc3\xa1-\xc3\xa8\\p{Any}]", "abc" },
{ MUP, A, 0, 0 | F_PROPERTY, "[^\xc3\xa1-\xc3\xa8\\P{Any}]", "abc" },
{ MUP, A, 0, 0, "[b-\xc3\xa9\\s]", "a\xc\xe6\x92\xad" },
{ CMUP, A, 0, 0, "[\xc2\x85-\xc2\x89\xc3\x89]", "\xc2\x84\xc3\xa9" },
{ MUP, A, 0, 0, "[^b-d^&\\s]{3,}", "db^ !a\xe2\x80\xa8_ae" },
{ MUP, A, 0, 0 | F_PROPERTY, "[^\\S\\P{Any}][\\sN]{1,3}[\\P{N}]{4}", "\xe2\x80\xaa\xa N\x9\xc3\xa9_0" },
{ MU, A, 0, 0 | F_PROPERTY, "[^\\P{L}\x9!D-F\xa]{2,3}", "\x9,.DF\xa.CG\xc3\x81" },
{ CMUP, A, 0, 0, "[\xc3\xa1-\xc3\xa9_\xe2\x80\xa0-\xe2\x80\xaf]{1,5}[^\xe2\x80\xa0-\xe2\x80\xaf]", "\xc2\xa1\xc3\x89\xc3\x89\xe2\x80\xaf_\xe2\x80\xa0" },
{ MUP, A, 0, 0 | F_PROPERTY, "[\xc3\xa2-\xc3\xa6\xc3\x81-\xc3\x84\xe2\x80\xa8-\xe2\x80\xa9\xe6\x92\xad\\p{Zs}]{2,}", "\xe2\x80\xa7\xe2\x80\xa9\xe6\x92\xad \xe6\x92\xae" },
{ MUP, A, 0, 0 | F_PROPERTY, "[\\P{L&}]{2}[^\xc2\x85-\xc2\x89\\p{Ll}\\p{Lu}]{2}", "\xc3\xa9\xe6\x92\xad.a\xe6\x92\xad|\xc2\x8a#" },
{ PCRE2_UCP, 0, 0, 0 | F_PROPERTY, "[a-b\\s]{2,5}[^a]", "AB baaa" },
{ MUP, 0, 0, 0 | F_NOMATCH | F_PROPERTY, "[^\\p{Hangul}\\p{Z}]", " " },
{ MUP, 0, 0, 0, "[\\p{Lu}\\P{Latin}]+", "c\xEA\xA4\xAE,A,b" },
{ MUP, 0, 0, 0, "[\\x{a92e}\\p{Lu}\\P{Latin}]+", "c\xEA\xA4\xAE,A,b" },
{ CMUP, 0, 0, 0, "[^S]\\B", "\xe2\x80\x8a" },
{ MUP, 0, 0, 0 | F_NOMATCH, "[^[:print:]\\x{f6f6}]", "\xef\x9b\xb6" },
{ MUP, 0, 0, 0, "[[:xdigit:]\\x{6500}]#", "\xe6\x94\x80#" },
{ MUP, 0, 0, 0 | F_PROPERTY, "[\\pC\\PC]#", "A#" },
{ MUP, 0, 0, 0 | F_PROPERTY, "[\\x80-\\xff\\x{800}\\x{802}\\x{804}\\p{Cc}]", "\xdf\xbf\xe0\xa0\x80" },
/* Possible empty brackets. */
{ MU, A, 0, 0, "(?:|ab||bc|a)+d", "abcxabcabd" },
{ MU, A, 0, 0, "(|ab||bc|a)+d", "abcxabcabd" },
{ MU, A, 0, 0, "(?:|ab||bc|a)*d", "abcxabcabd" },
{ MU, A, 0, 0, "(|ab||bc|a)*d", "abcxabcabd" },
{ MU, A, 0, 0, "(?:|ab||bc|a)+?d", "abcxabcabd" },
{ MU, A, 0, 0, "(|ab||bc|a)+?d", "abcxabcabd" },
{ MU, A, 0, 0, "(?:|ab||bc|a)*?d", "abcxabcabd" },
{ MU, A, 0, 0, "(|ab||bc|a)*?d", "abcxabcabd" },
{ MU, A, 0, 0, "(((a)*?|(?:ba)+)+?|(?:|c|ca)*)*m", "abaacaccabacabalabaacaccabacabamm" },
{ MU, A, 0, 0, "(?:((?:a)*|(ba)+?)+|(|c|ca)*?)*?m", "abaacaccabacabalabaacaccabacabamm" },
/* Start offset. */
{ MU, A, 0, 3, "(\\d|(?:\\w)*\\w)+", "0ac01Hb" },
{ MU, A, 0, 4 | F_NOMATCH, "(\\w\\W\\w)+", "ab#d" },
{ MU, A, 0, 2 | F_NOMATCH, "(\\w\\W\\w)+", "ab#d" },
{ MU, A, 0, 1, "(\\w\\W\\w)+", "ab#d" },
/* Newline. */
{ M, PCRE2_NEWLINE_CRLF, 0, 0, "\\W{0,2}[^#]{3}", "\r\n#....." },
{ M, PCRE2_NEWLINE_CR, 0, 0, "\\W{0,2}[^#]{3}", "\r\n#....." },
{ M, PCRE2_NEWLINE_CRLF, 0, 0, "\\W{1,3}[^#]", "\r\n##...." },
{ MU, A, PCRE2_NO_UTF_CHECK, 1, "^.a", "\n\x80\nxa" },
{ MU, A, 0, 1, "^", "\r\n" },
{ M, PCRE2_NEWLINE_CRLF, 0, 1 | F_NOMATCH, "^", "\r\n" },
{ M, PCRE2_NEWLINE_CRLF, 0, 1, "^", "\r\na" },
/* Any character except newline or any newline. */
{ 0, PCRE2_NEWLINE_CRLF, 0, 0, ".", "\r" },
{ U, PCRE2_NEWLINE_CRLF, 0, 0, ".(.).", "a\xc3\xa1\r\n\n\r\r" },
{ 0, PCRE2_NEWLINE_ANYCRLF, 0, 0, ".(.)", "a\rb\nc\r\n\xc2\x85\xe2\x80\xa8" },
{ U, PCRE2_NEWLINE_ANYCRLF, 0, 0, ".(.)", "a\rb\nc\r\n\xc2\x85\xe2\x80\xa8" },
{ U, PCRE2_NEWLINE_ANY, 0, 0, "(.).", "a\rb\nc\r\n\xc2\x85\xe2\x80\xa9$de" },
{ U, PCRE2_NEWLINE_ANYCRLF, 0, 0 | F_NOMATCH, ".(.).", "\xe2\x80\xa8\nb\r" },
{ 0, PCRE2_NEWLINE_ANY, 0, 0, "(.)(.)", "#\x85#\r#\n#\r\n#\x84" },
{ U, PCRE2_NEWLINE_ANY, 0, 0, "(.+)#", "#\rMn\xc2\x85#\n###" },
{ 0, BSR(PCRE2_BSR_ANYCRLF), 0, 0, "\\R", "\r" },
{ 0, BSR(PCRE2_BSR_ANYCRLF), 0, 0, "\\R", "\x85#\r\n#" },
{ U, BSR(PCRE2_BSR_UNICODE), 0, 0, "\\R", "ab\xe2\x80\xa8#c" },
{ U, BSR(PCRE2_BSR_UNICODE), 0, 0, "\\R", "ab\r\nc" },
{ U, PCRE2_NEWLINE_CRLF | BSR(PCRE2_BSR_UNICODE), 0, 0, "(\\R.)+", "\xc2\x85\r\n#\xe2\x80\xa8\n\r\n\r" },
{ MU, A, 0, 0 | F_NOMATCH, "\\R+", "ab" },
{ MU, A, 0, 0, "\\R+", "ab\r\n\r" },
{ MU, A, 0, 0, "\\R*", "ab\r\n\r" },
{ MU, A, 0, 0, "\\R*", "\r\n\r" },
{ M, A, 0, 0, "\\R+\x85", "\r\n\n\r#\r\x85\n" },
{ MU, A, 0, 0, "\\R{2,4}", "\r\nab\r\r" },
{ MU, A, 0, 0, "\\R{2,4}", "\r\nab\n\n\n\r\r\r" },
{ MU, A, 0, 0, "\\R{2,}", "\r\nab\n\n\n\r\r\r" },
{ MU, A, 0, 0, "\\R{0,3}", "\r\n\r\n\r\n\r\n\r\n" },
{ MU, A, 0, 0, "\\R{2,4}\n", "\r\n\nab\r\r\nab\r\r\n\n" },
{ MU, A, 0, 0, "\\R{2,4}\n", "\r\n\nab\n\n\n\r\r\n" },
{ MU, A, 0, 0, "\\R{3,}\n", "\r\n\r\n\nab\n\n\n\r\r\n\n" },
{ MU, A, 0, 0, "\\R{0,3}\n", "\r\n\r\n\r\n\n" },
{ MU, A, 0, 0, "\\R{0,3}\n", "\r\n\r\n\r\n\r" },
{ MU, A, 0, 0, "(\\R{0,3}\n;)+", "\r\n\r\n\r\n\r\n\n;\n;\n\n;\n\n\n;\n\n\n\n\n;" },
{ MU, A, 0, 0 | F_NOMATCH, "\\R+\\R\\R", "\r\n\r\n" },
{ MU, A, 0, 0, "\\R+\\R\\R", "\r\r\r" },
{ MU, A, 0, 0, "\\R*\\R\\R", "\n\r" },
{ MU, A, 0, 0 | F_NOMATCH, "\\R{2,4}\\R\\R", "\r\r\r" },
{ MU, A, 0, 0, "\\R{2,4}\\R\\R", "\r\r\r\r" },
/* Atomic groups (no fallback from "next" direction). */
{ MU, A, 0, 0 | F_NOMATCH, "(?>ab)ab", "bab" },
{ MU, A, 0, 0 | F_NOMATCH, "(?>(ab))ab", "bab" },
{ MU, A, 0, 0, "(?>ab)+abc(?>de)*def(?>gh)?ghe(?>ij)+?k(?>lm)*?n(?>op)?\?op",
"bababcdedefgheijijklmlmnop" },
{ MU, A, 0, 0, "(?>a(b)+a|(ab)?\?(b))an", "abban" },
{ MU, A, 0, 0, "(?>ab+a|(?:ab)?\?b)an", "abban" },
{ MU, A, 0, 0, "((?>ab|ad|)*?)(?>|c)*abad", "abababcababad" },
{ MU, A, 0, 0, "(?>(aa|b|)*+(?>(##)|###)*d|(aa)(?>(baa)?)m)", "aabaa#####da" },
{ MU, A, 0, 0, "((?>a|)+?)b", "aaacaaab" },
{ MU, A, 0, 0, "(?>x|)*$", "aaa" },
{ MU, A, 0, 0, "(?>(x)|)*$", "aaa" },
{ MU, A, 0, 0, "(?>x|())*$", "aaa" },
{ MU, A, 0, 0, "((?>[cxy]a|[a-d])*?)b", "aaa+ aaab" },
{ MU, A, 0, 0, "((?>[cxy](a)|[a-d])*?)b", "aaa+ aaab" },
{ MU, A, 0, 0, "(?>((?>(a+))))bab|(?>((?>(a+))))bb", "aaaabaaabaabab" },
{ MU, A, 0, 0, "(?>(?>a+))bab|(?>(?>a+))bb", "aaaabaaabaabab" },
{ MU, A, 0, 0, "(?>(a)c|(?>(c)|(a))a)b*?bab", "aaaabaaabaabab" },
{ MU, A, 0, 0, "(?>ac|(?>c|a)a)b*?bab", "aaaabaaabaabab" },
{ MU, A, 0, 0, "(?>(b)b|(a))*b(?>(c)|d)?x", "ababcaaabdbx" },
{ MU, A, 0, 0, "(?>bb|a)*b(?>c|d)?x", "ababcaaabdbx" },
{ MU, A, 0, 0, "(?>(bb)|a)*b(?>c|(d))?x", "ababcaaabdbx" },
{ MU, A, 0, 0, "(?>(a))*?(?>(a))+?(?>(a))??x", "aaaaaacccaaaaabax" },
{ MU, A, 0, 0, "(?>a)*?(?>a)+?(?>a)??x", "aaaaaacccaaaaabax" },
{ MU, A, 0, 0, "(?>(a)|)*?(?>(a)|)+?(?>(a)|)??x", "aaaaaacccaaaaabax" },
{ MU, A, 0, 0, "(?>a|)*?(?>a|)+?(?>a|)??x", "aaaaaacccaaaaabax" },
{ MU, A, 0, 0, "(?>a(?>(a{0,2}))*?b|aac)+b", "aaaaaaacaaaabaaaaacaaaabaacaaabb" },
{ CM, A, 0, 0, "(?>((?>a{32}|b+|(a*))?(?>c+|d*)?\?)+e)+?f", "aaccebbdde bbdaaaccebbdee bbdaaaccebbdeef" },
{ MU, A, 0, 0, "(?>(?:(?>aa|a||x)+?b|(?>aa|a||(x))+?c)?(?>[ad]{0,2})*?d)+d", "aaacdbaabdcabdbaaacd aacaabdbdcdcaaaadaabcbaadd" },
{ MU, A, 0, 0, "(?>(?:(?>aa|a||(x))+?b|(?>aa|a||x)+?c)?(?>[ad]{0,2})*?d)+d", "aaacdbaabdcabdbaaacd aacaabdbdcdcaaaadaabcbaadd" },
{ MU, A, 0, 0 | F_PROPERTY, "\\X", "\xcc\x8d\xcc\x8d" },
{ MU, A, 0, 0 | F_PROPERTY, "\\X", "\xcc\x8d\xcc\x8d#\xcc\x8d\xcc\x8d" },
{ MU, A, 0, 0 | F_PROPERTY, "\\X+..", "\xcc\x8d#\xcc\x8d#\xcc\x8d\xcc\x8d" },
{ MU, A, 0, 0 | F_PROPERTY, "\\X{2,4}", "abcdef" },
{ MU, A, 0, 0 | F_PROPERTY, "\\X{2,4}?", "abcdef" },
{ MU, A, 0, 0 | F_NOMATCH | F_PROPERTY, "\\X{2,4}..", "#\xcc\x8d##" },
{ MU, A, 0, 0 | F_PROPERTY, "\\X{2,4}..", "#\xcc\x8d#\xcc\x8d##" },
{ MU, A, 0, 0, "(c(ab)?+ab)+", "cabcababcab" },
{ MU, A, 0, 0, "(?>(a+)b)+aabab", "aaaabaaabaabab" },
{ MU, A, 0, 0 | F_NOMATCH, "(?>a*|)a", "aaa" },
/* Possessive quantifiers. */
{ MU, A, 0, 0, "(?:a|b)++m", "mababbaaxababbaam" },
{ MU, A, 0, 0, "(?:a|b)*+m", "mababbaaxababbaam" },
{ MU, A, 0, 0, "(?:a|b)*+m", "ababbaaxababbaam" },
{ MU, A, 0, 0, "(a|b)++m", "mababbaaxababbaam" },
{ MU, A, 0, 0, "(a|b)*+m", "mababbaaxababbaam" },
{ MU, A, 0, 0, "(a|b)*+m", "ababbaaxababbaam" },
{ MU, A, 0, 0, "(a|b(*ACCEPT))++m", "maaxab" },
{ MU, A, 0, 0, "(?:b*)++m", "bxbbxbbbxm" },
{ MU, A, 0, 0, "(?:b*)++m", "bxbbxbbbxbbm" },
{ MU, A, 0, 0, "(?:b*)*+m", "bxbbxbbbxm" },
{ MU, A, 0, 0, "(?:b*)*+m", "bxbbxbbbxbbm" },
{ MU, A, 0, 0, "(b*)++m", "bxbbxbbbxm" },
{ MU, A, 0, 0, "(b*)++m", "bxbbxbbbxbbm" },
{ MU, A, 0, 0, "(b*)*+m", "bxbbxbbbxm" },
{ MU, A, 0, 0, "(b*)*+m", "bxbbxbbbxbbm" },
{ MU, A, 0, 0, "(?:a|(b))++m", "mababbaaxababbaam" },
{ MU, A, 0, 0, "(?:(a)|b)*+m", "mababbaaxababbaam" },
{ MU, A, 0, 0, "(?:(a)|(b))*+m", "ababbaaxababbaam" },
{ MU, A, 0, 0, "(a|(b))++m", "mababbaaxababbaam" },
{ MU, A, 0, 0, "((a)|b)*+m", "mababbaaxababbaam" },
{ MU, A, 0, 0, "((a)|(b))*+m", "ababbaaxababbaam" },
{ MU, A, 0, 0, "(a|(b)(*ACCEPT))++m", "maaxab" },
{ MU, A, 0, 0, "(?:(b*))++m", "bxbbxbbbxm" },
{ MU, A, 0, 0, "(?:(b*))++m", "bxbbxbbbxbbm" },
{ MU, A, 0, 0, "(?:(b*))*+m", "bxbbxbbbxm" },
{ MU, A, 0, 0, "(?:(b*))*+m", "bxbbxbbbxbbm" },
{ MU, A, 0, 0, "((b*))++m", "bxbbxbbbxm" },
{ MU, A, 0, 0, "((b*))++m", "bxbbxbbbxbbm" },
{ MU, A, 0, 0, "((b*))*+m", "bxbbxbbbxm" },
{ MU, A, 0, 0, "((b*))*+m", "bxbbxbbbxbbm" },
{ MU, A, 0, 0, "(A)*+$", "ABC" },
{ MU, A, 0, 0 | F_NOMATCH, "(?>(b{2,4}))(?:(?:(aa|c))++m|(?:(aa|c))+n)", "bbaacaaccaaaacxbbbmbn" },
{ MU, A, 0, 0, "((?:b)++a)+(cd)*+m", "bbababbacdcdnbbababbacdcdm" },
{ MU, A, 0, 0, "((?:(b))++a)+((c)d)*+m", "bbababbacdcdnbbababbacdcdm" },
{ MU, A, 0, 0, "(?:(?:(?:ab)*+k)++(?:n(?:cd)++)*+)*+m", "ababkkXababkkabkncXababkkabkncdcdncdXababkkabkncdcdncdkkabkncdXababkkabkncdcdncdkkabkncdm" },
{ MU, A, 0, 0, "(?:((ab)*+(k))++(n(?:c(d))++)*+)*+m", "ababkkXababkkabkncXababkkabkncdcdncdXababkkabkncdcdncdkkabkncdXababkkabkncdcdncdkkabkncdm" },
/* Back references. */
{ MU, A, 0, 0, "(aa|bb)(\\1*)(ll|)(\\3*)bbbbbbc", "aaaaaabbbbbbbbc" },
{ CMU, A, 0, 0, "(aa|bb)(\\1+)(ll|)(\\3+)bbbbbbc", "bBbbBbCbBbbbBbbcbbBbbbBBbbC" },
{ CM, A, 0, 0, "(a{2,4})\\1", "AaAaaAaA" },
{ MU, A, 0, 0, "(aa|bb)(\\1?)aa(\\1?)(ll|)(\\4+)bbc", "aaaaaaaabbaabbbbaabbbbc" },
{ MU, A, 0, 0, "(aa|bb)(\\1{0,5})(ll|)(\\3{0,5})cc", "bbxxbbbbxxaaaaaaaaaaaaaaaacc" },
{ MU, A, 0, 0, "(aa|bb)(\\1{3,5})(ll|)(\\3{3,5})cc", "bbbbbbbbbbbbaaaaaaccbbbbbbbbbbbbbbcc" },
{ MU, A, 0, 0, "(aa|bb)(\\1{3,})(ll|)(\\3{3,})cc", "bbbbbbbbbbbbaaaaaaccbbbbbbbbbbbbbbcc" },
{ MU, A, 0, 0, "(\\w+)b(\\1+)c", "GabGaGaDbGaDGaDc" },
{ MU, A, 0, 0, "(?:(aa)|b)\\1?b", "bb" },
{ CMU, A, 0, 0, "(aa|bb)(\\1*?)aa(\\1+?)", "bBBbaaAAaaAAaa" },
{ MU, A, 0, 0, "(aa|bb)(\\1*?)(dd|)cc(\\3+?)", "aaaaaccdd" },
{ CMU, A, 0, 0, "(?:(aa|bb)(\\1?\?)cc){2}(\\1?\?)", "aAaABBbbAAaAcCaAcCaA" },
{ MU, A, 0, 0, "(?:(aa|bb)(\\1{3,5}?)){2}(dd|)(\\3{3,5}?)", "aaaaaabbbbbbbbbbaaaaaaaaaaaaaa" },
{ CM, A, 0, 0, "(?:(aa|bb)(\\1{3,}?)){2}(dd|)(\\3{3,}?)", "aaaaaabbbbbbbbbbaaaaaaaaaaaaaa" },
{ MU, A, 0, 0, "(?:(aa|bb)(\\1{0,3}?)){2}(dd|)(\\3{0,3}?)b(\\1{0,3}?)(\\1{0,3})", "aaaaaaaaaaaaaaabaaaaa" },
{ MU, A, 0, 0, "(a(?:\\1|)a){3}b", "aaaaaaaaaaab" },
{ M, A, 0, 0, "(a?)b(\\1\\1*\\1+\\1?\\1*?\\1+?\\1??\\1*+\\1++\\1?+\\1{4}\\1{3,5}\\1{4,}\\1{0,5}\\1{3,5}?\\1{4,}?\\1{0,5}?\\1{3,5}+\\1{4,}+\\1{0,5}+#){2}d", "bb#b##d" },
{ MUP, A, 0, 0 | F_PROPERTY, "(\\P{N})\\1{2,}", ".www." },
{ MUP, A, 0, 0 | F_PROPERTY, "(\\P{N})\\1{0,2}", "wwwww." },
{ MUP, A, 0, 0 | F_PROPERTY, "(\\P{N})\\1{1,2}ww", "wwww" },
{ MUP, A, 0, 0 | F_PROPERTY, "(\\P{N})\\1{1,2}ww", "wwwww" },
{ PCRE2_UCP, 0, 0, 0 | F_PROPERTY, "(\\P{N})\\1{2,}", ".www." },
{ CMUP, A, 0, 0, "(\xf0\x90\x90\x80)\\1", "\xf0\x90\x90\xa8\xf0\x90\x90\xa8" },
{ MU | PCRE2_DUPNAMES, A, 0, 0 | F_NOMATCH, "\\k
{1,3}(?aa)(?bb)", "aabb" },
{ MU | PCRE2_DUPNAMES | PCRE2_MATCH_UNSET_BACKREF, A, 0, 0, "\\k{1,3}(?aa)(?bb)", "aabb" },
{ MU | PCRE2_DUPNAMES | PCRE2_MATCH_UNSET_BACKREF, A, 0, 0, "\\k*(?aa)(?bb)", "aabb" },
{ MU | PCRE2_DUPNAMES, A, 0, 0, "(?aa)(?bb)\\k{0,3}aaaaaa", "aabbaaaaaa" },
{ MU | PCRE2_DUPNAMES, A, 0, 0, "(?aa)(?bb)\\k{2,5}bb", "aabbaaaabb" },
{ MU | PCRE2_DUPNAMES, A, 0, 0, "(?:(?aa)|(?bb))\\k{0,3}m", "aaaaaaaabbbbaabbbbm" },
{ MU | PCRE2_DUPNAMES, A, 0, 0 | F_NOMATCH, "\\k{1,3}?(?aa)(?bb)", "aabb" },
{ MU | PCRE2_DUPNAMES | PCRE2_MATCH_UNSET_BACKREF, A, 0, 0, "\\k{1,3}?(?aa)(?bb)", "aabb" },
{ MU | PCRE2_DUPNAMES, A, 0, 0, "\\k*?(?aa)(?bb)", "aabb" },
{ MU | PCRE2_DUPNAMES, A, 0, 0, "(?:(?aa)|(?bb))\\k{0,3}?m", "aaaaaabbbbbbaabbbbbbbbbbm" },
{ MU | PCRE2_DUPNAMES, A, 0, 0, "(?:(?aa)|(?bb))\\k*?m", "aaaaaabbbbbbaabbbbbbbbbbm" },
{ MU | PCRE2_DUPNAMES, A, 0, 0, "(?:(?aa)|(?bb))\\k{2,3}?", "aaaabbbbaaaabbbbbbbbbb" },
{ CMU | PCRE2_DUPNAMES, A, 0, 0, "(?:(?AA)|(?BB))\\k{0,3}M", "aaaaaaaabbbbaabbbbm" },
{ CMU | PCRE2_DUPNAMES, A, 0, 0, "(?:(?AA)|(?BB))\\k{1,3}M", "aaaaaaaabbbbaabbbbm" },
{ CMU | PCRE2_DUPNAMES, A, 0, 0, "(?:(?AA)|(?BB))\\k{0,3}?M", "aaaaaabbbbbbaabbbbbbbbbbm" },
{ CMU | PCRE2_DUPNAMES, A, 0, 0, "(?:(?AA)|(?BB))\\k{2,3}?", "aaaabbbbaaaabbbbbbbbbb" },
{ MU | PCRE2_DUPNAMES, A, 0, 0, "^(?P..)(?P..)\\k{2,4}", "AaAAAaAaAaaA" },
{ MU | PCRE2_MATCH_UNSET_BACKREF, A, 0, 0, "(a)|\\1+c", "xxc" },
{ MU | PCRE2_MATCH_UNSET_BACKREF, A, 0, 0, "\\1+?()", "" },
/* Assertions. */
{ MU, A, 0, 0, "(?=xx|yy|zz)\\w{4}", "abczzdefg" },
{ MU, A, 0, 0, "(?=((\\w+)b){3}|ab)", "dbbbb ab" },
{ MU, A, 0, 0, "(?!ab|bc|cd)[a-z]{2}", "Xabcdef" },
{ MU, A, 0, 0, "(?<=aaa|aa|a)a", "aaa" },
{ MU, A, 0, 2, "(?<=aaa|aa|a)a", "aaa" },
{ M, A, 0, 0, "(?<=aaa|aa|a)a", "aaa" },
{ M, A, 0, 2, "(?<=aaa|aa|a)a", "aaa" },
{ MU, A, 0, 0, "(\\d{2})(?!\\w+c|(((\\w?)m){2}n)+|\\1)", "x5656" },
{ MU, A, 0, 0, "((?=((\\d{2,6}\\w){2,}))\\w{5,20}K){2,}", "567v09708K12l00M00 567v09708K12l00M00K45K" },
{ MU, A, 0, 0, "(?=(?:(?=\\S+a)\\w*(b)){3})\\w+\\d", "bba bbab nbbkba nbbkba0kl" },
{ MU, A, 0, 0, "(?>a(?>(b+))a(?=(..)))*?k", "acabbcabbaabacabaabbakk" },
{ MU, A, 0, 0, "((?(?=(a))a)+k)", "bbak" },
{ MU, A, 0, 0, "((?(?=a)a)+k)", "bbak" },
{ MU, A, 0, 0 | F_NOMATCH, "(?=(?>(a))m)amk", "a k" },
{ MU, A, 0, 0 | F_NOMATCH, "(?!(?>(a))m)amk", "a k" },
{ MU, A, 0, 0 | F_NOMATCH, "(?>(?=(a))am)amk", "a k" },
{ MU, A, 0, 0, "(?=(?>a|(?=(?>(b+))a|c)[a-c]+)*?m)[a-cm]+k", "aaam bbam baaambaam abbabba baaambaamk" },
{ MU, A, 0, 0, "(?> ?\?\\b(?(?=\\w{1,4}(a))m)\\w{0,8}bc){2,}?", "bca ssbc mabd ssbc mabc" },
{ MU, A, 0, 0, "(?:(?=ab)?[^n][^n])+m", "ababcdabcdcdabnababcdabcdcdabm" },
{ MU, A, 0, 0, "(?:(?=a(b))?[^n][^n])+m", "ababcdabcdcdabnababcdabcdcdabm" },
{ MU, A, 0, 0, "(?:(?=.(.))??\\1.)+m", "aabbbcbacccanaabbbcbacccam" },
{ MU, A, 0, 0, "(?:(?=.)??[a-c])+m", "abacdcbacacdcaccam" },
{ MU, A, 0, 0, "((?!a)?(?!([^a]))?)+$", "acbab" },
{ MU, A, 0, 0, "((?!a)?\?(?!([^a]))?\?)+$", "acbab" },
{ MU, A, 0, 0, "a(?=(?C)\\B(?C`x`))b", "ab" },
{ MU, A, 0, 0, "a(?!(?C)\\B(?C`x`))bb|ab", "abb" },
{ MU, A, 0, 0, "a(?=\\b|(?C)\\B(?C`x`))b", "ab" },
{ MU, A, 0, 0, "a(?!\\b|(?C)\\B(?C`x`))bb|ab", "abb" },
{ MU, A, 0, 0, "c(?(?=(?C)\\B(?C`x`))ab|a)", "cab" },
{ MU, A, 0, 0, "c(?(?!(?C)\\B(?C`x`))ab|a)", "cab" },
{ MU, A, 0, 0, "c(?(?=\\b|(?C)\\B(?C`x`))ab|a)", "cab" },
{ MU, A, 0, 0, "c(?(?!\\b|(?C)\\B(?C`x`))ab|a)", "cab" },
{ MU, A, 0, 0, "a(?=)b", "ab" },
{ MU, A, 0, 0 | F_NOMATCH, "a(?!)b", "ab" },
{ MU, A, 0, 0, "(?(?a)?(?Pb)?(?(Name)c|d)*l", "bc ddd abccabccl" },
{ MU, A, 0, 0, "(?Pa)?(?Pb)?(?(Name)c|d)+?dd", "bcabcacdb bdddd" },
{ MU, A, 0, 0, "(?Pa)?(?Pb)?(?(Name)c|d)+l", "ababccddabdbccd abcccl" },
{ MU, A, 0, 0, "((?:a|aa)(?(1)aaa))x", "aax" },
{ MU, A, 0, 0, "(?(?!)a|b)", "ab" },
{ MU, A, 0, 0, "(?(?!)a)", "ab" },
{ MU, A, 0, 0 | F_NOMATCH, "(?(?!)a|b)", "ac" },
/* Set start of match. */
{ MU, A, 0, 0, "(?:\\Ka)*aaaab", "aaaaaaaa aaaaaaabb" },
{ MU, A, 0, 0, "(?>\\Ka\\Ka)*aaaab", "aaaaaaaa aaaaaaaaaabb" },
{ MU, A, 0, 0, "a+\\K(?<=\\Gaa)a", "aaaaaa" },
{ MU, A, PCRE2_NOTEMPTY, 0 | F_NOMATCH, "a\\K(*ACCEPT)b", "aa" },
{ MU, A, PCRE2_NOTEMPTY_ATSTART, 0, "a\\K(*ACCEPT)b", "aa" },
/* First line. */
{ MU | PCRE2_FIRSTLINE, A, 0, 0 | F_PROPERTY, "\\p{Any}a", "bb\naaa" },
{ MU | PCRE2_FIRSTLINE, A, 0, 0 | F_NOMATCH | F_PROPERTY, "\\p{Any}a", "bb\r\naaa" },
{ MU | PCRE2_FIRSTLINE, A, 0, 0, "(?<=a)", "a" },
{ MU | PCRE2_FIRSTLINE, A, 0, 0 | F_NOMATCH, "[^a][^b]", "ab" },
{ MU | PCRE2_FIRSTLINE, A, 0, 0 | F_NOMATCH, "a", "\na" },
{ MU | PCRE2_FIRSTLINE, A, 0, 0 | F_NOMATCH, "[abc]", "\na" },
{ MU | PCRE2_FIRSTLINE, A, 0, 0 | F_NOMATCH, "^a", "\na" },
{ MU | PCRE2_FIRSTLINE, A, 0, 0 | F_NOMATCH, "^(?<=\n)", "\na" },
{ MU | PCRE2_FIRSTLINE, A, 0, 0, "\xf0\x90\x90\x80", "\xf0\x90\x90\x80" },
{ MU | PCRE2_FIRSTLINE, PCRE2_NEWLINE_ANY, 0, 0 | F_NOMATCH, "#", "\xc2\x85#" },
{ M | PCRE2_FIRSTLINE, PCRE2_NEWLINE_ANY, 0, 0 | F_NOMATCH, "#", "\x85#" },
{ MU | PCRE2_FIRSTLINE, PCRE2_NEWLINE_ANY, 0, 0 | F_NOMATCH, "^#", "\xe2\x80\xa8#" },
{ MU | PCRE2_FIRSTLINE, PCRE2_NEWLINE_CRLF, 0, 0 | F_PROPERTY, "\\p{Any}", "\r\na" },
{ MU | PCRE2_FIRSTLINE, PCRE2_NEWLINE_CRLF, 0, 0, ".", "\r" },
{ MU | PCRE2_FIRSTLINE, PCRE2_NEWLINE_CRLF, 0, 0, "a", "\ra" },
{ MU | PCRE2_FIRSTLINE, PCRE2_NEWLINE_CRLF, 0, 0 | F_NOMATCH, "ba", "bbb\r\nba" },
{ MU | PCRE2_FIRSTLINE, PCRE2_NEWLINE_CRLF, 0, 0 | F_NOMATCH | F_PROPERTY, "\\p{Any}{4}|a", "\r\na" },
{ MU | PCRE2_FIRSTLINE, PCRE2_NEWLINE_CRLF, 0, 1, ".", "\r\n" },
{ PCRE2_FIRSTLINE | PCRE2_DOTALL, PCRE2_NEWLINE_LF, 0, 0 | F_NOMATCH, "ab.", "ab" },
{ MU | PCRE2_FIRSTLINE, A, 0, 1 | F_NOMATCH, "^[a-d0-9]", "\nxx\nd" },
{ PCRE2_FIRSTLINE | PCRE2_DOTALL, PCRE2_NEWLINE_ANY, 0, 0, "....a", "012\n0a" },
{ MU | PCRE2_FIRSTLINE, A, 0, 0, "[aC]", "a" },
/* Recurse. */
{ MU, A, 0, 0, "(a)(?1)", "aa" },
{ MU, A, 0, 0, "((a))(?1)", "aa" },
{ MU, A, 0, 0, "(b|a)(?1)", "aa" },
{ MU, A, 0, 0, "(b|(a))(?1)", "aa" },
{ MU, A, 0, 0 | F_NOMATCH, "((a)(b)(?:a*))(?1)", "aba" },
{ MU, A, 0, 0, "((a)(b)(?:a*))(?1)", "abab" },
{ MU, A, 0, 0, "((a+)c(?2))b(?1)", "aacaabaca" },
{ MU, A, 0, 0, "((?2)b|(a)){2}(?1)", "aabab" },
{ MU, A, 0, 0, "(?1)(a)*+(?2)(b(?1))", "aababa" },
{ MU, A, 0, 0, "(?1)(((a(*ACCEPT)))b)", "axaa" },
{ MU, A, 0, 0, "(?1)(?(DEFINE) (((ac(*ACCEPT)))b) )", "akaac" },
{ MU, A, 0, 0, "(a+)b(?1)b\\1", "abaaabaaaaa" },
{ MU, A, 0, 0, "(?(DEFINE)(aa|a))(?1)ab", "aab" },
{ MU, A, 0, 0, "(?(DEFINE)(a\\Kb))(?1)+ababc", "abababxabababc" },
{ MU, A, 0, 0, "(a\\Kb)(?1)+ababc", "abababxababababc" },
{ MU, A, 0, 0 | F_NOMATCH, "(a\\Kb)(?1)+ababc", "abababxababababxc" },
{ MU, A, 0, 0, "b|<(?R)*>", "<" },
{ MU, A, 0, 0, "(a\\K){0}(?:(?1)b|ac)", "ac" },
{ MU, A, 0, 0, "(?(DEFINE)(a(?2)|b)(b(?1)|(a)))(?:(?1)|(?2))m", "ababababnababababaam" },
{ MU, A, 0, 0, "(a)((?(R)a|b))(?2)", "aabbabaa" },
{ MU, A, 0, 0, "(a)((?(R2)a|b))(?2)", "aabbabaa" },
{ MU, A, 0, 0, "(a)((?(R1)a|b))(?2)", "ababba" },
{ MU, A, 0, 0, "(?(R0)aa|bb(?R))", "abba aabb bbaa" },
{ MU, A, 0, 0, "((?(R)(?:aaaa|a)|(?:(aaaa)|(a)))+)(?1)$", "aaaaaaaaaa aaaa" },
{ MU, A, 0, 0, "(?Pa(?(R&Name)a|b))(?1)", "aab abb abaa" },
{ MU, A, 0, 0, "((?(R)a|(?1)){3})", "XaaaaaaaaaX" },
{ MU, A, 0, 0, "((?:(?(R)a|(?1))){3})", "XaaaaaaaaaX" },
{ MU, A, 0, 0, "((?(R)a|(?1)){1,3})aaaaaa", "aaaaaaaaXaaaaaaaaa" },
{ MU, A, 0, 0, "((?(R)a|(?1)){1,3}?)M", "aaaM" },
{ MU, A, 0, 0, "((.)(?:.|\\2(?1))){0}#(?1)#", "#aabbccdde# #aabbccddee#" },
{ MU, A, 0, 0, "((.)(?:\\2|\\2{4}b)){0}#(?:(?1))+#", "#aaaab# #aaaaab#" },
{ MU, A, 0, 0 | F_NOMATCH, "(?1)$((.|\\2xx){1,2})", "abc" },
/* 16 bit specific tests. */
{ CM, A, 0, 0 | F_FORCECONV, "\xc3\xa1", "\xc3\x81\xc3\xa1" },
{ CM, A, 0, 0 | F_FORCECONV, "\xe1\xbd\xb8", "\xe1\xbf\xb8\xe1\xbd\xb8" },
{ CM, A, 0, 0 | F_FORCECONV, "[\xc3\xa1]", "\xc3\x81\xc3\xa1" },
{ CM, A, 0, 0 | F_FORCECONV, "[\xe1\xbd\xb8]", "\xe1\xbf\xb8\xe1\xbd\xb8" },
{ CM, A, 0, 0 | F_FORCECONV, "[a-\xed\xb0\x80]", "A" },
{ CM, A, 0, 0 | F_NO8 | F_FORCECONV, "[a-\\x{dc00}]", "B" },
{ CM, A, 0, 0 | F_NO8 | F_NOMATCH | F_FORCECONV, "[b-\\x{dc00}]", "a" },
{ CM, A, 0, 0 | F_NO8 | F_FORCECONV, "\xed\xa0\x80\\x{d800}\xed\xb0\x80\\x{dc00}", "\xed\xa0\x80\xed\xa0\x80\xed\xb0\x80\xed\xb0\x80" },
{ CM, A, 0, 0 | F_NO8 | F_FORCECONV, "[\xed\xa0\x80\\x{d800}]{1,2}?[\xed\xb0\x80\\x{dc00}]{1,2}?#", "\xed\xa0\x80\xed\xa0\x80\xed\xb0\x80\xed\xb0\x80#" },
{ CM, A, 0, 0 | F_FORCECONV, "[\xed\xa0\x80\xed\xb0\x80#]{0,3}(?<=\xed\xb0\x80.)", "\xed\xa0\x80#\xed\xa0\x80##\xed\xb0\x80\xed\xa0\x80" },
{ CM, A, 0, 0 | F_FORCECONV, "[\xed\xa0\x80-\xed\xb3\xbf]", "\xed\x9f\xbf\xed\xa0\x83" },
{ CM, A, 0, 0 | F_FORCECONV, "[\xed\xa0\x80-\xed\xb3\xbf]", "\xed\xb4\x80\xed\xb3\xb0" },
{ CM, A, 0, 0 | F_NO8 | F_FORCECONV, "[\\x{d800}-\\x{dcff}]", "\xed\x9f\xbf\xed\xa0\x83" },
{ CM, A, 0, 0 | F_NO8 | F_FORCECONV, "[\\x{d800}-\\x{dcff}]", "\xed\xb4\x80\xed\xb3\xb0" },
{ CM, A, 0, 0 | F_FORCECONV, "[\xed\xa0\x80-\xef\xbf\xbf]+[\x1-\xed\xb0\x80]+#", "\xed\xa0\x85\xc3\x81\xed\xa0\x85\xef\xbf\xb0\xc2\x85\xed\xa9\x89#" },
{ CM, A, 0, 0 | F_FORCECONV, "[\xed\xa0\x80][\xed\xb0\x80]{2,}", "\xed\xa0\x80\xed\xb0\x80\xed\xa0\x80\xed\xb0\x80\xed\xb0\x80\xed\xb0\x80" },
{ M, A, 0, 0 | F_FORCECONV, "[^\xed\xb0\x80]{3,}?", "##\xed\xb0\x80#\xed\xb0\x80#\xc3\x89#\xed\xb0\x80" },
{ M, A, 0, 0 | F_NO8 | F_FORCECONV, "[^\\x{dc00}]{3,}?", "##\xed\xb0\x80#\xed\xb0\x80#\xc3\x89#\xed\xb0\x80" },
{ CM, A, 0, 0 | F_FORCECONV, ".\\B.", "\xed\xa0\x80\xed\xb0\x80" },
{ CM, A, 0, 0 | F_FORCECONV, "\\D+(?:\\d+|.)\\S+(?:\\s+|.)\\W+(?:\\w+|.)\xed\xa0\x80\xed\xa0\x80", "\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80" },
{ CM, A, 0, 0 | F_FORCECONV, "\\d*\\s*\\w*\xed\xa0\x80\xed\xa0\x80", "\xed\xa0\x80\xed\xa0\x80" },
{ CM, A, 0, 0 | F_FORCECONV | F_NOMATCH, "\\d*?\\D*?\\s*?\\S*?\\w*?\\W*?##", "\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80\xed\xa0\x80#" },
{ CM | PCRE2_EXTENDED, A, 0, 0 | F_FORCECONV, "\xed\xa0\x80 \xed\xb0\x80 !", "\xed\xa0\x80\xed\xb0\x80!" },
{ CM, A, 0, 0 | F_FORCECONV, "\xed\xa0\x80+#[^#]+\xed\xa0\x80", "\xed\xa0\x80#a\xed\xa0\x80" },
{ CM, A, 0, 0 | F_FORCECONV, "(\xed\xa0\x80+)#\\1", "\xed\xa0\x80\xed\xa0\x80#\xed\xa0\x80\xed\xa0\x80" },
{ M, PCRE2_NEWLINE_ANY, 0, 0 | F_NO8 | F_FORCECONV, "^-", "a--\xe2\x80\xa8--" },
{ 0, BSR(PCRE2_BSR_UNICODE), 0, 0 | F_NO8 | F_FORCECONV, "\\R", "ab\xe2\x80\xa8" },
{ 0, 0, 0, 0 | F_NO8 | F_FORCECONV, "\\v", "ab\xe2\x80\xa9" },
{ 0, 0, 0, 0 | F_NO8 | F_FORCECONV, "\\h", "ab\xe1\xa0\x8e" },
{ 0, 0, 0, 0 | F_NO8 | F_FORCECONV, "\\v+?\\V+?#", "\xe2\x80\xa9\xe2\x80\xa9\xef\xbf\xbf\xef\xbf\xbf#" },
{ 0, 0, 0, 0 | F_NO8 | F_FORCECONV, "\\h+?\\H+?#", "\xe1\xa0\x8e\xe1\xa0\x8e\xef\xbf\xbf\xef\xbf\xbf#" },
/* Partial matching. */
{ MU, A, PCRE2_PARTIAL_SOFT, 0, "ab", "a" },
{ MU, A, PCRE2_PARTIAL_SOFT, 0, "ab|a", "a" },
{ MU, A, PCRE2_PARTIAL_HARD, 0, "ab|a", "a" },
{ MU, A, PCRE2_PARTIAL_SOFT, 0, "\\b#", "a" },
{ MU, A, PCRE2_PARTIAL_SOFT, 0, "(?<=a)b", "a" },
{ MU, A, PCRE2_PARTIAL_SOFT, 0, "abc|(?<=xxa)bc", "xxab" },
{ MU, A, PCRE2_PARTIAL_SOFT, 0, "a\\B", "a" },
{ MU, A, PCRE2_PARTIAL_HARD, 0, "a\\b", "a" },
{ M | PCRE2_DUPNAMES, A, PCRE2_PARTIAL_HARD, 0, "^(?P..)(?P..)\\k{2,4}", "AaAAAaAaAaA" },
{ M | PCRE2_DUPNAMES, A, PCRE2_PARTIAL_HARD, 0, "^(?P..)(?P..)\\k{2,4}", "AaAAAaAaAaa" },
/* (*MARK) verb. */
{ MU, A, 0, 0, "a(*MARK:aa)a", "ababaa" },
{ MU, A, 0, 0 | F_NOMATCH, "a(*:aa)a", "abab" },
{ MU, A, 0, 0, "a(*:aa)(b(*:bb)b|bc)", "abc" },
{ MU, A, 0, 0 | F_NOMATCH, "a(*:1)x|b(*:2)y", "abc" },
{ MU, A, 0, 0, "(?>a(*:aa))b|ac", "ac" },
{ MU, A, 0, 0, "(?(DEFINE)(a(*:aa)))(?1)", "a" },
{ MU, A, 0, 0 | F_NOMATCH, "(?(DEFINE)((a)(*:aa)))(?1)b", "aa" },
{ MU, A, 0, 0, "(?(DEFINE)(a(*:aa)))a(?1)b|aac", "aac" },
{ MU, A, 0, 0, "(a(*:aa)){0}(?:b(?1)b|c)+c", "babbab cc" },
{ MU, A, 0, 0, "(a(*:aa)){0}(?:b(?1)b)+", "babba" },
{ MU, A, 0, 0 | F_NOMATCH, "(a(*:aa)){0}(?:b(?1)b)+", "ba" },
{ MU, A, 0, 0, "(a\\K(*:aa)){0}(?:b(?1)b|c)+c", "babbab cc" },
{ MU, A, 0, 0, "(a\\K(*:aa)){0}(?:b(?1)b)+", "babba" },
{ MU, A, 0, 0 | F_NOMATCH, "(a\\K(*:aa)){0}(?:b(?1)b)+", "ba" },
{ MU, A, 0, 0 | F_NOMATCH, "(*:mark)m", "a" },
/* (*COMMIT) verb. */
{ MU, A, 0, 0 | F_NOMATCH, "a(*COMMIT)b", "ac" },
{ MU, A, 0, 0, "aa(*COMMIT)b", "xaxaab" },
{ MU, A, 0, 0 | F_NOMATCH, "a(*COMMIT)(*:msg)b|ac", "ac" },
{ MU, A, 0, 0 | F_NOMATCH, "(a(*COMMIT)b)++", "abac" },
{ MU, A, 0, 0 | F_NOMATCH, "((a)(*COMMIT)b)++", "abac" },
{ MU, A, 0, 0 | F_NOMATCH, "(?=a(*COMMIT)b)ab|ad", "ad" },
/* (*PRUNE) verb. */
{ MU, A, 0, 0, "aa\\K(*PRUNE)b", "aaab" },
{ MU, A, 0, 0, "aa(*PRUNE:bb)b|a", "aa" },
{ MU, A, 0, 0, "(a)(a)(*PRUNE)b|(a)", "aa" },
{ MU, A, 0, 0, "(a)(a)(a)(a)(a)(a)(a)(a)(*PRUNE)b|(a)", "aaaaaaaa" },
{ MU, A, PCRE2_PARTIAL_SOFT, 0, "a(*PRUNE)a|", "a" },
{ MU, A, PCRE2_PARTIAL_SOFT, 0, "a(*PRUNE)a|m", "a" },
{ MU, A, 0, 0 | F_NOMATCH, "(?=a(*PRUNE)b)ab|ad", "ad" },
{ MU, A, 0, 0, "a(*COMMIT)(*PRUNE)d|bc", "abc" },
{ MU, A, 0, 0, "(?=a(*COMMIT)b)a(*PRUNE)c|bc", "abc" },
{ MU, A, 0, 0 | F_NOMATCH, "(*COMMIT)(?=a(*COMMIT)b)a(*PRUNE)c|bc", "abc" },
{ MU, A, 0, 0, "(?=(a)(*COMMIT)b)a(*PRUNE)c|bc", "abc" },
{ MU, A, 0, 0 | F_NOMATCH, "(*COMMIT)(?=(a)(*COMMIT)b)a(*PRUNE)c|bc", "abc" },
{ MU, A, 0, 0, "(a(*COMMIT)b){0}a(?1)(*PRUNE)c|bc", "abc" },
{ MU, A, 0, 0 | F_NOMATCH, "(a(*COMMIT)b){0}a(*COMMIT)(?1)(*PRUNE)c|bc", "abc" },
{ MU, A, 0, 0, "(a(*COMMIT)b)++(*PRUNE)d|c", "ababc" },
{ MU, A, 0, 0 | F_NOMATCH, "(*COMMIT)(a(*COMMIT)b)++(*PRUNE)d|c", "ababc" },
{ MU, A, 0, 0, "((a)(*COMMIT)b)++(*PRUNE)d|c", "ababc" },
{ MU, A, 0, 0 | F_NOMATCH, "(*COMMIT)((a)(*COMMIT)b)++(*PRUNE)d|c", "ababc" },
{ MU, A, 0, 0, "(?>a(*COMMIT)b)*abab(*PRUNE)d|ba", "ababab" },
{ MU, A, 0, 0 | F_NOMATCH, "(*COMMIT)(?>a(*COMMIT)b)*abab(*PRUNE)d|ba", "ababab" },
{ MU, A, 0, 0, "(?>a(*COMMIT)b)+abab(*PRUNE)d|ba", "ababab" },
{ MU, A, 0, 0 | F_NOMATCH, "(*COMMIT)(?>a(*COMMIT)b)+abab(*PRUNE)d|ba", "ababab" },
{ MU, A, 0, 0, "(?>a(*COMMIT)b)?ab(*PRUNE)d|ba", "aba" },
{ MU, A, 0, 0 | F_NOMATCH, "(*COMMIT)(?>a(*COMMIT)b)?ab(*PRUNE)d|ba", "aba" },
{ MU, A, 0, 0, "(?>a(*COMMIT)b)*?n(*PRUNE)d|ba", "abababn" },
{ MU, A, 0, 0 | F_NOMATCH, "(*COMMIT)(?>a(*COMMIT)b)*?n(*PRUNE)d|ba", "abababn" },
{ MU, A, 0, 0, "(?>a(*COMMIT)b)+?n(*PRUNE)d|ba", "abababn" },
{ MU, A, 0, 0 | F_NOMATCH, "(*COMMIT)(?>a(*COMMIT)b)+?n(*PRUNE)d|ba", "abababn" },
{ MU, A, 0, 0, "(?>a(*COMMIT)b)??n(*PRUNE)d|bn", "abn" },
{ MU, A, 0, 0 | F_NOMATCH, "(*COMMIT)(?>a(*COMMIT)b)??n(*PRUNE)d|bn", "abn" },
/* (*SKIP) verb. */
{ MU, A, 0, 0 | F_NOMATCH, "(?=a(*SKIP)b)ab|ad", "ad" },
{ MU, A, 0, 0, "(\\w+(*SKIP)#)", "abcd,xyz#," },
{ MU, A, 0, 0, "\\w+(*SKIP)#|mm", "abcd,xyz#," },
{ MU, A, 0, 0 | F_NOMATCH, "b+(?<=(*SKIP)#c)|b+", "#bbb" },
/* (*THEN) verb. */
{ MU, A, 0, 0, "((?:a(*THEN)|aab)(*THEN)c|a+)+m", "aabcaabcaabcaabcnacm" },
{ MU, A, 0, 0 | F_NOMATCH, "((?:a(*THEN)|aab)(*THEN)c|a+)+m", "aabcm" },
{ MU, A, 0, 0, "((?:a(*THEN)|aab)c|a+)+m", "aabcaabcnmaabcaabcm" },
{ MU, A, 0, 0, "((?:a|aab)(*THEN)c|a+)+m", "aam" },
{ MU, A, 0, 0, "((?:a(*COMMIT)|aab)(*THEN)c|a+)+m", "aam" },
{ MU, A, 0, 0, "(?(?=a(*THEN)b)ab|ad)", "ad" },
{ MU, A, 0, 0, "(?(?!a(*THEN)b)ad|add)", "add" },
{ MU, A, 0, 0 | F_NOMATCH, "(?(?=a)a(*THEN)b|ad)", "ad" },
{ MU, A, 0, 0, "(?!(?(?=a)ab|b(*THEN)d))bn|bnn", "bnn" },
{ MU, A, 0, 0, "(?=(*THEN: ))* ", " " },
{ MU, A, 0, 0, "a(*THEN)(?R) |", "a" },
{ MU, A, 0, 0 | F_NOMATCH, "(?\\w)+?)+|(?>\\w)?\?)*)?\\s", "aaaaa+ " },
{ MU, A, 0, 0, "(?:((?:(?:(?:\\w*?)+)??|(?>\\w)?|\\w*+)*)+)+?\\s", "aa+ " },
{ MU, A, 0, 0, "((a?)+)+b", "aaaaaaaaaaaa b" },
/* Deep recursion: Stack limit reached. */
{ M, A, 0, 0 | F_NOMATCH, "a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?a?aaaaaaaaaaaaaaaaaaaaaaa", "aaaaaaaaaaaaaaaaaaaaaaa" },
{ M, A, 0, 0 | F_NOMATCH, "(?:a+)+b", "aaaaaaaaaaaaaaaaaaaaaaaa b" },
{ M, A, 0, 0 | F_NOMATCH, "(?:a+?)+?b", "aaaaaaaaaaaaaaaaaaaaaaaa b" },
{ M, A, 0, 0 | F_NOMATCH, "(?:a*)*b", "aaaaaaaaaaaaaaaaaaaaaaaa b" },
{ M, A, 0, 0 | F_NOMATCH, "(?:a*?)*?b", "aaaaaaaaaaaaaaaaaaaaaaaa b" },
{ 0, 0, 0, 0, NULL, NULL }
};
#ifdef SUPPORT_PCRE2_8
static pcre2_jit_stack_8* callback8(void *arg)
{
return (pcre2_jit_stack_8 *)arg;
}
#endif
#ifdef SUPPORT_PCRE2_16
static pcre2_jit_stack_16* callback16(void *arg)
{
return (pcre2_jit_stack_16 *)arg;
}
#endif
#ifdef SUPPORT_PCRE2_32
static pcre2_jit_stack_32* callback32(void *arg)
{
return (pcre2_jit_stack_32 *)arg;
}
#endif
#ifdef SUPPORT_PCRE2_8
static pcre2_jit_stack_8 *stack8;
static pcre2_jit_stack_8 *getstack8(void)
{
if (!stack8)
stack8 = pcre2_jit_stack_create_8(1, 1024 * 1024, NULL);
return stack8;
}
static void setstack8(pcre2_match_context_8 *mcontext)
{
if (!mcontext) {
if (stack8)
pcre2_jit_stack_free_8(stack8);
stack8 = NULL;
return;
}
pcre2_jit_stack_assign_8(mcontext, callback8, getstack8());
}
#endif /* SUPPORT_PCRE2_8 */
#ifdef SUPPORT_PCRE2_16
static pcre2_jit_stack_16 *stack16;
static pcre2_jit_stack_16 *getstack16(void)
{
if (!stack16)
stack16 = pcre2_jit_stack_create_16(1, 1024 * 1024, NULL);
return stack16;
}
static void setstack16(pcre2_match_context_16 *mcontext)
{
if (!mcontext) {
if (stack16)
pcre2_jit_stack_free_16(stack16);
stack16 = NULL;
return;
}
pcre2_jit_stack_assign_16(mcontext, callback16, getstack16());
}
#endif /* SUPPORT_PCRE2_16 */
#ifdef SUPPORT_PCRE2_32
static pcre2_jit_stack_32 *stack32;
static pcre2_jit_stack_32 *getstack32(void)
{
if (!stack32)
stack32 = pcre2_jit_stack_create_32(1, 1024 * 1024, NULL);
return stack32;
}
static void setstack32(pcre2_match_context_32 *mcontext)
{
if (!mcontext) {
if (stack32)
pcre2_jit_stack_free_32(stack32);
stack32 = NULL;
return;
}
pcre2_jit_stack_assign_32(mcontext, callback32, getstack32());
}
#endif /* SUPPORT_PCRE2_32 */
#ifdef SUPPORT_PCRE2_16
static int convert_utf8_to_utf16(PCRE2_SPTR8 input, PCRE2_UCHAR16 *output, int *offsetmap, int max_length)
{
PCRE2_SPTR8 iptr = input;
PCRE2_UCHAR16 *optr = output;
unsigned int c;
if (max_length == 0)
return 0;
while (*iptr && max_length > 1) {
c = 0;
if (offsetmap)
*offsetmap++ = (int)(iptr - (unsigned char*)input);
if (*iptr < 0xc0)
c = *iptr++;
else if (!(*iptr & 0x20)) {
c = ((iptr[0] & 0x1f) << 6) | (iptr[1] & 0x3f);
iptr += 2;
} else if (!(*iptr & 0x10)) {
c = ((iptr[0] & 0x0f) << 12) | ((iptr[1] & 0x3f) << 6) | (iptr[2] & 0x3f);
iptr += 3;
} else if (!(*iptr & 0x08)) {
c = ((iptr[0] & 0x07) << 18) | ((iptr[1] & 0x3f) << 12) | ((iptr[2] & 0x3f) << 6) | (iptr[3] & 0x3f);
iptr += 4;
}
if (c < 65536) {
*optr++ = c;
max_length--;
} else if (max_length <= 2) {
*optr = '\0';
return (int)(optr - output);
} else {
c -= 0x10000;
*optr++ = 0xd800 | ((c >> 10) & 0x3ff);
*optr++ = 0xdc00 | (c & 0x3ff);
max_length -= 2;
if (offsetmap)
offsetmap++;
}
}
if (offsetmap)
*offsetmap = (int)(iptr - (unsigned char*)input);
*optr = '\0';
return (int)(optr - output);
}
static int copy_char8_to_char16(PCRE2_SPTR8 input, PCRE2_UCHAR16 *output, int max_length)
{
PCRE2_SPTR8 iptr = input;
PCRE2_UCHAR16 *optr = output;
if (max_length == 0)
return 0;
while (*iptr && max_length > 1) {
*optr++ = *iptr++;
max_length--;
}
*optr = '\0';
return (int)(optr - output);
}
#define REGTEST_MAX_LENGTH16 4096
static PCRE2_UCHAR16 regtest_buf16[REGTEST_MAX_LENGTH16];
static int regtest_offsetmap16[REGTEST_MAX_LENGTH16];
#endif /* SUPPORT_PCRE2_16 */
#ifdef SUPPORT_PCRE2_32
static int convert_utf8_to_utf32(PCRE2_SPTR8 input, PCRE2_UCHAR32 *output, int *offsetmap, int max_length)
{
PCRE2_SPTR8 iptr = input;
PCRE2_UCHAR32 *optr = output;
unsigned int c;
if (max_length == 0)
return 0;
while (*iptr && max_length > 1) {
c = 0;
if (offsetmap)
*offsetmap++ = (int)(iptr - (unsigned char*)input);
if (*iptr < 0xc0)
c = *iptr++;
else if (!(*iptr & 0x20)) {
c = ((iptr[0] & 0x1f) << 6) | (iptr[1] & 0x3f);
iptr += 2;
} else if (!(*iptr & 0x10)) {
c = ((iptr[0] & 0x0f) << 12) | ((iptr[1] & 0x3f) << 6) | (iptr[2] & 0x3f);
iptr += 3;
} else if (!(*iptr & 0x08)) {
c = ((iptr[0] & 0x07) << 18) | ((iptr[1] & 0x3f) << 12) | ((iptr[2] & 0x3f) << 6) | (iptr[3] & 0x3f);
iptr += 4;
}
*optr++ = c;
max_length--;
}
if (offsetmap)
*offsetmap = (int)(iptr - (unsigned char*)input);
*optr = 0;
return (int)(optr - output);
}
static int copy_char8_to_char32(PCRE2_SPTR8 input, PCRE2_UCHAR32 *output, int max_length)
{
PCRE2_SPTR8 iptr = input;
PCRE2_UCHAR32 *optr = output;
if (max_length == 0)
return 0;
while (*iptr && max_length > 1) {
*optr++ = *iptr++;
max_length--;
}
*optr = '\0';
return (int)(optr - output);
}
#define REGTEST_MAX_LENGTH32 4096
static PCRE2_UCHAR32 regtest_buf32[REGTEST_MAX_LENGTH32];
static int regtest_offsetmap32[REGTEST_MAX_LENGTH32];
#endif /* SUPPORT_PCRE2_32 */
static int check_ascii(const char *input)
{
const unsigned char *ptr = (unsigned char *)input;
while (*ptr) {
if (*ptr > 127)
return 0;
ptr++;
}
return 1;
}
#define OVECTOR_SIZE 15
static int regression_tests(void)
{
struct regression_test_case *current = regression_test_cases;
int error;
PCRE2_SIZE err_offs;
int is_successful;
int is_ascii;
int total = 0;
int successful = 0;
int successful_row = 0;
int counter = 0;
int jit_compile_mode;
int utf = 0;
uint32_t disabled_options = 0;
int i;
#ifdef SUPPORT_PCRE2_8
pcre2_code_8 *re8;
pcre2_compile_context_8 *ccontext8;
pcre2_match_data_8 *mdata8_1;
pcre2_match_data_8 *mdata8_2;
pcre2_match_context_8 *mcontext8;
PCRE2_SIZE *ovector8_1 = NULL;
PCRE2_SIZE *ovector8_2 = NULL;
int return_value8[2];
#endif
#ifdef SUPPORT_PCRE2_16
pcre2_code_16 *re16;
pcre2_compile_context_16 *ccontext16;
pcre2_match_data_16 *mdata16_1;
pcre2_match_data_16 *mdata16_2;
pcre2_match_context_16 *mcontext16;
PCRE2_SIZE *ovector16_1 = NULL;
PCRE2_SIZE *ovector16_2 = NULL;
int return_value16[2];
int length16;
#endif
#ifdef SUPPORT_PCRE2_32
pcre2_code_32 *re32;
pcre2_compile_context_32 *ccontext32;
pcre2_match_data_32 *mdata32_1;
pcre2_match_data_32 *mdata32_2;
pcre2_match_context_32 *mcontext32;
PCRE2_SIZE *ovector32_1 = NULL;
PCRE2_SIZE *ovector32_2 = NULL;
int return_value32[2];
int length32;
#endif
#if defined SUPPORT_PCRE2_8
PCRE2_UCHAR8 cpu_info[128];
#elif defined SUPPORT_PCRE2_16
PCRE2_UCHAR16 cpu_info[128];
#elif defined SUPPORT_PCRE2_32
PCRE2_UCHAR32 cpu_info[128];
#endif
#if defined SUPPORT_UNICODE && ((defined(SUPPORT_PCRE2_8) + defined(SUPPORT_PCRE2_16) + defined(SUPPORT_PCRE2_32)) >= 2)
int return_value;
#endif
/* This test compares the behaviour of interpreter and JIT. Although disabling
utf or ucp may make tests fail, if the pcre2_match result is the SAME, it is
still considered successful from pcre2_jit_test point of view. */
#if defined SUPPORT_PCRE2_8
pcre2_config_8(PCRE2_CONFIG_JITTARGET, &cpu_info);
#elif defined SUPPORT_PCRE2_16
pcre2_config_16(PCRE2_CONFIG_JITTARGET, &cpu_info);
#elif defined SUPPORT_PCRE2_32
pcre2_config_32(PCRE2_CONFIG_JITTARGET, &cpu_info);
#endif
printf("Running JIT regression tests\n");
printf(" target CPU of SLJIT compiler: ");
for (i = 0; cpu_info[i]; i++)
printf("%c", (char)(cpu_info[i]));
printf("\n");
#if defined SUPPORT_PCRE2_8
pcre2_config_8(PCRE2_CONFIG_UNICODE, &utf);
#elif defined SUPPORT_PCRE2_16
pcre2_config_16(PCRE2_CONFIG_UNICODE, &utf);
#elif defined SUPPORT_PCRE2_32
pcre2_config_32(PCRE2_CONFIG_UNICODE, &utf);
#endif
if (!utf)
disabled_options |= PCRE2_UTF;
#ifdef SUPPORT_PCRE2_8
printf(" in 8 bit mode with UTF-8 %s:\n", utf ? "enabled" : "disabled");
#endif
#ifdef SUPPORT_PCRE2_16
printf(" in 16 bit mode with UTF-16 %s:\n", utf ? "enabled" : "disabled");
#endif
#ifdef SUPPORT_PCRE2_32
printf(" in 32 bit mode with UTF-32 %s:\n", utf ? "enabled" : "disabled");
#endif
while (current->pattern) {
/* printf("\nPattern: %s :\n", current->pattern); */
total++;
is_ascii = 0;
if (!(current->start_offset & F_PROPERTY))
is_ascii = check_ascii(current->pattern) && check_ascii(current->input);
if (current->match_options & PCRE2_PARTIAL_SOFT)
jit_compile_mode = PCRE2_JIT_PARTIAL_SOFT;
else if (current->match_options & PCRE2_PARTIAL_HARD)
jit_compile_mode = PCRE2_JIT_PARTIAL_HARD;
else
jit_compile_mode = PCRE2_JIT_COMPLETE;
error = 0;
#ifdef SUPPORT_PCRE2_8
re8 = NULL;
ccontext8 = pcre2_compile_context_create_8(NULL);
if (ccontext8) {
if (GET_NEWLINE(current->newline))
pcre2_set_newline_8(ccontext8, GET_NEWLINE(current->newline));
if (GET_BSR(current->newline))
pcre2_set_bsr_8(ccontext8, GET_BSR(current->newline));
if (!(current->start_offset & F_NO8)) {
re8 = pcre2_compile_8((PCRE2_SPTR8)current->pattern, PCRE2_ZERO_TERMINATED,
current->compile_options & ~disabled_options,
&error, &err_offs, ccontext8);
if (!re8 && (utf || is_ascii))
printf("\n8 bit: Cannot compile pattern \"%s\": %d\n", current->pattern, error);
}
pcre2_compile_context_free_8(ccontext8);
}
else
printf("\n8 bit: Cannot allocate compile context\n");
#endif
#ifdef SUPPORT_PCRE2_16
if ((current->compile_options & PCRE2_UTF) || (current->start_offset & F_FORCECONV))
convert_utf8_to_utf16((PCRE2_SPTR8)current->pattern, regtest_buf16, NULL, REGTEST_MAX_LENGTH16);
else
copy_char8_to_char16((PCRE2_SPTR8)current->pattern, regtest_buf16, REGTEST_MAX_LENGTH16);
re16 = NULL;
ccontext16 = pcre2_compile_context_create_16(NULL);
if (ccontext16) {
if (GET_NEWLINE(current->newline))
pcre2_set_newline_16(ccontext16, GET_NEWLINE(current->newline));
if (GET_BSR(current->newline))
pcre2_set_bsr_16(ccontext16, GET_BSR(current->newline));
if (!(current->start_offset & F_NO16)) {
re16 = pcre2_compile_16(regtest_buf16, PCRE2_ZERO_TERMINATED,
current->compile_options & ~disabled_options,
&error, &err_offs, ccontext16);
if (!re16 && (utf || is_ascii))
printf("\n16 bit: Cannot compile pattern \"%s\": %d\n", current->pattern, error);
}
pcre2_compile_context_free_16(ccontext16);
}
else
printf("\n16 bit: Cannot allocate compile context\n");
#endif
#ifdef SUPPORT_PCRE2_32
if ((current->compile_options & PCRE2_UTF) || (current->start_offset & F_FORCECONV))
convert_utf8_to_utf32((PCRE2_SPTR8)current->pattern, regtest_buf32, NULL, REGTEST_MAX_LENGTH32);
else
copy_char8_to_char32((PCRE2_SPTR8)current->pattern, regtest_buf32, REGTEST_MAX_LENGTH32);
re32 = NULL;
ccontext32 = pcre2_compile_context_create_32(NULL);
if (ccontext32) {
if (GET_NEWLINE(current->newline))
pcre2_set_newline_32(ccontext32, GET_NEWLINE(current->newline));
if (GET_BSR(current->newline))
pcre2_set_bsr_32(ccontext32, GET_BSR(current->newline));
if (!(current->start_offset & F_NO32)) {
re32 = pcre2_compile_32(regtest_buf32, PCRE2_ZERO_TERMINATED,
current->compile_options & ~disabled_options,
&error, &err_offs, ccontext32);
if (!re32 && (utf || is_ascii))
printf("\n32 bit: Cannot compile pattern \"%s\": %d\n", current->pattern, error);
}
pcre2_compile_context_free_32(ccontext32);
}
else
printf("\n32 bit: Cannot allocate compile context\n");
#endif
counter++;
if ((counter & 0x3) != 0) {
#ifdef SUPPORT_PCRE2_8
setstack8(NULL);
#endif
#ifdef SUPPORT_PCRE2_16
setstack16(NULL);
#endif
#ifdef SUPPORT_PCRE2_32
setstack32(NULL);
#endif
}
#ifdef SUPPORT_PCRE2_8
return_value8[0] = -1000;
return_value8[1] = -1000;
mdata8_1 = pcre2_match_data_create_8(OVECTOR_SIZE, NULL);
mdata8_2 = pcre2_match_data_create_8(OVECTOR_SIZE, NULL);
mcontext8 = pcre2_match_context_create_8(NULL);
if (!mdata8_1 || !mdata8_2 || !mcontext8) {
printf("\n8 bit: Cannot allocate match data\n");
pcre2_match_data_free_8(mdata8_1);
pcre2_match_data_free_8(mdata8_2);
pcre2_match_context_free_8(mcontext8);
pcre2_code_free_8(re8);
re8 = NULL;
} else {
ovector8_1 = pcre2_get_ovector_pointer_8(mdata8_1);
ovector8_2 = pcre2_get_ovector_pointer_8(mdata8_2);
for (i = 0; i < OVECTOR_SIZE * 2; ++i)
ovector8_1[i] = (PCRE2_SIZE)(-2);
for (i = 0; i < OVECTOR_SIZE * 2; ++i)
ovector8_2[i] = (PCRE2_SIZE)(-2);
pcre2_set_match_limit_8(mcontext8, 10000000);
}
if (re8) {
return_value8[1] = pcre2_match_8(re8, (PCRE2_SPTR8)current->input, strlen(current->input),
current->start_offset & OFFSET_MASK, current->match_options, mdata8_2, mcontext8);
if (pcre2_jit_compile_8(re8, jit_compile_mode)) {
printf("\n8 bit: JIT compiler does not support \"%s\"\n", current->pattern);
} else if ((counter & 0x1) != 0) {
setstack8(mcontext8);
return_value8[0] = pcre2_match_8(re8, (PCRE2_SPTR8)current->input, strlen(current->input),
current->start_offset & OFFSET_MASK, current->match_options, mdata8_1, mcontext8);
} else {
pcre2_jit_stack_assign_8(mcontext8, NULL, getstack8());
return_value8[0] = pcre2_jit_match_8(re8, (PCRE2_SPTR8)current->input, strlen(current->input),
current->start_offset & OFFSET_MASK, current->match_options, mdata8_1, mcontext8);
}
}
#endif
#ifdef SUPPORT_PCRE2_16
return_value16[0] = -1000;
return_value16[1] = -1000;
mdata16_1 = pcre2_match_data_create_16(OVECTOR_SIZE, NULL);
mdata16_2 = pcre2_match_data_create_16(OVECTOR_SIZE, NULL);
mcontext16 = pcre2_match_context_create_16(NULL);
if (!mdata16_1 || !mdata16_2 || !mcontext16) {
printf("\n16 bit: Cannot allocate match data\n");
pcre2_match_data_free_16(mdata16_1);
pcre2_match_data_free_16(mdata16_2);
pcre2_match_context_free_16(mcontext16);
pcre2_code_free_16(re16);
re16 = NULL;
} else {
ovector16_1 = pcre2_get_ovector_pointer_16(mdata16_1);
ovector16_2 = pcre2_get_ovector_pointer_16(mdata16_2);
for (i = 0; i < OVECTOR_SIZE * 2; ++i)
ovector16_1[i] = (PCRE2_SIZE)(-2);
for (i = 0; i < OVECTOR_SIZE * 2; ++i)
ovector16_2[i] = (PCRE2_SIZE)(-2);
pcre2_set_match_limit_16(mcontext16, 10000000);
}
if (re16) {
if ((current->compile_options & PCRE2_UTF) || (current->start_offset & F_FORCECONV))
length16 = convert_utf8_to_utf16((PCRE2_SPTR8)current->input, regtest_buf16, regtest_offsetmap16, REGTEST_MAX_LENGTH16);
else
length16 = copy_char8_to_char16((PCRE2_SPTR8)current->input, regtest_buf16, REGTEST_MAX_LENGTH16);
return_value16[1] = pcre2_match_16(re16, regtest_buf16, length16,
current->start_offset & OFFSET_MASK, current->match_options, mdata16_2, mcontext16);
if (pcre2_jit_compile_16(re16, jit_compile_mode)) {
printf("\n16 bit: JIT compiler does not support \"%s\"\n", current->pattern);
} else if ((counter & 0x1) != 0) {
setstack16(mcontext16);
return_value16[0] = pcre2_match_16(re16, regtest_buf16, length16,
current->start_offset & OFFSET_MASK, current->match_options, mdata16_1, mcontext16);
} else {
pcre2_jit_stack_assign_16(mcontext16, NULL, getstack16());
return_value16[0] = pcre2_jit_match_16(re16, regtest_buf16, length16,
current->start_offset & OFFSET_MASK, current->match_options, mdata16_1, mcontext16);
}
}
#endif
#ifdef SUPPORT_PCRE2_32
return_value32[0] = -1000;
return_value32[1] = -1000;
mdata32_1 = pcre2_match_data_create_32(OVECTOR_SIZE, NULL);
mdata32_2 = pcre2_match_data_create_32(OVECTOR_SIZE, NULL);
mcontext32 = pcre2_match_context_create_32(NULL);
if (!mdata32_1 || !mdata32_2 || !mcontext32) {
printf("\n32 bit: Cannot allocate match data\n");
pcre2_match_data_free_32(mdata32_1);
pcre2_match_data_free_32(mdata32_2);
pcre2_match_context_free_32(mcontext32);
pcre2_code_free_32(re32);
re32 = NULL;
} else {
ovector32_1 = pcre2_get_ovector_pointer_32(mdata32_1);
ovector32_2 = pcre2_get_ovector_pointer_32(mdata32_2);
for (i = 0; i < OVECTOR_SIZE * 2; ++i)
ovector32_1[i] = (PCRE2_SIZE)(-2);
for (i = 0; i < OVECTOR_SIZE * 2; ++i)
ovector32_2[i] = (PCRE2_SIZE)(-2);
pcre2_set_match_limit_32(mcontext32, 10000000);
}
if (re32) {
if ((current->compile_options & PCRE2_UTF) || (current->start_offset & F_FORCECONV))
length32 = convert_utf8_to_utf32((PCRE2_SPTR8)current->input, regtest_buf32, regtest_offsetmap32, REGTEST_MAX_LENGTH32);
else
length32 = copy_char8_to_char32((PCRE2_SPTR8)current->input, regtest_buf32, REGTEST_MAX_LENGTH32);
return_value32[1] = pcre2_match_32(re32, regtest_buf32, length32,
current->start_offset & OFFSET_MASK, current->match_options, mdata32_2, mcontext32);
if (pcre2_jit_compile_32(re32, jit_compile_mode)) {
printf("\n32 bit: JIT compiler does not support \"%s\"\n", current->pattern);
} else if ((counter & 0x1) != 0) {
setstack32(mcontext32);
return_value32[0] = pcre2_match_32(re32, regtest_buf32, length32,
current->start_offset & OFFSET_MASK, current->match_options, mdata32_1, mcontext32);
} else {
pcre2_jit_stack_assign_32(mcontext32, NULL, getstack32());
return_value32[0] = pcre2_jit_match_32(re32, regtest_buf32, length32,
current->start_offset & OFFSET_MASK, current->match_options, mdata32_1, mcontext32);
}
}
#endif
/* printf("[%d-%d-%d|%d-%d|%d-%d|%d-%d]%s",
return_value8[0], return_value16[0], return_value32[0],
(int)ovector8_1[0], (int)ovector8_1[1],
(int)ovector16_1[0], (int)ovector16_1[1],
(int)ovector32_1[0], (int)ovector32_1[1],
(current->compile_options & PCRE2_CASELESS) ? "C" : ""); */
/* If F_DIFF is set, just run the test, but do not compare the results.
Segfaults can still be captured. */
is_successful = 1;
if (!(current->start_offset & F_DIFF)) {
#if defined SUPPORT_UNICODE && ((defined(SUPPORT_PCRE2_8) + defined(SUPPORT_PCRE2_16) + defined(SUPPORT_PCRE2_32)) >= 2)
if (!(current->start_offset & F_FORCECONV)) {
/* All results must be the same. */
#ifdef SUPPORT_PCRE2_8
if ((return_value = return_value8[0]) != return_value8[1]) {
printf("\n8 bit: Return value differs(J8:%d,I8:%d): [%d] '%s' @ '%s'\n",
return_value8[0], return_value8[1], total, current->pattern, current->input);
is_successful = 0;
} else
#endif
#ifdef SUPPORT_PCRE2_16
if ((return_value = return_value16[0]) != return_value16[1]) {
printf("\n16 bit: Return value differs(J16:%d,I16:%d): [%d] '%s' @ '%s'\n",
return_value16[0], return_value16[1], total, current->pattern, current->input);
is_successful = 0;
} else
#endif
#ifdef SUPPORT_PCRE2_32
if ((return_value = return_value32[0]) != return_value32[1]) {
printf("\n32 bit: Return value differs(J32:%d,I32:%d): [%d] '%s' @ '%s'\n",
return_value32[0], return_value32[1], total, current->pattern, current->input);
is_successful = 0;
} else
#endif
#if defined SUPPORT_PCRE2_8 && defined SUPPORT_PCRE2_16
if (return_value8[0] != return_value16[0]) {
printf("\n8 and 16 bit: Return value differs(J8:%d,J16:%d): [%d] '%s' @ '%s'\n",
return_value8[0], return_value16[0],
total, current->pattern, current->input);
is_successful = 0;
} else
#endif
#if defined SUPPORT_PCRE2_8 && defined SUPPORT_PCRE2_32
if (return_value8[0] != return_value32[0]) {
printf("\n8 and 32 bit: Return value differs(J8:%d,J32:%d): [%d] '%s' @ '%s'\n",
return_value8[0], return_value32[0],
total, current->pattern, current->input);
is_successful = 0;
} else
#endif
#if defined SUPPORT_PCRE2_16 && defined SUPPORT_PCRE2_32
if (return_value16[0] != return_value32[0]) {
printf("\n16 and 32 bit: Return value differs(J16:%d,J32:%d): [%d] '%s' @ '%s'\n",
return_value16[0], return_value32[0],
total, current->pattern, current->input);
is_successful = 0;
} else
#endif
if (return_value >= 0 || return_value == PCRE2_ERROR_PARTIAL) {
if (return_value == PCRE2_ERROR_PARTIAL) {
return_value = 2;
} else {
return_value *= 2;
}
#ifdef SUPPORT_PCRE2_8
return_value8[0] = return_value;
#endif
#ifdef SUPPORT_PCRE2_16
return_value16[0] = return_value;
#endif
#ifdef SUPPORT_PCRE2_32
return_value32[0] = return_value;
#endif
/* Transform back the results. */
if (current->compile_options & PCRE2_UTF) {
#ifdef SUPPORT_PCRE2_16
for (i = 0; i < return_value; ++i) {
if (ovector16_1[i] != PCRE2_UNSET)
ovector16_1[i] = regtest_offsetmap16[ovector16_1[i]];
if (ovector16_2[i] != PCRE2_UNSET)
ovector16_2[i] = regtest_offsetmap16[ovector16_2[i]];
}
#endif
#ifdef SUPPORT_PCRE2_32
for (i = 0; i < return_value; ++i) {
if (ovector32_1[i] != PCRE2_UNSET)
ovector32_1[i] = regtest_offsetmap32[ovector32_1[i]];
if (ovector32_2[i] != PCRE2_UNSET)
ovector32_2[i] = regtest_offsetmap32[ovector32_2[i]];
}
#endif
}
for (i = 0; i < return_value; ++i) {
#if defined SUPPORT_PCRE2_8 && defined SUPPORT_PCRE2_16
if (ovector8_1[i] != ovector8_2[i] || ovector8_1[i] != ovector16_1[i] || ovector8_1[i] != ovector16_2[i]) {
printf("\n8 and 16 bit: Ovector[%d] value differs(J8:%d,I8:%d,J16:%d,I16:%d): [%d] '%s' @ '%s' \n",
i, (int)ovector8_1[i], (int)ovector8_2[i], (int)ovector16_1[i], (int)ovector16_2[i],
total, current->pattern, current->input);
is_successful = 0;
}
#endif
#if defined SUPPORT_PCRE2_8 && defined SUPPORT_PCRE2_32
if (ovector8_1[i] != ovector8_2[i] || ovector8_1[i] != ovector32_1[i] || ovector8_1[i] != ovector32_2[i]) {
printf("\n8 and 32 bit: Ovector[%d] value differs(J8:%d,I8:%d,J32:%d,I32:%d): [%d] '%s' @ '%s' \n",
i, (int)ovector8_1[i], (int)ovector8_2[i], (int)ovector32_1[i], (int)ovector32_2[i],
total, current->pattern, current->input);
is_successful = 0;
}
#endif
#if defined SUPPORT_PCRE2_16 && defined SUPPORT_PCRE2_32
if (ovector16_1[i] != ovector16_2[i] || ovector16_1[i] != ovector32_1[i] || ovector16_1[i] != ovector32_2[i]) {
printf("\n16 and 32 bit: Ovector[%d] value differs(J16:%d,I16:%d,J32:%d,I32:%d): [%d] '%s' @ '%s' \n",
i, (int)ovector16_1[i], (int)ovector16_2[i], (int)ovector32_1[i], (int)ovector32_2[i],
total, current->pattern, current->input);
is_successful = 0;
}
#endif
}
}
} else
#endif /* more than one of SUPPORT_PCRE2_8, SUPPORT_PCRE2_16 and SUPPORT_PCRE2_32 */
{
#ifdef SUPPORT_PCRE2_8
if (return_value8[0] != return_value8[1]) {
printf("\n8 bit: Return value differs(%d:%d): [%d] '%s' @ '%s'\n",
return_value8[0], return_value8[1], total, current->pattern, current->input);
is_successful = 0;
} else if (return_value8[0] >= 0 || return_value8[0] == PCRE2_ERROR_PARTIAL) {
if (return_value8[0] == PCRE2_ERROR_PARTIAL)
return_value8[0] = 2;
else
return_value8[0] *= 2;
for (i = 0; i < return_value8[0]; ++i)
if (ovector8_1[i] != ovector8_2[i]) {
printf("\n8 bit: Ovector[%d] value differs(%d:%d): [%d] '%s' @ '%s'\n",
i, (int)ovector8_1[i], (int)ovector8_2[i], total, current->pattern, current->input);
is_successful = 0;
}
}
#endif
#ifdef SUPPORT_PCRE2_16
if (return_value16[0] != return_value16[1]) {
printf("\n16 bit: Return value differs(%d:%d): [%d] '%s' @ '%s'\n",
return_value16[0], return_value16[1], total, current->pattern, current->input);
is_successful = 0;
} else if (return_value16[0] >= 0 || return_value16[0] == PCRE2_ERROR_PARTIAL) {
if (return_value16[0] == PCRE2_ERROR_PARTIAL)
return_value16[0] = 2;
else
return_value16[0] *= 2;
for (i = 0; i < return_value16[0]; ++i)
if (ovector16_1[i] != ovector16_2[i]) {
printf("\n16 bit: Ovector[%d] value differs(%d:%d): [%d] '%s' @ '%s'\n",
i, (int)ovector16_1[i], (int)ovector16_2[i], total, current->pattern, current->input);
is_successful = 0;
}
}
#endif
#ifdef SUPPORT_PCRE2_32
if (return_value32[0] != return_value32[1]) {
printf("\n32 bit: Return value differs(%d:%d): [%d] '%s' @ '%s'\n",
return_value32[0], return_value32[1], total, current->pattern, current->input);
is_successful = 0;
} else if (return_value32[0] >= 0 || return_value32[0] == PCRE2_ERROR_PARTIAL) {
if (return_value32[0] == PCRE2_ERROR_PARTIAL)
return_value32[0] = 2;
else
return_value32[0] *= 2;
for (i = 0; i < return_value32[0]; ++i)
if (ovector32_1[i] != ovector32_2[i]) {
printf("\n32 bit: Ovector[%d] value differs(%d:%d): [%d] '%s' @ '%s'\n",
i, (int)ovector32_1[i], (int)ovector32_2[i], total, current->pattern, current->input);
is_successful = 0;
}
}
#endif
}
}
if (is_successful) {
#ifdef SUPPORT_PCRE2_8
if (!(current->start_offset & F_NO8) && (utf || is_ascii)) {
if (return_value8[0] < 0 && !(current->start_offset & F_NOMATCH)) {
printf("8 bit: Test should match: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
if (return_value8[0] >= 0 && (current->start_offset & F_NOMATCH)) {
printf("8 bit: Test should not match: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
}
#endif
#ifdef SUPPORT_PCRE2_16
if (!(current->start_offset & F_NO16) && (utf || is_ascii)) {
if (return_value16[0] < 0 && !(current->start_offset & F_NOMATCH)) {
printf("16 bit: Test should match: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
if (return_value16[0] >= 0 && (current->start_offset & F_NOMATCH)) {
printf("16 bit: Test should not match: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
}
#endif
#ifdef SUPPORT_PCRE2_32
if (!(current->start_offset & F_NO32) && (utf || is_ascii)) {
if (return_value32[0] < 0 && !(current->start_offset & F_NOMATCH)) {
printf("32 bit: Test should match: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
if (return_value32[0] >= 0 && (current->start_offset & F_NOMATCH)) {
printf("32 bit: Test should not match: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
}
#endif
}
if (is_successful) {
#ifdef SUPPORT_PCRE2_8
if (re8 && !(current->start_offset & F_NO8) && pcre2_get_mark_8(mdata8_1) != pcre2_get_mark_8(mdata8_2)) {
printf("8 bit: Mark value mismatch: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
#endif
#ifdef SUPPORT_PCRE2_16
if (re16 && !(current->start_offset & F_NO16) && pcre2_get_mark_16(mdata16_1) != pcre2_get_mark_16(mdata16_2)) {
printf("16 bit: Mark value mismatch: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
#endif
#ifdef SUPPORT_PCRE2_32
if (re32 && !(current->start_offset & F_NO32) && pcre2_get_mark_32(mdata32_1) != pcre2_get_mark_32(mdata32_2)) {
printf("32 bit: Mark value mismatch: [%d] '%s' @ '%s'\n",
total, current->pattern, current->input);
is_successful = 0;
}
#endif
}
#ifdef SUPPORT_PCRE2_8
pcre2_code_free_8(re8);
pcre2_match_data_free_8(mdata8_1);
pcre2_match_data_free_8(mdata8_2);
pcre2_match_context_free_8(mcontext8);
#endif
#ifdef SUPPORT_PCRE2_16
pcre2_code_free_16(re16);
pcre2_match_data_free_16(mdata16_1);
pcre2_match_data_free_16(mdata16_2);
pcre2_match_context_free_16(mcontext16);
#endif
#ifdef SUPPORT_PCRE2_32
pcre2_code_free_32(re32);
pcre2_match_data_free_32(mdata32_1);
pcre2_match_data_free_32(mdata32_2);
pcre2_match_context_free_32(mcontext32);
#endif
if (is_successful) {
successful++;
successful_row++;
printf(".");
if (successful_row >= 60) {
successful_row = 0;
printf("\n");
}
} else
successful_row = 0;
fflush(stdout);
current++;
}
#ifdef SUPPORT_PCRE2_8
setstack8(NULL);
#endif
#ifdef SUPPORT_PCRE2_16
setstack16(NULL);
#endif
#ifdef SUPPORT_PCRE2_32
setstack32(NULL);
#endif
if (total == successful) {
printf("\nAll JIT regression tests are successfully passed.\n");
return 0;
} else {
printf("\nSuccessful test ratio: %d%% (%d failed)\n", successful * 100 / total, total - successful);
return 1;
}
}
#if defined SUPPORT_UNICODE
static int check_invalid_utf_result(int pattern_index, const char *type, int result,
int match_start, int match_end, PCRE2_SIZE *ovector)
{
if (match_start < 0) {
if (result != -1) {
printf("Pattern[%d] %s result is not -1.\n", pattern_index, type);
return 1;
}
return 0;
}
if (result <= 0) {
printf("Pattern[%d] %s result (%d) is not greater than 0.\n", pattern_index, type, result);
return 1;
}
if (ovector[0] != (PCRE2_SIZE)match_start) {
printf("Pattern[%d] %s ovector[0] is unexpected (%d instead of %d)\n",
pattern_index, type, (int)ovector[0], match_start);
return 1;
}
if (ovector[1] != (PCRE2_SIZE)match_end) {
printf("Pattern[%d] %s ovector[1] is unexpected (%d instead of %d)\n",
pattern_index, type, (int)ovector[1], match_end);
return 1;
}
return 0;
}
#endif /* SUPPORT_UNICODE */
#if defined SUPPORT_UNICODE && defined SUPPORT_PCRE2_8
#define UDA (PCRE2_UTF | PCRE2_DOTALL | PCRE2_ANCHORED)
#define CI (PCRE2_JIT_COMPLETE | PCRE2_JIT_INVALID_UTF)
#define CPI (PCRE2_JIT_COMPLETE | PCRE2_JIT_PARTIAL_SOFT | PCRE2_JIT_INVALID_UTF)
struct invalid_utf8_regression_test_case {
uint32_t compile_options;
int jit_compile_options;
int start_offset;
int skip_left;
int skip_right;
int match_start;
int match_end;
const char *pattern[2];
const char *input;
};
static const char invalid_utf8_newline_cr;
static const struct invalid_utf8_regression_test_case invalid_utf8_regression_test_cases[] = {
{ UDA, CI, 0, 0, 0, 0, 4, { ".", NULL }, "\xf4\x8f\xbf\xbf" },
{ UDA, CI, 0, 0, 0, 0, 4, { ".", NULL }, "\xf0\x90\x80\x80" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xf4\x90\x80\x80" },
{ UDA, CI, 0, 0, 1, -1, -1, { ".", NULL }, "\xf4\x8f\xbf\xbf" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xf0\x90\x80\x7f" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xf0\x90\x80\xc0" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xf0\x8f\xbf\xbf" },
{ UDA, CI, 0, 0, 0, 0, 3, { ".", NULL }, "\xef\xbf\xbf#" },
{ UDA, CI, 0, 0, 0, 0, 3, { ".", NULL }, "\xef\xbf\xbf" },
{ UDA, CI, 0, 0, 0, 0, 3, { ".", NULL }, "\xe0\xa0\x80#" },
{ UDA, CI, 0, 0, 0, 0, 3, { ".", NULL }, "\xe0\xa0\x80" },
{ UDA, CI, 0, 0, 2, -1, -1, { ".", NULL }, "\xef\xbf\xbf#" },
{ UDA, CI, 0, 0, 1, -1, -1, { ".", NULL }, "\xef\xbf\xbf" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xef\xbf\x7f#" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xef\xbf\xc0" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xe0\x9f\xbf#" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xe0\x9f\xbf" },
{ UDA, CI, 0, 0, 0, 0, 3, { ".", NULL }, "\xed\x9f\xbf#" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xed\xa0\x80#" },
{ UDA, CI, 0, 0, 0, 0, 3, { ".", NULL }, "\xee\x80\x80#" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xed\xbf\xbf#" },
{ UDA, CI, 0, 0, 0, 0, 2, { ".", NULL }, "\xdf\xbf##" },
{ UDA, CI, 0, 0, 0, 0, 2, { ".", NULL }, "\xdf\xbf#" },
{ UDA, CI, 0, 0, 0, 0, 2, { ".", NULL }, "\xdf\xbf" },
{ UDA, CI, 0, 0, 0, 0, 2, { ".", NULL }, "\xc2\x80##" },
{ UDA, CI, 0, 0, 0, 0, 2, { ".", NULL }, "\xc2\x80#" },
{ UDA, CI, 0, 0, 0, 0, 2, { ".", NULL }, "\xc2\x80" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xe0\x80##" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xdf\xc0##" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xe0\x80" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xdf\xc0" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xc1\xbf##" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xc1\xbf" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\x80###" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\x80" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xf8###" },
{ UDA, CI, 0, 0, 0, -1, -1, { ".", NULL }, "\xf8" },
{ UDA, CI, 0, 0, 0, 0, 1, { ".", NULL }, "\x7f" },
{ UDA, CPI, 4, 0, 0, 4, 4, { "\\B", NULL }, "\xf4\x8f\xbf\xbf#" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "\xf4\xa0\x80\x80\xf4\xa0\x80\x80" },
{ UDA, CPI, 4, 1, 1, -1, -1, { "\\B", "\\b" }, "\xf4\x8f\xbf\xbf\xf4\x8f\xbf\xbf" },
{ UDA, CPI, 4, 0, 0, 4, 4, { "\\B", NULL }, "#\xef\xbf\xbf#" },
{ UDA, CPI, 4, 0, 0, 4, 4, { "\\B", NULL }, "#\xe0\xa0\x80#" },
{ UDA, CPI, 4, 0, 0, 4, 4, { "\\B", NULL }, "\xf0\x90\x80\x80#" },
{ UDA, CPI, 4, 0, 0, 4, 4, { "\\B", NULL }, "\xf3\xbf\xbf\xbf#" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "\xf0\x8f\xbf\xbf\xf0\x8f\xbf\xbf" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "\xf5\x80\x80\x80\xf5\x80\x80\x80" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "\xf4\x90\x80\x80\xf4\x90\x80\x80" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "\xf4\x8f\xbf\xff\xf4\x8f\xbf\xff" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "\xf4\x8f\xff\xbf\xf4\x8f\xff\xbf" },
{ UDA, CPI, 4, 0, 1, -1, -1, { "\\B", "\\b" }, "\xef\x80\x80\x80\xef\x80\x80" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "\x80\x80\x80\x80\x80\x80\x80\x80" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "#\xe0\x9f\xbf\xe0\x9f\xbf#" },
{ UDA, CPI, 4, 2, 2, -1, -1, { "\\B", "\\b" }, "#\xe0\xa0\x80\xe0\xa0\x80#" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "#\xf0\x80\x80\xf0\x80\x80#" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "#\xed\xa0\x80\xed\xa0\x80#" },
{ UDA, CPI, 4, 0, 0, 4, 4, { "\\B", NULL }, "##\xdf\xbf#" },
{ UDA, CPI, 4, 2, 0, 2, 2, { "\\B", NULL }, "##\xdf\xbf#" },
{ UDA, CPI, 4, 0, 0, 4, 4, { "\\B", NULL }, "##\xc2\x80#" },
{ UDA, CPI, 4, 2, 0, 2, 2, { "\\B", NULL }, "##\xc2\x80#" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "##\xc1\xbf\xc1\xbf##" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "##\xdf\xc0\xdf\xc0##" },
{ UDA, CPI, 4, 0, 0, -1, -1, { "\\B", "\\b" }, "##\xe0\x80\xe0\x80##" },
{ UDA, CPI, 3, 0, 0, 3, 3, { "\\B", NULL }, "\xef\xbf\xbf#" },
{ UDA, CPI, 3, 0, 0, 3, 3, { "\\B", NULL }, "\xe0\xa0\x80#" },
{ UDA, CPI, 3, 0, 0, -1, -1, { "\\B", "\\b" }, "\xe0\x9f\xbf\xe0\x9f\xbf" },
{ UDA, CPI, 3, 1, 1, -1, -1, { "\\B", "\\b" }, "\xef\xbf\xbf\xef\xbf\xbf" },
{ UDA, CPI, 3, 0, 1, -1, -1, { "\\B", "\\b" }, "\xdf\x80\x80\xdf\x80" },
{ UDA, CPI, 3, 0, 0, -1, -1, { "\\B", "\\b" }, "\xef\xbf\xff\xef\xbf\xff" },
{ UDA, CPI, 3, 0, 0, -1, -1, { "\\B", "\\b" }, "\xef\xff\xbf\xef\xff\xbf" },
{ UDA, CPI, 3, 0, 0, -1, -1, { "\\B", "\\b" }, "\xed\xbf\xbf\xed\xbf\xbf" },
{ UDA, CPI, 2, 0, 0, 2, 2, { "\\B", NULL }, "\xdf\xbf#" },
{ UDA, CPI, 2, 0, 0, 2, 2, { "\\B", NULL }, "\xc2\x80#" },
{ UDA, CPI, 2, 1, 1, -1, -1, { "\\B", "\\b" }, "\xdf\xbf\xdf\xbf" },
{ UDA, CPI, 2, 0, 0, -1, -1, { "\\B", "\\b" }, "\xc1\xbf\xc1\xbf" },
{ UDA, CPI, 2, 0, 0, -1, -1, { "\\B", "\\b" }, "\xe0\x80\xe0\x80" },
{ UDA, CPI, 2, 0, 0, -1, -1, { "\\B", "\\b" }, "\xdf\xff\xdf\xff" },
{ UDA, CPI, 2, 0, 0, -1, -1, { "\\B", "\\b" }, "\xff\xbf\xff\xbf" },
{ UDA, CPI, 1, 0, 0, 1, 1, { "\\B", NULL }, "\x7f#" },
{ UDA, CPI, 1, 0, 0, 1, 1, { "\\B", NULL }, "\x01#" },
{ UDA, CPI, 1, 0, 0, -1, -1, { "\\B", "\\b" }, "\x80\x80" },
{ UDA, CPI, 1, 0, 0, -1, -1, { "\\B", "\\b" }, "\xb0\xb0" },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 0, 0, 2, { "(.)\\1", NULL }, "aA" },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 0, -1, -1, { "(.)\\1", NULL }, "a\xff" },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 0, 0, 4, { "(.)\\1", NULL }, "\xc3\xa1\xc3\x81" },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 1, -1, -1, { "(.)\\1", NULL }, "\xc3\xa1\xc3\x81" },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 0, -1, -1, { "(.)\\1", NULL }, "\xc2\x80\x80" },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 0, 0, 6, { "(.)\\1", NULL }, "\xe1\xbd\xb8\xe1\xbf\xb8" },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 1, -1, -1, { "(.)\\1", NULL }, "\xe1\xbd\xb8\xe1\xbf\xb8" },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 0, 0, 8, { "(.)\\1", NULL }, "\xf0\x90\x90\x80\xf0\x90\x90\xa8" },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 1, -1, -1, { "(.)\\1", NULL }, "\xf0\x90\x90\x80\xf0\x90\x90\xa8" },
{ UDA, CPI, 0, 0, 0, 0, 1, { "\\X", NULL }, "A" },
{ UDA, CPI, 0, 0, 0, -1, -1, { "\\X", NULL }, "\xff" },
{ UDA, CPI, 0, 0, 0, 0, 2, { "\\X", NULL }, "\xc3\xa1" },
{ UDA, CPI, 0, 0, 1, -1, -1, { "\\X", NULL }, "\xc3\xa1" },
{ UDA, CPI, 0, 0, 0, -1, -1, { "\\X", NULL }, "\xc3\x7f" },
{ UDA, CPI, 0, 0, 0, 0, 3, { "\\X", NULL }, "\xe1\xbd\xb8" },
{ UDA, CPI, 0, 0, 1, -1, -1, { "\\X", NULL }, "\xe1\xbd\xb8" },
{ UDA, CPI, 0, 0, 0, 0, 4, { "\\X", NULL }, "\xf0\x90\x90\x80" },
{ UDA, CPI, 0, 0, 1, -1, -1, { "\\X", NULL }, "\xf0\x90\x90\x80" },
{ UDA, CPI, 0, 0, 0, -1, -1, { "[^#]", NULL }, "#" },
{ UDA, CPI, 0, 0, 0, 0, 4, { "[^#]", NULL }, "\xf4\x8f\xbf\xbf" },
{ UDA, CPI, 0, 0, 0, -1, -1, { "[^#]", NULL }, "\xf4\x90\x80\x80" },
{ UDA, CPI, 0, 0, 0, -1, -1, { "[^#]", NULL }, "\xc1\x80" },
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 2, 3, { "^\\W", NULL }, " \x0a#"},
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 14, 15, { "^\\W", NULL }, " \xc0\x8a#\xe0\x80\x8a#\xf0\x80\x80\x8a#\x0a#"},
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 3, 4, { "^\\W", NULL }, " \xf8\x0a#"},
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 3, 4, { "^\\W", NULL }, " \xc3\x0a#"},
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 3, 4, { "^\\W", NULL }, " \xf1\x0a#"},
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 4, 5, { "^\\W", NULL }, " \xf2\xbf\x0a#"},
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 5, 6, { "^\\W", NULL }, " \xf2\xbf\xbf\x0a#"},
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 3, 4, { "^\\W", NULL }, " \xef\x0a#"},
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 4, 5, { "^\\W", NULL }, " \xef\xbf\x0a#"},
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 5, 6, { "^\\W", NULL }, " \x85#\xc2\x85#"},
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 7, 8, { "^\\W", NULL }, " \xe2\x80\xf8\xe2\x80\xa8#"},
{ PCRE2_UTF | PCRE2_FIRSTLINE, CI, 0, 0, 0, -1, -1, { "#", NULL }, "\xe2\x80\xf8\xe2\x80\xa8#"},
{ PCRE2_UTF | PCRE2_FIRSTLINE, CI, 0, 0, 0, 3, 4, { "#", NULL }, "\xe2\x80\xf8#\xe2\x80\xa8#"},
{ PCRE2_UTF | PCRE2_FIRSTLINE, CI, 0, 0, 0, -1, -1, { "#", NULL }, "abcd\xc2\x85#"},
{ PCRE2_UTF | PCRE2_FIRSTLINE, CI, 0, 0, 0, 1, 2, { "#", NULL }, "\x85#\xc2\x85#"},
{ PCRE2_UTF | PCRE2_FIRSTLINE, CI, 0, 0, 0, 5, 6, { "#", NULL }, "\xef,\x80,\xf8#\x0a"},
{ PCRE2_UTF | PCRE2_FIRSTLINE, CI, 0, 0, 0, -1, -1, { "#", NULL }, "\xef,\x80,\xf8\x0a#"},
{ PCRE2_UTF | PCRE2_NO_START_OPTIMIZE, CI, 0, 0, 0, 4, 8, { "#\xc7\x85#", NULL }, "\x80\x80#\xc7#\xc7\x85#" },
{ PCRE2_UTF | PCRE2_NO_START_OPTIMIZE, CI, 0, 0, 0, 7, 11, { "#\xc7\x85#", NULL }, "\x80\x80#\xc7\x80\x80\x80#\xc7\x85#" },
{ PCRE2_UTF, CI, 0, 0, 0, 4, 8, { "#\xc7\x85#", NULL }, "\x80\x80#\xc7#\xc7\x85#" },
{ PCRE2_UTF, CI, 0, 0, 0, 7, 11, { "#\xc7\x85#", NULL }, "\x80\x80#\xc7\x80\x80\x80#\xc7\x85#" },
{ PCRE2_UTF | PCRE2_UCP, CI, 0, 0, 0, -1, -1, { "[\\s]", NULL }, "\xed\xa0\x80" },
{ PCRE2_UTF, CI, 0, 0, 0, 0, 3, { "[\\D]", NULL }, "\xe0\xab\xaa@" },
{ PCRE2_UTF, CI, 0, 0, 0, 0, 3, { "\\D+", NULL }, "n\xc3\xb1" },
{ PCRE2_UTF, CI, 0, 0, 0, 0, 5, { "\\W+", NULL }, "@\xf0\x9d\x84\x9e" },
/* These two are not invalid UTF tests, but this infrastructure fits better for them. */
{ 0, PCRE2_JIT_COMPLETE, 0, 0, 1, -1, -1, { "\\X{2}", NULL }, "\r\n\n" },
{ 0, PCRE2_JIT_COMPLETE, 0, 0, 1, -1, -1, { "\\R{2}", NULL }, "\r\n\n" },
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 0, 0, 0, -1, -1, { "^.a", &invalid_utf8_newline_cr }, "\xc3\xa7#a" },
{ 0, 0, 0, 0, 0, 0, 0, { NULL, NULL }, NULL }
};
#undef UDA
#undef CI
#undef CPI
static int run_invalid_utf8_test(const struct invalid_utf8_regression_test_case *current,
int pattern_index, int i, pcre2_compile_context_8 *ccontext, pcre2_match_data_8 *mdata)
{
pcre2_code_8 *code;
int result, errorcode;
PCRE2_SIZE length, erroroffset;
PCRE2_SIZE *ovector = pcre2_get_ovector_pointer_8(mdata);
if (current->pattern[i] == NULL)
return 1;
code = pcre2_compile_8((PCRE2_UCHAR8*)current->pattern[i], PCRE2_ZERO_TERMINATED,
current->compile_options, &errorcode, &erroroffset, ccontext);
if (!code) {
printf("Pattern[%d:0] cannot be compiled. Error offset: %d\n", pattern_index, (int)erroroffset);
return 0;
}
if (pcre2_jit_compile_8(code, current->jit_compile_options) != 0) {
printf("Pattern[%d:0] cannot be compiled by the JIT compiler.\n", pattern_index);
pcre2_code_free_8(code);
return 0;
}
length = (PCRE2_SIZE)(strlen(current->input) - current->skip_left - current->skip_right);
if (current->jit_compile_options & PCRE2_JIT_COMPLETE) {
result = pcre2_jit_match_8(code, (PCRE2_UCHAR8*)(current->input + current->skip_left),
length, current->start_offset - current->skip_left, 0, mdata, NULL);
if (check_invalid_utf_result(pattern_index, "match", result, current->match_start, current->match_end, ovector)) {
pcre2_code_free_8(code);
return 0;
}
}
if (current->jit_compile_options & PCRE2_JIT_PARTIAL_SOFT) {
result = pcre2_jit_match_8(code, (PCRE2_UCHAR8*)(current->input + current->skip_left),
length, current->start_offset - current->skip_left, PCRE2_PARTIAL_SOFT, mdata, NULL);
if (check_invalid_utf_result(pattern_index, "partial match", result, current->match_start, current->match_end, ovector)) {
pcre2_code_free_8(code);
return 0;
}
}
pcre2_code_free_8(code);
return 1;
}
static int invalid_utf8_regression_tests(void)
{
const struct invalid_utf8_regression_test_case *current;
pcre2_compile_context_8 *ccontext;
pcre2_match_data_8 *mdata;
int total = 0, successful = 0;
int result;
printf("\nRunning invalid-utf8 JIT regression tests\n");
ccontext = pcre2_compile_context_create_8(NULL);
pcre2_set_newline_8(ccontext, PCRE2_NEWLINE_ANY);
mdata = pcre2_match_data_create_8(4, NULL);
for (current = invalid_utf8_regression_test_cases; current->pattern[0]; current++) {
/* printf("\nPattern: %s :\n", current->pattern); */
total++;
result = 1;
if (current->pattern[1] != &invalid_utf8_newline_cr)
{
if (!run_invalid_utf8_test(current, total - 1, 0, ccontext, mdata))
result = 0;
if (!run_invalid_utf8_test(current, total - 1, 1, ccontext, mdata))
result = 0;
} else {
pcre2_set_newline_8(ccontext, PCRE2_NEWLINE_CR);
if (!run_invalid_utf8_test(current, total - 1, 0, ccontext, mdata))
result = 0;
pcre2_set_newline_8(ccontext, PCRE2_NEWLINE_ANY);
}
if (result) {
successful++;
}
printf(".");
if ((total % 60) == 0)
printf("\n");
}
if ((total % 60) != 0)
printf("\n");
pcre2_match_data_free_8(mdata);
pcre2_compile_context_free_8(ccontext);
if (total == successful) {
printf("\nAll invalid UTF8 JIT regression tests are successfully passed.\n");
return 0;
} else {
printf("\nInvalid UTF8 successful test ratio: %d%% (%d failed)\n", successful * 100 / total, total - successful);
return 1;
}
}
#else /* !SUPPORT_UNICODE || !SUPPORT_PCRE2_8 */
static int invalid_utf8_regression_tests(void)
{
return 0;
}
#endif /* SUPPORT_UNICODE && SUPPORT_PCRE2_8 */
#if defined SUPPORT_UNICODE && defined SUPPORT_PCRE2_16
#define UDA (PCRE2_UTF | PCRE2_DOTALL | PCRE2_ANCHORED)
#define CI (PCRE2_JIT_COMPLETE | PCRE2_JIT_INVALID_UTF)
#define CPI (PCRE2_JIT_COMPLETE | PCRE2_JIT_PARTIAL_SOFT | PCRE2_JIT_INVALID_UTF)
struct invalid_utf16_regression_test_case {
uint32_t compile_options;
int jit_compile_options;
int start_offset;
int skip_left;
int skip_right;
int match_start;
int match_end;
const PCRE2_UCHAR16 *pattern[2];
const PCRE2_UCHAR16 *input;
};
static PCRE2_UCHAR16 allany16[] = { '.', 0 };
static PCRE2_UCHAR16 non_word_boundary16[] = { '\\', 'B', 0 };
static PCRE2_UCHAR16 word_boundary16[] = { '\\', 'b', 0 };
static PCRE2_UCHAR16 backreference16[] = { '(', '.', ')', '\\', '1', 0 };
static PCRE2_UCHAR16 grapheme16[] = { '\\', 'X', 0 };
static PCRE2_UCHAR16 nothashmark16[] = { '[', '^', '#', ']', 0 };
static PCRE2_UCHAR16 afternl16[] = { '^', '\\', 'W', 0 };
static PCRE2_UCHAR16 generic16[] = { '#', 0xd800, 0xdc00, '#', 0 };
static PCRE2_UCHAR16 test16_1[] = { 0xd7ff, 0xe000, 0xffff, 0x01, '#', 0 };
static PCRE2_UCHAR16 test16_2[] = { 0xd800, 0xdc00, 0xd800, 0xdc00, 0 };
static PCRE2_UCHAR16 test16_3[] = { 0xdbff, 0xdfff, 0xdbff, 0xdfff, 0 };
static PCRE2_UCHAR16 test16_4[] = { 0xd800, 0xdbff, 0xd800, 0xdbff, 0 };
static PCRE2_UCHAR16 test16_5[] = { '#', 0xd800, 0xdc00, '#', 0 };
static PCRE2_UCHAR16 test16_6[] = { 'a', 'A', 0xdc28, 0 };
static PCRE2_UCHAR16 test16_7[] = { 0xd801, 0xdc00, 0xd801, 0xdc28, 0 };
static PCRE2_UCHAR16 test16_8[] = { '#', 0xd800, 0xdc00, 0 };
static PCRE2_UCHAR16 test16_9[] = { ' ', 0x2028, '#', 0 };
static PCRE2_UCHAR16 test16_10[] = { ' ', 0xdc00, 0xd800, 0x2028, '#', 0 };
static PCRE2_UCHAR16 test16_11[] = { 0xdc00, 0xdc00, 0xd800, 0xdc00, 0xdc00, '#', 0xd800, 0xdc00, '#', 0 };
static PCRE2_UCHAR16 test16_12[] = { '#', 0xd800, 0xdc00, 0xd800, '#', 0xd800, 0xdc00, 0xdc00, 0xdc00, '#', 0xd800, 0xdc00, '#', 0 };
static const struct invalid_utf16_regression_test_case invalid_utf16_regression_test_cases[] = {
{ UDA, CI, 0, 0, 0, 0, 1, { allany16, NULL }, test16_1 },
{ UDA, CI, 1, 0, 0, 1, 2, { allany16, NULL }, test16_1 },
{ UDA, CI, 2, 0, 0, 2, 3, { allany16, NULL }, test16_1 },
{ UDA, CI, 3, 0, 0, 3, 4, { allany16, NULL }, test16_1 },
{ UDA, CI, 0, 0, 0, 0, 2, { allany16, NULL }, test16_2 },
{ UDA, CI, 0, 0, 3, -1, -1, { allany16, NULL }, test16_2 },
{ UDA, CI, 1, 0, 0, -1, -1, { allany16, NULL }, test16_2 },
{ UDA, CI, 0, 0, 0, 0, 2, { allany16, NULL }, test16_3 },
{ UDA, CI, 0, 0, 3, -1, -1, { allany16, NULL }, test16_3 },
{ UDA, CI, 1, 0, 0, -1, -1, { allany16, NULL }, test16_3 },
{ UDA, CPI, 1, 0, 0, 1, 1, { non_word_boundary16, NULL }, test16_1 },
{ UDA, CPI, 2, 0, 0, 2, 2, { non_word_boundary16, NULL }, test16_1 },
{ UDA, CPI, 3, 0, 0, 3, 3, { non_word_boundary16, NULL }, test16_1 },
{ UDA, CPI, 4, 0, 0, 4, 4, { non_word_boundary16, NULL }, test16_1 },
{ UDA, CPI, 2, 0, 0, 2, 2, { non_word_boundary16, NULL }, test16_2 },
{ UDA, CPI, 2, 0, 0, 2, 2, { non_word_boundary16, NULL }, test16_3 },
{ UDA, CPI, 2, 1, 1, -1, -1, { non_word_boundary16, word_boundary16 }, test16_2 },
{ UDA, CPI, 2, 1, 1, -1, -1, { non_word_boundary16, word_boundary16 }, test16_3 },
{ UDA, CPI, 2, 0, 0, -1, -1, { non_word_boundary16, word_boundary16 }, test16_4 },
{ UDA, CPI, 2, 0, 0, -1, -1, { non_word_boundary16, word_boundary16 }, test16_5 },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 0, 0, 2, { backreference16, NULL }, test16_6 },
{ UDA | PCRE2_CASELESS, CPI, 1, 0, 0, -1, -1, { backreference16, NULL }, test16_6 },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 0, 0, 4, { backreference16, NULL }, test16_7 },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 1, -1, -1, { backreference16, NULL }, test16_7 },
{ UDA, CPI, 0, 0, 0, 0, 1, { grapheme16, NULL }, test16_6 },
{ UDA, CPI, 1, 0, 0, 1, 2, { grapheme16, NULL }, test16_6 },
{ UDA, CPI, 2, 0, 0, -1, -1, { grapheme16, NULL }, test16_6 },
{ UDA, CPI, 0, 0, 0, 0, 2, { grapheme16, NULL }, test16_7 },
{ UDA, CPI, 2, 0, 0, 2, 4, { grapheme16, NULL }, test16_7 },
{ UDA, CPI, 1, 0, 0, -1, -1, { grapheme16, NULL }, test16_7 },
{ UDA, CPI, 0, 0, 0, -1, -1, { nothashmark16, NULL }, test16_8 },
{ UDA, CPI, 1, 0, 0, 1, 3, { nothashmark16, NULL }, test16_8 },
{ UDA, CPI, 2, 0, 0, -1, -1, { nothashmark16, NULL }, test16_8 },
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 2, 3, { afternl16, NULL }, test16_9 },
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 4, 5, { afternl16, NULL }, test16_10 },
{ PCRE2_UTF | PCRE2_NO_START_OPTIMIZE, CI, 0, 0, 0, 5, 9, { generic16, NULL }, test16_11 },
{ PCRE2_UTF | PCRE2_NO_START_OPTIMIZE, CI, 0, 0, 0, 9, 13, { generic16, NULL }, test16_12 },
{ PCRE2_UTF, CI, 0, 0, 0, 5, 9, { generic16, NULL }, test16_11 },
{ PCRE2_UTF, CI, 0, 0, 0, 9, 13, { generic16, NULL }, test16_12 },
{ 0, 0, 0, 0, 0, 0, 0, { NULL, NULL }, NULL }
};
#undef UDA
#undef CI
#undef CPI
static int run_invalid_utf16_test(const struct invalid_utf16_regression_test_case *current,
int pattern_index, int i, pcre2_compile_context_16 *ccontext, pcre2_match_data_16 *mdata)
{
pcre2_code_16 *code;
int result, errorcode;
PCRE2_SIZE length, erroroffset;
const PCRE2_UCHAR16 *input;
PCRE2_SIZE *ovector = pcre2_get_ovector_pointer_16(mdata);
if (current->pattern[i] == NULL)
return 1;
code = pcre2_compile_16(current->pattern[i], PCRE2_ZERO_TERMINATED,
current->compile_options, &errorcode, &erroroffset, ccontext);
if (!code) {
printf("Pattern[%d:0] cannot be compiled. Error offset: %d\n", pattern_index, (int)erroroffset);
return 0;
}
if (pcre2_jit_compile_16(code, current->jit_compile_options) != 0) {
printf("Pattern[%d:0] cannot be compiled by the JIT compiler.\n", pattern_index);
pcre2_code_free_16(code);
return 0;
}
input = current->input;
length = 0;
while (*input++ != 0)
length++;
length -= current->skip_left + current->skip_right;
if (current->jit_compile_options & PCRE2_JIT_COMPLETE) {
result = pcre2_jit_match_16(code, (current->input + current->skip_left),
length, current->start_offset - current->skip_left, 0, mdata, NULL);
if (check_invalid_utf_result(pattern_index, "match", result, current->match_start, current->match_end, ovector)) {
pcre2_code_free_16(code);
return 0;
}
}
if (current->jit_compile_options & PCRE2_JIT_PARTIAL_SOFT) {
result = pcre2_jit_match_16(code, (current->input + current->skip_left),
length, current->start_offset - current->skip_left, PCRE2_PARTIAL_SOFT, mdata, NULL);
if (check_invalid_utf_result(pattern_index, "partial match", result, current->match_start, current->match_end, ovector)) {
pcre2_code_free_16(code);
return 0;
}
}
pcre2_code_free_16(code);
return 1;
}
static int invalid_utf16_regression_tests(void)
{
const struct invalid_utf16_regression_test_case *current;
pcre2_compile_context_16 *ccontext;
pcre2_match_data_16 *mdata;
int total = 0, successful = 0;
int result;
printf("\nRunning invalid-utf16 JIT regression tests\n");
ccontext = pcre2_compile_context_create_16(NULL);
pcre2_set_newline_16(ccontext, PCRE2_NEWLINE_ANY);
mdata = pcre2_match_data_create_16(4, NULL);
for (current = invalid_utf16_regression_test_cases; current->pattern[0]; current++) {
/* printf("\nPattern: %s :\n", current->pattern); */
total++;
result = 1;
if (!run_invalid_utf16_test(current, total - 1, 0, ccontext, mdata))
result = 0;
if (!run_invalid_utf16_test(current, total - 1, 1, ccontext, mdata))
result = 0;
if (result) {
successful++;
}
printf(".");
if ((total % 60) == 0)
printf("\n");
}
if ((total % 60) != 0)
printf("\n");
pcre2_match_data_free_16(mdata);
pcre2_compile_context_free_16(ccontext);
if (total == successful) {
printf("\nAll invalid UTF16 JIT regression tests are successfully passed.\n");
return 0;
} else {
printf("\nInvalid UTF16 successful test ratio: %d%% (%d failed)\n", successful * 100 / total, total - successful);
return 1;
}
}
#else /* !SUPPORT_UNICODE || !SUPPORT_PCRE2_16 */
static int invalid_utf16_regression_tests(void)
{
return 0;
}
#endif /* SUPPORT_UNICODE && SUPPORT_PCRE2_16 */
#if defined SUPPORT_UNICODE && defined SUPPORT_PCRE2_32
#define UDA (PCRE2_UTF | PCRE2_DOTALL | PCRE2_ANCHORED)
#define CI (PCRE2_JIT_COMPLETE | PCRE2_JIT_INVALID_UTF)
#define CPI (PCRE2_JIT_COMPLETE | PCRE2_JIT_PARTIAL_SOFT | PCRE2_JIT_INVALID_UTF)
struct invalid_utf32_regression_test_case {
uint32_t compile_options;
int jit_compile_options;
int start_offset;
int skip_left;
int skip_right;
int match_start;
int match_end;
const PCRE2_UCHAR32 *pattern[2];
const PCRE2_UCHAR32 *input;
};
static PCRE2_UCHAR32 allany32[] = { '.', 0 };
static PCRE2_UCHAR32 non_word_boundary32[] = { '\\', 'B', 0 };
static PCRE2_UCHAR32 word_boundary32[] = { '\\', 'b', 0 };
static PCRE2_UCHAR32 backreference32[] = { '(', '.', ')', '\\', '1', 0 };
static PCRE2_UCHAR32 grapheme32[] = { '\\', 'X', 0 };
static PCRE2_UCHAR32 nothashmark32[] = { '[', '^', '#', ']', 0 };
static PCRE2_UCHAR32 afternl32[] = { '^', '\\', 'W', 0 };
static PCRE2_UCHAR32 test32_1[] = { 0x10ffff, 0x10ffff, 0x110000, 0x110000, 0x10ffff, 0 };
static PCRE2_UCHAR32 test32_2[] = { 0xd7ff, 0xe000, 0xd800, 0xdfff, 0xe000, 0xdfff, 0xd800, 0 };
static PCRE2_UCHAR32 test32_3[] = { 'a', 'A', 0x110000, 0 };
static PCRE2_UCHAR32 test32_4[] = { '#', 0x10ffff, 0x110000, 0 };
static PCRE2_UCHAR32 test32_5[] = { ' ', 0x2028, '#', 0 };
static PCRE2_UCHAR32 test32_6[] = { ' ', 0x110000, 0x2028, '#', 0 };
static const struct invalid_utf32_regression_test_case invalid_utf32_regression_test_cases[] = {
{ UDA, CI, 0, 0, 0, 0, 1, { allany32, NULL }, test32_1 },
{ UDA, CI, 2, 0, 0, -1, -1, { allany32, NULL }, test32_1 },
{ UDA, CI, 0, 0, 0, 0, 1, { allany32, NULL }, test32_2 },
{ UDA, CI, 1, 0, 0, 1, 2, { allany32, NULL }, test32_2 },
{ UDA, CI, 2, 0, 0, -1, -1, { allany32, NULL }, test32_2 },
{ UDA, CI, 3, 0, 0, -1, -1, { allany32, NULL }, test32_2 },
{ UDA, CPI, 1, 0, 0, 1, 1, { non_word_boundary32, NULL }, test32_1 },
{ UDA, CPI, 3, 0, 0, -1, -1, { non_word_boundary32, word_boundary32 }, test32_1 },
{ UDA, CPI, 1, 0, 0, 1, 1, { non_word_boundary32, NULL }, test32_2 },
{ UDA, CPI, 3, 0, 0, -1, -1, { non_word_boundary32, word_boundary32 }, test32_2 },
{ UDA, CPI, 6, 0, 0, -1, -1, { non_word_boundary32, word_boundary32 }, test32_2 },
{ UDA | PCRE2_CASELESS, CPI, 0, 0, 0, 0, 2, { backreference32, NULL }, test32_3 },
{ UDA | PCRE2_CASELESS, CPI, 1, 0, 0, -1, -1, { backreference32, NULL }, test32_3 },
{ UDA, CPI, 0, 0, 0, 0, 1, { grapheme32, NULL }, test32_1 },
{ UDA, CPI, 2, 0, 0, -1, -1, { grapheme32, NULL }, test32_1 },
{ UDA, CPI, 1, 0, 0, 1, 2, { grapheme32, NULL }, test32_2 },
{ UDA, CPI, 2, 0, 0, -1, -1, { grapheme32, NULL }, test32_2 },
{ UDA, CPI, 3, 0, 0, -1, -1, { grapheme32, NULL }, test32_2 },
{ UDA, CPI, 4, 0, 0, 4, 5, { grapheme32, NULL }, test32_2 },
{ UDA, CPI, 0, 0, 0, -1, -1, { nothashmark32, NULL }, test32_4 },
{ UDA, CPI, 1, 0, 0, 1, 2, { nothashmark32, NULL }, test32_4 },
{ UDA, CPI, 2, 0, 0, -1, -1, { nothashmark32, NULL }, test32_4 },
{ UDA, CPI, 1, 0, 0, 1, 2, { nothashmark32, NULL }, test32_2 },
{ UDA, CPI, 2, 0, 0, -1, -1, { nothashmark32, NULL }, test32_2 },
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 2, 3, { afternl32, NULL }, test32_5 },
{ PCRE2_UTF | PCRE2_MULTILINE, CI, 1, 0, 0, 3, 4, { afternl32, NULL }, test32_6 },
{ 0, 0, 0, 0, 0, 0, 0, { NULL, NULL }, NULL }
};
#undef UDA
#undef CI
#undef CPI
static int run_invalid_utf32_test(const struct invalid_utf32_regression_test_case *current,
int pattern_index, int i, pcre2_compile_context_32 *ccontext, pcre2_match_data_32 *mdata)
{
pcre2_code_32 *code;
int result, errorcode;
PCRE2_SIZE length, erroroffset;
const PCRE2_UCHAR32 *input;
PCRE2_SIZE *ovector = pcre2_get_ovector_pointer_32(mdata);
if (current->pattern[i] == NULL)
return 1;
code = pcre2_compile_32(current->pattern[i], PCRE2_ZERO_TERMINATED,
current->compile_options, &errorcode, &erroroffset, ccontext);
if (!code) {
printf("Pattern[%d:0] cannot be compiled. Error offset: %d\n", pattern_index, (int)erroroffset);
return 0;
}
if (pcre2_jit_compile_32(code, current->jit_compile_options) != 0) {
printf("Pattern[%d:0] cannot be compiled by the JIT compiler.\n", pattern_index);
pcre2_code_free_32(code);
return 0;
}
input = current->input;
length = 0;
while (*input++ != 0)
length++;
length -= current->skip_left + current->skip_right;
if (current->jit_compile_options & PCRE2_JIT_COMPLETE) {
result = pcre2_jit_match_32(code, (current->input + current->skip_left),
length, current->start_offset - current->skip_left, 0, mdata, NULL);
if (check_invalid_utf_result(pattern_index, "match", result, current->match_start, current->match_end, ovector)) {
pcre2_code_free_32(code);
return 0;
}
}
if (current->jit_compile_options & PCRE2_JIT_PARTIAL_SOFT) {
result = pcre2_jit_match_32(code, (current->input + current->skip_left),
length, current->start_offset - current->skip_left, PCRE2_PARTIAL_SOFT, mdata, NULL);
if (check_invalid_utf_result(pattern_index, "partial match", result, current->match_start, current->match_end, ovector)) {
pcre2_code_free_32(code);
return 0;
}
}
pcre2_code_free_32(code);
return 1;
}
static int invalid_utf32_regression_tests(void)
{
const struct invalid_utf32_regression_test_case *current;
pcre2_compile_context_32 *ccontext;
pcre2_match_data_32 *mdata;
int total = 0, successful = 0;
int result;
printf("\nRunning invalid-utf32 JIT regression tests\n");
ccontext = pcre2_compile_context_create_32(NULL);
pcre2_set_newline_32(ccontext, PCRE2_NEWLINE_ANY);
mdata = pcre2_match_data_create_32(4, NULL);
for (current = invalid_utf32_regression_test_cases; current->pattern[0]; current++) {
/* printf("\nPattern: %s :\n", current->pattern); */
total++;
result = 1;
if (!run_invalid_utf32_test(current, total - 1, 0, ccontext, mdata))
result = 0;
if (!run_invalid_utf32_test(current, total - 1, 1, ccontext, mdata))
result = 0;
if (result) {
successful++;
}
printf(".");
if ((total % 60) == 0)
printf("\n");
}
if ((total % 60) != 0)
printf("\n");
pcre2_match_data_free_32(mdata);
pcre2_compile_context_free_32(ccontext);
if (total == successful) {
printf("\nAll invalid UTF32 JIT regression tests are successfully passed.\n");
return 0;
} else {
printf("\nInvalid UTF32 successful test ratio: %d%% (%d failed)\n", successful * 100 / total, total - successful);
return 1;
}
}
#else /* !SUPPORT_UNICODE || !SUPPORT_PCRE2_32 */
static int invalid_utf32_regression_tests(void)
{
return 0;
}
#endif /* SUPPORT_UNICODE && SUPPORT_PCRE2_32 */
/* End of pcre2_jit_test.c */
pcre2-10.45/src/pcre2posix.c 0000644 0001751 0000166 00000034505 14750415257 011272 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This module is a wrapper that provides a POSIX API to the underlying PCRE2
functions. The functions are called pcre2_regcomp(), pcre2_regexec(), etc.
pcre2posix.h defines the POSIX names as macros for the corresonding pcre2_xxx
functions, so any program that includes it and uses the POSIX names will call
the PCRE2 implementations instead. */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#ifdef PCRE2POSIX_SHARED
#undef PCRE2_STATIC
#endif
/* Ensure that the PCRE2POSIX_EXP_xxx macros are set appropriately for
compiling these functions. This must come before including pcre2posix.h, where
they are set for an application (using these functions) if they have not
previously been set. */
#if defined(_WIN32) && (defined(PCRE2POSIX_SHARED) || !defined(PCRE2_STATIC))
# define PCRE2POSIX_EXP_DECL extern __declspec(dllexport)
# define PCRE2POSIX_EXP_DEFN __declspec(dllexport)
#endif
/* Older versions of MSVC lack snprintf(). This define allows for
warning/error-free compilation and testing with MSVC compilers back to at least
MSVC 10/2010. Except for VC6 (which is missing some fundamentals and fails). */
#if defined(_MSC_VER) && (_MSC_VER < 1900)
#define snprintf _snprintf
#define BROKEN_SNPRINTF
#endif
/* Compile-time error numbers start at this value. It should probably never be
changed. This #define is a copy of the one in pcre2_internal.h. */
#define COMPILE_ERROR_BASE 100
/* Standard C headers */
#include
#include
#include
#include
#include
#include
/* PCRE2 headers */
#include "pcre2.h"
#include "pcre2posix.h"
#include "pcre2_util.h"
/* Table to translate PCRE2 compile time error codes into POSIX error codes.
Only a few PCRE2 errors with a value greater than 23 turn into special POSIX
codes: most go to REG_BADPAT. The second table lists, in pairs, those that
don't, even though some of them cannot currently be provoked from within the
POSIX wrapper. */
static const int eint1[] = {
0, /* No error */
REG_EESCAPE, /* \ at end of pattern */
REG_EESCAPE, /* \c at end of pattern */
REG_EESCAPE, /* unrecognized character follows \ */
REG_BADBR, /* numbers out of order in {} quantifier */
/* 5 */
REG_BADBR, /* number too big in {} quantifier */
REG_EBRACK, /* missing terminating ] for character class */
REG_ECTYPE, /* invalid escape sequence in character class */
REG_ERANGE, /* range out of order in character class */
REG_BADRPT, /* nothing to repeat */
/* 10 */
REG_ASSERT, /* internal error: unexpected repeat */
REG_BADPAT, /* unrecognized character after (? or (?- */
REG_BADPAT, /* POSIX named classes are supported only within a class */
REG_BADPAT, /* POSIX collating elements are not supported */
REG_EPAREN, /* missing ) */
/* 15 */
REG_ESUBREG, /* reference to non-existent subpattern */
REG_INVARG, /* pattern passed as NULL */
REG_INVARG, /* unknown compile-time option bit(s) */
REG_EPAREN, /* missing ) after (?# comment */
REG_ESIZE, /* parentheses nested too deeply */
/* 20 */
REG_ESIZE, /* regular expression too large */
REG_ESPACE, /* failed to get memory */
REG_EPAREN, /* unmatched closing parenthesis */
REG_ASSERT /* internal error: code overflow */
};
static const int eint2[] = {
30, REG_ECTYPE, /* unknown POSIX class name */
32, REG_INVARG, /* this version of PCRE2 does not have Unicode support */
37, REG_EESCAPE, /* PCRE2 does not support \L, \l, \N{name}, \U, or \u */
56, REG_INVARG, /* internal error: unknown newline setting */
92, REG_INVARG, /* invalid option bits with PCRE2_LITERAL */
98, REG_EESCAPE, /* missing digit after \0 in NO_BS0 mode */
99, REG_EESCAPE, /* \K in lookaround */
102, REG_EESCAPE /* \ddd octal > \377 in PYTHON_OCTAL mode */
};
/* Table of texts corresponding to POSIX error codes */
static const char *const pstring[] = {
"", /* Dummy for value 0 */
"internal error", /* REG_ASSERT */
"invalid repeat counts in {}", /* BADBR */
"pattern error", /* BADPAT */
"? * + invalid", /* BADRPT */
"unbalanced {}", /* EBRACE */
"unbalanced []", /* EBRACK */
"collation error - not relevant", /* ECOLLATE */
"bad class", /* ECTYPE */
"bad escape sequence", /* EESCAPE */
"empty expression", /* EMPTY */
"unbalanced ()", /* EPAREN */
"bad range inside []", /* ERANGE */
"expression too big", /* ESIZE */
"failed to get memory", /* ESPACE */
"bad back reference", /* ESUBREG */
"bad argument", /* INVARG */
"match failed" /* NOMATCH */
};
static int message_len(const char *message, int offset)
{
char buf[12];
/* 11 magic number comes from the format below */
return (int)strlen(message) + 11 + snprintf(buf, sizeof(buf), "%d", offset);
}
/*************************************************
* Translate error code to string *
*************************************************/
PCRE2POSIX_EXP_DEFN size_t PCRE2_CALL_CONVENTION
pcre2_regerror(int errcode, const regex_t *preg, char *errbuf,
size_t errbuf_size)
{
int ret;
const char *message;
size_t len = 0; /* keeps 0 if snprintf is used */
message = (errcode <= 0 || errcode >= (int)(sizeof(pstring)/sizeof(char *)))?
"unknown error code" : pstring[errcode];
if (preg != NULL && (int)preg->re_erroffset != -1)
{
/* no need to deal with UB in snprintf */
if (errbuf_size > INT_MAX) errbuf_size = INT_MAX;
/* there are 11 characters between message and offset;
update message_len() if changed */
ret = snprintf(errbuf, errbuf_size, "%s at offset %d", message,
(int)preg->re_erroffset);
}
else
{
len = strlen(message);
if (errbuf_size != 0)
{
strncpy(errbuf, message, errbuf_size);
if (errbuf_size <= len) errbuf[errbuf_size - 1] = '\0';
}
ret = (int)len;
}
PCRE2_ASSERT(len > 0 || preg != NULL);
do {
if (ret < 0)
{
#ifdef BROKEN_SNPRINTF
/* _snprintf returns -1 on overflow and doesn't zero terminate */
if (!len)
{
if (ret == -1 && errbuf_size != 0) errbuf[errbuf_size - 1] = '\0';
ret = message_len(message, (int)preg->re_erroffset);
break;
}
#endif
/* snprintf failed, will use a 14 char long message if possible */
ret = 14;
if (errbuf_size != 0)
{
strncpy(errbuf, "internal error", errbuf_size);
if ((int)errbuf_size <= ret) errbuf[errbuf_size - 1] = '\0';
}
}
else if (ret == (int)errbuf_size && !len)
{
/* pre C99 snprintf returns used, so redo ret to fix that */
ret = message_len(message, (int)preg->re_erroffset);
}
} while (0);
return ret + 1;
}
/*************************************************
* Free store held by a regex *
*************************************************/
PCRE2POSIX_EXP_DEFN void PCRE2_CALL_CONVENTION
pcre2_regfree(regex_t *preg)
{
pcre2_match_data_free(preg->re_match_data);
pcre2_code_free(preg->re_pcre2_code);
}
/*************************************************
* Compile a regular expression *
*************************************************/
/*
Arguments:
preg points to a structure for recording the compiled expression
pattern the pattern to compile
cflags compilation flags
Returns: 0 on success
various non-zero codes on failure
*/
PCRE2POSIX_EXP_DEFN int PCRE2_CALL_CONVENTION
pcre2_regcomp(regex_t *preg, const char *pattern, int cflags)
{
PCRE2_SIZE erroffset;
PCRE2_SIZE patlen;
int errorcode;
int options = 0;
int re_nsub = 0;
patlen = ((cflags & REG_PEND) != 0)? (PCRE2_SIZE)(preg->re_endp - pattern) :
PCRE2_ZERO_TERMINATED;
if ((cflags & REG_ICASE) != 0) options |= PCRE2_CASELESS;
if ((cflags & REG_NEWLINE) != 0) options |= PCRE2_MULTILINE;
if ((cflags & REG_DOTALL) != 0) options |= PCRE2_DOTALL;
if ((cflags & REG_NOSPEC) != 0) options |= PCRE2_LITERAL;
if ((cflags & REG_UTF) != 0) options |= PCRE2_UTF;
if ((cflags & REG_UCP) != 0) options |= PCRE2_UCP;
if ((cflags & REG_UNGREEDY) != 0) options |= PCRE2_UNGREEDY;
preg->re_cflags = cflags;
preg->re_pcre2_code = pcre2_compile((PCRE2_SPTR)pattern, patlen, options,
&errorcode, &erroffset, NULL);
preg->re_erroffset = erroffset;
if (preg->re_pcre2_code == NULL)
{
unsigned int i;
/* A negative value is a UTF error; otherwise all error codes are greater
than COMPILE_ERROR_BASE, but check, just in case. */
if (errorcode < COMPILE_ERROR_BASE) return REG_BADPAT;
errorcode -= COMPILE_ERROR_BASE;
if (errorcode < (int)(sizeof(eint1)/sizeof(const int)))
return eint1[errorcode];
for (i = 0; i < sizeof(eint2)/sizeof(const int); i += 2)
if (errorcode == eint2[i]) return eint2[i+1];
return REG_BADPAT;
}
(void)pcre2_pattern_info((const pcre2_code *)preg->re_pcre2_code,
PCRE2_INFO_CAPTURECOUNT, &re_nsub);
preg->re_nsub = (size_t)re_nsub;
preg->re_match_data = pcre2_match_data_create(re_nsub + 1, NULL);
preg->re_erroffset = (size_t)(-1); /* No meaning after successful compile */
if (preg->re_match_data == NULL)
{
/* LCOV_EXCL_START */
pcre2_code_free(preg->re_pcre2_code);
return REG_ESPACE;
/* LCOV_EXCL_STOP */
}
return 0;
}
/*************************************************
* Match a regular expression *
*************************************************/
/* A suitable match_data block, large enough to hold all possible captures, was
obtained when the pattern was compiled, to save having to allocate and free it
for each match. If REG_NOSUB was specified at compile time, the nmatch and
pmatch arguments are ignored, and the only result is yes/no/error. */
PCRE2POSIX_EXP_DEFN int PCRE2_CALL_CONVENTION
pcre2_regexec(const regex_t *preg, const char *string, size_t nmatch,
regmatch_t pmatch[], int eflags)
{
int rc, so, eo;
int options = 0;
pcre2_match_data *md = (pcre2_match_data *)preg->re_match_data;
if (string == NULL) return REG_INVARG;
if ((eflags & REG_NOTBOL) != 0) options |= PCRE2_NOTBOL;
if ((eflags & REG_NOTEOL) != 0) options |= PCRE2_NOTEOL;
if ((eflags & REG_NOTEMPTY) != 0) options |= PCRE2_NOTEMPTY;
/* When REG_NOSUB was specified, or if no vector has been passed in which to
put captured strings, ensure that nmatch is zero. This will stop any attempt to
write to pmatch. */
if ((preg->re_cflags & REG_NOSUB) != 0 || pmatch == NULL) nmatch = 0;
/* REG_STARTEND is a BSD extension, to allow for non-NUL-terminated strings.
The man page from OS X says "REG_STARTEND affects only the location of the
string, not how it is matched". That is why the "so" value is used to bump the
start location rather than being passed as a PCRE2 "starting offset". */
if ((eflags & REG_STARTEND) != 0)
{
if (pmatch == NULL) return REG_INVARG;
so = pmatch[0].rm_so;
eo = pmatch[0].rm_eo;
}
else
{
so = 0;
eo = (int)strlen(string);
}
rc = pcre2_match((const pcre2_code *)preg->re_pcre2_code,
(PCRE2_SPTR)string + so, (eo - so), 0, options, md, NULL);
/* Successful match */
if (rc >= 0)
{
size_t i;
PCRE2_SIZE *ovector = pcre2_get_ovector_pointer(md);
if ((size_t)rc > nmatch) rc = (int)nmatch;
for (i = 0; i < (size_t)rc; i++)
{
pmatch[i].rm_so = (ovector[i*2] == PCRE2_UNSET)? -1 :
(int)(ovector[i*2] + so);
pmatch[i].rm_eo = (ovector[i*2+1] == PCRE2_UNSET)? -1 :
(int)(ovector[i*2+1] + so);
}
for (; i < nmatch; i++) pmatch[i].rm_so = pmatch[i].rm_eo = -1;
return 0;
}
/* Unsuccessful match */
if (rc <= PCRE2_ERROR_UTF8_ERR1 && rc >= PCRE2_ERROR_UTF8_ERR21)
return REG_INVARG;
/* Most of these are events that won't occur during testing, so exclude them
from coverage. */
switch(rc)
{
case PCRE2_ERROR_HEAPLIMIT: return REG_ESPACE;
case PCRE2_ERROR_NOMATCH: return REG_NOMATCH;
/* LCOV_EXCL_START */
case PCRE2_ERROR_BADMODE: return REG_INVARG;
case PCRE2_ERROR_BADMAGIC: return REG_INVARG;
case PCRE2_ERROR_BADOPTION: return REG_INVARG;
case PCRE2_ERROR_BADUTFOFFSET: return REG_INVARG;
case PCRE2_ERROR_MATCHLIMIT: return REG_ESPACE;
case PCRE2_ERROR_NOMEMORY: return REG_ESPACE;
case PCRE2_ERROR_NULL: return REG_INVARG;
default: return REG_ASSERT;
/* LCOV_EXCL_STOP */
}
}
/* End of pcre2posix.c */
pcre2-10.45/src/pcre2_jit_neon_inc.h 0000644 0001751 0000166 00000020757 14750415257 012736 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
This module by Zoltan Herczeg and Sebastian Pop
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2019 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
# if defined(FFCS)
# if defined(FF_UTF)
# define FF_FUN ffcs_utf
# else
# define FF_FUN ffcs
# endif
# elif defined(FFCS_2)
# if defined(FF_UTF)
# define FF_FUN ffcs_2_utf
# else
# define FF_FUN ffcs_2
# endif
# elif defined(FFCS_MASK)
# if defined(FF_UTF)
# define FF_FUN ffcs_mask_utf
# else
# define FF_FUN ffcs_mask
# endif
# elif defined(FFCPS_0)
# if defined (FF_UTF)
# define FF_FUN ffcps_0_utf
# else
# define FF_FUN ffcps_0
# endif
# elif defined (FFCPS_1)
# if defined (FF_UTF)
# define FF_FUN ffcps_1_utf
# else
# define FF_FUN ffcps_1
# endif
# elif defined (FFCPS_DEFAULT)
# if defined (FF_UTF)
# define FF_FUN ffcps_default_utf
# else
# define FF_FUN ffcps_default
# endif
# endif
#if (defined(__GNUC__) && defined(__SANITIZE_ADDRESS__) && __SANITIZE_ADDRESS__ ) \
|| (defined(__clang__) \
&& ((__clang_major__ == 3 && __clang_minor__ >= 3) || (__clang_major__ > 3)))
__attribute__((no_sanitize_address))
#endif
static sljit_u8* SLJIT_FUNC FF_FUN(sljit_u8 *str_end, sljit_u8 **str_ptr, sljit_uw offs1, sljit_uw offs2, sljit_uw chars)
#undef FF_FUN
{
quad_word qw;
int_char ic;
SLJIT_UNUSED_ARG(offs1);
SLJIT_UNUSED_ARG(offs2);
ic.x = chars;
#if defined(FFCS)
sljit_u8 c1 = ic.c.c1;
vect_t vc1 = VDUPQ(c1);
#elif defined(FFCS_2)
sljit_u8 c1 = ic.c.c1;
vect_t vc1 = VDUPQ(c1);
sljit_u8 c2 = ic.c.c2;
vect_t vc2 = VDUPQ(c2);
#elif defined(FFCS_MASK)
sljit_u8 c1 = ic.c.c1;
vect_t vc1 = VDUPQ(c1);
sljit_u8 mask = ic.c.c2;
vect_t vmask = VDUPQ(mask);
#endif
#if defined(FFCPS)
compare_type compare1_type = compare_match1;
compare_type compare2_type = compare_match1;
vect_t cmp1a, cmp1b, cmp2a, cmp2b;
const sljit_u32 diff = IN_UCHARS(offs1 - offs2);
PCRE2_UCHAR char1a = ic.c.c1;
PCRE2_UCHAR char2a = ic.c.c3;
# ifdef FFCPS_CHAR1A2A
cmp1a = VDUPQ(char1a);
cmp2a = VDUPQ(char2a);
cmp1b = VDUPQ(0); /* to avoid errors on older compilers -Werror=maybe-uninitialized */
cmp2b = VDUPQ(0); /* to avoid errors on older compilers -Werror=maybe-uninitialized */
# else
PCRE2_UCHAR char1b = ic.c.c2;
PCRE2_UCHAR char2b = ic.c.c4;
if (char1a == char1b)
{
cmp1a = VDUPQ(char1a);
cmp1b = VDUPQ(0); /* to avoid errors on older compilers -Werror=maybe-uninitialized */
}
else
{
sljit_u32 bit1 = char1a ^ char1b;
if (is_powerof2(bit1))
{
compare1_type = compare_match1i;
cmp1a = VDUPQ(char1a | bit1);
cmp1b = VDUPQ(bit1);
}
else
{
compare1_type = compare_match2;
cmp1a = VDUPQ(char1a);
cmp1b = VDUPQ(char1b);
}
}
if (char2a == char2b)
{
cmp2a = VDUPQ(char2a);
cmp2b = VDUPQ(0); /* to avoid errors on older compilers -Werror=maybe-uninitialized */
}
else
{
sljit_u32 bit2 = char2a ^ char2b;
if (is_powerof2(bit2))
{
compare2_type = compare_match1i;
cmp2a = VDUPQ(char2a | bit2);
cmp2b = VDUPQ(bit2);
}
else
{
compare2_type = compare_match2;
cmp2a = VDUPQ(char2a);
cmp2b = VDUPQ(char2b);
}
}
# endif
*str_ptr += IN_UCHARS(offs1);
#endif
#if PCRE2_CODE_UNIT_WIDTH != 8
vect_t char_mask = VDUPQ(0xff);
#endif
#if defined(FF_UTF)
restart:;
#endif
#if defined(FFCPS)
if (*str_ptr >= str_end)
return NULL;
sljit_u8 *p1 = *str_ptr - diff;
#endif
sljit_s32 align_offset = ((uint64_t)*str_ptr & 0xf);
*str_ptr = (sljit_u8 *) ((uint64_t)*str_ptr & ~0xf);
vect_t data = VLD1Q(*str_ptr);
#if PCRE2_CODE_UNIT_WIDTH != 8
data = VANDQ(data, char_mask);
#endif
#if defined(FFCS)
vect_t eq = VCEQQ(data, vc1);
#elif defined(FFCS_2)
vect_t eq1 = VCEQQ(data, vc1);
vect_t eq2 = VCEQQ(data, vc2);
vect_t eq = VORRQ(eq1, eq2);
#elif defined(FFCS_MASK)
vect_t eq = VORRQ(data, vmask);
eq = VCEQQ(eq, vc1);
#elif defined(FFCPS)
# if defined(FFCPS_DIFF1)
vect_t prev_data = data;
# endif
vect_t data2;
if (p1 < *str_ptr)
{
data2 = VLD1Q(*str_ptr - diff);
#if PCRE2_CODE_UNIT_WIDTH != 8
data2 = VANDQ(data2, char_mask);
#endif
}
else
data2 = shift_left_n_lanes(data, offs1 - offs2);
if (compare1_type == compare_match1)
data = VCEQQ(data, cmp1a);
else
data = fast_forward_char_pair_compare(compare1_type, data, cmp1a, cmp1b);
if (compare2_type == compare_match1)
data2 = VCEQQ(data2, cmp2a);
else
data2 = fast_forward_char_pair_compare(compare2_type, data2, cmp2a, cmp2b);
vect_t eq = VANDQ(data, data2);
#endif
VST1Q(qw.mem, eq);
/* Ignore matches before the first STR_PTR. */
if (align_offset < 8)
{
qw.dw[0] >>= align_offset * 8;
if (qw.dw[0])
{
*str_ptr += align_offset + __builtin_ctzll(qw.dw[0]) / 8;
goto match;
}
if (qw.dw[1])
{
*str_ptr += 8 + __builtin_ctzll(qw.dw[1]) / 8;
goto match;
}
}
else
{
qw.dw[1] >>= (align_offset - 8) * 8;
if (qw.dw[1])
{
*str_ptr += align_offset + __builtin_ctzll(qw.dw[1]) / 8;
goto match;
}
}
*str_ptr += 16;
while (*str_ptr < str_end)
{
vect_t orig_data = VLD1Q(*str_ptr);
#if PCRE2_CODE_UNIT_WIDTH != 8
orig_data = VANDQ(orig_data, char_mask);
#endif
data = orig_data;
#if defined(FFCS)
eq = VCEQQ(data, vc1);
#elif defined(FFCS_2)
eq1 = VCEQQ(data, vc1);
eq2 = VCEQQ(data, vc2);
eq = VORRQ(eq1, eq2);
#elif defined(FFCS_MASK)
eq = VORRQ(data, vmask);
eq = VCEQQ(eq, vc1);
#endif
#if defined(FFCPS)
# if defined (FFCPS_DIFF1)
data2 = VEXTQ(prev_data, data, VECTOR_FACTOR - 1);
# else
data2 = VLD1Q(*str_ptr - diff);
# if PCRE2_CODE_UNIT_WIDTH != 8
data2 = VANDQ(data2, char_mask);
# endif
# endif
# ifdef FFCPS_CHAR1A2A
data = VCEQQ(data, cmp1a);
data2 = VCEQQ(data2, cmp2a);
# else
if (compare1_type == compare_match1)
data = VCEQQ(data, cmp1a);
else
data = fast_forward_char_pair_compare(compare1_type, data, cmp1a, cmp1b);
if (compare2_type == compare_match1)
data2 = VCEQQ(data2, cmp2a);
else
data2 = fast_forward_char_pair_compare(compare2_type, data2, cmp2a, cmp2b);
# endif
eq = VANDQ(data, data2);
#endif
VST1Q(qw.mem, eq);
if (qw.dw[0])
*str_ptr += __builtin_ctzll(qw.dw[0]) / 8;
else if (qw.dw[1])
*str_ptr += 8 + __builtin_ctzll(qw.dw[1]) / 8;
else {
*str_ptr += 16;
#if defined (FFCPS_DIFF1)
prev_data = orig_data;
#endif
continue;
}
match:;
if (*str_ptr >= str_end)
/* Failed match. */
return NULL;
#if defined(FF_UTF)
if (utf_continue((PCRE2_SPTR)*str_ptr - offs1))
{
/* Not a match. */
*str_ptr += IN_UCHARS(1);
goto restart;
}
#endif
/* Match. */
#if defined (FFCPS)
*str_ptr -= IN_UCHARS(offs1);
#endif
return *str_ptr;
}
/* Failed match. */
return NULL;
}
pcre2-10.45/src/pcre2_dfa_match.c 0000644 0001751 0000166 00000427412 14750415257 012200 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This module contains the external function pcre2_dfa_match(), which is an
alternative matching function that uses a sort of DFA algorithm (not a true
FSM). This is NOT Perl-compatible, but it has advantages in certain
applications. */
/* NOTE ABOUT PERFORMANCE: A user of this function sent some code that improved
the performance of his patterns greatly. I could not use it as it stood, as it
was not thread safe, and made assumptions about pattern sizes. Also, it caused
test 7 to loop, and test 9 to crash with a segfault.
The issue is the check for duplicate states, which is done by a simple linear
search up the state list. (Grep for "duplicate" below to find the code.) For
many patterns, there will never be many states active at one time, so a simple
linear search is fine. In patterns that have many active states, it might be a
bottleneck. The suggested code used an indexing scheme to remember which states
had previously been used for each character, and avoided the linear search when
it knew there was no chance of a duplicate. This was implemented when adding
states to the state lists.
I wrote some thread-safe, not-limited code to try something similar at the time
of checking for duplicates (instead of when adding states), using index vectors
on the stack. It did give a 13% improvement with one specially constructed
pattern for certain subject strings, but on other strings and on many of the
simpler patterns in the test suite it did worse. The major problem, I think,
was the extra time to initialize the index. This had to be done for each call
of internal_dfa_match(). (The supplied patch used a static vector, initialized
only once - I suspect this was the cause of the problems with the tests.)
Overall, I concluded that the gains in some cases did not outweigh the losses
in others, so I abandoned this code. */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#define NLBLOCK mb /* Block containing newline information */
#define PSSTART start_subject /* Field containing processed string start */
#define PSEND end_subject /* Field containing processed string end */
#include "pcre2_internal.h"
#define PUBLIC_DFA_MATCH_OPTIONS \
(PCRE2_ANCHORED|PCRE2_ENDANCHORED|PCRE2_NOTBOL|PCRE2_NOTEOL|PCRE2_NOTEMPTY| \
PCRE2_NOTEMPTY_ATSTART|PCRE2_NO_UTF_CHECK|PCRE2_PARTIAL_HARD| \
PCRE2_PARTIAL_SOFT|PCRE2_DFA_SHORTEST|PCRE2_DFA_RESTART| \
PCRE2_COPY_MATCHED_SUBJECT)
/*************************************************
* Code parameters and static tables *
*************************************************/
/* These are offsets that are used to turn the OP_TYPESTAR and friends opcodes
into others, under special conditions. A gap of 20 between the blocks should be
enough. The resulting opcodes don't have to be less than 256 because they are
never stored, so we push them well clear of the normal opcodes. */
#define OP_PROP_EXTRA 300
#define OP_EXTUNI_EXTRA 320
#define OP_ANYNL_EXTRA 340
#define OP_HSPACE_EXTRA 360
#define OP_VSPACE_EXTRA 380
/* This table identifies those opcodes that are followed immediately by a
character that is to be tested in some way. This makes it possible to
centralize the loading of these characters. In the case of Type * etc, the
"character" is the opcode for \D, \d, \S, \s, \W, or \w, which will always be a
small value. Non-zero values in the table are the offsets from the opcode where
the character is to be found. ***NOTE*** If the start of this table is
modified, the three tables that follow must also be modified. */
static const uint8_t coptable[] = {
0, /* End */
0, 0, 0, 0, 0, /* \A, \G, \K, \B, \b */
0, 0, 0, 0, 0, 0, /* \D, \d, \S, \s, \W, \w */
0, 0, 0, /* Any, AllAny, Anybyte */
0, 0, /* \P, \p */
0, 0, 0, 0, 0, /* \R, \H, \h, \V, \v */
0, /* \X */
0, 0, 0, 0, 0, 0, /* \Z, \z, $, $M, ^, ^M */
1, /* Char */
1, /* Chari */
1, /* not */
1, /* noti */
/* Positive single-char repeats */
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
1+IMM2_SIZE, 1+IMM2_SIZE, /* upto, minupto */
1+IMM2_SIZE, /* exact */
1, 1, 1, 1+IMM2_SIZE, /* *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* *I, *?I, +I, +?I, ?I, ??I */
1+IMM2_SIZE, 1+IMM2_SIZE, /* upto I, minupto I */
1+IMM2_SIZE, /* exact I */
1, 1, 1, 1+IMM2_SIZE, /* *+I, ++I, ?+I, upto+I */
/* Negative single-char repeats - only for chars < 256 */
1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */
1+IMM2_SIZE, 1+IMM2_SIZE, /* NOT upto, minupto */
1+IMM2_SIZE, /* NOT exact */
1, 1, 1, 1+IMM2_SIZE, /* NOT *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* NOT *I, *?I, +I, +?I, ?I, ??I */
1+IMM2_SIZE, 1+IMM2_SIZE, /* NOT upto I, minupto I */
1+IMM2_SIZE, /* NOT exact I */
1, 1, 1, 1+IMM2_SIZE, /* NOT *+I, ++I, ?+I, upto+I */
/* Positive type repeats */
1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */
1+IMM2_SIZE, 1+IMM2_SIZE, /* Type upto, minupto */
1+IMM2_SIZE, /* Type exact */
1, 1, 1, 1+IMM2_SIZE, /* Type *+, ++, ?+, upto+ */
/* Character class & ref repeats */
0, 0, 0, 0, 0, 0, /* *, *?, +, +?, ?, ?? */
0, 0, /* CRRANGE, CRMINRANGE */
0, 0, 0, 0, /* Possessive *+, ++, ?+, CRPOSRANGE */
0, /* CLASS */
0, /* NCLASS */
0, /* XCLASS - variable length */
0, /* ECLASS - variable length */
0, /* REF */
0, /* REFI */
0, /* DNREF */
0, /* DNREFI */
0, /* RECURSE */
0, /* CALLOUT */
0, /* CALLOUT_STR */
0, /* Alt */
0, /* Ket */
0, /* KetRmax */
0, /* KetRmin */
0, /* KetRpos */
0, 0, /* Reverse, Vreverse */
0, /* Assert */
0, /* Assert not */
0, /* Assert behind */
0, /* Assert behind not */
0, /* NA assert */
0, /* NA assert behind */
0, /* Assert scan substring */
0, /* ONCE */
0, /* SCRIPT_RUN */
0, 0, 0, 0, 0, /* BRA, BRAPOS, CBRA, CBRAPOS, COND */
0, 0, 0, 0, 0, /* SBRA, SBRAPOS, SCBRA, SCBRAPOS, SCOND */
0, 0, /* CREF, DNCREF */
0, 0, /* RREF, DNRREF */
0, 0, /* FALSE, TRUE */
0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */
0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */
0, 0, 0, 0, /* SKIP, SKIP_ARG, THEN, THEN_ARG */
0, 0, /* COMMIT, COMMIT_ARG */
0, 0, 0, /* FAIL, ACCEPT, ASSERT_ACCEPT */
0, 0, 0, /* CLOSE, SKIPZERO, DEFINE */
0, 0, /* \B and \b in UCP mode */
};
/* This table identifies those opcodes that inspect a character. It is used to
remember the fact that a character could have been inspected when the end of
the subject is reached. ***NOTE*** If the start of this table is modified, the
two tables that follow must also be modified. */
static const uint8_t poptable[] = {
0, /* End */
0, 0, 0, 1, 1, /* \A, \G, \K, \B, \b */
1, 1, 1, 1, 1, 1, /* \D, \d, \S, \s, \W, \w */
1, 1, 1, /* Any, AllAny, Anybyte */
1, 1, /* \P, \p */
1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */
1, /* \X */
0, 0, 0, 0, 0, 0, /* \Z, \z, $, $M, ^, ^M */
1, /* Char */
1, /* Chari */
1, /* not */
1, /* noti */
/* Positive single-char repeats */
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
1, 1, 1, /* upto, minupto, exact */
1, 1, 1, 1, /* *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* *I, *?I, +I, +?I, ?I, ??I */
1, 1, 1, /* upto I, minupto I, exact I */
1, 1, 1, 1, /* *+I, ++I, ?+I, upto+I */
/* Negative single-char repeats - only for chars < 256 */
1, 1, 1, 1, 1, 1, /* NOT *, *?, +, +?, ?, ?? */
1, 1, 1, /* NOT upto, minupto, exact */
1, 1, 1, 1, /* NOT *+, ++, ?+, upto+ */
1, 1, 1, 1, 1, 1, /* NOT *I, *?I, +I, +?I, ?I, ??I */
1, 1, 1, /* NOT upto I, minupto I, exact I */
1, 1, 1, 1, /* NOT *+I, ++I, ?+I, upto+I */
/* Positive type repeats */
1, 1, 1, 1, 1, 1, /* Type *, *?, +, +?, ?, ?? */
1, 1, 1, /* Type upto, minupto, exact */
1, 1, 1, 1, /* Type *+, ++, ?+, upto+ */
/* Character class & ref repeats */
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */
1, 1, /* CRRANGE, CRMINRANGE */
1, 1, 1, 1, /* Possessive *+, ++, ?+, CRPOSRANGE */
1, /* CLASS */
1, /* NCLASS */
1, /* XCLASS - variable length */
1, /* ECLASS - variable length */
0, /* REF */
0, /* REFI */
0, /* DNREF */
0, /* DNREFI */
0, /* RECURSE */
0, /* CALLOUT */
0, /* CALLOUT_STR */
0, /* Alt */
0, /* Ket */
0, /* KetRmax */
0, /* KetRmin */
0, /* KetRpos */
0, 0, /* Reverse, Vreverse */
0, /* Assert */
0, /* Assert not */
0, /* Assert behind */
0, /* Assert behind not */
0, /* NA assert */
0, /* NA assert behind */
0, /* Assert scan substring */
0, /* ONCE */
0, /* SCRIPT_RUN */
0, 0, 0, 0, 0, /* BRA, BRAPOS, CBRA, CBRAPOS, COND */
0, 0, 0, 0, 0, /* SBRA, SBRAPOS, SCBRA, SCBRAPOS, SCOND */
0, 0, /* CREF, DNCREF */
0, 0, /* RREF, DNRREF */
0, 0, /* FALSE, TRUE */
0, 0, 0, /* BRAZERO, BRAMINZERO, BRAPOSZERO */
0, 0, 0, /* MARK, PRUNE, PRUNE_ARG */
0, 0, 0, 0, /* SKIP, SKIP_ARG, THEN, THEN_ARG */
0, 0, /* COMMIT, COMMIT_ARG */
0, 0, 0, /* FAIL, ACCEPT, ASSERT_ACCEPT */
0, 0, 0, /* CLOSE, SKIPZERO, DEFINE */
1, 1, /* \B and \b in UCP mode */
};
/* Compile-time check that these tables have the correct size. */
STATIC_ASSERT(sizeof(coptable) == OP_TABLE_LENGTH, coptable);
STATIC_ASSERT(sizeof(poptable) == OP_TABLE_LENGTH, poptable);
/* These 2 tables allow for compact code for testing for \D, \d, \S, \s, \W,
and \w */
static const uint8_t toptable1[] = {
0, 0, 0, 0, 0, 0,
ctype_digit, ctype_digit,
ctype_space, ctype_space,
ctype_word, ctype_word,
0, 0 /* OP_ANY, OP_ALLANY */
};
static const uint8_t toptable2[] = {
0, 0, 0, 0, 0, 0,
ctype_digit, 0,
ctype_space, 0,
ctype_word, 0,
1, 1 /* OP_ANY, OP_ALLANY */
};
/* Structure for holding data about a particular state, which is in effect the
current data for an active path through the match tree. It must consist
entirely of ints because the working vector we are passed, and which we put
these structures in, is a vector of ints. */
typedef struct stateblock {
int offset; /* Offset to opcode (-ve has meaning) */
int count; /* Count for repeats */
int data; /* Some use extra data */
} stateblock;
#define INTS_PER_STATEBLOCK (int)(sizeof(stateblock)/sizeof(int))
/* Before version 10.32 the recursive calls of internal_dfa_match() were passed
local working space and output vectors that were created on the stack. This has
caused issues for some patterns, especially in small-stack environments such as
Windows. A new scheme is now in use which sets up a vector on the stack, but if
this is too small, heap memory is used, up to the heap_limit. The main
parameters are all numbers of ints because the workspace is a vector of ints.
The size of the starting stack vector, DFA_START_RWS_SIZE, is in bytes, and is
defined in pcre2_internal.h so as to be available to pcre2test when it is
finding the minimum heap requirement for a match. */
#define OVEC_UNIT (sizeof(PCRE2_SIZE)/sizeof(int))
#define RWS_BASE_SIZE (DFA_START_RWS_SIZE/sizeof(int)) /* Stack vector */
#define RWS_RSIZE 1000 /* Work size for recursion */
#define RWS_OVEC_RSIZE (1000*OVEC_UNIT) /* Ovector for recursion */
#define RWS_OVEC_OSIZE (2*OVEC_UNIT) /* Ovector in other cases */
/* This structure is at the start of each workspace block. */
typedef struct RWS_anchor {
struct RWS_anchor *next;
uint32_t size; /* Number of ints */
uint32_t free; /* Number of ints */
} RWS_anchor;
#define RWS_ANCHOR_SIZE (sizeof(RWS_anchor)/sizeof(int))
/*************************************************
* Process a callout *
*************************************************/
/* This function is called to perform a callout.
Arguments:
code current code pointer
offsets points to current capture offsets
current_subject start of current subject match
ptr current position in subject
mb the match block
extracode extra code offset when called from condition
lengthptr where to return the callout length
Returns: the return from the callout
*/
static int
do_callout_dfa(PCRE2_SPTR code, PCRE2_SIZE *offsets, PCRE2_SPTR current_subject,
PCRE2_SPTR ptr, dfa_match_block *mb, PCRE2_SIZE extracode,
PCRE2_SIZE *lengthptr)
{
pcre2_callout_block *cb = mb->cb;
*lengthptr = (code[extracode] == OP_CALLOUT)?
(PCRE2_SIZE)PRIV(OP_lengths)[OP_CALLOUT] :
(PCRE2_SIZE)GET(code, 1 + 2*LINK_SIZE + extracode);
if (mb->callout == NULL) return 0; /* No callout provided */
/* Fixed fields in the callout block are set once and for all at the start of
matching. */
cb->offset_vector = offsets;
cb->start_match = (PCRE2_SIZE)(current_subject - mb->start_subject);
cb->current_position = (PCRE2_SIZE)(ptr - mb->start_subject);
cb->pattern_position = GET(code, 1 + extracode);
cb->next_item_length = GET(code, 1 + LINK_SIZE + extracode);
if (code[extracode] == OP_CALLOUT)
{
cb->callout_number = code[1 + 2*LINK_SIZE + extracode];
cb->callout_string_offset = 0;
cb->callout_string = NULL;
cb->callout_string_length = 0;
}
else
{
cb->callout_number = 0;
cb->callout_string_offset = GET(code, 1 + 3*LINK_SIZE + extracode);
cb->callout_string = code + (1 + 4*LINK_SIZE + extracode) + 1;
cb->callout_string_length = *lengthptr - (1 + 4*LINK_SIZE) - 2;
}
return (mb->callout)(cb, mb->callout_data);
}
/*************************************************
* Expand local workspace memory *
*************************************************/
/* This function is called when internal_dfa_match() is about to be called
recursively and there is insufficient working space left in the current
workspace block. If there's an existing next block, use it; otherwise get a new
block unless the heap limit is reached.
Arguments:
rwsptr pointer to block pointer (updated)
ovecsize space needed for an ovector
mb the match block
Returns: 0 rwsptr has been updated
!0 an error code
*/
static int
more_workspace(RWS_anchor **rwsptr, unsigned int ovecsize, dfa_match_block *mb)
{
RWS_anchor *rws = *rwsptr;
RWS_anchor *new;
if (rws->next != NULL)
{
new = rws->next;
}
/* Sizes in the RWS_anchor blocks are in units of sizeof(int), but
mb->heap_limit and mb->heap_used are in kibibytes. Play carefully, to avoid
overflow. */
else
{
uint32_t newsize = (rws->size >= UINT32_MAX/(sizeof(int)*2))? UINT32_MAX/sizeof(int) : rws->size * 2;
uint32_t newsizeK = newsize/(1024/sizeof(int));
if (newsizeK + mb->heap_used > mb->heap_limit)
newsizeK = (uint32_t)(mb->heap_limit - mb->heap_used);
newsize = newsizeK*(1024/sizeof(int));
if (newsize < RWS_RSIZE + ovecsize + RWS_ANCHOR_SIZE)
return PCRE2_ERROR_HEAPLIMIT;
new = mb->memctl.malloc(newsize*sizeof(int), mb->memctl.memory_data);
if (new == NULL) return PCRE2_ERROR_NOMEMORY;
mb->heap_used += newsizeK;
new->next = NULL;
new->size = newsize;
rws->next = new;
}
new->free = new->size - RWS_ANCHOR_SIZE;
*rwsptr = new;
return 0;
}
/*************************************************
* Match a Regular Expression - DFA engine *
*************************************************/
/* This internal function applies a compiled pattern to a subject string,
starting at a given point, using a DFA engine. This function is called from the
external one, possibly multiple times if the pattern is not anchored. The
function calls itself recursively for some kinds of subpattern.
Arguments:
mb the match_data block with fixed information
this_start_code the opening bracket of this subexpression's code
current_subject where we currently are in the subject string
start_offset start offset in the subject string
offsets vector to contain the matching string offsets
offsetcount size of same
workspace vector of workspace
wscount size of same
rlevel function call recursion level
Returns: > 0 => number of match offset pairs placed in offsets
= 0 => offsets overflowed; longest matches are present
-1 => failed to match
< -1 => some kind of unexpected problem
The following macros are used for adding states to the two state vectors (one
for the current character, one for the following character). */
#define ADD_ACTIVE(x,y) \
if (active_count++ < wscount) \
{ \
next_active_state->offset = (x); \
next_active_state->count = (y); \
next_active_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
#define ADD_ACTIVE_DATA(x,y,z) \
if (active_count++ < wscount) \
{ \
next_active_state->offset = (x); \
next_active_state->count = (y); \
next_active_state->data = (z); \
next_active_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
#define ADD_NEW(x,y) \
if (new_count++ < wscount) \
{ \
next_new_state->offset = (x); \
next_new_state->count = (y); \
next_new_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
#define ADD_NEW_DATA(x,y,z) \
if (new_count++ < wscount) \
{ \
next_new_state->offset = (x); \
next_new_state->count = (y); \
next_new_state->data = (z); \
next_new_state++; \
} \
else return PCRE2_ERROR_DFA_WSSIZE
/* And now, here is the code */
static int
internal_dfa_match(
dfa_match_block *mb,
PCRE2_SPTR this_start_code,
PCRE2_SPTR current_subject,
PCRE2_SIZE start_offset,
PCRE2_SIZE *offsets,
uint32_t offsetcount,
int *workspace,
int wscount,
uint32_t rlevel,
int *RWS)
{
stateblock *active_states, *new_states, *temp_states;
stateblock *next_active_state, *next_new_state;
const uint8_t *ctypes, *lcc, *fcc;
PCRE2_SPTR ptr;
PCRE2_SPTR end_code;
dfa_recursion_info new_recursive;
int active_count, new_count, match_count;
/* Some fields in the mb block are frequently referenced, so we load them into
independent variables in the hope that this will perform better. */
PCRE2_SPTR start_subject = mb->start_subject;
PCRE2_SPTR end_subject = mb->end_subject;
PCRE2_SPTR start_code = mb->start_code;
#ifdef SUPPORT_UNICODE
BOOL utf = (mb->poptions & PCRE2_UTF) != 0;
BOOL utf_or_ucp = utf || (mb->poptions & PCRE2_UCP) != 0;
#else
BOOL utf = FALSE;
#endif
BOOL reset_could_continue = FALSE;
if (mb->match_call_count++ >= mb->match_limit) return PCRE2_ERROR_MATCHLIMIT;
if (rlevel++ > mb->match_limit_depth) return PCRE2_ERROR_DEPTHLIMIT;
offsetcount &= (uint32_t)(-2); /* Round down */
wscount -= 2;
wscount = (wscount - (wscount % (INTS_PER_STATEBLOCK * 2))) /
(2 * INTS_PER_STATEBLOCK);
ctypes = mb->tables + ctypes_offset;
lcc = mb->tables + lcc_offset;
fcc = mb->tables + fcc_offset;
match_count = PCRE2_ERROR_NOMATCH; /* A negative number */
active_states = (stateblock *)(workspace + 2);
next_new_state = new_states = active_states + wscount;
new_count = 0;
/* The first thing in any (sub) pattern is a bracket of some sort. Push all
the alternative states onto the list, and find out where the end is. This
makes is possible to use this function recursively, when we want to stop at a
matching internal ket rather than at the end.
If we are dealing with a backward assertion we have to find out the maximum
amount to move back, and set up each alternative appropriately. */
if (*this_start_code == OP_ASSERTBACK || *this_start_code == OP_ASSERTBACK_NOT)
{
size_t max_back = 0;
size_t gone_back;
end_code = this_start_code;
do
{
size_t back = (size_t)GET2(end_code, 2+LINK_SIZE);
if (back > max_back) max_back = back;
end_code += GET(end_code, 1);
}
while (*end_code == OP_ALT);
/* If we can't go back the amount required for the longest lookbehind
pattern, go back as far as we can; some alternatives may still be viable. */
#ifdef SUPPORT_UNICODE
/* In character mode we have to step back character by character */
if (utf)
{
for (gone_back = 0; gone_back < max_back; gone_back++)
{
if (current_subject <= start_subject) break;
current_subject--;
ACROSSCHAR(current_subject > start_subject, current_subject,
current_subject--);
}
}
else
#endif
/* In byte-mode we can do this quickly. */
{
size_t current_offset = (size_t)(current_subject - start_subject);
gone_back = (current_offset < max_back)? current_offset : max_back;
current_subject -= gone_back;
}
/* Save the earliest consulted character */
if (current_subject < mb->start_used_ptr)
mb->start_used_ptr = current_subject;
/* Now we can process the individual branches. There will be an OP_REVERSE at
the start of each branch, except when the length of the branch is zero. */
end_code = this_start_code;
do
{
uint32_t revlen = (end_code[1+LINK_SIZE] == OP_REVERSE)? 1 + IMM2_SIZE : 0;
size_t back = (revlen == 0)? 0 : (size_t)GET2(end_code, 2+LINK_SIZE);
if (back <= gone_back)
{
int bstate = (int)(end_code - start_code + 1 + LINK_SIZE + revlen);
ADD_NEW_DATA(-bstate, 0, (int)(gone_back - back));
}
end_code += GET(end_code, 1);
}
while (*end_code == OP_ALT);
}
/* This is the code for a "normal" subpattern (not a backward assertion). The
start of a whole pattern is always one of these. If we are at the top level,
we may be asked to restart matching from the same point that we reached for a
previous partial match. We still have to scan through the top-level branches to
find the end state. */
else
{
end_code = this_start_code;
/* Restarting */
if (rlevel == 1 && (mb->moptions & PCRE2_DFA_RESTART) != 0)
{
do { end_code += GET(end_code, 1); } while (*end_code == OP_ALT);
new_count = workspace[1];
if (!workspace[0])
memcpy(new_states, active_states, (size_t)new_count * sizeof(stateblock));
}
/* Not restarting */
else
{
int length = 1 + LINK_SIZE +
((*this_start_code == OP_CBRA || *this_start_code == OP_SCBRA ||
*this_start_code == OP_CBRAPOS || *this_start_code == OP_SCBRAPOS)
? IMM2_SIZE:0);
do
{
ADD_NEW((int)(end_code - start_code + length), 0);
end_code += GET(end_code, 1);
length = 1 + LINK_SIZE;
}
while (*end_code == OP_ALT);
}
}
workspace[0] = 0; /* Bit indicating which vector is current */
/* Loop for scanning the subject */
ptr = current_subject;
for (;;)
{
int i, j;
int clen, dlen;
uint32_t c, d;
BOOL partial_newline = FALSE;
BOOL could_continue = reset_could_continue;
reset_could_continue = FALSE;
if (ptr > mb->last_used_ptr) mb->last_used_ptr = ptr;
/* Make the new state list into the active state list and empty the
new state list. */
temp_states = active_states;
active_states = new_states;
new_states = temp_states;
active_count = new_count;
new_count = 0;
workspace[0] ^= 1; /* Remember for the restarting feature */
workspace[1] = active_count;
/* Set the pointers for adding new states */
next_active_state = active_states + active_count;
next_new_state = new_states;
/* Load the current character from the subject outside the loop, as many
different states may want to look at it, and we assume that at least one
will. */
if (ptr < end_subject)
{
clen = 1; /* Number of data items in the character */
#ifdef SUPPORT_UNICODE
GETCHARLENTEST(c, ptr, clen);
#else
c = *ptr;
#endif /* SUPPORT_UNICODE */
}
else
{
clen = 0; /* This indicates the end of the subject */
c = NOTACHAR; /* This value should never actually be used */
}
/* Scan up the active states and act on each one. The result of an action
may be to add more states to the currently active list (e.g. on hitting a
parenthesis) or it may be to put states on the new list, for considering
when we move the character pointer on. */
for (i = 0; i < active_count; i++)
{
stateblock *current_state = active_states + i;
BOOL caseless = FALSE;
PCRE2_SPTR code;
uint32_t codevalue;
int state_offset = current_state->offset;
int rrc;
int count;
/* A negative offset is a special case meaning "hold off going to this
(negated) state until the number of characters in the data field have
been skipped". If the could_continue flag was passed over from a previous
state, arrange for it to passed on. */
if (state_offset < 0)
{
if (current_state->data > 0)
{
ADD_NEW_DATA(state_offset, current_state->count,
current_state->data - 1);
if (could_continue) reset_could_continue = TRUE;
continue;
}
else
{
current_state->offset = state_offset = -state_offset;
}
}
/* Check for a duplicate state with the same count, and skip if found.
See the note at the head of this module about the possibility of improving
performance here. */
for (j = 0; j < i; j++)
{
if (active_states[j].offset == state_offset &&
active_states[j].count == current_state->count)
goto NEXT_ACTIVE_STATE;
}
/* The state offset is the offset to the opcode */
code = start_code + state_offset;
codevalue = *code;
/* If this opcode inspects a character, but we are at the end of the
subject, remember the fact for use when testing for a partial match. */
if (clen == 0 && poptable[codevalue] != 0)
could_continue = TRUE;
/* If this opcode is followed by an inline character, load it. It is
tempting to test for the presence of a subject character here, but that
is wrong, because sometimes zero repetitions of the subject are
permitted.
We also use this mechanism for opcodes such as OP_TYPEPLUS that take an
argument that is not a data character - but is always one byte long because
the values are small. We have to take special action to deal with \P, \p,
\H, \h, \V, \v and \X in this case. To keep the other cases fast, convert
these ones to new opcodes. */
if (coptable[codevalue] > 0)
{
dlen = 1;
#ifdef SUPPORT_UNICODE
if (utf) { GETCHARLEN(d, (code + coptable[codevalue]), dlen); } else
#endif /* SUPPORT_UNICODE */
d = code[coptable[codevalue]];
if (codevalue >= OP_TYPESTAR)
{
switch(d)
{
case OP_ANYBYTE: return PCRE2_ERROR_DFA_UITEM;
case OP_NOTPROP:
case OP_PROP: codevalue += OP_PROP_EXTRA; break;
case OP_ANYNL: codevalue += OP_ANYNL_EXTRA; break;
case OP_EXTUNI: codevalue += OP_EXTUNI_EXTRA; break;
case OP_NOT_HSPACE:
case OP_HSPACE: codevalue += OP_HSPACE_EXTRA; break;
case OP_NOT_VSPACE:
case OP_VSPACE: codevalue += OP_VSPACE_EXTRA; break;
default: break;
}
}
}
else
{
dlen = 0; /* Not strictly necessary, but compilers moan */
d = NOTACHAR; /* if these variables are not set. */
}
/* Now process the individual opcodes */
switch (codevalue)
{
/* ========================================================================== */
/* Reached a closing bracket. If not at the end of the pattern, carry
on with the next opcode. For repeating opcodes, also add the repeat
state. Note that KETRPOS will always be encountered at the end of the
subpattern, because the possessive subpattern repeats are always handled
using recursive calls. Thus, it never adds any new states.
At the end of the (sub)pattern, unless we have an empty string and
PCRE2_NOTEMPTY is set, or PCRE2_NOTEMPTY_ATSTART is set and we are at the
start of the subject, save the match data, shifting up all previous
matches so we always have the longest first. */
case OP_KET:
case OP_KETRMIN:
case OP_KETRMAX:
case OP_KETRPOS:
if (code != end_code)
{
ADD_ACTIVE(state_offset + 1 + LINK_SIZE, 0);
if (codevalue != OP_KET)
{
ADD_ACTIVE(state_offset - (int)GET(code, 1), 0);
}
}
else
{
if (ptr > current_subject ||
((mb->moptions & PCRE2_NOTEMPTY) == 0 &&
((mb->moptions & PCRE2_NOTEMPTY_ATSTART) == 0 ||
current_subject > start_subject + mb->start_offset)))
{
if (match_count < 0) match_count = (offsetcount >= 2)? 1 : 0;
else if (match_count > 0 && ++match_count * 2 > (int)offsetcount)
match_count = 0;
count = ((match_count == 0)? (int)offsetcount : match_count * 2) - 2;
if (count > 0) (void)memmove(offsets + 2, offsets,
(size_t)count * sizeof(PCRE2_SIZE));
if (offsetcount >= 2)
{
offsets[0] = (PCRE2_SIZE)(current_subject - start_subject);
offsets[1] = (PCRE2_SIZE)(ptr - start_subject);
}
if ((mb->moptions & PCRE2_DFA_SHORTEST) != 0) return match_count;
}
}
break;
/* ========================================================================== */
/* These opcodes add to the current list of states without looking
at the current character. */
/*-----------------------------------------------------------------*/
case OP_ALT:
do { code += GET(code, 1); } while (*code == OP_ALT);
ADD_ACTIVE((int)(code - start_code), 0);
break;
/*-----------------------------------------------------------------*/
case OP_BRA:
case OP_SBRA:
do
{
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
code += GET(code, 1);
}
while (*code == OP_ALT);
break;
/*-----------------------------------------------------------------*/
case OP_CBRA:
case OP_SCBRA:
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE + IMM2_SIZE), 0);
code += GET(code, 1);
while (*code == OP_ALT)
{
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
code += GET(code, 1);
}
break;
/*-----------------------------------------------------------------*/
case OP_BRAZERO:
case OP_BRAMINZERO:
ADD_ACTIVE(state_offset + 1, 0);
code += 1 + GET(code, 2);
while (*code == OP_ALT) code += GET(code, 1);
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
break;
/*-----------------------------------------------------------------*/
case OP_SKIPZERO:
code += 1 + GET(code, 2);
while (*code == OP_ALT) code += GET(code, 1);
ADD_ACTIVE((int)(code - start_code + 1 + LINK_SIZE), 0);
break;
/*-----------------------------------------------------------------*/
case OP_CIRC:
if (ptr == start_subject && (mb->moptions & PCRE2_NOTBOL) == 0)
{ ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_CIRCM:
if ((ptr == start_subject && (mb->moptions & PCRE2_NOTBOL) == 0) ||
((ptr != end_subject || (mb->poptions & PCRE2_ALT_CIRCUMFLEX) != 0 )
&& WAS_NEWLINE(ptr)))
{ ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_EOD:
if (ptr >= end_subject)
{
if ((mb->moptions & PCRE2_PARTIAL_HARD) != 0)
return PCRE2_ERROR_PARTIAL;
else { ADD_ACTIVE(state_offset + 1, 0); }
}
break;
/*-----------------------------------------------------------------*/
case OP_SOD:
if (ptr == start_subject) { ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_SOM:
if (ptr == start_subject + start_offset) { ADD_ACTIVE(state_offset + 1, 0); }
break;
/* ========================================================================== */
/* These opcodes inspect the next subject character, and sometimes
the previous one as well, but do not have an argument. The variable
clen contains the length of the current character and is zero if we are
at the end of the subject. */
/*-----------------------------------------------------------------*/
case OP_ANY:
if (clen > 0 && !IS_NEWLINE(ptr))
{
if (ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else
{
ADD_NEW(state_offset + 1, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_ALLANY:
if (clen > 0)
{ ADD_NEW(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_EODN:
if (clen == 0 || (IS_NEWLINE(ptr) && ptr == end_subject - mb->nllen))
{
if ((mb->moptions & PCRE2_PARTIAL_HARD) != 0)
return PCRE2_ERROR_PARTIAL;
ADD_ACTIVE(state_offset + 1, 0);
}
break;
/*-----------------------------------------------------------------*/
case OP_DOLL:
if ((mb->moptions & PCRE2_NOTEOL) == 0)
{
if (clen == 0 && (mb->moptions & PCRE2_PARTIAL_HARD) != 0)
could_continue = TRUE;
else if (clen == 0 ||
((mb->poptions & PCRE2_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr) &&
(ptr == end_subject - mb->nllen)
))
{ ADD_ACTIVE(state_offset + 1, 0); }
else if (ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
if ((mb->moptions & PCRE2_PARTIAL_HARD) != 0)
{
reset_could_continue = TRUE;
ADD_NEW_DATA(-(state_offset + 1), 0, 1);
}
else could_continue = partial_newline = TRUE;
}
}
break;
/*-----------------------------------------------------------------*/
case OP_DOLLM:
if ((mb->moptions & PCRE2_NOTEOL) == 0)
{
if (clen == 0 && (mb->moptions & PCRE2_PARTIAL_HARD) != 0)
could_continue = TRUE;
else if (clen == 0 ||
((mb->poptions & PCRE2_DOLLAR_ENDONLY) == 0 && IS_NEWLINE(ptr)))
{ ADD_ACTIVE(state_offset + 1, 0); }
else if (ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
if ((mb->moptions & PCRE2_PARTIAL_HARD) != 0)
{
reset_could_continue = TRUE;
ADD_NEW_DATA(-(state_offset + 1), 0, 1);
}
else could_continue = partial_newline = TRUE;
}
}
else if (IS_NEWLINE(ptr))
{ ADD_ACTIVE(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_DIGIT:
case OP_WHITESPACE:
case OP_WORDCHAR:
if (clen > 0 && c < 256 &&
((ctypes[c] & toptable1[codevalue]) ^ toptable2[codevalue]) != 0)
{ ADD_NEW(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_NOT_DIGIT:
case OP_NOT_WHITESPACE:
case OP_NOT_WORDCHAR:
if (clen > 0 && (c >= 256 ||
((ctypes[c] & toptable1[codevalue]) ^ toptable2[codevalue]) != 0))
{ ADD_NEW(state_offset + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_WORD_BOUNDARY:
case OP_NOT_WORD_BOUNDARY:
case OP_NOT_UCP_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
{
int left_word, right_word;
if (ptr > start_subject)
{
PCRE2_SPTR temp = ptr - 1;
if (temp < mb->start_used_ptr) mb->start_used_ptr = temp;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (utf) { BACKCHAR(temp); }
#endif
GETCHARTEST(d, temp);
#ifdef SUPPORT_UNICODE
if (codevalue == OP_UCP_WORD_BOUNDARY ||
codevalue == OP_NOT_UCP_WORD_BOUNDARY)
{
int chartype = UCD_CHARTYPE(d);
int category = PRIV(ucp_gentype)[chartype];
left_word = (category == ucp_L || category == ucp_N ||
chartype == ucp_Mn || chartype == ucp_Pc);
}
else
#endif
left_word = d < 256 && (ctypes[d] & ctype_word) != 0;
}
else left_word = FALSE;
if (clen > 0)
{
if (ptr >= mb->last_used_ptr)
{
PCRE2_SPTR temp = ptr + 1;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (utf) { FORWARDCHARTEST(temp, mb->end_subject); }
#endif
mb->last_used_ptr = temp;
}
#ifdef SUPPORT_UNICODE
if (codevalue == OP_UCP_WORD_BOUNDARY ||
codevalue == OP_NOT_UCP_WORD_BOUNDARY)
{
int chartype = UCD_CHARTYPE(c);
int category = PRIV(ucp_gentype)[chartype];
right_word = (category == ucp_L || category == ucp_N ||
chartype == ucp_Mn || chartype == ucp_Pc);
}
else
#endif
right_word = c < 256 && (ctypes[c] & ctype_word) != 0;
}
else right_word = FALSE;
if ((left_word == right_word) ==
(codevalue == OP_NOT_WORD_BOUNDARY ||
codevalue == OP_NOT_UCP_WORD_BOUNDARY))
{ ADD_ACTIVE(state_offset + 1, 0); }
}
break;
/*-----------------------------------------------------------------*/
/* Check the next character by Unicode property. We will get here only
if the support is in the binary; otherwise a compile-time error occurs.
*/
#ifdef SUPPORT_UNICODE
case OP_PROP:
case OP_NOTPROP:
if (clen > 0)
{
BOOL OK;
int chartype;
const uint32_t *cp;
const ucd_record * prop = GET_UCD(c);
switch(code[1])
{
case PT_LAMP:
chartype = prop->chartype;
OK = chartype == ucp_Lu || chartype == ucp_Ll ||
chartype == ucp_Lt;
break;
case PT_GC:
OK = PRIV(ucp_gentype)[prop->chartype] == code[2];
break;
case PT_PC:
OK = prop->chartype == code[2];
break;
case PT_SC:
OK = prop->script == code[2];
break;
case PT_SCX:
OK = (prop->script == code[2] ||
MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), code[2]) != 0);
break;
/* These are specials for combination cases. */
case PT_ALNUM:
chartype = prop->chartype;
OK = PRIV(ucp_gentype)[chartype] == ucp_L ||
PRIV(ucp_gentype)[chartype] == ucp_N;
break;
/* Perl space used to exclude VT, but from Perl 5.18 it is included,
which means that Perl space and POSIX space are now identical. PCRE
was changed at release 8.34. */
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_CASES:
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z;
break;
}
break;
case PT_WORD:
chartype = prop->chartype;
OK = PRIV(ucp_gentype)[chartype] == ucp_L ||
PRIV(ucp_gentype)[chartype] == ucp_N ||
chartype == ucp_Mn || chartype == ucp_Pc;
break;
case PT_CLIST:
#if PCRE2_CODE_UNIT_WIDTH == 32
if (c > MAX_UTF_CODE_POINT)
{
OK = FALSE;
break;
}
#endif
cp = PRIV(ucd_caseless_sets) + code[2];
for (;;)
{
if (c < *cp) { OK = FALSE; break; }
if (c == *cp++) { OK = TRUE; break; }
}
break;
case PT_UCNC:
OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
c >= 0xe000;
break;
case PT_BIDICL:
OK = UCD_BIDICLASS(c) == code[2];
break;
case PT_BOOL:
OK = MAPBIT(PRIV(ucd_boolprop_sets) +
UCD_BPROPS_PROP(prop), code[2]) != 0;
break;
/* Should never occur, but keep compilers from grumbling. */
default:
OK = codevalue != OP_PROP;
break;
}
if (OK == (codevalue == OP_PROP)) { ADD_NEW(state_offset + 3, 0); }
}
break;
#endif
/* ========================================================================== */
/* These opcodes likewise inspect the subject character, but have an
argument that is not a data character. It is one of these opcodes:
OP_ANY, OP_ALLANY, OP_DIGIT, OP_NOT_DIGIT, OP_WHITESPACE, OP_NOT_SPACE,
OP_WORDCHAR, OP_NOT_WORDCHAR. The value is loaded into d. */
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
if (clen > 0)
{
if (d == OP_ANY && ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
(c < 256 &&
(d != OP_ANY || !IS_NEWLINE(ptr)) &&
((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
{
if (count > 0 && codevalue == OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW(state_offset, count);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSQUERY:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
if (d == OP_ANY && ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
(c < 256 &&
(d != OP_ANY || !IS_NEWLINE(ptr)) &&
((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
{
if (codevalue == OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset + 2, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPOSSTAR:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
if (d == OP_ANY && ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
(c < 256 &&
(d != OP_ANY || !IS_NEWLINE(ptr)) &&
((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
{
if (codevalue == OP_TYPEPOSSTAR)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_TYPEEXACT:
count = current_state->count; /* Number already matched */
if (clen > 0)
{
if (d == OP_ANY && ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
(c < 256 &&
(d != OP_ANY || !IS_NEWLINE(ptr)) &&
((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
{
if (++count >= (int)GET2(code, 1))
{ ADD_NEW(state_offset + 1 + IMM2_SIZE + 1, 0); }
else
{ ADD_NEW(state_offset, count); }
}
}
break;
/*-----------------------------------------------------------------*/
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0);
count = current_state->count; /* Number already matched */
if (clen > 0)
{
if (d == OP_ANY && ptr + 1 >= mb->end_subject &&
(mb->moptions & (PCRE2_PARTIAL_HARD)) != 0 &&
NLBLOCK->nltype == NLTYPE_FIXED &&
NLBLOCK->nllen == 2 &&
c == NLBLOCK->nl[0])
{
could_continue = partial_newline = TRUE;
}
else if ((c >= 256 && d != OP_DIGIT && d != OP_WHITESPACE && d != OP_WORDCHAR) ||
(c < 256 &&
(d != OP_ANY || !IS_NEWLINE(ptr)) &&
((ctypes[c] & toptable1[d]) ^ toptable2[d]) != 0))
{
if (codevalue == OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW(state_offset + 2 + IMM2_SIZE, 0); }
else
{ ADD_NEW(state_offset, count); }
}
}
break;
/* ========================================================================== */
/* These are virtual opcodes that are used when something like
OP_TYPEPLUS has OP_PROP, OP_NOTPROP, OP_ANYNL, or OP_EXTUNI as its
argument. It keeps the code above fast for the other cases. The argument
is in the d variable. */
#ifdef SUPPORT_UNICODE
case OP_PROP_EXTRA + OP_TYPEPLUS:
case OP_PROP_EXTRA + OP_TYPEMINPLUS:
case OP_PROP_EXTRA + OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 4, 0); }
if (clen > 0)
{
BOOL OK;
int chartype;
const uint32_t *cp;
const ucd_record * prop = GET_UCD(c);
switch(code[2])
{
case PT_LAMP:
chartype = prop->chartype;
OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;
break;
case PT_GC:
OK = PRIV(ucp_gentype)[prop->chartype] == code[3];
break;
case PT_PC:
OK = prop->chartype == code[3];
break;
case PT_SC:
OK = prop->script == code[3];
break;
case PT_SCX:
OK = (prop->script == code[3] ||
MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), code[3]) != 0);
break;
/* These are specials for combination cases. */
case PT_ALNUM:
chartype = prop->chartype;
OK = PRIV(ucp_gentype)[chartype] == ucp_L ||
PRIV(ucp_gentype)[chartype] == ucp_N;
break;
/* Perl space used to exclude VT, but from Perl 5.18 it is included,
which means that Perl space and POSIX space are now identical. PCRE
was changed at release 8.34. */
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_CASES:
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z;
break;
}
break;
case PT_WORD:
chartype = prop->chartype;
OK = PRIV(ucp_gentype)[chartype] == ucp_L ||
PRIV(ucp_gentype)[chartype] == ucp_N ||
chartype == ucp_Mn || chartype == ucp_Pc;
break;
case PT_CLIST:
#if PCRE2_CODE_UNIT_WIDTH == 32
if (c > MAX_UTF_CODE_POINT)
{
OK = FALSE;
break;
}
#endif
cp = PRIV(ucd_caseless_sets) + code[3];
for (;;)
{
if (c < *cp) { OK = FALSE; break; }
if (c == *cp++) { OK = TRUE; break; }
}
break;
case PT_UCNC:
OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
c >= 0xe000;
break;
case PT_BIDICL:
OK = UCD_BIDICLASS(c) == code[3];
break;
case PT_BOOL:
OK = MAPBIT(PRIV(ucd_boolprop_sets) +
UCD_BPROPS_PROP(prop), code[3]) != 0;
break;
/* Should never occur, but keep compilers from grumbling. */
default:
OK = codevalue != OP_PROP;
break;
}
if (OK == (d == OP_PROP))
{
if (count > 0 && codevalue == OP_PROP_EXTRA + OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW(state_offset, count);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_EXTUNI_EXTRA + OP_TYPEPLUS:
case OP_EXTUNI_EXTRA + OP_TYPEMINPLUS:
case OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
if (clen > 0)
{
int ncount = 0;
if (count > 0 && codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
(void)PRIV(extuni)(c, ptr + clen, mb->start_subject, end_subject, utf,
&ncount);
count++;
ADD_NEW_DATA(-state_offset, count, ncount);
}
break;
#endif
/*-----------------------------------------------------------------*/
case OP_ANYNL_EXTRA + OP_TYPEPLUS:
case OP_ANYNL_EXTRA + OP_TYPEMINPLUS:
case OP_ANYNL_EXTRA + OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
if (clen > 0)
{
int ncount = 0;
switch (c)
{
case CHAR_VT:
case CHAR_FF:
case CHAR_NEL:
#ifndef EBCDIC
case 0x2028:
case 0x2029:
#endif /* Not EBCDIC */
if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) break;
goto ANYNL01;
case CHAR_CR:
if (ptr + 1 < end_subject && UCHAR21TEST(ptr + 1) == CHAR_LF) ncount = 1;
/* Fall through */
ANYNL01:
case CHAR_LF:
if (count > 0 && codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW_DATA(-state_offset, count, ncount);
break;
default:
break;
}
}
break;
/*-----------------------------------------------------------------*/
case OP_VSPACE_EXTRA + OP_TYPEPLUS:
case OP_VSPACE_EXTRA + OP_TYPEMINPLUS:
case OP_VSPACE_EXTRA + OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
if (clen > 0)
{
BOOL OK;
switch (c)
{
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
break;
}
if (OK == (d == OP_VSPACE))
{
if (count > 0 && codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW_DATA(-state_offset, count, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_HSPACE_EXTRA + OP_TYPEPLUS:
case OP_HSPACE_EXTRA + OP_TYPEMINPLUS:
case OP_HSPACE_EXTRA + OP_TYPEPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + 2, 0); }
if (clen > 0)
{
BOOL OK;
switch (c)
{
HSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
break;
}
if (OK == (d == OP_HSPACE))
{
if (count > 0 && codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW_DATA(-state_offset, count, 0);
}
}
break;
/*-----------------------------------------------------------------*/
#ifdef SUPPORT_UNICODE
case OP_PROP_EXTRA + OP_TYPEQUERY:
case OP_PROP_EXTRA + OP_TYPEMINQUERY:
case OP_PROP_EXTRA + OP_TYPEPOSQUERY:
count = 4;
goto QS1;
case OP_PROP_EXTRA + OP_TYPESTAR:
case OP_PROP_EXTRA + OP_TYPEMINSTAR:
case OP_PROP_EXTRA + OP_TYPEPOSSTAR:
count = 0;
QS1:
ADD_ACTIVE(state_offset + 4, 0);
if (clen > 0)
{
BOOL OK;
int chartype;
const uint32_t *cp;
const ucd_record * prop = GET_UCD(c);
switch(code[2])
{
case PT_LAMP:
chartype = prop->chartype;
OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;
break;
case PT_GC:
OK = PRIV(ucp_gentype)[prop->chartype] == code[3];
break;
case PT_PC:
OK = prop->chartype == code[3];
break;
case PT_SC:
OK = prop->script == code[3];
break;
case PT_SCX:
OK = (prop->script == code[3] ||
MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), code[3]) != 0);
break;
/* These are specials for combination cases. */
case PT_ALNUM:
chartype = prop->chartype;
OK = PRIV(ucp_gentype)[chartype] == ucp_L ||
PRIV(ucp_gentype)[chartype] == ucp_N;
break;
/* Perl space used to exclude VT, but from Perl 5.18 it is included,
which means that Perl space and POSIX space are now identical. PCRE
was changed at release 8.34. */
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_CASES:
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z;
break;
}
break;
case PT_WORD:
chartype = prop->chartype;
OK = PRIV(ucp_gentype)[chartype] == ucp_L ||
PRIV(ucp_gentype)[chartype] == ucp_N ||
chartype == ucp_Mn || chartype == ucp_Pc;
break;
case PT_CLIST:
#if PCRE2_CODE_UNIT_WIDTH == 32
if (c > MAX_UTF_CODE_POINT)
{
OK = FALSE;
break;
}
#endif
cp = PRIV(ucd_caseless_sets) + code[3];
for (;;)
{
if (c < *cp) { OK = FALSE; break; }
if (c == *cp++) { OK = TRUE; break; }
}
break;
case PT_UCNC:
OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
c >= 0xe000;
break;
case PT_BIDICL:
OK = UCD_BIDICLASS(c) == code[3];
break;
case PT_BOOL:
OK = MAPBIT(PRIV(ucd_boolprop_sets) +
UCD_BPROPS_PROP(prop), code[3]) != 0;
break;
/* Should never occur, but keep compilers from grumbling. */
default:
OK = codevalue != OP_PROP;
break;
}
if (OK == (d == OP_PROP))
{
if (codevalue == OP_PROP_EXTRA + OP_TYPEPOSSTAR ||
codevalue == OP_PROP_EXTRA + OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset + count, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_EXTUNI_EXTRA + OP_TYPEQUERY:
case OP_EXTUNI_EXTRA + OP_TYPEMINQUERY:
case OP_EXTUNI_EXTRA + OP_TYPEPOSQUERY:
count = 2;
goto QS2;
case OP_EXTUNI_EXTRA + OP_TYPESTAR:
case OP_EXTUNI_EXTRA + OP_TYPEMINSTAR:
case OP_EXTUNI_EXTRA + OP_TYPEPOSSTAR:
count = 0;
QS2:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
int ncount = 0;
if (codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSSTAR ||
codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
(void)PRIV(extuni)(c, ptr + clen, mb->start_subject, end_subject, utf,
&ncount);
ADD_NEW_DATA(-(state_offset + count), 0, ncount);
}
break;
#endif
/*-----------------------------------------------------------------*/
case OP_ANYNL_EXTRA + OP_TYPEQUERY:
case OP_ANYNL_EXTRA + OP_TYPEMINQUERY:
case OP_ANYNL_EXTRA + OP_TYPEPOSQUERY:
count = 2;
goto QS3;
case OP_ANYNL_EXTRA + OP_TYPESTAR:
case OP_ANYNL_EXTRA + OP_TYPEMINSTAR:
case OP_ANYNL_EXTRA + OP_TYPEPOSSTAR:
count = 0;
QS3:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
int ncount = 0;
switch (c)
{
case CHAR_VT:
case CHAR_FF:
case CHAR_NEL:
#ifndef EBCDIC
case 0x2028:
case 0x2029:
#endif /* Not EBCDIC */
if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) break;
goto ANYNL02;
case CHAR_CR:
if (ptr + 1 < end_subject && UCHAR21TEST(ptr + 1) == CHAR_LF) ncount = 1;
/* Fall through */
ANYNL02:
case CHAR_LF:
if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSSTAR ||
codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW_DATA(-(state_offset + (int)count), 0, ncount);
break;
default:
break;
}
}
break;
/*-----------------------------------------------------------------*/
case OP_VSPACE_EXTRA + OP_TYPEQUERY:
case OP_VSPACE_EXTRA + OP_TYPEMINQUERY:
case OP_VSPACE_EXTRA + OP_TYPEPOSQUERY:
count = 2;
goto QS4;
case OP_VSPACE_EXTRA + OP_TYPESTAR:
case OP_VSPACE_EXTRA + OP_TYPEMINSTAR:
case OP_VSPACE_EXTRA + OP_TYPEPOSSTAR:
count = 0;
QS4:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
BOOL OK;
switch (c)
{
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
break;
}
if (OK == (d == OP_VSPACE))
{
if (codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSSTAR ||
codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW_DATA(-(state_offset + (int)count), 0, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_HSPACE_EXTRA + OP_TYPEQUERY:
case OP_HSPACE_EXTRA + OP_TYPEMINQUERY:
case OP_HSPACE_EXTRA + OP_TYPEPOSQUERY:
count = 2;
goto QS5;
case OP_HSPACE_EXTRA + OP_TYPESTAR:
case OP_HSPACE_EXTRA + OP_TYPEMINSTAR:
case OP_HSPACE_EXTRA + OP_TYPEPOSSTAR:
count = 0;
QS5:
ADD_ACTIVE(state_offset + 2, 0);
if (clen > 0)
{
BOOL OK;
switch (c)
{
HSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
break;
}
if (OK == (d == OP_HSPACE))
{
if (codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSSTAR ||
codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW_DATA(-(state_offset + (int)count), 0, 0);
}
}
break;
/*-----------------------------------------------------------------*/
#ifdef SUPPORT_UNICODE
case OP_PROP_EXTRA + OP_TYPEEXACT:
case OP_PROP_EXTRA + OP_TYPEUPTO:
case OP_PROP_EXTRA + OP_TYPEMINUPTO:
case OP_PROP_EXTRA + OP_TYPEPOSUPTO:
if (codevalue != OP_PROP_EXTRA + OP_TYPEEXACT)
{ ADD_ACTIVE(state_offset + 1 + IMM2_SIZE + 3, 0); }
count = current_state->count; /* Number already matched */
if (clen > 0)
{
BOOL OK;
int chartype;
const uint32_t *cp;
const ucd_record * prop = GET_UCD(c);
switch(code[1 + IMM2_SIZE + 1])
{
case PT_LAMP:
chartype = prop->chartype;
OK = chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt;
break;
case PT_GC:
OK = PRIV(ucp_gentype)[prop->chartype] == code[1 + IMM2_SIZE + 2];
break;
case PT_PC:
OK = prop->chartype == code[1 + IMM2_SIZE + 2];
break;
case PT_SC:
OK = prop->script == code[1 + IMM2_SIZE + 2];
break;
case PT_SCX:
OK = (prop->script == code[1 + IMM2_SIZE + 2] ||
MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop),
code[1 + IMM2_SIZE + 2]) != 0);
break;
/* These are specials for combination cases. */
case PT_ALNUM:
chartype = prop->chartype;
OK = PRIV(ucp_gentype)[chartype] == ucp_L ||
PRIV(ucp_gentype)[chartype] == ucp_N;
break;
/* Perl space used to exclude VT, but from Perl 5.18 it is included,
which means that Perl space and POSIX space are now identical. PCRE
was changed at release 8.34. */
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_CASES:
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = PRIV(ucp_gentype)[prop->chartype] == ucp_Z;
break;
}
break;
case PT_WORD:
chartype = prop->chartype;
OK = PRIV(ucp_gentype)[chartype] == ucp_L ||
PRIV(ucp_gentype)[chartype] == ucp_N ||
chartype == ucp_Mn || chartype == ucp_Pc;
break;
case PT_CLIST:
#if PCRE2_CODE_UNIT_WIDTH == 32
if (c > MAX_UTF_CODE_POINT)
{
OK = FALSE;
break;
}
#endif
cp = PRIV(ucd_caseless_sets) + code[1 + IMM2_SIZE + 2];
for (;;)
{
if (c < *cp) { OK = FALSE; break; }
if (c == *cp++) { OK = TRUE; break; }
}
break;
case PT_UCNC:
OK = c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT || (c >= 0xa0 && c <= 0xd7ff) ||
c >= 0xe000;
break;
case PT_BIDICL:
OK = UCD_BIDICLASS(c) == code[1 + IMM2_SIZE + 2];
break;
case PT_BOOL:
OK = MAPBIT(PRIV(ucd_boolprop_sets) +
UCD_BPROPS_PROP(prop), code[1 + IMM2_SIZE + 2]) != 0;
break;
/* Should never occur, but keep compilers from grumbling. */
default:
OK = codevalue != OP_PROP;
break;
}
if (OK == (d == OP_PROP))
{
if (codevalue == OP_PROP_EXTRA + OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW(state_offset + 1 + IMM2_SIZE + 3, 0); }
else
{ ADD_NEW(state_offset, count); }
}
}
break;
/*-----------------------------------------------------------------*/
case OP_EXTUNI_EXTRA + OP_TYPEEXACT:
case OP_EXTUNI_EXTRA + OP_TYPEUPTO:
case OP_EXTUNI_EXTRA + OP_TYPEMINUPTO:
case OP_EXTUNI_EXTRA + OP_TYPEPOSUPTO:
if (codevalue != OP_EXTUNI_EXTRA + OP_TYPEEXACT)
{ ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0); }
count = current_state->count; /* Number already matched */
if (clen > 0)
{
PCRE2_SPTR nptr;
int ncount = 0;
if (codevalue == OP_EXTUNI_EXTRA + OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
nptr = PRIV(extuni)(c, ptr + clen, mb->start_subject, end_subject, utf,
&ncount);
if (nptr >= end_subject && (mb->moptions & PCRE2_PARTIAL_HARD) != 0)
reset_could_continue = TRUE;
if (++count >= (int)GET2(code, 1))
{ ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, ncount); }
else
{ ADD_NEW_DATA(-state_offset, count, ncount); }
}
break;
#endif
/*-----------------------------------------------------------------*/
case OP_ANYNL_EXTRA + OP_TYPEEXACT:
case OP_ANYNL_EXTRA + OP_TYPEUPTO:
case OP_ANYNL_EXTRA + OP_TYPEMINUPTO:
case OP_ANYNL_EXTRA + OP_TYPEPOSUPTO:
if (codevalue != OP_ANYNL_EXTRA + OP_TYPEEXACT)
{ ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0); }
count = current_state->count; /* Number already matched */
if (clen > 0)
{
int ncount = 0;
switch (c)
{
case CHAR_VT:
case CHAR_FF:
case CHAR_NEL:
#ifndef EBCDIC
case 0x2028:
case 0x2029:
#endif /* Not EBCDIC */
if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) break;
goto ANYNL03;
case CHAR_CR:
if (ptr + 1 < end_subject && UCHAR21TEST(ptr + 1) == CHAR_LF) ncount = 1;
/* Fall through */
ANYNL03:
case CHAR_LF:
if (codevalue == OP_ANYNL_EXTRA + OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, ncount); }
else
{ ADD_NEW_DATA(-state_offset, count, ncount); }
break;
default:
break;
}
}
break;
/*-----------------------------------------------------------------*/
case OP_VSPACE_EXTRA + OP_TYPEEXACT:
case OP_VSPACE_EXTRA + OP_TYPEUPTO:
case OP_VSPACE_EXTRA + OP_TYPEMINUPTO:
case OP_VSPACE_EXTRA + OP_TYPEPOSUPTO:
if (codevalue != OP_VSPACE_EXTRA + OP_TYPEEXACT)
{ ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0); }
count = current_state->count; /* Number already matched */
if (clen > 0)
{
BOOL OK;
switch (c)
{
VSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
}
if (OK == (d == OP_VSPACE))
{
if (codevalue == OP_VSPACE_EXTRA + OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, 0); }
else
{ ADD_NEW_DATA(-state_offset, count, 0); }
}
}
break;
/*-----------------------------------------------------------------*/
case OP_HSPACE_EXTRA + OP_TYPEEXACT:
case OP_HSPACE_EXTRA + OP_TYPEUPTO:
case OP_HSPACE_EXTRA + OP_TYPEMINUPTO:
case OP_HSPACE_EXTRA + OP_TYPEPOSUPTO:
if (codevalue != OP_HSPACE_EXTRA + OP_TYPEEXACT)
{ ADD_ACTIVE(state_offset + 2 + IMM2_SIZE, 0); }
count = current_state->count; /* Number already matched */
if (clen > 0)
{
BOOL OK;
switch (c)
{
HSPACE_CASES:
OK = TRUE;
break;
default:
OK = FALSE;
break;
}
if (OK == (d == OP_HSPACE))
{
if (codevalue == OP_HSPACE_EXTRA + OP_TYPEPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW_DATA(-(state_offset + 2 + IMM2_SIZE), 0, 0); }
else
{ ADD_NEW_DATA(-state_offset, count, 0); }
}
}
break;
/* ========================================================================== */
/* These opcodes are followed by a character that is usually compared
to the current subject character; it is loaded into d. We still get
here even if there is no subject character, because in some cases zero
repetitions are permitted. */
/*-----------------------------------------------------------------*/
case OP_CHAR:
if (clen > 0 && c == d) { ADD_NEW(state_offset + dlen + 1, 0); }
break;
/*-----------------------------------------------------------------*/
case OP_CHARI:
if (clen == 0) break;
#ifdef SUPPORT_UNICODE
if (utf_or_ucp)
{
if (c == d) { ADD_NEW(state_offset + dlen + 1, 0); } else
{
unsigned int othercase;
if (c < 128)
othercase = fcc[c];
else
othercase = UCD_OTHERCASE(c);
if (d == othercase) { ADD_NEW(state_offset + dlen + 1, 0); }
}
}
else
#endif /* SUPPORT_UNICODE */
/* Not UTF or UCP mode */
{
if (TABLE_GET(c, lcc, c) == TABLE_GET(d, lcc, d))
{ ADD_NEW(state_offset + 2, 0); }
}
break;
#ifdef SUPPORT_UNICODE
/*-----------------------------------------------------------------*/
/* This is a tricky one because it can match more than one character.
Find out how many characters to skip, and then set up a negative state
to wait for them to pass before continuing. */
case OP_EXTUNI:
if (clen > 0)
{
int ncount = 0;
PCRE2_SPTR nptr = PRIV(extuni)(c, ptr + clen, mb->start_subject,
end_subject, utf, &ncount);
if (nptr >= end_subject && (mb->moptions & PCRE2_PARTIAL_HARD) != 0)
reset_could_continue = TRUE;
ADD_NEW_DATA(-(state_offset + 1), 0, ncount);
}
break;
#endif
/*-----------------------------------------------------------------*/
/* This is a tricky like EXTUNI because it too can match more than one
character (when CR is followed by LF). In this case, set up a negative
state to wait for one character to pass before continuing. */
case OP_ANYNL:
if (clen > 0) switch(c)
{
case CHAR_VT:
case CHAR_FF:
case CHAR_NEL:
#ifndef EBCDIC
case 0x2028:
case 0x2029:
#endif /* Not EBCDIC */
if (mb->bsr_convention == PCRE2_BSR_ANYCRLF) break;
/* Fall through */
case CHAR_LF:
ADD_NEW(state_offset + 1, 0);
break;
case CHAR_CR:
if (ptr + 1 >= end_subject)
{
ADD_NEW(state_offset + 1, 0);
if ((mb->moptions & PCRE2_PARTIAL_HARD) != 0)
reset_could_continue = TRUE;
}
else if (UCHAR21TEST(ptr + 1) == CHAR_LF)
{
ADD_NEW_DATA(-(state_offset + 1), 0, 1);
}
else
{
ADD_NEW(state_offset + 1, 0);
}
break;
}
break;
/*-----------------------------------------------------------------*/
case OP_NOT_VSPACE:
if (clen > 0) switch(c)
{
VSPACE_CASES:
break;
default:
ADD_NEW(state_offset + 1, 0);
break;
}
break;
/*-----------------------------------------------------------------*/
case OP_VSPACE:
if (clen > 0) switch(c)
{
VSPACE_CASES:
ADD_NEW(state_offset + 1, 0);
break;
default:
break;
}
break;
/*-----------------------------------------------------------------*/
case OP_NOT_HSPACE:
if (clen > 0) switch(c)
{
HSPACE_CASES:
break;
default:
ADD_NEW(state_offset + 1, 0);
break;
}
break;
/*-----------------------------------------------------------------*/
case OP_HSPACE:
if (clen > 0) switch(c)
{
HSPACE_CASES:
ADD_NEW(state_offset + 1, 0);
break;
default:
break;
}
break;
/*-----------------------------------------------------------------*/
/* Match a negated single character casefully. */
case OP_NOT:
if (clen > 0 && c != d) { ADD_NEW(state_offset + dlen + 1, 0); }
break;
/*-----------------------------------------------------------------*/
/* Match a negated single character caselessly. */
case OP_NOTI:
if (clen > 0)
{
uint32_t otherd;
#ifdef SUPPORT_UNICODE
if (utf_or_ucp && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
if (c != d && c != otherd)
{ ADD_NEW(state_offset + dlen + 1, 0); }
}
break;
/*-----------------------------------------------------------------*/
case OP_PLUSI:
case OP_MINPLUSI:
case OP_POSPLUSI:
case OP_NOTPLUSI:
case OP_NOTMINPLUSI:
case OP_NOTPOSPLUSI:
caseless = TRUE;
codevalue -= OP_STARI - OP_STAR;
/* Fall through */
case OP_PLUS:
case OP_MINPLUS:
case OP_POSPLUS:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(state_offset + dlen + 1, 0); }
if (clen > 0)
{
uint32_t otherd = NOTACHAR;
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf_or_ucp && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
}
if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
{
if (count > 0 &&
(codevalue == OP_POSPLUS || codevalue == OP_NOTPOSPLUS))
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW(state_offset, count);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_QUERYI:
case OP_MINQUERYI:
case OP_POSQUERYI:
case OP_NOTQUERYI:
case OP_NOTMINQUERYI:
case OP_NOTPOSQUERYI:
caseless = TRUE;
codevalue -= OP_STARI - OP_STAR;
/* Fall through */
case OP_QUERY:
case OP_MINQUERY:
case OP_POSQUERY:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
case OP_NOTPOSQUERY:
ADD_ACTIVE(state_offset + dlen + 1, 0);
if (clen > 0)
{
uint32_t otherd = NOTACHAR;
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf_or_ucp && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
}
if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
{
if (codevalue == OP_POSQUERY || codevalue == OP_NOTPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset + dlen + 1, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_STARI:
case OP_MINSTARI:
case OP_POSSTARI:
case OP_NOTSTARI:
case OP_NOTMINSTARI:
case OP_NOTPOSSTARI:
caseless = TRUE;
codevalue -= OP_STARI - OP_STAR;
/* Fall through */
case OP_STAR:
case OP_MINSTAR:
case OP_POSSTAR:
case OP_NOTSTAR:
case OP_NOTMINSTAR:
case OP_NOTPOSSTAR:
ADD_ACTIVE(state_offset + dlen + 1, 0);
if (clen > 0)
{
uint32_t otherd = NOTACHAR;
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf_or_ucp && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
}
if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
{
if (codevalue == OP_POSSTAR || codevalue == OP_NOTPOSSTAR)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset, 0);
}
}
break;
/*-----------------------------------------------------------------*/
case OP_EXACTI:
case OP_NOTEXACTI:
caseless = TRUE;
codevalue -= OP_STARI - OP_STAR;
/* Fall through */
case OP_EXACT:
case OP_NOTEXACT:
count = current_state->count; /* Number already matched */
if (clen > 0)
{
uint32_t otherd = NOTACHAR;
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf_or_ucp && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
}
if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
{
if (++count >= (int)GET2(code, 1))
{ ADD_NEW(state_offset + dlen + 1 + IMM2_SIZE, 0); }
else
{ ADD_NEW(state_offset, count); }
}
}
break;
/*-----------------------------------------------------------------*/
case OP_UPTOI:
case OP_MINUPTOI:
case OP_POSUPTOI:
case OP_NOTUPTOI:
case OP_NOTMINUPTOI:
case OP_NOTPOSUPTOI:
caseless = TRUE;
codevalue -= OP_STARI - OP_STAR;
/* Fall through */
case OP_UPTO:
case OP_MINUPTO:
case OP_POSUPTO:
case OP_NOTUPTO:
case OP_NOTMINUPTO:
case OP_NOTPOSUPTO:
ADD_ACTIVE(state_offset + dlen + 1 + IMM2_SIZE, 0);
count = current_state->count; /* Number already matched */
if (clen > 0)
{
uint32_t otherd = NOTACHAR;
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf_or_ucp && d >= 128)
otherd = UCD_OTHERCASE(d);
else
#endif /* SUPPORT_UNICODE */
otherd = TABLE_GET(d, fcc, d);
}
if ((c == d || c == otherd) == (codevalue < OP_NOTSTAR))
{
if (codevalue == OP_POSUPTO || codevalue == OP_NOTPOSUPTO)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= (int)GET2(code, 1))
{ ADD_NEW(state_offset + dlen + 1 + IMM2_SIZE, 0); }
else
{ ADD_NEW(state_offset, count); }
}
}
break;
/* ========================================================================== */
/* These are the class-handling opcodes */
case OP_CLASS:
case OP_NCLASS:
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
case OP_ECLASS:
#endif
{
BOOL isinclass = FALSE;
int next_state_offset;
PCRE2_SPTR ecode;
#ifdef SUPPORT_WIDE_CHARS
/* An extended class may have a table or a list of single characters,
ranges, or both, and it may be positive or negative. There's a
function that sorts all this out. */
if (codevalue == OP_XCLASS)
{
ecode = code + GET(code, 1);
if (clen > 0)
isinclass = PRIV(xclass)(c, code + 1 + LINK_SIZE,
(const uint8_t*)mb->start_code, utf);
}
/* A nested set-based class has internal opcodes for performing
set operations. */
else if (codevalue == OP_ECLASS)
{
ecode = code + GET(code, 1);
if (clen > 0)
isinclass = PRIV(eclass)(c, code + 1 + LINK_SIZE, ecode,
(const uint8_t*)mb->start_code, utf);
}
else
#endif /* SUPPORT_WIDE_CHARS */
/* For a simple class, there is always just a 32-byte table, and we
can set isinclass from it. */
{
ecode = code + 1 + (32 / sizeof(PCRE2_UCHAR));
if (clen > 0)
{
isinclass = (c > 255)? (codevalue == OP_NCLASS) :
((((const uint8_t *)(code + 1))[c/8] & (1u << (c&7))) != 0);
}
}
/* At this point, isinclass is set for all kinds of class, and ecode
points to the byte after the end of the class. If there is a
quantifier, this is where it will be. */
next_state_offset = (int)(ecode - start_code);
switch (*ecode)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRPOSSTAR:
ADD_ACTIVE(next_state_offset + 1, 0);
if (isinclass)
{
if (*ecode == OP_CRPOSSTAR)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(state_offset, 0);
}
break;
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRPOSPLUS:
count = current_state->count; /* Already matched */
if (count > 0) { ADD_ACTIVE(next_state_offset + 1, 0); }
if (isinclass)
{
if (count > 0 && *ecode == OP_CRPOSPLUS)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
count++;
ADD_NEW(state_offset, count);
}
break;
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSQUERY:
ADD_ACTIVE(next_state_offset + 1, 0);
if (isinclass)
{
if (*ecode == OP_CRPOSQUERY)
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
ADD_NEW(next_state_offset + 1, 0);
}
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
count = current_state->count; /* Already matched */
if (count >= (int)GET2(ecode, 1))
{ ADD_ACTIVE(next_state_offset + 1 + 2 * IMM2_SIZE, 0); }
if (isinclass)
{
int max = (int)GET2(ecode, 1 + IMM2_SIZE);
if (*ecode == OP_CRPOSRANGE && count >= (int)GET2(ecode, 1))
{
active_count--; /* Remove non-match possibility */
next_active_state--;
}
if (++count >= max && max != 0) /* Max 0 => no limit */
{ ADD_NEW(next_state_offset + 1 + 2 * IMM2_SIZE, 0); }
else
{ ADD_NEW(state_offset, count); }
}
break;
default:
if (isinclass) { ADD_NEW(next_state_offset, 0); }
break;
}
}
break;
/* ========================================================================== */
/* These are the opcodes for fancy brackets of various kinds. We have
to use recursion in order to handle them. The "always failing" assertion
(?!) is optimised to OP_FAIL when compiling, so we have to support that,
though the other "backtracking verbs" are not supported. */
case OP_FAIL:
break;
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
{
int rc;
int *local_workspace;
PCRE2_SIZE *local_offsets;
PCRE2_SPTR endasscode = code + GET(code, 1);
RWS_anchor *rws = (RWS_anchor *)RWS;
if (rws->free < RWS_RSIZE + RWS_OVEC_OSIZE)
{
rc = more_workspace(&rws, RWS_OVEC_OSIZE, mb);
if (rc != 0) return rc;
RWS = (int *)rws;
}
local_offsets = (PCRE2_SIZE *)(RWS + rws->size - rws->free);
local_workspace = ((int *)local_offsets) + RWS_OVEC_OSIZE;
rws->free -= RWS_RSIZE + RWS_OVEC_OSIZE;
while (*endasscode == OP_ALT) endasscode += GET(endasscode, 1);
rc = internal_dfa_match(
mb, /* static match data */
code, /* this subexpression's code */
ptr, /* where we currently are */
(PCRE2_SIZE)(ptr - start_subject), /* start offset */
local_offsets, /* offset vector */
RWS_OVEC_OSIZE/OVEC_UNIT, /* size of same */
local_workspace, /* workspace vector */
RWS_RSIZE, /* size of same */
rlevel, /* function recursion level */
RWS); /* recursion workspace */
rws->free += RWS_RSIZE + RWS_OVEC_OSIZE;
if (rc < 0 && rc != PCRE2_ERROR_NOMATCH) return rc;
if ((rc >= 0) == (codevalue == OP_ASSERT || codevalue == OP_ASSERTBACK))
{ ADD_ACTIVE((int)(endasscode + LINK_SIZE + 1 - start_code), 0); }
}
break;
/*-----------------------------------------------------------------*/
case OP_COND:
case OP_SCOND:
{
int codelink = (int)GET(code, 1);
PCRE2_UCHAR condcode;
/* Because of the way auto-callout works during compile, a callout item
is inserted between OP_COND and an assertion condition. This does not
happen for the other conditions. */
if (code[LINK_SIZE + 1] == OP_CALLOUT
|| code[LINK_SIZE + 1] == OP_CALLOUT_STR)
{
PCRE2_SIZE callout_length;
rrc = do_callout_dfa(code, offsets, current_subject, ptr, mb,
1 + LINK_SIZE, &callout_length);
if (rrc < 0) return rrc; /* Abandon */
if (rrc > 0) break; /* Fail this thread */
code += callout_length; /* Skip callout data */
}
condcode = code[LINK_SIZE+1];
/* Back reference conditions and duplicate named recursion conditions
are not supported */
if (condcode == OP_CREF || condcode == OP_DNCREF ||
condcode == OP_DNRREF)
return PCRE2_ERROR_DFA_UCOND;
/* The DEFINE condition is always false, and the assertion (?!) is
converted to OP_FAIL. */
if (condcode == OP_FALSE || condcode == OP_FAIL)
{ ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
/* There is also an always-true condition */
else if (condcode == OP_TRUE)
{ ADD_ACTIVE(state_offset + LINK_SIZE + 2, 0); }
/* The only supported version of OP_RREF is for the value RREF_ANY,
which means "test if in any recursion". We can't test for specifically
recursed groups. */
else if (condcode == OP_RREF)
{
unsigned int value = GET2(code, LINK_SIZE + 2);
if (value != RREF_ANY) return PCRE2_ERROR_DFA_UCOND;
if (mb->recursive != NULL)
{ ADD_ACTIVE(state_offset + LINK_SIZE + 2 + IMM2_SIZE, 0); }
else { ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
}
/* Otherwise, the condition is an assertion */
else
{
int rc;
int *local_workspace;
PCRE2_SIZE *local_offsets;
PCRE2_SPTR asscode = code + LINK_SIZE + 1;
PCRE2_SPTR endasscode = asscode + GET(asscode, 1);
RWS_anchor *rws = (RWS_anchor *)RWS;
if (rws->free < RWS_RSIZE + RWS_OVEC_OSIZE)
{
rc = more_workspace(&rws, RWS_OVEC_OSIZE, mb);
if (rc != 0) return rc;
RWS = (int *)rws;
}
local_offsets = (PCRE2_SIZE *)(RWS + rws->size - rws->free);
local_workspace = ((int *)local_offsets) + RWS_OVEC_OSIZE;
rws->free -= RWS_RSIZE + RWS_OVEC_OSIZE;
while (*endasscode == OP_ALT) endasscode += GET(endasscode, 1);
rc = internal_dfa_match(
mb, /* fixed match data */
asscode, /* this subexpression's code */
ptr, /* where we currently are */
(PCRE2_SIZE)(ptr - start_subject), /* start offset */
local_offsets, /* offset vector */
RWS_OVEC_OSIZE/OVEC_UNIT, /* size of same */
local_workspace, /* workspace vector */
RWS_RSIZE, /* size of same */
rlevel, /* function recursion level */
RWS); /* recursion workspace */
rws->free += RWS_RSIZE + RWS_OVEC_OSIZE;
if (rc < 0 && rc != PCRE2_ERROR_NOMATCH) return rc;
if ((rc >= 0) ==
(condcode == OP_ASSERT || condcode == OP_ASSERTBACK))
{ ADD_ACTIVE((int)(endasscode + LINK_SIZE + 1 - start_code), 0); }
else
{ ADD_ACTIVE(state_offset + codelink + LINK_SIZE + 1, 0); }
}
}
break;
/*-----------------------------------------------------------------*/
case OP_RECURSE:
{
int rc;
int *local_workspace;
PCRE2_SIZE *local_offsets;
RWS_anchor *rws = (RWS_anchor *)RWS;
PCRE2_SPTR callpat = start_code + GET(code, 1);
uint32_t recno = (callpat == mb->start_code)? 0 :
GET2(callpat, 1 + LINK_SIZE);
if (rws->free < RWS_RSIZE + RWS_OVEC_RSIZE)
{
rc = more_workspace(&rws, RWS_OVEC_RSIZE, mb);
if (rc != 0) return rc;
RWS = (int *)rws;
}
local_offsets = (PCRE2_SIZE *)(RWS + rws->size - rws->free);
local_workspace = ((int *)local_offsets) + RWS_OVEC_RSIZE;
rws->free -= RWS_RSIZE + RWS_OVEC_RSIZE;
/* Check for repeating a recursion without advancing the subject
pointer or last used character. This should catch convoluted mutual
recursions. (Some simple cases are caught at compile time.) */
for (dfa_recursion_info *ri = mb->recursive;
ri != NULL;
ri = ri->prevrec)
{
if (recno == ri->group_num && ptr == ri->subject_position &&
mb->last_used_ptr == ri->last_used_ptr)
return PCRE2_ERROR_RECURSELOOP;
}
/* Remember this recursion and where we started it so as to
catch infinite loops. */
new_recursive.group_num = recno;
new_recursive.subject_position = ptr;
new_recursive.last_used_ptr = mb->last_used_ptr;
new_recursive.prevrec = mb->recursive;
mb->recursive = &new_recursive;
rc = internal_dfa_match(
mb, /* fixed match data */
callpat, /* this subexpression's code */
ptr, /* where we currently are */
(PCRE2_SIZE)(ptr - start_subject), /* start offset */
local_offsets, /* offset vector */
RWS_OVEC_RSIZE/OVEC_UNIT, /* size of same */
local_workspace, /* workspace vector */
RWS_RSIZE, /* size of same */
rlevel, /* function recursion level */
RWS); /* recursion workspace */
rws->free += RWS_RSIZE + RWS_OVEC_RSIZE;
mb->recursive = new_recursive.prevrec; /* Done this recursion */
/* Ran out of internal offsets */
if (rc == 0) return PCRE2_ERROR_DFA_RECURSE;
/* For each successful matched substring, set up the next state with a
count of characters to skip before trying it. Note that the count is in
characters, not bytes. */
if (rc > 0)
{
for (rc = rc*2 - 2; rc >= 0; rc -= 2)
{
PCRE2_SIZE charcount = local_offsets[rc+1] - local_offsets[rc];
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (utf)
{
PCRE2_SPTR p = start_subject + local_offsets[rc];
PCRE2_SPTR pp = start_subject + local_offsets[rc+1];
while (p < pp) if (NOT_FIRSTCU(*p++)) charcount--;
}
#endif
if (charcount > 0)
{
ADD_NEW_DATA(-(state_offset + LINK_SIZE + 1), 0,
(int)(charcount - 1));
}
else
{
ADD_ACTIVE(state_offset + LINK_SIZE + 1, 0);
}
}
}
else if (rc != PCRE2_ERROR_NOMATCH) return rc;
}
break;
/*-----------------------------------------------------------------*/
case OP_BRAPOS:
case OP_SBRAPOS:
case OP_CBRAPOS:
case OP_SCBRAPOS:
case OP_BRAPOSZERO:
{
int rc;
int *local_workspace;
PCRE2_SIZE *local_offsets;
PCRE2_SIZE charcount, matched_count;
PCRE2_SPTR local_ptr = ptr;
RWS_anchor *rws = (RWS_anchor *)RWS;
BOOL allow_zero;
if (rws->free < RWS_RSIZE + RWS_OVEC_OSIZE)
{
rc = more_workspace(&rws, RWS_OVEC_OSIZE, mb);
if (rc != 0) return rc;
RWS = (int *)rws;
}
local_offsets = (PCRE2_SIZE *)(RWS + rws->size - rws->free);
local_workspace = ((int *)local_offsets) + RWS_OVEC_OSIZE;
rws->free -= RWS_RSIZE + RWS_OVEC_OSIZE;
if (codevalue == OP_BRAPOSZERO)
{
allow_zero = TRUE;
++code; /* The following opcode will be one of the above BRAs */
}
else allow_zero = FALSE;
/* Loop to match the subpattern as many times as possible as if it were
a complete pattern. */
for (matched_count = 0;; matched_count++)
{
rc = internal_dfa_match(
mb, /* fixed match data */
code, /* this subexpression's code */
local_ptr, /* where we currently are */
(PCRE2_SIZE)(ptr - start_subject), /* start offset */
local_offsets, /* offset vector */
RWS_OVEC_OSIZE/OVEC_UNIT, /* size of same */
local_workspace, /* workspace vector */
RWS_RSIZE, /* size of same */
rlevel, /* function recursion level */
RWS); /* recursion workspace */
/* Failed to match */
if (rc < 0)
{
if (rc != PCRE2_ERROR_NOMATCH) return rc;
break;
}
/* Matched: break the loop if zero characters matched. */
charcount = local_offsets[1] - local_offsets[0];
if (charcount == 0) break;
local_ptr += charcount; /* Advance temporary position ptr */
}
rws->free += RWS_RSIZE + RWS_OVEC_OSIZE;
/* At this point we have matched the subpattern matched_count
times, and local_ptr is pointing to the character after the end of the
last match. */
if (matched_count > 0 || allow_zero)
{
PCRE2_SPTR end_subpattern = code;
int next_state_offset;
do { end_subpattern += GET(end_subpattern, 1); }
while (*end_subpattern == OP_ALT);
next_state_offset =
(int)(end_subpattern - start_code + LINK_SIZE + 1);
/* Optimization: if there are no more active states, and there
are no new states yet set up, then skip over the subject string
right here, to save looping. Otherwise, set up the new state to swing
into action when the end of the matched substring is reached. */
if (i + 1 >= active_count && new_count == 0)
{
ptr = local_ptr;
clen = 0;
ADD_NEW(next_state_offset, 0);
}
else
{
PCRE2_SPTR p = ptr;
PCRE2_SPTR pp = local_ptr;
charcount = (PCRE2_SIZE)(pp - p);
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (utf) while (p < pp) if (NOT_FIRSTCU(*p++)) charcount--;
#endif
ADD_NEW_DATA(-next_state_offset, 0, (int)(charcount - 1));
}
}
}
break;
/*-----------------------------------------------------------------*/
case OP_ONCE:
{
int rc;
int *local_workspace;
PCRE2_SIZE *local_offsets;
RWS_anchor *rws = (RWS_anchor *)RWS;
if (rws->free < RWS_RSIZE + RWS_OVEC_OSIZE)
{
rc = more_workspace(&rws, RWS_OVEC_OSIZE, mb);
if (rc != 0) return rc;
RWS = (int *)rws;
}
local_offsets = (PCRE2_SIZE *)(RWS + rws->size - rws->free);
local_workspace = ((int *)local_offsets) + RWS_OVEC_OSIZE;
rws->free -= RWS_RSIZE + RWS_OVEC_OSIZE;
rc = internal_dfa_match(
mb, /* fixed match data */
code, /* this subexpression's code */
ptr, /* where we currently are */
(PCRE2_SIZE)(ptr - start_subject), /* start offset */
local_offsets, /* offset vector */
RWS_OVEC_OSIZE/OVEC_UNIT, /* size of same */
local_workspace, /* workspace vector */
RWS_RSIZE, /* size of same */
rlevel, /* function recursion level */
RWS); /* recursion workspace */
rws->free += RWS_RSIZE + RWS_OVEC_OSIZE;
if (rc >= 0)
{
PCRE2_SPTR end_subpattern = code;
PCRE2_SIZE charcount = local_offsets[1] - local_offsets[0];
int next_state_offset, repeat_state_offset;
do { end_subpattern += GET(end_subpattern, 1); }
while (*end_subpattern == OP_ALT);
next_state_offset =
(int)(end_subpattern - start_code + LINK_SIZE + 1);
/* If the end of this subpattern is KETRMAX or KETRMIN, we must
arrange for the repeat state also to be added to the relevant list.
Calculate the offset, or set -1 for no repeat. */
repeat_state_offset = (*end_subpattern == OP_KETRMAX ||
*end_subpattern == OP_KETRMIN)?
(int)(end_subpattern - start_code - GET(end_subpattern, 1)) : -1;
/* If we have matched an empty string, add the next state at the
current character pointer. This is important so that the duplicate
checking kicks in, which is what breaks infinite loops that match an
empty string. */
if (charcount == 0)
{
ADD_ACTIVE(next_state_offset, 0);
}
/* Optimization: if there are no more active states, and there
are no new states yet set up, then skip over the subject string
right here, to save looping. Otherwise, set up the new state to swing
into action when the end of the matched substring is reached. */
else if (i + 1 >= active_count && new_count == 0)
{
ptr += charcount;
clen = 0;
ADD_NEW(next_state_offset, 0);
/* If we are adding a repeat state at the new character position,
we must fudge things so that it is the only current state.
Otherwise, it might be a duplicate of one we processed before, and
that would cause it to be skipped. */
if (repeat_state_offset >= 0)
{
next_active_state = active_states;
active_count = 0;
i = -1;
ADD_ACTIVE(repeat_state_offset, 0);
}
}
else
{
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (utf)
{
PCRE2_SPTR p = start_subject + local_offsets[0];
PCRE2_SPTR pp = start_subject + local_offsets[1];
while (p < pp) if (NOT_FIRSTCU(*p++)) charcount--;
}
#endif
ADD_NEW_DATA(-next_state_offset, 0, (int)(charcount - 1));
if (repeat_state_offset >= 0)
{ ADD_NEW_DATA(-repeat_state_offset, 0, (int)(charcount - 1)); }
}
}
else if (rc != PCRE2_ERROR_NOMATCH) return rc;
}
break;
/* ========================================================================== */
/* Handle callouts */
case OP_CALLOUT:
case OP_CALLOUT_STR:
{
PCRE2_SIZE callout_length;
rrc = do_callout_dfa(code, offsets, current_subject, ptr, mb, 0,
&callout_length);
if (rrc < 0) return rrc; /* Abandon */
if (rrc == 0)
{ ADD_ACTIVE(state_offset + (int)callout_length, 0); }
}
break;
/* ========================================================================== */
default: /* Unsupported opcode */
return PCRE2_ERROR_DFA_UITEM;
}
NEXT_ACTIVE_STATE: continue;
} /* End of loop scanning active states */
/* We have finished the processing at the current subject character. If no
new states have been set for the next character, we have found all the
matches that we are going to find. If partial matching has been requested,
check for appropriate conditions.
The "could_continue" variable is true if a state could have continued but
for the fact that the end of the subject was reached. */
if (new_count <= 0)
{
if (could_continue && /* Some could go on, and */
( /* either... */
(mb->moptions & PCRE2_PARTIAL_HARD) != 0 /* Hard partial */
|| /* or... */
((mb->moptions & PCRE2_PARTIAL_SOFT) != 0 && /* Soft partial and */
match_count < 0) /* no matches */
) && /* And... */
(
partial_newline || /* Either partial NL */
( /* or ... */
ptr >= end_subject && /* End of subject and */
( /* either */
ptr > mb->start_used_ptr || /* Inspected non-empty string */
mb->allowemptypartial /* or pattern has lookbehind */
) /* or could match empty */
)
))
match_count = PCRE2_ERROR_PARTIAL;
break; /* Exit from loop along the subject string */
}
/* One or more states are active for the next character. */
ptr += clen; /* Advance to next subject character */
} /* Loop to move along the subject string */
/* Control gets here from "break" a few lines above. If we have a match and
PCRE2_ENDANCHORED is set, the match fails. */
if (match_count >= 0 &&
((mb->moptions | mb->poptions) & PCRE2_ENDANCHORED) != 0 &&
ptr < end_subject)
match_count = PCRE2_ERROR_NOMATCH;
return match_count;
}
/*************************************************
* Match a pattern using the DFA algorithm *
*************************************************/
/* This function matches a compiled pattern to a subject string, using the
alternate matching algorithm that finds all matches at once.
Arguments:
code points to the compiled pattern
subject subject string
length length of subject string
startoffset where to start matching in the subject
options option bits
match_data points to a match data structure
gcontext points to a match context
workspace pointer to workspace
wscount size of workspace
Returns: > 0 => number of match offset pairs placed in offsets
= 0 => offsets overflowed; longest matches are present
-1 => failed to match
< -1 => some kind of unexpected problem
*/
PCRE2_EXP_DEFN int PCRE2_CALL_CONVENTION
pcre2_dfa_match(const pcre2_code *code, PCRE2_SPTR subject, PCRE2_SIZE length,
PCRE2_SIZE start_offset, uint32_t options, pcre2_match_data *match_data,
pcre2_match_context *mcontext, int *workspace, PCRE2_SIZE wscount)
{
int rc;
int was_zero_terminated = 0;
const pcre2_real_code *re = (const pcre2_real_code *)code;
PCRE2_SPTR start_match;
PCRE2_SPTR end_subject;
PCRE2_SPTR bumpalong_limit;
PCRE2_SPTR req_cu_ptr;
BOOL utf, anchored, startline, firstline;
BOOL has_first_cu = FALSE;
BOOL has_req_cu = FALSE;
#if PCRE2_CODE_UNIT_WIDTH == 8
PCRE2_SPTR memchr_found_first_cu = NULL;
PCRE2_SPTR memchr_found_first_cu2 = NULL;
#endif
PCRE2_UCHAR first_cu = 0;
PCRE2_UCHAR first_cu2 = 0;
PCRE2_UCHAR req_cu = 0;
PCRE2_UCHAR req_cu2 = 0;
const uint8_t *start_bits = NULL;
/* We need to have mb pointing to a match block, because the IS_NEWLINE macro
is used below, and it expects NLBLOCK to be defined as a pointer. */
pcre2_callout_block cb;
dfa_match_block actual_match_block;
dfa_match_block *mb = &actual_match_block;
/* Set up a starting block of memory for use during recursive calls to
internal_dfa_match(). By putting this on the stack, it minimizes resource use
in the case when it is not needed. If this is too small, more memory is
obtained from the heap. At the start of each block is an anchor structure.*/
int base_recursion_workspace[RWS_BASE_SIZE];
RWS_anchor *rws = (RWS_anchor *)base_recursion_workspace;
rws->next = NULL;
rws->size = RWS_BASE_SIZE;
rws->free = RWS_BASE_SIZE - RWS_ANCHOR_SIZE;
/* Recognize NULL, length 0 as an empty string. */
if (subject == NULL && length == 0) subject = (PCRE2_SPTR)"";
/* Plausibility checks */
if ((options & ~PUBLIC_DFA_MATCH_OPTIONS) != 0) return PCRE2_ERROR_BADOPTION;
if (re == NULL || subject == NULL || workspace == NULL || match_data == NULL)
return PCRE2_ERROR_NULL;
if (length == PCRE2_ZERO_TERMINATED)
{
length = PRIV(strlen)(subject);
was_zero_terminated = 1;
}
if (wscount < 20) return PCRE2_ERROR_DFA_WSSIZE;
if (start_offset > length) return PCRE2_ERROR_BADOFFSET;
/* Partial matching and PCRE2_ENDANCHORED are currently not allowed at the same
time. */
if ((options & (PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT)) != 0 &&
((re->overall_options | options) & PCRE2_ENDANCHORED) != 0)
return PCRE2_ERROR_BADOPTION;
/* Invalid UTF support is not available for DFA matching. */
if ((re->overall_options & PCRE2_MATCH_INVALID_UTF) != 0)
return PCRE2_ERROR_DFA_UINVALID_UTF;
/* Check that the first field in the block is the magic number. If it is not,
return with PCRE2_ERROR_BADMAGIC. */
if (re->magic_number != MAGIC_NUMBER) return PCRE2_ERROR_BADMAGIC;
/* Check the code unit width. */
if ((re->flags & PCRE2_MODE_MASK) != PCRE2_CODE_UNIT_WIDTH/8)
return PCRE2_ERROR_BADMODE;
/* PCRE2_NOTEMPTY and PCRE2_NOTEMPTY_ATSTART are match-time flags in the
options variable for this function. Users of PCRE2 who are not calling the
function directly would like to have a way of setting these flags, in the same
way that they can set pcre2_compile() flags like PCRE2_NO_AUTO_POSSESS with
constructions like (*NO_AUTOPOSSESS). To enable this, (*NOTEMPTY) and
(*NOTEMPTY_ATSTART) set bits in the pattern's "flag" function which can now be
transferred to the options for this function. The bits are guaranteed to be
adjacent, but do not have the same values. This bit of Boolean trickery assumes
that the match-time bits are not more significant than the flag bits. If by
accident this is not the case, a compile-time division by zero error will
occur. */
#define FF (PCRE2_NOTEMPTY_SET|PCRE2_NE_ATST_SET)
#define OO (PCRE2_NOTEMPTY|PCRE2_NOTEMPTY_ATSTART)
options |= (re->flags & FF) / ((FF & (~FF+1)) / (OO & (~OO+1)));
#undef FF
#undef OO
/* If restarting after a partial match, do some sanity checks on the contents
of the workspace. */
if ((options & PCRE2_DFA_RESTART) != 0)
{
if ((workspace[0] & (-2)) != 0 || workspace[1] < 1 ||
workspace[1] > (int)((wscount - 2)/INTS_PER_STATEBLOCK))
return PCRE2_ERROR_DFA_BADRESTART;
}
/* Set some local values */
utf = (re->overall_options & PCRE2_UTF) != 0;
start_match = subject + start_offset;
end_subject = subject + length;
req_cu_ptr = start_match - 1;
anchored = (options & (PCRE2_ANCHORED|PCRE2_DFA_RESTART)) != 0 ||
(re->overall_options & PCRE2_ANCHORED) != 0;
/* The "must be at the start of a line" flags are used in a loop when finding
where to start. */
startline = (re->flags & PCRE2_STARTLINE) != 0;
firstline = !anchored && (re->overall_options & PCRE2_FIRSTLINE) != 0;
bumpalong_limit = end_subject;
/* Initialize and set up the fixed fields in the callout block, with a pointer
in the match block. */
mb->cb = &cb;
cb.version = 2;
cb.subject = subject;
cb.subject_length = (PCRE2_SIZE)(end_subject - subject);
cb.callout_flags = 0;
cb.capture_top = 1; /* No capture support */
cb.capture_last = 0;
cb.mark = NULL; /* No (*MARK) support */
/* Get data from the match context, if present, and fill in the remaining
fields in the match block. It is an error to set an offset limit without
setting the flag at compile time. */
if (mcontext == NULL)
{
mb->callout = NULL;
mb->memctl = re->memctl;
mb->match_limit = PRIV(default_match_context).match_limit;
mb->match_limit_depth = PRIV(default_match_context).depth_limit;
mb->heap_limit = PRIV(default_match_context).heap_limit;
}
else
{
if (mcontext->offset_limit != PCRE2_UNSET)
{
if ((re->overall_options & PCRE2_USE_OFFSET_LIMIT) == 0)
return PCRE2_ERROR_BADOFFSETLIMIT;
bumpalong_limit = subject + mcontext->offset_limit;
}
mb->callout = mcontext->callout;
mb->callout_data = mcontext->callout_data;
mb->memctl = mcontext->memctl;
mb->match_limit = mcontext->match_limit;
mb->match_limit_depth = mcontext->depth_limit;
mb->heap_limit = mcontext->heap_limit;
}
if (mb->match_limit > re->limit_match)
mb->match_limit = re->limit_match;
if (mb->match_limit_depth > re->limit_depth)
mb->match_limit_depth = re->limit_depth;
if (mb->heap_limit > re->limit_heap)
mb->heap_limit = re->limit_heap;
mb->start_code = (PCRE2_SPTR)((const uint8_t *)re + re->code_start);
mb->tables = re->tables;
mb->start_subject = subject;
mb->end_subject = end_subject;
mb->start_offset = start_offset;
mb->allowemptypartial = (re->max_lookbehind > 0) ||
(re->flags & PCRE2_MATCH_EMPTY) != 0;
mb->moptions = options;
mb->poptions = re->overall_options;
mb->match_call_count = 0;
mb->heap_used = 0;
/* Process the \R and newline settings. */
mb->bsr_convention = re->bsr_convention;
mb->nltype = NLTYPE_FIXED;
switch(re->newline_convention)
{
case PCRE2_NEWLINE_CR:
mb->nllen = 1;
mb->nl[0] = CHAR_CR;
break;
case PCRE2_NEWLINE_LF:
mb->nllen = 1;
mb->nl[0] = CHAR_NL;
break;
case PCRE2_NEWLINE_NUL:
mb->nllen = 1;
mb->nl[0] = CHAR_NUL;
break;
case PCRE2_NEWLINE_CRLF:
mb->nllen = 2;
mb->nl[0] = CHAR_CR;
mb->nl[1] = CHAR_NL;
break;
case PCRE2_NEWLINE_ANY:
mb->nltype = NLTYPE_ANY;
break;
case PCRE2_NEWLINE_ANYCRLF:
mb->nltype = NLTYPE_ANYCRLF;
break;
default:
PCRE2_DEBUG_UNREACHABLE();
return PCRE2_ERROR_INTERNAL;
}
/* Check a UTF string for validity if required. For 8-bit and 16-bit strings,
we must also check that a starting offset does not point into the middle of a
multiunit character. We check only the portion of the subject that is going to
be inspected during matching - from the offset minus the maximum back reference
to the given length. This saves time when a small part of a large subject is
being matched by the use of a starting offset. Note that the maximum lookbehind
is a number of characters, not code units. */
#ifdef SUPPORT_UNICODE
if (utf && (options & PCRE2_NO_UTF_CHECK) == 0)
{
PCRE2_SPTR check_subject = start_match; /* start_match includes offset */
if (start_offset > 0)
{
#if PCRE2_CODE_UNIT_WIDTH != 32
unsigned int i;
if (start_match < end_subject && NOT_FIRSTCU(*start_match))
return PCRE2_ERROR_BADUTFOFFSET;
for (i = re->max_lookbehind; i > 0 && check_subject > subject; i--)
{
check_subject--;
while (check_subject > subject &&
#if PCRE2_CODE_UNIT_WIDTH == 8
(*check_subject & 0xc0) == 0x80)
#else /* 16-bit */
(*check_subject & 0xfc00) == 0xdc00)
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */
check_subject--;
}
#else /* In the 32-bit library, one code unit equals one character. */
check_subject -= re->max_lookbehind;
if (check_subject < subject) check_subject = subject;
#endif /* PCRE2_CODE_UNIT_WIDTH != 32 */
}
/* Validate the relevant portion of the subject. After an error, adjust the
offset to be an absolute offset in the whole string. */
match_data->rc = PRIV(valid_utf)(check_subject,
length - (PCRE2_SIZE)(check_subject - subject), &(match_data->startchar));
if (match_data->rc != 0)
{
match_data->startchar += (PCRE2_SIZE)(check_subject - subject);
return match_data->rc;
}
}
#endif /* SUPPORT_UNICODE */
/* Set up the first code unit to match, if available. If there's no first code
unit there may be a bitmap of possible first characters. */
if ((re->flags & PCRE2_FIRSTSET) != 0)
{
has_first_cu = TRUE;
first_cu = first_cu2 = (PCRE2_UCHAR)(re->first_codeunit);
if ((re->flags & PCRE2_FIRSTCASELESS) != 0)
{
first_cu2 = TABLE_GET(first_cu, mb->tables + fcc_offset, first_cu);
#ifdef SUPPORT_UNICODE
#if PCRE2_CODE_UNIT_WIDTH == 8
if (first_cu > 127 && !utf && (re->overall_options & PCRE2_UCP) != 0)
first_cu2 = (PCRE2_UCHAR)UCD_OTHERCASE(first_cu);
#else
if (first_cu > 127 && (utf || (re->overall_options & PCRE2_UCP) != 0))
first_cu2 = (PCRE2_UCHAR)UCD_OTHERCASE(first_cu);
#endif
#endif /* SUPPORT_UNICODE */
}
}
else
if (!startline && (re->flags & PCRE2_FIRSTMAPSET) != 0)
start_bits = re->start_bitmap;
/* There may be a "last known required code unit" set. */
if ((re->flags & PCRE2_LASTSET) != 0)
{
has_req_cu = TRUE;
req_cu = req_cu2 = (PCRE2_UCHAR)(re->last_codeunit);
if ((re->flags & PCRE2_LASTCASELESS) != 0)
{
req_cu2 = TABLE_GET(req_cu, mb->tables + fcc_offset, req_cu);
#ifdef SUPPORT_UNICODE
#if PCRE2_CODE_UNIT_WIDTH == 8
if (req_cu > 127 && !utf && (re->overall_options & PCRE2_UCP) != 0)
req_cu2 = (PCRE2_UCHAR)UCD_OTHERCASE(req_cu);
#else
if (req_cu > 127 && (utf || (re->overall_options & PCRE2_UCP) != 0))
req_cu2 = (PCRE2_UCHAR)UCD_OTHERCASE(req_cu);
#endif
#endif /* SUPPORT_UNICODE */
}
}
/* If the match data block was previously used with PCRE2_COPY_MATCHED_SUBJECT,
free the memory that was obtained. */
if ((match_data->flags & PCRE2_MD_COPIED_SUBJECT) != 0)
{
match_data->memctl.free((void *)match_data->subject,
match_data->memctl.memory_data);
match_data->flags &= ~PCRE2_MD_COPIED_SUBJECT;
}
/* Fill in fields that are always returned in the match data. */
match_data->code = re;
match_data->subject = NULL; /* Default for no match */
match_data->mark = NULL;
match_data->matchedby = PCRE2_MATCHEDBY_DFA_INTERPRETER;
/* Call the main matching function, looping for a non-anchored regex after a
failed match. If not restarting, perform certain optimizations at the start of
a match. */
for (;;)
{
/* ----------------- Start of match optimizations ---------------- */
/* There are some optimizations that avoid running the match if a known
starting point is not found, or if a known later code unit is not present.
However, there is an option (settable at compile time) that disables
these, for testing and for ensuring that all callouts do actually occur.
The optimizations must also be avoided when restarting a DFA match. */
if ((re->optimization_flags & PCRE2_OPTIM_START_OPTIMIZE) != 0 &&
(options & PCRE2_DFA_RESTART) == 0)
{
/* If firstline is TRUE, the start of the match is constrained to the first
line of a multiline string. That is, the match must be before or at the
first newline following the start of matching. Temporarily adjust
end_subject so that we stop the optimization scans for a first code unit
immediately after the first character of a newline (the first code unit can
legitimately be a newline). If the match fails at the newline, later code
breaks this loop. */
if (firstline)
{
PCRE2_SPTR t = start_match;
#ifdef SUPPORT_UNICODE
if (utf)
{
while (t < end_subject && !IS_NEWLINE(t))
{
t++;
ACROSSCHAR(t < end_subject, t, t++);
}
}
else
#endif
while (t < end_subject && !IS_NEWLINE(t)) t++;
end_subject = t;
}
/* Anchored: check the first code unit if one is recorded. This may seem
pointless but it can help in detecting a no match case without scanning for
the required code unit. */
if (anchored)
{
if (has_first_cu || start_bits != NULL)
{
BOOL ok = start_match < end_subject;
if (ok)
{
PCRE2_UCHAR c = UCHAR21TEST(start_match);
ok = has_first_cu && (c == first_cu || c == first_cu2);
if (!ok && start_bits != NULL)
{
#if PCRE2_CODE_UNIT_WIDTH != 8
if (c > 255) c = 255;
#endif
ok = (start_bits[c/8] & (1u << (c&7))) != 0;
}
}
if (!ok) break;
}
}
/* Not anchored. Advance to a unique first code unit if there is one. */
else
{
if (has_first_cu)
{
if (first_cu != first_cu2) /* Caseless */
{
/* In 16-bit and 32_bit modes we have to do our own search, so can
look for both cases at once. */
#if PCRE2_CODE_UNIT_WIDTH != 8
PCRE2_UCHAR smc;
while (start_match < end_subject &&
(smc = UCHAR21TEST(start_match)) != first_cu &&
smc != first_cu2)
start_match++;
#else
/* In 8-bit mode, the use of memchr() gives a big speed up, even
though we have to call it twice in order to find the earliest
occurrence of the code unit in either of its cases. Caching is used
to remember the positions of previously found code units. This can
make a huge difference when the strings are very long and only one
case is actually present. */
PCRE2_SPTR pp1 = NULL;
PCRE2_SPTR pp2 = NULL;
PCRE2_SIZE searchlength = end_subject - start_match;
/* If we haven't got a previously found position for first_cu, or if
the current starting position is later, we need to do a search. If
the code unit is not found, set it to the end. */
if (memchr_found_first_cu == NULL ||
start_match > memchr_found_first_cu)
{
pp1 = memchr(start_match, first_cu, searchlength);
memchr_found_first_cu = (pp1 == NULL)? end_subject : pp1;
}
/* If the start is before a previously found position, use the
previous position, or NULL if a previous search failed. */
else pp1 = (memchr_found_first_cu == end_subject)? NULL :
memchr_found_first_cu;
/* Do the same thing for the other case. */
if (memchr_found_first_cu2 == NULL ||
start_match > memchr_found_first_cu2)
{
pp2 = memchr(start_match, first_cu2, searchlength);
memchr_found_first_cu2 = (pp2 == NULL)? end_subject : pp2;
}
else pp2 = (memchr_found_first_cu2 == end_subject)? NULL :
memchr_found_first_cu2;
/* Set the start to the end of the subject if neither case was found.
Otherwise, use the earlier found point. */
if (pp1 == NULL)
start_match = (pp2 == NULL)? end_subject : pp2;
else
start_match = (pp2 == NULL || pp1 < pp2)? pp1 : pp2;
#endif /* 8-bit handling */
}
/* The caseful case is much simpler. */
else
{
#if PCRE2_CODE_UNIT_WIDTH != 8
while (start_match < end_subject && UCHAR21TEST(start_match) !=
first_cu)
start_match++;
#else /* 8-bit code units */
start_match = memchr(start_match, first_cu, end_subject - start_match);
if (start_match == NULL) start_match = end_subject;
#endif
}
/* If we can't find the required code unit, having reached the true end
of the subject, break the bumpalong loop, to force a match failure,
except when doing partial matching, when we let the next cycle run at
the end of the subject. To see why, consider the pattern /(?<=abc)def/,
which partially matches "abc", even though the string does not contain
the starting character "d". If we have not reached the true end of the
subject (PCRE2_FIRSTLINE caused end_subject to be temporarily modified)
we also let the cycle run, because the matching string is legitimately
allowed to start with the first code unit of a newline. */
if ((mb->moptions & (PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT)) == 0 &&
start_match >= mb->end_subject)
break;
}
/* If there's no first code unit, advance to just after a linebreak for a
multiline match if required. */
else if (startline)
{
if (start_match > mb->start_subject + start_offset)
{
#ifdef SUPPORT_UNICODE
if (utf)
{
while (start_match < end_subject && !WAS_NEWLINE(start_match))
{
start_match++;
ACROSSCHAR(start_match < end_subject, start_match, start_match++);
}
}
else
#endif
while (start_match < end_subject && !WAS_NEWLINE(start_match))
start_match++;
/* If we have just passed a CR and the newline option is ANY or
ANYCRLF, and we are now at a LF, advance the match position by one
more code unit. */
if (start_match[-1] == CHAR_CR &&
(mb->nltype == NLTYPE_ANY || mb->nltype == NLTYPE_ANYCRLF) &&
start_match < end_subject &&
UCHAR21TEST(start_match) == CHAR_NL)
start_match++;
}
}
/* If there's no first code unit or a requirement for a multiline line
start, advance to a non-unique first code unit if any have been
identified. The bitmap contains only 256 bits. When code units are 16 or
32 bits wide, all code units greater than 254 set the 255 bit. */
else if (start_bits != NULL)
{
while (start_match < end_subject)
{
uint32_t c = UCHAR21TEST(start_match);
#if PCRE2_CODE_UNIT_WIDTH != 8
if (c > 255) c = 255;
#endif
if ((start_bits[c/8] & (1u << (c&7))) != 0) break;
start_match++;
}
/* See comment above in first_cu checking about the next line. */
if ((mb->moptions & (PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT)) == 0 &&
start_match >= mb->end_subject)
break;
}
} /* End of first code unit handling */
/* Restore fudged end_subject */
end_subject = mb->end_subject;
/* The following two optimizations are disabled for partial matching. */
if ((mb->moptions & (PCRE2_PARTIAL_HARD|PCRE2_PARTIAL_SOFT)) == 0)
{
PCRE2_SPTR p;
/* The minimum matching length is a lower bound; no actual string of that
length may actually match the pattern. Although the value is, strictly,
in characters, we treat it as code units to avoid spending too much time
in this optimization. */
if (end_subject - start_match < re->minlength) goto NOMATCH_EXIT;
/* If req_cu is set, we know that that code unit must appear in the
subject for the match to succeed. If the first code unit is set, req_cu
must be later in the subject; otherwise the test starts at the match
point. This optimization can save a huge amount of backtracking in
patterns with nested unlimited repeats that aren't going to match.
Writing separate code for cased/caseless versions makes it go faster, as
does using an autoincrement and backing off on a match. As in the case of
the first code unit, using memchr() in the 8-bit library gives a big
speed up. Unlike the first_cu check above, we do not need to call
memchr() twice in the caseless case because we only need to check for the
presence of the character in either case, not find the first occurrence.
The search can be skipped if the code unit was found later than the
current starting point in a previous iteration of the bumpalong loop.
HOWEVER: when the subject string is very, very long, searching to its end
can take a long time, and give bad performance on quite ordinary
patterns. This showed up when somebody was matching something like
/^\d+C/ on a 32-megabyte string... so we don't do this when the string is
sufficiently long, but it's worth searching a lot more for unanchored
patterns. */
p = start_match + (has_first_cu? 1:0);
if (has_req_cu && p > req_cu_ptr)
{
PCRE2_SIZE check_length = end_subject - start_match;
if (check_length < REQ_CU_MAX ||
(!anchored && check_length < REQ_CU_MAX * 1000))
{
if (req_cu != req_cu2) /* Caseless */
{
#if PCRE2_CODE_UNIT_WIDTH != 8
while (p < end_subject)
{
uint32_t pp = UCHAR21INCTEST(p);
if (pp == req_cu || pp == req_cu2) { p--; break; }
}
#else /* 8-bit code units */
PCRE2_SPTR pp = p;
p = memchr(pp, req_cu, end_subject - pp);
if (p == NULL)
{
p = memchr(pp, req_cu2, end_subject - pp);
if (p == NULL) p = end_subject;
}
#endif /* PCRE2_CODE_UNIT_WIDTH != 8 */
}
/* The caseful case */
else
{
#if PCRE2_CODE_UNIT_WIDTH != 8
while (p < end_subject)
{
if (UCHAR21INCTEST(p) == req_cu) { p--; break; }
}
#else /* 8-bit code units */
p = memchr(p, req_cu, end_subject - p);
if (p == NULL) p = end_subject;
#endif
}
/* If we can't find the required code unit, break the matching loop,
forcing a match failure. */
if (p >= end_subject) break;
/* If we have found the required code unit, save the point where we
found it, so that we don't search again next time round the loop if
the start hasn't passed this code unit yet. */
req_cu_ptr = p;
}
}
}
}
/* ------------ End of start of match optimizations ------------ */
/* Give no match if we have passed the bumpalong limit. */
if (start_match > bumpalong_limit) break;
/* OK, now we can do the business */
mb->start_used_ptr = start_match;
mb->last_used_ptr = start_match;
mb->recursive = NULL;
rc = internal_dfa_match(
mb, /* fixed match data */
mb->start_code, /* this subexpression's code */
start_match, /* where we currently are */
start_offset, /* start offset in subject */
match_data->ovector, /* offset vector */
(uint32_t)match_data->oveccount * 2, /* actual size of same */
workspace, /* workspace vector */
(int)wscount, /* size of same */
0, /* function recurse level */
base_recursion_workspace); /* initial workspace for recursion */
/* Anything other than "no match" means we are done, always; otherwise, carry
on only if not anchored. */
if (rc != PCRE2_ERROR_NOMATCH || anchored)
{
if (rc == PCRE2_ERROR_PARTIAL && match_data->oveccount > 0)
{
match_data->ovector[0] = (PCRE2_SIZE)(start_match - subject);
match_data->ovector[1] = (PCRE2_SIZE)(end_subject - subject);
}
match_data->subject_length = length;
match_data->leftchar = (PCRE2_SIZE)(mb->start_used_ptr - subject);
match_data->rightchar = (PCRE2_SIZE)(mb->last_used_ptr - subject);
match_data->startchar = (PCRE2_SIZE)(start_match - subject);
match_data->rc = rc;
if (rc >= 0 &&(options & PCRE2_COPY_MATCHED_SUBJECT) != 0)
{
length = CU2BYTES(length + was_zero_terminated);
match_data->subject = match_data->memctl.malloc(length,
match_data->memctl.memory_data);
if (match_data->subject == NULL) return PCRE2_ERROR_NOMEMORY;
memcpy((void *)match_data->subject, subject, length);
match_data->flags |= PCRE2_MD_COPIED_SUBJECT;
}
else
{
if (rc >= 0 || rc == PCRE2_ERROR_PARTIAL) match_data->subject = subject;
}
goto EXIT;
}
/* Advance to the next subject character unless we are at the end of a line
and firstline is set. */
if (firstline && IS_NEWLINE(start_match)) break;
start_match++;
#ifdef SUPPORT_UNICODE
if (utf)
{
ACROSSCHAR(start_match < end_subject, start_match, start_match++);
}
#endif
if (start_match > end_subject) break;
/* If we have just passed a CR and we are now at a LF, and the pattern does
not contain any explicit matches for \r or \n, and the newline option is CRLF
or ANY or ANYCRLF, advance the match position by one more character. */
if (UCHAR21TEST(start_match - 1) == CHAR_CR &&
start_match < end_subject &&
UCHAR21TEST(start_match) == CHAR_NL &&
(re->flags & PCRE2_HASCRORLF) == 0 &&
(mb->nltype == NLTYPE_ANY ||
mb->nltype == NLTYPE_ANYCRLF ||
mb->nllen == 2))
start_match++;
} /* "Bumpalong" loop */
NOMATCH_EXIT:
rc = PCRE2_ERROR_NOMATCH;
EXIT:
while (rws->next != NULL)
{
RWS_anchor *next = rws->next;
rws->next = next->next;
mb->memctl.free(next, mb->memctl.memory_data);
}
return rc;
}
/* These #undefs are here to enable unity builds with CMake. */
#undef NLBLOCK /* Block containing newline information */
#undef PSSTART /* Field containing processed string start */
#undef PSEND /* Field containing processed string end */
/* End of pcre2_dfa_match.c */
pcre2-10.45/src/pcre2_compile_class.c 0000644 0001751 0000166 00000224301 14750415257 013077 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "pcre2_compile.h"
typedef struct {
/* Option bits for eclass. */
uint32_t options;
uint32_t xoptions;
/* Rarely used members. */
int *errorcodeptr;
compile_block *cb;
/* Bitmap is needed. */
BOOL needs_bitmap;
} eclass_context;
/* Checks the allowed tokens at the end of a class structure in debug mode.
When a new token is not processed by all loops, and the token is equals to
a) one of the cases here:
the compiler will complain about a duplicated case value.
b) none of the cases here:
the loop without the handler will stop with an assertion failure. */
#ifdef PCRE2_DEBUG
#define CLASS_END_CASES(meta) \
default: \
PCRE2_ASSERT((meta) <= META_END); \
/* Fall through */ \
case META_CLASS: \
case META_CLASS_NOT: \
case META_CLASS_EMPTY: \
case META_CLASS_EMPTY_NOT: \
case META_CLASS_END: \
case META_ECLASS_AND: \
case META_ECLASS_OR: \
case META_ECLASS_SUB: \
case META_ECLASS_XOR: \
case META_ECLASS_NOT:
#else
#define CLASS_END_CASES(meta) \
default:
#endif
#ifdef SUPPORT_WIDE_CHARS
/* Heapsort algorithm. */
static void do_heapify(uint32_t *buffer, size_t size, size_t i)
{
size_t max;
size_t left;
size_t right;
uint32_t tmp1, tmp2;
while (TRUE)
{
max = i;
left = (i << 1) + 2;
right = left + 2;
if (left < size && buffer[left] > buffer[max]) max = left;
if (right < size && buffer[right] > buffer[max]) max = right;
if (i == max) return;
/* Swap items. */
tmp1 = buffer[i];
tmp2 = buffer[i + 1];
buffer[i] = buffer[max];
buffer[i + 1] = buffer[max + 1];
buffer[max] = tmp1;
buffer[max + 1] = tmp2;
i = max;
}
}
#ifdef SUPPORT_UNICODE
#define PARSE_CLASS_UTF 0x1
#define PARSE_CLASS_CASELESS_UTF 0x2
#define PARSE_CLASS_RESTRICTED_UTF 0x4
#define PARSE_CLASS_TURKISH_UTF 0x8
/* Get the range of nocase characters which includes the
'c' character passed as argument, or directly follows 'c'. */
static const uint32_t*
get_nocase_range(uint32_t c)
{
uint32_t left = 0;
uint32_t right = PRIV(ucd_nocase_ranges_size);
uint32_t middle;
if (c > MAX_UTF_CODE_POINT) return PRIV(ucd_nocase_ranges) + right;
while (TRUE)
{
/* Range end of the middle element. */
middle = ((left + right) >> 1) | 0x1;
if (PRIV(ucd_nocase_ranges)[middle] <= c)
left = middle + 1;
else if (middle > 1 && PRIV(ucd_nocase_ranges)[middle - 2] > c)
right = middle - 1;
else
return PRIV(ucd_nocase_ranges) + (middle - 1);
}
}
/* Get the list of othercase characters, which belongs to the passed range.
Create ranges from these characters, and append them to the buffer argument. */
static size_t
utf_caseless_extend(uint32_t start, uint32_t end, uint32_t options,
uint32_t *buffer)
{
uint32_t new_start = start;
uint32_t new_end = end;
uint32_t c = start;
const uint32_t *list;
uint32_t tmp[3];
size_t result = 2;
const uint32_t *skip_range = get_nocase_range(c);
uint32_t skip_start = skip_range[0];
#if PCRE2_CODE_UNIT_WIDTH == 8
PCRE2_ASSERT(options & PARSE_CLASS_UTF);
#endif
#if PCRE2_CODE_UNIT_WIDTH == 32
if (end > MAX_UTF_CODE_POINT) end = MAX_UTF_CODE_POINT;
#endif
while (c <= end)
{
uint32_t co;
if (c > skip_start)
{
c = skip_range[1];
skip_range += 2;
skip_start = skip_range[0];
continue;
}
/* Compute caseless set. */
if ((options & (PARSE_CLASS_TURKISH_UTF|PARSE_CLASS_RESTRICTED_UTF)) ==
PARSE_CLASS_TURKISH_UTF &&
UCD_ANY_I(c))
{
co = PRIV(ucd_turkish_dotted_i_caseset) + (UCD_DOTTED_I(c)? 0 : 3);
}
else if ((co = UCD_CASESET(c)) != 0 &&
(options & PARSE_CLASS_RESTRICTED_UTF) != 0 &&
PRIV(ucd_caseless_sets)[co] < 128)
{
co = 0; /* Ignore the caseless set if it's restricted. */
}
if (co != 0)
list = PRIV(ucd_caseless_sets) + co;
else
{
co = UCD_OTHERCASE(c);
list = tmp;
tmp[0] = c;
tmp[1] = NOTACHAR;
if (co != c)
{
tmp[1] = co;
tmp[2] = NOTACHAR;
}
}
c++;
/* Add characters. */
do
{
#if PCRE2_CODE_UNIT_WIDTH == 16
if (!(options & PARSE_CLASS_UTF) && *list > 0xffff) continue;
#endif
if (*list < new_start)
{
if (*list + 1 == new_start)
{
new_start--;
continue;
}
}
else if (*list > new_end)
{
if (*list - 1 == new_end)
{
new_end++;
continue;
}
}
else continue;
result += 2;
if (buffer != NULL)
{
buffer[0] = *list;
buffer[1] = *list;
buffer += 2;
}
}
while (*(++list) != NOTACHAR);
}
if (buffer != NULL)
{
buffer[0] = new_start;
buffer[1] = new_end;
buffer += 2;
(void)buffer;
}
return result;
}
#endif
/* Add a character list to a buffer. */
static size_t
append_char_list(const uint32_t *p, uint32_t *buffer)
{
const uint32_t *n;
size_t result = 0;
while (*p != NOTACHAR)
{
n = p;
while (n[0] == n[1] - 1) n++;
PCRE2_ASSERT(*p < 0xffff);
if (buffer != NULL)
{
buffer[0] = *p;
buffer[1] = *n;
buffer += 2;
}
result += 2;
p = n + 1;
}
return result;
}
static uint32_t
get_highest_char(uint32_t options)
{
(void)options; /* Avoid compiler warning. */
#if PCRE2_CODE_UNIT_WIDTH == 8
return MAX_UTF_CODE_POINT;
#else
#ifdef SUPPORT_UNICODE
return GET_MAX_CHAR_VALUE((options & PARSE_CLASS_UTF) != 0);
#else
return MAX_UCHAR_VALUE;
#endif
#endif
}
/* Add a negated character list to a buffer. */
static size_t
append_negated_char_list(const uint32_t *p, uint32_t options, uint32_t *buffer)
{
const uint32_t *n;
uint32_t start = 0;
size_t result = 2;
PCRE2_ASSERT(*p > 0);
while (*p != NOTACHAR)
{
n = p;
while (n[0] == n[1] - 1) n++;
PCRE2_ASSERT(*p < 0xffff);
if (buffer != NULL)
{
buffer[0] = start;
buffer[1] = *p - 1;
buffer += 2;
}
result += 2;
start = *n + 1;
p = n + 1;
}
if (buffer != NULL)
{
buffer[0] = start;
buffer[1] = get_highest_char(options);
buffer += 2;
(void)buffer;
}
return result;
}
static uint32_t *
append_non_ascii_range(uint32_t options, uint32_t *buffer)
{
if (buffer == NULL) return NULL;
buffer[0] = 0x100;
buffer[1] = get_highest_char(options);
return buffer + 2;
}
static size_t
parse_class(uint32_t *ptr, uint32_t options, uint32_t *buffer)
{
size_t total_size = 0;
size_t size;
uint32_t meta_arg;
uint32_t start_char;
while (TRUE)
{
switch (META_CODE(*ptr))
{
case META_ESCAPE:
meta_arg = META_DATA(*ptr);
switch (meta_arg)
{
case ESC_D:
case ESC_W:
case ESC_S:
buffer = append_non_ascii_range(options, buffer);
total_size += 2;
break;
case ESC_h:
size = append_char_list(PRIV(hspace_list), buffer);
total_size += size;
if (buffer != NULL) buffer += size;
break;
case ESC_H:
size = append_negated_char_list(PRIV(hspace_list), options, buffer);
total_size += size;
if (buffer != NULL) buffer += size;
break;
case ESC_v:
size = append_char_list(PRIV(vspace_list), buffer);
total_size += size;
if (buffer != NULL) buffer += size;
break;
case ESC_V:
size = append_negated_char_list(PRIV(vspace_list), options, buffer);
total_size += size;
if (buffer != NULL) buffer += size;
break;
case ESC_p:
case ESC_P:
ptr++;
if (meta_arg == ESC_p && (*ptr >> 16) == PT_ANY)
{
if (buffer != NULL)
{
buffer[0] = 0;
buffer[1] = get_highest_char(options);
buffer += 2;
}
total_size += 2;
}
break;
}
ptr++;
continue;
case META_POSIX_NEG:
buffer = append_non_ascii_range(options, buffer);
total_size += 2;
ptr += 2;
continue;
case META_POSIX:
ptr += 2;
continue;
case META_BIGVALUE:
/* Character literal */
ptr++;
break;
CLASS_END_CASES(*ptr)
if (*ptr >= META_END) return total_size;
break;
}
start_char = *ptr;
if (ptr[1] == META_RANGE_LITERAL || ptr[1] == META_RANGE_ESCAPED)
{
ptr += 2;
PCRE2_ASSERT(*ptr < META_END || *ptr == META_BIGVALUE);
if (*ptr == META_BIGVALUE) ptr++;
#ifdef EBCDIC
#error "Missing EBCDIC support"
#endif
}
#ifdef SUPPORT_UNICODE
if (options & PARSE_CLASS_CASELESS_UTF)
{
size = utf_caseless_extend(start_char, *ptr++, options, buffer);
if (buffer != NULL) buffer += size;
total_size += size;
continue;
}
#endif
if (buffer != NULL)
{
buffer[0] = start_char;
buffer[1] = *ptr;
buffer += 2;
}
ptr++;
total_size += 2;
}
return total_size;
}
/* Extra uint32_t values for storing the lengths of range lists in
the worst case. Two uint32_t lengths and a range end for a range
starting before 255 */
#define CHAR_LIST_EXTRA_SIZE 3
/* Starting character values for each character list. */
static const uint32_t char_list_starts[] = {
#if PCRE2_CODE_UNIT_WIDTH == 32
XCL_CHAR_LIST_HIGH_32_START,
#endif
#if PCRE2_CODE_UNIT_WIDTH == 32 || defined SUPPORT_UNICODE
XCL_CHAR_LIST_LOW_32_START,
#endif
XCL_CHAR_LIST_HIGH_16_START,
/* Must be terminated by XCL_CHAR_LIST_LOW_16_START,
which also represents the end of the bitset. */
XCL_CHAR_LIST_LOW_16_START,
};
static class_ranges *
compile_optimize_class(uint32_t *start_ptr, uint32_t options,
uint32_t xoptions, compile_block *cb)
{
class_ranges* cranges;
uint32_t *ptr;
uint32_t *buffer;
uint32_t *dst;
uint32_t class_options = 0;
size_t range_list_size = 0, total_size, i;
uint32_t tmp1, tmp2;
const uint32_t *char_list_next;
uint16_t *next_char;
uint32_t char_list_start, char_list_end;
uint32_t range_start, range_end;
#ifdef SUPPORT_UNICODE
if (options & PCRE2_UTF)
class_options |= PARSE_CLASS_UTF;
if ((options & PCRE2_CASELESS) && (options & (PCRE2_UTF|PCRE2_UCP)))
class_options |= PARSE_CLASS_CASELESS_UTF;
if (xoptions & PCRE2_EXTRA_CASELESS_RESTRICT)
class_options |= PARSE_CLASS_RESTRICTED_UTF;
if (xoptions & PCRE2_EXTRA_TURKISH_CASING)
class_options |= PARSE_CLASS_TURKISH_UTF;
#endif
/* Compute required space for the range. */
range_list_size = parse_class(start_ptr, class_options, NULL);
PCRE2_ASSERT((range_list_size & 0x1) == 0);
/* Allocate buffer. The total_size also represents the end of the buffer. */
total_size = range_list_size +
((range_list_size >= 2) ? CHAR_LIST_EXTRA_SIZE : 0);
cranges = cb->cx->memctl.malloc(
sizeof(class_ranges) + total_size * sizeof(uint32_t),
cb->cx->memctl.memory_data);
if (cranges == NULL) return NULL;
cranges->next = NULL;
cranges->range_list_size = (uint16_t)range_list_size;
cranges->char_lists_types = 0;
cranges->char_lists_size = 0;
cranges->char_lists_start = 0;
if (range_list_size == 0) return cranges;
buffer = (uint32_t*)(cranges + 1);
parse_class(start_ptr, class_options, buffer);
/* Using <= instead of == to help static analysis. */
if (range_list_size <= 2) return cranges;
/* In-place sorting of ranges. */
i = (((range_list_size >> 2) - 1) << 1);
while (TRUE)
{
do_heapify(buffer, range_list_size, i);
if (i == 0) break;
i -= 2;
}
i = range_list_size - 2;
while (TRUE)
{
tmp1 = buffer[i];
tmp2 = buffer[i + 1];
buffer[i] = buffer[0];
buffer[i + 1] = buffer[1];
buffer[0] = tmp1;
buffer[1] = tmp2;
do_heapify(buffer, i, 0);
if (i == 0) break;
i -= 2;
}
/* Merge ranges whenever possible. */
dst = buffer;
ptr = buffer + 2;
range_list_size -= 2;
/* The second condition is a very rare corner case, where the end of the last
range is the maximum character. This range cannot be extended further. */
while (range_list_size > 0 && dst[1] != ~(uint32_t)0)
{
if (dst[1] + 1 < ptr[0])
{
dst += 2;
dst[0] = ptr[0];
dst[1] = ptr[1];
}
else if (dst[1] < ptr[1]) dst[1] = ptr[1];
ptr += 2;
range_list_size -= 2;
}
PCRE2_ASSERT(dst[1] <= get_highest_char(class_options));
/* When the number of ranges are less than six,
they are not converted to range lists. */
ptr = buffer;
while (ptr < dst && ptr[1] < 0x100) ptr += 2;
if (dst - ptr < (2 * (6 - 1)))
{
cranges->range_list_size = (uint16_t)(dst + 2 - buffer);
return cranges;
}
/* Compute character lists structures. */
char_list_next = char_list_starts;
char_list_start = *char_list_next++;
#if PCRE2_CODE_UNIT_WIDTH == 32
char_list_end = XCL_CHAR_LIST_HIGH_32_END;
#elif defined SUPPORT_UNICODE
char_list_end = XCL_CHAR_LIST_LOW_32_END;
#else
char_list_end = XCL_CHAR_LIST_HIGH_16_END;
#endif
next_char = (uint16_t*)(buffer + total_size);
tmp1 = 0;
tmp2 = ((sizeof(char_list_starts) / sizeof(uint32_t)) - 1) * XCL_TYPE_BIT_LEN;
PCRE2_ASSERT(tmp2 <= 3 * XCL_TYPE_BIT_LEN && tmp2 >= XCL_TYPE_BIT_LEN);
range_start = dst[0];
range_end = dst[1];
while (TRUE)
{
if (range_start >= char_list_start)
{
if (range_start == range_end || range_end < char_list_end)
{
tmp1++;
next_char--;
if (char_list_start < XCL_CHAR_LIST_LOW_32_START)
*next_char = (uint16_t)((range_end << XCL_CHAR_SHIFT) | XCL_CHAR_END);
else
*(uint32_t*)(--next_char) =
(range_end << XCL_CHAR_SHIFT) | XCL_CHAR_END;
}
if (range_start < range_end)
{
if (range_start > char_list_start)
{
tmp1++;
next_char--;
if (char_list_start < XCL_CHAR_LIST_LOW_32_START)
*next_char = (uint16_t)(range_start << XCL_CHAR_SHIFT);
else
*(uint32_t*)(--next_char) = (range_start << XCL_CHAR_SHIFT);
}
else
cranges->char_lists_types |= XCL_BEGIN_WITH_RANGE << tmp2;
}
PCRE2_ASSERT((uint32_t*)next_char >= dst + 2);
if (dst > buffer)
{
dst -= 2;
range_start = dst[0];
range_end = dst[1];
continue;
}
range_start = 0;
range_end = 0;
}
if (range_end >= char_list_start)
{
PCRE2_ASSERT(range_start < char_list_start);
if (range_end < char_list_end)
{
tmp1++;
next_char--;
if (char_list_start < XCL_CHAR_LIST_LOW_32_START)
*next_char = (uint16_t)((range_end << XCL_CHAR_SHIFT) | XCL_CHAR_END);
else
*(uint32_t*)(--next_char) =
(range_end << XCL_CHAR_SHIFT) | XCL_CHAR_END;
PCRE2_ASSERT((uint32_t*)next_char >= dst + 2);
}
cranges->char_lists_types |= XCL_BEGIN_WITH_RANGE << tmp2;
}
if (tmp1 >= XCL_ITEM_COUNT_MASK)
{
cranges->char_lists_types |= XCL_ITEM_COUNT_MASK << tmp2;
next_char--;
if (char_list_start < XCL_CHAR_LIST_LOW_32_START)
*next_char = (uint16_t)tmp1;
else
*(uint32_t*)(--next_char) = tmp1;
}
else
cranges->char_lists_types |= tmp1 << tmp2;
if (range_start < XCL_CHAR_LIST_LOW_16_START) break;
PCRE2_ASSERT(tmp2 >= XCL_TYPE_BIT_LEN);
char_list_end = char_list_start - 1;
char_list_start = *char_list_next++;
tmp1 = 0;
tmp2 -= XCL_TYPE_BIT_LEN;
}
if (dst[0] < XCL_CHAR_LIST_LOW_16_START) dst += 2;
PCRE2_ASSERT((uint16_t*)dst <= next_char);
cranges->char_lists_size =
(size_t)((uint8_t*)(buffer + total_size) - (uint8_t*)next_char);
cranges->char_lists_start = (size_t)((uint8_t*)next_char - (uint8_t*)buffer);
cranges->range_list_size = (uint16_t)(dst - buffer);
return cranges;
}
#endif /* SUPPORT_WIDE_CHARS */
#ifdef SUPPORT_UNICODE
void PRIV(update_classbits)(uint32_t ptype, uint32_t pdata, BOOL negated,
uint8_t *classbits)
{
/* Update PRIV(xclass) when this function is changed. */
int c, chartype;
const ucd_record *prop;
uint32_t gentype;
BOOL set_bit;
if (ptype == PT_ANY)
{
if (!negated) memset(classbits, 0xff, 32);
return;
}
for (c = 0; c < 256; c++)
{
prop = GET_UCD(c);
set_bit = FALSE;
(void)set_bit;
switch (ptype)
{
case PT_LAMP:
chartype = prop->chartype;
set_bit = (chartype == ucp_Lu || chartype == ucp_Ll || chartype == ucp_Lt);
break;
case PT_GC:
set_bit = (PRIV(ucp_gentype)[prop->chartype] == pdata);
break;
case PT_PC:
set_bit = (prop->chartype == pdata);
break;
case PT_SC:
set_bit = (prop->script == pdata);
break;
case PT_SCX:
set_bit = (prop->script == pdata ||
MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), pdata) != 0);
break;
case PT_ALNUM:
gentype = PRIV(ucp_gentype)[prop->chartype];
set_bit = (gentype == ucp_L || gentype == ucp_N);
break;
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_BYTE_CASES:
VSPACE_BYTE_CASES:
set_bit = TRUE;
break;
default:
set_bit = (PRIV(ucp_gentype)[prop->chartype] == ucp_Z);
break;
}
break;
case PT_WORD:
chartype = prop->chartype;
gentype = PRIV(ucp_gentype)[chartype];
set_bit = (gentype == ucp_L || gentype == ucp_N ||
chartype == ucp_Mn || chartype == ucp_Pc);
break;
case PT_UCNC:
set_bit = (c == CHAR_DOLLAR_SIGN || c == CHAR_COMMERCIAL_AT ||
c == CHAR_GRAVE_ACCENT || c >= 0xa0);
break;
case PT_BIDICL:
set_bit = (UCD_BIDICLASS_PROP(prop) == pdata);
break;
case PT_BOOL:
set_bit = MAPBIT(PRIV(ucd_boolprop_sets) +
UCD_BPROPS_PROP(prop), pdata) != 0;
break;
case PT_PXGRAPH:
chartype = prop->chartype;
gentype = PRIV(ucp_gentype)[chartype];
set_bit = (gentype != ucp_Z && (gentype != ucp_C || chartype == ucp_Cf));
break;
case PT_PXPRINT:
chartype = prop->chartype;
set_bit = (chartype != ucp_Zl && chartype != ucp_Zp &&
(PRIV(ucp_gentype)[chartype] != ucp_C || chartype == ucp_Cf));
break;
case PT_PXPUNCT:
gentype = PRIV(ucp_gentype)[prop->chartype];
set_bit = (gentype == ucp_P || (c < 128 && gentype == ucp_S));
break;
default:
PCRE2_ASSERT(ptype == PT_PXXDIGIT);
set_bit = (c >= CHAR_0 && c <= CHAR_9) ||
(c >= CHAR_A && c <= CHAR_F) ||
(c >= CHAR_a && c <= CHAR_f);
break;
}
if (negated) set_bit = !set_bit;
if (set_bit) *classbits |= (uint8_t)(1 << (c & 0x7));
if ((c & 0x7) == 0x7) classbits++;
}
}
#endif /* SUPPORT_UNICODE */
#ifdef SUPPORT_WIDE_CHARS
/*************************************************
* XClass related properties *
*************************************************/
/* XClass needs to be generated. */
#define XCLASS_REQUIRED 0x1
/* XClass has 8 bit character. */
#define XCLASS_HAS_8BIT_CHARS 0x2
/* XClass has properties. */
#define XCLASS_HAS_PROPS 0x4
/* XClass has character lists. */
#define XCLASS_HAS_CHAR_LISTS 0x8
/* XClass matches to all >= 256 characters. */
#define XCLASS_HIGH_ANY 0x10
#endif
/*************************************************
* Internal entry point for add range to class *
*************************************************/
/* This function sets the overall range for characters < 256.
It also handles non-utf case folding.
Arguments:
options the options bits
xoptions the extra options bits
cb compile data
start start of range character
end end of range character
Returns: cb->classbits is updated
*/
static void
add_to_class(uint32_t options, uint32_t xoptions, compile_block *cb,
uint32_t start, uint32_t end)
{
uint8_t *classbits = cb->classbits.classbits;
uint32_t c, byte_start, byte_end;
uint32_t classbits_end = (end <= 0xff ? end : 0xff);
/* If caseless matching is required, scan the range and process alternate
cases. In Unicode, there are 8-bit characters that have alternate cases that
are greater than 255 and vice-versa (though these may be ignored if caseless
restriction is in force). Sometimes we can just extend the original range. */
if ((options & PCRE2_CASELESS) != 0)
{
#ifdef SUPPORT_UNICODE
/* UTF mode. This branch is taken if we don't support wide characters (e.g.
8-bit library, without UTF), but we do treat those characters as Unicode
(if UCP flag is set). In this case, we only need to expand the character class
set to include the case pairs which are in the 0-255 codepoint range. */
if ((options & (PCRE2_UTF|PCRE2_UCP)) != 0)
{
BOOL turkish_i = (xoptions & (PCRE2_EXTRA_TURKISH_CASING|PCRE2_EXTRA_CASELESS_RESTRICT)) ==
PCRE2_EXTRA_TURKISH_CASING;
if (start < 128)
{
uint32_t lo_end = (classbits_end < 127 ? classbits_end : 127);
for (c = start; c <= lo_end; c++)
{
if (turkish_i && UCD_ANY_I(c)) continue;
SETBIT(classbits, cb->fcc[c]);
}
}
if (classbits_end >= 128)
{
uint32_t hi_start = (start > 128 ? start : 128);
for (c = hi_start; c <= classbits_end; c++)
{
uint32_t co = UCD_OTHERCASE(c);
if (co <= 0xff) SETBIT(classbits, co);
}
}
}
else
#endif /* SUPPORT_UNICODE */
/* Not UTF mode */
{
for (c = start; c <= classbits_end; c++)
SETBIT(classbits, cb->fcc[c]);
}
}
/* Use the bitmap for characters < 256. Otherwise use extra data. */
byte_start = (start + 7) >> 3;
byte_end = (classbits_end + 1) >> 3;
if (byte_start >= byte_end)
{
for (c = start; c <= classbits_end; c++)
/* Regardless of start, c will always be <= 255. */
SETBIT(classbits, c);
return;
}
for (c = byte_start; c < byte_end; c++)
classbits[c] = 0xff;
byte_start <<= 3;
byte_end <<= 3;
for (c = start; c < byte_start; c++)
SETBIT(classbits, c);
for (c = byte_end; c <= classbits_end; c++)
SETBIT(classbits, c);
}
#if PCRE2_CODE_UNIT_WIDTH == 8
/*************************************************
* Internal entry point for add list to class *
*************************************************/
/* This function is used for adding a list of horizontal or vertical whitespace
characters to a class. The list must be in order so that ranges of characters
can be detected and handled appropriately. This function sets the overall range
so that the internal functions can try to avoid duplication when handling
case-independence.
Arguments:
options the options bits
xoptions the extra options bits
cb contains pointers to tables etc.
p points to row of 32-bit values, terminated by NOTACHAR
Returns: cb->classbits is updated
*/
static void
add_list_to_class(uint32_t options, uint32_t xoptions, compile_block *cb,
const uint32_t *p)
{
while (p[0] < 256)
{
unsigned int n = 0;
while(p[n+1] == p[0] + n + 1) n++;
add_to_class(options, xoptions, cb, p[0], p[n]);
p += n + 1;
}
}
/*************************************************
* Add characters not in a list to a class *
*************************************************/
/* This function is used for adding the complement of a list of horizontal or
vertical whitespace to a class. The list must be in order.
Arguments:
options the options bits
xoptions the extra options bits
cb contains pointers to tables etc.
p points to row of 32-bit values, terminated by NOTACHAR
Returns: cb->classbits is updated
*/
static void
add_not_list_to_class(uint32_t options, uint32_t xoptions, compile_block *cb,
const uint32_t *p)
{
if (p[0] > 0)
add_to_class(options, xoptions, cb, 0, p[0] - 1);
while (p[0] < 256)
{
while (p[1] == p[0] + 1) p++;
add_to_class(options, xoptions, cb, p[0] + 1, (p[1] > 255) ? 255 : p[1] - 1);
p++;
}
}
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */
/*************************************************
* Main entry-point to compile a character class *
*************************************************/
/* This function consumes a "leaf", which is a set of characters that will
become a single OP_CLASS OP_NCLASS, OP_XCLASS, or OP_ALLANY. */
uint32_t *
PRIV(compile_class_not_nested)(uint32_t options, uint32_t xoptions,
uint32_t *start_ptr, PCRE2_UCHAR **pcode, BOOL negate_class, BOOL* has_bitmap,
int *errorcodeptr, compile_block *cb, PCRE2_SIZE *lengthptr)
{
uint32_t *pptr = start_ptr;
PCRE2_UCHAR *code = *pcode;
BOOL should_flip_negation;
const uint8_t *cbits = cb->cbits;
/* Some functions such as add_to_class() or eclass processing
expects that the bitset is stored in cb->classbits.classbits. */
uint8_t *const classbits = cb->classbits.classbits;
#ifdef SUPPORT_UNICODE
BOOL utf = (options & PCRE2_UTF) != 0;
#else /* No Unicode support */
BOOL utf = FALSE;
#endif
/* Helper variables for OP_XCLASS opcode (for characters > 255). */
#ifdef SUPPORT_WIDE_CHARS
uint32_t xclass_props;
PCRE2_UCHAR *class_uchardata;
class_ranges* cranges;
#endif
/* If an XClass contains a negative special such as \S, we need to flip the
negation flag at the end, so that support for characters > 255 works correctly
(they are all included in the class). An XClass may need to insert specific
matching or non-matching code for wide characters.
*/
should_flip_negation = FALSE;
/* XClass will be used when characters > 255 might match. */
#ifdef SUPPORT_WIDE_CHARS
xclass_props = 0;
#if PCRE2_CODE_UNIT_WIDTH == 8
cranges = NULL;
if (utf)
#endif
{
if (lengthptr != NULL)
{
cranges = compile_optimize_class(pptr, options, xoptions, cb);
if (cranges == NULL)
{
*errorcodeptr = ERR21;
return NULL;
}
/* Caching the pre-processed character ranges. */
if (cb->next_cranges != NULL)
cb->next_cranges->next = cranges;
else
cb->cranges = cranges;
cb->next_cranges = cranges;
}
else
{
/* Reuse the pre-processed character ranges. */
cranges = cb->cranges;
PCRE2_ASSERT(cranges != NULL);
cb->cranges = cranges->next;
}
if (cranges->range_list_size > 0)
{
const uint32_t *ranges = (const uint32_t*)(cranges + 1);
if (ranges[0] <= 255)
xclass_props |= XCLASS_HAS_8BIT_CHARS;
if (ranges[cranges->range_list_size - 1] == GET_MAX_CHAR_VALUE(utf) &&
ranges[cranges->range_list_size - 2] <= 256)
xclass_props |= XCLASS_HIGH_ANY;
}
}
class_uchardata = code + LINK_SIZE + 2; /* For XCLASS items */
#endif /* SUPPORT_WIDE_CHARS */
/* Initialize the 256-bit (32-byte) bit map to all zeros. We build the map
in a temporary bit of memory, in case the class contains fewer than two
8-bit characters because in that case the compiled code doesn't use the bit
map. */
memset(classbits, 0, 32);
/* Process items until end_ptr is reached. */
while (TRUE)
{
uint32_t meta = *(pptr++);
BOOL local_negate;
int posix_class;
int taboffset, tabopt;
class_bits_storage pbits;
uint32_t escape, c;
/* Handle POSIX classes such as [:alpha:] etc. */
switch (META_CODE(meta))
{
case META_POSIX:
case META_POSIX_NEG:
local_negate = (meta == META_POSIX_NEG);
posix_class = *(pptr++);
if (local_negate) should_flip_negation = TRUE; /* Note negative special */
/* If matching is caseless, upper and lower are converted to alpha.
This relies on the fact that the class table starts with alpha,
lower, upper as the first 3 entries. */
if ((options & PCRE2_CASELESS) != 0 && posix_class <= 2)
posix_class = 0;
/* When PCRE2_UCP is set, some of the POSIX classes are converted to
different escape sequences that use Unicode properties \p or \P.
Others that are not available via \p or \P have to generate
XCL_PROP/XCL_NOTPROP directly, which is done here. */
#ifdef SUPPORT_UNICODE
/* TODO This entire block of code here appears to be unreachable!? I simply
can't see how it can be hit, given that the frontend parser doesn't emit
META_POSIX for GRAPH/PRINT/PUNCT when UCP is set. */
if ((options & PCRE2_UCP) != 0 &&
(xoptions & PCRE2_EXTRA_ASCII_POSIX) == 0)
{
uint32_t ptype;
switch(posix_class)
{
case PC_GRAPH:
case PC_PRINT:
case PC_PUNCT:
ptype = (posix_class == PC_GRAPH)? PT_PXGRAPH :
(posix_class == PC_PRINT)? PT_PXPRINT : PT_PXPUNCT;
PRIV(update_classbits)(ptype, 0, local_negate, classbits);
if ((xclass_props & XCLASS_HIGH_ANY) == 0)
{
if (lengthptr != NULL)
*lengthptr += 3;
else
{
*class_uchardata++ = local_negate? XCL_NOTPROP : XCL_PROP;
*class_uchardata++ = (PCRE2_UCHAR)ptype;
*class_uchardata++ = 0;
}
xclass_props |= XCLASS_REQUIRED | XCLASS_HAS_PROPS;
}
continue;
/* For the other POSIX classes (ex: ascii) we are going to
fall through to the non-UCP case and build a bit map for
characters with code points less than 256. However, if we are in
a negated POSIX class, characters with code points greater than
255 must either all match or all not match, depending on whether
the whole class is not or is negated. For example, for
[[:^ascii:]... they must all match, whereas for [^[:^ascii:]...
they must not.
In the special case where there are no xclass items, this is
automatically handled by the use of OP_CLASS or OP_NCLASS, but an
explicit range is needed for OP_XCLASS. Setting a flag here
causes the range to be generated later when it is known that
OP_XCLASS is required. In the 8-bit library this is relevant only in
utf mode, since no wide characters can exist otherwise. */
default:
break;
}
}
#endif /* SUPPORT_UNICODE */
/* In the non-UCP case, or when UCP makes no difference, we build the
bit map for the POSIX class in a chunk of local store because we may
be adding and subtracting from it, and we don't want to subtract bits
that may be in the main map already. At the end we or the result into
the bit map that is being built. */
posix_class *= 3;
/* Copy in the first table (always present) */
memcpy(pbits.classbits, cbits + PRIV(posix_class_maps)[posix_class], 32);
/* If there is a second table, add or remove it as required. */
taboffset = PRIV(posix_class_maps)[posix_class + 1];
tabopt = PRIV(posix_class_maps)[posix_class + 2];
if (taboffset >= 0)
{
if (tabopt >= 0)
for (int i = 0; i < 32; i++)
pbits.classbits[i] |= cbits[i + taboffset];
else
for (int i = 0; i < 32; i++)
pbits.classbits[i] &= (uint8_t)(~cbits[i + taboffset]);
}
/* Now see if we need to remove any special characters. An option
value of 1 removes vertical space and 2 removes underscore. */
if (tabopt < 0) tabopt = -tabopt;
if (tabopt == 1) pbits.classbits[1] &= ~0x3c;
else if (tabopt == 2) pbits.classbits[11] &= 0x7f;
/* Add the POSIX table or its complement into the main table that is
being built and we are done. */
{
uint32_t *classwords = cb->classbits.classwords;
if (local_negate)
for (int i = 0; i < 8; i++)
classwords[i] |= (uint32_t)(~pbits.classwords[i]);
else
for (int i = 0; i < 8; i++)
classwords[i] |= pbits.classwords[i];
}
#ifdef SUPPORT_WIDE_CHARS
/* Every class contains at least one < 256 character. */
xclass_props |= XCLASS_HAS_8BIT_CHARS;
#endif
continue; /* End of POSIX handling */
/* Other than POSIX classes, the only items we should encounter are
\d-type escapes and literal characters (possibly as ranges). */
case META_BIGVALUE:
meta = *(pptr++);
break;
case META_ESCAPE:
escape = META_DATA(meta);
switch(escape)
{
case ESC_d:
for (int i = 0; i < 32; i++) classbits[i] |= cbits[i+cbit_digit];
break;
case ESC_D:
should_flip_negation = TRUE;
for (int i = 0; i < 32; i++)
classbits[i] |= (uint8_t)(~cbits[i+cbit_digit]);
break;
case ESC_w:
for (int i = 0; i < 32; i++) classbits[i] |= cbits[i+cbit_word];
break;
case ESC_W:
should_flip_negation = TRUE;
for (int i = 0; i < 32; i++)
classbits[i] |= (uint8_t)(~cbits[i+cbit_word]);
break;
/* Perl 5.004 onwards omitted VT from \s, but restored it at Perl
5.18. Before PCRE 8.34, we had to preserve the VT bit if it was
previously set by something earlier in the character class.
Luckily, the value of CHAR_VT is 0x0b in both ASCII and EBCDIC, so
we could just adjust the appropriate bit. From PCRE 8.34 we no
longer treat \s and \S specially. */
case ESC_s:
for (int i = 0; i < 32; i++) classbits[i] |= cbits[i+cbit_space];
break;
case ESC_S:
should_flip_negation = TRUE;
for (int i = 0; i < 32; i++)
classbits[i] |= (uint8_t)(~cbits[i+cbit_space]);
break;
/* When adding the horizontal or vertical space lists to a class, or
their complements, disable PCRE2_CASELESS, because it justs wastes
time, and in the "not-x" UTF cases can create unwanted duplicates in
the XCLASS list (provoked by characters that have more than one other
case and by both cases being in the same "not-x" sublist). */
case ESC_h:
#if PCRE2_CODE_UNIT_WIDTH == 8
#ifdef SUPPORT_UNICODE
if (cranges != NULL) break;
#endif
add_list_to_class(options & ~PCRE2_CASELESS, xoptions,
cb, PRIV(hspace_list));
#else
PCRE2_ASSERT(cranges != NULL);
#endif
break;
case ESC_H:
#if PCRE2_CODE_UNIT_WIDTH == 8
#ifdef SUPPORT_UNICODE
if (cranges != NULL) break;
#endif
add_not_list_to_class(options & ~PCRE2_CASELESS, xoptions,
cb, PRIV(hspace_list));
#else
PCRE2_ASSERT(cranges != NULL);
#endif
break;
case ESC_v:
#if PCRE2_CODE_UNIT_WIDTH == 8
#ifdef SUPPORT_UNICODE
if (cranges != NULL) break;
#endif
add_list_to_class(options & ~PCRE2_CASELESS, xoptions,
cb, PRIV(vspace_list));
#else
PCRE2_ASSERT(cranges != NULL);
#endif
break;
case ESC_V:
#if PCRE2_CODE_UNIT_WIDTH == 8
#ifdef SUPPORT_UNICODE
if (cranges != NULL) break;
#endif
add_not_list_to_class(options & ~PCRE2_CASELESS, xoptions,
cb, PRIV(vspace_list));
#else
PCRE2_ASSERT(cranges != NULL);
#endif
break;
/* If Unicode is not supported, \P and \p are not allowed and are
faulted at parse time, so will never appear here. */
#ifdef SUPPORT_UNICODE
case ESC_p:
case ESC_P:
{
uint32_t ptype = *pptr >> 16;
uint32_t pdata = *(pptr++) & 0xffff;
/* The "Any" is processed by PRIV(update_classbits)(). */
if (ptype == PT_ANY)
{
#if PCRE2_CODE_UNIT_WIDTH == 8
if (!utf && escape == ESC_p) memset(classbits, 0xff, 32);
#endif
continue;
}
PRIV(update_classbits)(ptype, pdata, (escape == ESC_P), classbits);
if ((xclass_props & XCLASS_HIGH_ANY) == 0)
{
if (lengthptr != NULL)
*lengthptr += 3;
else
{
*class_uchardata++ = (escape == ESC_p)? XCL_PROP : XCL_NOTPROP;
*class_uchardata++ = ptype;
*class_uchardata++ = pdata;
}
xclass_props |= XCLASS_REQUIRED | XCLASS_HAS_PROPS;
}
}
continue;
#endif
}
#ifdef SUPPORT_WIDE_CHARS
/* Every non-property class contains at least one < 256 character. */
xclass_props |= XCLASS_HAS_8BIT_CHARS;
#endif
/* End handling \d-type escapes */
continue;
CLASS_END_CASES(meta)
/* Literals. */
if (meta < META_END) break;
/* Non-literals: end of class contents. */
goto END_PROCESSING;
}
/* A literal character may be followed by a range meta. At parse time
there are checks for out-of-order characters, for ranges where the two
characters are equal, and for hyphens that cannot indicate a range. At
this point, therefore, no checking is needed. */
c = meta;
/* Remember if \r or \n were explicitly used */
if (c == CHAR_CR || c == CHAR_NL) cb->external_flags |= PCRE2_HASCRORLF;
/* Process a character range */
if (*pptr == META_RANGE_LITERAL || *pptr == META_RANGE_ESCAPED)
{
uint32_t d;
#ifdef EBCDIC
BOOL range_is_literal = (*pptr == META_RANGE_LITERAL);
#endif
++pptr;
d = *(pptr++);
if (d == META_BIGVALUE) d = *(pptr++);
/* Remember an explicit \r or \n, and add the range to the class. */
if (d == CHAR_CR || d == CHAR_NL) cb->external_flags |= PCRE2_HASCRORLF;
#if PCRE2_CODE_UNIT_WIDTH == 8
#ifdef SUPPORT_UNICODE
if (cranges != NULL) continue;
xclass_props |= XCLASS_HAS_8BIT_CHARS;
#endif
/* In an EBCDIC environment, Perl treats alphabetic ranges specially
because there are holes in the encoding, and simply using the range
A-Z (for example) would include the characters in the holes. This
applies only to literal ranges; [\xC1-\xE9] is different to [A-Z]. */
#ifdef EBCDIC
if (range_is_literal &&
(cb->ctypes[c] & ctype_letter) != 0 &&
(cb->ctypes[d] & ctype_letter) != 0 &&
(c <= CHAR_z) == (d <= CHAR_z))
{
uint32_t uc = (d <= CHAR_z)? 0 : 64;
uint32_t C = c - uc;
uint32_t D = d - uc;
if (C <= CHAR_i)
{
add_to_class(options, xoptions, cb, C + uc,
((D < CHAR_i)? D : CHAR_i) + uc);
C = CHAR_j;
}
if (C <= D && C <= CHAR_r)
{
add_to_class(options, xoptions, cb, C + uc,
((D < CHAR_r)? D : CHAR_r) + uc);
C = CHAR_s;
}
if (C <= D)
add_to_class(options, xoptions, cb, C + uc, D + uc);
}
else
#endif
/* Not an EBCDIC special range */
add_to_class(options, xoptions, cb, c, d);
#else
PCRE2_ASSERT(cranges != NULL);
#endif
continue;
} /* End of range handling */
/* Character ranges are ignored when class_ranges is present. */
#if PCRE2_CODE_UNIT_WIDTH == 8
#ifdef SUPPORT_UNICODE
if (cranges != NULL) continue;
xclass_props |= XCLASS_HAS_8BIT_CHARS;
#endif
/* Handle a single character. */
add_to_class(options, xoptions, cb, meta, meta);
#else
PCRE2_ASSERT(cranges != NULL);
#endif
} /* End of main class-processing loop */
END_PROCESSING:
#ifdef SUPPORT_WIDE_CHARS
PCRE2_ASSERT((xclass_props & XCLASS_HAS_PROPS) == 0 ||
(xclass_props & XCLASS_HIGH_ANY) == 0);
if (cranges != NULL)
{
uint32_t *range = (uint32_t*)(cranges + 1);
uint32_t *end = range + cranges->range_list_size;
while (range < end && range[0] < 256)
{
PCRE2_ASSERT((xclass_props & XCLASS_HAS_8BIT_CHARS) != 0);
/* Add range to bitset. If we are in UTF or UCP mode, then clear the
caseless bit, because the cranges handle caselessness (only) in this
condition; see the condition for PARSE_CLASS_CASELESS_UTF in
compile_optimize_class(). */
add_to_class(((options & (PCRE2_UTF|PCRE2_UCP)) != 0)?
(options & ~PCRE2_CASELESS) : options, xoptions, cb, range[0], range[1]);
if (range[1] > 255) break;
range += 2;
}
if (cranges->char_lists_size > 0)
{
/* The cranges structure is still used and freed later. */
PCRE2_ASSERT((xclass_props & XCLASS_HIGH_ANY) == 0);
xclass_props |= XCLASS_REQUIRED | XCLASS_HAS_CHAR_LISTS;
}
else
{
if ((xclass_props & XCLASS_HIGH_ANY) != 0)
{
PCRE2_ASSERT(range + 2 == end && range[0] <= 256 &&
range[1] >= GET_MAX_CHAR_VALUE(utf));
should_flip_negation = TRUE;
range = end;
}
while (range < end)
{
uint32_t range_start = range[0];
uint32_t range_end = range[1];
range += 2;
xclass_props |= XCLASS_REQUIRED;
if (range_start < 256) range_start = 256;
if (lengthptr != NULL)
{
#ifdef SUPPORT_UNICODE
if (utf)
{
*lengthptr += 1;
if (range_start < range_end)
*lengthptr += PRIV(ord2utf)(range_start, class_uchardata);
*lengthptr += PRIV(ord2utf)(range_end, class_uchardata);
continue;
}
#endif /* SUPPORT_UNICODE */
*lengthptr += range_start < range_end ? 3 : 2;
continue;
}
#ifdef SUPPORT_UNICODE
if (utf)
{
if (range_start < range_end)
{
*class_uchardata++ = XCL_RANGE;
class_uchardata += PRIV(ord2utf)(range_start, class_uchardata);
}
else
*class_uchardata++ = XCL_SINGLE;
class_uchardata += PRIV(ord2utf)(range_end, class_uchardata);
continue;
}
#endif /* SUPPORT_UNICODE */
/* Without UTF support, character values are constrained
by the bit length, and can only be > 256 for 16-bit and
32-bit libraries. */
#if PCRE2_CODE_UNIT_WIDTH != 8
if (range_start < range_end)
{
*class_uchardata++ = XCL_RANGE;
*class_uchardata++ = range_start;
}
else
*class_uchardata++ = XCL_SINGLE;
*class_uchardata++ = range_end;
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */
}
if (lengthptr == NULL)
cb->cx->memctl.free(cranges, cb->cx->memctl.memory_data);
}
}
#endif /* SUPPORT_WIDE_CHARS */
/* If there are characters with values > 255, or Unicode property settings
(\p or \P), we have to compile an extended class, with its own opcode,
unless there were no property settings and there was a negated special such
as \S in the class, and PCRE2_UCP is not set, because in that case all
characters > 255 are in or not in the class, so any that were explicitly
given as well can be ignored.
In the UCP case, if certain negated POSIX classes (ex: [:^ascii:]) were
were present in a class, we either have to match or not match all wide
characters (depending on whether the whole class is or is not negated).
This requirement is indicated by match_all_or_no_wide_chars being true.
We do this by including an explicit range, which works in both cases.
This applies only in UTF and 16-bit and 32-bit non-UTF modes, since there
cannot be any wide characters in 8-bit non-UTF mode.
When there *are* properties in a positive UTF-8 or any 16-bit or 32_bit
class where \S etc is present without PCRE2_UCP, causing an extended class
to be compiled, we make sure that all characters > 255 are included by
forcing match_all_or_no_wide_chars to be true.
If, when generating an xclass, there are no characters < 256, we can omit
the bitmap in the actual compiled code. */
#ifdef SUPPORT_WIDE_CHARS /* Defined for 16/32 bits, or 8-bit with Unicode */
if ((xclass_props & XCLASS_REQUIRED) != 0)
{
PCRE2_UCHAR *previous = code;
if ((xclass_props & XCLASS_HAS_CHAR_LISTS) == 0)
*class_uchardata++ = XCL_END; /* Marks the end of extra data */
*code++ = OP_XCLASS;
code += LINK_SIZE;
*code = negate_class? XCL_NOT:0;
if ((xclass_props & XCLASS_HAS_PROPS) != 0) *code |= XCL_HASPROP;
/* If the map is required, move up the extra data to make room for it;
otherwise just move the code pointer to the end of the extra data. */
if ((xclass_props & XCLASS_HAS_8BIT_CHARS) != 0 || has_bitmap != NULL)
{
if (negate_class)
{
uint32_t *classwords = cb->classbits.classwords;
for (int i = 0; i < 8; i++) classwords[i] = ~classwords[i];
}
if (has_bitmap == NULL)
{
*code++ |= XCL_MAP;
(void)memmove(code + (32 / sizeof(PCRE2_UCHAR)), code,
CU2BYTES(class_uchardata - code));
memcpy(code, classbits, 32);
code = class_uchardata + (32 / sizeof(PCRE2_UCHAR));
}
else
{
code = class_uchardata;
if ((xclass_props & XCLASS_HAS_8BIT_CHARS) != 0)
*has_bitmap = TRUE;
}
}
else code = class_uchardata;
if ((xclass_props & XCLASS_HAS_CHAR_LISTS) != 0)
{
/* Char lists size is an even number, because all items are 16 or 32
bit values. The character list data is always aligned to 32 bits. */
size_t char_lists_size = cranges->char_lists_size;
PCRE2_ASSERT((char_lists_size & 0x1) == 0 &&
(cb->char_lists_size & 0x3) == 0);
if (lengthptr != NULL)
{
char_lists_size = CLIST_ALIGN_TO(char_lists_size, sizeof(uint32_t));
#if PCRE2_CODE_UNIT_WIDTH == 8
*lengthptr += 2 + LINK_SIZE;
#else
*lengthptr += 1 + LINK_SIZE;
#endif
cb->char_lists_size += char_lists_size;
char_lists_size /= sizeof(PCRE2_UCHAR);
/* Storage space for character lists is included
in the maximum pattern size. */
if (*lengthptr > MAX_PATTERN_SIZE ||
MAX_PATTERN_SIZE - *lengthptr < char_lists_size)
{
*errorcodeptr = ERR20; /* Pattern is too large */
return NULL;
}
}
else
{
uint8_t *data;
PCRE2_ASSERT(cranges->char_lists_types <= XCL_TYPE_MASK);
#if PCRE2_CODE_UNIT_WIDTH == 8
/* Encode as high / low bytes. */
code[0] = (uint8_t)(XCL_LIST |
(cranges->char_lists_types >> 8));
code[1] = (uint8_t)cranges->char_lists_types;
code += 2;
#else
*code++ = (PCRE2_UCHAR)(XCL_LIST | cranges->char_lists_types);
#endif
/* Character lists are stored in backwards direction from
byte code start. The non-dfa/dfa matchers can access these
lists using the byte code start stored in match blocks.
Each list is aligned to 32 bit with an optional unused
16 bit value at the beginning of the character list. */
cb->char_lists_size += char_lists_size;
data = (uint8_t*)cb->start_code - cb->char_lists_size;
memcpy(data, (uint8_t*)(cranges + 1) + cranges->char_lists_start,
char_lists_size);
/* Since character lists total size is less than MAX_PATTERN_SIZE,
their starting offset fits into a value which size is LINK_SIZE. */
char_lists_size = cb->char_lists_size;
PUT(code, 0, (uint32_t)(char_lists_size >> 1));
code += LINK_SIZE;
#if defined PCRE2_DEBUG || defined SUPPORT_VALGRIND
if ((char_lists_size & 0x2) != 0)
{
/* In debug the unused 16 bit value is set
to a fixed value and marked unused. */
((uint16_t*)data)[-1] = 0x5555;
#ifdef SUPPORT_VALGRIND
VALGRIND_MAKE_MEM_NOACCESS(data - 2, 2);
#endif
}
#endif
cb->char_lists_size =
CLIST_ALIGN_TO(char_lists_size, sizeof(uint32_t));
cb->cx->memctl.free(cranges, cb->cx->memctl.memory_data);
}
}
/* Now fill in the complete length of the item */
PUT(previous, 1, (int)(code - previous));
goto DONE; /* End of class handling */
}
#endif /* SUPPORT_WIDE_CHARS */
/* If there are no characters > 255, or they are all to be included or
excluded, set the opcode to OP_CLASS or OP_NCLASS, depending on whether the
whole class was negated and whether there were negative specials such as \S
(non-UCP) in the class. Then copy the 32-byte map into the code vector,
negating it if necessary. */
if (negate_class)
{
uint32_t *classwords = cb->classbits.classwords;
for (int i = 0; i < 8; i++) classwords[i] = ~classwords[i];
}
if ((SELECT_VALUE8(!utf, 0) || negate_class != should_flip_negation) &&
cb->classbits.classwords[0] == ~(uint32_t)0)
{
const uint32_t *classwords = cb->classbits.classwords;
int i;
for (i = 0; i < 8; i++)
if (classwords[i] != ~(uint32_t)0) break;
if (i == 8)
{
*code++ = OP_ALLANY;
goto DONE; /* End of class handling */
}
}
*code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
memcpy(code, classbits, 32);
code += 32 / sizeof(PCRE2_UCHAR);
DONE:
*pcode = code;
return pptr - 1;
}
/* ===================================================================*/
/* Here follows a block of ECLASS-compiling functions. You may well want to
read them from top to bottom; they are ordered from leafmost (at the top) to
outermost parser (at the bottom of the file). */
/* This function folds one operand using the negation operator.
The new, combined chunk of stack code is written out to *pop_info. */
static void
fold_negation(eclass_op_info *pop_info, PCRE2_SIZE *lengthptr,
BOOL preserve_classbits)
{
/* If the chunk of stack code is already composed of multiple ops, we won't
descend in and try and propagate the negation down the tree. (That would lead
to O(n^2) compile-time, which could be exploitable with a malicious regex -
although maybe that's not really too much of a worry in a library that offers
an exponential-time matching function!) */
if (pop_info->op_single_type == 0)
{
if (lengthptr != NULL)
*lengthptr += 1;
else
pop_info->code_start[pop_info->length] = ECL_NOT;
pop_info->length += 1;
}
/* Otherwise, it's a nice single-op item, so we can easily fold in the negation
without needing to produce an ECL_NOT. */
else if (pop_info->op_single_type == ECL_ANY ||
pop_info->op_single_type == ECL_NONE)
{
pop_info->op_single_type = (pop_info->op_single_type == ECL_NONE)?
ECL_ANY : ECL_NONE;
if (lengthptr == NULL)
*(pop_info->code_start) = pop_info->op_single_type;
}
else
{
PCRE2_ASSERT(pop_info->op_single_type == ECL_XCLASS &&
pop_info->length >= 1 + LINK_SIZE + 1);
if (lengthptr == NULL)
pop_info->code_start[1 + LINK_SIZE] ^= XCL_NOT;
}
if (!preserve_classbits)
{
for (int i = 0; i < 8; i++)
pop_info->bits.classwords[i] = ~pop_info->bits.classwords[i];
}
}
/* This function folds together two operands using a binary operator.
The new, combined chunk of stack code is written out to *lhs_op_info. */
static void
fold_binary(int op, eclass_op_info *lhs_op_info, eclass_op_info *rhs_op_info,
PCRE2_SIZE *lengthptr)
{
switch (op)
{
/* ECL_AND truth table:
LHS RHS RESULT
----------------
ANY * RHS
* ANY LHS
NONE * NONE
* NONE NONE
X Y X & Y
*/
case ECL_AND:
if (rhs_op_info->op_single_type == ECL_ANY)
{
/* no-op: drop the RHS */
}
else if (lhs_op_info->op_single_type == ECL_ANY)
{
/* no-op: drop the LHS, and memmove the RHS into its place */
if (lengthptr == NULL)
memmove(lhs_op_info->code_start, rhs_op_info->code_start,
CU2BYTES(rhs_op_info->length));
lhs_op_info->length = rhs_op_info->length;
lhs_op_info->op_single_type = rhs_op_info->op_single_type;
}
else if (rhs_op_info->op_single_type == ECL_NONE)
{
/* the result is ECL_NONE: write into the LHS */
if (lengthptr == NULL)
lhs_op_info->code_start[0] = ECL_NONE;
lhs_op_info->length = 1;
lhs_op_info->op_single_type = ECL_NONE;
}
else if (lhs_op_info->op_single_type == ECL_NONE)
{
/* the result is ECL_NONE: drop the RHS */
}
else
{
/* Both of LHS & RHS are either ECL_XCLASS, or compound operations. */
if (lengthptr != NULL)
*lengthptr += 1;
else
{
PCRE2_ASSERT(rhs_op_info->code_start ==
lhs_op_info->code_start + lhs_op_info->length);
rhs_op_info->code_start[rhs_op_info->length] = ECL_AND;
}
lhs_op_info->length += rhs_op_info->length + 1;
lhs_op_info->op_single_type = 0;
}
for (int i = 0; i < 8; i++)
lhs_op_info->bits.classwords[i] &= rhs_op_info->bits.classwords[i];
break;
/* ECL_OR truth table:
LHS RHS RESULT
----------------
ANY * ANY
* ANY ANY
NONE * RHS
* NONE LHS
X Y X | Y
*/
case ECL_OR:
if (rhs_op_info->op_single_type == ECL_NONE)
{
/* no-op: drop the RHS */
}
else if (lhs_op_info->op_single_type == ECL_NONE)
{
/* no-op: drop the LHS, and memmove the RHS into its place */
if (lengthptr == NULL)
memmove(lhs_op_info->code_start, rhs_op_info->code_start,
CU2BYTES(rhs_op_info->length));
lhs_op_info->length = rhs_op_info->length;
lhs_op_info->op_single_type = rhs_op_info->op_single_type;
}
else if (rhs_op_info->op_single_type == ECL_ANY)
{
/* the result is ECL_ANY: write into the LHS */
if (lengthptr == NULL)
lhs_op_info->code_start[0] = ECL_ANY;
lhs_op_info->length = 1;
lhs_op_info->op_single_type = ECL_ANY;
}
else if (lhs_op_info->op_single_type == ECL_ANY)
{
/* the result is ECL_ANY: drop the RHS */
}
else
{
/* Both of LHS & RHS are either ECL_XCLASS, or compound operations. */
if (lengthptr != NULL)
*lengthptr += 1;
else
{
PCRE2_ASSERT(rhs_op_info->code_start ==
lhs_op_info->code_start + lhs_op_info->length);
rhs_op_info->code_start[rhs_op_info->length] = ECL_OR;
}
lhs_op_info->length += rhs_op_info->length + 1;
lhs_op_info->op_single_type = 0;
}
for (int i = 0; i < 8; i++)
lhs_op_info->bits.classwords[i] |= rhs_op_info->bits.classwords[i];
break;
/* ECL_XOR truth table:
LHS RHS RESULT
----------------
ANY * !RHS
* ANY !LHS
NONE * RHS
* NONE LHS
X Y X ^ Y
*/
case ECL_XOR:
if (rhs_op_info->op_single_type == ECL_NONE)
{
/* no-op: drop the RHS */
}
else if (lhs_op_info->op_single_type == ECL_NONE)
{
/* no-op: drop the LHS, and memmove the RHS into its place */
if (lengthptr == NULL)
memmove(lhs_op_info->code_start, rhs_op_info->code_start,
CU2BYTES(rhs_op_info->length));
lhs_op_info->length = rhs_op_info->length;
lhs_op_info->op_single_type = rhs_op_info->op_single_type;
}
else if (rhs_op_info->op_single_type == ECL_ANY)
{
/* the result is !LHS: fold in the negation, and drop the RHS */
/* Preserve the classbits, because we promise to deal with them later. */
fold_negation(lhs_op_info, lengthptr, TRUE);
}
else if (lhs_op_info->op_single_type == ECL_ANY)
{
/* the result is !RHS: drop the LHS, memmove the RHS into its place, and
fold in the negation */
if (lengthptr == NULL)
memmove(lhs_op_info->code_start, rhs_op_info->code_start,
CU2BYTES(rhs_op_info->length));
lhs_op_info->length = rhs_op_info->length;
lhs_op_info->op_single_type = rhs_op_info->op_single_type;
/* Preserve the classbits, because we promise to deal with them later. */
fold_negation(lhs_op_info, lengthptr, TRUE);
}
else
{
/* Both of LHS & RHS are either ECL_XCLASS, or compound operations. */
if (lengthptr != NULL)
*lengthptr += 1;
else
{
PCRE2_ASSERT(rhs_op_info->code_start ==
lhs_op_info->code_start + lhs_op_info->length);
rhs_op_info->code_start[rhs_op_info->length] = ECL_XOR;
}
lhs_op_info->length += rhs_op_info->length + 1;
lhs_op_info->op_single_type = 0;
}
for (int i = 0; i < 8; i++)
lhs_op_info->bits.classwords[i] ^= rhs_op_info->bits.classwords[i];
break;
default:
PCRE2_DEBUG_UNREACHABLE();
break;
}
}
static BOOL
compile_eclass_nested(eclass_context *context, BOOL negated,
uint32_t **pptr, PCRE2_UCHAR **pcode,
eclass_op_info *pop_info, PCRE2_SIZE *lengthptr);
/* This function consumes a group of implicitly-unioned class elements.
These can be characters, ranges, properties, or nested classes, as long
as they are all joined by being placed adjacently. */
static BOOL
compile_class_operand(eclass_context *context, BOOL negated,
uint32_t **pptr, PCRE2_UCHAR **pcode, eclass_op_info *pop_info,
PCRE2_SIZE *lengthptr)
{
uint32_t *ptr = *pptr;
uint32_t *prev_ptr;
PCRE2_UCHAR *code = *pcode;
PCRE2_UCHAR *code_start = code;
PCRE2_SIZE prev_length = (lengthptr != NULL)? *lengthptr : 0;
PCRE2_SIZE extra_length;
uint32_t meta = META_CODE(*ptr);
switch (meta)
{
case META_CLASS_EMPTY_NOT:
case META_CLASS_EMPTY:
++ptr;
pop_info->length = 1;
if ((meta == META_CLASS_EMPTY) == negated)
{
*code++ = pop_info->op_single_type = ECL_ANY;
memset(pop_info->bits.classbits, 0xff, 32);
}
else
{
*code++ = pop_info->op_single_type = ECL_NONE;
memset(pop_info->bits.classbits, 0, 32);
}
break;
case META_CLASS:
case META_CLASS_NOT:
if ((*ptr & CLASS_IS_ECLASS) != 0)
{
if (!compile_eclass_nested(context, negated, &ptr, &code,
pop_info, lengthptr))
return FALSE;
PCRE2_ASSERT(*ptr == META_CLASS_END);
ptr++;
goto DONE;
}
ptr++;
/* Fall through */
default:
/* Scan forward characters, ranges, and properties.
For example: inside [a-z_ -- m] we don't have brackets around "a-z_" but
we still need to collect that fragment up into a "leaf" OP_CLASS. */
prev_ptr = ptr;
ptr = PRIV(compile_class_not_nested)(
context->options, context->xoptions, ptr, &code,
(meta != META_CLASS_NOT) == negated, &context->needs_bitmap,
context->errorcodeptr, context->cb, lengthptr);
if (ptr == NULL) return FALSE;
/* We must have a 100% guarantee that ptr increases when
compile_class_operand() returns, even on Release builds, so that we can
statically prove our loops terminate. */
if (ptr <= prev_ptr)
{
PCRE2_DEBUG_UNREACHABLE();
return FALSE;
}
/* If we fell through above, consume the closing ']'. */
if (meta == META_CLASS || meta == META_CLASS_NOT)
{
PCRE2_ASSERT(*ptr == META_CLASS_END);
ptr++;
}
/* Regardless of whether (lengthptr == NULL), some data will still be written
out to *pcode, which we need: we have to peek at it, to transform the opcode
into the ECLASS version (since we need to hoist up the bitmaps). */
PCRE2_ASSERT(code > code_start);
extra_length = (lengthptr != NULL)? *lengthptr - prev_length : 0;
/* Easiest case: convert OP_ALLANY to ECL_ANY */
if (*code_start == OP_ALLANY)
{
PCRE2_ASSERT(code - code_start == 1 && extra_length == 0);
pop_info->length = 1;
*code_start = pop_info->op_single_type = ECL_ANY;
memset(pop_info->bits.classbits, 0xff, 32);
}
/* For OP_CLASS and OP_NCLASS, we hoist out the bitmap and convert to
ECL_NONE / ECL_ANY respectively. */
else if (*code_start == OP_CLASS || *code_start == OP_NCLASS)
{
PCRE2_ASSERT(code - code_start == 1 + 32 / sizeof(PCRE2_UCHAR) &&
extra_length == 0);
pop_info->length = 1;
*code_start = pop_info->op_single_type =
(*code_start == OP_CLASS)? ECL_NONE : ECL_ANY;
memcpy(pop_info->bits.classbits, code_start + 1, 32);
/* Rewind the code pointer, but make sure we adjust *lengthptr, because we
do need to reserve that space (even though we only use it temporarily). */
if (lengthptr != NULL)
*lengthptr += code - (code_start + 1);
code = code_start + 1;
if (!context->needs_bitmap && *code_start == ECL_NONE)
{
uint32_t *classwords = pop_info->bits.classwords;
for (int i = 0; i < 8; i++)
if (classwords[i] != 0)
{
context->needs_bitmap = TRUE;
break;
}
}
else
context->needs_bitmap = TRUE;
}
/* Finally, for OP_XCLASS we hoist out the bitmap (if any), and convert to
ECL_XCLASS. */
else
{
PCRE2_ASSERT(*code_start == OP_XCLASS);
*code_start = pop_info->op_single_type = ECL_XCLASS;
PCRE2_ASSERT(code - code_start >= 1 + LINK_SIZE + 1);
memcpy(pop_info->bits.classbits, context->cb->classbits.classbits, 32);
pop_info->length = (code - code_start) + extra_length;
}
break;
} /* End of switch(meta) */
pop_info->code_start = (lengthptr == NULL)? code_start : NULL;
if (lengthptr != NULL)
{
*lengthptr += code - code_start;
code = code_start;
}
DONE:
PCRE2_ASSERT(lengthptr == NULL || (code == code_start));
*pptr = ptr;
*pcode = code;
return TRUE;
}
/* This function consumes a group of implicitly-unioned class elements.
These can be characters, ranges, properties, or nested classes, as long
as they are all joined by being placed adjacently. */
static BOOL
compile_class_juxtaposition(eclass_context *context, BOOL negated,
uint32_t **pptr, PCRE2_UCHAR **pcode, eclass_op_info *pop_info,
PCRE2_SIZE *lengthptr)
{
uint32_t *ptr = *pptr;
PCRE2_UCHAR *code = *pcode;
#ifdef PCRE2_DEBUG
PCRE2_UCHAR *start_code = *pcode;
#endif
/* See compile_class_binary_loose() for comments on compile-time folding of
the "negated" flag. */
/* Because it's a non-empty class, there must be an operand at the start. */
if (!compile_class_operand(context, negated, &ptr, &code, pop_info, lengthptr))
return FALSE;
while (*ptr != META_CLASS_END &&
!(*ptr >= META_ECLASS_AND && *ptr <= META_ECLASS_NOT))
{
uint32_t op;
BOOL rhs_negated;
eclass_op_info rhs_op_info;
if (negated)
{
/* !(A juxtapose B) -> !A && !B */
op = ECL_AND;
rhs_negated = TRUE;
}
else
{
/* A juxtapose B -> A || B */
op = ECL_OR;
rhs_negated = FALSE;
}
/* An operand must follow the operator. */
if (!compile_class_operand(context, rhs_negated, &ptr, &code,
&rhs_op_info, lengthptr))
return FALSE;
/* Convert infix to postfix (RPN). */
fold_binary(op, pop_info, &rhs_op_info, lengthptr);
if (lengthptr == NULL)
code = pop_info->code_start + pop_info->length;
}
PCRE2_ASSERT(lengthptr == NULL || code == start_code);
*pptr = ptr;
*pcode = code;
return TRUE;
}
/* This function consumes unary prefix operators. */
static BOOL
compile_class_unary(eclass_context *context, BOOL negated,
uint32_t **pptr, PCRE2_UCHAR **pcode, eclass_op_info *pop_info,
PCRE2_SIZE *lengthptr)
{
uint32_t *ptr = *pptr;
#ifdef PCRE2_DEBUG
PCRE2_UCHAR *start_code = *pcode;
#endif
while (*ptr == META_ECLASS_NOT)
{
++ptr;
negated = !negated;
}
*pptr = ptr;
/* Because it's a non-empty class, there must be an operand. */
if (!compile_class_juxtaposition(context, negated, pptr, pcode,
pop_info, lengthptr))
return FALSE;
PCRE2_ASSERT(lengthptr == NULL || *pcode == start_code);
return TRUE;
}
/* This function consumes tightly-binding binary operators. */
static BOOL
compile_class_binary_tight(eclass_context *context, BOOL negated,
uint32_t **pptr, PCRE2_UCHAR **pcode, eclass_op_info *pop_info,
PCRE2_SIZE *lengthptr)
{
uint32_t *ptr = *pptr;
PCRE2_UCHAR *code = *pcode;
#ifdef PCRE2_DEBUG
PCRE2_UCHAR *start_code = *pcode;
#endif
/* See compile_class_binary_loose() for comments on compile-time folding of
the "negated" flag. */
/* Because it's a non-empty class, there must be an operand at the start. */
if (!compile_class_unary(context, negated, &ptr, &code, pop_info, lengthptr))
return FALSE;
while (*ptr == META_ECLASS_AND)
{
uint32_t op;
BOOL rhs_negated;
eclass_op_info rhs_op_info;
if (negated)
{
/* !(A && B) -> !A || !B */
op = ECL_OR;
rhs_negated = TRUE;
}
else
{
/* A && B -> A && B */
op = ECL_AND;
rhs_negated = FALSE;
}
++ptr;
/* An operand must follow the operator. */
if (!compile_class_unary(context, rhs_negated, &ptr, &code,
&rhs_op_info, lengthptr))
return FALSE;
/* Convert infix to postfix (RPN). */
fold_binary(op, pop_info, &rhs_op_info, lengthptr);
if (lengthptr == NULL)
code = pop_info->code_start + pop_info->length;
}
PCRE2_ASSERT(lengthptr == NULL || code == start_code);
*pptr = ptr;
*pcode = code;
return TRUE;
}
/* This function consumes loosely-binding binary operators. */
static BOOL
compile_class_binary_loose(eclass_context *context, BOOL negated,
uint32_t **pptr, PCRE2_UCHAR **pcode, eclass_op_info *pop_info,
PCRE2_SIZE *lengthptr)
{
uint32_t *ptr = *pptr;
PCRE2_UCHAR *code = *pcode;
#ifdef PCRE2_DEBUG
PCRE2_UCHAR *start_code = *pcode;
#endif
/* We really want to fold the negation operator, if at all possible, so that
simple cases can be reduced down. In particular, in 8-bit no-UTF mode, we want
to produce a fully-folded expression, so that we can guarantee not to emit any
OP_ECLASS codes (in the same way that we never emit OP_XCLASS in this mode).
This has the consequence that with a little ingenuity, we can in fact avoid
emitting (nearly...) all cases of the "NOT" operator. Imagine that we have:
!(A ...
We have parsed the preceding "!", and we are about to parse the "A" operand. We
don't know yet whether there will even be a following binary operand! Both of
these are possibilities for what follows:
!(A && B)
!(A)
However, we can still fold the "!" into the "A" operand, because no matter what
the following binary operator will be, we can produce an expression which is
equivalent. */
/* Because it's a non-empty class, there must be an operand at the start. */
if (!compile_class_binary_tight(context, negated, &ptr, &code,
pop_info, lengthptr))
return FALSE;
while (*ptr >= META_ECLASS_OR && *ptr <= META_ECLASS_XOR)
{
uint32_t op;
BOOL op_neg;
BOOL rhs_negated;
eclass_op_info rhs_op_info;
if (negated)
{
/* The whole expression is being negated; we respond by unconditionally
negating the LHS A, before seeing what follows. And hooray! We can recover,
no matter what follows. */
/* !(A || B) -> !A && !B */
/* !(A -- B) -> !(A && !B) -> !A || B */
/* !(A XOR B) -> !(!A XOR !B) -> !A XNOR !B */
op = (*ptr == META_ECLASS_OR )? ECL_AND :
(*ptr == META_ECLASS_SUB)? ECL_OR :
/*ptr == META_ECLASS_XOR*/ ECL_XOR;
op_neg = (*ptr == META_ECLASS_XOR);
rhs_negated = *ptr != META_ECLASS_SUB;
}
else
{
/* A || B -> A || B */
/* A -- B -> A && !B */
/* A XOR B -> A XOR B */
op = (*ptr == META_ECLASS_OR )? ECL_OR :
(*ptr == META_ECLASS_SUB)? ECL_AND :
/*ptr == META_ECLASS_XOR*/ ECL_XOR;
op_neg = FALSE;
rhs_negated = *ptr == META_ECLASS_SUB;
}
++ptr;
/* An operand must follow the operator. */
if (!compile_class_binary_tight(context, rhs_negated, &ptr, &code,
&rhs_op_info, lengthptr))
return FALSE;
/* Convert infix to postfix (RPN). */
fold_binary(op, pop_info, &rhs_op_info, lengthptr);
if (op_neg) fold_negation(pop_info, lengthptr, FALSE);
if (lengthptr == NULL)
code = pop_info->code_start + pop_info->length;
}
PCRE2_ASSERT(lengthptr == NULL || code == start_code);
*pptr = ptr;
*pcode = code;
return TRUE;
}
/* This function converts the META codes in pptr into opcodes written to
pcode. The pptr must start at a META_CLASS or META_CLASS_NOT.
The class is compiled as a left-associative sequence of operator
applications.
The pptr will be left pointing at the matching META_CLASS_END. */
static BOOL
compile_eclass_nested(eclass_context *context, BOOL negated,
uint32_t **pptr, PCRE2_UCHAR **pcode,
eclass_op_info *pop_info, PCRE2_SIZE *lengthptr)
{
uint32_t *ptr = *pptr;
#ifdef PCRE2_DEBUG
PCRE2_UCHAR *start_code = *pcode;
#endif
/* The CLASS_IS_ECLASS bit must be set since it is a nested class. */
PCRE2_ASSERT(*ptr == (META_CLASS | CLASS_IS_ECLASS) ||
*ptr == (META_CLASS_NOT | CLASS_IS_ECLASS));
if (*ptr++ == (META_CLASS_NOT | CLASS_IS_ECLASS))
negated = !negated;
(*pptr)++;
/* Because it's a non-empty class, there must be an operand at the start. */
if (!compile_class_binary_loose(context, negated, pptr, pcode,
pop_info, lengthptr))
return FALSE;
PCRE2_ASSERT(**pptr == META_CLASS_END);
PCRE2_ASSERT(lengthptr == NULL || *pcode == start_code);
return TRUE;
}
BOOL
PRIV(compile_class_nested)(uint32_t options, uint32_t xoptions,
uint32_t **pptr, PCRE2_UCHAR **pcode, int *errorcodeptr,
compile_block *cb, PCRE2_SIZE *lengthptr)
{
eclass_context context;
eclass_op_info op_info;
PCRE2_SIZE previous_length = (lengthptr != NULL)? *lengthptr : 0;
PCRE2_UCHAR *code = *pcode;
PCRE2_UCHAR *previous;
BOOL allbitsone = TRUE;
context.needs_bitmap = FALSE;
context.options = options;
context.xoptions = xoptions;
context.errorcodeptr = errorcodeptr;
context.cb = cb;
previous = code;
*code++ = OP_ECLASS;
code += LINK_SIZE;
*code++ = 0; /* Flags, currently zero. */
if (!compile_eclass_nested(&context, FALSE, pptr, &code, &op_info, lengthptr))
return FALSE;
if (lengthptr != NULL)
{
*lengthptr += code - previous;
code = previous;
/* (*lengthptr - previous_length) now holds the amount of buffer that
we require to make the call to compile_class_nested() with
lengthptr = NULL, and including the (1+LINK_SIZE+1) that we write out
before that call. */
}
/* Do some useful counting of what's in the bitmap. */
for (int i = 0; i < 8; i++)
if (op_info.bits.classwords[i] != 0xffffffff)
{
allbitsone = FALSE;
break;
}
/* After constant-folding the extended class syntax, it may turn out to be
a simple class after all. In that case, we can unwrap it from the
OP_ECLASS container - and in fact, we must do so, because in 8-bit
no-Unicode mode the matcher is compiled without support for OP_ECLASS. */
#ifndef SUPPORT_WIDE_CHARS
PCRE2_ASSERT(op_info.op_single_type != 0);
#else
if (op_info.op_single_type != 0)
#endif
{
/* Rewind back over the OP_ECLASS. */
code = previous;
/* If the bits are all ones, and the "high characters" are all matched
too, we use a special-cased encoding of OP_ALLANY. */
if (op_info.op_single_type == ECL_ANY && allbitsone)
{
/* Advancing code means rewinding lengthptr, at this point. */
if (lengthptr != NULL) *lengthptr -= 1;
*code++ = OP_ALLANY;
}
/* If the high bits are all matched / all not-matched, then we emit an
OP_NCLASS/OP_CLASS respectively. */
else if (op_info.op_single_type == ECL_ANY ||
op_info.op_single_type == ECL_NONE)
{
PCRE2_SIZE required_len = 1 + (32 / sizeof(PCRE2_UCHAR));
if (lengthptr != NULL)
{
if (required_len > (*lengthptr - previous_length))
*lengthptr = previous_length + required_len;
}
/* Advancing code means rewinding lengthptr, at this point. */
if (lengthptr != NULL) *lengthptr -= required_len;
*code++ = (op_info.op_single_type == ECL_ANY)? OP_NCLASS : OP_CLASS;
memcpy(code, op_info.bits.classbits, 32);
code += 32 / sizeof(PCRE2_UCHAR);
}
/* Otherwise, we have an ECL_XCLASS, so we have the OP_XCLASS data
there, but, we pulled out its bitmap into op_info, so now we have to
put that back into the OP_XCLASS. */
else
{
#ifndef SUPPORT_WIDE_CHARS
PCRE2_DEBUG_UNREACHABLE();
#else
BOOL need_map = context.needs_bitmap;
PCRE2_SIZE required_len;
PCRE2_ASSERT(op_info.op_single_type == ECL_XCLASS);
required_len = op_info.length + (need_map? 32/sizeof(PCRE2_UCHAR) : 0);
if (lengthptr != NULL)
{
/* Don't unconditionally request all the space we need - we may
already have asked for more during processing of the ECLASS. */
if (required_len > (*lengthptr - previous_length))
*lengthptr = previous_length + required_len;
/* The code we write out here won't be ignored, even during the
(lengthptr != NULL) phase, because if there's a following quantifier
it will peek backwards. So we do have to write out a (truncated)
OP_XCLASS, even on this branch. */
*lengthptr -= 1 + LINK_SIZE + 1;
*code++ = OP_XCLASS;
PUT(code, 0, 1 + LINK_SIZE + 1);
code += LINK_SIZE;
*code++ = 0;
}
else
{
PCRE2_UCHAR *rest;
PCRE2_SIZE rest_len;
PCRE2_UCHAR flags;
/* 1 unit: OP_XCLASS | LINK_SIZE units | 1 unit: flags | ...rest */
PCRE2_ASSERT(op_info.length >= 1 + LINK_SIZE + 1);
rest = op_info.code_start + 1 + LINK_SIZE + 1;
rest_len = (op_info.code_start + op_info.length) - rest;
/* First read any data we use, before memmove splats it. */
flags = op_info.code_start[1 + LINK_SIZE];
PCRE2_ASSERT((flags & XCL_MAP) == 0);
/* Next do the memmove before any writes. */
memmove(code + 1 + LINK_SIZE + 1 + (need_map? 32/sizeof(PCRE2_UCHAR) : 0),
rest, CU2BYTES(rest_len));
/* Finally write the header data. */
*code++ = OP_XCLASS;
PUT(code, 0, (int)required_len);
code += LINK_SIZE;
*code++ = flags | (need_map? XCL_MAP : 0);
if (need_map)
{
memcpy(code, op_info.bits.classbits, 32);
code += 32 / sizeof(PCRE2_UCHAR);
}
code += rest_len;
}
#endif /* SUPPORT_WIDE_CHARS */
}
}
/* Otherwise, we're going to keep the OP_ECLASS. However, again we need
to do some adjustment to insert the bitmap if we have one. */
#ifdef SUPPORT_WIDE_CHARS
else
{
BOOL need_map = context.needs_bitmap;
PCRE2_SIZE required_len =
1 + LINK_SIZE + 1 + (need_map? 32/sizeof(PCRE2_UCHAR) : 0) + op_info.length;
if (lengthptr != NULL)
{
if (required_len > (*lengthptr - previous_length))
*lengthptr = previous_length + required_len;
/* As for the XCLASS branch above, we do have to write out a dummy
OP_ECLASS, because of the backwards peek by the quantifier code. Write
out a (truncated) OP_ECLASS, even on this branch. */
*lengthptr -= 1 + LINK_SIZE + 1;
*code++ = OP_ECLASS;
PUT(code, 0, 1 + LINK_SIZE + 1);
code += LINK_SIZE;
*code++ = 0;
}
else
{
if (need_map)
{
PCRE2_UCHAR *map_start = previous + 1 + LINK_SIZE + 1;
previous[1 + LINK_SIZE] |= ECL_MAP;
memmove(map_start + 32/sizeof(PCRE2_UCHAR), map_start,
CU2BYTES(code - map_start));
memcpy(map_start, op_info.bits.classbits, 32);
code += 32 / sizeof(PCRE2_UCHAR);
}
PUT(previous, 1, (int)(code - previous));
}
}
#endif /* SUPPORT_WIDE_CHARS */
*pcode = code;
return TRUE;
}
/* End of pcre2_compile_class.c */
pcre2-10.45/src/pcre2posix.h 0000644 0001751 0000166 00000016273 14750415257 011301 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE2 is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language. This is
the public header file to be #included by applications that call PCRE2 via the
POSIX wrapper interface.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2023 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
#ifndef PCRE2POSIX_H_IDEMPOTENT_GUARD
#define PCRE2POSIX_H_IDEMPOTENT_GUARD
/* Have to include stdlib.h in order to ensure that size_t is defined. */
#include
/* Allow for C++ users */
#ifdef __cplusplus
extern "C" {
#endif
/* Options, mostly defined by POSIX, but with some extras. */
#define REG_ICASE 0x0001 /* Maps to PCRE2_CASELESS */
#define REG_NEWLINE 0x0002 /* Maps to PCRE2_MULTILINE */
#define REG_NOTBOL 0x0004 /* Maps to PCRE2_NOTBOL */
#define REG_NOTEOL 0x0008 /* Maps to PCRE2_NOTEOL */
#define REG_DOTALL 0x0010 /* NOT defined by POSIX; maps to PCRE2_DOTALL */
#define REG_NOSUB 0x0020 /* Do not report what was matched */
#define REG_UTF 0x0040 /* NOT defined by POSIX; maps to PCRE2_UTF */
#define REG_STARTEND 0x0080 /* BSD feature: pass subject string by so,eo */
#define REG_NOTEMPTY 0x0100 /* NOT defined by POSIX; maps to PCRE2_NOTEMPTY */
#define REG_UNGREEDY 0x0200 /* NOT defined by POSIX; maps to PCRE2_UNGREEDY */
#define REG_UCP 0x0400 /* NOT defined by POSIX; maps to PCRE2_UCP */
#define REG_PEND 0x0800 /* GNU feature: pass end pattern by re_endp */
#define REG_NOSPEC 0x1000 /* Maps to PCRE2_LITERAL */
/* This is not used by PCRE2, but by defining it we make it easier
to slot PCRE2 into existing programs that make POSIX calls. */
#define REG_EXTENDED 0
/* Error values. Not all these are relevant or used by the wrapper. */
enum {
REG_ASSERT = 1, /* internal error ? */
REG_BADBR, /* invalid repeat counts in {} */
REG_BADPAT, /* pattern error */
REG_BADRPT, /* ? * + invalid */
REG_EBRACE, /* unbalanced {} */
REG_EBRACK, /* unbalanced [] */
REG_ECOLLATE, /* collation error - not relevant */
REG_ECTYPE, /* bad class */
REG_EESCAPE, /* bad escape sequence */
REG_EMPTY, /* empty expression */
REG_EPAREN, /* unbalanced () */
REG_ERANGE, /* bad range inside [] */
REG_ESIZE, /* expression too big */
REG_ESPACE, /* failed to get memory */
REG_ESUBREG, /* bad back reference */
REG_INVARG, /* bad argument */
REG_NOMATCH /* match failed */
};
/* The structure representing a compiled regular expression. It is also used
for passing the pattern end pointer when REG_PEND is set. */
typedef struct {
void *re_pcre2_code;
void *re_match_data;
const char *re_endp;
size_t re_nsub;
size_t re_erroffset;
int re_cflags;
} regex_t;
/* The structure in which a captured offset is returned. */
typedef int regoff_t;
typedef struct {
regoff_t rm_so;
regoff_t rm_eo;
} regmatch_t;
/* When compiling with the MSVC compiler, it is sometimes necessary to include
a "calling convention" before exported function names. (This is secondhand
information; I know nothing about MSVC myself). For example, something like
void __cdecl function(....)
might be needed. In order to make this easy, all the exported functions have
PCRE2_CALL_CONVENTION just before their names. It is rarely needed; if not
set, we ensure here that it has no effect. */
#ifndef PCRE2_CALL_CONVENTION
#define PCRE2_CALL_CONVENTION
#endif
#ifndef PCRE2_EXPORT
#define PCRE2_EXPORT
#endif
/* When an application links to a PCRE2 DLL in Windows, the symbols that are
imported have to be identified as such. When building PCRE2, the appropriate
export settings are needed, and are set in pcre2posix.c before including this
file. */
/* By default, we use the standard "extern" declarations. */
#ifndef PCRE2POSIX_EXP_DECL
# if defined(_WIN32) && defined(PCRE2POSIX_SHARED) && !defined(PCRE2_STATIC)
# define PCRE2POSIX_EXP_DECL extern __declspec(dllimport)
# define PCRE2POSIX_EXP_DEFN __declspec(dllimport)
# else
# define PCRE2POSIX_EXP_DECL extern PCRE2_EXPORT
# define PCRE2POSIX_EXP_DEFN
# endif
#endif
/* The functions. The actual code is in functions with pcre2_xxx names for
uniqueness. POSIX names are provided as macros for API compatibility with POSIX
regex functions. It's done this way to ensure to they are always linked from
the PCRE2 library and not by accident from elsewhere (regex_t differs in size
elsewhere). */
PCRE2POSIX_EXP_DECL int PCRE2_CALL_CONVENTION pcre2_regcomp(regex_t *, const char *, int);
PCRE2POSIX_EXP_DECL int PCRE2_CALL_CONVENTION pcre2_regexec(const regex_t *, const char *, size_t,
regmatch_t *, int);
PCRE2POSIX_EXP_DECL size_t PCRE2_CALL_CONVENTION pcre2_regerror(int, const regex_t *, char *, size_t);
PCRE2POSIX_EXP_DECL void PCRE2_CALL_CONVENTION pcre2_regfree(regex_t *);
#define regcomp pcre2_regcomp
#define regexec pcre2_regexec
#define regerror pcre2_regerror
#define regfree pcre2_regfree
/* Debian had a patch that used different names. These are now here to save
them having to maintain their own patch, but are not documented by PCRE2. */
#define PCRE2regcomp pcre2_regcomp
#define PCRE2regexec pcre2_regexec
#define PCRE2regerror pcre2_regerror
#define PCRE2regfree pcre2_regfree
#ifdef __cplusplus
} /* extern "C" */
#endif
#endif /* PCRE2POSIX_H_IDEMPOTENT_GUARD */
/* End of pcre2posix.h */
pcre2-10.45/src/pcre2_extuni.c 0000644 0001751 0000166 00000012463 14750415257 011602 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This module contains an internal function that is used to match a Unicode
extended grapheme sequence. It is used by both pcre2_match() and
pcre2_dfa_match(). However, it is called only when Unicode support is being
compiled. Nevertheless, we provide a dummy function when there is no Unicode
support, because some compilers do not like functionless source files. */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "pcre2_internal.h"
/* Dummy function */
#ifndef SUPPORT_UNICODE
PCRE2_SPTR
PRIV(extuni)(uint32_t c, PCRE2_SPTR eptr, PCRE2_SPTR start_subject,
PCRE2_SPTR end_subject, BOOL utf, int *xcount)
{
(void)c;
(void)eptr;
(void)start_subject;
(void)end_subject;
(void)utf;
(void)xcount;
return NULL;
}
#else
/*************************************************
* Match an extended grapheme sequence *
*************************************************/
/* NOTE: The logic contained in this function is replicated in three special-
purpose functions in the pcre2_jit_compile.c module. If the logic below is
changed, they must be kept in step so that the interpreter and the JIT have the
same behaviour.
Arguments:
c the first character
eptr pointer to next character
start_subject pointer to start of subject
end_subject pointer to end of subject
utf TRUE if in UTF mode
xcount pointer to count of additional characters,
or NULL if count not needed
Returns: pointer after the end of the sequence
*/
PCRE2_SPTR
PRIV(extuni)(uint32_t c, PCRE2_SPTR eptr, PCRE2_SPTR start_subject,
PCRE2_SPTR end_subject, BOOL utf, int *xcount)
{
BOOL was_ep_ZWJ = FALSE;
int lgb = UCD_GRAPHBREAK(c);
while (eptr < end_subject)
{
int rgb;
int len = 1;
if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); }
rgb = UCD_GRAPHBREAK(c);
if ((PRIV(ucp_gbtable)[lgb] & (1u << rgb)) == 0) break;
/* ZWJ followed by Extended Pictographic is allowed only if the ZWJ was
preceded by Extended Pictographic. */
if (lgb == ucp_gbZWJ && rgb == ucp_gbExtended_Pictographic && !was_ep_ZWJ)
break;
/* Not breaking between Regional Indicators is allowed only if there
are an even number of preceding RIs. */
if (lgb == ucp_gbRegional_Indicator && rgb == ucp_gbRegional_Indicator)
{
int ricount = 0;
PCRE2_SPTR bptr = eptr - 1;
if (utf) BACKCHAR(bptr);
/* bptr is pointing to the left-hand character */
while (bptr > start_subject)
{
bptr--;
if (utf)
{
BACKCHAR(bptr);
GETCHAR(c, bptr);
}
else
c = *bptr;
if (UCD_GRAPHBREAK(c) != ucp_gbRegional_Indicator) break;
ricount++;
}
if ((ricount & 1) != 0) break; /* Grapheme break required */
}
/* Set a flag when ZWJ follows Extended Pictographic (with optional Extend in
between; see next statement). */
was_ep_ZWJ = (lgb == ucp_gbExtended_Pictographic && rgb == ucp_gbZWJ);
/* If Extend follows Extended_Pictographic, do not update lgb; this allows
any number of them before a following ZWJ. */
if (rgb != ucp_gbExtend || lgb != ucp_gbExtended_Pictographic) lgb = rgb;
eptr += len;
if (xcount != NULL) *xcount += 1;
}
return eptr;
}
#endif /* SUPPORT_UNICODE */
/* End of pcre2_extuni.c */
pcre2-10.45/src/pcre2_find_bracket.c 0000644 0001751 0000166 00000015376 14750415257 012707 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This module contains a single function that scans through a compiled pattern
until it finds a capturing bracket with the given number, or, if the number is
negative, an instance of OP_REVERSE or OP_VREVERSE for a lookbehind. The
function is called from pcre2_compile.c and also from pcre2_study.c when
finding the minimum matching length. */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "pcre2_internal.h"
/*************************************************
* Scan compiled regex for specific bracket *
*************************************************/
/*
Arguments:
code points to start of expression
utf TRUE in UTF mode
number the required bracket number or negative to find a lookbehind
Returns: pointer to the opcode for the bracket, or NULL if not found
*/
PCRE2_SPTR
PRIV(find_bracket)(PCRE2_SPTR code, BOOL utf, int number)
{
for (;;)
{
PCRE2_UCHAR c = *code;
if (c == OP_END) return NULL;
/* XCLASS is used for classes that cannot be represented just by a bit map.
This includes negated single high-valued characters. ECLASS is used for
classes that use set operations internally. CALLOUT_STR is used for
callouts with string arguments. In each case the length in the table is
zero; the actual length is stored in the compiled code. */
if (c == OP_XCLASS || c == OP_ECLASS) code += GET(code, 1);
else if (c == OP_CALLOUT_STR) code += GET(code, 1 + 2*LINK_SIZE);
/* Handle lookbehind */
else if (c == OP_REVERSE || c == OP_VREVERSE)
{
if (number < 0) return code;
code += PRIV(OP_lengths)[c];
}
/* Handle capturing bracket */
else if (c == OP_CBRA || c == OP_SCBRA ||
c == OP_CBRAPOS || c == OP_SCBRAPOS)
{
int n = (int)GET2(code, 1+LINK_SIZE);
if (n == number) return code;
code += PRIV(OP_lengths)[c];
}
/* Otherwise, we can get the item's length from the table, except that for
repeated character types, we have to test for \p and \P, which have an extra
two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we
must add in its length. */
else
{
switch(c)
{
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSPLUS:
case OP_TYPEPOSQUERY:
if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
break;
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEEXACT:
case OP_TYPEPOSUPTO:
if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
code += 2;
break;
case OP_MARK:
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_SKIP_ARG:
case OP_THEN_ARG:
code += code[1];
break;
}
/* Add in the fixed length from the table */
code += PRIV(OP_lengths)[c];
/* In UTF-8 and UTF-16 modes, opcodes that are followed by a character may be
followed by a multi-byte character. The length in the table is a minimum, so
we have to arrange to skip the extra bytes. */
#ifdef MAYBE_UTF_MULTI
if (utf) switch(c)
{
case OP_CHAR:
case OP_CHARI:
case OP_NOT:
case OP_NOTI:
case OP_EXACT:
case OP_EXACTI:
case OP_NOTEXACT:
case OP_NOTEXACTI:
case OP_UPTO:
case OP_UPTOI:
case OP_NOTUPTO:
case OP_NOTUPTOI:
case OP_MINUPTO:
case OP_MINUPTOI:
case OP_NOTMINUPTO:
case OP_NOTMINUPTOI:
case OP_POSUPTO:
case OP_POSUPTOI:
case OP_NOTPOSUPTO:
case OP_NOTPOSUPTOI:
case OP_STAR:
case OP_STARI:
case OP_NOTSTAR:
case OP_NOTSTARI:
case OP_MINSTAR:
case OP_MINSTARI:
case OP_NOTMINSTAR:
case OP_NOTMINSTARI:
case OP_POSSTAR:
case OP_POSSTARI:
case OP_NOTPOSSTAR:
case OP_NOTPOSSTARI:
case OP_PLUS:
case OP_PLUSI:
case OP_NOTPLUS:
case OP_NOTPLUSI:
case OP_MINPLUS:
case OP_MINPLUSI:
case OP_NOTMINPLUS:
case OP_NOTMINPLUSI:
case OP_POSPLUS:
case OP_POSPLUSI:
case OP_NOTPOSPLUS:
case OP_NOTPOSPLUSI:
case OP_QUERY:
case OP_QUERYI:
case OP_NOTQUERY:
case OP_NOTQUERYI:
case OP_MINQUERY:
case OP_MINQUERYI:
case OP_NOTMINQUERY:
case OP_NOTMINQUERYI:
case OP_POSQUERY:
case OP_POSQUERYI:
case OP_NOTPOSQUERY:
case OP_NOTPOSQUERYI:
if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
break;
}
#else
(void)(utf); /* Keep compiler happy by referencing function argument */
#endif /* MAYBE_UTF_MULTI */
}
}
}
/* End of pcre2_find_bracket.c */
pcre2-10.45/src/pcre2posix_test.c 0000644 0001751 0000166 00000012663 14750415257 012332 /*************************************************
* PCRE2 POSIX interface test program *
*************************************************/
/*
Written by Philip Hazel, December 2022
Copyright (c) 2022
File last edited: December 2022
This program tests the POSIX wrapper to the PCRE2 regular expression library.
The main PCRE2 test program is pcre2test, which also tests these function
calls. This little program is needed to test the case where the client includes
pcre2posix.h but not pcre2.h, mainly to make sure that it builds successfully.
However, the code is written as a flexible test program to which extra tests
can be added.
Compile with -lpcre2-posix -lpcre2-8
If run with no options, there is no output on success, and the return code is
zero. If any test fails there is output to stderr, and the return code is 1.
For testing purposes, the "-v" option causes verification output to be written
to stdout. */
#include
#include
#include
#define CAPCOUNT 5 /* Number of captures supported */
#define PRINTF if (v) printf /* Shorthand for testing output */
/* This vector contains compiler flags for each pattern that is tested. */
static int cflags[] = {
0, /* Test 0 */
REG_ICASE, /* Test 1 */
0, /* Test 2 */
REG_NEWLINE, /* Test 3 */
0 /* Test 4 */
};
/* This vector contains match flags for each pattern that is tested. */
static int mflags[] = {
0, /* Test 0 */
0, /* Test 1 */
0, /* Test 2 */
REG_NOTBOL, /* Test 3 */
0 /* Test 4 */
};
/* Automate the number of patterns */
#define count (int)(sizeof(cflags)/sizeof(int))
/* The data for each pattern consists of a pattern string, followed by any
number of subject strings, terminated by NULL. Some tests share data, but use
different flags. */
static const char *data0_1[] = { "posix", "lower posix", "upper POSIX", NULL };
static const char *data2_3[] = { "(*LF)^(cat|dog)", "catastrophic\ncataclysm",
"dogfight", "no animals", NULL };
static const char *data4[] = { "*badpattern", NULL };
/* Index the data strings */
static char **data[] = {
(char **)(&data0_1),
(char **)(&data0_1),
(char **)(&data2_3),
(char **)(&data2_3),
(char **)(&data4)
};
/* The expected results for each pattern consist of a compiler return code,
optionally followed, for each subject string, by a match return code and, for a
successful match, up to CAPCOUNT pairs of returned match data. */
static int results0[] = {
0, /* Compiler rc */
0, 6, 11, /* 1st match */
REG_NOMATCH /* 2nd match */
};
static int results1[] = {
0, /* Compiler rc */
0, 6, 11, /* 1st match */
0, 6, 11 /* 2nd match */
};
static int results2[] = {
0, /* Compiler rc */
0, 0, 3, 0, 3, /* 1st match */
0, 0, 3, 0, 3, /* 2nd match */
REG_NOMATCH /* 3rd match */
};
static int results3[] = {
0, /* Compiler rc */
0, 13, 16, 13, 16, /* 1st match */
REG_NOMATCH, /* 2nd match */
REG_NOMATCH /* 3rd match */
};
static int results4[] = {
REG_BADRPT /* Compiler rc */
};
/* Index the result vectors */
static int *results[] = {
(int *)(&results0),
(int *)(&results1),
(int *)(&results2),
(int *)(&results3),
(int *)(&results4)
};
/* And here is the program */
int main(int argc, char **argv)
{
regex_t re;
regmatch_t match[CAPCOUNT];
int v = argc > 1 && strcmp(argv[1], "-v") == 0;
PRINTF("Test of pcre2posix.h without pcre2.h\n");
for (int i = 0; i < count; i++)
{
char *pattern = data[i][0];
char **subjects = data[i] + 1;
int *rd = results[i];
int rc = regcomp(&re, pattern, cflags[i]);
PRINTF("Pattern: %s flags=0x%02x\n", pattern, cflags[i]);
if (rc != *rd)
{
fprintf(stderr, "Unexpected compile error %d (expected %d)\n", rc, *rd);
fprintf(stderr, "Pattern is: %s\n", pattern);
return 1;
}
if (rc != 0)
{
if (v)
{
char buffer[256];
(void)regerror(rc, &re, buffer, sizeof(buffer));
PRINTF("Compile error %d: %s (expected)\n", rc, buffer);
}
continue;
}
for (; *subjects != NULL; subjects++)
{
rc = regexec(&re, *subjects, CAPCOUNT, match, mflags[i]);
PRINTF("Subject: %s\n", *subjects);
PRINTF("Return: %d", rc);
if (rc != *(++rd))
{
PRINTF("\n");
fprintf(stderr, "Unexpected match error %d (expected %d)\n", rc, *rd);
fprintf(stderr, "Pattern is: %s\n", pattern);
fprintf(stderr, "Subject is: %s\n", *subjects);
return 1;
}
if (rc == 0)
{
for (int j = 0; j < CAPCOUNT; j++)
{
regmatch_t *m = match + j;
if (m->rm_so < 0) continue;
if (m->rm_so != *(++rd) || m->rm_eo != *(++rd))
{
PRINTF("\n");
fprintf(stderr, "Mismatched results for successful match\n");
fprintf(stderr, "Pattern is: %s\n", pattern);
fprintf(stderr, "Subject is: %s\n", *subjects);
fprintf(stderr, "Result %d: expected %d %d received %d %d\n",
j, rd[-1], rd[0], m->rm_so, m->rm_eo);
return 1;
}
PRINTF(" (%d %d %d)", j, m->rm_so, m->rm_eo);
}
}
else if (v)
{
char buffer[256];
(void)regerror(rc, &re, buffer, sizeof(buffer));
PRINTF(": %s (expected)", buffer);
}
PRINTF("\n");
}
regfree(&re);
}
PRINTF("End of test\n");
return 0;
}
/* End of pcre2posix_test.c */
pcre2-10.45/src/pcre2_chkdint.c 0000644 0001751 0000166 00000006161 14750415257 011710 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Copyright (c) 2023 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This file contains functions to implement checked integer operation */
#ifndef PCRE2_PCRE2TEST
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "pcre2_internal.h"
#endif
/*************************************************
* Checked Integer Multiplication *
*************************************************/
/*
Arguments:
r A pointer to PCRE2_SIZE to store the answer
a, b Two integers
Returns: Bool indicating if the operation overflows
It is modeled after C23's interface
The INT64_OR_DOUBLE type is a 64-bit integer type when available,
otherwise double. */
BOOL
PRIV(ckd_smul)(PCRE2_SIZE *r, int a, int b)
{
#ifdef HAVE_BUILTIN_MUL_OVERFLOW
PCRE2_SIZE m;
if (__builtin_mul_overflow(a, b, &m)) return TRUE;
*r = m;
#else
INT64_OR_DOUBLE m;
PCRE2_ASSERT(a >= 0 && b >= 0);
m = (INT64_OR_DOUBLE)a * (INT64_OR_DOUBLE)b;
#if defined INT64_MAX || defined int64_t
if (sizeof(m) > sizeof(*r) && m > (INT64_OR_DOUBLE)PCRE2_SIZE_MAX) return TRUE;
*r = (PCRE2_SIZE)m;
#else
if (m > PCRE2_SIZE_MAX) return TRUE;
*r = m;
#endif
#endif
return FALSE;
}
/* End of pcre2_chkdint.c */
pcre2-10.45/src/pcre2_jit_compile.c 0000644 0001751 0000166 00001515604 14750415257 012572 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
This module by Zoltan Herczeg
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#if defined(__has_feature)
#if __has_feature(memory_sanitizer)
#include
#endif /* __has_feature(memory_sanitizer) */
#endif /* defined(__has_feature) */
#include "pcre2_internal.h"
#ifdef SUPPORT_JIT
/* All-in-one: Since we use the JIT compiler only from here,
we just include it. This way we don't need to touch the build
system files. */
#define SLJIT_CONFIG_AUTO 1
#define SLJIT_CONFIG_STATIC 1
#define SLJIT_VERBOSE 0
#ifdef PCRE2_DEBUG
#define SLJIT_DEBUG 1
#else
#define SLJIT_DEBUG 0
#endif
#define SLJIT_MALLOC(size, allocator_data) pcre2_jit_malloc(size, allocator_data)
#define SLJIT_FREE(ptr, allocator_data) pcre2_jit_free(ptr, allocator_data)
static void * pcre2_jit_malloc(size_t size, void *allocator_data)
{
pcre2_memctl *allocator = ((pcre2_memctl*)allocator_data);
return allocator->malloc(size, allocator->memory_data);
}
static void pcre2_jit_free(void *ptr, void *allocator_data)
{
pcre2_memctl *allocator = ((pcre2_memctl*)allocator_data);
allocator->free(ptr, allocator->memory_data);
}
#include "../deps/sljit/sljit_src/sljitLir.c"
#if defined SLJIT_CONFIG_UNSUPPORTED && SLJIT_CONFIG_UNSUPPORTED
#error Unsupported architecture
#endif
/* Defines for debugging purposes. */
/* 1 - Use unoptimized capturing brackets.
2 - Enable capture_last_ptr (includes option 1). */
/* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */
/* 1 - Always have a control head. */
/* #define DEBUG_FORCE_CONTROL_HEAD 1 */
/* Allocate memory for the regex stack on the real machine stack.
Fast, but limited size. */
#define MACHINE_STACK_SIZE 32768
/* Growth rate for stack allocated by the OS. Should be the multiply
of page size. */
#define STACK_GROWTH_RATE 8192
/* Enable to check that the allocation could destroy temporaries. */
#if defined SLJIT_DEBUG && SLJIT_DEBUG
#define DESTROY_REGISTERS 1
#endif
/*
Short summary about the backtracking mechanism empolyed by the jit code generator:
The code generator follows the recursive nature of the PERL compatible regular
expressions. The basic blocks of regular expressions are condition checkers
whose execute different commands depending on the result of the condition check.
The relationship between the operators can be horizontal (concatenation) and
vertical (sub-expression) (See struct backtrack_common for more details).
'ab' - 'a' and 'b' regexps are concatenated
'a+' - 'a' is the sub-expression of the '+' operator
The condition checkers are boolean (true/false) checkers. Machine code is generated
for the checker itself and for the actions depending on the result of the checker.
The 'true' case is called as the matching path (expected path), and the other is called as
the 'backtrack' path. Branch instructions are expesive for all CPUs, so we avoid taken
branches on the matching path.
Greedy star operator (*) :
Matching path: match happens.
Backtrack path: match failed.
Non-greedy star operator (*?) :
Matching path: no need to perform a match.
Backtrack path: match is required.
The following example shows how the code generated for a capturing bracket
with two alternatives. Let A, B, C, D are arbirary regular expressions, and
we have the following regular expression:
A(B|C)D
The generated code will be the following:
A matching path
'(' matching path (pushing arguments to the stack)
B matching path
')' matching path (pushing arguments to the stack)
D matching path
return with successful match
D backtrack path
')' backtrack path (If we arrived from "C" jump to the backtrack of "C")
B backtrack path
C expected path
jump to D matching path
C backtrack path
A backtrack path
Notice, that the order of backtrack code paths are the opposite of the fast
code paths. In this way the topmost value on the stack is always belong
to the current backtrack code path. The backtrack path must check
whether there is a next alternative. If so, it needs to jump back to
the matching path eventually. Otherwise it needs to clear out its own stack
frame and continue the execution on the backtrack code paths.
*/
/*
Saved stack frames:
Atomic blocks and asserts require reloading the values of private data
when the backtrack mechanism performed. Because of OP_RECURSE, the data
are not necessarly known in compile time, thus we need a dynamic restore
mechanism.
The stack frames are stored in a chain list, and have the following format:
([ capturing bracket offset ][ start value ][ end value ])+ ... [ 0 ] [ previous head ]
Thus we can restore the private data to a particular point in the stack.
*/
typedef struct jit_arguments {
/* Pointers first. */
struct sljit_stack *stack;
PCRE2_SPTR str;
PCRE2_SPTR begin;
PCRE2_SPTR end;
pcre2_match_data *match_data;
PCRE2_SPTR startchar_ptr;
PCRE2_UCHAR *mark_ptr;
int (*callout)(pcre2_callout_block *, void *);
void *callout_data;
/* Everything else after. */
sljit_uw offset_limit;
sljit_u32 limit_match;
sljit_u32 oveccount;
sljit_u32 options;
} jit_arguments;
#define JIT_NUMBER_OF_COMPILE_MODES 3
typedef struct executable_functions {
void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES];
void *read_only_data_heads[JIT_NUMBER_OF_COMPILE_MODES];
sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES];
sljit_u32 top_bracket;
sljit_u32 limit_match;
} executable_functions;
typedef struct jump_list {
struct sljit_jump *jump;
struct jump_list *next;
} jump_list;
typedef struct stub_list {
struct sljit_jump *start;
struct sljit_label *quit;
struct stub_list *next;
} stub_list;
enum frame_types {
no_frame = -1,
no_stack = -2
};
enum control_types {
type_mark = 0,
type_then_trap = 1
};
enum early_fail_types {
type_skip = 0,
type_fail = 1,
type_fail_range = 2
};
typedef int (SLJIT_FUNC *jit_function)(jit_arguments *args);
/* The following structure is the key data type for the recursive
code generator. It is allocated by compile_matchingpath, and contains
the arguments for compile_backtrackingpath. Must be the first member
of its descendants. */
typedef struct backtrack_common {
/* Backtracking path of an opcode, which falls back
to our opcode, if it cannot resume matching. */
struct backtrack_common *prev;
/* Backtracks for opcodes without backtracking path.
These opcodes are between 'prev' and the current
opcode, and they never resume the match. */
jump_list *simple_backtracks;
/* Internal backtracking list for block constructs
which contains other opcodes, such as brackets,
asserts, conditionals, etc. */
struct backtrack_common *top;
/* Backtracks used internally by the opcode. For component
opcodes, this list is also used by those opcodes without
backtracking path which follows the 'top' backtrack. */
jump_list *own_backtracks;
/* Opcode pointer. */
PCRE2_SPTR cc;
} backtrack_common;
typedef struct assert_backtrack {
backtrack_common common;
jump_list *condfailed;
/* Less than 0 if a frame is not needed. */
int framesize;
/* Points to our private memory word on the stack. */
int private_data_ptr;
/* For iterators. */
struct sljit_label *matchingpath;
} assert_backtrack;
typedef struct bracket_backtrack {
backtrack_common common;
/* Where to coninue if an alternative is successfully matched. */
struct sljit_label *alternative_matchingpath;
/* For rmin and rmax iterators. */
struct sljit_label *recursive_matchingpath;
/* For greedy ? operator. */
struct sljit_label *zero_matchingpath;
/* Contains the branches of a failed condition. */
union {
/* Both for OP_COND, OP_SCOND, OP_ASSERT_SCS. */
jump_list *no_capture;
assert_backtrack *assert;
/* For OP_ONCE. Less than 0 if not needed. */
int framesize;
} u;
/* For brackets with >3 alternatives. */
struct sljit_jump *matching_mov_addr;
/* Points to our private memory word on the stack. */
int private_data_ptr;
} bracket_backtrack;
typedef struct bracketpos_backtrack {
backtrack_common common;
/* Points to our private memory word on the stack. */
int private_data_ptr;
/* Reverting stack is needed. */
int framesize;
/* Allocated stack size. */
int stacksize;
} bracketpos_backtrack;
typedef struct braminzero_backtrack {
backtrack_common common;
struct sljit_label *matchingpath;
} braminzero_backtrack;
typedef struct char_iterator_backtrack {
backtrack_common common;
/* Next iteration. */
struct sljit_label *matchingpath;
/* Creating a range based on the next character. */
struct {
unsigned int othercasebit;
PCRE2_UCHAR chr;
BOOL charpos_enabled;
} charpos;
} char_iterator_backtrack;
typedef struct ref_iterator_backtrack {
backtrack_common common;
/* Next iteration. */
struct sljit_label *matchingpath;
} ref_iterator_backtrack;
typedef struct recurse_entry {
struct recurse_entry *next;
/* Contains the function entry label. */
struct sljit_label *entry_label;
/* Contains the function entry label. */
struct sljit_label *backtrack_label;
/* Collects the entry calls until the function is not created. */
jump_list *entry_calls;
/* Collects the backtrack calls until the function is not created. */
jump_list *backtrack_calls;
/* Points to the starting opcode. */
sljit_sw start;
} recurse_entry;
typedef struct recurse_backtrack {
backtrack_common common;
/* Return to the matching path. */
struct sljit_label *matchingpath;
/* Recursive pattern. */
recurse_entry *entry;
/* Pattern is inlined. */
BOOL inlined_pattern;
} recurse_backtrack;
typedef struct vreverse_backtrack {
backtrack_common common;
/* Return to the matching path. */
struct sljit_label *matchingpath;
} vreverse_backtrack;
#define OP_THEN_TRAP OP_TABLE_LENGTH
typedef struct then_trap_backtrack {
backtrack_common common;
/* If then_trap is not NULL, this structure contains the real
then_trap for the backtracking path. */
struct then_trap_backtrack *then_trap;
/* Points to the starting opcode. */
sljit_sw start;
/* Exit point for the then opcodes of this alternative. */
jump_list *quit;
/* Frame size of the current alternative. */
int framesize;
} then_trap_backtrack;
#define MAX_N_CHARS 12
#define MAX_DIFF_CHARS 5
typedef struct fast_forward_char_data {
/* Number of characters in the chars array, 255 for any character. */
sljit_u8 count;
/* Number of last UTF-8 characters in the chars array. */
sljit_u8 last_count;
/* Available characters in the current position. */
PCRE2_UCHAR chars[MAX_DIFF_CHARS];
} fast_forward_char_data;
#define MAX_CLASS_RANGE_SIZE 4
#define MAX_CLASS_CHARS_SIZE 3
typedef struct compiler_common {
/* The sljit ceneric compiler. */
struct sljit_compiler *compiler;
/* Compiled regular expression. */
pcre2_real_code *re;
/* First byte code. */
PCRE2_SPTR start;
/* Maps private data offset to each opcode. */
sljit_s32 *private_data_ptrs;
/* Chain list of read-only data ptrs. */
void *read_only_data_head;
/* Tells whether the capturing bracket is optimized. */
sljit_u8 *optimized_cbracket;
/* Tells whether the starting offset is a target of then. */
sljit_u8 *then_offsets;
/* Current position where a THEN must jump. */
then_trap_backtrack *then_trap;
/* Starting offset of private data for capturing brackets. */
sljit_s32 cbra_ptr;
#if defined SLJIT_DEBUG && SLJIT_DEBUG
/* End offset of locals for assertions. */
sljit_s32 locals_size;
#endif
/* Output vector starting point. Must be divisible by 2. */
sljit_s32 ovector_start;
/* Points to the starting character of the current match. */
sljit_s32 start_ptr;
/* Last known position of the requested byte. */
sljit_s32 req_char_ptr;
/* Head of the last recursion. */
sljit_s32 recursive_head_ptr;
/* First inspected character for partial matching.
(Needed for avoiding zero length partial matches.) */
sljit_s32 start_used_ptr;
/* Starting pointer for partial soft matches. */
sljit_s32 hit_start;
/* Pointer of the match end position. */
sljit_s32 match_end_ptr;
/* Points to the marked string. */
sljit_s32 mark_ptr;
/* Head of the recursive control verb management chain.
Each item must have a previous offset and type
(see control_types) values. See do_search_mark. */
sljit_s32 control_head_ptr;
/* The offset of the saved STR_END in the outermost
scan substring block. Since scan substring restores
STR_END after a match, it is enough to restore
STR_END inside a scan substring block. */
sljit_s32 restore_end_ptr;
/* Points to the last matched capture block index. */
sljit_s32 capture_last_ptr;
/* Fast forward skipping byte code pointer. */
PCRE2_SPTR fast_forward_bc_ptr;
/* Locals used by fast fail optimization. */
sljit_s32 early_fail_start_ptr;
sljit_s32 early_fail_end_ptr;
/* Variables used by recursive call generator. */
sljit_s32 recurse_bitset_size;
uint8_t *recurse_bitset;
/* Flipped and lower case tables. */
const sljit_u8 *fcc;
sljit_sw lcc;
/* Mode can be PCRE2_JIT_COMPLETE and others. */
int mode;
/* TRUE, when empty match is accepted for partial matching. */
BOOL allow_empty_partial;
/* TRUE, when minlength is greater than 0. */
BOOL might_be_empty;
/* \K is found in the pattern. */
BOOL has_set_som;
/* (*SKIP:arg) is found in the pattern. */
BOOL has_skip_arg;
/* (*THEN) is found in the pattern. */
BOOL has_then;
/* (*SKIP) or (*SKIP:arg) is found in lookbehind assertion. */
BOOL has_skip_in_assert_back;
/* Quit is redirected by recurse, negative assertion, or positive assertion in conditional block. */
BOOL local_quit_available;
/* Currently in a positive assertion. */
BOOL in_positive_assertion;
/* Newline control. */
int nltype;
sljit_u32 nlmax;
sljit_u32 nlmin;
int newline;
int bsr_nltype;
sljit_u32 bsr_nlmax;
sljit_u32 bsr_nlmin;
/* Dollar endonly. */
int endonly;
/* Tables. */
sljit_sw ctypes;
/* Named capturing brackets. */
PCRE2_SPTR name_table;
sljit_sw name_count;
sljit_sw name_entry_size;
/* Labels and jump lists. */
struct sljit_label *partialmatchlabel;
struct sljit_label *quit_label;
struct sljit_label *abort_label;
struct sljit_label *accept_label;
struct sljit_label *ff_newline_shortcut;
stub_list *stubs;
recurse_entry *entries;
recurse_entry *currententry;
jump_list *partialmatch;
jump_list *quit;
jump_list *positive_assertion_quit;
jump_list *abort;
jump_list *failed_match;
jump_list *accept;
jump_list *calllimit;
jump_list *stackalloc;
jump_list *revertframes;
jump_list *wordboundary;
jump_list *ucp_wordboundary;
jump_list *anynewline;
jump_list *hspace;
jump_list *vspace;
jump_list *casefulcmp;
jump_list *caselesscmp;
jump_list *reset_match;
/* Same as reset_match, but resets the STR_PTR as well. */
jump_list *restart_match;
BOOL unset_backref;
BOOL alt_circumflex;
#ifdef SUPPORT_UNICODE
BOOL utf;
BOOL invalid_utf;
BOOL ucp;
/* Points to saving area for iref. */
jump_list *getucd;
jump_list *getucdtype;
#if PCRE2_CODE_UNIT_WIDTH == 8
jump_list *utfreadchar;
jump_list *utfreadtype8;
jump_list *utfpeakcharback;
#endif
#if PCRE2_CODE_UNIT_WIDTH == 8 || PCRE2_CODE_UNIT_WIDTH == 16
jump_list *utfreadchar_invalid;
jump_list *utfreadnewline_invalid;
jump_list *utfmoveback_invalid;
jump_list *utfpeakcharback_invalid;
#endif
#endif /* SUPPORT_UNICODE */
} compiler_common;
/* For byte_sequence_compare. */
typedef struct compare_context {
int length;
int sourcereg;
#if defined SLJIT_UNALIGNED && SLJIT_UNALIGNED
int ucharptr;
union {
sljit_s32 asint;
sljit_u16 asushort;
#if PCRE2_CODE_UNIT_WIDTH == 8
sljit_u8 asbyte;
sljit_u8 asuchars[4];
#elif PCRE2_CODE_UNIT_WIDTH == 16
sljit_u16 asuchars[2];
#elif PCRE2_CODE_UNIT_WIDTH == 32
sljit_u32 asuchars[1];
#endif
} c;
union {
sljit_s32 asint;
sljit_u16 asushort;
#if PCRE2_CODE_UNIT_WIDTH == 8
sljit_u8 asbyte;
sljit_u8 asuchars[4];
#elif PCRE2_CODE_UNIT_WIDTH == 16
sljit_u16 asuchars[2];
#elif PCRE2_CODE_UNIT_WIDTH == 32
sljit_u32 asuchars[1];
#endif
} oc;
#endif
} compare_context;
/* Undefine sljit macros. */
#undef CMP
/* Used for accessing the elements of the stack. */
#define STACK(i) ((i) * SSIZE_OF(sw))
#ifdef SLJIT_PREF_SHIFT_REG
#if SLJIT_PREF_SHIFT_REG == SLJIT_R2
/* Nothing. */
#elif SLJIT_PREF_SHIFT_REG == SLJIT_R3
#define SHIFT_REG_IS_R3
#else
#error "Unsupported shift register"
#endif
#endif
#define TMP1 SLJIT_R0
#ifdef SHIFT_REG_IS_R3
#define TMP2 SLJIT_R3
#define TMP3 SLJIT_R2
#else
#define TMP2 SLJIT_R2
#define TMP3 SLJIT_R3
#endif
#define STR_PTR SLJIT_R1
#define STR_END SLJIT_S0
#define STACK_TOP SLJIT_S1
#define STACK_LIMIT SLJIT_S2
#define COUNT_MATCH SLJIT_S3
#define ARGUMENTS SLJIT_S4
#define RETURN_ADDR SLJIT_R4
#if (defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32)
#define HAS_VIRTUAL_REGISTERS 1
#else
#define HAS_VIRTUAL_REGISTERS 0
#endif
/* Local space layout. */
/* Max limit of recursions. */
#define LIMIT_MATCH (0 * sizeof(sljit_sw))
/* Local variables. Their number is computed by check_opcode_types. */
#define LOCAL0 (1 * sizeof(sljit_sw))
#define LOCAL1 (2 * sizeof(sljit_sw))
#define LOCAL2 (3 * sizeof(sljit_sw))
#define LOCAL3 (4 * sizeof(sljit_sw))
#define LOCAL4 (5 * sizeof(sljit_sw))
/* The output vector is stored on the stack, and contains pointers
to characters. The vector data is divided into two groups: the first
group contains the start / end character pointers, and the second is
the start pointers when the end of the capturing group has not yet reached. */
#define OVECTOR_START (common->ovector_start)
#define OVECTOR(i) (OVECTOR_START + (i) * SSIZE_OF(sw))
#define OVECTOR_PRIV(i) (common->cbra_ptr + (i) * SSIZE_OF(sw))
#define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start])
#if PCRE2_CODE_UNIT_WIDTH == 8
#define MOV_UCHAR SLJIT_MOV_U8
#define IN_UCHARS(x) (x)
#elif PCRE2_CODE_UNIT_WIDTH == 16
#define MOV_UCHAR SLJIT_MOV_U16
#define UCHAR_SHIFT (1)
#define IN_UCHARS(x) ((x) * 2)
#elif PCRE2_CODE_UNIT_WIDTH == 32
#define MOV_UCHAR SLJIT_MOV_U32
#define UCHAR_SHIFT (2)
#define IN_UCHARS(x) ((x) * 4)
#else
#error Unsupported compiling mode
#endif
/* Shortcuts. */
#define DEFINE_COMPILER \
struct sljit_compiler *compiler = common->compiler
#define OP1(op, dst, dstw, src, srcw) \
sljit_emit_op1(compiler, (op), (dst), (dstw), (src), (srcw))
#define OP2(op, dst, dstw, src1, src1w, src2, src2w) \
sljit_emit_op2(compiler, (op), (dst), (dstw), (src1), (src1w), (src2), (src2w))
#define OP2U(op, src1, src1w, src2, src2w) \
sljit_emit_op2u(compiler, (op), (src1), (src1w), (src2), (src2w))
#define OP_SRC(op, src, srcw) \
sljit_emit_op_src(compiler, (op), (src), (srcw))
#define LABEL() \
sljit_emit_label(compiler)
#define JUMP(type) \
sljit_emit_jump(compiler, (type))
#define JUMPTO(type, label) \
sljit_set_label(sljit_emit_jump(compiler, (type)), (label))
#define JUMPHERE(jump) \
sljit_set_label((jump), sljit_emit_label(compiler))
#define SET_LABEL(jump, label) \
sljit_set_label((jump), (label))
#define CMP(type, src1, src1w, src2, src2w) \
sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w))
#define CMPTO(type, src1, src1w, src2, src2w, label) \
sljit_set_label(sljit_emit_cmp(compiler, (type), (src1), (src1w), (src2), (src2w)), (label))
#define OP_FLAGS(op, dst, dstw, type) \
sljit_emit_op_flags(compiler, (op), (dst), (dstw), (type))
#define SELECT(type, dst_reg, src1, src1w, src2_reg) \
sljit_emit_select(compiler, (type), (dst_reg), (src1), (src1w), (src2_reg))
#define GET_LOCAL_BASE(dst, dstw, offset) \
sljit_get_local_base(compiler, (dst), (dstw), (offset))
#define READ_CHAR_MAX ((sljit_u32)0xffffffff)
#define INVALID_UTF_CHAR -1
#define UNASSIGNED_UTF_CHAR 888
#if defined SUPPORT_UNICODE
#if PCRE2_CODE_UNIT_WIDTH == 8
#define GETCHARINC_INVALID(c, ptr, end, invalid_action) \
{ \
if (ptr[0] <= 0x7f) \
c = *ptr++; \
else if (ptr + 1 < end && ptr[1] >= 0x80 && ptr[1] < 0xc0) \
{ \
c = ptr[1] - 0x80; \
\
if (ptr[0] >= 0xc2 && ptr[0] <= 0xdf) \
{ \
c |= (ptr[0] - 0xc0) << 6; \
ptr += 2; \
} \
else if (ptr + 2 < end && ptr[2] >= 0x80 && ptr[2] < 0xc0) \
{ \
c = c << 6 | (ptr[2] - 0x80); \
\
if (ptr[0] >= 0xe0 && ptr[0] <= 0xef) \
{ \
c |= (ptr[0] - 0xe0) << 12; \
ptr += 3; \
\
if (c < 0x800 || (c >= 0xd800 && c < 0xe000)) \
{ \
invalid_action; \
} \
} \
else if (ptr + 3 < end && ptr[3] >= 0x80 && ptr[3] < 0xc0) \
{ \
c = c << 6 | (ptr[3] - 0x80); \
\
if (ptr[0] >= 0xf0 && ptr[0] <= 0xf4) \
{ \
c |= (ptr[0] - 0xf0) << 18; \
ptr += 4; \
\
if (c >= 0x110000 || c < 0x10000) \
{ \
invalid_action; \
} \
} \
else \
{ \
invalid_action; \
} \
} \
else \
{ \
invalid_action; \
} \
} \
else \
{ \
invalid_action; \
} \
} \
else \
{ \
invalid_action; \
} \
}
#define GETCHARBACK_INVALID(c, ptr, start, invalid_action) \
{ \
c = ptr[-1]; \
if (c <= 0x7f) \
ptr--; \
else if (ptr - 1 > start && ptr[-1] >= 0x80 && ptr[-1] < 0xc0) \
{ \
c -= 0x80; \
\
if (ptr[-2] >= 0xc2 && ptr[-2] <= 0xdf) \
{ \
c |= (ptr[-2] - 0xc0) << 6; \
ptr -= 2; \
} \
else if (ptr - 2 > start && ptr[-2] >= 0x80 && ptr[-2] < 0xc0) \
{ \
c = c << 6 | (ptr[-2] - 0x80); \
\
if (ptr[-3] >= 0xe0 && ptr[-3] <= 0xef) \
{ \
c |= (ptr[-3] - 0xe0) << 12; \
ptr -= 3; \
\
if (c < 0x800 || (c >= 0xd800 && c < 0xe000)) \
{ \
invalid_action; \
} \
} \
else if (ptr - 3 > start && ptr[-3] >= 0x80 && ptr[-3] < 0xc0) \
{ \
c = c << 6 | (ptr[-3] - 0x80); \
\
if (ptr[-4] >= 0xf0 && ptr[-4] <= 0xf4) \
{ \
c |= (ptr[-4] - 0xf0) << 18; \
ptr -= 4; \
\
if (c >= 0x110000 || c < 0x10000) \
{ \
invalid_action; \
} \
} \
else \
{ \
invalid_action; \
} \
} \
else \
{ \
invalid_action; \
} \
} \
else \
{ \
invalid_action; \
} \
} \
else \
{ \
invalid_action; \
} \
}
#elif PCRE2_CODE_UNIT_WIDTH == 16
#define GETCHARINC_INVALID(c, ptr, end, invalid_action) \
{ \
if (ptr[0] < 0xd800 || ptr[0] >= 0xe000) \
c = *ptr++; \
else if (ptr[0] < 0xdc00 && ptr + 1 < end && ptr[1] >= 0xdc00 && ptr[1] < 0xe000) \
{ \
c = (((ptr[0] - 0xd800) << 10) | (ptr[1] - 0xdc00)) + 0x10000; \
ptr += 2; \
} \
else \
{ \
invalid_action; \
} \
}
#define GETCHARBACK_INVALID(c, ptr, start, invalid_action) \
{ \
c = ptr[-1]; \
if (c < 0xd800 || c >= 0xe000) \
ptr--; \
else if (c >= 0xdc00 && ptr - 1 > start && ptr[-2] >= 0xd800 && ptr[-2] < 0xdc00) \
{ \
c = (((ptr[-2] - 0xd800) << 10) | (c - 0xdc00)) + 0x10000; \
ptr -= 2; \
} \
else \
{ \
invalid_action; \
} \
}
#elif PCRE2_CODE_UNIT_WIDTH == 32
#define GETCHARINC_INVALID(c, ptr, end, invalid_action) \
{ \
if (ptr[0] < 0xd800 || (ptr[0] >= 0xe000 && ptr[0] < 0x110000)) \
c = *ptr++; \
else \
{ \
invalid_action; \
} \
}
#define GETCHARBACK_INVALID(c, ptr, start, invalid_action) \
{ \
c = ptr[-1]; \
if (ptr[-1] < 0xd800 || (ptr[-1] >= 0xe000 && ptr[-1] < 0x110000)) \
ptr--; \
else \
{ \
invalid_action; \
} \
}
#endif /* PCRE2_CODE_UNIT_WIDTH == [8|16|32] */
#endif /* SUPPORT_UNICODE */
static PCRE2_SPTR bracketend(PCRE2_SPTR cc)
{
SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERT_SCS) || (*cc >= OP_ONCE && *cc <= OP_SCOND));
do cc += GET(cc, 1); while (*cc == OP_ALT);
SLJIT_ASSERT(*cc >= OP_KET && *cc <= OP_KETRPOS);
cc += 1 + LINK_SIZE;
return cc;
}
static int no_alternatives(PCRE2_SPTR cc)
{
int count = 0;
SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERT_SCS) || (*cc >= OP_ONCE && *cc <= OP_SCOND));
do
{
cc += GET(cc, 1);
count++;
}
while (*cc == OP_ALT);
SLJIT_ASSERT(*cc >= OP_KET && *cc <= OP_KETRPOS);
return count;
}
static BOOL find_vreverse(PCRE2_SPTR cc)
{
SLJIT_ASSERT(*cc == OP_ASSERTBACK || *cc == OP_ASSERTBACK_NOT || *cc == OP_ASSERTBACK_NA);
do
{
if (cc[1 + LINK_SIZE] == OP_VREVERSE)
return TRUE;
cc += GET(cc, 1);
}
while (*cc == OP_ALT);
return FALSE;
}
/* Functions whose might need modification for all new supported opcodes:
next_opcode
check_opcode_types
set_private_data_ptrs
get_framesize
init_frame
get_recurse_data_length
copy_recurse_data
compile_matchingpath
compile_backtrackingpath
*/
static PCRE2_SPTR next_opcode(compiler_common *common, PCRE2_SPTR cc)
{
SLJIT_UNUSED_ARG(common);
switch(*cc)
{
case OP_SOD:
case OP_SOM:
case OP_SET_SOM:
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
case OP_ALLANY:
case OP_NOTPROP:
case OP_PROP:
case OP_ANYNL:
case OP_NOT_HSPACE:
case OP_HSPACE:
case OP_NOT_VSPACE:
case OP_VSPACE:
case OP_EXTUNI:
case OP_EODN:
case OP_EOD:
case OP_CIRC:
case OP_CIRCM:
case OP_DOLL:
case OP_DOLLM:
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSSTAR:
case OP_CRPOSPLUS:
case OP_CRPOSQUERY:
case OP_CRPOSRANGE:
case OP_CLASS:
case OP_NCLASS:
case OP_REF:
case OP_REFI:
case OP_DNREF:
case OP_DNREFI:
case OP_RECURSE:
case OP_CALLOUT:
case OP_ALT:
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_KETRPOS:
case OP_REVERSE:
case OP_VREVERSE:
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
case OP_ASSERT_NA:
case OP_ASSERTBACK_NA:
case OP_ASSERT_SCS:
case OP_ONCE:
case OP_SCRIPT_RUN:
case OP_BRA:
case OP_BRAPOS:
case OP_CBRA:
case OP_CBRAPOS:
case OP_COND:
case OP_SBRA:
case OP_SBRAPOS:
case OP_SCBRA:
case OP_SCBRAPOS:
case OP_SCOND:
case OP_CREF:
case OP_DNCREF:
case OP_RREF:
case OP_DNRREF:
case OP_FALSE:
case OP_TRUE:
case OP_BRAZERO:
case OP_BRAMINZERO:
case OP_BRAPOSZERO:
case OP_PRUNE:
case OP_SKIP:
case OP_THEN:
case OP_COMMIT:
case OP_FAIL:
case OP_ACCEPT:
case OP_ASSERT_ACCEPT:
case OP_CLOSE:
case OP_SKIPZERO:
case OP_NOT_UCP_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
return cc + PRIV(OP_lengths)[*cc];
case OP_CHAR:
case OP_CHARI:
case OP_NOT:
case OP_NOTI:
case OP_STAR:
case OP_MINSTAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_QUERY:
case OP_MINQUERY:
case OP_UPTO:
case OP_MINUPTO:
case OP_EXACT:
case OP_POSSTAR:
case OP_POSPLUS:
case OP_POSQUERY:
case OP_POSUPTO:
case OP_STARI:
case OP_MINSTARI:
case OP_PLUSI:
case OP_MINPLUSI:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_UPTOI:
case OP_MINUPTOI:
case OP_EXACTI:
case OP_POSSTARI:
case OP_POSPLUSI:
case OP_POSQUERYI:
case OP_POSUPTOI:
case OP_NOTSTAR:
case OP_NOTMINSTAR:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
case OP_NOTUPTO:
case OP_NOTMINUPTO:
case OP_NOTEXACT:
case OP_NOTPOSSTAR:
case OP_NOTPOSPLUS:
case OP_NOTPOSQUERY:
case OP_NOTPOSUPTO:
case OP_NOTSTARI:
case OP_NOTMINSTARI:
case OP_NOTPLUSI:
case OP_NOTMINPLUSI:
case OP_NOTQUERYI:
case OP_NOTMINQUERYI:
case OP_NOTUPTOI:
case OP_NOTMINUPTOI:
case OP_NOTEXACTI:
case OP_NOTPOSSTARI:
case OP_NOTPOSPLUSI:
case OP_NOTPOSQUERYI:
case OP_NOTPOSUPTOI:
cc += PRIV(OP_lengths)[*cc];
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
return cc;
/* Special cases. */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEEXACT:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSPLUS:
case OP_TYPEPOSQUERY:
case OP_TYPEPOSUPTO:
return cc + PRIV(OP_lengths)[*cc] - 1;
case OP_ANYBYTE:
#ifdef SUPPORT_UNICODE
if (common->utf) return NULL;
#endif
return cc + 1;
case OP_CALLOUT_STR:
return cc + GET(cc, 1 + 2*LINK_SIZE);
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH != 8
case OP_ECLASS:
case OP_XCLASS:
SLJIT_COMPILE_ASSERT(OP_XCLASS + 1 == OP_ECLASS && OP_CLASS + 1 == OP_NCLASS && OP_NCLASS < OP_XCLASS, class_byte_code_order);
return cc + GET(cc, 1);
#endif
case OP_MARK:
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_SKIP_ARG:
case OP_THEN_ARG:
return cc + 1 + 2 + cc[1];
default:
SLJIT_UNREACHABLE();
return NULL;
}
}
static sljit_s32 ref_update_local_size(compiler_common *common, PCRE2_SPTR cc, sljit_s32 current_locals_size)
{
/* Depends on do_casefulcmp(), do_caselesscmp(), and compile_ref_matchingpath() */
int locals_size = 2 * SSIZE_OF(sw);
SLJIT_UNUSED_ARG(common);
#ifdef SUPPORT_UNICODE
if ((*cc == OP_REFI || *cc == OP_DNREFI) && (common->utf || common->ucp))
locals_size = 3 * SSIZE_OF(sw);
#endif
cc += PRIV(OP_lengths)[*cc];
/* Although do_casefulcmp() uses only one local, the allocate_stack()
calls during the repeat destroys LOCAL1 variables. */
if (*cc >= OP_CRSTAR && *cc <= OP_CRPOSRANGE)
locals_size += 2 * SSIZE_OF(sw);
return (current_locals_size >= locals_size) ? current_locals_size : locals_size;
}
static BOOL check_opcode_types(compiler_common *common, PCRE2_SPTR cc, PCRE2_SPTR ccend)
{
int count;
PCRE2_SPTR slot;
PCRE2_SPTR assert_back_end = cc - 1;
PCRE2_SPTR assert_na_end = cc - 1;
sljit_s32 locals_size = 2 * SSIZE_OF(sw);
BOOL set_recursive_head = FALSE;
BOOL set_capture_last = FALSE;
BOOL set_mark = FALSE;
/* Calculate important variables (like stack size) and checks whether all opcodes are supported. */
while (cc < ccend)
{
switch(*cc)
{
case OP_SET_SOM:
common->has_set_som = TRUE;
common->might_be_empty = TRUE;
cc += 1;
break;
case OP_TYPEUPTO:
case OP_TYPEEXACT:
if (cc[1 + IMM2_SIZE] == OP_EXTUNI && locals_size <= 3 * SSIZE_OF(sw))
locals_size = 3 * SSIZE_OF(sw);
cc += (2 + IMM2_SIZE) - 1;
break;
case OP_TYPEPOSSTAR:
case OP_TYPEPOSPLUS:
case OP_TYPEPOSQUERY:
if (cc[1] == OP_EXTUNI && locals_size <= 3 * SSIZE_OF(sw))
locals_size = 3 * SSIZE_OF(sw);
cc += 2 - 1;
break;
case OP_TYPEPOSUPTO:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf && locals_size <= 3 * SSIZE_OF(sw))
locals_size = 3 * SSIZE_OF(sw);
#endif
if (cc[1 + IMM2_SIZE] == OP_EXTUNI && locals_size <= 3 * SSIZE_OF(sw))
locals_size = 3 * SSIZE_OF(sw);
cc += (2 + IMM2_SIZE) - 1;
break;
case OP_REFI:
case OP_REF:
locals_size = ref_update_local_size(common, cc, locals_size);
common->optimized_cbracket[GET2(cc, 1)] = 0;
cc += PRIV(OP_lengths)[*cc];
break;
case OP_ASSERT_NA:
case OP_ASSERTBACK_NA:
case OP_ASSERT_SCS:
slot = bracketend(cc);
if (slot > assert_na_end)
assert_na_end = slot;
cc += 1 + LINK_SIZE;
break;
case OP_CBRAPOS:
case OP_SCBRAPOS:
common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0;
cc += 1 + LINK_SIZE + IMM2_SIZE;
break;
case OP_COND:
case OP_SCOND:
/* Only AUTO_CALLOUT can insert this opcode. We do
not intend to support this case. */
if (cc[1 + LINK_SIZE] == OP_CALLOUT || cc[1 + LINK_SIZE] == OP_CALLOUT_STR)
return FALSE;
cc += 1 + LINK_SIZE;
break;
case OP_CREF:
common->optimized_cbracket[GET2(cc, 1)] = 0;
cc += 1 + IMM2_SIZE;
break;
case OP_DNREFI:
case OP_DNREF:
locals_size = ref_update_local_size(common, cc, locals_size);
/* Fall through */
case OP_DNCREF:
count = GET2(cc, 1 + IMM2_SIZE);
slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
while (count-- > 0)
{
common->optimized_cbracket[GET2(slot, 0)] = 0;
slot += common->name_entry_size;
}
cc += PRIV(OP_lengths)[*cc];
break;
case OP_RECURSE:
/* Set its value only once. */
set_recursive_head = TRUE;
cc += 1 + LINK_SIZE;
break;
case OP_CALLOUT:
case OP_CALLOUT_STR:
set_capture_last = TRUE;
cc += (*cc == OP_CALLOUT) ? PRIV(OP_lengths)[OP_CALLOUT] : GET(cc, 1 + 2*LINK_SIZE);
break;
case OP_ASSERTBACK:
slot = bracketend(cc);
if (slot > assert_back_end)
assert_back_end = slot;
cc += 1 + LINK_SIZE;
break;
case OP_THEN_ARG:
common->has_then = TRUE;
common->control_head_ptr = 1;
/* Fall through. */
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_MARK:
set_mark = TRUE;
cc += 1 + 2 + cc[1];
break;
case OP_THEN:
common->has_then = TRUE;
common->control_head_ptr = 1;
cc += 1;
break;
case OP_SKIP:
if (cc < assert_back_end)
common->has_skip_in_assert_back = TRUE;
cc += 1;
break;
case OP_SKIP_ARG:
common->control_head_ptr = 1;
common->has_skip_arg = TRUE;
if (cc < assert_back_end)
common->has_skip_in_assert_back = TRUE;
cc += 1 + 2 + cc[1];
break;
case OP_ASSERT_ACCEPT:
if (cc < assert_na_end)
return FALSE;
cc++;
break;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
case OP_CRPOSRANGE:
/* The second value can be 0 for infinite repeats. */
if (common->utf && GET2(cc, 1) != GET2(cc, 1 + IMM2_SIZE) && locals_size <= 3 * SSIZE_OF(sw))
locals_size = 3 * SSIZE_OF(sw);
cc += 1 + 2 * IMM2_SIZE;
break;
case OP_POSUPTO:
case OP_POSUPTOI:
case OP_NOTPOSUPTO:
case OP_NOTPOSUPTOI:
if (common->utf && locals_size <= 3 * SSIZE_OF(sw))
locals_size = 3 * SSIZE_OF(sw);
#endif
/* Fall through */
default:
cc = next_opcode(common, cc);
if (cc == NULL)
return FALSE;
break;
}
}
SLJIT_ASSERT((locals_size & (SSIZE_OF(sw) - 1)) == 0);
#if defined SLJIT_DEBUG && SLJIT_DEBUG
common->locals_size = locals_size;
#endif
if (locals_size > 0)
common->ovector_start += locals_size;
if (set_mark)
{
SLJIT_ASSERT(common->mark_ptr == 0);
common->mark_ptr = common->ovector_start;
common->ovector_start += sizeof(sljit_sw);
}
if (set_recursive_head)
{
SLJIT_ASSERT(common->recursive_head_ptr == 0);
common->recursive_head_ptr = common->ovector_start;
common->ovector_start += sizeof(sljit_sw);
}
if (set_capture_last)
{
SLJIT_ASSERT(common->capture_last_ptr == 0);
common->capture_last_ptr = common->ovector_start;
common->ovector_start += sizeof(sljit_sw);
}
return TRUE;
}
#define EARLY_FAIL_ENHANCE_MAX (3 + 3)
/*
Start represent the number of allowed early fail enhancements
The 0-2 values has a special meaning:
0 - skip is allowed for all iterators
1 - fail is allowed for all iterators
2 - fail is allowed for greedy iterators
3 - only ranged early fail is allowed
>3 - (start - 3) number of remaining ranged early fails allowed
return: the updated value of start
*/
static int detect_early_fail(compiler_common *common, PCRE2_SPTR cc,
int *private_data_start, sljit_s32 depth, int start)
{
PCRE2_SPTR begin = cc;
PCRE2_SPTR next_alt;
PCRE2_SPTR end;
PCRE2_SPTR accelerated_start;
int result = 0;
int count, prev_count;
SLJIT_ASSERT(*cc == OP_ONCE || *cc == OP_BRA || *cc == OP_CBRA);
SLJIT_ASSERT(*cc != OP_CBRA || common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] != 0);
SLJIT_ASSERT(start < EARLY_FAIL_ENHANCE_MAX);
next_alt = cc + GET(cc, 1);
if (*next_alt == OP_ALT && start < 1)
start = 1;
do
{
count = start;
cc += 1 + LINK_SIZE + ((*cc == OP_CBRA) ? IMM2_SIZE : 0);
while (TRUE)
{
accelerated_start = NULL;
switch(*cc)
{
case OP_SOD:
case OP_SOM:
case OP_SET_SOM:
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
case OP_EODN:
case OP_EOD:
case OP_CIRC:
case OP_CIRCM:
case OP_DOLL:
case OP_DOLLM:
case OP_NOT_UCP_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
/* Zero width assertions. */
cc++;
continue;
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
case OP_ALLANY:
case OP_ANYBYTE:
case OP_NOT_HSPACE:
case OP_HSPACE:
case OP_NOT_VSPACE:
case OP_VSPACE:
if (count < 1)
count = 1;
cc++;
continue;
case OP_ANYNL:
case OP_EXTUNI:
if (count < 3)
count = 3;
cc++;
continue;
case OP_NOTPROP:
case OP_PROP:
if (count < 1)
count = 1;
cc += 1 + 2;
continue;
case OP_CHAR:
case OP_CHARI:
case OP_NOT:
case OP_NOTI:
if (count < 1)
count = 1;
cc += 2;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
continue;
case OP_TYPEMINSTAR:
case OP_TYPEMINPLUS:
if (count == 2)
count = 3;
/* Fall through */
case OP_TYPESTAR:
case OP_TYPEPLUS:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSPLUS:
/* The type or prop opcode is skipped in the next iteration. */
cc += 1;
if (cc[0] != OP_ANYNL && cc[0] != OP_EXTUNI)
{
accelerated_start = cc - 1;
break;
}
if (count < 3)
count = 3;
continue;
case OP_TYPEEXACT:
if (count < 1)
count = 1;
cc += 1 + IMM2_SIZE;
continue;
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
cc += IMM2_SIZE;
/* Fall through */
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSQUERY:
/* The type or prop opcode is skipped in the next iteration. */
if (count < 3)
count = 3;
cc += 1;
continue;
case OP_MINSTAR:
case OP_MINPLUS:
case OP_MINSTARI:
case OP_MINPLUSI:
case OP_NOTMINSTAR:
case OP_NOTMINPLUS:
case OP_NOTMINSTARI:
case OP_NOTMINPLUSI:
if (count == 2)
count = 3;
/* Fall through */
case OP_STAR:
case OP_PLUS:
case OP_POSSTAR:
case OP_POSPLUS:
case OP_STARI:
case OP_PLUSI:
case OP_POSSTARI:
case OP_POSPLUSI:
case OP_NOTSTAR:
case OP_NOTPLUS:
case OP_NOTPOSSTAR:
case OP_NOTPOSPLUS:
case OP_NOTSTARI:
case OP_NOTPLUSI:
case OP_NOTPOSSTARI:
case OP_NOTPOSPLUSI:
accelerated_start = cc;
cc += 2;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
case OP_EXACT:
if (count < 1)
count = 1;
cc += 2 + IMM2_SIZE;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
continue;
case OP_UPTO:
case OP_MINUPTO:
case OP_POSUPTO:
case OP_UPTOI:
case OP_MINUPTOI:
case OP_EXACTI:
case OP_POSUPTOI:
case OP_NOTUPTO:
case OP_NOTMINUPTO:
case OP_NOTEXACT:
case OP_NOTPOSUPTO:
case OP_NOTUPTOI:
case OP_NOTMINUPTOI:
case OP_NOTEXACTI:
case OP_NOTPOSUPTOI:
cc += IMM2_SIZE;
/* Fall through */
case OP_QUERY:
case OP_MINQUERY:
case OP_POSQUERY:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_POSQUERYI:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
case OP_NOTPOSQUERY:
case OP_NOTQUERYI:
case OP_NOTMINQUERYI:
case OP_NOTPOSQUERYI:
if (count < 3)
count = 3;
cc += 2;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
continue;
case OP_CLASS:
case OP_NCLASS:
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH != 8
case OP_XCLASS:
case OP_ECLASS:
accelerated_start = cc;
cc += (*cc >= OP_XCLASS) ? GET(cc, 1) : (unsigned int)(1 + (32 / sizeof(PCRE2_UCHAR)));
#else
accelerated_start = cc;
cc += (1 + (32 / sizeof(PCRE2_UCHAR)));
#endif
switch (*cc)
{
case OP_CRMINSTAR:
case OP_CRMINPLUS:
if (count == 2)
count = 3;
/* Fall through */
case OP_CRSTAR:
case OP_CRPLUS:
case OP_CRPOSSTAR:
case OP_CRPOSPLUS:
cc++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE))
{
/* Exact repeat. */
cc += 1 + 2 * IMM2_SIZE;
if (count < 1)
count = 1;
continue;
}
cc += 2 * IMM2_SIZE;
/* Fall through */
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSQUERY:
cc++;
if (count < 3)
count = 3;
continue;
default:
/* No repeat. */
if (count < 1)
count = 1;
continue;
}
break;
case OP_BRA:
case OP_CBRA:
prev_count = count;
if (count < 1)
count = 1;
if (depth >= 4)
break;
if (count < 3 && cc[GET(cc, 1)] == OP_ALT)
count = 3;
end = bracketend(cc);
if (end[-1 - LINK_SIZE] != OP_KET || (*cc == OP_CBRA && common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] == 0))
break;
prev_count = detect_early_fail(common, cc, private_data_start, depth + 1, prev_count);
if (prev_count > count)
count = prev_count;
if (PRIVATE_DATA(cc) != 0)
common->private_data_ptrs[begin - common->start] = 1;
if (count < EARLY_FAIL_ENHANCE_MAX)
{
cc = end;
continue;
}
break;
case OP_KET:
SLJIT_ASSERT(PRIVATE_DATA(cc) == 0);
if (cc >= next_alt)
break;
cc += 1 + LINK_SIZE;
continue;
}
if (accelerated_start == NULL)
break;
if (count == 0)
{
common->fast_forward_bc_ptr = accelerated_start;
common->private_data_ptrs[(accelerated_start + 1) - common->start] = ((*private_data_start) << 3) | type_skip;
*private_data_start += sizeof(sljit_sw);
count = 4;
}
else if (count < 3)
{
common->private_data_ptrs[(accelerated_start + 1) - common->start] = ((*private_data_start) << 3) | type_fail;
if (common->early_fail_start_ptr == 0)
common->early_fail_start_ptr = *private_data_start;
*private_data_start += sizeof(sljit_sw);
common->early_fail_end_ptr = *private_data_start;
if (*private_data_start > SLJIT_MAX_LOCAL_SIZE)
return EARLY_FAIL_ENHANCE_MAX;
count = 4;
}
else
{
common->private_data_ptrs[(accelerated_start + 1) - common->start] = ((*private_data_start) << 3) | type_fail_range;
if (common->early_fail_start_ptr == 0)
common->early_fail_start_ptr = *private_data_start;
*private_data_start += 2 * sizeof(sljit_sw);
common->early_fail_end_ptr = *private_data_start;
if (*private_data_start > SLJIT_MAX_LOCAL_SIZE)
return EARLY_FAIL_ENHANCE_MAX;
count++;
}
/* Cannot be part of a repeat. */
common->private_data_ptrs[begin - common->start] = 1;
if (count >= EARLY_FAIL_ENHANCE_MAX)
break;
}
if (*cc != OP_ALT && *cc != OP_KET)
result = EARLY_FAIL_ENHANCE_MAX;
else if (result < count)
result = count;
cc = next_alt;
next_alt = cc + GET(cc, 1);
}
while (*cc == OP_ALT);
return result;
}
static int get_class_iterator_size(PCRE2_SPTR cc)
{
sljit_u32 min;
sljit_u32 max;
switch(*cc)
{
case OP_CRSTAR:
case OP_CRPLUS:
return 2;
case OP_CRMINSTAR:
case OP_CRMINPLUS:
case OP_CRQUERY:
case OP_CRMINQUERY:
return 1;
case OP_CRRANGE:
case OP_CRMINRANGE:
min = GET2(cc, 1);
max = GET2(cc, 1 + IMM2_SIZE);
if (max == 0)
return (*cc == OP_CRRANGE) ? 2 : 1;
max -= min;
if (max > (sljit_u32)(*cc == OP_CRRANGE ? 0 : 1))
max = 2;
return max;
default:
return 0;
}
}
static BOOL detect_repeat(compiler_common *common, PCRE2_SPTR begin)
{
PCRE2_SPTR end = bracketend(begin);
PCRE2_SPTR next;
PCRE2_SPTR next_end;
PCRE2_SPTR max_end;
PCRE2_UCHAR type;
sljit_sw length = end - begin;
sljit_s32 min, max, i;
/* Detect fixed iterations first. */
if (end[-(1 + LINK_SIZE)] != OP_KET || PRIVATE_DATA(begin) != 0)
return FALSE;
/* /(?:AB){4,6}/ is currently converted to /(?:AB){3}(?AB){1,3}/
* Skip the check of the second part. */
if (PRIVATE_DATA(end - LINK_SIZE) != 0)
return TRUE;
next = end;
min = 1;
while (1)
{
if (*next != *begin)
break;
next_end = bracketend(next);
if (next_end - next != length || memcmp(begin, next, IN_UCHARS(length)) != 0)
break;
next = next_end;
min++;
}
if (min == 2)
return FALSE;
max = 0;
max_end = next;
if (*next == OP_BRAZERO || *next == OP_BRAMINZERO)
{
type = *next;
while (1)
{
if (next[0] != type || next[1] != OP_BRA || next[2 + LINK_SIZE] != *begin)
break;
next_end = bracketend(next + 2 + LINK_SIZE);
if (next_end - next != (length + 2 + LINK_SIZE) || memcmp(begin, next + 2 + LINK_SIZE, IN_UCHARS(length)) != 0)
break;
next = next_end;
max++;
}
if (next[0] == type && next[1] == *begin && max >= 1)
{
next_end = bracketend(next + 1);
if (next_end - next == (length + 1) && memcmp(begin, next + 1, IN_UCHARS(length)) == 0)
{
for (i = 0; i < max; i++, next_end += 1 + LINK_SIZE)
if (*next_end != OP_KET)
break;
if (i == max)
{
common->private_data_ptrs[max_end - common->start - LINK_SIZE] = next_end - max_end;
common->private_data_ptrs[max_end - common->start - LINK_SIZE + 1] = (type == OP_BRAZERO) ? OP_UPTO : OP_MINUPTO;
/* +2 the original and the last. */
common->private_data_ptrs[max_end - common->start - LINK_SIZE + 2] = max + 2;
if (min == 1)
return TRUE;
min--;
max_end -= (1 + LINK_SIZE) + GET(max_end, -LINK_SIZE);
}
}
}
}
if (min >= 3)
{
common->private_data_ptrs[end - common->start - LINK_SIZE] = max_end - end;
common->private_data_ptrs[end - common->start - LINK_SIZE + 1] = OP_EXACT;
common->private_data_ptrs[end - common->start - LINK_SIZE + 2] = min;
return TRUE;
}
return FALSE;
}
#define CASE_ITERATOR_PRIVATE_DATA_1 \
case OP_MINSTAR: \
case OP_MINPLUS: \
case OP_QUERY: \
case OP_MINQUERY: \
case OP_MINSTARI: \
case OP_MINPLUSI: \
case OP_QUERYI: \
case OP_MINQUERYI: \
case OP_NOTMINSTAR: \
case OP_NOTMINPLUS: \
case OP_NOTQUERY: \
case OP_NOTMINQUERY: \
case OP_NOTMINSTARI: \
case OP_NOTMINPLUSI: \
case OP_NOTQUERYI: \
case OP_NOTMINQUERYI:
#define CASE_ITERATOR_PRIVATE_DATA_2A \
case OP_STAR: \
case OP_PLUS: \
case OP_STARI: \
case OP_PLUSI: \
case OP_NOTSTAR: \
case OP_NOTPLUS: \
case OP_NOTSTARI: \
case OP_NOTPLUSI:
#define CASE_ITERATOR_PRIVATE_DATA_2B \
case OP_UPTO: \
case OP_MINUPTO: \
case OP_UPTOI: \
case OP_MINUPTOI: \
case OP_NOTUPTO: \
case OP_NOTMINUPTO: \
case OP_NOTUPTOI: \
case OP_NOTMINUPTOI:
#define CASE_ITERATOR_TYPE_PRIVATE_DATA_1 \
case OP_TYPEMINSTAR: \
case OP_TYPEMINPLUS: \
case OP_TYPEQUERY: \
case OP_TYPEMINQUERY:
#define CASE_ITERATOR_TYPE_PRIVATE_DATA_2A \
case OP_TYPESTAR: \
case OP_TYPEPLUS:
#define CASE_ITERATOR_TYPE_PRIVATE_DATA_2B \
case OP_TYPEUPTO: \
case OP_TYPEMINUPTO:
static void set_private_data_ptrs(compiler_common *common, int *private_data_start, PCRE2_SPTR ccend)
{
PCRE2_SPTR cc = common->start;
PCRE2_SPTR alternative;
PCRE2_SPTR end = NULL;
int private_data_ptr = *private_data_start;
int space, size, bracketlen;
BOOL repeat_check = TRUE;
while (cc < ccend)
{
space = 0;
size = 0;
bracketlen = 0;
if (private_data_ptr > SLJIT_MAX_LOCAL_SIZE)
break;
/* When the bracket is prefixed by a zero iteration, skip the repeat check (at this point). */
if (repeat_check && (*cc == OP_ONCE || *cc == OP_BRA || *cc == OP_CBRA || *cc == OP_COND))
{
if (detect_repeat(common, cc))
{
/* These brackets are converted to repeats, so no global
based single character repeat is allowed. */
if (cc >= end)
end = bracketend(cc);
}
}
repeat_check = TRUE;
switch(*cc)
{
case OP_KET:
if (common->private_data_ptrs[cc + 1 - common->start] != 0)
{
common->private_data_ptrs[cc - common->start] = private_data_ptr;
private_data_ptr += sizeof(sljit_sw);
cc += common->private_data_ptrs[cc + 1 - common->start];
}
cc += 1 + LINK_SIZE;
break;
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
case OP_ASSERT_NA:
case OP_ONCE:
case OP_SCRIPT_RUN:
case OP_BRAPOS:
case OP_SBRA:
case OP_SBRAPOS:
case OP_SCOND:
common->private_data_ptrs[cc - common->start] = private_data_ptr;
private_data_ptr += sizeof(sljit_sw);
bracketlen = 1 + LINK_SIZE;
break;
case OP_ASSERTBACK_NA:
common->private_data_ptrs[cc - common->start] = private_data_ptr;
private_data_ptr += sizeof(sljit_sw);
if (find_vreverse(cc))
{
common->private_data_ptrs[cc + 1 - common->start] = 1;
private_data_ptr += sizeof(sljit_sw);
}
bracketlen = 1 + LINK_SIZE;
break;
case OP_ASSERT_SCS:
common->private_data_ptrs[cc - common->start] = private_data_ptr;
private_data_ptr += 2 * sizeof(sljit_sw);
bracketlen = 1 + LINK_SIZE;
break;
case OP_CBRAPOS:
case OP_SCBRAPOS:
common->private_data_ptrs[cc - common->start] = private_data_ptr;
private_data_ptr += sizeof(sljit_sw);
bracketlen = 1 + LINK_SIZE + IMM2_SIZE;
break;
case OP_COND:
/* Might be a hidden SCOND. */
common->private_data_ptrs[cc - common->start] = 0;
alternative = cc + GET(cc, 1);
if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN)
{
common->private_data_ptrs[cc - common->start] = private_data_ptr;
private_data_ptr += sizeof(sljit_sw);
}
bracketlen = 1 + LINK_SIZE;
break;
case OP_BRA:
bracketlen = 1 + LINK_SIZE;
break;
case OP_CBRA:
case OP_SCBRA:
bracketlen = 1 + LINK_SIZE + IMM2_SIZE;
break;
case OP_BRAZERO:
case OP_BRAMINZERO:
case OP_BRAPOSZERO:
size = 1;
repeat_check = FALSE;
break;
CASE_ITERATOR_PRIVATE_DATA_1
size = -2;
space = 1;
break;
CASE_ITERATOR_PRIVATE_DATA_2A
size = -2;
space = 2;
break;
CASE_ITERATOR_PRIVATE_DATA_2B
size = -(2 + IMM2_SIZE);
space = 2;
break;
CASE_ITERATOR_TYPE_PRIVATE_DATA_1
size = 1;
space = 1;
break;
CASE_ITERATOR_TYPE_PRIVATE_DATA_2A
size = 1;
if (cc[1] != OP_EXTUNI)
space = 2;
break;
case OP_TYPEUPTO:
size = 1 + IMM2_SIZE;
if (cc[1 + IMM2_SIZE] != OP_EXTUNI)
space = 2;
break;
case OP_TYPEMINUPTO:
size = 1 + IMM2_SIZE;
space = 2;
break;
case OP_CLASS:
case OP_NCLASS:
size = 1 + 32 / sizeof(PCRE2_UCHAR);
space = get_class_iterator_size(cc + size);
break;
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH != 8
case OP_XCLASS:
case OP_ECLASS:
size = GET(cc, 1);
space = get_class_iterator_size(cc + size);
break;
#endif
default:
cc = next_opcode(common, cc);
SLJIT_ASSERT(cc != NULL);
break;
}
/* Character iterators, which are not inside a repeated bracket,
gets a private slot instead of allocating it on the stack. */
if (space > 0 && cc >= end)
{
common->private_data_ptrs[cc - common->start] = private_data_ptr;
private_data_ptr += sizeof(sljit_sw) * space;
}
if (size != 0)
{
if (size < 0)
{
cc += -size;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
}
else
cc += size;
}
if (bracketlen > 0)
{
if (cc >= end)
{
end = bracketend(cc);
if (end[-1 - LINK_SIZE] == OP_KET)
end = NULL;
}
cc += bracketlen;
}
}
*private_data_start = private_data_ptr;
}
/* Returns with a frame_types (always < 0) if no need for frame. */
static int get_framesize(compiler_common *common, PCRE2_SPTR cc, PCRE2_SPTR ccend, BOOL recursive, BOOL *needs_control_head)
{
int length = 0;
int possessive = 0;
BOOL stack_restore = FALSE;
BOOL setsom_found = recursive;
BOOL setmark_found = recursive;
/* The last capture is a local variable even for recursions. */
BOOL capture_last_found = FALSE;
#if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD
SLJIT_ASSERT(common->control_head_ptr != 0);
*needs_control_head = TRUE;
#else
*needs_control_head = FALSE;
#endif
if (ccend == NULL)
{
ccend = bracketend(cc) - (1 + LINK_SIZE);
if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS))
{
possessive = length = (common->capture_last_ptr != 0) ? 5 : 3;
/* This is correct regardless of common->capture_last_ptr. */
capture_last_found = TRUE;
}
cc = next_opcode(common, cc);
}
SLJIT_ASSERT(cc != NULL);
while (cc < ccend)
switch(*cc)
{
case OP_SET_SOM:
SLJIT_ASSERT(common->has_set_som);
stack_restore = TRUE;
if (!setsom_found)
{
length += 2;
setsom_found = TRUE;
}
cc += 1;
break;
case OP_MARK:
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_THEN_ARG:
SLJIT_ASSERT(common->mark_ptr != 0);
stack_restore = TRUE;
if (!setmark_found)
{
length += 2;
setmark_found = TRUE;
}
if (common->control_head_ptr != 0)
*needs_control_head = TRUE;
cc += 1 + 2 + cc[1];
break;
case OP_RECURSE:
stack_restore = TRUE;
if (common->has_set_som && !setsom_found)
{
length += 2;
setsom_found = TRUE;
}
if (common->mark_ptr != 0 && !setmark_found)
{
length += 2;
setmark_found = TRUE;
}
if (common->capture_last_ptr != 0 && !capture_last_found)
{
length += 2;
capture_last_found = TRUE;
}
cc += 1 + LINK_SIZE;
break;
case OP_CBRA:
case OP_CBRAPOS:
case OP_SCBRA:
case OP_SCBRAPOS:
stack_restore = TRUE;
if (common->capture_last_ptr != 0 && !capture_last_found)
{
length += 2;
capture_last_found = TRUE;
}
length += 3;
cc += 1 + LINK_SIZE + IMM2_SIZE;
break;
case OP_THEN:
stack_restore = TRUE;
if (common->control_head_ptr != 0)
*needs_control_head = TRUE;
cc ++;
break;
default:
stack_restore = TRUE;
/* Fall through. */
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
case OP_ALLANY:
case OP_ANYBYTE:
case OP_NOTPROP:
case OP_PROP:
case OP_ANYNL:
case OP_NOT_HSPACE:
case OP_HSPACE:
case OP_NOT_VSPACE:
case OP_VSPACE:
case OP_EXTUNI:
case OP_EODN:
case OP_EOD:
case OP_CIRC:
case OP_CIRCM:
case OP_DOLL:
case OP_DOLLM:
case OP_CHAR:
case OP_CHARI:
case OP_NOT:
case OP_NOTI:
case OP_EXACT:
case OP_POSSTAR:
case OP_POSPLUS:
case OP_POSQUERY:
case OP_POSUPTO:
case OP_EXACTI:
case OP_POSSTARI:
case OP_POSPLUSI:
case OP_POSQUERYI:
case OP_POSUPTOI:
case OP_NOTEXACT:
case OP_NOTPOSSTAR:
case OP_NOTPOSPLUS:
case OP_NOTPOSQUERY:
case OP_NOTPOSUPTO:
case OP_NOTEXACTI:
case OP_NOTPOSSTARI:
case OP_NOTPOSPLUSI:
case OP_NOTPOSQUERYI:
case OP_NOTPOSUPTOI:
case OP_TYPEEXACT:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSPLUS:
case OP_TYPEPOSQUERY:
case OP_TYPEPOSUPTO:
case OP_CLASS:
case OP_NCLASS:
case OP_XCLASS:
case OP_ECLASS:
case OP_CALLOUT:
case OP_CALLOUT_STR:
case OP_NOT_UCP_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
cc = next_opcode(common, cc);
SLJIT_ASSERT(cc != NULL);
break;
}
/* Possessive quantifiers can use a special case. */
if (SLJIT_UNLIKELY(possessive == length))
return stack_restore ? no_frame : no_stack;
if (length > 0)
return length + 1;
return stack_restore ? no_frame : no_stack;
}
static void init_frame(compiler_common *common, PCRE2_SPTR cc, PCRE2_SPTR ccend, int stackpos, int stacktop)
{
DEFINE_COMPILER;
BOOL setsom_found = FALSE;
BOOL setmark_found = FALSE;
/* The last capture is a local variable even for recursions. */
BOOL capture_last_found = FALSE;
int offset;
/* >= 1 + shortest item size (2) */
SLJIT_UNUSED_ARG(stacktop);
SLJIT_ASSERT(stackpos >= stacktop + 2);
stackpos = STACK(stackpos);
if (ccend == NULL)
{
ccend = bracketend(cc) - (1 + LINK_SIZE);
if (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS)
cc = next_opcode(common, cc);
}
/* The data is restored by do_revertframes(). */
SLJIT_ASSERT(cc != NULL);
while (cc < ccend)
switch(*cc)
{
case OP_SET_SOM:
SLJIT_ASSERT(common->has_set_som);
if (!setsom_found)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(0));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0));
stackpos -= SSIZE_OF(sw);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
stackpos -= SSIZE_OF(sw);
setsom_found = TRUE;
}
cc += 1;
break;
case OP_MARK:
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_THEN_ARG:
SLJIT_ASSERT(common->mark_ptr != 0);
if (!setmark_found)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->mark_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr);
stackpos -= SSIZE_OF(sw);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
stackpos -= SSIZE_OF(sw);
setmark_found = TRUE;
}
cc += 1 + 2 + cc[1];
break;
case OP_RECURSE:
if (common->has_set_som && !setsom_found)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(0));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -OVECTOR(0));
stackpos -= SSIZE_OF(sw);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
stackpos -= SSIZE_OF(sw);
setsom_found = TRUE;
}
if (common->mark_ptr != 0 && !setmark_found)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->mark_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->mark_ptr);
stackpos -= SSIZE_OF(sw);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
stackpos -= SSIZE_OF(sw);
setmark_found = TRUE;
}
if (common->capture_last_ptr != 0 && !capture_last_found)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr);
stackpos -= SSIZE_OF(sw);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
stackpos -= SSIZE_OF(sw);
capture_last_found = TRUE;
}
cc += 1 + LINK_SIZE;
break;
case OP_CBRA:
case OP_CBRAPOS:
case OP_SCBRA:
case OP_SCBRAPOS:
if (common->capture_last_ptr != 0 && !capture_last_found)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, -common->capture_last_ptr);
stackpos -= SSIZE_OF(sw);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
stackpos -= SSIZE_OF(sw);
capture_last_found = TRUE;
}
offset = (GET2(cc, 1 + LINK_SIZE)) << 1;
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, OVECTOR(offset));
stackpos -= SSIZE_OF(sw);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset));
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP1, 0);
stackpos -= SSIZE_OF(sw);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, TMP2, 0);
stackpos -= SSIZE_OF(sw);
cc += 1 + LINK_SIZE + IMM2_SIZE;
break;
default:
cc = next_opcode(common, cc);
SLJIT_ASSERT(cc != NULL);
break;
}
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), stackpos, SLJIT_IMM, 0);
SLJIT_ASSERT(stackpos == STACK(stacktop));
}
#define RECURSE_TMP_REG_COUNT 3
typedef struct delayed_mem_copy_status {
struct sljit_compiler *compiler;
int store_bases[RECURSE_TMP_REG_COUNT];
int store_offsets[RECURSE_TMP_REG_COUNT];
int tmp_regs[RECURSE_TMP_REG_COUNT];
int saved_tmp_regs[RECURSE_TMP_REG_COUNT];
int next_tmp_reg;
} delayed_mem_copy_status;
static void delayed_mem_copy_init(delayed_mem_copy_status *status, compiler_common *common)
{
int i;
for (i = 0; i < RECURSE_TMP_REG_COUNT; i++)
{
SLJIT_ASSERT(status->tmp_regs[i] >= 0);
SLJIT_ASSERT(sljit_get_register_index(SLJIT_GP_REGISTER, status->saved_tmp_regs[i]) < 0 || status->tmp_regs[i] == status->saved_tmp_regs[i]);
status->store_bases[i] = -1;
}
status->next_tmp_reg = 0;
status->compiler = common->compiler;
}
static void delayed_mem_copy_move(delayed_mem_copy_status *status, int load_base, sljit_sw load_offset,
int store_base, sljit_sw store_offset)
{
struct sljit_compiler *compiler = status->compiler;
int next_tmp_reg = status->next_tmp_reg;
int tmp_reg = status->tmp_regs[next_tmp_reg];
SLJIT_ASSERT(load_base > 0 && store_base > 0);
if (status->store_bases[next_tmp_reg] == -1)
{
/* Preserve virtual registers. */
if (sljit_get_register_index(SLJIT_GP_REGISTER, status->saved_tmp_regs[next_tmp_reg]) < 0)
OP1(SLJIT_MOV, status->saved_tmp_regs[next_tmp_reg], 0, tmp_reg, 0);
}
else
OP1(SLJIT_MOV, SLJIT_MEM1(status->store_bases[next_tmp_reg]), status->store_offsets[next_tmp_reg], tmp_reg, 0);
OP1(SLJIT_MOV, tmp_reg, 0, SLJIT_MEM1(load_base), load_offset);
status->store_bases[next_tmp_reg] = store_base;
status->store_offsets[next_tmp_reg] = store_offset;
status->next_tmp_reg = (next_tmp_reg + 1) % RECURSE_TMP_REG_COUNT;
}
static void delayed_mem_copy_finish(delayed_mem_copy_status *status)
{
struct sljit_compiler *compiler = status->compiler;
int next_tmp_reg = status->next_tmp_reg;
int tmp_reg, saved_tmp_reg, i;
for (i = 0; i < RECURSE_TMP_REG_COUNT; i++)
{
if (status->store_bases[next_tmp_reg] != -1)
{
tmp_reg = status->tmp_regs[next_tmp_reg];
saved_tmp_reg = status->saved_tmp_regs[next_tmp_reg];
OP1(SLJIT_MOV, SLJIT_MEM1(status->store_bases[next_tmp_reg]), status->store_offsets[next_tmp_reg], tmp_reg, 0);
/* Restore virtual registers. */
if (sljit_get_register_index(SLJIT_GP_REGISTER, saved_tmp_reg) < 0)
OP1(SLJIT_MOV, tmp_reg, 0, saved_tmp_reg, 0);
}
next_tmp_reg = (next_tmp_reg + 1) % RECURSE_TMP_REG_COUNT;
}
}
#undef RECURSE_TMP_REG_COUNT
static BOOL recurse_check_bit(compiler_common *common, sljit_sw bit_index)
{
uint8_t *byte;
uint8_t mask;
SLJIT_ASSERT((bit_index & (sizeof(sljit_sw) - 1)) == 0);
bit_index >>= SLJIT_WORD_SHIFT;
SLJIT_ASSERT((bit_index >> 3) < common->recurse_bitset_size);
mask = 1 << (bit_index & 0x7);
byte = common->recurse_bitset + (bit_index >> 3);
if (*byte & mask)
return FALSE;
*byte |= mask;
return TRUE;
}
enum get_recurse_flags {
recurse_flag_quit_found = (1 << 0),
recurse_flag_accept_found = (1 << 1),
recurse_flag_setsom_found = (1 << 2),
recurse_flag_setmark_found = (1 << 3),
recurse_flag_control_head_found = (1 << 4),
};
static int get_recurse_data_length(compiler_common *common, PCRE2_SPTR cc, PCRE2_SPTR ccend, uint32_t *result_flags)
{
int length = 1;
int size, offset;
PCRE2_SPTR alternative;
uint32_t recurse_flags = 0;
memset(common->recurse_bitset, 0, common->recurse_bitset_size);
#if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD
SLJIT_ASSERT(common->control_head_ptr != 0);
recurse_flags |= recurse_flag_control_head_found;
#endif
/* Calculate the sum of the private machine words. */
while (cc < ccend)
{
size = 0;
switch(*cc)
{
case OP_SET_SOM:
SLJIT_ASSERT(common->has_set_som);
recurse_flags |= recurse_flag_setsom_found;
cc += 1;
break;
case OP_RECURSE:
if (common->has_set_som)
recurse_flags |= recurse_flag_setsom_found;
if (common->mark_ptr != 0)
recurse_flags |= recurse_flag_setmark_found;
if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr))
length++;
cc += 1 + LINK_SIZE;
break;
case OP_KET:
offset = PRIVATE_DATA(cc);
if (offset != 0)
{
if (recurse_check_bit(common, offset))
length++;
SLJIT_ASSERT(PRIVATE_DATA(cc + 1) != 0);
cc += PRIVATE_DATA(cc + 1);
}
cc += 1 + LINK_SIZE;
break;
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
case OP_ASSERT_NA:
case OP_ASSERTBACK_NA:
case OP_ONCE:
case OP_SCRIPT_RUN:
case OP_BRAPOS:
case OP_SBRA:
case OP_SBRAPOS:
case OP_SCOND:
SLJIT_ASSERT(PRIVATE_DATA(cc) != 0);
if (recurse_check_bit(common, PRIVATE_DATA(cc)))
length++;
cc += 1 + LINK_SIZE;
break;
case OP_ASSERT_SCS:
SLJIT_ASSERT(PRIVATE_DATA(cc) != 0);
if (recurse_check_bit(common, PRIVATE_DATA(cc)))
length += 2;
cc += 1 + LINK_SIZE;
break;
case OP_CBRA:
case OP_SCBRA:
offset = GET2(cc, 1 + LINK_SIZE);
if (recurse_check_bit(common, OVECTOR(offset << 1)))
{
SLJIT_ASSERT(recurse_check_bit(common, OVECTOR((offset << 1) + 1)));
length += 2;
}
if (common->optimized_cbracket[offset] == 0 && recurse_check_bit(common, OVECTOR_PRIV(offset)))
length++;
if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr))
length++;
cc += 1 + LINK_SIZE + IMM2_SIZE;
break;
case OP_CBRAPOS:
case OP_SCBRAPOS:
offset = GET2(cc, 1 + LINK_SIZE);
if (recurse_check_bit(common, OVECTOR(offset << 1)))
{
SLJIT_ASSERT(recurse_check_bit(common, OVECTOR((offset << 1) + 1)));
length += 2;
}
if (recurse_check_bit(common, OVECTOR_PRIV(offset)))
length++;
if (recurse_check_bit(common, PRIVATE_DATA(cc)))
length++;
if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr))
length++;
cc += 1 + LINK_SIZE + IMM2_SIZE;
break;
case OP_COND:
/* Might be a hidden SCOND. */
alternative = cc + GET(cc, 1);
if ((*alternative == OP_KETRMAX || *alternative == OP_KETRMIN) && recurse_check_bit(common, PRIVATE_DATA(cc)))
length++;
cc += 1 + LINK_SIZE;
break;
CASE_ITERATOR_PRIVATE_DATA_1
offset = PRIVATE_DATA(cc);
if (offset != 0 && recurse_check_bit(common, offset))
length++;
cc += 2;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
CASE_ITERATOR_PRIVATE_DATA_2A
offset = PRIVATE_DATA(cc);
if (offset != 0 && recurse_check_bit(common, offset))
{
SLJIT_ASSERT(recurse_check_bit(common, offset + sizeof(sljit_sw)));
length += 2;
}
cc += 2;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
CASE_ITERATOR_PRIVATE_DATA_2B
offset = PRIVATE_DATA(cc);
if (offset != 0 && recurse_check_bit(common, offset))
{
SLJIT_ASSERT(recurse_check_bit(common, offset + sizeof(sljit_sw)));
length += 2;
}
cc += 2 + IMM2_SIZE;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
CASE_ITERATOR_TYPE_PRIVATE_DATA_1
offset = PRIVATE_DATA(cc);
if (offset != 0 && recurse_check_bit(common, offset))
length++;
cc += 1;
break;
CASE_ITERATOR_TYPE_PRIVATE_DATA_2A
offset = PRIVATE_DATA(cc);
if (offset != 0 && recurse_check_bit(common, offset))
{
SLJIT_ASSERT(recurse_check_bit(common, offset + sizeof(sljit_sw)));
length += 2;
}
cc += 1;
break;
CASE_ITERATOR_TYPE_PRIVATE_DATA_2B
offset = PRIVATE_DATA(cc);
if (offset != 0 && recurse_check_bit(common, offset))
{
SLJIT_ASSERT(recurse_check_bit(common, offset + sizeof(sljit_sw)));
length += 2;
}
cc += 1 + IMM2_SIZE;
break;
case OP_CLASS:
case OP_NCLASS:
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH != 8
case OP_XCLASS:
case OP_ECLASS:
size = (*cc >= OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(PCRE2_UCHAR);
#else
size = 1 + 32 / (int)sizeof(PCRE2_UCHAR);
#endif
offset = PRIVATE_DATA(cc);
if (offset != 0 && recurse_check_bit(common, offset))
length += get_class_iterator_size(cc + size);
cc += size;
break;
case OP_MARK:
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_THEN_ARG:
SLJIT_ASSERT(common->mark_ptr != 0);
recurse_flags |= recurse_flag_setmark_found;
if (common->control_head_ptr != 0)
recurse_flags |= recurse_flag_control_head_found;
if (*cc != OP_MARK)
recurse_flags |= recurse_flag_quit_found;
cc += 1 + 2 + cc[1];
break;
case OP_PRUNE:
case OP_SKIP:
case OP_COMMIT:
recurse_flags |= recurse_flag_quit_found;
cc++;
break;
case OP_SKIP_ARG:
recurse_flags |= recurse_flag_quit_found;
cc += 1 + 2 + cc[1];
break;
case OP_THEN:
SLJIT_ASSERT(common->control_head_ptr != 0);
recurse_flags |= recurse_flag_quit_found | recurse_flag_control_head_found;
cc++;
break;
case OP_ACCEPT:
case OP_ASSERT_ACCEPT:
recurse_flags |= recurse_flag_accept_found;
cc++;
break;
default:
cc = next_opcode(common, cc);
SLJIT_ASSERT(cc != NULL);
break;
}
}
SLJIT_ASSERT(cc == ccend);
if (recurse_flags & recurse_flag_control_head_found)
length++;
if (recurse_flags & recurse_flag_quit_found)
{
if (recurse_flags & recurse_flag_setsom_found)
length++;
if (recurse_flags & recurse_flag_setmark_found)
length++;
}
*result_flags = recurse_flags;
return length;
}
enum copy_recurse_data_types {
recurse_copy_from_global,
recurse_copy_private_to_global,
recurse_copy_shared_to_global,
recurse_copy_kept_shared_to_global,
recurse_swap_global
};
static void copy_recurse_data(compiler_common *common, PCRE2_SPTR cc, PCRE2_SPTR ccend,
int type, int stackptr, int stacktop, uint32_t recurse_flags)
{
delayed_mem_copy_status status;
PCRE2_SPTR alternative;
sljit_sw private_srcw[2];
sljit_sw shared_srcw[3];
sljit_sw kept_shared_srcw[2];
int private_count, shared_count, kept_shared_count;
int from_sp, base_reg, offset, i;
memset(common->recurse_bitset, 0, common->recurse_bitset_size);
#if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD
SLJIT_ASSERT(common->control_head_ptr != 0);
recurse_check_bit(common, common->control_head_ptr);
#endif
switch (type)
{
case recurse_copy_from_global:
from_sp = TRUE;
base_reg = STACK_TOP;
break;
case recurse_copy_private_to_global:
case recurse_copy_shared_to_global:
case recurse_copy_kept_shared_to_global:
from_sp = FALSE;
base_reg = STACK_TOP;
break;
default:
SLJIT_ASSERT(type == recurse_swap_global);
from_sp = FALSE;
base_reg = TMP2;
break;
}
stackptr = STACK(stackptr);
stacktop = STACK(stacktop);
status.tmp_regs[0] = TMP1;
status.saved_tmp_regs[0] = TMP1;
if (base_reg != TMP2)
{
status.tmp_regs[1] = TMP2;
status.saved_tmp_regs[1] = TMP2;
}
else
{
status.saved_tmp_regs[1] = RETURN_ADDR;
if (HAS_VIRTUAL_REGISTERS)
status.tmp_regs[1] = STR_PTR;
else
status.tmp_regs[1] = RETURN_ADDR;
}
status.saved_tmp_regs[2] = TMP3;
if (HAS_VIRTUAL_REGISTERS)
status.tmp_regs[2] = STR_END;
else
status.tmp_regs[2] = TMP3;
delayed_mem_copy_init(&status, common);
if (type != recurse_copy_shared_to_global && type != recurse_copy_kept_shared_to_global)
{
SLJIT_ASSERT(type == recurse_copy_from_global || type == recurse_copy_private_to_global || type == recurse_swap_global);
if (!from_sp)
delayed_mem_copy_move(&status, base_reg, stackptr, SLJIT_SP, common->recursive_head_ptr);
if (from_sp || type == recurse_swap_global)
delayed_mem_copy_move(&status, SLJIT_SP, common->recursive_head_ptr, base_reg, stackptr);
}
stackptr += sizeof(sljit_sw);
#if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD
if (type != recurse_copy_shared_to_global)
{
if (!from_sp)
delayed_mem_copy_move(&status, base_reg, stackptr, SLJIT_SP, common->control_head_ptr);
if (from_sp || type == recurse_swap_global)
delayed_mem_copy_move(&status, SLJIT_SP, common->control_head_ptr, base_reg, stackptr);
}
stackptr += sizeof(sljit_sw);
#endif
while (cc < ccend)
{
private_count = 0;
shared_count = 0;
kept_shared_count = 0;
switch(*cc)
{
case OP_SET_SOM:
SLJIT_ASSERT(common->has_set_som);
if ((recurse_flags & recurse_flag_quit_found) && recurse_check_bit(common, OVECTOR(0)))
{
kept_shared_srcw[0] = OVECTOR(0);
kept_shared_count = 1;
}
cc += 1;
break;
case OP_RECURSE:
if (recurse_flags & recurse_flag_quit_found)
{
if (common->has_set_som && recurse_check_bit(common, OVECTOR(0)))
{
kept_shared_srcw[0] = OVECTOR(0);
kept_shared_count = 1;
}
if (common->mark_ptr != 0 && recurse_check_bit(common, common->mark_ptr))
{
kept_shared_srcw[kept_shared_count] = common->mark_ptr;
kept_shared_count++;
}
}
if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr))
{
shared_srcw[0] = common->capture_last_ptr;
shared_count = 1;
}
cc += 1 + LINK_SIZE;
break;
case OP_KET:
private_srcw[0] = PRIVATE_DATA(cc);
if (private_srcw[0] != 0)
{
if (recurse_check_bit(common, private_srcw[0]))
private_count = 1;
SLJIT_ASSERT(PRIVATE_DATA(cc + 1) != 0);
cc += PRIVATE_DATA(cc + 1);
}
cc += 1 + LINK_SIZE;
break;
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
case OP_ASSERT_NA:
case OP_ASSERTBACK_NA:
case OP_ONCE:
case OP_SCRIPT_RUN:
case OP_BRAPOS:
case OP_SBRA:
case OP_SBRAPOS:
case OP_SCOND:
private_srcw[0] = PRIVATE_DATA(cc);
if (recurse_check_bit(common, private_srcw[0]))
private_count = 1;
cc += 1 + LINK_SIZE;
break;
case OP_ASSERT_SCS:
private_srcw[0] = PRIVATE_DATA(cc);
private_srcw[1] = private_srcw[0] + sizeof(sljit_sw);
if (recurse_check_bit(common, private_srcw[0]))
private_count = 2;
cc += 1 + LINK_SIZE;
break;
case OP_CBRA:
case OP_SCBRA:
offset = GET2(cc, 1 + LINK_SIZE);
shared_srcw[0] = OVECTOR(offset << 1);
if (recurse_check_bit(common, shared_srcw[0]))
{
shared_srcw[1] = shared_srcw[0] + sizeof(sljit_sw);
SLJIT_ASSERT(recurse_check_bit(common, shared_srcw[1]));
shared_count = 2;
}
if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr))
{
shared_srcw[shared_count] = common->capture_last_ptr;
shared_count++;
}
if (common->optimized_cbracket[offset] == 0)
{
private_srcw[0] = OVECTOR_PRIV(offset);
if (recurse_check_bit(common, private_srcw[0]))
private_count = 1;
}
cc += 1 + LINK_SIZE + IMM2_SIZE;
break;
case OP_CBRAPOS:
case OP_SCBRAPOS:
offset = GET2(cc, 1 + LINK_SIZE);
shared_srcw[0] = OVECTOR(offset << 1);
if (recurse_check_bit(common, shared_srcw[0]))
{
shared_srcw[1] = shared_srcw[0] + sizeof(sljit_sw);
SLJIT_ASSERT(recurse_check_bit(common, shared_srcw[1]));
shared_count = 2;
}
if (common->capture_last_ptr != 0 && recurse_check_bit(common, common->capture_last_ptr))
{
shared_srcw[shared_count] = common->capture_last_ptr;
shared_count++;
}
private_srcw[0] = PRIVATE_DATA(cc);
if (recurse_check_bit(common, private_srcw[0]))
private_count = 1;
offset = OVECTOR_PRIV(offset);
if (recurse_check_bit(common, offset))
{
private_srcw[private_count] = offset;
private_count++;
}
cc += 1 + LINK_SIZE + IMM2_SIZE;
break;
case OP_COND:
/* Might be a hidden SCOND. */
alternative = cc + GET(cc, 1);
if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN)
{
private_srcw[0] = PRIVATE_DATA(cc);
if (recurse_check_bit(common, private_srcw[0]))
private_count = 1;
}
cc += 1 + LINK_SIZE;
break;
CASE_ITERATOR_PRIVATE_DATA_1
private_srcw[0] = PRIVATE_DATA(cc);
if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0]))
private_count = 1;
cc += 2;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
CASE_ITERATOR_PRIVATE_DATA_2A
private_srcw[0] = PRIVATE_DATA(cc);
if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0]))
{
private_count = 2;
private_srcw[1] = private_srcw[0] + sizeof(sljit_sw);
SLJIT_ASSERT(recurse_check_bit(common, private_srcw[1]));
}
cc += 2;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
CASE_ITERATOR_PRIVATE_DATA_2B
private_srcw[0] = PRIVATE_DATA(cc);
if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0]))
{
private_count = 2;
private_srcw[1] = private_srcw[0] + sizeof(sljit_sw);
SLJIT_ASSERT(recurse_check_bit(common, private_srcw[1]));
}
cc += 2 + IMM2_SIZE;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
CASE_ITERATOR_TYPE_PRIVATE_DATA_1
private_srcw[0] = PRIVATE_DATA(cc);
if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0]))
private_count = 1;
cc += 1;
break;
CASE_ITERATOR_TYPE_PRIVATE_DATA_2A
private_srcw[0] = PRIVATE_DATA(cc);
if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0]))
{
private_count = 2;
private_srcw[1] = private_srcw[0] + sizeof(sljit_sw);
SLJIT_ASSERT(recurse_check_bit(common, private_srcw[1]));
}
cc += 1;
break;
CASE_ITERATOR_TYPE_PRIVATE_DATA_2B
private_srcw[0] = PRIVATE_DATA(cc);
if (private_srcw[0] != 0 && recurse_check_bit(common, private_srcw[0]))
{
private_count = 2;
private_srcw[1] = private_srcw[0] + sizeof(sljit_sw);
SLJIT_ASSERT(recurse_check_bit(common, private_srcw[1]));
}
cc += 1 + IMM2_SIZE;
break;
case OP_CLASS:
case OP_NCLASS:
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH != 8
case OP_XCLASS:
case OP_ECLASS:
i = (*cc >= OP_XCLASS) ? GET(cc, 1) : 1 + 32 / (int)sizeof(PCRE2_UCHAR);
#else
i = 1 + 32 / (int)sizeof(PCRE2_UCHAR);
#endif
if (PRIVATE_DATA(cc) != 0)
{
private_count = 1;
private_srcw[0] = PRIVATE_DATA(cc);
switch(get_class_iterator_size(cc + i))
{
case 1:
break;
case 2:
if (recurse_check_bit(common, private_srcw[0]))
{
private_count = 2;
private_srcw[1] = private_srcw[0] + sizeof(sljit_sw);
SLJIT_ASSERT(recurse_check_bit(common, private_srcw[1]));
}
break;
default:
SLJIT_UNREACHABLE();
break;
}
}
cc += i;
break;
case OP_MARK:
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_THEN_ARG:
SLJIT_ASSERT(common->mark_ptr != 0);
if ((recurse_flags & recurse_flag_quit_found) && recurse_check_bit(common, common->mark_ptr))
{
kept_shared_srcw[0] = common->mark_ptr;
kept_shared_count = 1;
}
if (common->control_head_ptr != 0 && recurse_check_bit(common, common->control_head_ptr))
{
private_srcw[0] = common->control_head_ptr;
private_count = 1;
}
cc += 1 + 2 + cc[1];
break;
case OP_THEN:
SLJIT_ASSERT(common->control_head_ptr != 0);
if (recurse_check_bit(common, common->control_head_ptr))
{
private_srcw[0] = common->control_head_ptr;
private_count = 1;
}
cc++;
break;
default:
cc = next_opcode(common, cc);
SLJIT_ASSERT(cc != NULL);
continue;
}
if (type != recurse_copy_shared_to_global && type != recurse_copy_kept_shared_to_global)
{
SLJIT_ASSERT(type == recurse_copy_from_global || type == recurse_copy_private_to_global || type == recurse_swap_global);
for (i = 0; i < private_count; i++)
{
SLJIT_ASSERT(private_srcw[i] != 0);
if (!from_sp)
delayed_mem_copy_move(&status, base_reg, stackptr, SLJIT_SP, private_srcw[i]);
if (from_sp || type == recurse_swap_global)
delayed_mem_copy_move(&status, SLJIT_SP, private_srcw[i], base_reg, stackptr);
stackptr += sizeof(sljit_sw);
}
}
else
stackptr += sizeof(sljit_sw) * private_count;
if (type != recurse_copy_private_to_global && type != recurse_copy_kept_shared_to_global)
{
SLJIT_ASSERT(type == recurse_copy_from_global || type == recurse_copy_shared_to_global || type == recurse_swap_global);
for (i = 0; i < shared_count; i++)
{
SLJIT_ASSERT(shared_srcw[i] != 0);
if (!from_sp)
delayed_mem_copy_move(&status, base_reg, stackptr, SLJIT_SP, shared_srcw[i]);
if (from_sp || type == recurse_swap_global)
delayed_mem_copy_move(&status, SLJIT_SP, shared_srcw[i], base_reg, stackptr);
stackptr += sizeof(sljit_sw);
}
}
else
stackptr += sizeof(sljit_sw) * shared_count;
if (type != recurse_copy_private_to_global && type != recurse_swap_global)
{
SLJIT_ASSERT(type == recurse_copy_from_global || type == recurse_copy_shared_to_global || type == recurse_copy_kept_shared_to_global);
for (i = 0; i < kept_shared_count; i++)
{
SLJIT_ASSERT(kept_shared_srcw[i] != 0);
if (!from_sp)
delayed_mem_copy_move(&status, base_reg, stackptr, SLJIT_SP, kept_shared_srcw[i]);
if (from_sp || type == recurse_swap_global)
delayed_mem_copy_move(&status, SLJIT_SP, kept_shared_srcw[i], base_reg, stackptr);
stackptr += sizeof(sljit_sw);
}
}
else
stackptr += sizeof(sljit_sw) * kept_shared_count;
}
SLJIT_ASSERT(cc == ccend && stackptr == stacktop);
delayed_mem_copy_finish(&status);
}
static SLJIT_INLINE PCRE2_SPTR set_then_offsets(compiler_common *common, PCRE2_SPTR cc, sljit_u8 *current_offset)
{
PCRE2_SPTR end = bracketend(cc);
BOOL has_alternatives = cc[GET(cc, 1)] == OP_ALT;
/* Assert captures *THEN verb even if it has no alternatives. */
if (*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT)
current_offset = NULL;
else if (*cc >= OP_ASSERT_NA && *cc <= OP_ASSERT_SCS)
has_alternatives = TRUE;
/* Conditional block does never capture. */
else if (*cc == OP_COND || *cc == OP_SCOND)
has_alternatives = FALSE;
cc = next_opcode(common, cc);
if (has_alternatives)
{
switch (*cc)
{
case OP_REVERSE:
case OP_CREF:
cc += 1 + IMM2_SIZE;
break;
case OP_VREVERSE:
case OP_DNCREF:
cc += 1 + 2 * IMM2_SIZE;
break;
}
current_offset = common->then_offsets + (cc - common->start);
}
while (cc < end)
{
if (*cc >= OP_ASSERT && *cc <= OP_SCOND)
{
cc = set_then_offsets(common, cc, current_offset);
continue;
}
if (*cc == OP_ALT && has_alternatives)
{
cc += 1 + LINK_SIZE;
if (*cc == OP_REVERSE)
cc += 1 + IMM2_SIZE;
else if (*cc == OP_VREVERSE)
cc += 1 + 2 * IMM2_SIZE;
current_offset = common->then_offsets + (cc - common->start);
continue;
}
if (*cc >= OP_THEN && *cc <= OP_THEN_ARG && current_offset != NULL)
*current_offset = 1;
cc = next_opcode(common, cc);
}
cc = end - 1 - LINK_SIZE;
/* Ignore repeats. */
if (*cc == OP_KET && PRIVATE_DATA(cc) != 0)
end += PRIVATE_DATA(cc + 1);
return end;
}
#undef CASE_ITERATOR_PRIVATE_DATA_1
#undef CASE_ITERATOR_PRIVATE_DATA_2A
#undef CASE_ITERATOR_PRIVATE_DATA_2B
#undef CASE_ITERATOR_TYPE_PRIVATE_DATA_1
#undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2A
#undef CASE_ITERATOR_TYPE_PRIVATE_DATA_2B
static SLJIT_INLINE BOOL is_powerof2(unsigned int value)
{
return (value & (value - 1)) == 0;
}
static SLJIT_INLINE void set_jumps(jump_list *list, struct sljit_label *label)
{
while (list != NULL)
{
/* sljit_set_label is clever enough to do nothing
if either the jump or the label is NULL. */
SET_LABEL(list->jump, label);
list = list->next;
}
}
static SLJIT_INLINE void add_jump(struct sljit_compiler *compiler, jump_list **list, struct sljit_jump *jump)
{
jump_list *list_item = sljit_alloc_memory(compiler, sizeof(jump_list));
if (list_item)
{
list_item->next = *list;
list_item->jump = jump;
*list = list_item;
}
}
static void add_stub(compiler_common *common, struct sljit_jump *start)
{
DEFINE_COMPILER;
stub_list *list_item = sljit_alloc_memory(compiler, sizeof(stub_list));
if (list_item)
{
list_item->start = start;
list_item->quit = LABEL();
list_item->next = common->stubs;
common->stubs = list_item;
}
}
static void flush_stubs(compiler_common *common)
{
DEFINE_COMPILER;
stub_list *list_item = common->stubs;
while (list_item)
{
JUMPHERE(list_item->start);
add_jump(compiler, &common->stackalloc, JUMP(SLJIT_FAST_CALL));
JUMPTO(SLJIT_JUMP, list_item->quit);
list_item = list_item->next;
}
common->stubs = NULL;
}
static SLJIT_INLINE void count_match(compiler_common *common)
{
DEFINE_COMPILER;
OP2(SLJIT_SUB | SLJIT_SET_Z, COUNT_MATCH, 0, COUNT_MATCH, 0, SLJIT_IMM, 1);
add_jump(compiler, &common->calllimit, JUMP(SLJIT_ZERO));
}
static SLJIT_INLINE void allocate_stack(compiler_common *common, int size)
{
/* May destroy all locals and registers except TMP2. */
DEFINE_COMPILER;
SLJIT_ASSERT(size > 0);
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * SSIZE_OF(sw));
#ifdef DESTROY_REGISTERS
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 12345);
OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0);
#if defined SLJIT_DEBUG && SLJIT_DEBUG
SLJIT_ASSERT(common->locals_size >= 2 * SSIZE_OF(sw));
/* These two are also used by the stackalloc calls. */
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL0, TMP1, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL1, TMP1, 0);
#endif
#endif
add_stub(common, CMP(SLJIT_LESS, STACK_TOP, 0, STACK_LIMIT, 0));
}
static SLJIT_INLINE void free_stack(compiler_common *common, int size)
{
DEFINE_COMPILER;
SLJIT_ASSERT(size > 0);
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, size * SSIZE_OF(sw));
}
static sljit_uw * allocate_read_only_data(compiler_common *common, sljit_uw size)
{
DEFINE_COMPILER;
sljit_uw *result;
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
return NULL;
result = (sljit_uw *)SLJIT_MALLOC(size + sizeof(sljit_uw), compiler->allocator_data);
if (SLJIT_UNLIKELY(result == NULL))
{
sljit_set_compiler_memory_error(compiler);
return NULL;
}
*(void**)result = common->read_only_data_head;
common->read_only_data_head = (void *)result;
return result + 1;
}
static SLJIT_INLINE void reset_ovector(compiler_common *common, int length)
{
DEFINE_COMPILER;
struct sljit_label *loop;
sljit_s32 i;
/* At this point we can freely use all temporary registers. */
SLJIT_ASSERT(length > 1);
/* TMP1 returns with begin - 1. */
OP2(SLJIT_SUB, SLJIT_R0, 0, SLJIT_MEM1(SLJIT_S0), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1));
if (length < 8)
{
for (i = 1; i < length; i++)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(i), SLJIT_R0, 0);
}
else
{
if (sljit_emit_mem_update(compiler, SLJIT_MOV | SLJIT_MEM_SUPP | SLJIT_MEM_STORE | SLJIT_MEM_PRE, SLJIT_R0, SLJIT_MEM1(SLJIT_R1), sizeof(sljit_sw)) == SLJIT_SUCCESS)
{
GET_LOCAL_BASE(SLJIT_R1, 0, OVECTOR_START);
OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_IMM, length - 1);
loop = LABEL();
sljit_emit_mem_update(compiler, SLJIT_MOV | SLJIT_MEM_STORE | SLJIT_MEM_PRE, SLJIT_R0, SLJIT_MEM1(SLJIT_R1), sizeof(sljit_sw));
OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, loop);
}
else
{
GET_LOCAL_BASE(SLJIT_R1, 0, OVECTOR_START + sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_IMM, length - 1);
loop = LABEL();
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_R1), 0, SLJIT_R0, 0);
OP2(SLJIT_ADD, SLJIT_R1, 0, SLJIT_R1, 0, SLJIT_IMM, sizeof(sljit_sw));
OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, loop);
}
}
}
static SLJIT_INLINE void reset_early_fail(compiler_common *common)
{
DEFINE_COMPILER;
sljit_u32 size = (sljit_u32)(common->early_fail_end_ptr - common->early_fail_start_ptr);
sljit_u32 uncleared_size;
sljit_s32 src = SLJIT_IMM;
sljit_s32 i;
struct sljit_label *loop;
SLJIT_ASSERT(common->early_fail_start_ptr < common->early_fail_end_ptr);
if (size == sizeof(sljit_sw))
{
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->early_fail_start_ptr, SLJIT_IMM, 0);
return;
}
if (sljit_get_register_index(SLJIT_GP_REGISTER, TMP3) >= 0 && !sljit_has_cpu_feature(SLJIT_HAS_ZERO_REGISTER))
{
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 0);
src = TMP3;
}
if (size <= 6 * sizeof(sljit_sw))
{
for (i = common->early_fail_start_ptr; i < common->early_fail_end_ptr; i += sizeof(sljit_sw))
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), i, src, 0);
return;
}
GET_LOCAL_BASE(TMP1, 0, common->early_fail_start_ptr);
uncleared_size = ((size / sizeof(sljit_sw)) % 3) * sizeof(sljit_sw);
OP2(SLJIT_ADD, TMP2, 0, TMP1, 0, SLJIT_IMM, size - uncleared_size);
loop = LABEL();
OP1(SLJIT_MOV, SLJIT_MEM1(TMP1), 0, src, 0);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 3 * sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(TMP1), -2 * SSIZE_OF(sw), src, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(TMP1), -1 * SSIZE_OF(sw), src, 0);
CMPTO(SLJIT_LESS, TMP1, 0, TMP2, 0, loop);
if (uncleared_size >= sizeof(sljit_sw))
OP1(SLJIT_MOV, SLJIT_MEM1(TMP1), 0, src, 0);
if (uncleared_size >= 2 * sizeof(sljit_sw))
OP1(SLJIT_MOV, SLJIT_MEM1(TMP1), sizeof(sljit_sw), src, 0);
}
static SLJIT_INLINE void do_reset_match(compiler_common *common, int length)
{
DEFINE_COMPILER;
struct sljit_label *loop;
int i;
SLJIT_ASSERT(length > 1);
/* OVECTOR(1) contains the "string begin - 1" constant. */
if (length > 2)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1));
if (length < 8)
{
for (i = 2; i < length; i++)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(i), TMP1, 0);
}
else
{
if (sljit_emit_mem_update(compiler, SLJIT_MOV | SLJIT_MEM_SUPP | SLJIT_MEM_STORE | SLJIT_MEM_PRE, TMP1, SLJIT_MEM1(TMP2), sizeof(sljit_sw)) == SLJIT_SUCCESS)
{
GET_LOCAL_BASE(TMP2, 0, OVECTOR_START + sizeof(sljit_sw));
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_IMM, length - 2);
loop = LABEL();
sljit_emit_mem_update(compiler, SLJIT_MOV | SLJIT_MEM_STORE | SLJIT_MEM_PRE, TMP1, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
OP2(SLJIT_SUB | SLJIT_SET_Z, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, loop);
}
else
{
GET_LOCAL_BASE(TMP2, 0, OVECTOR_START + 2 * sizeof(sljit_sw));
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_IMM, length - 2);
loop = LABEL();
OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, TMP1, 0);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, sizeof(sljit_sw));
OP2(SLJIT_SUB | SLJIT_SET_Z, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, loop);
}
}
if (!HAS_VIRTUAL_REGISTERS)
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, stack));
else
OP1(SLJIT_MOV, STACK_TOP, 0, ARGUMENTS, 0);
if (common->mark_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->mark_ptr, SLJIT_IMM, 0);
if (common->control_head_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_IMM, 0);
if (HAS_VIRTUAL_REGISTERS)
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(jit_arguments, stack));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->start_ptr);
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(struct sljit_stack, end));
}
static sljit_sw SLJIT_FUNC do_search_mark(sljit_sw *current, PCRE2_SPTR skip_arg)
{
while (current != NULL)
{
switch (current[1])
{
case type_then_trap:
break;
case type_mark:
if (PRIV(strcmp)(skip_arg, (PCRE2_SPTR)current[2]) == 0)
return current[3];
break;
default:
SLJIT_UNREACHABLE();
break;
}
SLJIT_ASSERT(current[0] == 0 || current < (sljit_sw*)current[0]);
current = (sljit_sw*)current[0];
}
return 0;
}
static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket)
{
DEFINE_COMPILER;
struct sljit_label *loop;
BOOL has_pre;
/* At this point we can freely use all registers. */
OP1(SLJIT_MOV, SLJIT_S2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(1), STR_PTR, 0);
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, SLJIT_R0, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, SLJIT_S0, 0, SLJIT_MEM1(SLJIT_SP), common->start_ptr);
if (common->mark_ptr != 0)
OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_SP), common->mark_ptr);
OP1(SLJIT_MOV_U32, SLJIT_R1, 0, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, oveccount));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, startchar_ptr), SLJIT_S0, 0);
if (common->mark_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_R2, 0);
OP2(SLJIT_ADD, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, match_data),
SLJIT_IMM, SLJIT_OFFSETOF(pcre2_match_data, ovector) - sizeof(PCRE2_SIZE));
}
else
{
OP1(SLJIT_MOV, SLJIT_S0, 0, SLJIT_MEM1(SLJIT_SP), common->start_ptr);
OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, match_data));
if (common->mark_ptr != 0)
OP1(SLJIT_MOV, SLJIT_R0, 0, SLJIT_MEM1(SLJIT_SP), common->mark_ptr);
OP1(SLJIT_MOV_U32, SLJIT_R1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, oveccount));
OP1(SLJIT_MOV, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, startchar_ptr), SLJIT_S0, 0);
if (common->mark_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, mark_ptr), SLJIT_R0, 0);
OP2(SLJIT_ADD, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, SLJIT_OFFSETOF(pcre2_match_data, ovector) - sizeof(PCRE2_SIZE));
}
has_pre = sljit_emit_mem_update(compiler, SLJIT_MOV | SLJIT_MEM_SUPP | SLJIT_MEM_PRE, SLJIT_S1, SLJIT_MEM1(SLJIT_S0), sizeof(sljit_sw)) == SLJIT_SUCCESS;
GET_LOCAL_BASE(SLJIT_S0, 0, OVECTOR_START - (has_pre ? sizeof(sljit_sw) : 0));
OP1(SLJIT_MOV, SLJIT_R0, 0, SLJIT_MEM1(HAS_VIRTUAL_REGISTERS ? SLJIT_R0 : ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin));
loop = LABEL();
if (has_pre)
sljit_emit_mem_update(compiler, SLJIT_MOV | SLJIT_MEM_PRE, SLJIT_S1, SLJIT_MEM1(SLJIT_S0), sizeof(sljit_sw));
else
{
OP1(SLJIT_MOV, SLJIT_S1, 0, SLJIT_MEM1(SLJIT_S0), 0);
OP2(SLJIT_ADD, SLJIT_S0, 0, SLJIT_S0, 0, SLJIT_IMM, sizeof(sljit_sw));
}
OP2(SLJIT_ADD, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, sizeof(PCRE2_SIZE));
OP2(SLJIT_SUB, SLJIT_S1, 0, SLJIT_S1, 0, SLJIT_R0, 0);
/* Copy the integer value to the output buffer */
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_ASHR, SLJIT_S1, 0, SLJIT_S1, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif
SLJIT_ASSERT(sizeof(PCRE2_SIZE) == 4 || sizeof(PCRE2_SIZE) == 8);
OP1(((sizeof(PCRE2_SIZE) == 4) ? SLJIT_MOV_U32 : SLJIT_MOV), SLJIT_MEM1(SLJIT_R2), 0, SLJIT_S1, 0);
OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_R1, 0, SLJIT_R1, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, loop);
/* Calculate the return value, which is the maximum ovector value. */
if (topbracket > 1)
{
if (sljit_emit_mem_update(compiler, SLJIT_MOV | SLJIT_MEM_SUPP | SLJIT_MEM_PRE, SLJIT_R2, SLJIT_MEM1(SLJIT_R0), -(2 * SSIZE_OF(sw))) == SLJIT_SUCCESS)
{
GET_LOCAL_BASE(SLJIT_R0, 0, OVECTOR_START + topbracket * 2 * sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_R1, 0, SLJIT_IMM, topbracket + 1);
/* OVECTOR(0) is never equal to SLJIT_S2. */
loop = LABEL();
sljit_emit_mem_update(compiler, SLJIT_MOV | SLJIT_MEM_PRE, SLJIT_R2, SLJIT_MEM1(SLJIT_R0), -(2 * SSIZE_OF(sw)));
OP2(SLJIT_SUB, SLJIT_R1, 0, SLJIT_R1, 0, SLJIT_IMM, 1);
CMPTO(SLJIT_EQUAL, SLJIT_R2, 0, SLJIT_S2, 0, loop);
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_R1, 0);
}
else
{
GET_LOCAL_BASE(SLJIT_R0, 0, OVECTOR_START + (topbracket - 1) * 2 * sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_R1, 0, SLJIT_IMM, topbracket + 1);
/* OVECTOR(0) is never equal to SLJIT_S2. */
loop = LABEL();
OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_R0), 0);
OP2(SLJIT_SUB, SLJIT_R0, 0, SLJIT_R0, 0, SLJIT_IMM, 2 * SSIZE_OF(sw));
OP2(SLJIT_SUB, SLJIT_R1, 0, SLJIT_R1, 0, SLJIT_IMM, 1);
CMPTO(SLJIT_EQUAL, SLJIT_R2, 0, SLJIT_S2, 0, loop);
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_R1, 0);
}
}
else
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, 1);
}
static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit)
{
DEFINE_COMPILER;
sljit_s32 mov_opcode;
sljit_s32 arguments_reg = !HAS_VIRTUAL_REGISTERS ? ARGUMENTS : SLJIT_R1;
SLJIT_COMPILE_ASSERT(STR_END == SLJIT_S0, str_end_must_be_saved_reg0);
SLJIT_ASSERT(common->start_used_ptr != 0 && common->start_ptr != 0
&& (common->mode == PCRE2_JIT_PARTIAL_SOFT ? common->hit_start != 0 : common->hit_start == 0));
if (arguments_reg != ARGUMENTS)
OP1(SLJIT_MOV, arguments_reg, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, SLJIT_R2, 0, SLJIT_MEM1(SLJIT_SP),
common->mode == PCRE2_JIT_PARTIAL_SOFT ? common->hit_start : common->start_ptr);
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE2_ERROR_PARTIAL);
/* Store match begin and end. */
OP1(SLJIT_MOV, SLJIT_S1, 0, SLJIT_MEM1(arguments_reg), SLJIT_OFFSETOF(jit_arguments, begin));
OP1(SLJIT_MOV, SLJIT_MEM1(arguments_reg), SLJIT_OFFSETOF(jit_arguments, startchar_ptr), SLJIT_R2, 0);
OP1(SLJIT_MOV, SLJIT_R1, 0, SLJIT_MEM1(arguments_reg), SLJIT_OFFSETOF(jit_arguments, match_data));
mov_opcode = (sizeof(PCRE2_SIZE) == 4) ? SLJIT_MOV_U32 : SLJIT_MOV;
OP2(SLJIT_SUB, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_S1, 0);
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_ASHR, SLJIT_R2, 0, SLJIT_R2, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif
OP1(mov_opcode, SLJIT_MEM1(SLJIT_R1), SLJIT_OFFSETOF(pcre2_match_data, ovector), SLJIT_R2, 0);
OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_S1, 0);
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_ASHR, STR_END, 0, STR_END, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif
OP1(mov_opcode, SLJIT_MEM1(SLJIT_R1), SLJIT_OFFSETOF(pcre2_match_data, ovector) + sizeof(PCRE2_SIZE), STR_END, 0);
JUMPTO(SLJIT_JUMP, quit);
}
static SLJIT_INLINE void check_start_used_ptr(compiler_common *common)
{
/* May destroy TMP1. */
DEFINE_COMPILER;
struct sljit_jump *jump;
if (common->mode == PCRE2_JIT_PARTIAL_SOFT)
{
/* The value of -1 must be kept for start_used_ptr! */
OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, SLJIT_IMM, 1);
/* Jumps if start_used_ptr < STR_PTR, or start_used_ptr == -1. Although overwriting
is not necessary if start_used_ptr == STR_PTR, it does not hurt as well. */
jump = CMP(SLJIT_LESS_EQUAL, TMP1, 0, STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0);
JUMPHERE(jump);
}
else if (common->mode == PCRE2_JIT_PARTIAL_HARD)
{
jump = CMP(SLJIT_LESS_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0);
JUMPHERE(jump);
}
}
static SLJIT_INLINE BOOL char_has_othercase(compiler_common *common, PCRE2_SPTR cc)
{
/* Detects if the character has an othercase. */
unsigned int c;
#ifdef SUPPORT_UNICODE
if (common->utf || common->ucp)
{
if (common->utf)
{
GETCHAR(c, cc);
}
else
c = *cc;
if (c > 127)
return c != UCD_OTHERCASE(c);
return common->fcc[c] != c;
}
else
#endif
c = *cc;
return MAX_255(c) ? common->fcc[c] != c : FALSE;
}
static SLJIT_INLINE unsigned int char_othercase(compiler_common *common, unsigned int c)
{
/* Returns with the othercase. */
#ifdef SUPPORT_UNICODE
if ((common->utf || common->ucp) && c > 127)
return UCD_OTHERCASE(c);
#endif
return TABLE_GET(c, common->fcc, c);
}
static unsigned int char_get_othercase_bit(compiler_common *common, PCRE2_SPTR cc)
{
/* Detects if the character and its othercase has only 1 bit difference. */
unsigned int c, oc, bit;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
int n;
#endif
#ifdef SUPPORT_UNICODE
if (common->utf || common->ucp)
{
if (common->utf)
{
GETCHAR(c, cc);
}
else
c = *cc;
if (c <= 127)
oc = common->fcc[c];
else
oc = UCD_OTHERCASE(c);
}
else
{
c = *cc;
oc = TABLE_GET(c, common->fcc, c);
}
#else
c = *cc;
oc = TABLE_GET(c, common->fcc, c);
#endif
SLJIT_ASSERT(c != oc);
bit = c ^ oc;
/* Optimized for English alphabet. */
if (c <= 127 && bit == 0x20)
return (0 << 8) | 0x20;
/* Since c != oc, they must have at least 1 bit difference. */
if (!is_powerof2(bit))
return 0;
#if PCRE2_CODE_UNIT_WIDTH == 8
#ifdef SUPPORT_UNICODE
if (common->utf && c > 127)
{
n = GET_EXTRALEN(*cc);
while ((bit & 0x3f) == 0)
{
n--;
bit >>= 6;
}
return (n << 8) | bit;
}
#endif /* SUPPORT_UNICODE */
return (0 << 8) | bit;
#elif PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
#ifdef SUPPORT_UNICODE
if (common->utf && c > 65535)
{
if (bit >= (1u << 10))
bit >>= 10;
else
return (bit < 256) ? ((2 << 8) | bit) : ((3 << 8) | (bit >> 8));
}
#endif /* SUPPORT_UNICODE */
return (bit < 256) ? ((0u << 8) | bit) : ((1u << 8) | (bit >> 8));
#endif /* PCRE2_CODE_UNIT_WIDTH == [8|16|32] */
}
static void check_partial(compiler_common *common, BOOL force)
{
/* Checks whether a partial matching is occurred. Does not modify registers. */
DEFINE_COMPILER;
struct sljit_jump *jump = NULL;
SLJIT_ASSERT(!force || common->mode != PCRE2_JIT_COMPLETE);
if (common->mode == PCRE2_JIT_COMPLETE)
return;
if (!force && !common->allow_empty_partial)
jump = CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0);
else if (common->mode == PCRE2_JIT_PARTIAL_SOFT)
jump = CMP(SLJIT_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, SLJIT_IMM, -1);
if (common->mode == PCRE2_JIT_PARTIAL_SOFT)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, 0);
else
{
if (common->partialmatchlabel != NULL)
JUMPTO(SLJIT_JUMP, common->partialmatchlabel);
else
add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP));
}
if (jump != NULL)
JUMPHERE(jump);
}
static void check_str_end(compiler_common *common, jump_list **end_reached)
{
/* Does not affect registers. Usually used in a tight spot. */
DEFINE_COMPILER;
struct sljit_jump *jump;
if (common->mode == PCRE2_JIT_COMPLETE)
{
add_jump(compiler, end_reached, CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
return;
}
jump = CMP(SLJIT_LESS, STR_PTR, 0, STR_END, 0);
if (common->mode == PCRE2_JIT_PARTIAL_SOFT)
{
add_jump(compiler, end_reached, CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, 0);
add_jump(compiler, end_reached, JUMP(SLJIT_JUMP));
}
else
{
add_jump(compiler, end_reached, CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0));
if (common->partialmatchlabel != NULL)
JUMPTO(SLJIT_JUMP, common->partialmatchlabel);
else
add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP));
}
JUMPHERE(jump);
}
static void detect_partial_match(compiler_common *common, jump_list **backtracks)
{
DEFINE_COMPILER;
struct sljit_jump *jump;
if (common->mode == PCRE2_JIT_COMPLETE)
{
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
return;
}
/* Partial matching mode. */
jump = CMP(SLJIT_LESS, STR_PTR, 0, STR_END, 0);
if (!common->allow_empty_partial)
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0));
else if (common->mode == PCRE2_JIT_PARTIAL_SOFT)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, SLJIT_IMM, -1));
if (common->mode == PCRE2_JIT_PARTIAL_SOFT)
{
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, 0);
add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
}
else
{
if (common->partialmatchlabel != NULL)
JUMPTO(SLJIT_JUMP, common->partialmatchlabel);
else
add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP));
}
JUMPHERE(jump);
}
static void process_partial_match(compiler_common *common)
{
DEFINE_COMPILER;
struct sljit_jump *jump;
/* Partial matching mode. */
if (common->mode == PCRE2_JIT_PARTIAL_SOFT)
{
jump = CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, 0);
JUMPHERE(jump);
}
else if (common->mode == PCRE2_JIT_PARTIAL_HARD)
{
if (common->partialmatchlabel != NULL)
CMPTO(SLJIT_LESS, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0, common->partialmatchlabel);
else
add_jump(compiler, &common->partialmatch, CMP(SLJIT_LESS, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0));
}
}
static void detect_partial_match_to(compiler_common *common, struct sljit_label *label)
{
DEFINE_COMPILER;
CMPTO(SLJIT_LESS, STR_PTR, 0, STR_END, 0, label);
process_partial_match(common);
}
static void peek_char(compiler_common *common, sljit_u32 max, sljit_s32 dst, sljit_sw dstw, jump_list **backtracks)
{
/* Reads the character into TMP1, keeps STR_PTR.
Does not check STR_END. TMP2, dst, RETURN_ADDR Destroyed. */
DEFINE_COMPILER;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
struct sljit_jump *jump;
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32 */
SLJIT_UNUSED_ARG(max);
SLJIT_UNUSED_ARG(dst);
SLJIT_UNUSED_ARG(dstw);
SLJIT_UNUSED_ARG(backtracks);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
#ifdef SUPPORT_UNICODE
#if PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf)
{
if (max < 128) return;
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x80);
OP1(SLJIT_MOV, dst, dstw, STR_PTR, 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
add_jump(compiler, common->invalid_utf ? &common->utfreadchar_invalid : &common->utfreadchar, JUMP(SLJIT_FAST_CALL));
OP1(SLJIT_MOV, STR_PTR, 0, dst, dstw);
if (backtracks && common->invalid_utf)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR));
JUMPHERE(jump);
}
#elif PCRE2_CODE_UNIT_WIDTH == 16
if (common->utf)
{
if (max < 0xd800) return;
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
if (common->invalid_utf)
{
jump = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800);
OP1(SLJIT_MOV, dst, dstw, STR_PTR, 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
add_jump(compiler, &common->utfreadchar_invalid, JUMP(SLJIT_FAST_CALL));
OP1(SLJIT_MOV, STR_PTR, 0, dst, dstw);
if (backtracks && common->invalid_utf)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR));
}
else
{
/* TMP2 contains the high surrogate. */
jump = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000 - 0xdc00);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
}
JUMPHERE(jump);
}
#elif PCRE2_CODE_UNIT_WIDTH == 32
if (common->invalid_utf)
{
if (max < 0xd800) return;
if (backtracks != NULL)
{
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x110000));
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800));
}
else
{
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x110000);
SELECT(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, INVALID_UTF_CHAR, TMP1);
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800);
SELECT(SLJIT_LESS, TMP1, SLJIT_IMM, INVALID_UTF_CHAR, TMP1);
}
}
#endif /* PCRE2_CODE_UNIT_WIDTH == [8|16|32] */
#endif /* SUPPORT_UNICODE */
}
static void peek_char_back(compiler_common *common, sljit_u32 max, jump_list **backtracks)
{
/* Reads one character back without moving STR_PTR. TMP2 must
contain the start of the subject buffer. Affects TMP1, TMP2, and RETURN_ADDR. */
DEFINE_COMPILER;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
struct sljit_jump *jump;
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32 */
SLJIT_UNUSED_ARG(max);
SLJIT_UNUSED_ARG(backtracks);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
#ifdef SUPPORT_UNICODE
#if PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf)
{
if (max < 128) return;
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x80);
if (common->invalid_utf)
{
add_jump(compiler, &common->utfpeakcharback_invalid, JUMP(SLJIT_FAST_CALL));
if (backtracks != NULL)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR));
}
else
add_jump(compiler, &common->utfpeakcharback, JUMP(SLJIT_FAST_CALL));
JUMPHERE(jump);
}
#elif PCRE2_CODE_UNIT_WIDTH == 16
if (common->utf)
{
if (max < 0xd800) return;
if (common->invalid_utf)
{
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xd800);
add_jump(compiler, &common->utfpeakcharback_invalid, JUMP(SLJIT_FAST_CALL));
if (backtracks != NULL)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR));
}
else
{
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xdc00);
jump = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0xe000 - 0xdc00);
/* TMP2 contains the low surrogate. */
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x10000);
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xd800);
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 10);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
}
JUMPHERE(jump);
}
#elif PCRE2_CODE_UNIT_WIDTH == 32
if (common->invalid_utf)
{
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x110000));
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800));
}
#endif /* PCRE2_CODE_UNIT_WIDTH == [8|16|32] */
#endif /* SUPPORT_UNICODE */
}
#define READ_CHAR_UPDATE_STR_PTR 0x1
#define READ_CHAR_UTF8_NEWLINE 0x2
#define READ_CHAR_NEWLINE (READ_CHAR_UPDATE_STR_PTR | READ_CHAR_UTF8_NEWLINE)
#define READ_CHAR_VALID_UTF 0x4
static void read_char(compiler_common *common, sljit_u32 min, sljit_u32 max,
jump_list **backtracks, sljit_u32 options)
{
/* Reads the precise value of a character into TMP1, if the character is
between min and max (c >= min && c <= max). Otherwise it returns with a value
outside the range. Does not check STR_END. */
DEFINE_COMPILER;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
struct sljit_jump *jump;
#endif
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
struct sljit_jump *jump2;
#endif
SLJIT_UNUSED_ARG(min);
SLJIT_UNUSED_ARG(max);
SLJIT_UNUSED_ARG(backtracks);
SLJIT_UNUSED_ARG(options);
SLJIT_ASSERT(min <= max);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
#ifdef SUPPORT_UNICODE
#if PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf)
{
if (max < 128 && !(options & READ_CHAR_UPDATE_STR_PTR)) return;
if (common->invalid_utf && !(options & READ_CHAR_VALID_UTF))
{
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x80);
if (options & READ_CHAR_UTF8_NEWLINE)
add_jump(compiler, &common->utfreadnewline_invalid, JUMP(SLJIT_FAST_CALL));
else
add_jump(compiler, &common->utfreadchar_invalid, JUMP(SLJIT_FAST_CALL));
if (backtracks != NULL)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR));
JUMPHERE(jump);
return;
}
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
if (min >= 0x10000)
{
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xf0);
if (options & READ_CHAR_UPDATE_STR_PTR)
OP1(SLJIT_MOV_U8, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
jump2 = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 0x7);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));
if (!(options & READ_CHAR_UPDATE_STR_PTR))
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
JUMPHERE(jump2);
if (options & READ_CHAR_UPDATE_STR_PTR)
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0);
}
else if (min >= 0x800 && max <= 0xffff)
{
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xe0);
if (options & READ_CHAR_UPDATE_STR_PTR)
OP1(SLJIT_MOV_U8, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
jump2 = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 0xf);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
if (!(options & READ_CHAR_UPDATE_STR_PTR))
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
JUMPHERE(jump2);
if (options & READ_CHAR_UPDATE_STR_PTR)
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0);
}
else if (max >= 0x800)
{
add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));
}
else if (max < 128)
{
OP1(SLJIT_MOV_U8, TMP2, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
}
else
{
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
if (!(options & READ_CHAR_UPDATE_STR_PTR))
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
else
OP1(SLJIT_MOV_U8, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
if (options & READ_CHAR_UPDATE_STR_PTR)
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0);
}
JUMPHERE(jump);
}
#elif PCRE2_CODE_UNIT_WIDTH == 16
if (common->utf)
{
if (max < 0xd800 && !(options & READ_CHAR_UPDATE_STR_PTR)) return;
if (common->invalid_utf && !(options & READ_CHAR_VALID_UTF))
{
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
jump = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800);
if (options & READ_CHAR_UTF8_NEWLINE)
add_jump(compiler, &common->utfreadnewline_invalid, JUMP(SLJIT_FAST_CALL));
else
add_jump(compiler, &common->utfreadchar_invalid, JUMP(SLJIT_FAST_CALL));
if (backtracks != NULL)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR));
JUMPHERE(jump);
return;
}
if (max >= 0x10000)
{
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
jump = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800);
/* TMP2 contains the high surrogate. */
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000 - 0xdc00);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
JUMPHERE(jump);
return;
}
/* Skip low surrogate if necessary. */
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
if (sljit_has_cpu_feature(SLJIT_HAS_CMOV) && !HAS_VIRTUAL_REGISTERS)
{
if (options & READ_CHAR_UPDATE_STR_PTR)
OP2(SLJIT_ADD, RETURN_ADDR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, SLJIT_IMM, 0x400);
if (options & READ_CHAR_UPDATE_STR_PTR)
SELECT(SLJIT_LESS, STR_PTR, RETURN_ADDR, 0, STR_PTR);
if (max >= 0xd800)
SELECT(SLJIT_LESS, TMP1, SLJIT_IMM, 0x10000, TMP1);
}
else
{
jump = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x400);
if (options & READ_CHAR_UPDATE_STR_PTR)
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (max >= 0xd800)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0x10000);
JUMPHERE(jump);
}
}
#elif PCRE2_CODE_UNIT_WIDTH == 32
if (common->invalid_utf)
{
if (backtracks != NULL)
{
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x110000));
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800));
}
else
{
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x110000);
SELECT(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, INVALID_UTF_CHAR, TMP1);
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800);
SELECT(SLJIT_LESS, TMP1, SLJIT_IMM, INVALID_UTF_CHAR, TMP1);
}
}
#endif /* PCRE2_CODE_UNIT_WIDTH == [8|16|32] */
#endif /* SUPPORT_UNICODE */
}
static void skip_valid_char(compiler_common *common)
{
DEFINE_COMPILER;
#if (defined SUPPORT_UNICODE) && (PCRE2_CODE_UNIT_WIDTH == 8 || PCRE2_CODE_UNIT_WIDTH == 16)
struct sljit_jump *jump;
#endif
#if (defined SUPPORT_UNICODE) && (PCRE2_CODE_UNIT_WIDTH == 8 || PCRE2_CODE_UNIT_WIDTH == 16)
if (common->utf)
{
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
#if PCRE2_CODE_UNIT_WIDTH == 8
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
#elif PCRE2_CODE_UNIT_WIDTH == 16
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xd800);
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0xd800);
OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_EQUAL);
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */
JUMPHERE(jump);
return;
}
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == [8|16] */
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
}
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
static BOOL is_char7_bitset(const sljit_u8 *bitset, BOOL nclass)
{
/* Tells whether the character codes below 128 are enough
to determine a match. */
const sljit_u8 value = nclass ? 0xff : 0;
const sljit_u8 *end = bitset + 32;
bitset += 16;
do
{
if (*bitset++ != value)
return FALSE;
}
while (bitset < end);
return TRUE;
}
static void read_char7_type(compiler_common *common, jump_list **backtracks, BOOL negated)
{
/* Reads the precise character type of a character into TMP1, if the character
is less than 128. Otherwise it returns with zero. Does not check STR_END. The
full_read argument tells whether characters above max are accepted or not. */
DEFINE_COMPILER;
struct sljit_jump *jump;
SLJIT_ASSERT(common->utf);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
/* All values > 127 are zero in ctypes. */
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
if (negated)
{
jump = CMP(SLJIT_LESS, TMP2, 0, SLJIT_IMM, 0x80);
if (common->invalid_utf)
{
OP1(SLJIT_MOV, TMP1, 0, TMP2, 0);
add_jump(compiler, &common->utfreadchar_invalid, JUMP(SLJIT_FAST_CALL));
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
}
else
{
OP1(SLJIT_MOV_U8, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
}
JUMPHERE(jump);
}
}
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8 */
static void read_char8_type(compiler_common *common, jump_list **backtracks, BOOL negated)
{
/* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */
DEFINE_COMPILER;
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH != 8
struct sljit_jump *jump;
#endif
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
struct sljit_jump *jump2;
#endif
SLJIT_UNUSED_ARG(backtracks);
SLJIT_UNUSED_ARG(negated);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf)
{
/* The result of this read may be unused, but saves an "else" part. */
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
jump = CMP(SLJIT_LESS, TMP2, 0, SLJIT_IMM, 0x80);
if (!negated)
{
if (common->invalid_utf)
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xc2);
if (common->invalid_utf)
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0xe0 - 0xc2));
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP1, 0);
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x80);
if (common->invalid_utf)
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
jump2 = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 255);
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
JUMPHERE(jump2);
}
else if (common->invalid_utf)
{
add_jump(compiler, &common->utfreadchar_invalid, JUMP(SLJIT_FAST_CALL));
OP1(SLJIT_MOV, TMP2, 0, TMP1, 0);
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
jump2 = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 255);
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
JUMPHERE(jump2);
}
else
add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL));
JUMPHERE(jump);
return;
}
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8 */
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 32
if (common->invalid_utf && negated)
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x110000));
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 32 */
#if PCRE2_CODE_UNIT_WIDTH != 8
/* The ctypes array contains only 256 values. */
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
jump = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 255);
#endif /* PCRE2_CODE_UNIT_WIDTH != 8 */
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
#if PCRE2_CODE_UNIT_WIDTH != 8
JUMPHERE(jump);
#endif /* PCRE2_CODE_UNIT_WIDTH != 8 */
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 16
if (common->utf && negated)
{
/* Skip low surrogate if necessary. */
if (!common->invalid_utf)
{
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xd800);
if (sljit_has_cpu_feature(SLJIT_HAS_CMOV) && !HAS_VIRTUAL_REGISTERS)
{
OP2(SLJIT_ADD, RETURN_ADDR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, SLJIT_IMM, 0x400);
SELECT(SLJIT_LESS, STR_PTR, RETURN_ADDR, 0, STR_PTR);
}
else
{
jump = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x400);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
JUMPHERE(jump);
}
return;
}
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xd800);
jump = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0xe000 - 0xd800);
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x400));
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xdc00);
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x400));
JUMPHERE(jump);
return;
}
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 16 */
}
static void move_back(compiler_common *common, jump_list **backtracks, BOOL must_be_valid)
{
/* Goes one character back. Affects STR_PTR and TMP1. If must_be_valid is TRUE,
TMP2 is not used. Otherwise TMP2 must contain the start of the subject buffer,
and it is destroyed. Does not modify STR_PTR for invalid character sequences. */
DEFINE_COMPILER;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
struct sljit_jump *jump;
#endif
#ifdef SUPPORT_UNICODE
#if PCRE2_CODE_UNIT_WIDTH == 8
struct sljit_label *label;
if (common->utf)
{
if (!must_be_valid && common->invalid_utf)
{
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1));
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x80);
add_jump(compiler, &common->utfmoveback_invalid, JUMP(SLJIT_FAST_CALL));
if (backtracks != NULL)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, 0));
JUMPHERE(jump);
return;
}
label = LABEL();
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1));
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0);
CMPTO(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, 0x80, label);
return;
}
#elif PCRE2_CODE_UNIT_WIDTH == 16
if (common->utf)
{
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1));
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (!must_be_valid && common->invalid_utf)
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xd800);
jump = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0xe000 - 0xd800);
add_jump(compiler, &common->utfmoveback_invalid, JUMP(SLJIT_FAST_CALL));
if (backtracks != NULL)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, 0));
JUMPHERE(jump);
return;
}
/* Skip low surrogate if necessary. */
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xfc00);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0xdc00);
OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_EQUAL);
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
return;
}
#elif PCRE2_CODE_UNIT_WIDTH == 32
if (common->invalid_utf && !must_be_valid)
{
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), -IN_UCHARS(1));
if (backtracks != NULL)
{
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x110000));
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
return;
}
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x110000);
OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_LESS);
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
return;
}
#endif /* PCRE2_CODE_UNIT_WIDTH == [8|16|32] */
#endif /* SUPPORT_UNICODE */
SLJIT_UNUSED_ARG(backtracks);
SLJIT_UNUSED_ARG(must_be_valid);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
}
static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpifmatch)
{
/* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */
DEFINE_COMPILER;
struct sljit_jump *jump;
if (nltype == NLTYPE_ANY)
{
add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));
sljit_set_current_flags(compiler, SLJIT_SET_Z);
add_jump(compiler, backtracks, JUMP(jumpifmatch ? SLJIT_NOT_ZERO : SLJIT_ZERO));
}
else if (nltype == NLTYPE_ANYCRLF)
{
if (jumpifmatch)
{
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR));
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL));
}
else
{
jump = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL));
JUMPHERE(jump);
}
}
else
{
SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256);
add_jump(compiler, backtracks, CMP(jumpifmatch ? SLJIT_EQUAL : SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline));
}
}
#ifdef SUPPORT_UNICODE
#if PCRE2_CODE_UNIT_WIDTH == 8
static void do_utfreadchar(compiler_common *common)
{
/* Fast decoding a UTF-8 character. TMP1 contains the first byte
of the character (>= 0xc0). Return char value in TMP1. */
DEFINE_COMPILER;
struct sljit_jump *jump;
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
/* Searching for the first zero. */
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x800);
jump = JUMP(SLJIT_NOT_ZERO);
/* Two byte sequence. */
OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3000);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(jump);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x10000);
jump = JUMP(SLJIT_NOT_ZERO);
/* Three byte sequence. */
OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xe0000);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
/* Four byte sequence. */
JUMPHERE(jump);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));
OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xf0000);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_utfreadtype8(compiler_common *common)
{
/* Fast decoding a UTF-8 character type. TMP2 contains the first byte
of the character (>= 0xc0). Return value in TMP1. */
DEFINE_COMPILER;
struct sljit_jump *jump;
struct sljit_jump *compare;
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, 0x20);
jump = JUMP(SLJIT_NOT_ZERO);
/* Two byte sequence. */
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f);
/* The upper 5 bits are known at this point. */
compare = CMP(SLJIT_GREATER, TMP2, 0, SLJIT_IMM, 0x3);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0);
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(compare);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
/* We only have types for characters less than 256. */
JUMPHERE(jump);
OP1(SLJIT_MOV_U8, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_utfreadchar_invalid(compiler_common *common)
{
/* Slow decoding a UTF-8 character. TMP1 contains the first byte
of the character (>= 0xc0). Return char value in TMP1. STR_PTR is
undefined for invalid characters. */
DEFINE_COMPILER;
sljit_s32 i;
sljit_s32 has_cmov = sljit_has_cpu_feature(SLJIT_HAS_CMOV);
struct sljit_jump *jump;
struct sljit_jump *buffer_end_close;
struct sljit_label *three_byte_entry;
struct sljit_label *exit_invalid_label;
struct sljit_jump *exit_invalid[11];
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc2);
/* Usually more than 3 characters remained in the subject buffer. */
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));
/* Not a valid start of a multi-byte sequence, no more bytes read. */
exit_invalid[0] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0xf5 - 0xc2);
buffer_end_close = CMP(SLJIT_GREATER, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-3));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
/* If TMP2 is in 0x80-0xbf range, TMP1 is also increased by (0x2 << 6). */
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80);
exit_invalid[1] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x800);
jump = JUMP(SLJIT_NOT_ZERO);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(jump);
/* Three-byte sequence. */
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
if (has_cmov)
{
OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
SELECT(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, 0x20000, TMP1);
exit_invalid[2] = NULL;
}
else
exit_invalid[2] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x10000);
jump = JUMP(SLJIT_NOT_ZERO);
three_byte_entry = LABEL();
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2d800);
if (has_cmov)
{
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x800);
SELECT(SLJIT_LESS, TMP1, SLJIT_IMM, INVALID_UTF_CHAR - 0xd800, TMP1);
exit_invalid[3] = NULL;
}
else
exit_invalid[3] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x800);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xd800);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (has_cmov)
{
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x800);
SELECT(SLJIT_LESS, TMP1, SLJIT_IMM, INVALID_UTF_CHAR, TMP1);
exit_invalid[4] = NULL;
}
else
exit_invalid[4] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x800);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(jump);
/* Four-byte sequence. */
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
if (has_cmov)
{
OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
SELECT(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, 0, TMP1);
exit_invalid[5] = NULL;
}
else
exit_invalid[5] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc10000);
if (has_cmov)
{
OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x100000);
SELECT(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, INVALID_UTF_CHAR - 0x10000, TMP1);
exit_invalid[6] = NULL;
}
else
exit_invalid[6] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x100000);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(buffer_end_close);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
exit_invalid[7] = CMP(SLJIT_GREATER, STR_PTR, 0, STR_END, 0);
/* Two-byte sequence. */
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
/* If TMP2 is in 0x80-0xbf range, TMP1 is also increased by (0x2 << 6). */
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80);
exit_invalid[8] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x800);
jump = JUMP(SLJIT_NOT_ZERO);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
/* Three-byte sequence. */
JUMPHERE(jump);
exit_invalid[9] = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
if (has_cmov)
{
OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
SELECT(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, INVALID_UTF_CHAR, TMP1);
exit_invalid[10] = NULL;
}
else
exit_invalid[10] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
/* One will be substracted from STR_PTR later. */
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
/* Four byte sequences are not possible. */
CMPTO(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x30000, three_byte_entry);
exit_invalid_label = LABEL();
for (i = 0; i < 11; i++)
sljit_set_label(exit_invalid[i], exit_invalid_label);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_utfreadnewline_invalid(compiler_common *common)
{
/* Slow decoding a UTF-8 character, specialized for newlines.
TMP1 contains the first byte of the character (>= 0xc0). Return
char value in TMP1. */
DEFINE_COMPILER;
struct sljit_label *loop;
struct sljit_label *skip_start;
struct sljit_label *three_byte_exit;
struct sljit_jump *jump[5];
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
if (common->nltype != NLTYPE_ANY)
{
SLJIT_ASSERT(common->nltype != NLTYPE_FIXED || common->newline < 128);
/* All newlines are ascii, just skip intermediate octets. */
jump[0] = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
loop = LABEL();
if (sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_SUPP | SLJIT_MEM_POST, TMP2, SLJIT_MEM1(STR_PTR), IN_UCHARS(1)) == SLJIT_SUCCESS)
sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_POST, TMP2, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
else
{
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
}
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xc0);
CMPTO(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, 0x80, loop);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
JUMPHERE(jump[0]);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
return;
}
jump[0] = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
jump[1] = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, 0xc2);
jump[2] = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, 0xe2);
skip_start = LABEL();
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xc0);
jump[3] = CMP(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0x80);
/* Skip intermediate octets. */
loop = LABEL();
jump[4] = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xc0);
CMPTO(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, 0x80, loop);
JUMPHERE(jump[3]);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
three_byte_exit = LABEL();
JUMPHERE(jump[0]);
JUMPHERE(jump[4]);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
/* Two byte long newline: 0x85. */
JUMPHERE(jump[1]);
CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0x85, skip_start);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0x85);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
/* Three byte long newlines: 0x2028 and 0x2029. */
JUMPHERE(jump[2]);
CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0x80, skip_start);
CMPTO(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0, three_byte_exit);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_SUB, TMP1, 0, TMP2, 0, SLJIT_IMM, 0x80);
CMPTO(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x40, skip_start);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0x2000);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_utfmoveback_invalid(compiler_common *common)
{
/* Goes one character back. */
DEFINE_COMPILER;
sljit_s32 i;
struct sljit_jump *jump;
struct sljit_jump *buffer_start_close;
struct sljit_label *exit_ok_label;
struct sljit_label *exit_invalid_label;
struct sljit_jump *exit_invalid[7];
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));
exit_invalid[0] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0xc0);
/* Two-byte sequence. */
buffer_start_close = CMP(SLJIT_LESS, STR_PTR, 0, TMP2, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0);
jump = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x20);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 1);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
/* Three-byte sequence. */
JUMPHERE(jump);
exit_invalid[1] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, -0x40);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xe0);
jump = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x10);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 1);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
/* Four-byte sequence. */
JUMPHERE(jump);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xe0 - 0x80);
exit_invalid[2] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x40);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xf0);
exit_invalid[3] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x05);
exit_ok_label = LABEL();
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 1);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
/* Two-byte sequence. */
JUMPHERE(buffer_start_close);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
exit_invalid[4] = CMP(SLJIT_LESS, STR_PTR, 0, TMP2, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0);
CMPTO(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x20, exit_ok_label);
/* Three-byte sequence. */
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
exit_invalid[5] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, -0x40);
exit_invalid[6] = CMP(SLJIT_LESS, STR_PTR, 0, TMP2, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xe0);
CMPTO(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x10, exit_ok_label);
/* Four-byte sequences are not possible. */
exit_invalid_label = LABEL();
sljit_set_label(exit_invalid[5], exit_invalid_label);
sljit_set_label(exit_invalid[6], exit_invalid_label);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(exit_invalid[4]);
/* -2 + 4 = 2 */
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
exit_invalid_label = LABEL();
for (i = 0; i < 4; i++)
sljit_set_label(exit_invalid[i], exit_invalid_label);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(4));
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_utfpeakcharback(compiler_common *common)
{
/* Peak a character back. Does not modify STR_PTR. */
DEFINE_COMPILER;
struct sljit_jump *jump[2];
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xc0);
jump[0] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x20);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-3));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xe0);
jump[1] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x10);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-4));
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xe0 - 0x80);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xf0);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
JUMPHERE(jump[1]);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
JUMPHERE(jump[0]);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x80);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_utfpeakcharback_invalid(compiler_common *common)
{
/* Peak a character back. Does not modify STR_PTR. */
DEFINE_COMPILER;
sljit_s32 i;
sljit_s32 has_cmov = sljit_has_cpu_feature(SLJIT_HAS_CMOV);
struct sljit_jump *jump[2];
struct sljit_label *two_byte_entry;
struct sljit_label *three_byte_entry;
struct sljit_label *exit_invalid_label;
struct sljit_jump *exit_invalid[8];
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(3));
exit_invalid[0] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0xc0);
jump[0] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, STR_PTR, 0);
/* Two-byte sequence. */
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xc2);
jump[1] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x1e);
two_byte_entry = LABEL();
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
/* If TMP1 is in 0x80-0xbf range, TMP1 is also increased by (0x2 << 6). */
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(jump[1]);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xc2 - 0x80);
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x80);
exit_invalid[1] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
/* Three-byte sequence. */
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-3));
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xe0);
jump[1] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x10);
three_byte_entry = LABEL();
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 12);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xd800);
if (has_cmov)
{
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x800);
SELECT(SLJIT_LESS, TMP1, SLJIT_IMM, -0xd800, TMP1);
exit_invalid[2] = NULL;
}
else
exit_invalid[2] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x800);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xd800);
if (has_cmov)
{
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x800);
SELECT(SLJIT_LESS, TMP1, SLJIT_IMM, INVALID_UTF_CHAR, TMP1);
exit_invalid[3] = NULL;
}
else
exit_invalid[3] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x800);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(jump[1]);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xe0 - 0x80);
exit_invalid[4] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 12);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
/* Four-byte sequence. */
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-4));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xf0);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 18);
/* ADD is used instead of OR because of the SUB 0x10000 above. */
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
if (has_cmov)
{
OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x100000);
SELECT(SLJIT_GREATER_EQUAL, TMP1, SLJIT_IMM, INVALID_UTF_CHAR - 0x10000, TMP1);
exit_invalid[5] = NULL;
}
else
exit_invalid[5] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x100000);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(jump[0]);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1));
jump[0] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, STR_PTR, 0);
/* Two-byte sequence. */
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xc2);
CMPTO(SLJIT_LESS, TMP2, 0, SLJIT_IMM, 0x1e, two_byte_entry);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xc2 - 0x80);
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x80);
exit_invalid[6] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x40);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
/* Three-byte sequence. */
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-3));
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xe0);
CMPTO(SLJIT_LESS, TMP2, 0, SLJIT_IMM, 0x10, three_byte_entry);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(jump[0]);
exit_invalid[7] = CMP(SLJIT_GREATER, TMP2, 0, STR_PTR, 0);
/* Two-byte sequence. */
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xc2);
CMPTO(SLJIT_LESS, TMP2, 0, SLJIT_IMM, 0x1e, two_byte_entry);
exit_invalid_label = LABEL();
for (i = 0; i < 8; i++)
sljit_set_label(exit_invalid[i], exit_invalid_label);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */
#if PCRE2_CODE_UNIT_WIDTH == 16
static void do_utfreadchar_invalid(compiler_common *common)
{
/* Slow decoding a UTF-16 character. TMP1 contains the first half
of the character (>= 0xd800). Return char value in TMP1. STR_PTR is
undefined for invalid characters. */
DEFINE_COMPILER;
struct sljit_jump *exit_invalid[3];
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
/* TMP2 contains the high surrogate. */
exit_invalid[0] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0xdc00);
exit_invalid[1] = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xdc00);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x10000);
exit_invalid[2] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x400);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(exit_invalid[0]);
JUMPHERE(exit_invalid[1]);
JUMPHERE(exit_invalid[2]);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_utfreadnewline_invalid(compiler_common *common)
{
/* Slow decoding a UTF-16 character, specialized for newlines.
TMP1 contains the first half of the character (>= 0xd800). Return
char value in TMP1. */
DEFINE_COMPILER;
struct sljit_jump *exit_invalid[2];
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
/* TMP2 contains the high surrogate. */
exit_invalid[0] = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
exit_invalid[1] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0xdc00);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xdc00);
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, SLJIT_IMM, 0x400);
OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0x10000);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(exit_invalid[0]);
JUMPHERE(exit_invalid[1]);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_utfmoveback_invalid(compiler_common *common)
{
/* Goes one character back. */
DEFINE_COMPILER;
struct sljit_jump *exit_invalid[3];
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
exit_invalid[0] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x400);
exit_invalid[1] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, STR_PTR, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xd800);
exit_invalid[2] = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0x400);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 1);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(exit_invalid[0]);
JUMPHERE(exit_invalid[1]);
JUMPHERE(exit_invalid[2]);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_utfpeakcharback_invalid(compiler_common *common)
{
/* Peak a character back. Does not modify STR_PTR. */
DEFINE_COMPILER;
struct sljit_jump *jump;
struct sljit_jump *exit_invalid[3];
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
jump = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 0xe000);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1));
exit_invalid[0] = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xdc00);
exit_invalid[1] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, STR_PTR, 0);
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000 - 0xdc00);
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xd800);
exit_invalid[2] = CMP(SLJIT_GREATER_EQUAL, TMP2, 0, SLJIT_IMM, 0x400);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
JUMPHERE(jump);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(exit_invalid[0]);
JUMPHERE(exit_invalid[1]);
JUMPHERE(exit_invalid[2]);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
#endif /* PCRE2_CODE_UNIT_WIDTH == 16 */
/* UCD_BLOCK_SIZE must be 128 (see the assert below). */
#define UCD_BLOCK_MASK 127
#define UCD_BLOCK_SHIFT 7
static void do_getucd(compiler_common *common)
{
/* Search the UCD record for the character comes in TMP1.
Returns chartype in TMP1 and UCD offset in TMP2. */
DEFINE_COMPILER;
#if PCRE2_CODE_UNIT_WIDTH == 32
struct sljit_jump *jump;
#endif
#if defined SLJIT_DEBUG && SLJIT_DEBUG
/* dummy_ucd_record */
const ucd_record *record = GET_UCD(UNASSIGNED_UTF_CHAR);
SLJIT_ASSERT(record->script == ucp_Unknown && record->chartype == ucp_Cn && record->gbprop == ucp_gbOther);
SLJIT_ASSERT(record->caseset == 0 && record->other_case == 0);
#endif
SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && sizeof(ucd_record) == 12);
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
#if PCRE2_CODE_UNIT_WIDTH == 32
if (!common->utf)
{
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, MAX_UTF_CODE_POINT + 1);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, UNASSIGNED_UTF_CHAR);
JUMPHERE(jump);
}
#endif
OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1);
OP1(SLJIT_MOV_U16, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_stage1));
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_stage2));
OP1(SLJIT_MOV_U16, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_getucdtype(compiler_common *common)
{
/* Search the UCD record for the character comes in TMP1.
Returns chartype in TMP1 and UCD offset in TMP2. */
DEFINE_COMPILER;
#if PCRE2_CODE_UNIT_WIDTH == 32
struct sljit_jump *jump;
#endif
#if defined SLJIT_DEBUG && SLJIT_DEBUG
/* dummy_ucd_record */
const ucd_record *record = GET_UCD(UNASSIGNED_UTF_CHAR);
SLJIT_ASSERT(record->script == ucp_Unknown && record->chartype == ucp_Cn && record->gbprop == ucp_gbOther);
SLJIT_ASSERT(record->caseset == 0 && record->other_case == 0);
#endif
SLJIT_ASSERT(UCD_BLOCK_SIZE == 128 && sizeof(ucd_record) == 12);
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
#if PCRE2_CODE_UNIT_WIDTH == 32
if (!common->utf)
{
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, MAX_UTF_CODE_POINT + 1);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, UNASSIGNED_UTF_CHAR);
JUMPHERE(jump);
}
#endif
OP2(SLJIT_LSHR, TMP2, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_SHIFT);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1);
OP1(SLJIT_MOV_U16, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(ucd_stage1));
OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, UCD_BLOCK_MASK);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCD_BLOCK_SHIFT);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_stage2));
OP1(SLJIT_MOV_U16, TMP2, 0, SLJIT_MEM2(TMP2, TMP1), 1);
/* TMP2 is multiplied by 12. Same as (TMP2 << 2) + ((TMP2 << 2) << 1). */
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records) + SLJIT_OFFSETOF(ucd_record, chartype));
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 2);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP2, 0);
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM2(TMP1, TMP2), 1);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
#endif /* SUPPORT_UNICODE */
static SLJIT_INLINE struct sljit_label *mainloop_entry(compiler_common *common)
{
DEFINE_COMPILER;
struct sljit_label *mainloop;
struct sljit_label *newlinelabel = NULL;
struct sljit_jump *start;
struct sljit_jump *end = NULL;
struct sljit_jump *end2 = NULL;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
struct sljit_label *loop;
struct sljit_jump *jump;
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32 */
jump_list *newline = NULL;
sljit_u32 overall_options = common->re->overall_options;
BOOL hascrorlf = (common->re->flags & PCRE2_HASCRORLF) != 0;
BOOL newlinecheck = FALSE;
BOOL readuchar = FALSE;
if (!(hascrorlf || (overall_options & PCRE2_FIRSTLINE) != 0)
&& (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF || common->newline > 255))
newlinecheck = TRUE;
SLJIT_ASSERT(common->abort_label == NULL);
if ((overall_options & PCRE2_FIRSTLINE) != 0)
{
/* Search for the end of the first line. */
SLJIT_ASSERT(common->match_end_ptr != 0);
OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0);
if (common->nltype == NLTYPE_FIXED && common->newline > 255)
{
mainloop = LABEL();
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
end = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, mainloop);
CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, mainloop);
JUMPHERE(end);
OP2(SLJIT_SUB, SLJIT_MEM1(SLJIT_SP), common->match_end_ptr, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
}
else
{
end = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
mainloop = LABEL();
/* Continual stores does not cause data dependency. */
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->match_end_ptr, STR_PTR, 0);
read_char(common, common->nlmin, common->nlmax, NULL, READ_CHAR_NEWLINE);
check_newlinechar(common, common->nltype, &newline, TRUE);
CMPTO(SLJIT_LESS, STR_PTR, 0, STR_END, 0, mainloop);
JUMPHERE(end);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->match_end_ptr, STR_PTR, 0);
set_jumps(newline, LABEL());
}
OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0);
}
else if ((overall_options & PCRE2_USE_OFFSET_LIMIT) != 0)
{
/* Check whether offset limit is set and valid. */
SLJIT_ASSERT(common->match_end_ptr != 0);
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, offset_limit));
}
else
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, offset_limit));
OP1(SLJIT_MOV, TMP2, 0, STR_END, 0);
end = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, (sljit_sw) PCRE2_UNSET);
if (HAS_VIRTUAL_REGISTERS)
OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0);
else
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin));
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif /* PCRE2_CODE_UNIT_WIDTH == [16|32] */
if (HAS_VIRTUAL_REGISTERS)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin));
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP1, 0);
end2 = CMP(SLJIT_LESS_EQUAL, TMP2, 0, STR_END, 0);
OP1(SLJIT_MOV, TMP2, 0, STR_END, 0);
JUMPHERE(end2);
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE2_ERROR_NOMATCH);
add_jump(compiler, &common->abort, CMP(SLJIT_LESS, TMP2, 0, STR_PTR, 0));
JUMPHERE(end);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->match_end_ptr, TMP2, 0);
}
start = JUMP(SLJIT_JUMP);
if (newlinecheck)
{
newlinelabel = LABEL();
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
end = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, common->newline & 0xff);
OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_EQUAL);
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif /* PCRE2_CODE_UNIT_WIDTH == [16|32] */
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
end2 = JUMP(SLJIT_JUMP);
}
mainloop = LABEL();
/* Increasing the STR_PTR here requires one less jump in the most common case. */
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf && !common->invalid_utf) readuchar = TRUE;
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32 */
if (newlinecheck) readuchar = TRUE;
if (readuchar)
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
if (newlinecheck)
CMPTO(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, newlinelabel);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
#if PCRE2_CODE_UNIT_WIDTH == 8
if (common->invalid_utf)
{
/* Skip continuation code units. */
loop = LABEL();
jump = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x80);
CMPTO(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x40, loop);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
JUMPHERE(jump);
}
else if (common->utf)
{
jump = CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
JUMPHERE(jump);
}
#elif PCRE2_CODE_UNIT_WIDTH == 16
if (common->invalid_utf)
{
/* Skip continuation code units. */
loop = LABEL();
jump = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xdc00);
CMPTO(SLJIT_LESS, TMP1, 0, SLJIT_IMM, 0x400, loop);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
JUMPHERE(jump);
}
else if (common->utf)
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0xd800);
if (sljit_has_cpu_feature(SLJIT_HAS_CMOV))
{
OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x400);
SELECT(SLJIT_LESS, STR_PTR, TMP2, 0, STR_PTR);
}
else
{
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP1, 0, SLJIT_IMM, 0x400);
OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_LESS);
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
}
}
#endif /* PCRE2_CODE_UNIT_WIDTH == [8|16] */
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32 */
JUMPHERE(start);
if (newlinecheck)
{
JUMPHERE(end);
JUMPHERE(end2);
}
return mainloop;
}
static SLJIT_INLINE void add_prefix_char(PCRE2_UCHAR chr, fast_forward_char_data *chars, BOOL last)
{
sljit_u32 i, count = chars->count;
if (count == 255)
return;
if (count == 0)
{
chars->count = 1;
chars->chars[0] = chr;
if (last)
chars->last_count = 1;
return;
}
for (i = 0; i < count; i++)
if (chars->chars[i] == chr)
return;
if (count >= MAX_DIFF_CHARS)
{
chars->count = 255;
return;
}
chars->chars[count] = chr;
chars->count = count + 1;
if (last)
chars->last_count++;
}
/* Value can be increased if needed. Patterns
such as /(a|){33}b/ can exhaust the stack.
Note: /(a|){29}b/ already stops scan_prefix()
because it reaches the maximum step_count. */
#define SCAN_PREFIX_STACK_END 32
/*
Scan prefix stores the prefix string in the chars array.
The elements of the chars array is either small character
sets or "any" (count is set to 255).
Examples (the chars array is represented by a simple regex):
/(abc|xbyd)/ prefix: /[ax]b[cy]/ (length: 3)
/a[a-z]b+c/ prefix: a.b (length: 3)
/ab?cd/ prefix: a[bc][cd] (length: 3)
/(ab|cd)|(ef|gh)/ prefix: [aceg][bdfh] (length: 2)
The length is returned by scan_prefix(). The length is
less than or equal than the minimum length of the pattern.
*/
static int scan_prefix(compiler_common *common, PCRE2_SPTR cc, fast_forward_char_data *chars)
{
fast_forward_char_data *chars_start = chars;
fast_forward_char_data *chars_end = chars + MAX_N_CHARS;
PCRE2_SPTR cc_stack[SCAN_PREFIX_STACK_END];
fast_forward_char_data *chars_stack[SCAN_PREFIX_STACK_END];
sljit_u8 next_alternative_stack[SCAN_PREFIX_STACK_END];
BOOL last, any, class, caseless;
int stack_ptr, step_count, repeat, len, len_save;
sljit_u32 chr; /* Any unicode character. */
sljit_u8 *bytes, *bytes_end, byte;
PCRE2_SPTR alternative, cc_save, oc;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
PCRE2_UCHAR othercase[4];
#elif defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 16
PCRE2_UCHAR othercase[2];
#else
PCRE2_UCHAR othercase[1];
#endif
repeat = 1;
stack_ptr = 0;
step_count = 10000;
while (TRUE)
{
if (--step_count == 0)
return 0;
SLJIT_ASSERT(chars <= chars_start + MAX_N_CHARS);
if (chars >= chars_end)
{
if (stack_ptr == 0)
return (int)(chars_end - chars_start);
--stack_ptr;
cc = cc_stack[stack_ptr];
chars = chars_stack[stack_ptr];
if (chars >= chars_end)
continue;
if (next_alternative_stack[stack_ptr] != 0)
{
/* When an alternative is processed, the
next alternative is pushed onto the stack. */
SLJIT_ASSERT(*cc == OP_ALT);
alternative = cc + GET(cc, 1);
if (*alternative == OP_ALT)
{
SLJIT_ASSERT(stack_ptr < SCAN_PREFIX_STACK_END);
SLJIT_ASSERT(chars_stack[stack_ptr] == chars);
SLJIT_ASSERT(next_alternative_stack[stack_ptr] == 1);
cc_stack[stack_ptr] = alternative;
stack_ptr++;
}
cc += 1 + LINK_SIZE;
}
}
last = TRUE;
any = FALSE;
class = FALSE;
caseless = FALSE;
switch (*cc)
{
case OP_CHARI:
caseless = TRUE;
/* Fall through */
case OP_CHAR:
last = FALSE;
cc++;
break;
case OP_SOD:
case OP_SOM:
case OP_SET_SOM:
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
case OP_EODN:
case OP_EOD:
case OP_CIRC:
case OP_CIRCM:
case OP_DOLL:
case OP_DOLLM:
case OP_NOT_UCP_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
/* Zero width assertions. */
cc++;
continue;
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
case OP_ASSERT_NA:
case OP_ASSERTBACK_NA:
case OP_ASSERT_SCS:
cc = bracketend(cc);
continue;
case OP_PLUSI:
case OP_MINPLUSI:
case OP_POSPLUSI:
caseless = TRUE;
/* Fall through */
case OP_PLUS:
case OP_MINPLUS:
case OP_POSPLUS:
cc++;
break;
case OP_EXACTI:
caseless = TRUE;
/* Fall through */
case OP_EXACT:
repeat = GET2(cc, 1);
last = FALSE;
cc += 1 + IMM2_SIZE;
break;
case OP_QUERYI:
case OP_MINQUERYI:
case OP_POSQUERYI:
caseless = TRUE;
/* Fall through */
case OP_QUERY:
case OP_MINQUERY:
case OP_POSQUERY:
len = 1;
cc++;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc);
#endif
if (stack_ptr >= SCAN_PREFIX_STACK_END)
{
chars_end = chars;
continue;
}
cc_stack[stack_ptr] = cc + len;
chars_stack[stack_ptr] = chars;
next_alternative_stack[stack_ptr] = 0;
stack_ptr++;
last = FALSE;
break;
case OP_KET:
cc += 1 + LINK_SIZE;
continue;
case OP_ALT:
cc += GET(cc, 1);
continue;
case OP_ONCE:
case OP_BRA:
case OP_BRAPOS:
case OP_CBRA:
case OP_CBRAPOS:
alternative = cc + GET(cc, 1);
if (*alternative == OP_ALT)
{
if (stack_ptr >= SCAN_PREFIX_STACK_END)
{
chars_end = chars;
continue;
}
cc_stack[stack_ptr] = alternative;
chars_stack[stack_ptr] = chars;
next_alternative_stack[stack_ptr] = 1;
stack_ptr++;
}
if (*cc == OP_CBRA || *cc == OP_CBRAPOS)
cc += IMM2_SIZE;
cc += 1 + LINK_SIZE;
continue;
case OP_CLASS:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf && !is_char7_bitset((const sljit_u8 *)(cc + 1), FALSE))
{
chars_end = chars;
continue;
}
#endif
class = TRUE;
break;
case OP_NCLASS:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
{
chars_end = chars;
continue;
}
#endif
class = TRUE;
break;
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH != 8
case OP_XCLASS:
case OP_ECLASS:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
{
chars_end = chars;
continue;
}
#endif
any = TRUE;
cc += GET(cc, 1);
break;
#endif
case OP_DIGIT:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf && !is_char7_bitset((const sljit_u8 *)common->ctypes - cbit_length + cbit_digit, FALSE))
{
chars_end = chars;
continue;
}
#endif
any = TRUE;
cc++;
break;
case OP_WHITESPACE:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf && !is_char7_bitset((const sljit_u8 *)common->ctypes - cbit_length + cbit_space, FALSE))
{
chars_end = chars;
continue;
}
#endif
any = TRUE;
cc++;
break;
case OP_WORDCHAR:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf && !is_char7_bitset((const sljit_u8 *)common->ctypes - cbit_length + cbit_word, FALSE))
{
chars_end = chars;
continue;
}
#endif
any = TRUE;
cc++;
break;
case OP_NOT:
case OP_NOTI:
cc++;
/* Fall through. */
case OP_NOT_DIGIT:
case OP_NOT_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_ANY:
case OP_ALLANY:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
{
chars_end = chars;
continue;
}
#endif
any = TRUE;
cc++;
break;
#ifdef SUPPORT_UNICODE
case OP_NOTPROP:
case OP_PROP:
#if PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
{
chars_end = chars;
continue;
}
#endif
any = TRUE;
cc += 1 + 2;
break;
#endif
case OP_TYPEEXACT:
repeat = GET2(cc, 1);
cc += 1 + IMM2_SIZE;
continue;
case OP_NOTEXACT:
case OP_NOTEXACTI:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
{
chars_end = chars;
continue;
}
#endif
any = TRUE;
repeat = GET2(cc, 1);
cc += 1 + IMM2_SIZE + 1;
break;
default:
chars_end = chars;
continue;
}
SLJIT_ASSERT(chars < chars_end);
if (any)
{
do
{
chars->count = 255;
chars++;
}
while (--repeat > 0 && chars < chars_end);
repeat = 1;
continue;
}
if (class)
{
bytes = (sljit_u8*) (cc + 1);
cc += 1 + 32 / sizeof(PCRE2_UCHAR);
SLJIT_ASSERT(last == TRUE && repeat == 1);
switch (*cc)
{
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSQUERY:
last = FALSE;
/* Fall through */
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRPOSSTAR:
if (stack_ptr >= SCAN_PREFIX_STACK_END)
{
chars_end = chars;
continue;
}
cc_stack[stack_ptr] = ++cc;
chars_stack[stack_ptr] = chars;
next_alternative_stack[stack_ptr] = 0;
stack_ptr++;
break;
default:
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRPOSPLUS:
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
repeat = GET2(cc, 1);
if (repeat <= 0)
{
chars_end = chars;
continue;
}
last = (repeat != (int)GET2(cc, 1 + IMM2_SIZE));
cc += 1 + 2 * IMM2_SIZE;
break;
}
do
{
if (bytes[31] & 0x80)
chars->count = 255;
else if (chars->count != 255)
{
bytes_end = bytes + 32;
chr = 0;
do
{
byte = *bytes++;
SLJIT_ASSERT((chr & 0x7) == 0);
if (byte == 0)
chr += 8;
else
{
do
{
if ((byte & 0x1) != 0)
add_prefix_char(chr, chars, TRUE);
byte >>= 1;
chr++;
}
while (byte != 0);
chr = (chr + 7) & (sljit_u32)(~7);
}
}
while (chars->count != 255 && bytes < bytes_end);
bytes = bytes_end - 32;
}
chars++;
}
while (--repeat > 0 && chars < chars_end);
repeat = 1;
if (last)
chars_end = chars;
continue;
}
len = 1;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc);
#endif
if (caseless && char_has_othercase(common, cc))
{
#ifdef SUPPORT_UNICODE
if (common->utf)
{
GETCHAR(chr, cc);
if ((int)PRIV(ord2utf)(char_othercase(common, chr), othercase) != len)
{
chars_end = chars;
continue;
}
}
else
#endif
{
chr = *cc;
#ifdef SUPPORT_UNICODE
if (common->ucp && chr > 127)
{
chr = UCD_OTHERCASE(chr);
othercase[0] = (chr == (PCRE2_UCHAR)chr) ? chr : *cc;
}
else
#endif
othercase[0] = TABLE_GET(chr, common->fcc, chr);
}
}
else
{
caseless = FALSE;
othercase[0] = 0; /* Stops compiler warning - PH */
}
len_save = len;
cc_save = cc;
while (TRUE)
{
oc = othercase;
do
{
len--;
chr = *cc;
add_prefix_char(*cc, chars, len == 0);
if (caseless)
add_prefix_char(*oc, chars, len == 0);
chars++;
cc++;
oc++;
}
while (len > 0 && chars < chars_end);
if (--repeat == 0 || chars >= chars_end)
break;
len = len_save;
cc = cc_save;
}
repeat = 1;
if (last)
chars_end = chars;
}
}
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
static void jumpto_if_not_utf_char_start(struct sljit_compiler *compiler, sljit_s32 reg, struct sljit_label *label)
{
#if PCRE2_CODE_UNIT_WIDTH == 8
OP2(SLJIT_AND, reg, 0, reg, 0, SLJIT_IMM, 0xc0);
CMPTO(SLJIT_EQUAL, reg, 0, SLJIT_IMM, 0x80, label);
#elif PCRE2_CODE_UNIT_WIDTH == 16
OP2(SLJIT_AND, reg, 0, reg, 0, SLJIT_IMM, 0xfc00);
CMPTO(SLJIT_EQUAL, reg, 0, SLJIT_IMM, 0xdc00, label);
#else
#error "Unknown code width"
#endif
}
#endif
#include "pcre2_jit_simd_inc.h"
#ifdef JIT_HAS_FAST_FORWARD_CHAR_PAIR_SIMD
static BOOL check_fast_forward_char_pair_simd(compiler_common *common, fast_forward_char_data *chars, int max)
{
sljit_s32 i, j, max_i = 0, max_j = 0;
sljit_u32 max_pri = 0;
sljit_s32 max_offset = max_fast_forward_char_pair_offset();
PCRE2_UCHAR a1, a2, a_pri, b1, b2, b_pri;
for (i = max - 1; i >= 1; i--)
{
if (chars[i].last_count > 2)
{
a1 = chars[i].chars[0];
a2 = chars[i].chars[1];
a_pri = chars[i].last_count;
j = i - max_offset;
if (j < 0)
j = 0;
while (j < i)
{
b_pri = chars[j].last_count;
if (b_pri > 2 && (sljit_u32)a_pri + (sljit_u32)b_pri >= max_pri)
{
b1 = chars[j].chars[0];
b2 = chars[j].chars[1];
if (a1 != b1 && a1 != b2 && a2 != b1 && a2 != b2)
{
max_pri = a_pri + b_pri;
max_i = i;
max_j = j;
}
}
j++;
}
}
}
if (max_pri == 0)
return FALSE;
fast_forward_char_pair_simd(common, max_i, chars[max_i].chars[0], chars[max_i].chars[1], max_j, chars[max_j].chars[0], chars[max_j].chars[1]);
return TRUE;
}
#endif /* JIT_HAS_FAST_FORWARD_CHAR_PAIR_SIMD */
static void fast_forward_first_char2(compiler_common *common, PCRE2_UCHAR char1, PCRE2_UCHAR char2, sljit_s32 offset)
{
DEFINE_COMPILER;
struct sljit_label *start;
struct sljit_jump *match;
struct sljit_jump *partial_quit;
PCRE2_UCHAR mask;
BOOL has_match_end = (common->match_end_ptr != 0);
SLJIT_ASSERT(common->mode == PCRE2_JIT_COMPLETE || offset == 0);
if (has_match_end)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->match_end_ptr);
if (offset > 0)
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(offset));
if (has_match_end)
{
OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(offset + 1));
OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_END, 0, TMP1, 0);
SELECT(SLJIT_GREATER, STR_END, TMP1, 0, STR_END);
}
#ifdef JIT_HAS_FAST_FORWARD_CHAR_SIMD
if (JIT_HAS_FAST_FORWARD_CHAR_SIMD)
{
fast_forward_char_simd(common, char1, char2, offset);
if (offset > 0)
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(offset));
if (has_match_end)
OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
return;
}
#endif
start = LABEL();
partial_quit = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
if (common->mode == PCRE2_JIT_COMPLETE)
add_jump(compiler, &common->failed_match, partial_quit);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (char1 == char2)
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, char1, start);
else
{
mask = char1 ^ char2;
if (is_powerof2(mask))
{
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, mask);
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, char1 | mask, start);
}
else
{
match = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, char1);
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, char2, start);
JUMPHERE(match);
}
}
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf && offset > 0)
{
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-(offset + 1)));
jumpto_if_not_utf_char_start(compiler, TMP1, start);
}
#endif
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(offset + 1));
if (common->mode != PCRE2_JIT_COMPLETE)
JUMPHERE(partial_quit);
if (has_match_end)
OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
}
static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common)
{
DEFINE_COMPILER;
struct sljit_label *start;
struct sljit_jump *match;
fast_forward_char_data chars[MAX_N_CHARS];
sljit_s32 offset;
PCRE2_UCHAR mask;
PCRE2_UCHAR *char_set, *char_set_end;
int i, max, from;
int range_right = -1, range_len;
sljit_u8 *update_table = NULL;
BOOL in_range;
for (i = 0; i < MAX_N_CHARS; i++)
{
chars[i].count = 0;
chars[i].last_count = 0;
}
max = scan_prefix(common, common->start, chars);
if (max < 1)
return FALSE;
/* Convert last_count to priority. */
for (i = 0; i < max; i++)
{
SLJIT_ASSERT(chars[i].last_count <= chars[i].count);
switch (chars[i].count)
{
case 0:
chars[i].count = 255;
chars[i].last_count = 0;
break;
case 1:
chars[i].last_count = (chars[i].last_count == 1) ? 7 : 5;
/* Simplifies algorithms later. */
chars[i].chars[1] = chars[i].chars[0];
break;
case 2:
SLJIT_ASSERT(chars[i].chars[0] != chars[i].chars[1]);
if (is_powerof2(chars[i].chars[0] ^ chars[i].chars[1]))
chars[i].last_count = (chars[i].last_count == 2) ? 6 : 4;
else
chars[i].last_count = (chars[i].last_count == 2) ? 3 : 2;
break;
default:
chars[i].last_count = (chars[i].count == 255) ? 0 : 1;
break;
}
}
#ifdef JIT_HAS_FAST_FORWARD_CHAR_PAIR_SIMD
if (JIT_HAS_FAST_FORWARD_CHAR_PAIR_SIMD && check_fast_forward_char_pair_simd(common, chars, max))
return TRUE;
#endif
in_range = FALSE;
/* Prevent compiler "uninitialized" warning */
from = 0;
range_len = 4 /* minimum length */ - 1;
for (i = 0; i <= max; i++)
{
if (in_range && (i - from) > range_len && (chars[i - 1].count < 255))
{
range_len = i - from;
range_right = i - 1;
}
if (i < max && chars[i].count < 255)
{
SLJIT_ASSERT(chars[i].count > 0);
if (!in_range)
{
in_range = TRUE;
from = i;
}
}
else
in_range = FALSE;
}
if (range_right >= 0)
{
update_table = (sljit_u8 *)allocate_read_only_data(common, 256);
if (update_table == NULL)
return TRUE;
memset(update_table, IN_UCHARS(range_len), 256);
for (i = 0; i < range_len; i++)
{
SLJIT_ASSERT(chars[range_right - i].count > 0 && chars[range_right - i].count < 255);
char_set = chars[range_right - i].chars;
char_set_end = char_set + chars[range_right - i].count;
do
{
if (update_table[(*char_set) & 0xff] > IN_UCHARS(i))
update_table[(*char_set) & 0xff] = IN_UCHARS(i);
char_set++;
}
while (char_set < char_set_end);
}
}
offset = -1;
/* Scan forward. */
for (i = 0; i < max; i++)
{
if (range_right == i)
continue;
if (offset == -1)
{
if (chars[i].last_count >= 2)
offset = i;
}
else if (chars[offset].last_count < chars[i].last_count)
offset = i;
}
SLJIT_ASSERT(offset == -1 || (chars[offset].count >= 1 && chars[offset].count <= 2));
if (range_right < 0)
{
if (offset < 0)
return FALSE;
/* Works regardless the value is 1 or 2. */
fast_forward_first_char2(common, chars[offset].chars[0], chars[offset].chars[1], offset);
return TRUE;
}
SLJIT_ASSERT(range_right != offset);
if (common->match_end_ptr != 0)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->match_end_ptr);
OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);
OP2(SLJIT_SUB | SLJIT_SET_LESS, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max));
add_jump(compiler, &common->failed_match, JUMP(SLJIT_LESS));
OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_END, 0, TMP1, 0);
SELECT(SLJIT_GREATER, STR_END, TMP1, 0, STR_END);
}
else
{
OP2(SLJIT_SUB | SLJIT_SET_LESS, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max));
add_jump(compiler, &common->failed_match, JUMP(SLJIT_LESS));
}
SLJIT_ASSERT(range_right >= 0);
if (!HAS_VIRTUAL_REGISTERS)
OP1(SLJIT_MOV, RETURN_ADDR, 0, SLJIT_IMM, (sljit_sw)update_table);
start = LABEL();
add_jump(compiler, &common->failed_match, CMP(SLJIT_GREATER, STR_PTR, 0, STR_END, 0));
#if PCRE2_CODE_UNIT_WIDTH == 8 || (defined SLJIT_LITTLE_ENDIAN && SLJIT_LITTLE_ENDIAN)
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(range_right));
#else
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(range_right + 1) - 1);
#endif
if (!HAS_VIRTUAL_REGISTERS)
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM2(RETURN_ADDR, TMP1), 0);
else
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)update_table);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, start);
if (offset >= 0)
{
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offset));
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (chars[offset].count == 1)
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[offset].chars[0], start);
else
{
mask = chars[offset].chars[0] ^ chars[offset].chars[1];
if (is_powerof2(mask))
{
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, mask);
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[offset].chars[0] | mask, start);
}
else
{
match = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, chars[offset].chars[0]);
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[offset].chars[1], start);
JUMPHERE(match);
}
}
}
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf && offset != 0)
{
if (offset < 0)
{
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
}
else
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
jumpto_if_not_utf_char_start(compiler, TMP1, start);
if (offset < 0)
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
}
#endif
if (offset >= 0)
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (common->match_end_ptr != 0)
OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
else
OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max));
return TRUE;
}
static SLJIT_INLINE void fast_forward_first_char(compiler_common *common)
{
PCRE2_UCHAR first_char = (PCRE2_UCHAR)(common->re->first_codeunit);
PCRE2_UCHAR oc;
oc = first_char;
if ((common->re->flags & PCRE2_FIRSTCASELESS) != 0)
{
oc = TABLE_GET(first_char, common->fcc, first_char);
#if defined SUPPORT_UNICODE
if (first_char > 127 && (common->utf || common->ucp))
oc = UCD_OTHERCASE(first_char);
#endif
}
fast_forward_first_char2(common, first_char, oc, 0);
}
static SLJIT_INLINE void fast_forward_newline(compiler_common *common)
{
DEFINE_COMPILER;
struct sljit_label *loop;
struct sljit_jump *lastchar = NULL;
struct sljit_jump *firstchar;
struct sljit_jump *quit = NULL;
struct sljit_jump *foundcr = NULL;
struct sljit_jump *notfoundnl;
jump_list *newline = NULL;
if (common->match_end_ptr != 0)
{
OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), common->match_end_ptr);
}
if (common->nltype == NLTYPE_FIXED && common->newline > 255)
{
#ifdef JIT_HAS_FAST_FORWARD_CHAR_PAIR_SIMD
if (JIT_HAS_FAST_FORWARD_CHAR_PAIR_SIMD && common->mode == PCRE2_JIT_COMPLETE)
{
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin));
}
else
{
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, str));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin));
}
firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP2U(SLJIT_SUB | SLJIT_SET_Z, STR_PTR, 0, TMP1, 0);
OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_NOT_EQUAL);
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
fast_forward_char_pair_simd(common, 1, common->newline & 0xff, common->newline & 0xff, 0, (common->newline >> 8) & 0xff, (common->newline >> 8) & 0xff);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
}
else
#endif /* JIT_HAS_FAST_FORWARD_CHAR_PAIR_SIMD */
{
lastchar = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin));
}
else
{
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, str));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin));
}
firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(2));
OP2U(SLJIT_SUB | SLJIT_SET_GREATER_EQUAL, STR_PTR, 0, TMP1, 0);
OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_GREATER_EQUAL);
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
loop = LABEL();
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
quit = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff, loop);
CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff, loop);
JUMPHERE(quit);
JUMPHERE(lastchar);
}
JUMPHERE(firstchar);
if (common->match_end_ptr != 0)
OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
return;
}
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str));
}
else
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, str));
/* Example: match /^/ to \r\n from offset 1. */
firstchar = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0);
if (common->nltype == NLTYPE_ANY)
move_back(common, NULL, FALSE);
else
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
loop = LABEL();
common->ff_newline_shortcut = loop;
#ifdef JIT_HAS_FAST_FORWARD_CHAR_SIMD
if (JIT_HAS_FAST_FORWARD_CHAR_SIMD && (common->nltype == NLTYPE_FIXED || common->nltype == NLTYPE_ANYCRLF))
{
if (common->nltype == NLTYPE_ANYCRLF)
{
fast_forward_char_simd(common, CHAR_CR, CHAR_LF, 0);
if (common->mode != PCRE2_JIT_COMPLETE)
lastchar = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
quit = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
}
else
{
fast_forward_char_simd(common, common->newline, common->newline, 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (common->mode != PCRE2_JIT_COMPLETE)
{
OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_PTR, 0, STR_END, 0);
SELECT(SLJIT_GREATER, STR_PTR, STR_END, 0, STR_PTR);
}
}
}
else
#endif /* JIT_HAS_FAST_FORWARD_CHAR_SIMD */
{
read_char(common, common->nlmin, common->nlmax, NULL, READ_CHAR_NEWLINE);
lastchar = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)
foundcr = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
check_newlinechar(common, common->nltype, &newline, FALSE);
set_jumps(newline, loop);
}
if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)
{
if (quit == NULL)
{
quit = JUMP(SLJIT_JUMP);
JUMPHERE(foundcr);
}
notfoundnl = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, CHAR_NL);
OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_EQUAL);
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
JUMPHERE(notfoundnl);
JUMPHERE(quit);
}
if (lastchar)
JUMPHERE(lastchar);
JUMPHERE(firstchar);
if (common->match_end_ptr != 0)
OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
}
static BOOL optimize_class(compiler_common *common, const sljit_u8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks);
static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common)
{
DEFINE_COMPILER;
const sljit_u8 *start_bits = common->re->start_bitmap;
struct sljit_label *start;
struct sljit_jump *partial_quit;
#if PCRE2_CODE_UNIT_WIDTH != 8
struct sljit_jump *found = NULL;
#endif
jump_list *matches = NULL;
if (common->match_end_ptr != 0)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->match_end_ptr);
OP1(SLJIT_MOV, RETURN_ADDR, 0, STR_END, 0);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1));
OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_END, 0, TMP1, 0);
SELECT(SLJIT_GREATER, STR_END, TMP1, 0, STR_END);
}
start = LABEL();
partial_quit = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
if (common->mode == PCRE2_JIT_COMPLETE)
add_jump(compiler, &common->failed_match, partial_quit);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (!optimize_class(common, start_bits, (start_bits[31] & 0x80) != 0, FALSE, &matches))
{
#if PCRE2_CODE_UNIT_WIDTH != 8
if ((start_bits[31] & 0x80) != 0)
found = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 255);
else
CMPTO(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, 255, start);
#elif defined SUPPORT_UNICODE
if (common->utf && is_char7_bitset(start_bits, FALSE))
CMPTO(SLJIT_GREATER, TMP1, 0, SLJIT_IMM, 127, start);
#endif
OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)start_bits);
if (!HAS_VIRTUAL_REGISTERS)
{
OP2(SLJIT_SHL, TMP3, 0, SLJIT_IMM, 1, TMP2, 0);
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, TMP3, 0);
}
else
{
OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP1, 0, TMP2, 0);
}
JUMPTO(SLJIT_ZERO, start);
}
else
set_jumps(matches, start);
#if PCRE2_CODE_UNIT_WIDTH != 8
if (found != NULL)
JUMPHERE(found);
#endif
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (common->mode != PCRE2_JIT_COMPLETE)
JUMPHERE(partial_quit);
if (common->match_end_ptr != 0)
OP1(SLJIT_MOV, STR_END, 0, RETURN_ADDR, 0);
}
static SLJIT_INLINE jump_list *search_requested_char(compiler_common *common, PCRE2_UCHAR req_char, BOOL caseless, BOOL has_firstchar)
{
DEFINE_COMPILER;
struct sljit_label *loop;
struct sljit_jump *toolong;
struct sljit_jump *already_found;
struct sljit_jump *found;
struct sljit_jump *found_oc = NULL;
jump_list *not_found = NULL;
sljit_u32 oc, bit;
SLJIT_ASSERT(common->req_char_ptr != 0);
OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(REQ_CU_MAX) * 100);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->req_char_ptr);
toolong = CMP(SLJIT_LESS, TMP2, 0, STR_END, 0);
already_found = CMP(SLJIT_LESS, STR_PTR, 0, TMP1, 0);
if (has_firstchar)
OP2(SLJIT_ADD, TMP1, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
else
OP1(SLJIT_MOV, TMP1, 0, STR_PTR, 0);
oc = req_char;
if (caseless)
{
oc = TABLE_GET(req_char, common->fcc, req_char);
#if defined SUPPORT_UNICODE
if (req_char > 127 && (common->utf || common->ucp))
oc = UCD_OTHERCASE(req_char);
#endif
}
#ifdef JIT_HAS_FAST_REQUESTED_CHAR_SIMD
if (JIT_HAS_FAST_REQUESTED_CHAR_SIMD)
{
not_found = fast_requested_char_simd(common, req_char, oc);
}
else
#endif
{
loop = LABEL();
add_jump(compiler, ¬_found, CMP(SLJIT_GREATER_EQUAL, TMP1, 0, STR_END, 0));
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(TMP1), 0);
if (req_char == oc)
found = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, req_char);
else
{
bit = req_char ^ oc;
if (is_powerof2(bit))
{
OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, bit);
found = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, req_char | bit);
}
else
{
found = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, req_char);
found_oc = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, oc);
}
}
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1));
JUMPTO(SLJIT_JUMP, loop);
JUMPHERE(found);
if (found_oc)
JUMPHERE(found_oc);
}
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->req_char_ptr, TMP1, 0);
JUMPHERE(already_found);
JUMPHERE(toolong);
return not_found;
}
static void do_revertframes(compiler_common *common)
{
DEFINE_COMPILER;
struct sljit_jump *jump;
struct sljit_label *mainloop;
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
GET_LOCAL_BASE(TMP1, 0, 0);
/* Drop frames until we reach STACK_TOP. */
mainloop = LABEL();
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), -SSIZE_OF(sw));
OP2U(SLJIT_SUB | SLJIT_SET_SIG_LESS_EQUAL | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, 0);
jump = JUMP(SLJIT_SIG_LESS_EQUAL);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP1, 0);
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(STACK_TOP), -(2 * SSIZE_OF(sw)));
OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_sw), SLJIT_MEM1(STACK_TOP), -(3 * SSIZE_OF(sw)));
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 3 * SSIZE_OF(sw));
}
else
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), -(2 * SSIZE_OF(sw)));
OP1(SLJIT_MOV, TMP3, 0, SLJIT_MEM1(STACK_TOP), -(3 * SSIZE_OF(sw)));
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 3 * SSIZE_OF(sw));
OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, TMP1, 0);
GET_LOCAL_BASE(TMP1, 0, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP3, 0);
}
JUMPTO(SLJIT_JUMP, mainloop);
JUMPHERE(jump);
sljit_set_current_flags(compiler, SLJIT_CURRENT_FLAGS_SUB | SLJIT_CURRENT_FLAGS_COMPARE | SLJIT_SET_SIG_LESS_EQUAL | SLJIT_SET_Z);
jump = JUMP(SLJIT_NOT_ZERO /* SIG_LESS */);
/* End of reverting values. */
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
JUMPHERE(jump);
OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, TMP2, 0);
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, SLJIT_MEM1(STACK_TOP), -(2 * SSIZE_OF(sw)));
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 2 * SSIZE_OF(sw));
}
else
{
OP1(SLJIT_MOV, TMP3, 0, SLJIT_MEM1(STACK_TOP), -(2 * SSIZE_OF(sw)));
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 2 * SSIZE_OF(sw));
OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), 0, TMP3, 0);
}
JUMPTO(SLJIT_JUMP, mainloop);
}
#ifdef SUPPORT_UNICODE
#define UCPCAT(bit) (1 << (bit))
#define UCPCAT2(bit1, bit2) (UCPCAT(bit1) | UCPCAT(bit2))
#define UCPCAT3(bit1, bit2, bit3) (UCPCAT(bit1) | UCPCAT(bit2) | UCPCAT(bit3))
#define UCPCAT_RANGE(start, end) (((1 << ((end) + 1)) - 1) - ((1 << (start)) - 1))
#define UCPCAT_L UCPCAT_RANGE(ucp_Ll, ucp_Lu)
#define UCPCAT_N UCPCAT_RANGE(ucp_Nd, ucp_No)
#define UCPCAT_ALL ((1 << (ucp_Zs + 1)) - 1)
#endif
static void check_wordboundary(compiler_common *common, BOOL ucp)
{
DEFINE_COMPILER;
struct sljit_jump *skipread;
jump_list *skipread_list = NULL;
#ifdef SUPPORT_UNICODE
struct sljit_label *valid_utf;
jump_list *invalid_utf1 = NULL;
#endif /* SUPPORT_UNICODE */
jump_list *invalid_utf2 = NULL;
#if PCRE2_CODE_UNIT_WIDTH != 8 || defined SUPPORT_UNICODE
struct sljit_jump *jump;
#endif /* PCRE2_CODE_UNIT_WIDTH != 8 || SUPPORT_UNICODE */
SLJIT_UNUSED_ARG(ucp);
SLJIT_COMPILE_ASSERT(ctype_word == 0x10, ctype_word_must_be_16);
SLJIT_ASSERT(common->locals_size >= 2 * SSIZE_OF(sw));
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, SLJIT_MEM1(SLJIT_SP), LOCAL0);
/* Get type of the previous char, and put it to TMP3. */
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin));
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 0);
skipread = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0);
#ifdef SUPPORT_UNICODE
if (common->invalid_utf)
{
peek_char_back(common, READ_CHAR_MAX, &invalid_utf1);
if (common->mode != PCRE2_JIT_COMPLETE)
{
OP1(SLJIT_MOV, RETURN_ADDR, 0, TMP1, 0);
OP1(SLJIT_MOV, TMP2, 0, STR_PTR, 0);
move_back(common, NULL, TRUE);
check_start_used_ptr(common);
OP1(SLJIT_MOV, TMP1, 0, RETURN_ADDR, 0);
OP1(SLJIT_MOV, STR_PTR, 0, TMP2, 0);
}
}
else
#endif /* SUPPORT_UNICODE */
{
if (common->mode == PCRE2_JIT_COMPLETE)
peek_char_back(common, READ_CHAR_MAX, NULL);
else
{
move_back(common, NULL, TRUE);
check_start_used_ptr(common);
read_char(common, 0, READ_CHAR_MAX, NULL, READ_CHAR_UPDATE_STR_PTR);
}
}
/* Testing char type. */
#ifdef SUPPORT_UNICODE
if (ucp)
{
add_jump(compiler, &common->getucdtype, JUMP(SLJIT_FAST_CALL));
OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP1, 0);
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, UCPCAT2(ucp_Mn, ucp_Pc) | UCPCAT_L | UCPCAT_N);
OP_FLAGS(SLJIT_MOV, TMP3, 0, SLJIT_NOT_ZERO);
}
else
#endif /* SUPPORT_UNICODE */
{
#if PCRE2_CODE_UNIT_WIDTH != 8
jump = CMP(SLJIT_GREATER, TMP1, 0, SLJIT_IMM, 255);
#elif defined SUPPORT_UNICODE
/* Here TMP3 has already been zeroed. */
jump = NULL;
if (common->utf)
jump = CMP(SLJIT_GREATER, TMP1, 0, SLJIT_IMM, 255);
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */
OP1(SLJIT_MOV_U8, TMP1, 0, SLJIT_MEM1(TMP1), common->ctypes);
OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 4 /* ctype_word */);
OP2(SLJIT_AND, TMP3, 0, TMP1, 0, SLJIT_IMM, 1);
#if PCRE2_CODE_UNIT_WIDTH != 8
JUMPHERE(jump);
#elif defined SUPPORT_UNICODE
if (jump != NULL)
JUMPHERE(jump);
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */
}
JUMPHERE(skipread);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);
check_str_end(common, &skipread_list);
peek_char(common, READ_CHAR_MAX, SLJIT_MEM1(SLJIT_SP), LOCAL1, &invalid_utf2);
/* Testing char type. This is a code duplication. */
#ifdef SUPPORT_UNICODE
valid_utf = LABEL();
if (ucp)
{
add_jump(compiler, &common->getucdtype, JUMP(SLJIT_FAST_CALL));
OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP1, 0);
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, UCPCAT2(ucp_Mn, ucp_Pc) | UCPCAT_L | UCPCAT_N);
OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_NOT_ZERO);
}
else
#endif /* SUPPORT_UNICODE */
{
#if PCRE2_CODE_UNIT_WIDTH != 8
/* TMP2 may be destroyed by peek_char. */
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);
jump = CMP(SLJIT_GREATER, TMP1, 0, SLJIT_IMM, 255);
#elif defined SUPPORT_UNICODE
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);
jump = NULL;
if (common->utf)
jump = CMP(SLJIT_GREATER, TMP1, 0, SLJIT_IMM, 255);
#endif
OP1(SLJIT_MOV_U8, TMP2, 0, SLJIT_MEM1(TMP1), common->ctypes);
OP2(SLJIT_LSHR, TMP2, 0, TMP2, 0, SLJIT_IMM, 4 /* ctype_word */);
OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 1);
#if PCRE2_CODE_UNIT_WIDTH != 8
JUMPHERE(jump);
#elif defined SUPPORT_UNICODE
if (jump != NULL)
JUMPHERE(jump);
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */
}
set_jumps(skipread_list, LABEL());
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
OP2(SLJIT_XOR | SLJIT_SET_Z, TMP2, 0, TMP2, 0, TMP3, 0);
OP_SRC(SLJIT_FAST_RETURN, TMP1, 0);
#ifdef SUPPORT_UNICODE
if (common->invalid_utf)
{
set_jumps(invalid_utf1, LABEL());
peek_char(common, READ_CHAR_MAX, SLJIT_MEM1(SLJIT_SP), LOCAL1, NULL);
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, INVALID_UTF_CHAR, valid_utf);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, -1);
OP_SRC(SLJIT_FAST_RETURN, TMP1, 0);
set_jumps(invalid_utf2, LABEL());
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
OP1(SLJIT_MOV, TMP2, 0, TMP3, 0);
OP_SRC(SLJIT_FAST_RETURN, TMP1, 0);
}
#endif /* SUPPORT_UNICODE */
}
static BOOL optimize_class_ranges(compiler_common *common, const sljit_u8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks)
{
/* May destroy TMP1. */
DEFINE_COMPILER;
int ranges[MAX_CLASS_RANGE_SIZE];
sljit_u8 bit, cbit, all;
int i, byte, length = 0;
bit = bits[0] & 0x1;
/* All bits will be zero or one (since bit is zero or one). */
all = (sljit_u8)-bit;
for (i = 0; i < 256; )
{
byte = i >> 3;
if ((i & 0x7) == 0 && bits[byte] == all)
i += 8;
else
{
cbit = (bits[byte] >> (i & 0x7)) & 0x1;
if (cbit != bit)
{
if (length >= MAX_CLASS_RANGE_SIZE)
return FALSE;
ranges[length] = i;
length++;
bit = cbit;
all = (sljit_u8)-cbit; /* sign extend bit into byte */
}
i++;
}
}
if (((bit == 0) && nclass) || ((bit == 1) && !nclass))
{
if (length >= MAX_CLASS_RANGE_SIZE)
return FALSE;
ranges[length] = 256;
length++;
}
if (length < 0 || length > 4)
return FALSE;
bit = bits[0] & 0x1;
if (invert) bit ^= 0x1;
/* No character is accepted. */
if (length == 0 && bit == 0)
add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
switch(length)
{
case 0:
/* When bit != 0, all characters are accepted. */
return TRUE;
case 1:
add_jump(compiler, backtracks, CMP(bit == 0 ? SLJIT_LESS : SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
return TRUE;
case 2:
if (ranges[0] + 1 != ranges[1])
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_LESS : SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
}
else
add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_EQUAL : SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
return TRUE;
case 3:
if (bit != 0)
{
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
if (ranges[0] + 1 != ranges[1])
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
}
else
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
return TRUE;
}
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, ranges[0]));
if (ranges[1] + 1 != ranges[2])
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1]);
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1]));
}
else
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1]));
return TRUE;
case 4:
if ((ranges[1] - ranges[0]) == (ranges[3] - ranges[2])
&& (ranges[0] | (ranges[2] - ranges[0])) == ranges[2]
&& (ranges[1] & (ranges[2] - ranges[0])) == 0
&& is_powerof2(ranges[2] - ranges[0]))
{
SLJIT_ASSERT((ranges[0] & (ranges[2] - ranges[0])) == 0 && (ranges[2] & ranges[3] & (ranges[2] - ranges[0])) != 0);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[0]);
if (ranges[2] + 1 != ranges[3])
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]);
add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_LESS : SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
}
else
add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_EQUAL : SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
return TRUE;
}
if (bit != 0)
{
i = 0;
if (ranges[0] + 1 != ranges[1])
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
i = ranges[0];
}
else
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
if (ranges[2] + 1 != ranges[3])
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2] - i);
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
}
else
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2] - i));
return TRUE;
}
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[0]));
if (ranges[1] + 1 != ranges[2])
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]);
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1]));
}
else
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
return TRUE;
default:
SLJIT_UNREACHABLE();
return FALSE;
}
}
static BOOL optimize_class_chars(compiler_common *common, const sljit_u8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks)
{
/* May destroy TMP1. */
DEFINE_COMPILER;
uint16_t char_list[MAX_CLASS_CHARS_SIZE];
uint8_t byte;
sljit_s32 type;
int i, j, k, len, c;
if (!sljit_has_cpu_feature(SLJIT_HAS_CMOV))
return FALSE;
len = 0;
for (i = 0; i < 32; i++)
{
byte = bits[i];
if (nclass)
byte = (sljit_u8)~byte;
j = 0;
while (byte != 0)
{
if (byte & 0x1)
{
c = i * 8 + j;
k = len;
if ((c & 0x20) != 0)
{
for (k = 0; k < len; k++)
if (char_list[k] == c - 0x20)
{
char_list[k] |= 0x120;
break;
}
}
if (k == len)
{
if (len >= MAX_CLASS_CHARS_SIZE)
return FALSE;
char_list[len++] = (uint16_t) c;
}
}
byte >>= 1;
j++;
}
}
if (len == 0) return FALSE; /* Should never occur, but stops analyzers complaining. */
i = 0;
j = 0;
if (char_list[0] == 0)
{
i++;
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0);
OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_ZERO);
}
else
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);
while (i < len)
{
if ((char_list[i] & 0x100) != 0)
j++;
else
{
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, char_list[i]);
SELECT(SLJIT_ZERO, TMP2, TMP1, 0, TMP2);
}
i++;
}
if (j != 0)
{
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x20);
for (i = 0; i < len; i++)
if ((char_list[i] & 0x100) != 0)
{
j--;
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, char_list[i] & 0xff);
SELECT(SLJIT_ZERO, TMP2, TMP1, 0, TMP2);
}
}
if (invert)
nclass = !nclass;
type = nclass ? SLJIT_NOT_EQUAL : SLJIT_EQUAL;
add_jump(compiler, backtracks, CMP(type, TMP2, 0, SLJIT_IMM, 0));
return TRUE;
}
static BOOL optimize_class(compiler_common *common, const sljit_u8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks)
{
/* May destroy TMP1. */
if (optimize_class_ranges(common, bits, nclass, invert, backtracks))
return TRUE;
return optimize_class_chars(common, bits, nclass, invert, backtracks);
}
static void check_anynewline(compiler_common *common)
{
/* Check whether TMP1 contains a newline character. TMP2 destroyed. */
DEFINE_COMPILER;
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a);
OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a);
OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS_EQUAL);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a);
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
#if PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf)
{
#endif
OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a);
#if PCRE2_CODE_UNIT_WIDTH == 8
}
#endif
#endif /* SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH == [16|32] */
OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_EQUAL);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void check_hspace(compiler_common *common)
{
/* Check whether TMP1 contains a newline character. TMP2 destroyed. */
DEFINE_COMPILER;
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x09);
OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_EQUAL);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x20);
OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0xa0);
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
#if PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf)
{
#endif
OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x1680);
OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x180e);
OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL);
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x2000);
OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0x200A - 0x2000);
OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_LESS_EQUAL);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x202f - 0x2000);
OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x205f - 0x2000);
OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x3000 - 0x2000);
#if PCRE2_CODE_UNIT_WIDTH == 8
}
#endif
#endif /* SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH == [16|32] */
OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_EQUAL);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void check_vspace(compiler_common *common)
{
/* Check whether TMP1 contains a newline character. TMP2 destroyed. */
DEFINE_COMPILER;
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, RETURN_ADDR, 0);
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0a);
OP2U(SLJIT_SUB | SLJIT_SET_LESS_EQUAL, TMP1, 0, SLJIT_IMM, 0x0d - 0x0a);
OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS_EQUAL);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x85 - 0x0a);
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
#if PCRE2_CODE_UNIT_WIDTH == 8
if (common->utf)
{
#endif
OP_FLAGS(SLJIT_OR, TMP2, 0, SLJIT_EQUAL);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, 0x2029 - 0x0a);
#if PCRE2_CODE_UNIT_WIDTH == 8
}
#endif
#endif /* SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH == [16|32] */
OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_EQUAL);
OP_SRC(SLJIT_FAST_RETURN, RETURN_ADDR, 0);
}
static void do_casefulcmp(compiler_common *common)
{
DEFINE_COMPILER;
struct sljit_jump *jump;
struct sljit_label *label;
int char1_reg;
int char2_reg;
if (HAS_VIRTUAL_REGISTERS)
{
char1_reg = STR_END;
char2_reg = STACK_TOP;
}
else
{
char1_reg = TMP3;
char2_reg = RETURN_ADDR;
}
/* Update ref_update_local_size() when this changes. */
SLJIT_ASSERT(common->locals_size >= SSIZE_OF(sw));
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, SLJIT_MEM1(SLJIT_SP), LOCAL0);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
if (char1_reg == STR_END)
{
OP1(SLJIT_MOV, TMP3, 0, char1_reg, 0);
OP1(SLJIT_MOV, RETURN_ADDR, 0, char2_reg, 0);
}
if (sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_SUPP | SLJIT_MEM_POST, char1_reg, SLJIT_MEM1(TMP1), IN_UCHARS(1)) == SLJIT_SUCCESS)
{
label = LABEL();
sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_POST, char1_reg, SLJIT_MEM1(TMP1), IN_UCHARS(1));
sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_POST, char2_reg, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
jump = CMP(SLJIT_NOT_EQUAL, char1_reg, 0, char2_reg, 0);
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1));
JUMPTO(SLJIT_NOT_ZERO, label);
JUMPHERE(jump);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
}
else if (sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_SUPP | SLJIT_MEM_PRE, char1_reg, SLJIT_MEM1(TMP1), IN_UCHARS(1)) == SLJIT_SUCCESS)
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
label = LABEL();
sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_PRE, char1_reg, SLJIT_MEM1(TMP1), IN_UCHARS(1));
sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_PRE, char2_reg, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
jump = CMP(SLJIT_NOT_EQUAL, char1_reg, 0, char2_reg, 0);
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1));
JUMPTO(SLJIT_NOT_ZERO, label);
JUMPHERE(jump);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
}
else
{
label = LABEL();
OP1(MOV_UCHAR, char1_reg, 0, SLJIT_MEM1(TMP1), 0);
OP1(MOV_UCHAR, char2_reg, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
jump = CMP(SLJIT_NOT_EQUAL, char1_reg, 0, char2_reg, 0);
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1));
JUMPTO(SLJIT_NOT_ZERO, label);
JUMPHERE(jump);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
}
if (char1_reg == STR_END)
{
OP1(SLJIT_MOV, char1_reg, 0, TMP3, 0);
OP1(SLJIT_MOV, char2_reg, 0, RETURN_ADDR, 0);
}
OP_SRC(SLJIT_FAST_RETURN, TMP1, 0);
}
static void do_caselesscmp(compiler_common *common)
{
DEFINE_COMPILER;
struct sljit_jump *jump;
struct sljit_label *label;
int char1_reg = STR_END;
int char2_reg;
int lcc_table;
int opt_type = 0;
if (HAS_VIRTUAL_REGISTERS)
{
char2_reg = STACK_TOP;
lcc_table = STACK_LIMIT;
}
else
{
char2_reg = RETURN_ADDR;
lcc_table = TMP3;
}
if (sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_SUPP | SLJIT_MEM_POST, char1_reg, SLJIT_MEM1(TMP1), IN_UCHARS(1)) == SLJIT_SUCCESS)
opt_type = 1;
else if (sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_SUPP | SLJIT_MEM_PRE, char1_reg, SLJIT_MEM1(TMP1), IN_UCHARS(1)) == SLJIT_SUCCESS)
opt_type = 2;
/* Update ref_update_local_size() when this changes. */
SLJIT_ASSERT(common->locals_size >= 2 * SSIZE_OF(sw));
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, SLJIT_MEM1(SLJIT_SP), LOCAL0);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL1, char1_reg, 0);
if (char2_reg == STACK_TOP)
{
OP1(SLJIT_MOV, TMP3, 0, char2_reg, 0);
OP1(SLJIT_MOV, RETURN_ADDR, 0, lcc_table, 0);
}
OP1(SLJIT_MOV, lcc_table, 0, SLJIT_IMM, common->lcc);
if (opt_type == 1)
{
label = LABEL();
sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_POST, char1_reg, SLJIT_MEM1(TMP1), IN_UCHARS(1));
sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_POST, char2_reg, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
}
else if (opt_type == 2)
{
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1));
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
label = LABEL();
sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_PRE, char1_reg, SLJIT_MEM1(TMP1), IN_UCHARS(1));
sljit_emit_mem_update(compiler, MOV_UCHAR | SLJIT_MEM_PRE, char2_reg, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
}
else
{
label = LABEL();
OP1(MOV_UCHAR, char1_reg, 0, SLJIT_MEM1(TMP1), 0);
OP1(MOV_UCHAR, char2_reg, 0, SLJIT_MEM1(STR_PTR), 0);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, IN_UCHARS(1));
}
#if PCRE2_CODE_UNIT_WIDTH != 8
jump = CMP(SLJIT_GREATER, char1_reg, 0, SLJIT_IMM, 255);
#endif
OP1(SLJIT_MOV_U8, char1_reg, 0, SLJIT_MEM2(lcc_table, char1_reg), 0);
#if PCRE2_CODE_UNIT_WIDTH != 8
JUMPHERE(jump);
jump = CMP(SLJIT_GREATER, char2_reg, 0, SLJIT_IMM, 255);
#endif
OP1(SLJIT_MOV_U8, char2_reg, 0, SLJIT_MEM2(lcc_table, char2_reg), 0);
#if PCRE2_CODE_UNIT_WIDTH != 8
JUMPHERE(jump);
#endif
if (opt_type == 0)
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
jump = CMP(SLJIT_NOT_EQUAL, char1_reg, 0, char2_reg, 0);
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP2, 0, TMP2, 0, SLJIT_IMM, IN_UCHARS(1));
JUMPTO(SLJIT_NOT_ZERO, label);
JUMPHERE(jump);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
if (opt_type == 2)
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
if (char2_reg == STACK_TOP)
{
OP1(SLJIT_MOV, char2_reg, 0, TMP3, 0);
OP1(SLJIT_MOV, lcc_table, 0, RETURN_ADDR, 0);
}
OP1(SLJIT_MOV, char1_reg, 0, SLJIT_MEM1(SLJIT_SP), LOCAL1);
OP_SRC(SLJIT_FAST_RETURN, TMP1, 0);
}
#include "pcre2_jit_char_inc.h"
static PCRE2_SPTR compile_simple_assertion_matchingpath(compiler_common *common, PCRE2_UCHAR type, PCRE2_SPTR cc, jump_list **backtracks)
{
DEFINE_COMPILER;
struct sljit_jump *jump[4];
switch(type)
{
case OP_SOD:
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin));
}
else
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, STR_PTR, 0, TMP1, 0));
return cc;
case OP_SOM:
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str));
}
else
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, str));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, STR_PTR, 0, TMP1, 0));
return cc;
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
case OP_NOT_UCP_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
add_jump(compiler, (type == OP_NOT_WORD_BOUNDARY || type == OP_WORD_BOUNDARY) ? &common->wordboundary : &common->ucp_wordboundary, JUMP(SLJIT_FAST_CALL));
#ifdef SUPPORT_UNICODE
if (common->invalid_utf)
{
add_jump(compiler, backtracks, CMP((type == OP_NOT_WORD_BOUNDARY || type == OP_NOT_UCP_WORD_BOUNDARY) ? SLJIT_NOT_EQUAL : SLJIT_SIG_LESS_EQUAL, TMP2, 0, SLJIT_IMM, 0));
return cc;
}
#endif /* SUPPORT_UNICODE */
sljit_set_current_flags(compiler, SLJIT_SET_Z);
add_jump(compiler, backtracks, JUMP((type == OP_NOT_WORD_BOUNDARY || type == OP_NOT_UCP_WORD_BOUNDARY) ? SLJIT_NOT_ZERO : SLJIT_ZERO));
return cc;
case OP_EODN:
/* Requires rather complex checks. */
jump[0] = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
if (common->nltype == NLTYPE_FIXED && common->newline > 255)
{
OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
if (common->mode == PCRE2_JIT_COMPLETE)
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP2, 0, STR_END, 0));
else
{
jump[1] = CMP(SLJIT_EQUAL, TMP2, 0, STR_END, 0);
OP2U(SLJIT_SUB | SLJIT_SET_LESS, TMP2, 0, STR_END, 0);
OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_LESS);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff);
OP_FLAGS(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, SLJIT_NOT_EQUAL);
add_jump(compiler, backtracks, JUMP(SLJIT_NOT_EQUAL));
check_partial(common, TRUE);
add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
JUMPHERE(jump[1]);
}
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff));
}
else if (common->nltype == NLTYPE_FIXED)
{
OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP2, 0, STR_END, 0));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline));
}
else
{
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
jump[1] = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
OP2U(SLJIT_SUB | SLJIT_SET_Z | SLJIT_SET_GREATER, TMP2, 0, STR_END, 0);
jump[2] = JUMP(SLJIT_GREATER);
add_jump(compiler, backtracks, JUMP(SLJIT_NOT_EQUAL) /* LESS */);
/* Equal. */
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
jump[3] = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL);
add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
JUMPHERE(jump[1]);
if (common->nltype == NLTYPE_ANYCRLF)
{
OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP2, 0, STR_END, 0));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL));
}
else
{
OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0);
read_char(common, common->nlmin, common->nlmax, backtracks, READ_CHAR_UPDATE_STR_PTR);
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, STR_PTR, 0, STR_END, 0));
add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));
sljit_set_current_flags(compiler, SLJIT_SET_Z);
add_jump(compiler, backtracks, JUMP(SLJIT_ZERO));
OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0);
}
JUMPHERE(jump[2]);
JUMPHERE(jump[3]);
}
JUMPHERE(jump[0]);
if (common->mode != PCRE2_JIT_COMPLETE)
check_partial(common, TRUE);
return cc;
case OP_EOD:
add_jump(compiler, backtracks, CMP(SLJIT_LESS, STR_PTR, 0, STR_END, 0));
if (common->mode != PCRE2_JIT_COMPLETE)
check_partial(common, TRUE);
return cc;
case OP_DOLL:
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0);
OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL);
}
else
OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL);
add_jump(compiler, backtracks, JUMP(SLJIT_NOT_ZERO));
if (!common->endonly)
compile_simple_assertion_matchingpath(common, OP_EODN, cc, backtracks);
else
{
add_jump(compiler, backtracks, CMP(SLJIT_LESS, STR_PTR, 0, STR_END, 0));
check_partial(common, FALSE);
}
return cc;
case OP_DOLLM:
jump[1] = CMP(SLJIT_LESS, STR_PTR, 0, STR_END, 0);
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0);
OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL);
}
else
OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTEOL);
add_jump(compiler, backtracks, JUMP(SLJIT_NOT_ZERO));
check_partial(common, FALSE);
jump[0] = JUMP(SLJIT_JUMP);
JUMPHERE(jump[1]);
if (common->nltype == NLTYPE_FIXED && common->newline > 255)
{
OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
if (common->mode == PCRE2_JIT_COMPLETE)
add_jump(compiler, backtracks, CMP(SLJIT_GREATER, TMP2, 0, STR_END, 0));
else
{
jump[1] = CMP(SLJIT_LESS_EQUAL, TMP2, 0, STR_END, 0);
/* STR_PTR = STR_END - IN_UCHARS(1) */
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff));
check_partial(common, TRUE);
add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
JUMPHERE(jump[1]);
}
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff));
}
else
{
peek_char(common, common->nlmax, TMP3, 0, NULL);
check_newlinechar(common, common->nltype, backtracks, FALSE);
}
JUMPHERE(jump[0]);
return cc;
case OP_CIRC:
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP2, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, begin));
add_jump(compiler, backtracks, CMP(SLJIT_GREATER, STR_PTR, 0, TMP1, 0));
OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL);
add_jump(compiler, backtracks, JUMP(SLJIT_NOT_ZERO));
}
else
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin));
add_jump(compiler, backtracks, CMP(SLJIT_GREATER, STR_PTR, 0, TMP1, 0));
OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL);
add_jump(compiler, backtracks, JUMP(SLJIT_NOT_ZERO));
}
return cc;
case OP_CIRCM:
/* TMP2 might be used by peek_char_back. */
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin));
jump[1] = CMP(SLJIT_GREATER, STR_PTR, 0, TMP2, 0);
OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL);
}
else
{
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin));
jump[1] = CMP(SLJIT_GREATER, STR_PTR, 0, TMP2, 0);
OP2U(SLJIT_AND32 | SLJIT_SET_Z, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options), SLJIT_IMM, PCRE2_NOTBOL);
}
add_jump(compiler, backtracks, JUMP(SLJIT_NOT_ZERO));
jump[0] = JUMP(SLJIT_JUMP);
JUMPHERE(jump[1]);
if (!common->alt_circumflex)
add_jump(compiler, backtracks, CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
if (common->nltype == NLTYPE_FIXED && common->newline > 255)
{
OP2(SLJIT_SUB, TMP1, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
add_jump(compiler, backtracks, CMP(SLJIT_LESS, TMP1, 0, TMP2, 0));
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, common->newline & 0xff));
}
else
{
peek_char_back(common, common->nlmax, backtracks);
check_newlinechar(common, common->nltype, backtracks, FALSE);
}
JUMPHERE(jump[0]);
return cc;
}
SLJIT_UNREACHABLE();
return cc;
}
/* Forward definitions. */
static void compile_matchingpath(compiler_common *, PCRE2_SPTR, PCRE2_SPTR, backtrack_common *);
static void compile_backtrackingpath(compiler_common *, struct backtrack_common *);
#define PUSH_BACKTRACK(size, ccstart, error) \
do \
{ \
backtrack = sljit_alloc_memory(compiler, (size)); \
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \
return error; \
memset(backtrack, 0, size); \
backtrack->prev = parent->top; \
backtrack->cc = (ccstart); \
parent->top = backtrack; \
} \
while (0)
#define PUSH_BACKTRACK_NOVALUE(size, ccstart) \
do \
{ \
backtrack = sljit_alloc_memory(compiler, (size)); \
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \
return; \
memset(backtrack, 0, size); \
backtrack->prev = parent->top; \
backtrack->cc = (ccstart); \
parent->top = backtrack; \
} \
while (0)
#define BACKTRACK_AS(type) ((type *)backtrack)
static void compile_dnref_search(compiler_common *common, PCRE2_SPTR cc, jump_list **backtracks)
{
/* The OVECTOR offset goes to TMP2. */
DEFINE_COMPILER;
int count = GET2(cc, 1 + IMM2_SIZE);
PCRE2_SPTR slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
unsigned int offset;
jump_list *found = NULL;
SLJIT_ASSERT(*cc == OP_DNREF || *cc == OP_DNREFI);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1));
count--;
while (count-- > 0)
{
offset = GET2(slot, 0) << 1;
GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
add_jump(compiler, &found, CMP(SLJIT_NOT_EQUAL, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset), TMP1, 0));
slot += common->name_entry_size;
}
offset = GET2(slot, 0) << 1;
GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
if (backtracks != NULL && !common->unset_backref)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset), TMP1, 0));
set_jumps(found, LABEL());
}
static void compile_ref_matchingpath(compiler_common *common, PCRE2_SPTR cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail)
{
DEFINE_COMPILER;
BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
int offset = 0;
struct sljit_jump *jump = NULL;
struct sljit_jump *partial;
struct sljit_jump *nopartial;
#if defined SUPPORT_UNICODE
struct sljit_label *loop;
struct sljit_label *caseless_loop;
struct sljit_jump *turkish_ascii_i = NULL;
struct sljit_jump *turkish_non_ascii_i = NULL;
jump_list *no_match = NULL;
int source_reg = COUNT_MATCH;
int source_end_reg = ARGUMENTS;
int char1_reg = STACK_LIMIT;
PCRE2_UCHAR refi_flag = 0;
if (*cc == OP_REFI || *cc == OP_DNREFI)
refi_flag = cc[PRIV(OP_lengths)[*cc] - 1];
#endif /* SUPPORT_UNICODE */
if (ref)
{
offset = GET2(cc, 1) << 1;
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset));
/* OVECTOR(1) contains the "string begin - 1" constant. */
if (withchecks && !common->unset_backref)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1)));
}
else
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
#if defined SUPPORT_UNICODE
if ((common->utf || common->ucp) && (*cc == OP_REFI || *cc == OP_DNREFI))
{
/* Update ref_update_local_size() when this changes. */
SLJIT_ASSERT(common->locals_size >= 3 * SSIZE_OF(sw));
if (ref)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1));
else
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
if (withchecks && emptyfail)
add_jump(compiler, backtracks, CMP(SLJIT_EQUAL, TMP1, 0, TMP2, 0));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL0, source_reg, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL1, source_end_reg, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL2, char1_reg, 0);
OP1(SLJIT_MOV, source_reg, 0, TMP1, 0);
OP1(SLJIT_MOV, source_end_reg, 0, TMP2, 0);
loop = LABEL();
jump = CMP(SLJIT_GREATER_EQUAL, source_reg, 0, source_end_reg, 0);
partial = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
/* Read original character. It must be a valid UTF character. */
OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0);
OP1(SLJIT_MOV, STR_PTR, 0, source_reg, 0);
read_char(common, 0, READ_CHAR_MAX, NULL, READ_CHAR_UPDATE_STR_PTR | READ_CHAR_VALID_UTF);
OP1(SLJIT_MOV, source_reg, 0, STR_PTR, 0);
OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0);
OP1(SLJIT_MOV, char1_reg, 0, TMP1, 0);
/* Read second character. */
read_char(common, 0, READ_CHAR_MAX, &no_match, READ_CHAR_UPDATE_STR_PTR);
CMPTO(SLJIT_EQUAL, TMP1, 0, char1_reg, 0, loop);
if ((refi_flag & (REFI_FLAG_TURKISH_CASING|REFI_FLAG_CASELESS_RESTRICT)) ==
REFI_FLAG_TURKISH_CASING)
{
OP2(SLJIT_OR, SLJIT_TMP_DEST_REG, 0, char1_reg, 0, SLJIT_IMM, 0x20);
turkish_ascii_i = CMP(SLJIT_EQUAL, SLJIT_TMP_DEST_REG, 0, SLJIT_IMM, 0x69);
OP2(SLJIT_OR, SLJIT_TMP_DEST_REG, 0, char1_reg, 0, SLJIT_IMM, 0x1);
turkish_non_ascii_i = CMP(SLJIT_EQUAL, SLJIT_TMP_DEST_REG, 0, SLJIT_IMM, 0x131);
}
OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
add_jump(compiler, &common->getucd, JUMP(SLJIT_FAST_CALL));
OP2(SLJIT_SHL, TMP1, 0, TMP2, 0, SLJIT_IMM, 2);
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 3);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, TMP1, 0);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_records));
OP1(SLJIT_MOV_S32, TMP1, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(ucd_record, other_case));
OP1(SLJIT_MOV_U8, TMP2, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(ucd_record, caseset));
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, TMP3, 0);
CMPTO(SLJIT_EQUAL, TMP1, 0, char1_reg, 0, loop);
add_jump(compiler, &no_match, CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, 0));
OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 2);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, (sljit_sw)PRIV(ucd_caseless_sets));
if (refi_flag & REFI_FLAG_CASELESS_RESTRICT)
add_jump(compiler, &no_match, CMP(SLJIT_LESS | SLJIT_32, SLJIT_MEM1(TMP2), 0, SLJIT_IMM, 128));
caseless_loop = LABEL();
OP1(SLJIT_MOV_U32, TMP1, 0, SLJIT_MEM1(TMP2), 0);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, sizeof(uint32_t));
OP2U(SLJIT_SUB | SLJIT_SET_Z | SLJIT_SET_LESS, TMP1, 0, char1_reg, 0);
JUMPTO(SLJIT_EQUAL, loop);
JUMPTO(SLJIT_LESS, caseless_loop);
if ((refi_flag & (REFI_FLAG_TURKISH_CASING|REFI_FLAG_CASELESS_RESTRICT)) ==
REFI_FLAG_TURKISH_CASING)
{
add_jump(compiler, &no_match, JUMP(SLJIT_JUMP));
JUMPHERE(turkish_ascii_i);
OP2(SLJIT_LSHR, char1_reg, 0, char1_reg, 0, SLJIT_IMM, 5);
OP2(SLJIT_AND, char1_reg, 0, char1_reg, 0, SLJIT_IMM, 1);
OP2(SLJIT_XOR, char1_reg, 0, char1_reg, 0, SLJIT_IMM, 1);
OP2(SLJIT_ADD, char1_reg, 0, char1_reg, 0, SLJIT_IMM, 0x130);
CMPTO(SLJIT_EQUAL, TMP1, 0, char1_reg, 0, loop);
add_jump(compiler, &no_match, JUMP(SLJIT_JUMP));
JUMPHERE(turkish_non_ascii_i);
OP2(SLJIT_AND, char1_reg, 0, char1_reg, 0, SLJIT_IMM, 1);
OP2(SLJIT_XOR, char1_reg, 0, char1_reg, 0, SLJIT_IMM, 1);
OP2(SLJIT_SHL, char1_reg, 0, char1_reg, 0, SLJIT_IMM, 5);
OP2(SLJIT_ADD, char1_reg, 0, char1_reg, 0, SLJIT_IMM, 0x49);
CMPTO(SLJIT_EQUAL, TMP1, 0, char1_reg, 0, loop);
}
set_jumps(no_match, LABEL());
if (common->mode == PCRE2_JIT_COMPLETE)
JUMPHERE(partial);
OP1(SLJIT_MOV, source_reg, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
OP1(SLJIT_MOV, source_end_reg, 0, SLJIT_MEM1(SLJIT_SP), LOCAL1);
OP1(SLJIT_MOV, char1_reg, 0, SLJIT_MEM1(SLJIT_SP), LOCAL2);
add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
if (common->mode != PCRE2_JIT_COMPLETE)
{
JUMPHERE(partial);
OP1(SLJIT_MOV, source_reg, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
OP1(SLJIT_MOV, source_end_reg, 0, SLJIT_MEM1(SLJIT_SP), LOCAL1);
OP1(SLJIT_MOV, char1_reg, 0, SLJIT_MEM1(SLJIT_SP), LOCAL2);
check_partial(common, FALSE);
add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
}
JUMPHERE(jump);
OP1(SLJIT_MOV, source_reg, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
OP1(SLJIT_MOV, source_end_reg, 0, SLJIT_MEM1(SLJIT_SP), LOCAL1);
OP1(SLJIT_MOV, char1_reg, 0, SLJIT_MEM1(SLJIT_SP), LOCAL2);
return;
}
else
#endif /* SUPPORT_UNICODE */
{
if (ref)
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), TMP1, 0);
else
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0);
if (withchecks)
jump = JUMP(SLJIT_ZERO);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
partial = CMP(SLJIT_GREATER, STR_PTR, 0, STR_END, 0);
if (common->mode == PCRE2_JIT_COMPLETE)
add_jump(compiler, backtracks, partial);
add_jump(compiler, (*cc == OP_REF || *cc == OP_DNREF) ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0));
if (common->mode != PCRE2_JIT_COMPLETE)
{
nopartial = JUMP(SLJIT_JUMP);
JUMPHERE(partial);
/* TMP2 -= STR_END - STR_PTR */
OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, STR_PTR, 0);
OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, STR_END, 0);
partial = CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_IMM, 0);
OP1(SLJIT_MOV, STR_PTR, 0, STR_END, 0);
add_jump(compiler, (*cc == OP_REF || *cc == OP_DNREF) ? &common->casefulcmp : &common->caselesscmp, JUMP(SLJIT_FAST_CALL));
add_jump(compiler, backtracks, CMP(SLJIT_NOT_EQUAL, TMP2, 0, SLJIT_IMM, 0));
JUMPHERE(partial);
check_partial(common, FALSE);
add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
JUMPHERE(nopartial);
}
}
if (jump != NULL)
{
if (emptyfail)
add_jump(compiler, backtracks, jump);
else
JUMPHERE(jump);
}
}
static SLJIT_INLINE PCRE2_SPTR compile_ref_iterator_matchingpath(compiler_common *common, PCRE2_SPTR cc, backtrack_common *parent)
{
DEFINE_COMPILER;
BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
backtrack_common *backtrack;
PCRE2_UCHAR type;
int local_start = LOCAL2;
int offset = 0;
struct sljit_label *label;
struct sljit_jump *zerolength;
struct sljit_jump *jump = NULL;
PCRE2_SPTR ccbegin = cc;
int min = 0, max = 0;
BOOL minimize;
PUSH_BACKTRACK(sizeof(ref_iterator_backtrack), cc, NULL);
if (ref)
offset = GET2(cc, 1) << 1;
else
cc += IMM2_SIZE;
if (*ccbegin == OP_REFI || *ccbegin == OP_DNREFI)
{
cc += 1;
#ifdef SUPPORT_UNICODE
if (common->utf || common->ucp)
local_start = LOCAL3;
#endif
}
type = cc[1 + IMM2_SIZE];
SLJIT_COMPILE_ASSERT((OP_CRSTAR & 0x1) == 0, crstar_opcode_must_be_even);
/* Update ref_update_local_size() when this changes. */
SLJIT_ASSERT(local_start + 2 * SSIZE_OF(sw) <= (int)LOCAL0 + common->locals_size);
minimize = (type & 0x1) != 0;
switch(type)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
min = 0;
max = 0;
cc += 1 + IMM2_SIZE + 1;
break;
case OP_CRPLUS:
case OP_CRMINPLUS:
min = 1;
max = 0;
cc += 1 + IMM2_SIZE + 1;
break;
case OP_CRQUERY:
case OP_CRMINQUERY:
min = 0;
max = 1;
cc += 1 + IMM2_SIZE + 1;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
min = GET2(cc, 1 + IMM2_SIZE + 1);
max = GET2(cc, 1 + IMM2_SIZE + 1 + IMM2_SIZE);
cc += 1 + IMM2_SIZE + 1 + 2 * IMM2_SIZE;
break;
default:
SLJIT_UNREACHABLE();
break;
}
if (!minimize)
{
if (min == 0)
{
allocate_stack(common, 2);
if (ref)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);
/* Temporary release of STR_PTR. */
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));
/* Handles both invalid and empty cases. Since the minimum repeat,
is zero the invalid case is basically the same as an empty case. */
if (ref)
zerolength = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1));
else
{
compile_dnref_search(common, ccbegin, NULL);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), local_start + SSIZE_OF(sw), TMP2, 0);
zerolength = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
}
/* Restore if not zero length. */
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));
}
else
{
allocate_stack(common, 1);
if (ref)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
if (ref)
{
if (!common->unset_backref)
add_jump(compiler, &backtrack->own_backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1)));
zerolength = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1));
}
else
{
compile_dnref_search(common, ccbegin, &backtrack->own_backtracks);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), local_start + SSIZE_OF(sw), TMP2, 0);
zerolength = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
}
}
if (min > 1 || max > 1)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), local_start, SLJIT_IMM, 0);
label = LABEL();
if (!ref)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), local_start + SSIZE_OF(sw));
compile_ref_matchingpath(common, ccbegin, &backtrack->own_backtracks, FALSE, FALSE);
if (min > 1 || max > 1)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), local_start);
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), local_start, TMP1, 0);
if (min > 1)
CMPTO(SLJIT_LESS, TMP1, 0, SLJIT_IMM, min, label);
if (max > 1)
{
jump = CMP(SLJIT_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, max);
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
JUMPTO(SLJIT_JUMP, label);
JUMPHERE(jump);
}
}
if (max == 0)
{
/* Includes min > 1 case as well. */
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
JUMPTO(SLJIT_JUMP, label);
}
JUMPHERE(zerolength);
BACKTRACK_AS(ref_iterator_backtrack)->matchingpath = LABEL();
count_match(common);
return cc;
}
allocate_stack(common, ref ? 2 : 3);
if (ref)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
if (type != OP_CRMINSTAR)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);
if (min == 0)
{
/* Handles both invalid and empty cases. Since the minimum repeat,
is zero the invalid case is basically the same as an empty case. */
if (ref)
zerolength = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1));
else
{
compile_dnref_search(common, ccbegin, NULL);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0);
zerolength = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
}
/* Length is non-zero, we can match real repeats. */
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
jump = JUMP(SLJIT_JUMP);
}
else
{
if (ref)
{
if (!common->unset_backref)
add_jump(compiler, &backtrack->own_backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1)));
zerolength = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1));
}
else
{
compile_dnref_search(common, ccbegin, &backtrack->own_backtracks);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0);
zerolength = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
}
}
BACKTRACK_AS(ref_iterator_backtrack)->matchingpath = LABEL();
if (max > 0)
add_jump(compiler, &backtrack->own_backtracks, CMP(SLJIT_GREATER_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, max));
if (!ref)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(2));
compile_ref_matchingpath(common, ccbegin, &backtrack->own_backtracks, TRUE, TRUE);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
if (min > 1)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0);
CMPTO(SLJIT_LESS, TMP1, 0, SLJIT_IMM, min, BACKTRACK_AS(ref_iterator_backtrack)->matchingpath);
}
else if (max > 0)
OP2(SLJIT_ADD, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 1);
if (jump != NULL)
JUMPHERE(jump);
JUMPHERE(zerolength);
count_match(common);
return cc;
}
static SLJIT_INLINE PCRE2_SPTR compile_recurse_matchingpath(compiler_common *common, PCRE2_SPTR cc, backtrack_common *parent)
{
DEFINE_COMPILER;
backtrack_common *backtrack;
recurse_entry *entry = common->entries;
recurse_entry *prev = NULL;
sljit_sw start = GET(cc, 1);
PCRE2_SPTR start_cc;
BOOL needs_control_head;
PUSH_BACKTRACK(sizeof(recurse_backtrack), cc, NULL);
/* Inlining simple patterns. */
if (get_framesize(common, common->start + start, NULL, TRUE, &needs_control_head) == no_stack)
{
start_cc = common->start + start;
compile_matchingpath(common, next_opcode(common, start_cc), bracketend(start_cc) - (1 + LINK_SIZE), backtrack);
BACKTRACK_AS(recurse_backtrack)->inlined_pattern = TRUE;
return cc + 1 + LINK_SIZE;
}
while (entry != NULL)
{
if (entry->start == start)
break;
prev = entry;
entry = entry->next;
}
if (entry == NULL)
{
entry = sljit_alloc_memory(compiler, sizeof(recurse_entry));
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
return NULL;
entry->next = NULL;
entry->entry_label = NULL;
entry->backtrack_label = NULL;
entry->entry_calls = NULL;
entry->backtrack_calls = NULL;
entry->start = start;
if (prev != NULL)
prev->next = entry;
else
common->entries = entry;
}
BACKTRACK_AS(recurse_backtrack)->entry = entry;
if (entry->entry_label == NULL)
add_jump(compiler, &entry->entry_calls, JUMP(SLJIT_FAST_CALL));
else
JUMPTO(SLJIT_FAST_CALL, entry->entry_label);
/* Leave if the match is failed. */
add_jump(compiler, &backtrack->own_backtracks, CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, 0));
BACKTRACK_AS(recurse_backtrack)->matchingpath = LABEL();
return cc + 1 + LINK_SIZE;
}
static sljit_s32 SLJIT_FUNC do_callout_jit(struct jit_arguments *arguments, pcre2_callout_block *callout_block, PCRE2_SPTR *jit_ovector)
{
PCRE2_SPTR begin;
PCRE2_SIZE *ovector;
sljit_u32 oveccount, capture_top;
if (arguments->callout == NULL)
return 0;
SLJIT_COMPILE_ASSERT(sizeof (PCRE2_SIZE) <= sizeof (sljit_sw), pcre2_size_must_be_lower_than_sljit_sw_size);
begin = arguments->begin;
ovector = (PCRE2_SIZE*)(callout_block + 1);
oveccount = callout_block->capture_top;
SLJIT_ASSERT(oveccount >= 1);
callout_block->version = 2;
callout_block->callout_flags = 0;
/* Offsets in subject. */
callout_block->subject_length = arguments->end - arguments->begin;
callout_block->start_match = jit_ovector[0] - begin;
callout_block->current_position = (PCRE2_SPTR)callout_block->offset_vector - begin;
callout_block->subject = begin;
/* Convert and copy the JIT offset vector to the ovector array. */
callout_block->capture_top = 1;
callout_block->offset_vector = ovector;
ovector[0] = PCRE2_UNSET;
ovector[1] = PCRE2_UNSET;
ovector += 2;
jit_ovector += 2;
capture_top = 1;
/* Convert pointers to sizes. */
while (--oveccount != 0)
{
capture_top++;
ovector[0] = (PCRE2_SIZE)(jit_ovector[0] - begin);
ovector[1] = (PCRE2_SIZE)(jit_ovector[1] - begin);
if (ovector[0] != PCRE2_UNSET)
callout_block->capture_top = capture_top;
ovector += 2;
jit_ovector += 2;
}
return (arguments->callout)(callout_block, arguments->callout_data);
}
#define CALLOUT_ARG_OFFSET(arg) \
SLJIT_OFFSETOF(pcre2_callout_block, arg)
static SLJIT_INLINE PCRE2_SPTR compile_callout_matchingpath(compiler_common *common, PCRE2_SPTR cc, backtrack_common *parent)
{
DEFINE_COMPILER;
backtrack_common *backtrack;
sljit_s32 mov_opcode;
unsigned int callout_length = (*cc == OP_CALLOUT)
? PRIV(OP_lengths)[OP_CALLOUT] : GET(cc, 1 + 2 * LINK_SIZE);
sljit_sw value1;
sljit_sw value2;
sljit_sw value3;
sljit_uw callout_arg_size = (common->re->top_bracket + 1) * 2 * SSIZE_OF(sw);
PUSH_BACKTRACK(sizeof(backtrack_common), cc, NULL);
callout_arg_size = (sizeof(pcre2_callout_block) + callout_arg_size + sizeof(sljit_sw) - 1) / sizeof(sljit_sw);
allocate_stack(common, callout_arg_size);
SLJIT_ASSERT(common->capture_last_ptr != 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr);
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
value1 = (*cc == OP_CALLOUT) ? cc[1 + 2 * LINK_SIZE] : 0;
OP1(SLJIT_MOV_U32, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(callout_number), SLJIT_IMM, value1);
OP1(SLJIT_MOV_U32, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(capture_last), TMP2, 0);
OP1(SLJIT_MOV_U32, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(capture_top), SLJIT_IMM, common->re->top_bracket + 1);
/* These pointer sized fields temporarly stores internal variables. */
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(offset_vector), STR_PTR, 0);
if (common->mark_ptr != 0)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, mark_ptr));
mov_opcode = (sizeof(PCRE2_SIZE) == 4) ? SLJIT_MOV_U32 : SLJIT_MOV;
OP1(mov_opcode, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(pattern_position), SLJIT_IMM, GET(cc, 1));
OP1(mov_opcode, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(next_item_length), SLJIT_IMM, GET(cc, 1 + LINK_SIZE));
if (*cc == OP_CALLOUT)
{
value1 = 0;
value2 = 0;
value3 = 0;
}
else
{
value1 = (sljit_sw) (cc + (1 + 4*LINK_SIZE) + 1);
value2 = (callout_length - (1 + 4*LINK_SIZE + 2));
value3 = (sljit_sw) (GET(cc, 1 + 3*LINK_SIZE));
}
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(callout_string), SLJIT_IMM, value1);
OP1(mov_opcode, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(callout_string_length), SLJIT_IMM, value2);
OP1(mov_opcode, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(callout_string_offset), SLJIT_IMM, value3);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), CALLOUT_ARG_OFFSET(mark), (common->mark_ptr != 0) ? TMP2 : SLJIT_IMM, 0);
SLJIT_ASSERT(TMP1 == SLJIT_R0 && STR_PTR == SLJIT_R1);
/* Needed to save important temporary registers. */
SLJIT_ASSERT(common->locals_size >= SSIZE_OF(sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL0, STR_PTR, 0);
/* SLJIT_R0 = arguments */
OP1(SLJIT_MOV, SLJIT_R1, 0, STACK_TOP, 0);
GET_LOCAL_BASE(SLJIT_R2, 0, OVECTOR_START);
sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS3(32, W, W, W), SLJIT_IMM, SLJIT_FUNC_ADDR(do_callout_jit));
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
free_stack(common, callout_arg_size);
/* Check return value. */
OP2U(SLJIT_SUB32 | SLJIT_SET_Z | SLJIT_SET_SIG_GREATER, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0);
add_jump(compiler, &backtrack->own_backtracks, JUMP(SLJIT_SIG_GREATER));
if (common->abort_label == NULL)
add_jump(compiler, &common->abort, JUMP(SLJIT_NOT_EQUAL) /* SIG_LESS */);
else
JUMPTO(SLJIT_NOT_EQUAL /* SIG_LESS */, common->abort_label);
return cc + callout_length;
}
#undef CALLOUT_ARG_SIZE
#undef CALLOUT_ARG_OFFSET
static PCRE2_SPTR compile_reverse_matchingpath(compiler_common *common, PCRE2_SPTR cc, backtrack_common *parent)
{
DEFINE_COMPILER;
backtrack_common *backtrack = NULL;
jump_list **reverse_failed;
unsigned int lmin, lmax;
#ifdef SUPPORT_UNICODE
struct sljit_jump *jump;
struct sljit_label *label;
#endif
SLJIT_ASSERT(parent->top == NULL);
if (*cc == OP_REVERSE)
{
reverse_failed = &parent->own_backtracks;
lmin = GET2(cc, 1);
lmax = lmin;
cc += 1 + IMM2_SIZE;
SLJIT_ASSERT(lmin > 0);
}
else
{
SLJIT_ASSERT(*cc == OP_VREVERSE);
PUSH_BACKTRACK(sizeof(vreverse_backtrack), cc, NULL);
reverse_failed = &backtrack->own_backtracks;
lmin = GET2(cc, 1);
lmax = GET2(cc, 1 + IMM2_SIZE);
cc += 1 + 2 * IMM2_SIZE;
SLJIT_ASSERT(lmin < lmax);
}
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, begin));
}
else
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, begin));
#ifdef SUPPORT_UNICODE
if (common->utf)
{
if (lmin > 0)
{
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, lmin);
label = LABEL();
add_jump(compiler, reverse_failed, CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0));
move_back(common, reverse_failed, FALSE);
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
}
if (lmin < lmax)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(3), STR_PTR, 0);
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, lmax - lmin);
label = LABEL();
jump = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0);
move_back(common, reverse_failed, FALSE);
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
JUMPHERE(jump);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), STR_PTR, 0);
}
}
else
#endif
{
if (lmin > 0)
{
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(lmin));
add_jump(compiler, reverse_failed, CMP(SLJIT_LESS, STR_PTR, 0, TMP2, 0));
}
if (lmin < lmax)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(3), STR_PTR, 0);
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(lmax - lmin));
OP2U(SLJIT_SUB | SLJIT_SET_LESS, STR_PTR, 0, TMP2, 0);
SELECT(SLJIT_LESS, STR_PTR, TMP2, 0, STR_PTR);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), STR_PTR, 0);
}
}
check_start_used_ptr(common);
if (lmin < lmax)
BACKTRACK_AS(vreverse_backtrack)->matchingpath = LABEL();
return cc;
}
static SLJIT_INLINE BOOL assert_needs_str_ptr_saving(PCRE2_SPTR cc)
{
while (TRUE)
{
switch (*cc)
{
case OP_CALLOUT_STR:
cc += GET(cc, 1 + 2*LINK_SIZE);
break;
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
case OP_CIRC:
case OP_CIRCM:
case OP_DOLL:
case OP_DOLLM:
case OP_CALLOUT:
case OP_ALT:
case OP_NOT_UCP_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
cc += PRIV(OP_lengths)[*cc];
break;
case OP_KET:
return FALSE;
default:
return TRUE;
}
}
}
static PCRE2_SPTR compile_assert_matchingpath(compiler_common *common, PCRE2_SPTR cc, assert_backtrack *backtrack, BOOL conditional)
{
DEFINE_COMPILER;
int framesize;
int extrasize;
BOOL local_quit_available = FALSE;
BOOL needs_control_head;
BOOL end_block_size = 0;
BOOL has_vreverse;
int private_data_ptr;
backtrack_common altbacktrack;
PCRE2_SPTR ccbegin;
PCRE2_UCHAR opcode;
PCRE2_UCHAR bra = OP_BRA;
jump_list *tmp = NULL;
jump_list **target = (conditional) ? &backtrack->condfailed : &backtrack->common.own_backtracks;
jump_list **found;
/* Saving previous accept variables. */
BOOL save_local_quit_available = common->local_quit_available;
BOOL save_in_positive_assertion = common->in_positive_assertion;
sljit_s32 save_restore_end_ptr = common->restore_end_ptr;
then_trap_backtrack *save_then_trap = common->then_trap;
struct sljit_label *save_quit_label = common->quit_label;
struct sljit_label *save_accept_label = common->accept_label;
jump_list *save_quit = common->quit;
jump_list *save_positive_assertion_quit = common->positive_assertion_quit;
jump_list *save_accept = common->accept;
struct sljit_jump *jump;
struct sljit_jump *brajump = NULL;
/* Assert captures then. */
common->then_trap = NULL;
if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO)
{
SLJIT_ASSERT(!conditional);
bra = *cc;
cc++;
}
private_data_ptr = PRIVATE_DATA(cc);
SLJIT_ASSERT(private_data_ptr != 0);
framesize = get_framesize(common, cc, NULL, FALSE, &needs_control_head);
backtrack->framesize = framesize;
backtrack->private_data_ptr = private_data_ptr;
opcode = *cc;
SLJIT_ASSERT(opcode >= OP_ASSERT && opcode <= OP_ASSERTBACK_NOT);
found = (opcode == OP_ASSERT || opcode == OP_ASSERTBACK) ? &tmp : target;
ccbegin = cc;
cc += GET(cc, 1);
if (bra == OP_BRAMINZERO)
{
/* This is a braminzero backtrack path. */
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
brajump = CMP(SLJIT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0);
}
if ((opcode == OP_ASSERTBACK || opcode == OP_ASSERTBACK_NOT) && find_vreverse(ccbegin))
end_block_size = 3;
if (framesize < 0)
{
extrasize = 1;
if (bra == OP_BRA && !assert_needs_str_ptr_saving(ccbegin + 1 + LINK_SIZE))
extrasize = 0;
extrasize += end_block_size;
if (needs_control_head)
extrasize++;
if (framesize == no_frame)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STACK_TOP, 0);
if (extrasize > 0)
allocate_stack(common, extrasize);
if (needs_control_head)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr);
if (extrasize > 0)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
if (needs_control_head)
{
SLJIT_ASSERT(extrasize == end_block_size + 2);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_IMM, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(end_block_size + 1), TMP1, 0);
}
}
else
{
extrasize = (needs_control_head ? 3 : 2) + end_block_size;
OP1(SLJIT_MOV, TMP2, 0, STACK_TOP, 0);
allocate_stack(common, framesize + extrasize);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP2, 0);
if (needs_control_head)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
if (needs_control_head)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(end_block_size + 2), TMP1, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(end_block_size + 1), TMP2, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_IMM, 0);
}
else
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(end_block_size + 1), TMP1, 0);
init_frame(common, ccbegin, NULL, framesize + extrasize - 1, extrasize);
}
if (end_block_size > 0)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), STR_END, 0);
OP1(SLJIT_MOV, STR_END, 0, STR_PTR, 0);
}
memset(&altbacktrack, 0, sizeof(backtrack_common));
if (conditional || (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT))
{
/* Control verbs cannot escape from these asserts. */
local_quit_available = TRUE;
common->restore_end_ptr = 0;
common->local_quit_available = TRUE;
common->quit_label = NULL;
common->quit = NULL;
}
common->in_positive_assertion = (opcode == OP_ASSERT || opcode == OP_ASSERTBACK);
common->positive_assertion_quit = NULL;
while (1)
{
common->accept_label = NULL;
common->accept = NULL;
altbacktrack.top = NULL;
altbacktrack.own_backtracks = NULL;
if (*ccbegin == OP_ALT && extrasize > 0)
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
altbacktrack.cc = ccbegin;
ccbegin += 1 + LINK_SIZE;
has_vreverse = (*ccbegin == OP_VREVERSE);
if (*ccbegin == OP_REVERSE || has_vreverse)
ccbegin = compile_reverse_matchingpath(common, ccbegin, &altbacktrack);
compile_matchingpath(common, ccbegin, cc, &altbacktrack);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
{
if (local_quit_available)
{
common->local_quit_available = save_local_quit_available;
common->quit_label = save_quit_label;
common->quit = save_quit;
}
common->in_positive_assertion = save_in_positive_assertion;
common->restore_end_ptr = save_restore_end_ptr;
common->then_trap = save_then_trap;
common->accept_label = save_accept_label;
common->positive_assertion_quit = save_positive_assertion_quit;
common->accept = save_accept;
return NULL;
}
if (has_vreverse)
{
SLJIT_ASSERT(altbacktrack.top != NULL);
add_jump(compiler, &altbacktrack.top->simple_backtracks, CMP(SLJIT_LESS, STR_PTR, 0, STR_END, 0));
}
common->accept_label = LABEL();
if (common->accept != NULL)
set_jumps(common->accept, common->accept_label);
/* Reset stack. */
if (framesize < 0)
{
if (framesize == no_frame)
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
else if (extrasize > 0)
free_stack(common, extrasize);
if (end_block_size > 0)
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(STACK_TOP), STACK(-extrasize + 1));
if (needs_control_head)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), STACK(-1));
}
else
{
if ((opcode != OP_ASSERT_NOT && opcode != OP_ASSERTBACK_NOT) || conditional)
{
/* We don't need to keep the STR_PTR, only the previous private_data_ptr. */
OP2(SLJIT_SUB, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_sw));
if (end_block_size > 0)
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(STACK_TOP), STACK(-extrasize + 2));
if (needs_control_head)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), STACK(-1));
}
else
{
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
if (end_block_size > 0)
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(STACK_TOP), STACK(-framesize - extrasize + 1));
if (needs_control_head)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), STACK(-framesize - 2));
add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL));
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize - 1) * sizeof(sljit_sw));
}
}
if (opcode == OP_ASSERT_NOT || opcode == OP_ASSERTBACK_NOT)
{
/* We know that STR_PTR was stored on the top of the stack. */
if (conditional)
{
if (extrasize > 0)
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(-end_block_size - (needs_control_head ? 2 : 1)));
}
else if (bra == OP_BRAZERO)
{
if (framesize < 0)
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(-extrasize));
else
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(-framesize - 1));
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(-framesize - extrasize));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP1, 0);
}
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
}
else if (framesize >= 0)
{
/* For OP_BRA and OP_BRAMINZERO. */
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, SLJIT_MEM1(STACK_TOP), STACK(-framesize - 1));
}
}
add_jump(compiler, found, JUMP(SLJIT_JUMP));
compile_backtrackingpath(common, altbacktrack.top);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
{
if (local_quit_available)
{
common->local_quit_available = save_local_quit_available;
common->quit_label = save_quit_label;
common->quit = save_quit;
}
common->in_positive_assertion = save_in_positive_assertion;
common->restore_end_ptr = save_restore_end_ptr;
common->then_trap = save_then_trap;
common->accept_label = save_accept_label;
common->positive_assertion_quit = save_positive_assertion_quit;
common->accept = save_accept;
return NULL;
}
set_jumps(altbacktrack.own_backtracks, LABEL());
if (*cc != OP_ALT)
break;
ccbegin = cc;
cc += GET(cc, 1);
}
if (local_quit_available)
{
SLJIT_ASSERT(common->positive_assertion_quit == NULL);
/* Makes the check less complicated below. */
common->positive_assertion_quit = common->quit;
}
/* None of them matched. */
if (common->positive_assertion_quit != NULL)
{
jump = JUMP(SLJIT_JUMP);
set_jumps(common->positive_assertion_quit, LABEL());
SLJIT_ASSERT(framesize != no_stack);
if (framesize < 0)
OP2(SLJIT_SUB, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr, SLJIT_IMM, extrasize * sizeof(sljit_sw));
else
{
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL));
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (extrasize + 1) * sizeof(sljit_sw));
}
JUMPHERE(jump);
}
if (end_block_size > 0)
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
if (needs_control_head)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), STACK(end_block_size + 1));
if (opcode == OP_ASSERT || opcode == OP_ASSERTBACK)
{
/* Assert is failed. */
if ((conditional && extrasize > 0) || bra == OP_BRAZERO)
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
if (framesize < 0)
{
/* The topmost item should be 0. */
if (bra == OP_BRAZERO)
{
if (extrasize >= 2)
free_stack(common, extrasize - 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
}
else if (extrasize > 0)
free_stack(common, extrasize);
}
else
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(extrasize - 1));
/* The topmost item should be 0. */
if (bra == OP_BRAZERO)
{
free_stack(common, framesize + extrasize - 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
}
else
free_stack(common, framesize + extrasize);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP1, 0);
}
jump = JUMP(SLJIT_JUMP);
if (bra != OP_BRAZERO)
add_jump(compiler, target, jump);
/* Assert is successful. */
set_jumps(tmp, LABEL());
if (framesize < 0)
{
/* We know that STR_PTR was stored on the top of the stack. */
if (extrasize > 0)
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(-extrasize));
/* Keep the STR_PTR on the top of the stack. */
if (bra == OP_BRAZERO)
{
/* This allocation is always successful. */
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));
if (extrasize >= 2)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
}
else if (bra == OP_BRAMINZERO)
{
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
}
}
else
{
if (bra == OP_BRA)
{
/* We don't need to keep the STR_PTR, only the previous private_data_ptr. */
OP2(SLJIT_SUB, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr, SLJIT_IMM, (framesize + 1) * sizeof(sljit_sw));
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(-extrasize + 1));
}
else
{
/* We don't need to keep the STR_PTR, only the previous private_data_ptr. */
OP2(SLJIT_SUB, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr, SLJIT_IMM, (framesize + end_block_size + 2) * sizeof(sljit_sw));
if (extrasize == 2 + end_block_size)
{
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
if (bra == OP_BRAMINZERO)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
}
else
{
SLJIT_ASSERT(extrasize == 3 + end_block_size);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(-1));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), bra == OP_BRAZERO ? STR_PTR : SLJIT_IMM, 0);
}
}
}
if (bra == OP_BRAZERO)
{
backtrack->matchingpath = LABEL();
SET_LABEL(jump, backtrack->matchingpath);
}
else if (bra == OP_BRAMINZERO)
{
JUMPTO(SLJIT_JUMP, backtrack->matchingpath);
JUMPHERE(brajump);
SLJIT_ASSERT(framesize != 0);
if (framesize > 0)
{
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(-2));
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize - 1) * sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP1, 0);
}
set_jumps(backtrack->common.own_backtracks, LABEL());
}
}
else
{
/* AssertNot is successful. */
if (framesize < 0)
{
if (extrasize > 0)
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
if (bra != OP_BRA)
{
if (extrasize >= 2)
free_stack(common, extrasize - 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
}
else if (extrasize > 0)
free_stack(common, extrasize);
}
else
{
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(extrasize - 1));
/* The topmost item should be 0. */
if (bra != OP_BRA)
{
free_stack(common, framesize + extrasize - 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
}
else
free_stack(common, framesize + extrasize);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP1, 0);
}
if (bra == OP_BRAZERO)
backtrack->matchingpath = LABEL();
else if (bra == OP_BRAMINZERO)
{
JUMPTO(SLJIT_JUMP, backtrack->matchingpath);
JUMPHERE(brajump);
}
if (bra != OP_BRA)
{
SLJIT_ASSERT(found == &backtrack->common.own_backtracks);
set_jumps(backtrack->common.own_backtracks, LABEL());
backtrack->common.own_backtracks = NULL;
}
}
if (local_quit_available)
{
common->local_quit_available = save_local_quit_available;
common->quit_label = save_quit_label;
common->quit = save_quit;
}
common->in_positive_assertion = save_in_positive_assertion;
common->restore_end_ptr = save_restore_end_ptr;
common->then_trap = save_then_trap;
common->accept_label = save_accept_label;
common->positive_assertion_quit = save_positive_assertion_quit;
common->accept = save_accept;
return cc + 1 + LINK_SIZE;
}
static SLJIT_INLINE void match_once_common(compiler_common *common, PCRE2_UCHAR ket, int framesize, int private_data_ptr, BOOL has_alternatives, BOOL needs_control_head)
{
DEFINE_COMPILER;
int stacksize;
if (framesize < 0)
{
if (framesize == no_frame)
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
else
{
stacksize = needs_control_head ? 1 : 0;
if (ket != OP_KET || has_alternatives)
stacksize++;
if (stacksize > 0)
free_stack(common, stacksize);
}
if (needs_control_head)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), (ket != OP_KET || has_alternatives) ? STACK(-2) : STACK(-1));
/* TMP2 which is set here used by OP_KETRMAX below. */
if (ket == OP_KETRMAX)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(-1));
else if (ket == OP_KETRMIN)
{
/* Move the STR_PTR to the private_data_ptr. */
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, SLJIT_MEM1(STACK_TOP), STACK(-1));
}
}
else
{
stacksize = (ket != OP_KET || has_alternatives) ? 2 : 1;
OP2(SLJIT_SUB, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr, SLJIT_IMM, (framesize + stacksize) * sizeof(sljit_sw));
if (needs_control_head)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(-1));
if (ket == OP_KETRMAX)
{
/* TMP2 which is set here used by OP_KETRMAX below. */
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
}
}
if (needs_control_head)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, TMP1, 0);
}
static SLJIT_INLINE int match_capture_common(compiler_common *common, int stacksize, int offset, int private_data_ptr)
{
DEFINE_COMPILER;
if (common->capture_last_ptr != 0)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr, SLJIT_IMM, offset >> 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP1, 0);
stacksize++;
}
if (common->optimized_cbracket[offset >> 1] == 0)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset));
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP1, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize + 1), TMP2, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset), TMP1, 0);
stacksize += 2;
}
return stacksize;
}
static PCRE2_SPTR SLJIT_FUNC do_script_run(PCRE2_SPTR ptr, PCRE2_SPTR endptr)
{
if (PRIV(script_run)(ptr, endptr, FALSE))
return endptr;
return NULL;
}
#ifdef SUPPORT_UNICODE
static PCRE2_SPTR SLJIT_FUNC do_script_run_utf(PCRE2_SPTR ptr, PCRE2_SPTR endptr)
{
if (PRIV(script_run)(ptr, endptr, TRUE))
return endptr;
return NULL;
}
#endif /* SUPPORT_UNICODE */
static void match_script_run_common(compiler_common *common, int private_data_ptr, backtrack_common *parent)
{
DEFINE_COMPILER;
SLJIT_ASSERT(TMP1 == SLJIT_R0 && STR_PTR == SLJIT_R1);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
#ifdef SUPPORT_UNICODE
sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS2(W, W, W), SLJIT_IMM,
common->utf ? SLJIT_FUNC_ADDR(do_script_run_utf) : SLJIT_FUNC_ADDR(do_script_run));
#else
sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS2(W, W, W), SLJIT_IMM, SLJIT_FUNC_ADDR(do_script_run));
#endif
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_RETURN_REG, 0);
add_jump(compiler, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks, CMP(SLJIT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0));
}
/*
Handling bracketed expressions is probably the most complex part.
Stack layout naming characters:
S - Push the current STR_PTR
0 - Push a 0 (NULL)
A - Push the current STR_PTR. Needed for restoring the STR_PTR
before the next alternative. Not pushed if there are no alternatives.
M - Any values pushed by the current alternative. Can be empty, or anything.
C - Push the previous OVECTOR(i), OVECTOR(i+1) and OVECTOR_PRIV(i) to the stack.
L - Push the previous local (pointed by localptr) to the stack
() - opional values stored on the stack
()* - optonal, can be stored multiple times
The following list shows the regular expression templates, their PCRE byte codes
and stack layout supported by pcre-sljit.
(?:) OP_BRA | OP_KET A M
() OP_CBRA | OP_KET C M
(?:)+ OP_BRA | OP_KETRMAX 0 A M S ( A M S )*
OP_SBRA | OP_KETRMAX 0 L M S ( L M S )*
(?:)+? OP_BRA | OP_KETRMIN 0 A M S ( A M S )*
OP_SBRA | OP_KETRMIN 0 L M S ( L M S )*
()+ OP_CBRA | OP_KETRMAX 0 C M S ( C M S )*
OP_SCBRA | OP_KETRMAX 0 C M S ( C M S )*
()+? OP_CBRA | OP_KETRMIN 0 C M S ( C M S )*
OP_SCBRA | OP_KETRMIN 0 C M S ( C M S )*
(?:)? OP_BRAZERO | OP_BRA | OP_KET S ( A M 0 )
(?:)?? OP_BRAMINZERO | OP_BRA | OP_KET S ( A M 0 )
()? OP_BRAZERO | OP_CBRA | OP_KET S ( C M 0 )
()?? OP_BRAMINZERO | OP_CBRA | OP_KET S ( C M 0 )
(?:)* OP_BRAZERO | OP_BRA | OP_KETRMAX S 0 ( A M S )*
OP_BRAZERO | OP_SBRA | OP_KETRMAX S 0 ( L M S )*
(?:)*? OP_BRAMINZERO | OP_BRA | OP_KETRMIN S 0 ( A M S )*
OP_BRAMINZERO | OP_SBRA | OP_KETRMIN S 0 ( L M S )*
()* OP_BRAZERO | OP_CBRA | OP_KETRMAX S 0 ( C M S )*
OP_BRAZERO | OP_SCBRA | OP_KETRMAX S 0 ( C M S )*
()*? OP_BRAMINZERO | OP_CBRA | OP_KETRMIN S 0 ( C M S )*
OP_BRAMINZERO | OP_SCBRA | OP_KETRMIN S 0 ( C M S )*
Stack layout naming characters:
A - Push the alternative index (starting from 0) on the stack.
Not pushed if there is no alternatives.
M - Any values pushed by the current alternative. Can be empty, or anything.
The next list shows the possible content of a bracket:
(|) OP_*BRA | OP_ALT ... M A
(?()|) OP_*COND | OP_ALT M A
(?>|) OP_ONCE | OP_ALT ... [stack trace] M A
Or nothing, if trace is unnecessary
*/
static PCRE2_SPTR compile_bracket_matchingpath(compiler_common *common, PCRE2_SPTR cc, backtrack_common *parent)
{
DEFINE_COMPILER;
backtrack_common *backtrack;
PCRE2_UCHAR opcode;
int private_data_ptr = 0;
int offset = 0;
int i, stacksize;
int repeat_ptr = 0, repeat_length = 0;
int repeat_type = 0, repeat_count = 0;
PCRE2_SPTR ccbegin;
PCRE2_SPTR matchingpath;
PCRE2_SPTR slot;
PCRE2_UCHAR bra = OP_BRA;
PCRE2_UCHAR ket;
assert_backtrack *assert;
BOOL has_alternatives;
BOOL needs_control_head = FALSE;
BOOL has_vreverse = FALSE;
struct sljit_jump *jump;
struct sljit_jump *skip;
jump_list *jumplist;
struct sljit_label *rmax_label = NULL;
struct sljit_jump *braminzero = NULL;
PUSH_BACKTRACK(sizeof(bracket_backtrack), cc, NULL);
if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO)
{
bra = *cc;
cc++;
opcode = *cc;
}
opcode = *cc;
ccbegin = cc;
matchingpath = bracketend(cc) - 1 - LINK_SIZE;
ket = *matchingpath;
if (ket == OP_KET && PRIVATE_DATA(matchingpath) != 0)
{
repeat_ptr = PRIVATE_DATA(matchingpath);
repeat_length = PRIVATE_DATA(matchingpath + 1);
repeat_type = PRIVATE_DATA(matchingpath + 2);
repeat_count = PRIVATE_DATA(matchingpath + 3);
SLJIT_ASSERT(repeat_length != 0 && repeat_type != 0 && repeat_count != 0);
if (repeat_type == OP_UPTO)
ket = OP_KETRMAX;
if (repeat_type == OP_MINUPTO)
ket = OP_KETRMIN;
}
matchingpath = ccbegin + 1 + LINK_SIZE;
SLJIT_ASSERT(ket == OP_KET || ket == OP_KETRMAX || ket == OP_KETRMIN);
SLJIT_ASSERT(!((bra == OP_BRAZERO && ket == OP_KETRMIN) || (bra == OP_BRAMINZERO && ket == OP_KETRMAX)));
cc += GET(cc, 1);
has_alternatives = *cc == OP_ALT;
if (SLJIT_UNLIKELY(opcode == OP_COND || opcode == OP_SCOND))
{
SLJIT_COMPILE_ASSERT(OP_DNRREF == OP_RREF + 1 && OP_FALSE == OP_RREF + 2 && OP_TRUE == OP_RREF + 3,
compile_time_checks_must_be_grouped_together);
has_alternatives = ((*matchingpath >= OP_RREF && *matchingpath <= OP_TRUE) || *matchingpath == OP_FAIL) ? FALSE : TRUE;
}
if (SLJIT_UNLIKELY(opcode == OP_COND) && (*cc == OP_KETRMAX || *cc == OP_KETRMIN))
opcode = OP_SCOND;
if (opcode == OP_CBRA || opcode == OP_SCBRA)
{
/* Capturing brackets has a pre-allocated space. */
offset = GET2(ccbegin, 1 + LINK_SIZE);
if (common->optimized_cbracket[offset] == 0)
{
private_data_ptr = OVECTOR_PRIV(offset);
offset <<= 1;
}
else
{
offset <<= 1;
private_data_ptr = OVECTOR(offset);
}
BACKTRACK_AS(bracket_backtrack)->private_data_ptr = private_data_ptr;
matchingpath += IMM2_SIZE;
}
else if (opcode == OP_ASSERT_NA || opcode == OP_ASSERTBACK_NA || opcode == OP_ONCE
|| opcode == OP_ASSERT_SCS || opcode == OP_SCRIPT_RUN || opcode == OP_SBRA || opcode == OP_SCOND)
{
/* Other brackets simply allocate the next entry. */
private_data_ptr = PRIVATE_DATA(ccbegin);
SLJIT_ASSERT(private_data_ptr != 0);
BACKTRACK_AS(bracket_backtrack)->private_data_ptr = private_data_ptr;
if (opcode == OP_ONCE)
BACKTRACK_AS(bracket_backtrack)->u.framesize = get_framesize(common, ccbegin, NULL, FALSE, &needs_control_head);
}
/* Instructions before the first alternative. */
stacksize = 0;
if (ket == OP_KETRMAX || (ket == OP_KETRMIN && bra != OP_BRAMINZERO))
stacksize++;
if (bra == OP_BRAZERO)
stacksize++;
if (stacksize > 0)
allocate_stack(common, stacksize);
stacksize = 0;
if (ket == OP_KETRMAX || (ket == OP_KETRMIN && bra != OP_BRAMINZERO))
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0);
stacksize++;
}
if (bra == OP_BRAZERO)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0);
if (bra == OP_BRAMINZERO)
{
/* This is a backtrack path! (Since the try-path of OP_BRAMINZERO matches to the empty string) */
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
if (ket != OP_KETRMIN)
{
free_stack(common, 1);
braminzero = CMP(SLJIT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0);
}
else if (opcode == OP_ONCE || opcode >= OP_SBRA)
{
jump = CMP(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
/* Nothing stored during the first run. */
skip = JUMP(SLJIT_JUMP);
JUMPHERE(jump);
/* Checking zero-length iteration. */
if (opcode != OP_ONCE || BACKTRACK_AS(bracket_backtrack)->u.framesize < 0)
{
/* When we come from outside, private_data_ptr contains the previous STR_PTR. */
braminzero = CMP(SLJIT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
}
else
{
/* Except when the whole stack frame must be saved. */
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
braminzero = CMP(SLJIT_EQUAL, STR_PTR, 0, SLJIT_MEM1(TMP1), STACK(-BACKTRACK_AS(bracket_backtrack)->u.framesize - 2));
}
JUMPHERE(skip);
}
else
{
jump = CMP(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
JUMPHERE(jump);
}
}
if (repeat_type != 0)
{
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), repeat_ptr, SLJIT_IMM, repeat_count);
if (repeat_type == OP_EXACT)
rmax_label = LABEL();
}
if (ket == OP_KETRMIN)
BACKTRACK_AS(bracket_backtrack)->recursive_matchingpath = LABEL();
if (ket == OP_KETRMAX)
{
rmax_label = LABEL();
if (has_alternatives && opcode >= OP_BRA && opcode < OP_SBRA && repeat_type == 0)
BACKTRACK_AS(bracket_backtrack)->alternative_matchingpath = rmax_label;
}
/* Handling capturing brackets and alternatives. */
if (opcode == OP_ONCE)
{
stacksize = 0;
if (needs_control_head)
{
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr);
stacksize++;
}
if (BACKTRACK_AS(bracket_backtrack)->u.framesize < 0)
{
/* Neither capturing brackets nor recursions are found in the block. */
if (ket == OP_KETRMIN)
{
stacksize += 2;
if (!needs_control_head)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
}
else
{
if (BACKTRACK_AS(bracket_backtrack)->u.framesize == no_frame)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STACK_TOP, 0);
if (ket == OP_KETRMAX || has_alternatives)
stacksize++;
}
if (stacksize > 0)
allocate_stack(common, stacksize);
stacksize = 0;
if (needs_control_head)
{
stacksize++;
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0);
}
if (ket == OP_KETRMIN)
{
if (needs_control_head)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0);
if (BACKTRACK_AS(bracket_backtrack)->u.framesize == no_frame)
OP2(SLJIT_ADD, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STACK_TOP, 0, SLJIT_IMM, needs_control_head ? (2 * sizeof(sljit_sw)) : sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize + 1), TMP2, 0);
}
else if (ket == OP_KETRMAX || has_alternatives)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0);
}
else
{
if (ket != OP_KET || has_alternatives)
stacksize++;
stacksize += BACKTRACK_AS(bracket_backtrack)->u.framesize + 1;
allocate_stack(common, stacksize);
if (needs_control_head)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP2(SLJIT_ADD, TMP2, 0, STACK_TOP, 0, SLJIT_IMM, stacksize * sizeof(sljit_sw));
stacksize = needs_control_head ? 1 : 0;
if (ket != OP_KET || has_alternatives)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP2, 0);
stacksize++;
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP1, 0);
}
else
{
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP2, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP1, 0);
}
init_frame(common, ccbegin, NULL, BACKTRACK_AS(bracket_backtrack)->u.framesize + stacksize, stacksize + 1);
}
}
else if (opcode == OP_CBRA || opcode == OP_SCBRA)
{
/* Saving the previous values. */
if (common->optimized_cbracket[offset >> 1] != 0)
{
SLJIT_ASSERT(private_data_ptr == OVECTOR(offset));
allocate_stack(common, 2);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP2, 0);
}
else
{
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0);
}
}
else if (opcode == OP_ASSERTBACK_NA && PRIVATE_DATA(ccbegin + 1))
{
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
allocate_stack(common, 4);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw), STR_END, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0);
OP1(SLJIT_MOV, STR_END, 0, STR_PTR, 0);
has_vreverse = (*matchingpath == OP_VREVERSE);
if (*matchingpath == OP_REVERSE || has_vreverse)
matchingpath = compile_reverse_matchingpath(common, matchingpath, backtrack);
}
else if (opcode == OP_ASSERT_NA || opcode == OP_ASSERTBACK_NA || opcode == OP_SCRIPT_RUN || opcode == OP_SBRA || opcode == OP_SCOND)
{
/* Saving the previous value. */
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0);
if (*matchingpath == OP_REVERSE)
matchingpath = compile_reverse_matchingpath(common, matchingpath, backtrack);
}
else if (opcode == OP_ASSERT_SCS)
{
/* Nested scs blocks will not update this variable. */
if (common->restore_end_ptr == 0)
common->restore_end_ptr = private_data_ptr + sizeof(sljit_sw);
if (*matchingpath == OP_CREF && (matchingpath[1 + IMM2_SIZE] != OP_CREF && matchingpath[1 + IMM2_SIZE] != OP_DNCREF))
{
/* Optimized case for a single capture reference. */
i = OVECTOR(GET2(matchingpath, 1) << 1);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), i);
add_jump(compiler, &(BACKTRACK_AS(bracket_backtrack)->u.no_capture), CMP(SLJIT_EQUAL, TMP2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1)));
matchingpath += 1 + IMM2_SIZE;
allocate_stack(common, has_alternatives ? 3 : 2);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP1(SLJIT_MOV, SLJIT_TMP_DEST_REG, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw), STR_END, 0);
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), i + sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STR_PTR, 0);
OP1(SLJIT_MOV, STR_PTR, 0, TMP2, 0);
}
else
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1));
jumplist = NULL;
while (TRUE)
{
if (*matchingpath == OP_CREF)
{
sljit_get_local_base(compiler, TMP2, 0, OVECTOR(GET2(matchingpath, 1) << 1));
matchingpath += 1 + IMM2_SIZE;
}
else
{
SLJIT_ASSERT(*matchingpath == OP_DNCREF);
i = GET2(matchingpath, 1 + IMM2_SIZE);
slot = common->name_table + GET2(matchingpath, 1) * common->name_entry_size;
while (i-- > 1)
{
sljit_get_local_base(compiler, TMP2, 0, OVECTOR(GET2(slot, 0) << 1));
add_jump(compiler, &jumplist, CMP(SLJIT_NOT_EQUAL, SLJIT_MEM1(TMP2), 0, TMP1, 0));
slot += common->name_entry_size;
}
sljit_get_local_base(compiler, TMP2, 0, OVECTOR(GET2(slot, 0) << 1));
matchingpath += 1 + 2 * IMM2_SIZE;
}
if (*matchingpath != OP_CREF && *matchingpath != OP_DNCREF)
break;
add_jump(compiler, &jumplist, CMP(SLJIT_NOT_EQUAL, SLJIT_MEM1(TMP2), 0, TMP1, 0));
}
add_jump(compiler, &(BACKTRACK_AS(bracket_backtrack)->u.no_capture),
CMP(SLJIT_EQUAL, SLJIT_MEM1(TMP2), 0, TMP1, 0));
set_jumps(jumplist, LABEL());
allocate_stack(common, has_alternatives ? 3 : 2);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP1(SLJIT_MOV, SLJIT_TMP_DEST_REG, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STR_PTR, 0);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(TMP2), 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw), STR_END, 0);
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
}
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_TMP_DEST_REG, 0);
if (has_alternatives)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), STR_PTR, 0);
}
else if (has_alternatives)
{
/* Pushing the starting string pointer. */
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
}
/* Generating code for the first alternative. */
if (opcode == OP_COND || opcode == OP_SCOND)
{
if (*matchingpath == OP_CREF)
{
SLJIT_ASSERT(has_alternatives);
add_jump(compiler, &(BACKTRACK_AS(bracket_backtrack)->u.no_capture),
CMP(SLJIT_EQUAL, SLJIT_MEM1(SLJIT_SP), OVECTOR(GET2(matchingpath, 1) << 1), SLJIT_MEM1(SLJIT_SP), OVECTOR(1)));
matchingpath += 1 + IMM2_SIZE;
}
else if (*matchingpath == OP_DNCREF)
{
SLJIT_ASSERT(has_alternatives);
i = GET2(matchingpath, 1 + IMM2_SIZE);
slot = common->name_table + GET2(matchingpath, 1) * common->name_entry_size;
OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(1));
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(GET2(slot, 0) << 1), TMP1, 0);
slot += common->name_entry_size;
i--;
while (i-- > 0)
{
OP2(SLJIT_SUB, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(GET2(slot, 0) << 1), TMP1, 0);
OP2(SLJIT_OR | SLJIT_SET_Z, TMP2, 0, TMP2, 0, STR_PTR, 0);
slot += common->name_entry_size;
}
OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0);
add_jump(compiler, &(BACKTRACK_AS(bracket_backtrack)->u.no_capture), JUMP(SLJIT_ZERO));
matchingpath += 1 + 2 * IMM2_SIZE;
}
else if ((*matchingpath >= OP_RREF && *matchingpath <= OP_TRUE) || *matchingpath == OP_FAIL)
{
/* Never has other case. */
BACKTRACK_AS(bracket_backtrack)->u.no_capture = NULL;
SLJIT_ASSERT(!has_alternatives);
if (*matchingpath == OP_TRUE)
{
stacksize = 1;
matchingpath++;
}
else if (*matchingpath == OP_FALSE || *matchingpath == OP_FAIL)
stacksize = 0;
else if (*matchingpath == OP_RREF)
{
stacksize = GET2(matchingpath, 1);
if (common->currententry == NULL)
stacksize = 0;
else if (stacksize == RREF_ANY)
stacksize = 1;
else if (common->currententry->start == 0)
stacksize = stacksize == 0;
else
stacksize = stacksize == (int)GET2(common->start, common->currententry->start + 1 + LINK_SIZE);
if (stacksize != 0)
matchingpath += 1 + IMM2_SIZE;
}
else
{
if (common->currententry == NULL || common->currententry->start == 0)
stacksize = 0;
else
{
stacksize = GET2(matchingpath, 1 + IMM2_SIZE);
slot = common->name_table + GET2(matchingpath, 1) * common->name_entry_size;
i = (int)GET2(common->start, common->currententry->start + 1 + LINK_SIZE);
while (stacksize > 0)
{
if ((int)GET2(slot, 0) == i)
break;
slot += common->name_entry_size;
stacksize--;
}
}
if (stacksize != 0)
matchingpath += 1 + 2 * IMM2_SIZE;
}
/* The stacksize == 0 is a common "else" case. */
if (stacksize == 0)
{
if (*cc == OP_ALT)
{
matchingpath = cc + 1 + LINK_SIZE;
cc += GET(cc, 1);
}
else
matchingpath = cc;
}
}
else
{
SLJIT_ASSERT(has_alternatives && *matchingpath >= OP_ASSERT && *matchingpath <= OP_ASSERTBACK_NOT);
/* Similar code as PUSH_BACKTRACK macro. */
assert = sljit_alloc_memory(compiler, sizeof(assert_backtrack));
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
return NULL;
memset(assert, 0, sizeof(assert_backtrack));
assert->common.cc = matchingpath;
BACKTRACK_AS(bracket_backtrack)->u.assert = assert;
matchingpath = compile_assert_matchingpath(common, matchingpath, assert, TRUE);
}
}
compile_matchingpath(common, matchingpath, cc, backtrack);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
return NULL;
switch (opcode)
{
case OP_ASSERTBACK_NA:
if (has_vreverse)
{
SLJIT_ASSERT(backtrack->top != NULL && PRIVATE_DATA(ccbegin + 1));
add_jump(compiler, &backtrack->top->simple_backtracks, CMP(SLJIT_LESS, STR_PTR, 0, STR_END, 0));
}
if (PRIVATE_DATA(ccbegin + 1))
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw));
break;
case OP_ONCE:
match_once_common(common, ket, BACKTRACK_AS(bracket_backtrack)->u.framesize, private_data_ptr, has_alternatives, needs_control_head);
break;
case OP_SCRIPT_RUN:
match_script_run_common(common, private_data_ptr, backtrack);
break;
}
stacksize = 0;
if (repeat_type == OP_MINUPTO)
{
/* We need to preserve the counter. TMP2 will be used below. */
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), repeat_ptr);
stacksize++;
}
if (ket != OP_KET || bra != OP_BRA)
stacksize++;
if (offset != 0)
{
if (common->capture_last_ptr != 0)
stacksize++;
if (common->optimized_cbracket[offset >> 1] == 0)
stacksize += 2;
}
if (has_alternatives && opcode != OP_ONCE)
stacksize++;
if (stacksize > 0)
allocate_stack(common, stacksize);
stacksize = 0;
if (repeat_type == OP_MINUPTO)
{
/* TMP2 was set above. */
OP2(SLJIT_SUB, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP2, 0, SLJIT_IMM, 1);
stacksize++;
}
if (ket != OP_KET || bra != OP_BRA)
{
if (ket != OP_KET)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0);
else
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0);
stacksize++;
}
if (offset != 0)
stacksize = match_capture_common(common, stacksize, offset, private_data_ptr);
/* Skip and count the other alternatives. */
i = 1;
while (*cc == OP_ALT)
{
cc += GET(cc, 1);
i++;
}
if (has_alternatives)
{
if (opcode != OP_ONCE)
{
if (i <= 3)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0);
else
BACKTRACK_AS(bracket_backtrack)->matching_mov_addr = sljit_emit_mov_addr(compiler, SLJIT_MEM1(STACK_TOP), STACK(stacksize));
}
if (ket != OP_KETRMAX)
BACKTRACK_AS(bracket_backtrack)->alternative_matchingpath = LABEL();
}
/* Must be after the matchingpath label. */
if (offset != 0 && common->optimized_cbracket[offset >> 1] != 0)
{
SLJIT_ASSERT(private_data_ptr == OVECTOR(offset + 0));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), STR_PTR, 0);
}
else switch (opcode)
{
case OP_ASSERT_NA:
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
break;
case OP_ASSERT_SCS:
OP1(SLJIT_MOV, TMP1, 0, STR_END, 0);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw), TMP1, 0);
/* Nested scs blocks will not update this variable. */
if (common->restore_end_ptr == private_data_ptr + SSIZE_OF(sw))
common->restore_end_ptr = 0;
break;
}
if (ket == OP_KETRMAX)
{
if (repeat_type != 0)
{
if (has_alternatives)
BACKTRACK_AS(bracket_backtrack)->alternative_matchingpath = LABEL();
OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_MEM1(SLJIT_SP), repeat_ptr, SLJIT_MEM1(SLJIT_SP), repeat_ptr, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, rmax_label);
/* Drop STR_PTR for greedy plus quantifier. */
if (opcode != OP_ONCE)
free_stack(common, 1);
}
else if (opcode < OP_BRA || opcode >= OP_SBRA)
{
if (has_alternatives)
BACKTRACK_AS(bracket_backtrack)->alternative_matchingpath = LABEL();
/* Checking zero-length iteration. */
if (opcode != OP_ONCE)
{
/* This case includes opcodes such as OP_SCRIPT_RUN. */
CMPTO(SLJIT_NOT_EQUAL, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STR_PTR, 0, rmax_label);
/* Drop STR_PTR for greedy plus quantifier. */
if (bra != OP_BRAZERO)
free_stack(common, 1);
}
else
/* TMP2 must contain the starting STR_PTR. */
CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, STR_PTR, 0, rmax_label);
}
else
JUMPTO(SLJIT_JUMP, rmax_label);
BACKTRACK_AS(bracket_backtrack)->recursive_matchingpath = LABEL();
}
if (repeat_type == OP_EXACT)
{
count_match(common);
OP2(SLJIT_SUB | SLJIT_SET_Z, SLJIT_MEM1(SLJIT_SP), repeat_ptr, SLJIT_MEM1(SLJIT_SP), repeat_ptr, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, rmax_label);
}
else if (repeat_type == OP_UPTO)
{
/* We need to preserve the counter. */
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), repeat_ptr);
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0);
}
if (bra == OP_BRAZERO)
BACKTRACK_AS(bracket_backtrack)->zero_matchingpath = LABEL();
if (bra == OP_BRAMINZERO)
{
/* This is a backtrack path! (From the viewpoint of OP_BRAMINZERO) */
JUMPTO(SLJIT_JUMP, ((braminzero_backtrack *)parent)->matchingpath);
if (braminzero != NULL)
{
JUMPHERE(braminzero);
/* We need to release the end pointer to perform the
backtrack for the zero-length iteration. When
framesize is < 0, OP_ONCE will do the release itself. */
if (opcode == OP_ONCE)
{
int framesize = BACKTRACK_AS(bracket_backtrack)->u.framesize;
SLJIT_ASSERT(framesize != 0);
if (framesize > 0)
{
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL));
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize - 1) * sizeof(sljit_sw));
}
}
else if (ket == OP_KETRMIN)
free_stack(common, 1);
}
/* Continue to the normal backtrack. */
}
if ((ket != OP_KET && bra != OP_BRAMINZERO) || bra == OP_BRAZERO || (has_alternatives && repeat_type != OP_EXACT))
count_match(common);
cc += 1 + LINK_SIZE;
if (opcode == OP_ONCE)
{
int data;
int framesize = BACKTRACK_AS(bracket_backtrack)->u.framesize;
SLJIT_ASSERT(SHRT_MIN <= framesize && framesize < SHRT_MAX/2);
/* We temporarily encode the needs_control_head in the lowest bit.
The real value should be short enough for this operation to work
without triggering Undefined Behaviour. */
data = (int)((short)((unsigned short)framesize << 1) | (needs_control_head ? 1 : 0));
BACKTRACK_AS(bracket_backtrack)->u.framesize = data;
}
return cc + repeat_length;
}
static PCRE2_SPTR compile_bracketpos_matchingpath(compiler_common *common, PCRE2_SPTR cc, backtrack_common *parent)
{
DEFINE_COMPILER;
backtrack_common *backtrack;
PCRE2_UCHAR opcode;
int private_data_ptr;
int cbraprivptr = 0;
BOOL needs_control_head;
int framesize;
int stacksize;
int offset = 0;
BOOL zero = FALSE;
PCRE2_SPTR ccbegin = NULL;
int stack; /* Also contains the offset of control head. */
struct sljit_label *loop = NULL;
struct jump_list *emptymatch = NULL;
PUSH_BACKTRACK(sizeof(bracketpos_backtrack), cc, NULL);
if (*cc == OP_BRAPOSZERO)
{
zero = TRUE;
cc++;
}
opcode = *cc;
private_data_ptr = PRIVATE_DATA(cc);
SLJIT_ASSERT(private_data_ptr != 0);
BACKTRACK_AS(bracketpos_backtrack)->private_data_ptr = private_data_ptr;
switch(opcode)
{
case OP_BRAPOS:
case OP_SBRAPOS:
ccbegin = cc + 1 + LINK_SIZE;
break;
case OP_CBRAPOS:
case OP_SCBRAPOS:
offset = GET2(cc, 1 + LINK_SIZE);
/* This case cannot be optimized in the same way as
normal capturing brackets. */
SLJIT_ASSERT(common->optimized_cbracket[offset] == 0);
cbraprivptr = OVECTOR_PRIV(offset);
offset <<= 1;
ccbegin = cc + 1 + LINK_SIZE + IMM2_SIZE;
break;
default:
SLJIT_UNREACHABLE();
break;
}
framesize = get_framesize(common, cc, NULL, FALSE, &needs_control_head);
BACKTRACK_AS(bracketpos_backtrack)->framesize = framesize;
if (framesize < 0)
{
if (offset != 0)
{
stacksize = 2;
if (common->capture_last_ptr != 0)
stacksize++;
}
else
stacksize = 1;
if (needs_control_head)
stacksize++;
if (!zero)
stacksize++;
BACKTRACK_AS(bracketpos_backtrack)->stacksize = stacksize;
allocate_stack(common, stacksize);
if (framesize == no_frame)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STACK_TOP, 0);
stack = 0;
if (offset != 0)
{
stack = 2;
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset));
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0);
if (common->capture_last_ptr != 0)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), TMP2, 0);
if (needs_control_head)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr);
if (common->capture_last_ptr != 0)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), TMP1, 0);
stack = 3;
}
}
else
{
if (needs_control_head)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
stack = 1;
}
if (needs_control_head)
stack++;
if (!zero)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), SLJIT_IMM, 1);
if (needs_control_head)
{
stack--;
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), TMP2, 0);
}
}
else
{
stacksize = framesize + 1;
if (!zero)
stacksize++;
if (needs_control_head)
stacksize++;
if (offset == 0)
stacksize++;
BACKTRACK_AS(bracketpos_backtrack)->stacksize = stacksize;
allocate_stack(common, stacksize);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
if (needs_control_head)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr);
OP2(SLJIT_ADD, SLJIT_MEM1(SLJIT_SP), private_data_ptr, STACK_TOP, 0, SLJIT_IMM, stacksize * sizeof(sljit_sw));
stack = 0;
if (!zero)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 1);
stack = 1;
}
if (needs_control_head)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), TMP2, 0);
stack++;
}
if (offset == 0)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), STR_PTR, 0);
stack++;
}
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stack), TMP1, 0);
init_frame(common, cc, NULL, stacksize - 1, stacksize - framesize);
stack -= 1 + (offset == 0);
}
if (offset != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), cbraprivptr, STR_PTR, 0);
loop = LABEL();
while (*cc != OP_KETRPOS)
{
backtrack->top = NULL;
backtrack->own_backtracks = NULL;
cc += GET(cc, 1);
compile_matchingpath(common, ccbegin, cc, backtrack);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
return NULL;
if (framesize < 0)
{
if (framesize == no_frame)
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
if (offset != 0)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), cbraprivptr);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), cbraprivptr, STR_PTR, 0);
if (common->capture_last_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr, SLJIT_IMM, offset >> 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset), TMP1, 0);
}
else
{
if (opcode == OP_SBRAPOS)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
}
/* Even if the match is empty, we need to reset the control head. */
if (needs_control_head)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), STACK(stack));
if (opcode == OP_SBRAPOS || opcode == OP_SCBRAPOS)
add_jump(compiler, &emptymatch, CMP(SLJIT_EQUAL, TMP1, 0, STR_PTR, 0));
if (!zero)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize - 1), SLJIT_IMM, 0);
}
else
{
if (offset != 0)
{
OP2(SLJIT_SUB, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr, SLJIT_IMM, stacksize * sizeof(sljit_sw));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), cbraprivptr);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), cbraprivptr, STR_PTR, 0);
if (common->capture_last_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr, SLJIT_IMM, offset >> 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset), TMP1, 0);
}
else
{
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP2(SLJIT_SUB, STACK_TOP, 0, TMP2, 0, SLJIT_IMM, stacksize * sizeof(sljit_sw));
if (opcode == OP_SBRAPOS)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), STACK(-framesize - 2));
OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), STACK(-framesize - 2), STR_PTR, 0);
}
/* Even if the match is empty, we need to reset the control head. */
if (needs_control_head)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_MEM1(STACK_TOP), STACK(stack));
if (opcode == OP_SBRAPOS || opcode == OP_SCBRAPOS)
add_jump(compiler, &emptymatch, CMP(SLJIT_EQUAL, TMP1, 0, STR_PTR, 0));
if (!zero)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
}
JUMPTO(SLJIT_JUMP, loop);
flush_stubs(common);
compile_backtrackingpath(common, backtrack->top);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
return NULL;
set_jumps(backtrack->own_backtracks, LABEL());
if (framesize < 0)
{
if (offset != 0)
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), cbraprivptr);
else
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
}
else
{
if (offset != 0)
{
/* Last alternative. */
if (*cc == OP_KETRPOS)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), cbraprivptr);
}
else
{
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(TMP2), STACK(-framesize - 2));
}
}
if (*cc == OP_KETRPOS)
break;
ccbegin = cc + 1 + LINK_SIZE;
}
/* We don't have to restore the control head in case of a failed match. */
backtrack->own_backtracks = NULL;
if (!zero)
{
if (framesize < 0)
add_jump(compiler, &backtrack->own_backtracks, CMP(SLJIT_NOT_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(stacksize - 1), SLJIT_IMM, 0));
else /* TMP2 is set to [private_data_ptr] above. */
add_jump(compiler, &backtrack->own_backtracks, CMP(SLJIT_NOT_EQUAL, SLJIT_MEM1(TMP2), STACK(-stacksize), SLJIT_IMM, 0));
}
/* None of them matched. */
set_jumps(emptymatch, LABEL());
count_match(common);
return cc + 1 + LINK_SIZE;
}
static SLJIT_INLINE PCRE2_SPTR get_iterator_parameters(compiler_common *common, PCRE2_SPTR cc, PCRE2_UCHAR *opcode, PCRE2_UCHAR *type, sljit_u32 *max, sljit_u32 *exact, PCRE2_SPTR *end)
{
int class_len;
*opcode = *cc;
*exact = 0;
if (*opcode >= OP_STAR && *opcode <= OP_POSUPTO)
{
cc++;
*type = OP_CHAR;
}
else if (*opcode >= OP_STARI && *opcode <= OP_POSUPTOI)
{
cc++;
*type = OP_CHARI;
*opcode -= OP_STARI - OP_STAR;
}
else if (*opcode >= OP_NOTSTAR && *opcode <= OP_NOTPOSUPTO)
{
cc++;
*type = OP_NOT;
*opcode -= OP_NOTSTAR - OP_STAR;
}
else if (*opcode >= OP_NOTSTARI && *opcode <= OP_NOTPOSUPTOI)
{
cc++;
*type = OP_NOTI;
*opcode -= OP_NOTSTARI - OP_STAR;
}
else if (*opcode >= OP_TYPESTAR && *opcode <= OP_TYPEPOSUPTO)
{
cc++;
*opcode -= OP_TYPESTAR - OP_STAR;
*type = OP_END;
}
else
{
SLJIT_ASSERT(*opcode == OP_CLASS || *opcode == OP_NCLASS || *opcode == OP_XCLASS || *opcode == OP_ECLASS);
*type = *opcode;
class_len = (*type < OP_XCLASS) ? (int)(1 + (32 / sizeof(PCRE2_UCHAR))) : GET(cc, 1);
*opcode = cc[class_len];
cc++;
if (*opcode >= OP_CRSTAR && *opcode <= OP_CRMINQUERY)
{
*opcode -= OP_CRSTAR - OP_STAR;
*end = cc + class_len;
if (*opcode == OP_PLUS || *opcode == OP_MINPLUS)
{
*exact = 1;
*opcode -= OP_PLUS - OP_STAR;
}
return cc;
}
if (*opcode >= OP_CRPOSSTAR && *opcode <= OP_CRPOSQUERY)
{
*opcode -= OP_CRPOSSTAR - OP_POSSTAR;
*end = cc + class_len;
if (*opcode == OP_POSPLUS)
{
*exact = 1;
*opcode = OP_POSSTAR;
}
return cc;
}
SLJIT_ASSERT(*opcode == OP_CRRANGE || *opcode == OP_CRMINRANGE || *opcode == OP_CRPOSRANGE);
*max = GET2(cc, (class_len + IMM2_SIZE));
*exact = GET2(cc, class_len);
*end = cc + class_len + 2 * IMM2_SIZE;
if (*max == 0)
{
SLJIT_ASSERT(*exact > 1);
if (*opcode == OP_CRRANGE)
*opcode = OP_UPTO;
else if (*opcode == OP_CRPOSRANGE)
*opcode = OP_POSUPTO;
else
*opcode = OP_MINSTAR;
return cc;
}
*max -= *exact;
if (*max == 0)
*opcode = OP_EXACT;
else
{
SLJIT_ASSERT(*exact > 0 || *max > 1);
if (*opcode == OP_CRRANGE)
*opcode = OP_UPTO;
else if (*opcode == OP_CRPOSRANGE)
*opcode = OP_POSUPTO;
else if (*max == 1)
*opcode = OP_MINQUERY;
else
*opcode = OP_MINUPTO;
}
return cc;
}
switch(*opcode)
{
case OP_EXACT:
*exact = GET2(cc, 0);
cc += IMM2_SIZE;
break;
case OP_PLUS:
case OP_MINPLUS:
*exact = 1;
*opcode -= OP_PLUS - OP_STAR;
break;
case OP_POSPLUS:
*exact = 1;
*opcode = OP_POSSTAR;
break;
case OP_UPTO:
case OP_MINUPTO:
case OP_POSUPTO:
*max = GET2(cc, 0);
cc += IMM2_SIZE;
break;
}
if (*type == OP_END)
{
*type = *cc;
*end = next_opcode(common, cc);
cc++;
return cc;
}
*end = cc + 1;
#ifdef SUPPORT_UNICODE
if (common->utf && HAS_EXTRALEN(*cc)) *end += GET_EXTRALEN(*cc);
#endif
return cc;
}
static PCRE2_SPTR compile_iterator_matchingpath(compiler_common *common, PCRE2_SPTR cc, backtrack_common *parent, jump_list **prev_backtracks)
{
DEFINE_COMPILER;
backtrack_common *backtrack = NULL;
PCRE2_SPTR begin = cc;
PCRE2_UCHAR opcode;
PCRE2_UCHAR type;
sljit_u32 max = 0, exact;
sljit_s32 early_fail_ptr = PRIVATE_DATA(cc + 1);
sljit_s32 early_fail_type;
BOOL charpos_enabled, use_tmp;
PCRE2_UCHAR charpos_char;
unsigned int charpos_othercasebit;
PCRE2_SPTR end;
jump_list *no_match = NULL;
jump_list *no_char1_match = NULL;
struct sljit_jump *jump = NULL;
struct sljit_label *label;
int private_data_ptr = PRIVATE_DATA(cc);
int base = (private_data_ptr == 0) ? SLJIT_MEM1(STACK_TOP) : SLJIT_MEM1(SLJIT_SP);
int offset0 = (private_data_ptr == 0) ? STACK(0) : private_data_ptr;
int offset1 = (private_data_ptr == 0) ? STACK(1) : private_data_ptr + SSIZE_OF(sw);
int tmp_base, tmp_offset;
early_fail_type = (early_fail_ptr & 0x7);
early_fail_ptr >>= 3;
/* During recursion, these optimizations are disabled. */
if (common->early_fail_start_ptr == 0 && common->fast_forward_bc_ptr == NULL)
{
early_fail_ptr = 0;
early_fail_type = type_skip;
}
SLJIT_ASSERT(common->fast_forward_bc_ptr != NULL || early_fail_ptr == 0
|| (early_fail_ptr >= common->early_fail_start_ptr && early_fail_ptr <= common->early_fail_end_ptr));
if (early_fail_type == type_fail)
add_jump(compiler, prev_backtracks, CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), early_fail_ptr));
cc = get_iterator_parameters(common, cc, &opcode, &type, &max, &exact, &end);
if (type != OP_EXTUNI)
{
tmp_base = TMP3;
tmp_offset = 0;
}
else
{
tmp_base = SLJIT_MEM1(SLJIT_SP);
tmp_offset = LOCAL2;
}
if (opcode == OP_EXACT)
{
SLJIT_ASSERT(early_fail_ptr == 0 && exact >= 2);
if (common->mode == PCRE2_JIT_COMPLETE
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
&& !common->utf
#endif
&& type != OP_ANYNL && type != OP_EXTUNI)
{
OP2(SLJIT_SUB, TMP1, 0, STR_END, 0, STR_PTR, 0);
add_jump(compiler, prev_backtracks, CMP(SLJIT_LESS, TMP1, 0, SLJIT_IMM, IN_UCHARS(exact)));
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 32
if (type == OP_ALLANY && !common->invalid_utf)
#else
if (type == OP_ALLANY)
#endif
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(exact));
else
{
OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, exact);
label = LABEL();
compile_char1_matchingpath(common, type, cc, prev_backtracks, FALSE);
OP2(SLJIT_SUB | SLJIT_SET_Z, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
}
}
else
{
SLJIT_ASSERT(tmp_base == TMP3 || common->locals_size >= 3 * SSIZE_OF(sw));
OP1(SLJIT_MOV, tmp_base, tmp_offset, SLJIT_IMM, exact);
label = LABEL();
compile_char1_matchingpath(common, type, cc, prev_backtracks, TRUE);
OP2(SLJIT_SUB | SLJIT_SET_Z, tmp_base, tmp_offset, tmp_base, tmp_offset, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
}
}
if (early_fail_type == type_fail_range)
{
/* Range end first, followed by range start. */
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), early_fail_ptr);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), early_fail_ptr + SSIZE_OF(sw));
OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, TMP2, 0);
OP2(SLJIT_SUB, TMP2, 0, STR_PTR, 0, TMP2, 0);
add_jump(compiler, prev_backtracks, CMP(SLJIT_LESS_EQUAL, TMP2, 0, TMP1, 0));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), early_fail_ptr, STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), early_fail_ptr + SSIZE_OF(sw), STR_PTR, 0);
}
if (opcode < OP_EXACT)
PUSH_BACKTRACK(sizeof(char_iterator_backtrack), begin, NULL);
switch(opcode)
{
case OP_STAR:
case OP_UPTO:
SLJIT_ASSERT(backtrack != NULL && (early_fail_ptr == 0 || opcode == OP_STAR));
max += exact;
if (type == OP_EXTUNI)
{
SLJIT_ASSERT(private_data_ptr == 0);
SLJIT_ASSERT(early_fail_ptr == 0);
if (exact == 1)
{
SLJIT_ASSERT(opcode == OP_STAR);
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
}
else
{
/* If OP_EXTUNI is present, it has a separate EXACT opcode. */
SLJIT_ASSERT(exact == 0);
allocate_stack(common, 2);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);
}
if (opcode == OP_UPTO)
{
SLJIT_ASSERT(common->locals_size >= 3 * SSIZE_OF(sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL2, SLJIT_IMM, max);
}
label = LABEL();
compile_char1_matchingpath(common, type, cc, &backtrack->own_backtracks, TRUE);
if (opcode == OP_UPTO)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), LOCAL2);
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);
jump = JUMP(SLJIT_ZERO);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL2, TMP1, 0);
}
/* We cannot use TMP3 because of allocate_stack. */
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
JUMPTO(SLJIT_JUMP, label);
if (jump != NULL)
JUMPHERE(jump);
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
break;
}
#ifdef SUPPORT_UNICODE
else if (type == OP_ALLANY && !common->invalid_utf)
#else
else if (type == OP_ALLANY)
#endif
{
if (opcode == OP_STAR)
{
if (exact == 1)
detect_partial_match(common, prev_backtracks);
if (private_data_ptr == 0)
allocate_stack(common, 2);
OP1(SLJIT_MOV, base, offset0, STR_END, 0);
OP1(SLJIT_MOV, base, offset1, STR_PTR, 0);
OP1(SLJIT_MOV, STR_PTR, 0, STR_END, 0);
process_partial_match(common);
if (early_fail_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), early_fail_ptr, STR_END, 0);
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
break;
}
#ifdef SUPPORT_UNICODE
else if (!common->utf)
#else
else
#endif
{
/* If OP_ALLANY is present, it has a separate EXACT opcode. */
SLJIT_ASSERT(exact == 0);
if (private_data_ptr == 0)
allocate_stack(common, 2);
OP1(SLJIT_MOV, base, offset1, STR_PTR, 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(max));
if (common->mode == PCRE2_JIT_COMPLETE)
{
OP2U(SLJIT_SUB | SLJIT_SET_GREATER, STR_PTR, 0, STR_END, 0);
SELECT(SLJIT_GREATER, STR_PTR, STR_END, 0, STR_PTR);
}
else
{
jump = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, STR_END, 0);
process_partial_match(common);
JUMPHERE(jump);
}
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
if (early_fail_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), early_fail_ptr, STR_PTR, 0);
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
break;
}
}
charpos_enabled = FALSE;
charpos_char = 0;
charpos_othercasebit = 0;
SLJIT_ASSERT(tmp_base == TMP3);
if ((type != OP_CHAR && type != OP_CHARI) && (*end == OP_CHAR || *end == OP_CHARI))
{
#ifdef SUPPORT_UNICODE
charpos_enabled = !common->utf || !HAS_EXTRALEN(end[1]);
#else
charpos_enabled = TRUE;
#endif
if (charpos_enabled && *end == OP_CHARI && char_has_othercase(common, end + 1))
{
charpos_othercasebit = char_get_othercase_bit(common, end + 1);
if (charpos_othercasebit == 0)
charpos_enabled = FALSE;
}
if (charpos_enabled)
{
charpos_char = end[1];
/* Consume the OP_CHAR opcode. */
end += 2;
#if PCRE2_CODE_UNIT_WIDTH == 8
SLJIT_ASSERT((charpos_othercasebit >> 8) == 0);
#elif PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
SLJIT_ASSERT((charpos_othercasebit >> 9) == 0);
if ((charpos_othercasebit & 0x100) != 0)
charpos_othercasebit = (charpos_othercasebit & 0xff) << 8;
#endif
if (charpos_othercasebit != 0)
charpos_char |= charpos_othercasebit;
BACKTRACK_AS(char_iterator_backtrack)->charpos.charpos_enabled = TRUE;
BACKTRACK_AS(char_iterator_backtrack)->charpos.chr = charpos_char;
BACKTRACK_AS(char_iterator_backtrack)->charpos.othercasebit = charpos_othercasebit;
if (private_data_ptr == 0)
allocate_stack(common, 2);
use_tmp = (opcode == OP_STAR);
if (use_tmp)
{
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 0);
OP1(SLJIT_MOV, base, offset0, TMP3, 0);
}
else
{
OP1(SLJIT_MOV, base, offset1, COUNT_MATCH, 0);
OP1(SLJIT_MOV, COUNT_MATCH, 0, SLJIT_IMM, 0);
OP1(SLJIT_MOV, base, offset0, COUNT_MATCH, 0);
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, exact == max ? 0 : (max + 1));
}
/* Search the first instance of charpos_char. */
if (exact > 0)
detect_partial_match(common, &no_match);
else
jump = JUMP(SLJIT_JUMP);
label = LABEL();
if (opcode == OP_UPTO)
{
if (exact == max)
OP2(SLJIT_ADD, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
else
{
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
add_jump(compiler, &no_match, JUMP(SLJIT_ZERO));
}
}
compile_char1_matchingpath(common, type, cc, &no_match, FALSE);
if (early_fail_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), early_fail_ptr, STR_PTR, 0);
if (exact == 0)
JUMPHERE(jump);
detect_partial_match(common, &no_match);
if (opcode == OP_UPTO && exact > 0)
{
if (exact == max)
CMPTO(SLJIT_LESS, TMP3, 0, SLJIT_IMM, exact, label);
else
CMPTO(SLJIT_GREATER, TMP3, 0, SLJIT_IMM, (max + 1) - exact, label);
}
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
if (charpos_othercasebit != 0)
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, charpos_othercasebit);
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, charpos_char, label);
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
if (use_tmp)
{
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP3, 0, SLJIT_IMM, 0);
SELECT(SLJIT_EQUAL, TMP3, STR_PTR, 0, TMP3);
}
else
{
OP2U(SLJIT_SUB | SLJIT_SET_Z, COUNT_MATCH, 0, SLJIT_IMM, 0);
SELECT(SLJIT_EQUAL, COUNT_MATCH, STR_PTR, 0, COUNT_MATCH);
}
JUMPTO(SLJIT_JUMP, label);
set_jumps(no_match, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
if (use_tmp)
OP1(SLJIT_MOV, base, offset1, TMP3, 0);
else
{
OP1(SLJIT_MOV, TMP1, 0, base, offset1);
OP1(SLJIT_MOV, base, offset1, COUNT_MATCH, 0);
OP1(SLJIT_MOV, COUNT_MATCH, 0, TMP1, 0);
}
add_jump(compiler, &backtrack->own_backtracks, CMP(SLJIT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0));
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
break;
}
}
if (private_data_ptr == 0)
allocate_stack(common, 2);
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
use_tmp = (opcode == OP_STAR);
if (common->utf)
{
if (!use_tmp)
OP1(SLJIT_MOV, base, offset0, COUNT_MATCH, 0);
OP1(SLJIT_MOV, use_tmp ? TMP3 : COUNT_MATCH, 0, STR_PTR, 0);
}
#endif
if (opcode == OP_UPTO)
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, exact == max ? -(sljit_sw)exact : (sljit_sw)max);
if (opcode == OP_UPTO && exact > 0)
{
label = LABEL();
detect_partial_match(common, &no_match);
compile_char1_matchingpath(common, type, cc, &no_char1_match, FALSE);
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
OP1(SLJIT_MOV, use_tmp ? TMP3 : COUNT_MATCH, 0, STR_PTR, 0);
#endif
if (exact == max)
{
OP2(SLJIT_ADD | SLJIT_SET_Z, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
}
else
{
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
add_jump(compiler, &no_match, JUMP(SLJIT_ZERO));
CMPTO(SLJIT_NOT_EQUAL, TMP3, 0, SLJIT_IMM, max - exact, label);
}
OP1(SLJIT_MOV, base, offset1, STR_PTR, 0);
JUMPTO(SLJIT_JUMP, label);
}
else
{
OP1(SLJIT_MOV, base, offset1, STR_PTR, 0);
detect_partial_match(common, &no_match);
label = LABEL();
compile_char1_matchingpath(common, type, cc, &no_char1_match, FALSE);
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
OP1(SLJIT_MOV, use_tmp ? TMP3 : COUNT_MATCH, 0, STR_PTR, 0);
#endif
if (opcode == OP_UPTO)
{
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
add_jump(compiler, &no_match, JUMP(SLJIT_ZERO));
}
detect_partial_match_to(common, label);
}
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
{
set_jumps(no_char1_match, LABEL());
set_jumps(no_match, LABEL());
if (use_tmp)
{
OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0);
OP1(SLJIT_MOV, base, offset0, TMP3, 0);
}
else
{
OP1(SLJIT_MOV, STR_PTR, 0, COUNT_MATCH, 0);
OP1(SLJIT_MOV, COUNT_MATCH, 0, base, offset0);
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
}
}
else
#endif
{
if (opcode != OP_UPTO || exact == 0)
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
set_jumps(no_char1_match, LABEL());
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
set_jumps(no_match, LABEL());
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
}
if (opcode == OP_UPTO)
{
if (exact > 0)
{
if (max == exact)
jump = CMP(SLJIT_GREATER_EQUAL, TMP3, 0, SLJIT_IMM, -(sljit_sw)exact);
else
jump = CMP(SLJIT_GREATER, TMP3, 0, SLJIT_IMM, max - exact);
add_jump(compiler, &backtrack->own_backtracks, jump);
}
}
else if (exact == 1)
add_jump(compiler, &backtrack->own_backtracks, CMP(SLJIT_EQUAL, base, offset1, STR_PTR, 0));
if (early_fail_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), early_fail_ptr, STR_PTR, 0);
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
break;
case OP_QUERY:
SLJIT_ASSERT(backtrack != NULL && early_fail_ptr == 0);
if (private_data_ptr == 0)
allocate_stack(common, 1);
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
compile_char1_matchingpath(common, type, cc, &backtrack->own_backtracks, TRUE);
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
break;
case OP_MINSTAR:
case OP_MINQUERY:
SLJIT_ASSERT(backtrack != NULL && (opcode == OP_MINSTAR || early_fail_ptr == 0));
if (private_data_ptr == 0)
allocate_stack(common, 1);
if (exact >= 1)
{
if (exact >= 2)
{
/* Extuni has a separate exact opcode. */
SLJIT_ASSERT(tmp_base == TMP3 && early_fail_ptr == 0);
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, exact);
}
if (opcode == OP_MINQUERY)
OP1(SLJIT_MOV, base, offset0, SLJIT_IMM, -1);
label = LABEL();
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = label;
compile_char1_matchingpath(common, type, cc, &backtrack->own_backtracks, TRUE);
if (exact >= 2)
{
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
}
if (opcode == OP_MINQUERY)
OP2(SLJIT_AND, base, offset0, base, offset0, STR_PTR, 0);
else
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
}
else
{
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
}
if (early_fail_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), early_fail_ptr, STR_PTR, 0);
break;
case OP_MINUPTO:
SLJIT_ASSERT(backtrack != NULL && early_fail_ptr == 0);
if (private_data_ptr == 0)
allocate_stack(common, 2);
OP1(SLJIT_MOV, base, offset1, SLJIT_IMM, max + 1);
if (exact == 0)
{
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = LABEL();
break;
}
if (exact >= 2)
{
/* Extuni has a separate exact opcode. */
SLJIT_ASSERT(tmp_base == TMP3);
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, exact);
}
label = LABEL();
BACKTRACK_AS(char_iterator_backtrack)->matchingpath = label;
compile_char1_matchingpath(common, type, cc, &backtrack->own_backtracks, TRUE);
if (exact >= 2)
{
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
}
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
break;
case OP_EXACT:
SLJIT_ASSERT(backtrack == NULL);
break;
case OP_POSSTAR:
SLJIT_ASSERT(backtrack == NULL);
#if defined SUPPORT_UNICODE
if (type == OP_ALLANY && !common->invalid_utf)
#else
if (type == OP_ALLANY)
#endif
{
if (exact == 1)
detect_partial_match(common, prev_backtracks);
OP1(SLJIT_MOV, STR_PTR, 0, STR_END, 0);
process_partial_match(common);
if (early_fail_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), early_fail_ptr, STR_END, 0);
break;
}
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (common->utf)
{
SLJIT_ASSERT(tmp_base == TMP3 || common->locals_size >= 3 * SSIZE_OF(sw));
if (tmp_base != TMP3)
{
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL2, COUNT_MATCH, 0);
tmp_base = COUNT_MATCH;
}
OP1(SLJIT_MOV, tmp_base, 0, exact == 1 ? SLJIT_IMM : STR_PTR, 0);
detect_partial_match(common, &no_match);
label = LABEL();
compile_char1_matchingpath(common, type, cc, &no_match, FALSE);
OP1(SLJIT_MOV, tmp_base, 0, STR_PTR, 0);
detect_partial_match_to(common, label);
set_jumps(no_match, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, tmp_base, 0);
if (tmp_base != TMP3)
OP1(SLJIT_MOV, COUNT_MATCH, 0, SLJIT_MEM1(SLJIT_SP), LOCAL2);
if (exact == 1)
add_jump(compiler, prev_backtracks, CMP(SLJIT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0));
if (early_fail_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), early_fail_ptr, STR_PTR, 0);
break;
}
#endif
if (exact == 1)
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0);
detect_partial_match(common, &no_match);
label = LABEL();
/* Extuni never fails, so no_char1_match is not used in that case.
Anynl optionally reads an extra character on success. */
compile_char1_matchingpath(common, type, cc, &no_char1_match, FALSE);
detect_partial_match_to(common, label);
if (type != OP_EXTUNI)
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
set_jumps(no_char1_match, LABEL());
if (type != OP_EXTUNI)
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
set_jumps(no_match, LABEL());
if (exact == 1)
add_jump(compiler, prev_backtracks, CMP(SLJIT_EQUAL, tmp_base, tmp_offset, STR_PTR, 0));
if (early_fail_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), early_fail_ptr, STR_PTR, 0);
break;
case OP_POSUPTO:
SLJIT_ASSERT(backtrack == NULL && early_fail_ptr == 0);
max += exact;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH != 32
if (type == OP_EXTUNI || common->utf)
#else
if (type == OP_EXTUNI)
#endif
{
SLJIT_ASSERT(common->locals_size >= 3 * SSIZE_OF(sw));
/* Count match is not modified by compile_char1_matchingpath. */
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL2, COUNT_MATCH, 0);
OP1(SLJIT_MOV, COUNT_MATCH, 0, SLJIT_IMM, exact == max ? 0 : max);
label = LABEL();
/* Extuni only modifies TMP3 on successful match. */
OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0);
compile_char1_matchingpath(common, type, cc, &no_match, TRUE);
if (exact == max)
{
OP2(SLJIT_ADD, COUNT_MATCH, 0, COUNT_MATCH, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_JUMP, label);
}
else
{
OP2(SLJIT_SUB | SLJIT_SET_Z, COUNT_MATCH, 0, COUNT_MATCH, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_NOT_ZERO, label);
OP1(SLJIT_MOV, TMP3, 0, STR_PTR, 0);
}
set_jumps(no_match, LABEL());
if (exact > 0)
{
if (exact == max)
OP2U(SLJIT_SUB | SLJIT_SET_LESS, COUNT_MATCH, 0, SLJIT_IMM, exact);
else
OP2U(SLJIT_SUB | SLJIT_SET_GREATER, COUNT_MATCH, 0, SLJIT_IMM, max - exact);
}
OP1(SLJIT_MOV, COUNT_MATCH, 0, SLJIT_MEM1(SLJIT_SP), LOCAL2);
if (exact > 0)
add_jump(compiler, prev_backtracks, JUMP(exact == max ? SLJIT_LESS : SLJIT_GREATER));
OP1(SLJIT_MOV, STR_PTR, 0, TMP3, 0);
break;
}
SLJIT_ASSERT(tmp_base == TMP3);
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, exact == max ? 0 : max);
detect_partial_match(common, &no_match);
label = LABEL();
compile_char1_matchingpath(common, type, cc, &no_char1_match, FALSE);
if (exact == max)
OP2(SLJIT_ADD, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
else
{
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP3, 0, TMP3, 0, SLJIT_IMM, 1);
add_jump(compiler, &no_match, JUMP(SLJIT_ZERO));
}
detect_partial_match_to(common, label);
OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
set_jumps(no_char1_match, LABEL());
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
set_jumps(no_match, LABEL());
if (exact > 0)
{
if (exact == max)
jump = CMP(SLJIT_LESS, TMP3, 0, SLJIT_IMM, exact);
else
jump = CMP(SLJIT_GREATER, TMP3, 0, SLJIT_IMM, max - exact);
add_jump(compiler, prev_backtracks, jump);
}
break;
case OP_POSQUERY:
SLJIT_ASSERT(backtrack == NULL && early_fail_ptr == 0);
SLJIT_ASSERT(tmp_base == TMP3 || common->locals_size >= 3 * SSIZE_OF(sw));
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0);
compile_char1_matchingpath(common, type, cc, &no_match, TRUE);
OP1(SLJIT_MOV, tmp_base, tmp_offset, STR_PTR, 0);
set_jumps(no_match, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, tmp_base, tmp_offset);
break;
default:
SLJIT_UNREACHABLE();
break;
}
count_match(common);
return end;
}
static SLJIT_INLINE PCRE2_SPTR compile_fail_accept_matchingpath(compiler_common *common, PCRE2_SPTR cc, backtrack_common *parent)
{
DEFINE_COMPILER;
backtrack_common *backtrack;
PUSH_BACKTRACK(sizeof(backtrack_common), cc, NULL);
if (*cc == OP_FAIL)
{
add_jump(compiler, &backtrack->own_backtracks, JUMP(SLJIT_JUMP));
return cc + 1;
}
if (*cc == OP_ACCEPT && common->currententry == NULL && (common->re->overall_options & PCRE2_ENDANCHORED) != 0)
add_jump(compiler, &common->restart_match, CMP(SLJIT_NOT_EQUAL, STR_PTR, 0, STR_END, 0));
if (*cc == OP_ASSERT_ACCEPT || common->currententry != NULL || !common->might_be_empty)
{
/* No need to check notempty conditions. */
if (common->accept_label == NULL)
add_jump(compiler, &common->accept, JUMP(SLJIT_JUMP));
else
JUMPTO(SLJIT_JUMP, common->accept_label);
return cc + 1;
}
if (common->accept_label == NULL)
add_jump(compiler, &common->accept, CMP(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(0)));
else
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(0), common->accept_label);
if (HAS_VIRTUAL_REGISTERS)
{
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV_U32, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, options));
}
else
OP1(SLJIT_MOV_U32, TMP2, 0, SLJIT_MEM1(ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, options));
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY);
add_jump(compiler, &backtrack->own_backtracks, JUMP(SLJIT_NOT_ZERO));
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY_ATSTART);
if (common->accept_label == NULL)
add_jump(compiler, &common->accept, JUMP(SLJIT_ZERO));
else
JUMPTO(SLJIT_ZERO, common->accept_label);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(HAS_VIRTUAL_REGISTERS ? TMP1 : ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, str));
if (common->accept_label == NULL)
add_jump(compiler, &common->accept, CMP(SLJIT_NOT_EQUAL, TMP2, 0, STR_PTR, 0));
else
CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, STR_PTR, 0, common->accept_label);
add_jump(compiler, &backtrack->own_backtracks, JUMP(SLJIT_JUMP));
return cc + 1;
}
static SLJIT_INLINE PCRE2_SPTR compile_close_matchingpath(compiler_common *common, PCRE2_SPTR cc)
{
DEFINE_COMPILER;
int offset = GET2(cc, 1);
BOOL optimized_cbracket = common->optimized_cbracket[offset] != 0;
/* Data will be discarded anyway... */
if (common->currententry != NULL)
return cc + 1 + IMM2_SIZE;
if (!optimized_cbracket)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR_PRIV(offset));
offset <<= 1;
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), STR_PTR, 0);
if (!optimized_cbracket)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset), TMP1, 0);
return cc + 1 + IMM2_SIZE;
}
static SLJIT_INLINE PCRE2_SPTR compile_control_verb_matchingpath(compiler_common *common, PCRE2_SPTR cc, backtrack_common *parent)
{
DEFINE_COMPILER;
backtrack_common *backtrack;
PCRE2_UCHAR opcode = *cc;
PCRE2_SPTR ccend = cc + 1;
if (opcode == OP_COMMIT_ARG || opcode == OP_PRUNE_ARG ||
opcode == OP_SKIP_ARG || opcode == OP_THEN_ARG)
ccend += 2 + cc[1];
PUSH_BACKTRACK(sizeof(backtrack_common), cc, NULL);
if (opcode == OP_SKIP)
{
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
return ccend;
}
if (opcode == OP_COMMIT_ARG || opcode == OP_PRUNE_ARG || opcode == OP_THEN_ARG)
{
if (HAS_VIRTUAL_REGISTERS)
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)(cc + 2));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->mark_ptr, TMP2, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(HAS_VIRTUAL_REGISTERS ? TMP1 : ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, mark_ptr), TMP2, 0);
}
return ccend;
}
static PCRE2_UCHAR then_trap_opcode[1] = { OP_THEN_TRAP };
static SLJIT_INLINE void compile_then_trap_matchingpath(compiler_common *common, PCRE2_SPTR cc, PCRE2_SPTR ccend, backtrack_common *parent)
{
DEFINE_COMPILER;
backtrack_common *backtrack;
BOOL needs_control_head;
int size;
PUSH_BACKTRACK_NOVALUE(sizeof(then_trap_backtrack), cc);
common->then_trap = BACKTRACK_AS(then_trap_backtrack);
BACKTRACK_AS(then_trap_backtrack)->common.cc = then_trap_opcode;
BACKTRACK_AS(then_trap_backtrack)->start = (sljit_sw)(cc - common->start);
BACKTRACK_AS(then_trap_backtrack)->framesize = get_framesize(common, cc, ccend, FALSE, &needs_control_head);
size = BACKTRACK_AS(then_trap_backtrack)->framesize;
size = 3 + (size < 0 ? 0 : size);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr);
allocate_stack(common, size);
if (size > 3)
OP2(SLJIT_ADD, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, STACK_TOP, 0, SLJIT_IMM, (size - 3) * sizeof(sljit_sw));
else
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, STACK_TOP, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(size - 1), SLJIT_IMM, BACKTRACK_AS(then_trap_backtrack)->start);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(size - 2), SLJIT_IMM, type_then_trap);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(size - 3), TMP2, 0);
size = BACKTRACK_AS(then_trap_backtrack)->framesize;
if (size >= 0)
init_frame(common, cc, ccend, size - 1, 0);
}
static void compile_matchingpath(compiler_common *common, PCRE2_SPTR cc, PCRE2_SPTR ccend, backtrack_common *parent)
{
DEFINE_COMPILER;
backtrack_common *backtrack;
BOOL has_then_trap = FALSE;
then_trap_backtrack *save_then_trap = NULL;
size_t op_len;
SLJIT_ASSERT(*ccend == OP_END || (*ccend >= OP_ALT && *ccend <= OP_KETRPOS));
if (common->has_then && common->then_offsets[cc - common->start] != 0)
{
SLJIT_ASSERT(*ccend != OP_END && common->control_head_ptr != 0);
has_then_trap = TRUE;
save_then_trap = common->then_trap;
/* Tail item on backtrack. */
compile_then_trap_matchingpath(common, cc, ccend, parent);
}
while (cc < ccend)
{
switch(*cc)
{
case OP_SOD:
case OP_SOM:
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
case OP_EODN:
case OP_EOD:
case OP_DOLL:
case OP_DOLLM:
case OP_CIRC:
case OP_CIRCM:
case OP_NOT_UCP_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
cc = compile_simple_assertion_matchingpath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks);
break;
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
case OP_ALLANY:
case OP_ANYBYTE:
case OP_NOTPROP:
case OP_PROP:
case OP_ANYNL:
case OP_NOT_HSPACE:
case OP_HSPACE:
case OP_NOT_VSPACE:
case OP_VSPACE:
case OP_EXTUNI:
case OP_NOT:
case OP_NOTI:
cc = compile_char1_matchingpath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks, TRUE);
break;
case OP_SET_SOM:
PUSH_BACKTRACK_NOVALUE(sizeof(backtrack_common), cc);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(0));
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(0), STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0);
cc++;
break;
case OP_CHAR:
case OP_CHARI:
if (common->mode == PCRE2_JIT_COMPLETE)
cc = compile_charn_matchingpath(common, cc, ccend, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks);
else
cc = compile_char1_matchingpath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks, TRUE);
break;
case OP_STAR:
case OP_MINSTAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_QUERY:
case OP_MINQUERY:
case OP_UPTO:
case OP_MINUPTO:
case OP_EXACT:
case OP_POSSTAR:
case OP_POSPLUS:
case OP_POSQUERY:
case OP_POSUPTO:
case OP_STARI:
case OP_MINSTARI:
case OP_PLUSI:
case OP_MINPLUSI:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_UPTOI:
case OP_MINUPTOI:
case OP_EXACTI:
case OP_POSSTARI:
case OP_POSPLUSI:
case OP_POSQUERYI:
case OP_POSUPTOI:
case OP_NOTSTAR:
case OP_NOTMINSTAR:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
case OP_NOTUPTO:
case OP_NOTMINUPTO:
case OP_NOTEXACT:
case OP_NOTPOSSTAR:
case OP_NOTPOSPLUS:
case OP_NOTPOSQUERY:
case OP_NOTPOSUPTO:
case OP_NOTSTARI:
case OP_NOTMINSTARI:
case OP_NOTPLUSI:
case OP_NOTMINPLUSI:
case OP_NOTQUERYI:
case OP_NOTMINQUERYI:
case OP_NOTUPTOI:
case OP_NOTMINUPTOI:
case OP_NOTEXACTI:
case OP_NOTPOSSTARI:
case OP_NOTPOSPLUSI:
case OP_NOTPOSQUERYI:
case OP_NOTPOSUPTOI:
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEEXACT:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSPLUS:
case OP_TYPEPOSQUERY:
case OP_TYPEPOSUPTO:
cc = compile_iterator_matchingpath(common, cc, parent, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks);
break;
case OP_CLASS:
case OP_NCLASS:
if (cc[1 + (32 / sizeof(PCRE2_UCHAR))] >= OP_CRSTAR && cc[1 + (32 / sizeof(PCRE2_UCHAR))] <= OP_CRPOSRANGE)
cc = compile_iterator_matchingpath(common, cc, parent, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks);
else
cc = compile_char1_matchingpath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks, TRUE);
break;
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
case OP_XCLASS:
case OP_ECLASS:
op_len = GET(cc, 1);
if (cc[op_len] >= OP_CRSTAR && cc[op_len] <= OP_CRPOSRANGE)
cc = compile_iterator_matchingpath(common, cc, parent, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks);
else
cc = compile_char1_matchingpath(common, *cc, cc + 1, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks, TRUE);
break;
#endif
case OP_REF:
case OP_REFI:
op_len = PRIV(OP_lengths)[*cc];
if (cc[op_len] >= OP_CRSTAR && cc[op_len] <= OP_CRPOSRANGE)
cc = compile_ref_iterator_matchingpath(common, cc, parent);
else
{
compile_ref_matchingpath(common, cc, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks, TRUE, FALSE);
cc += op_len;
}
break;
case OP_DNREF:
case OP_DNREFI:
op_len = PRIV(OP_lengths)[*cc];
if (cc[op_len] >= OP_CRSTAR && cc[op_len] <= OP_CRPOSRANGE)
cc = compile_ref_iterator_matchingpath(common, cc, parent);
else
{
compile_dnref_search(common, cc, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks);
compile_ref_matchingpath(common, cc, parent->top != NULL ? &parent->top->simple_backtracks : &parent->own_backtracks, TRUE, FALSE);
cc += op_len;
}
break;
case OP_RECURSE:
cc = compile_recurse_matchingpath(common, cc, parent);
break;
case OP_CALLOUT:
case OP_CALLOUT_STR:
cc = compile_callout_matchingpath(common, cc, parent);
break;
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
PUSH_BACKTRACK_NOVALUE(sizeof(assert_backtrack), cc);
cc = compile_assert_matchingpath(common, cc, BACKTRACK_AS(assert_backtrack), FALSE);
break;
case OP_BRAMINZERO:
PUSH_BACKTRACK_NOVALUE(sizeof(braminzero_backtrack), cc);
cc = bracketend(cc + 1);
if (*(cc - 1 - LINK_SIZE) != OP_KETRMIN)
{
allocate_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
}
else
{
allocate_stack(common, 2);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), STR_PTR, 0);
}
BACKTRACK_AS(braminzero_backtrack)->matchingpath = LABEL();
count_match(common);
break;
case OP_ASSERT_NA:
case OP_ASSERTBACK_NA:
case OP_ASSERT_SCS:
case OP_ONCE:
case OP_SCRIPT_RUN:
case OP_BRA:
case OP_CBRA:
case OP_COND:
case OP_SBRA:
case OP_SCBRA:
case OP_SCOND:
cc = compile_bracket_matchingpath(common, cc, parent);
break;
case OP_BRAZERO:
if (cc[1] > OP_ASSERTBACK_NOT)
cc = compile_bracket_matchingpath(common, cc, parent);
else
{
PUSH_BACKTRACK_NOVALUE(sizeof(assert_backtrack), cc);
cc = compile_assert_matchingpath(common, cc, BACKTRACK_AS(assert_backtrack), FALSE);
}
break;
case OP_BRAPOS:
case OP_CBRAPOS:
case OP_SBRAPOS:
case OP_SCBRAPOS:
case OP_BRAPOSZERO:
cc = compile_bracketpos_matchingpath(common, cc, parent);
break;
case OP_MARK:
PUSH_BACKTRACK_NOVALUE(sizeof(backtrack_common), cc);
SLJIT_ASSERT(common->mark_ptr != 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->mark_ptr);
allocate_stack(common, common->has_skip_arg ? 5 : 1);
if (HAS_VIRTUAL_REGISTERS)
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(common->has_skip_arg ? 4 : 0), TMP2, 0);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, (sljit_sw)(cc + 2));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->mark_ptr, TMP2, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(HAS_VIRTUAL_REGISTERS ? TMP1 : ARGUMENTS), SLJIT_OFFSETOF(jit_arguments, mark_ptr), TMP2, 0);
if (common->has_skip_arg)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, STACK_TOP, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, type_mark);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), SLJIT_IMM, (sljit_sw)(cc + 2));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(3), STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP1, 0);
}
cc += 1 + 2 + cc[1];
break;
case OP_PRUNE:
case OP_PRUNE_ARG:
case OP_SKIP:
case OP_SKIP_ARG:
case OP_THEN:
case OP_THEN_ARG:
case OP_COMMIT:
case OP_COMMIT_ARG:
cc = compile_control_verb_matchingpath(common, cc, parent);
break;
case OP_FAIL:
case OP_ACCEPT:
case OP_ASSERT_ACCEPT:
cc = compile_fail_accept_matchingpath(common, cc, parent);
break;
case OP_CLOSE:
cc = compile_close_matchingpath(common, cc);
break;
case OP_SKIPZERO:
cc = bracketend(cc + 1);
break;
default:
SLJIT_UNREACHABLE();
return;
}
if (cc == NULL)
return;
}
if (has_then_trap)
{
/* Head item on backtrack. */
PUSH_BACKTRACK_NOVALUE(sizeof(then_trap_backtrack), cc);
BACKTRACK_AS(then_trap_backtrack)->common.cc = then_trap_opcode;
BACKTRACK_AS(then_trap_backtrack)->then_trap = common->then_trap;
common->then_trap = save_then_trap;
}
SLJIT_ASSERT(cc == ccend);
}
#undef PUSH_BACKTRACK
#undef PUSH_BACKTRACK_NOVALUE
#undef BACKTRACK_AS
#define COMPILE_BACKTRACKINGPATH(current) \
do \
{ \
compile_backtrackingpath(common, (current)); \
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler))) \
return; \
} \
while (0)
#define CURRENT_AS(type) ((type *)current)
static void compile_newline_move_back(compiler_common *common)
{
DEFINE_COMPILER;
struct sljit_jump *jump;
OP2(SLJIT_SUB, TMP1, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
jump = CMP(SLJIT_LESS_EQUAL, TMP1, 0, TMP2, 0);
/* All newlines are single byte, or their last byte
is not equal to CHAR_NL/CHAR_CR even if UTF is enabled. */
OP1(MOV_UCHAR, SLJIT_TMP_DEST_REG, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-2));
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(-1));
OP2(SLJIT_SHL, SLJIT_TMP_DEST_REG, 0, SLJIT_TMP_DEST_REG, 0, SLJIT_IMM, 8);
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_TMP_DEST_REG, 0);
OP2U(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, SLJIT_IMM, CHAR_CR << 8 | CHAR_NL);
OP_FLAGS(SLJIT_MOV, TMP1, 0, SLJIT_EQUAL);
#if PCRE2_CODE_UNIT_WIDTH == 16 || PCRE2_CODE_UNIT_WIDTH == 32
OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, UCHAR_SHIFT);
#endif
OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
JUMPHERE(jump);
}
static void compile_iterator_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
DEFINE_COMPILER;
PCRE2_SPTR cc = current->cc;
PCRE2_UCHAR opcode;
PCRE2_UCHAR type;
sljit_u32 max = 0, exact;
struct sljit_label *label = NULL;
struct sljit_jump *jump = NULL;
jump_list *jumplist = NULL;
PCRE2_SPTR end;
int private_data_ptr = PRIVATE_DATA(cc);
int base = (private_data_ptr == 0) ? SLJIT_MEM1(STACK_TOP) : SLJIT_MEM1(SLJIT_SP);
int offset0 = (private_data_ptr == 0) ? STACK(0) : private_data_ptr;
int offset1 = (private_data_ptr == 0) ? STACK(1) : private_data_ptr + SSIZE_OF(sw);
cc = get_iterator_parameters(common, cc, &opcode, &type, &max, &exact, &end);
switch(opcode)
{
case OP_STAR:
case OP_UPTO:
if (type == OP_EXTUNI)
{
SLJIT_ASSERT(private_data_ptr == 0);
set_jumps(current->own_backtracks, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(char_iterator_backtrack)->matchingpath);
}
else
{
if (CURRENT_AS(char_iterator_backtrack)->charpos.charpos_enabled)
{
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
OP1(SLJIT_MOV, TMP2, 0, base, offset1);
jump = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0);
label = LABEL();
if (type == OP_ANYNL)
compile_newline_move_back(common);
move_back(common, NULL, TRUE);
OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
if (CURRENT_AS(char_iterator_backtrack)->charpos.othercasebit != 0)
OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, CURRENT_AS(char_iterator_backtrack)->charpos.othercasebit);
CMPTO(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, CURRENT_AS(char_iterator_backtrack)->charpos.chr, CURRENT_AS(char_iterator_backtrack)->matchingpath);
/* The range beginning must match, no need to compare. */
JUMPTO(SLJIT_JUMP, label);
set_jumps(current->own_backtracks, LABEL());
current->own_backtracks = NULL;
}
else
{
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
if (opcode == OP_STAR && exact == 1)
{
if (type == OP_ANYNL)
{
OP1(SLJIT_MOV, TMP2, 0, base, offset1);
compile_newline_move_back(common);
}
move_back(common, NULL, TRUE);
jump = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, base, offset1);
}
else
{
if (type == OP_ANYNL)
{
OP1(SLJIT_MOV, TMP2, 0, base, offset1);
jump = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, TMP2, 0);
compile_newline_move_back(common);
}
else
jump = CMP(SLJIT_LESS_EQUAL, STR_PTR, 0, base, offset1);
move_back(common, NULL, TRUE);
}
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
JUMPTO(SLJIT_JUMP, CURRENT_AS(char_iterator_backtrack)->matchingpath);
set_jumps(current->own_backtracks, LABEL());
}
JUMPHERE(jump);
if (private_data_ptr == 0)
free_stack(common, 2);
}
break;
case OP_QUERY:
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
OP1(SLJIT_MOV, base, offset0, SLJIT_IMM, 0);
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(char_iterator_backtrack)->matchingpath);
jump = JUMP(SLJIT_JUMP);
set_jumps(current->own_backtracks, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
OP1(SLJIT_MOV, base, offset0, SLJIT_IMM, 0);
JUMPTO(SLJIT_JUMP, CURRENT_AS(char_iterator_backtrack)->matchingpath);
JUMPHERE(jump);
if (private_data_ptr == 0)
free_stack(common, 1);
break;
case OP_MINSTAR:
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
if (exact == 0)
{
compile_char1_matchingpath(common, type, cc, &jumplist, TRUE);
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
}
else if (exact > 1)
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 1);
JUMPTO(SLJIT_JUMP, CURRENT_AS(char_iterator_backtrack)->matchingpath);
set_jumps(exact > 0 ? current->own_backtracks : jumplist, LABEL());
if (private_data_ptr == 0)
free_stack(common, 1);
break;
case OP_MINUPTO:
OP1(SLJIT_MOV, TMP1, 0, base, offset1);
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
OP2(SLJIT_SUB | SLJIT_SET_Z, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);
if (exact == 0)
{
add_jump(compiler, &jumplist, JUMP(SLJIT_ZERO));
OP1(SLJIT_MOV, base, offset1, TMP1, 0);
compile_char1_matchingpath(common, type, cc, &jumplist, TRUE);
OP1(SLJIT_MOV, base, offset0, STR_PTR, 0);
JUMPTO(SLJIT_JUMP, CURRENT_AS(char_iterator_backtrack)->matchingpath);
set_jumps(jumplist, LABEL());
}
else
{
if (exact > 1)
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 1);
OP1(SLJIT_MOV, base, offset1, TMP1, 0);
JUMPTO(SLJIT_NOT_ZERO, CURRENT_AS(char_iterator_backtrack)->matchingpath);
set_jumps(current->own_backtracks, LABEL());
}
if (private_data_ptr == 0)
free_stack(common, 2);
break;
case OP_MINQUERY:
OP1(SLJIT_MOV, STR_PTR, 0, base, offset0);
OP1(SLJIT_MOV, base, offset0, SLJIT_IMM, 0);
if (exact >= 1)
{
if (exact >= 2)
OP1(SLJIT_MOV, TMP3, 0, SLJIT_IMM, 1);
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(char_iterator_backtrack)->matchingpath);
set_jumps(current->own_backtracks, LABEL());
}
else
{
jump = CMP(SLJIT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0);
compile_char1_matchingpath(common, type, cc, &jumplist, TRUE);
JUMPTO(SLJIT_JUMP, CURRENT_AS(char_iterator_backtrack)->matchingpath);
set_jumps(jumplist, LABEL());
JUMPHERE(jump);
}
if (private_data_ptr == 0)
free_stack(common, 1);
break;
default:
SLJIT_UNREACHABLE();
break;
}
}
static SLJIT_INLINE void compile_ref_iterator_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
DEFINE_COMPILER;
PCRE2_SPTR cc = current->cc;
BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
PCRE2_UCHAR type;
type = cc[PRIV(OP_lengths)[*cc]];
if ((type & 0x1) == 0)
{
/* Maximize case. */
set_jumps(current->own_backtracks, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(ref_iterator_backtrack)->matchingpath);
return;
}
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(ref_iterator_backtrack)->matchingpath);
set_jumps(current->own_backtracks, LABEL());
free_stack(common, ref ? 2 : 3);
}
static SLJIT_INLINE void compile_recurse_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
DEFINE_COMPILER;
recurse_entry *entry;
if (!CURRENT_AS(recurse_backtrack)->inlined_pattern)
{
entry = CURRENT_AS(recurse_backtrack)->entry;
if (entry->backtrack_label == NULL)
add_jump(compiler, &entry->backtrack_calls, JUMP(SLJIT_FAST_CALL));
else
JUMPTO(SLJIT_FAST_CALL, entry->backtrack_label);
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, CURRENT_AS(recurse_backtrack)->matchingpath);
}
else
compile_backtrackingpath(common, current->top);
set_jumps(current->own_backtracks, LABEL());
}
static void compile_assert_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
DEFINE_COMPILER;
PCRE2_SPTR cc = current->cc;
PCRE2_UCHAR bra = OP_BRA;
struct sljit_jump *brajump = NULL;
SLJIT_ASSERT(*cc != OP_BRAMINZERO);
if (*cc == OP_BRAZERO)
{
bra = *cc;
cc++;
}
if (bra == OP_BRAZERO)
{
SLJIT_ASSERT(current->own_backtracks == NULL);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
}
if (CURRENT_AS(assert_backtrack)->framesize < 0)
{
set_jumps(current->own_backtracks, LABEL());
if (bra == OP_BRAZERO)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(assert_backtrack)->matchingpath);
free_stack(common, 1);
}
return;
}
if (bra == OP_BRAZERO)
{
if (*cc == OP_ASSERT_NOT || *cc == OP_ASSERTBACK_NOT)
{
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(assert_backtrack)->matchingpath);
free_stack(common, 1);
return;
}
free_stack(common, 1);
brajump = CMP(SLJIT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0);
}
if (*cc == OP_ASSERT || *cc == OP_ASSERTBACK)
{
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), CURRENT_AS(assert_backtrack)->private_data_ptr);
add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(-2));
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (CURRENT_AS(assert_backtrack)->framesize - 1) * sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), CURRENT_AS(assert_backtrack)->private_data_ptr, TMP1, 0);
set_jumps(current->own_backtracks, LABEL());
}
else
set_jumps(current->own_backtracks, LABEL());
if (bra == OP_BRAZERO)
{
/* We know there is enough place on the stack. */
OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
JUMPTO(SLJIT_JUMP, CURRENT_AS(assert_backtrack)->matchingpath);
JUMPHERE(brajump);
}
}
static void compile_bracket_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
DEFINE_COMPILER;
int opcode, stacksize, alt_count, alt_max;
int offset = 0;
int private_data_ptr = CURRENT_AS(bracket_backtrack)->private_data_ptr;
int repeat_ptr = 0, repeat_type = 0, repeat_count = 0;
PCRE2_SPTR cc = current->cc;
PCRE2_SPTR ccbegin;
PCRE2_SPTR ccprev;
PCRE2_UCHAR bra = OP_BRA;
PCRE2_UCHAR ket;
const assert_backtrack *assert;
BOOL has_alternatives;
BOOL needs_control_head = FALSE;
BOOL has_vreverse;
struct sljit_jump *brazero = NULL;
struct sljit_jump *next_alt = NULL;
struct sljit_jump *once = NULL;
struct sljit_jump *cond = NULL;
struct sljit_label *rmin_label = NULL;
struct sljit_label *exact_label = NULL;
struct sljit_jump *mov_addr = NULL;
if (*cc == OP_BRAZERO || *cc == OP_BRAMINZERO)
{
bra = *cc;
cc++;
}
opcode = *cc;
ccbegin = bracketend(cc) - 1 - LINK_SIZE;
ket = *ccbegin;
if (ket == OP_KET && PRIVATE_DATA(ccbegin) != 0)
{
repeat_ptr = PRIVATE_DATA(ccbegin);
repeat_type = PRIVATE_DATA(ccbegin + 2);
repeat_count = PRIVATE_DATA(ccbegin + 3);
SLJIT_ASSERT(repeat_type != 0 && repeat_count != 0);
if (repeat_type == OP_UPTO)
ket = OP_KETRMAX;
if (repeat_type == OP_MINUPTO)
ket = OP_KETRMIN;
}
ccbegin = cc;
cc += GET(cc, 1);
has_alternatives = *cc == OP_ALT;
if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND))
has_alternatives = (ccbegin[1 + LINK_SIZE] >= OP_ASSERT && ccbegin[1 + LINK_SIZE] <= OP_ASSERTBACK_NOT) || CURRENT_AS(bracket_backtrack)->u.no_capture != NULL;
if (opcode == OP_CBRA || opcode == OP_SCBRA)
offset = (GET2(ccbegin, 1 + LINK_SIZE)) << 1;
if (SLJIT_UNLIKELY(opcode == OP_COND) && (*cc == OP_KETRMAX || *cc == OP_KETRMIN))
opcode = OP_SCOND;
alt_max = has_alternatives ? no_alternatives(ccbegin) : 0;
/* Decoding the needs_control_head in framesize. */
if (opcode == OP_ONCE)
{
needs_control_head = (CURRENT_AS(bracket_backtrack)->u.framesize & 0x1) != 0;
CURRENT_AS(bracket_backtrack)->u.framesize >>= 1;
}
if (ket != OP_KET && repeat_type != 0)
{
/* TMP1 is used in OP_KETRMIN below. */
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
if (repeat_type == OP_UPTO)
OP2(SLJIT_ADD, SLJIT_MEM1(SLJIT_SP), repeat_ptr, TMP1, 0, SLJIT_IMM, 1);
else
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), repeat_ptr, TMP1, 0);
}
if (ket == OP_KETRMAX)
{
if (bra == OP_BRAZERO)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
brazero = CMP(SLJIT_EQUAL, TMP1, 0, SLJIT_IMM, 0);
}
}
else if (ket == OP_KETRMIN)
{
if (bra != OP_BRAMINZERO)
{
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
if (repeat_type != 0)
{
/* TMP1 was set a few lines above. */
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, CURRENT_AS(bracket_backtrack)->recursive_matchingpath);
/* Drop STR_PTR for non-greedy plus quantifier. */
if (opcode != OP_ONCE)
free_stack(common, 1);
}
else if (opcode >= OP_SBRA || opcode == OP_ONCE)
{
/* Checking zero-length iteration. */
if (opcode != OP_ONCE || CURRENT_AS(bracket_backtrack)->u.framesize < 0)
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr, CURRENT_AS(bracket_backtrack)->recursive_matchingpath);
else
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_MEM1(TMP1), STACK(-CURRENT_AS(bracket_backtrack)->u.framesize - 2), CURRENT_AS(bracket_backtrack)->recursive_matchingpath);
}
/* Drop STR_PTR for non-greedy plus quantifier. */
if (opcode != OP_ONCE)
free_stack(common, 1);
}
else
JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->recursive_matchingpath);
}
rmin_label = LABEL();
if (repeat_type != 0)
OP2(SLJIT_ADD, SLJIT_MEM1(SLJIT_SP), repeat_ptr, SLJIT_MEM1(SLJIT_SP), repeat_ptr, SLJIT_IMM, 1);
}
else if (bra == OP_BRAZERO)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
brazero = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0);
}
else if (repeat_type == OP_EXACT)
{
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), repeat_ptr, SLJIT_IMM, 1);
exact_label = LABEL();
}
if (offset != 0)
{
if (common->capture_last_ptr != 0)
{
SLJIT_ASSERT(common->optimized_cbracket[offset >> 1] == 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr, TMP1, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(2));
free_stack(common, 3);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset), TMP2, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), TMP1, 0);
}
else if (common->optimized_cbracket[offset >> 1] == 0)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
free_stack(common, 2);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset), TMP1, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), TMP2, 0);
}
}
else if (SLJIT_UNLIKELY(opcode == OP_ASSERT_SCS))
{
OP1(SLJIT_MOV, TMP1, 0, STR_END, 0);
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw), TMP1, 0);
/* Nested scs blocks will not update this variable. */
if (common->restore_end_ptr == 0)
common->restore_end_ptr = private_data_ptr + sizeof(sljit_sw);
}
if (SLJIT_UNLIKELY(opcode == OP_ONCE))
{
int framesize = CURRENT_AS(bracket_backtrack)->u.framesize;
SLJIT_ASSERT(framesize != 0);
if (framesize > 0)
{
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL));
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize - 1) * sizeof(sljit_sw));
}
once = JUMP(SLJIT_JUMP);
}
else if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND))
{
if (has_alternatives)
{
/* Always exactly one alternative. */
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
alt_max = 2;
next_alt = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0);
}
}
else if (has_alternatives)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
if (alt_max > 3)
{
sljit_emit_ijump(compiler, SLJIT_JUMP, TMP1, 0);
SLJIT_ASSERT(CURRENT_AS(bracket_backtrack)->matching_mov_addr != NULL);
sljit_set_label(CURRENT_AS(bracket_backtrack)->matching_mov_addr, LABEL());
sljit_emit_op0(compiler, SLJIT_ENDBR);
}
else
next_alt = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0);
}
COMPILE_BACKTRACKINGPATH(current->top);
if (current->own_backtracks)
set_jumps(current->own_backtracks, LABEL());
if (SLJIT_UNLIKELY(opcode == OP_COND) || SLJIT_UNLIKELY(opcode == OP_SCOND))
{
/* Conditional block always has at most one alternative. */
if (ccbegin[1 + LINK_SIZE] >= OP_ASSERT && ccbegin[1 + LINK_SIZE] <= OP_ASSERTBACK_NOT)
{
SLJIT_ASSERT(has_alternatives);
assert = CURRENT_AS(bracket_backtrack)->u.assert;
SLJIT_ASSERT(assert->framesize != 0);
if (assert->framesize > 0 && (ccbegin[1 + LINK_SIZE] == OP_ASSERT || ccbegin[1 + LINK_SIZE] == OP_ASSERTBACK))
{
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), assert->private_data_ptr);
add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(-2));
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (assert->framesize - 1) * sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), assert->private_data_ptr, TMP1, 0);
}
cond = JUMP(SLJIT_JUMP);
set_jumps(CURRENT_AS(bracket_backtrack)->u.assert->condfailed, LABEL());
}
else if (CURRENT_AS(bracket_backtrack)->u.no_capture != NULL)
{
SLJIT_ASSERT(has_alternatives);
cond = JUMP(SLJIT_JUMP);
set_jumps(CURRENT_AS(bracket_backtrack)->u.no_capture, LABEL());
}
else
SLJIT_ASSERT(!has_alternatives);
}
if (has_alternatives)
{
alt_count = 1;
do
{
current->top = NULL;
current->own_backtracks = NULL;
current->simple_backtracks = NULL;
/* Conditional blocks always have an additional alternative, even if it is empty. */
if (*cc == OP_ALT)
{
ccprev = cc + 1 + LINK_SIZE;
cc += GET(cc, 1);
has_vreverse = FALSE;
switch (opcode)
{
case OP_ASSERTBACK:
case OP_ASSERTBACK_NA:
SLJIT_ASSERT(private_data_ptr != 0);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
has_vreverse = (*ccprev == OP_VREVERSE);
if (*ccprev == OP_REVERSE || has_vreverse)
ccprev = compile_reverse_matchingpath(common, ccprev, current);
break;
case OP_ASSERT_SCS:
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(2));
break;
case OP_ONCE:
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(needs_control_head ? 1 : 0));
break;
case OP_COND:
case OP_SCOND:
break;
default:
if (private_data_ptr != 0)
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
else
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
break;
}
compile_matchingpath(common, ccprev, cc, current);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
return;
switch (opcode)
{
case OP_ASSERTBACK_NA:
if (has_vreverse)
{
SLJIT_ASSERT(current->top != NULL && PRIVATE_DATA(ccbegin + 1));
add_jump(compiler, ¤t->top->simple_backtracks, CMP(SLJIT_LESS, STR_PTR, 0, STR_END, 0));
}
if (PRIVATE_DATA(ccbegin + 1))
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw));
break;
case OP_ASSERT_NA:
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr);
break;
case OP_SCRIPT_RUN:
match_script_run_common(common, private_data_ptr, current);
break;
}
}
/* Instructions after the current alternative is successfully matched. */
/* There is a similar code in compile_bracket_matchingpath. */
if (opcode == OP_ONCE)
match_once_common(common, ket, CURRENT_AS(bracket_backtrack)->u.framesize, private_data_ptr, has_alternatives, needs_control_head);
stacksize = 0;
if (repeat_type == OP_MINUPTO)
{
/* We need to preserve the counter. TMP2 will be used below. */
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), repeat_ptr);
stacksize++;
}
if (ket != OP_KET || bra != OP_BRA)
stacksize++;
if (offset != 0)
{
if (common->capture_last_ptr != 0)
stacksize++;
if (common->optimized_cbracket[offset >> 1] == 0)
stacksize += 2;
}
if (opcode != OP_ONCE)
stacksize++;
if (stacksize > 0)
allocate_stack(common, stacksize);
stacksize = 0;
if (repeat_type == OP_MINUPTO)
{
/* TMP2 was set above. */
OP2(SLJIT_SUB, SLJIT_MEM1(STACK_TOP), STACK(stacksize), TMP2, 0, SLJIT_IMM, 1);
stacksize++;
}
if (ket != OP_KET || bra != OP_BRA)
{
if (ket != OP_KET)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), STR_PTR, 0);
else
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, 0);
stacksize++;
}
if (offset != 0)
stacksize = match_capture_common(common, stacksize, offset, private_data_ptr);
if (opcode != OP_ONCE)
{
if (alt_max <= 3)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(stacksize), SLJIT_IMM, alt_count);
else
mov_addr = sljit_emit_mov_addr(compiler, SLJIT_MEM1(STACK_TOP), STACK(stacksize));
}
if (offset != 0 && ket == OP_KETRMAX && common->optimized_cbracket[offset >> 1] != 0)
{
/* If ket is not OP_KETRMAX, this code path is executed after the jump to alternative_matchingpath. */
SLJIT_ASSERT(private_data_ptr == OVECTOR(offset + 0));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), STR_PTR, 0);
}
JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->alternative_matchingpath);
if (opcode != OP_ONCE)
{
if (alt_max <= 3)
{
JUMPHERE(next_alt);
alt_count++;
if (alt_count < alt_max)
{
SLJIT_ASSERT(alt_count == 2 && alt_max == 3);
next_alt = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 1);
}
}
else
{
sljit_set_label(mov_addr, LABEL());
sljit_emit_op0(compiler, SLJIT_ENDBR);
}
}
COMPILE_BACKTRACKINGPATH(current->top);
if (current->own_backtracks)
set_jumps(current->own_backtracks, LABEL());
SLJIT_ASSERT(!current->simple_backtracks);
}
while (*cc == OP_ALT);
if (cond != NULL)
{
SLJIT_ASSERT(opcode == OP_COND || opcode == OP_SCOND);
if (ccbegin[1 + LINK_SIZE] == OP_ASSERT_NOT || ccbegin[1 + LINK_SIZE] == OP_ASSERTBACK_NOT)
{
assert = CURRENT_AS(bracket_backtrack)->u.assert;
SLJIT_ASSERT(assert->framesize != 0);
if (assert->framesize > 0)
{
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), assert->private_data_ptr);
add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(-2));
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (assert->framesize - 1) * sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), assert->private_data_ptr, TMP1, 0);
}
}
JUMPHERE(cond);
}
/* Free the STR_PTR. */
if (private_data_ptr == 0)
free_stack(common, 1);
}
if (offset != 0)
{
/* Using both tmp register is better for instruction scheduling. */
if (common->optimized_cbracket[offset >> 1] != 0)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
free_stack(common, 2);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset), TMP1, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), TMP2, 0);
}
else
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP1, 0);
}
}
else if (opcode == OP_ASSERTBACK_NA && PRIVATE_DATA(ccbegin + 1))
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP1, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw), TMP2, 0);
free_stack(common, 4);
}
else if (opcode == OP_ASSERT_NA || opcode == OP_ASSERTBACK_NA || opcode == OP_SCRIPT_RUN || opcode == OP_SBRA || opcode == OP_SCOND)
{
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
}
else if (opcode == OP_ASSERT_SCS)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP1, 0);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr + sizeof(sljit_sw), TMP2, 0);
free_stack(common, has_alternatives ? 3 : 2);
set_jumps(CURRENT_AS(bracket_backtrack)->u.no_capture, LABEL());
/* Nested scs blocks will not update this variable. */
if (common->restore_end_ptr == private_data_ptr + SSIZE_OF(sw))
common->restore_end_ptr = 0;
}
else if (opcode == OP_ONCE)
{
cc = ccbegin + GET(ccbegin, 1);
stacksize = needs_control_head ? 1 : 0;
if (CURRENT_AS(bracket_backtrack)->u.framesize >= 0)
{
/* Reset head and drop saved frame. */
stacksize += CURRENT_AS(bracket_backtrack)->u.framesize + ((ket != OP_KET || *cc == OP_ALT) ? 2 : 1);
}
else if (ket == OP_KETRMAX || (*cc == OP_ALT && ket != OP_KETRMIN))
{
/* The STR_PTR must be released. */
stacksize++;
}
if (stacksize > 0)
free_stack(common, stacksize);
JUMPHERE(once);
/* Restore previous private_data_ptr */
if (CURRENT_AS(bracket_backtrack)->u.framesize >= 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, SLJIT_MEM1(STACK_TOP), STACK(-CURRENT_AS(bracket_backtrack)->u.framesize - 1));
else if (ket == OP_KETRMIN)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
/* See the comment below. */
free_stack(common, 2);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), private_data_ptr, TMP1, 0);
}
}
if (repeat_type == OP_EXACT)
{
OP2(SLJIT_ADD, TMP1, 0, SLJIT_MEM1(SLJIT_SP), repeat_ptr, SLJIT_IMM, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), repeat_ptr, TMP1, 0);
CMPTO(SLJIT_LESS_EQUAL, TMP1, 0, SLJIT_IMM, repeat_count, exact_label);
}
else if (ket == OP_KETRMAX)
{
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
if (bra != OP_BRAZERO)
free_stack(common, 1);
CMPTO(SLJIT_NOT_EQUAL, STR_PTR, 0, SLJIT_IMM, 0, CURRENT_AS(bracket_backtrack)->recursive_matchingpath);
if (bra == OP_BRAZERO)
{
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->zero_matchingpath);
JUMPHERE(brazero);
free_stack(common, 1);
}
}
else if (ket == OP_KETRMIN)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
/* OP_ONCE removes everything in case of a backtrack, so we don't
need to explicitly release the STR_PTR. The extra release would
affect badly the free_stack(2) above. */
if (opcode != OP_ONCE)
free_stack(common, 1);
CMPTO(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, rmin_label);
if (opcode == OP_ONCE)
free_stack(common, bra == OP_BRAMINZERO ? 2 : 1);
else if (bra == OP_BRAMINZERO)
free_stack(common, 1);
}
else if (bra == OP_BRAZERO)
{
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
JUMPTO(SLJIT_JUMP, CURRENT_AS(bracket_backtrack)->zero_matchingpath);
JUMPHERE(brazero);
}
}
static SLJIT_INLINE void compile_bracketpos_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
DEFINE_COMPILER;
int offset;
struct sljit_jump *jump;
PCRE2_SPTR cc;
/* No retry on backtrack, just drop everything. */
if (CURRENT_AS(bracketpos_backtrack)->framesize < 0)
{
cc = current->cc;
if (*cc == OP_BRAPOSZERO)
cc++;
if (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS)
{
offset = (GET2(cc, 1 + LINK_SIZE)) << 1;
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset), TMP1, 0);
if (common->capture_last_ptr != 0)
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(2));
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(offset + 1), TMP2, 0);
if (common->capture_last_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr, TMP1, 0);
}
set_jumps(current->own_backtracks, LABEL());
free_stack(common, CURRENT_AS(bracketpos_backtrack)->stacksize);
return;
}
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), CURRENT_AS(bracketpos_backtrack)->private_data_ptr);
add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL));
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (CURRENT_AS(bracketpos_backtrack)->framesize - 1) * sizeof(sljit_sw));
if (current->own_backtracks)
{
jump = JUMP(SLJIT_JUMP);
set_jumps(current->own_backtracks, LABEL());
/* Drop the stack frame. */
free_stack(common, CURRENT_AS(bracketpos_backtrack)->stacksize);
JUMPHERE(jump);
}
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), CURRENT_AS(bracketpos_backtrack)->private_data_ptr, SLJIT_MEM1(STACK_TOP), STACK(-CURRENT_AS(bracketpos_backtrack)->framesize - 1));
}
static SLJIT_INLINE void compile_braminzero_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
assert_backtrack backtrack;
current->top = NULL;
current->own_backtracks = NULL;
current->simple_backtracks = NULL;
if (current->cc[1] > OP_ASSERTBACK_NOT)
{
/* Manual call of compile_bracket_matchingpath and compile_bracket_backtrackingpath. */
compile_bracket_matchingpath(common, current->cc, current);
compile_bracket_backtrackingpath(common, current->top);
}
else
{
memset(&backtrack, 0, sizeof(backtrack));
backtrack.common.cc = current->cc;
backtrack.matchingpath = CURRENT_AS(braminzero_backtrack)->matchingpath;
/* Manual call of compile_assert_matchingpath. */
compile_assert_matchingpath(common, current->cc, &backtrack, FALSE);
}
SLJIT_ASSERT(!current->simple_backtracks && !current->own_backtracks);
}
static SLJIT_INLINE void compile_control_verb_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
DEFINE_COMPILER;
PCRE2_UCHAR opcode = *current->cc;
struct sljit_label *loop;
struct sljit_jump *jump;
if (opcode == OP_THEN || opcode == OP_THEN_ARG)
{
if (common->then_trap != NULL)
{
SLJIT_ASSERT(common->control_head_ptr != 0);
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, type_then_trap);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, common->then_trap->start);
jump = JUMP(SLJIT_JUMP);
loop = LABEL();
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
JUMPHERE(jump);
CMPTO(SLJIT_NOT_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(1), TMP1, 0, loop);
CMPTO(SLJIT_NOT_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(2), TMP2, 0, loop);
add_jump(compiler, &common->then_trap->quit, JUMP(SLJIT_JUMP));
return;
}
else if (!common->local_quit_available && common->in_positive_assertion)
{
add_jump(compiler, &common->positive_assertion_quit, JUMP(SLJIT_JUMP));
return;
}
}
if (common->restore_end_ptr != 0 && opcode != OP_SKIP_ARG)
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), common->restore_end_ptr);
if (common->local_quit_available)
{
/* Abort match with a fail. */
if (common->quit_label == NULL)
add_jump(compiler, &common->quit, JUMP(SLJIT_JUMP));
else
JUMPTO(SLJIT_JUMP, common->quit_label);
return;
}
if (opcode == OP_SKIP_ARG)
{
SLJIT_ASSERT(common->control_head_ptr != 0 && TMP1 == SLJIT_R0 && STR_PTR == SLJIT_R1);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr);
OP1(SLJIT_MOV, SLJIT_R1, 0, SLJIT_IMM, (sljit_sw)(current->cc + 2));
sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS2(W, W, W), SLJIT_IMM, SLJIT_FUNC_ADDR(do_search_mark));
if (common->restore_end_ptr == 0)
{
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_R0, 0);
add_jump(compiler, &common->reset_match, CMP(SLJIT_NOT_EQUAL, SLJIT_R0, 0, SLJIT_IMM, 0));
return;
}
jump = CMP(SLJIT_EQUAL, SLJIT_R0, 0, SLJIT_IMM, 0);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_R0, 0);
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), common->restore_end_ptr);
add_jump(compiler, &common->reset_match, JUMP(SLJIT_JUMP));
JUMPHERE(jump);
return;
}
if (opcode == OP_SKIP)
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
else
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_IMM, 0);
add_jump(compiler, &common->reset_match, JUMP(SLJIT_JUMP));
}
static SLJIT_INLINE void compile_vreverse_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
DEFINE_COMPILER;
struct sljit_jump *jump;
struct sljit_label *label;
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(2));
jump = CMP(SLJIT_GREATER_EQUAL, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(3));
skip_valid_char(common);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(2), STR_PTR, 0);
JUMPTO(SLJIT_JUMP, CURRENT_AS(vreverse_backtrack)->matchingpath);
label = LABEL();
sljit_set_label(jump, label);
set_jumps(current->own_backtracks, label);
}
static SLJIT_INLINE void compile_then_trap_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
DEFINE_COMPILER;
struct sljit_jump *jump;
int framesize;
int size;
if (CURRENT_AS(then_trap_backtrack)->then_trap)
{
common->then_trap = CURRENT_AS(then_trap_backtrack)->then_trap;
return;
}
size = CURRENT_AS(then_trap_backtrack)->framesize;
size = 3 + (size < 0 ? 0 : size);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(size - 3));
free_stack(common, size);
jump = JUMP(SLJIT_JUMP);
set_jumps(CURRENT_AS(then_trap_backtrack)->quit, LABEL());
framesize = CURRENT_AS(then_trap_backtrack)->framesize;
SLJIT_ASSERT(framesize != 0);
/* STACK_TOP is set by THEN. */
if (framesize > 0)
{
add_jump(compiler, &common->revertframes, JUMP(SLJIT_FAST_CALL));
OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, (framesize - 1) * sizeof(sljit_sw));
}
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 3);
JUMPHERE(jump);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, TMP1, 0);
}
static void compile_backtrackingpath(compiler_common *common, struct backtrack_common *current)
{
DEFINE_COMPILER;
then_trap_backtrack *save_then_trap = common->then_trap;
while (current)
{
if (current->simple_backtracks != NULL)
set_jumps(current->simple_backtracks, LABEL());
switch(*current->cc)
{
case OP_SET_SOM:
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(0), TMP1, 0);
break;
case OP_STAR:
case OP_MINSTAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_QUERY:
case OP_MINQUERY:
case OP_UPTO:
case OP_MINUPTO:
case OP_EXACT:
case OP_POSSTAR:
case OP_POSPLUS:
case OP_POSQUERY:
case OP_POSUPTO:
case OP_STARI:
case OP_MINSTARI:
case OP_PLUSI:
case OP_MINPLUSI:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_UPTOI:
case OP_MINUPTOI:
case OP_EXACTI:
case OP_POSSTARI:
case OP_POSPLUSI:
case OP_POSQUERYI:
case OP_POSUPTOI:
case OP_NOTSTAR:
case OP_NOTMINSTAR:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
case OP_NOTUPTO:
case OP_NOTMINUPTO:
case OP_NOTEXACT:
case OP_NOTPOSSTAR:
case OP_NOTPOSPLUS:
case OP_NOTPOSQUERY:
case OP_NOTPOSUPTO:
case OP_NOTSTARI:
case OP_NOTMINSTARI:
case OP_NOTPLUSI:
case OP_NOTMINPLUSI:
case OP_NOTQUERYI:
case OP_NOTMINQUERYI:
case OP_NOTUPTOI:
case OP_NOTMINUPTOI:
case OP_NOTEXACTI:
case OP_NOTPOSSTARI:
case OP_NOTPOSPLUSI:
case OP_NOTPOSQUERYI:
case OP_NOTPOSUPTOI:
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEEXACT:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSPLUS:
case OP_TYPEPOSQUERY:
case OP_TYPEPOSUPTO:
/* Since classes has no backtracking path, this
backtrackingpath was pushed by an iterator. */
case OP_CLASS:
case OP_NCLASS:
#if defined SUPPORT_UNICODE || PCRE2_CODE_UNIT_WIDTH != 8
case OP_XCLASS:
case OP_ECLASS:
#endif
compile_iterator_backtrackingpath(common, current);
break;
case OP_REF:
case OP_REFI:
case OP_DNREF:
case OP_DNREFI:
compile_ref_iterator_backtrackingpath(common, current);
break;
case OP_RECURSE:
compile_recurse_backtrackingpath(common, current);
break;
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
compile_assert_backtrackingpath(common, current);
break;
case OP_ASSERT_NA:
case OP_ASSERTBACK_NA:
case OP_ASSERT_SCS:
case OP_ONCE:
case OP_SCRIPT_RUN:
case OP_BRA:
case OP_CBRA:
case OP_COND:
case OP_SBRA:
case OP_SCBRA:
case OP_SCOND:
compile_bracket_backtrackingpath(common, current);
break;
case OP_BRAZERO:
if (current->cc[1] > OP_ASSERTBACK_NOT)
compile_bracket_backtrackingpath(common, current);
else
compile_assert_backtrackingpath(common, current);
break;
case OP_BRAPOS:
case OP_CBRAPOS:
case OP_SBRAPOS:
case OP_SCBRAPOS:
case OP_BRAPOSZERO:
compile_bracketpos_backtrackingpath(common, current);
break;
case OP_BRAMINZERO:
compile_braminzero_backtrackingpath(common, current);
break;
case OP_MARK:
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(common->has_skip_arg ? 4 : 0));
if (common->has_skip_arg)
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
free_stack(common, common->has_skip_arg ? 5 : 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->mark_ptr, TMP1, 0);
if (common->has_skip_arg)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, TMP2, 0);
break;
case OP_THEN:
case OP_THEN_ARG:
case OP_PRUNE:
case OP_PRUNE_ARG:
case OP_SKIP:
case OP_SKIP_ARG:
compile_control_verb_backtrackingpath(common, current);
break;
case OP_COMMIT:
case OP_COMMIT_ARG:
if (common->restore_end_ptr != 0)
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_SP), common->restore_end_ptr);
if (!common->local_quit_available)
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE2_ERROR_NOMATCH);
if (common->quit_label == NULL)
add_jump(compiler, &common->quit, JUMP(SLJIT_JUMP));
else
JUMPTO(SLJIT_JUMP, common->quit_label);
break;
case OP_CALLOUT:
case OP_CALLOUT_STR:
case OP_FAIL:
case OP_ACCEPT:
case OP_ASSERT_ACCEPT:
set_jumps(current->own_backtracks, LABEL());
break;
case OP_VREVERSE:
compile_vreverse_backtrackingpath(common, current);
break;
case OP_THEN_TRAP:
/* A virtual opcode for then traps. */
compile_then_trap_backtrackingpath(common, current);
break;
default:
SLJIT_UNREACHABLE();
break;
}
current = current->prev;
}
common->then_trap = save_then_trap;
}
static SLJIT_INLINE void compile_recurse(compiler_common *common)
{
DEFINE_COMPILER;
PCRE2_SPTR cc = common->start + common->currententry->start;
PCRE2_SPTR ccbegin = cc + 1 + LINK_SIZE + (*cc == OP_BRA ? 0 : IMM2_SIZE);
PCRE2_SPTR ccend = bracketend(cc) - (1 + LINK_SIZE);
uint32_t recurse_flags = 0;
int private_data_size = get_recurse_data_length(common, ccbegin, ccend, &recurse_flags);
int alt_count, alt_max, local_size;
backtrack_common altbacktrack;
jump_list *match = NULL;
struct sljit_jump *next_alt = NULL;
struct sljit_jump *accept_exit = NULL;
struct sljit_label *quit;
struct sljit_jump *mov_addr = NULL;
/* Recurse captures then. */
common->then_trap = NULL;
SLJIT_ASSERT(*cc == OP_BRA || *cc == OP_CBRA || *cc == OP_CBRAPOS || *cc == OP_SCBRA || *cc == OP_SCBRAPOS);
alt_max = no_alternatives(cc);
alt_count = 0;
/* Matching path. */
SLJIT_ASSERT(common->currententry->entry_label == NULL && common->recursive_head_ptr != 0);
common->currententry->entry_label = LABEL();
set_jumps(common->currententry->entry_calls, common->currententry->entry_label);
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, TMP2, 0);
count_match(common);
local_size = (alt_max > 1) ? 2 : 1;
/* (Reversed) stack layout:
[private data][return address][optional: str ptr] ... [optional: alternative index][recursive_head_ptr] */
allocate_stack(common, private_data_size + local_size);
/* Save return address. */
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(local_size - 1), TMP2, 0);
copy_recurse_data(common, ccbegin, ccend, recurse_copy_from_global, local_size, private_data_size + local_size, recurse_flags);
/* This variable is saved and restored all time when we enter or exit from a recursive context. */
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->recursive_head_ptr, STACK_TOP, 0);
if (recurse_flags & recurse_flag_control_head_found)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_IMM, 0);
if (alt_max > 1)
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
memset(&altbacktrack, 0, sizeof(backtrack_common));
common->quit_label = NULL;
common->accept_label = NULL;
common->quit = NULL;
common->accept = NULL;
altbacktrack.cc = ccbegin;
cc += GET(cc, 1);
while (1)
{
altbacktrack.top = NULL;
altbacktrack.own_backtracks = NULL;
if (altbacktrack.cc != ccbegin)
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
compile_matchingpath(common, altbacktrack.cc, cc, &altbacktrack);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
return;
allocate_stack(common, (alt_max > 1 || (recurse_flags & recurse_flag_accept_found)) ? 2 : 1);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_SP), common->recursive_head_ptr);
if (alt_max > 1 || (recurse_flags & recurse_flag_accept_found))
{
if (alt_max > 3)
mov_addr = sljit_emit_mov_addr(compiler, SLJIT_MEM1(STACK_TOP), STACK(1));
else
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, alt_count);
}
add_jump(compiler, &match, JUMP(SLJIT_JUMP));
if (alt_count == 0)
{
/* Backtracking path entry. */
SLJIT_ASSERT(common->currententry->backtrack_label == NULL);
common->currententry->backtrack_label = LABEL();
set_jumps(common->currententry->backtrack_calls, common->currententry->backtrack_label);
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, TMP1, 0);
if (recurse_flags & recurse_flag_accept_found)
accept_exit = CMP(SLJIT_EQUAL, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, -1);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(0));
/* Save return address. */
OP1(SLJIT_MOV, SLJIT_MEM1(TMP2), STACK(local_size - 1), TMP1, 0);
copy_recurse_data(common, ccbegin, ccend, recurse_swap_global, local_size, private_data_size + local_size, recurse_flags);
if (alt_max > 1)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), STACK(1));
free_stack(common, 2);
if (alt_max > 3)
{
sljit_emit_ijump(compiler, SLJIT_JUMP, TMP1, 0);
sljit_set_label(mov_addr, LABEL());
sljit_emit_op0(compiler, SLJIT_ENDBR);
}
else
next_alt = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0);
}
else
free_stack(common, (recurse_flags & recurse_flag_accept_found) ? 2 : 1);
}
else if (alt_max > 3)
{
sljit_set_label(mov_addr, LABEL());
sljit_emit_op0(compiler, SLJIT_ENDBR);
}
else
{
JUMPHERE(next_alt);
if (alt_count + 1 < alt_max)
{
SLJIT_ASSERT(alt_count == 1 && alt_max == 3);
next_alt = CMP(SLJIT_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 1);
}
}
alt_count++;
compile_backtrackingpath(common, altbacktrack.top);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
return;
set_jumps(altbacktrack.own_backtracks, LABEL());
if (*cc != OP_ALT)
break;
altbacktrack.cc = cc + 1 + LINK_SIZE;
cc += GET(cc, 1);
}
/* No alternative is matched. */
quit = LABEL();
copy_recurse_data(common, ccbegin, ccend, recurse_copy_private_to_global, local_size, private_data_size + local_size, recurse_flags);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(local_size - 1));
free_stack(common, private_data_size + local_size);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
OP_SRC(SLJIT_FAST_RETURN, TMP2, 0);
if (common->quit != NULL)
{
SLJIT_ASSERT(recurse_flags & recurse_flag_quit_found);
set_jumps(common->quit, LABEL());
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), common->recursive_head_ptr);
copy_recurse_data(common, ccbegin, ccend, recurse_copy_shared_to_global, local_size, private_data_size + local_size, recurse_flags);
JUMPTO(SLJIT_JUMP, quit);
}
if (recurse_flags & recurse_flag_accept_found)
{
JUMPHERE(accept_exit);
free_stack(common, 2);
/* Save return address. */
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(local_size - 1), TMP1, 0);
copy_recurse_data(common, ccbegin, ccend, recurse_copy_kept_shared_to_global, local_size, private_data_size + local_size, recurse_flags);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(STACK_TOP), STACK(local_size - 1));
free_stack(common, private_data_size + local_size);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
OP_SRC(SLJIT_FAST_RETURN, TMP2, 0);
}
if (common->accept != NULL)
{
SLJIT_ASSERT(recurse_flags & recurse_flag_accept_found);
set_jumps(common->accept, LABEL());
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(SLJIT_SP), common->recursive_head_ptr);
OP1(SLJIT_MOV, TMP2, 0, STACK_TOP, 0);
allocate_stack(common, 2);
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, -1);
}
set_jumps(match, LABEL());
OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), TMP2, 0);
copy_recurse_data(common, ccbegin, ccend, recurse_swap_global, local_size, private_data_size + local_size, recurse_flags);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), STACK(local_size - 1));
OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 1);
OP_SRC(SLJIT_FAST_RETURN, TMP2, 0);
}
#undef COMPILE_BACKTRACKINGPATH
#undef CURRENT_AS
#define PUBLIC_JIT_COMPILE_CONFIGURATION_OPTIONS \
(PCRE2_JIT_INVALID_UTF)
static int jit_compile(pcre2_code *code, sljit_u32 mode)
{
pcre2_real_code *re = (pcre2_real_code *)code;
struct sljit_compiler *compiler;
backtrack_common rootbacktrack;
compiler_common common_data;
compiler_common *common = &common_data;
const sljit_u8 *tables = re->tables;
void *allocator_data = &re->memctl;
int private_data_size;
PCRE2_SPTR ccend;
executable_functions *functions;
void *executable_func;
sljit_uw executable_size, private_data_length, total_length;
struct sljit_label *mainloop_label = NULL;
struct sljit_label *continue_match_label;
struct sljit_label *empty_match_found_label = NULL;
struct sljit_label *empty_match_backtrack_label = NULL;
struct sljit_label *reset_match_label;
struct sljit_label *quit_label;
struct sljit_jump *jump;
struct sljit_jump *minlength_check_failed = NULL;
struct sljit_jump *empty_match = NULL;
struct sljit_jump *end_anchor_failed = NULL;
jump_list *reqcu_not_found = NULL;
SLJIT_ASSERT(tables);
#if HAS_VIRTUAL_REGISTERS == 1
SLJIT_ASSERT(sljit_get_register_index(SLJIT_GP_REGISTER, TMP3) < 0 && sljit_get_register_index(SLJIT_GP_REGISTER, ARGUMENTS) < 0 && sljit_get_register_index(SLJIT_GP_REGISTER, RETURN_ADDR) < 0);
#elif HAS_VIRTUAL_REGISTERS == 0
SLJIT_ASSERT(sljit_get_register_index(SLJIT_GP_REGISTER, TMP3) >= 0 && sljit_get_register_index(SLJIT_GP_REGISTER, ARGUMENTS) >= 0 && sljit_get_register_index(SLJIT_GP_REGISTER, RETURN_ADDR) >= 0);
#else
#error "Invalid value for HAS_VIRTUAL_REGISTERS"
#endif
memset(&rootbacktrack, 0, sizeof(backtrack_common));
memset(common, 0, sizeof(compiler_common));
common->re = re;
common->name_table = (PCRE2_SPTR)((uint8_t *)re + sizeof(pcre2_real_code));
rootbacktrack.cc = (PCRE2_SPTR)((uint8_t *)re + re->code_start);
#ifdef SUPPORT_UNICODE
common->invalid_utf = (mode & PCRE2_JIT_INVALID_UTF) != 0;
#endif /* SUPPORT_UNICODE */
mode &= ~PUBLIC_JIT_COMPILE_CONFIGURATION_OPTIONS;
common->start = rootbacktrack.cc;
common->read_only_data_head = NULL;
common->fcc = tables + fcc_offset;
common->lcc = (sljit_sw)(tables + lcc_offset);
common->mode = mode;
common->might_be_empty = (re->minlength == 0) || (re->flags & PCRE2_MATCH_EMPTY);
common->allow_empty_partial = (re->max_lookbehind > 0) || (re->flags & PCRE2_MATCH_EMPTY);
common->nltype = NLTYPE_FIXED;
switch(re->newline_convention)
{
case PCRE2_NEWLINE_CR: common->newline = CHAR_CR; break;
case PCRE2_NEWLINE_LF: common->newline = CHAR_NL; break;
case PCRE2_NEWLINE_CRLF: common->newline = (CHAR_CR << 8) | CHAR_NL; break;
case PCRE2_NEWLINE_ANY: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANY; break;
case PCRE2_NEWLINE_ANYCRLF: common->newline = (CHAR_CR << 8) | CHAR_NL; common->nltype = NLTYPE_ANYCRLF; break;
case PCRE2_NEWLINE_NUL: common->newline = CHAR_NUL; break;
default: return PCRE2_ERROR_INTERNAL;
}
common->nlmax = READ_CHAR_MAX;
common->nlmin = 0;
if (re->bsr_convention == PCRE2_BSR_UNICODE)
common->bsr_nltype = NLTYPE_ANY;
else if (re->bsr_convention == PCRE2_BSR_ANYCRLF)
common->bsr_nltype = NLTYPE_ANYCRLF;
else
{
#ifdef BSR_ANYCRLF
common->bsr_nltype = NLTYPE_ANYCRLF;
#else
common->bsr_nltype = NLTYPE_ANY;
#endif
}
common->bsr_nlmax = READ_CHAR_MAX;
common->bsr_nlmin = 0;
common->endonly = (re->overall_options & PCRE2_DOLLAR_ENDONLY) != 0;
common->ctypes = (sljit_sw)(tables + ctypes_offset);
common->name_count = re->name_count;
common->name_entry_size = re->name_entry_size;
common->unset_backref = (re->overall_options & PCRE2_MATCH_UNSET_BACKREF) != 0;
common->alt_circumflex = (re->overall_options & PCRE2_ALT_CIRCUMFLEX) != 0;
#ifdef SUPPORT_UNICODE
/* PCRE2_UTF[16|32] have the same value as PCRE2_UTF8. */
common->utf = (re->overall_options & PCRE2_UTF) != 0;
common->ucp = (re->overall_options & PCRE2_UCP) != 0;
if (common->utf)
{
if (common->nltype == NLTYPE_ANY)
common->nlmax = 0x2029;
else if (common->nltype == NLTYPE_ANYCRLF)
common->nlmax = (CHAR_CR > CHAR_NL) ? CHAR_CR : CHAR_NL;
else
{
/* We only care about the first newline character. */
common->nlmax = common->newline & 0xff;
}
if (common->nltype == NLTYPE_FIXED)
common->nlmin = common->newline & 0xff;
else
common->nlmin = (CHAR_CR < CHAR_NL) ? CHAR_CR : CHAR_NL;
if (common->bsr_nltype == NLTYPE_ANY)
common->bsr_nlmax = 0x2029;
else
common->bsr_nlmax = (CHAR_CR > CHAR_NL) ? CHAR_CR : CHAR_NL;
common->bsr_nlmin = (CHAR_CR < CHAR_NL) ? CHAR_CR : CHAR_NL;
}
else
common->invalid_utf = FALSE;
#endif /* SUPPORT_UNICODE */
ccend = bracketend(common->start);
/* Calculate the local space size on the stack. */
common->ovector_start = LOCAL0;
/* Allocate space for temporary data structures. */
private_data_length = ccend - common->start;
/* The chance of overflow is very low, but might happen on 32 bit. */
if (private_data_length > ~(sljit_uw)0 / sizeof(sljit_s32))
return PCRE2_ERROR_NOMEMORY;
private_data_length *= sizeof(sljit_s32);
/* Align to 32 bit. */
total_length = ((re->top_bracket + 1) + (sljit_uw)(sizeof(sljit_s32) - 1)) & ~(sljit_uw)(sizeof(sljit_s32) - 1);
if (~(sljit_uw)0 - private_data_length < total_length)
return PCRE2_ERROR_NOMEMORY;
total_length += private_data_length;
common->private_data_ptrs = (sljit_s32*)SLJIT_MALLOC(total_length, allocator_data);
if (!common->private_data_ptrs)
return PCRE2_ERROR_NOMEMORY;
memset(common->private_data_ptrs, 0, private_data_length);
common->optimized_cbracket = ((sljit_u8 *)common->private_data_ptrs) + private_data_length;
#if defined DEBUG_FORCE_UNOPTIMIZED_CBRAS && DEBUG_FORCE_UNOPTIMIZED_CBRAS == 1
memset(common->optimized_cbracket, 0, re->top_bracket + 1);
#else
memset(common->optimized_cbracket, 1, re->top_bracket + 1);
#endif
SLJIT_ASSERT(*common->start == OP_BRA && ccend[-(1 + LINK_SIZE)] == OP_KET);
#if defined DEBUG_FORCE_UNOPTIMIZED_CBRAS && DEBUG_FORCE_UNOPTIMIZED_CBRAS == 2
common->capture_last_ptr = common->ovector_start;
common->ovector_start += sizeof(sljit_sw);
#endif
if (!check_opcode_types(common, common->start, ccend))
{
SLJIT_FREE(common->private_data_ptrs, allocator_data);
return PCRE2_ERROR_JIT_UNSUPPORTED;
}
/* Checking flags and updating ovector_start. */
if (mode == PCRE2_JIT_COMPLETE &&
(re->flags & PCRE2_LASTSET) != 0 &&
(re->optimization_flags & PCRE2_OPTIM_START_OPTIMIZE) != 0)
{
common->req_char_ptr = common->ovector_start;
common->ovector_start += sizeof(sljit_sw);
}
if (mode != PCRE2_JIT_COMPLETE)
{
common->start_used_ptr = common->ovector_start;
common->ovector_start += sizeof(sljit_sw);
if (mode == PCRE2_JIT_PARTIAL_SOFT)
{
common->hit_start = common->ovector_start;
common->ovector_start += sizeof(sljit_sw);
}
}
if ((re->overall_options & (PCRE2_FIRSTLINE | PCRE2_USE_OFFSET_LIMIT)) != 0)
{
common->match_end_ptr = common->ovector_start;
common->ovector_start += sizeof(sljit_sw);
}
#if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD
common->control_head_ptr = 1;
#endif
if (common->control_head_ptr != 0)
{
common->control_head_ptr = common->ovector_start;
common->ovector_start += sizeof(sljit_sw);
}
if (common->has_set_som)
{
/* Saving the real start pointer is necessary. */
common->start_ptr = common->ovector_start;
common->ovector_start += sizeof(sljit_sw);
}
/* Aligning ovector to even number of sljit words. */
if ((common->ovector_start & sizeof(sljit_sw)) != 0)
common->ovector_start += sizeof(sljit_sw);
if (common->start_ptr == 0)
common->start_ptr = OVECTOR(0);
/* Capturing brackets cannot be optimized if callouts are allowed. */
if (common->capture_last_ptr != 0)
memset(common->optimized_cbracket, 0, re->top_bracket + 1);
SLJIT_ASSERT(!(common->req_char_ptr != 0 && common->start_used_ptr != 0));
common->cbra_ptr = OVECTOR_START + (re->top_bracket + 1) * 2 * sizeof(sljit_sw);
private_data_size = common->cbra_ptr + (re->top_bracket + 1) * sizeof(sljit_sw);
if ((re->overall_options & PCRE2_ANCHORED) == 0 &&
(re->optimization_flags & PCRE2_OPTIM_START_OPTIMIZE) != 0 &&
!common->has_skip_in_assert_back)
detect_early_fail(common, common->start, &private_data_size, 0, 0);
set_private_data_ptrs(common, &private_data_size, ccend);
SLJIT_ASSERT(common->early_fail_start_ptr <= common->early_fail_end_ptr);
if (private_data_size > 65536)
{
SLJIT_FREE(common->private_data_ptrs, allocator_data);
return PCRE2_ERROR_JIT_UNSUPPORTED;
}
if (common->has_then)
{
total_length = ccend - common->start;
common->then_offsets = (sljit_u8 *)SLJIT_MALLOC(total_length, allocator_data);
if (!common->then_offsets)
{
SLJIT_FREE(common->private_data_ptrs, allocator_data);
return PCRE2_ERROR_NOMEMORY;
}
memset(common->then_offsets, 0, total_length);
set_then_offsets(common, common->start, NULL);
}
compiler = sljit_create_compiler(allocator_data);
if (!compiler)
{
SLJIT_FREE(common->private_data_ptrs, allocator_data);
if (common->has_then)
SLJIT_FREE(common->then_offsets, allocator_data);
return PCRE2_ERROR_NOMEMORY;
}
common->compiler = compiler;
/* Main pcre2_jit_exec entry. */
SLJIT_ASSERT((private_data_size & (sizeof(sljit_sw) - 1)) == 0);
sljit_emit_enter(compiler, 0, SLJIT_ARGS1(W, W), 5 | SLJIT_ENTER_VECTOR(SLJIT_NUMBER_OF_SCRATCH_VECTOR_REGISTERS), 5, private_data_size);
/* Register init. */
reset_ovector(common, (re->top_bracket + 1) * 2);
if (common->req_char_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->req_char_ptr, SLJIT_R0, 0);
OP1(SLJIT_MOV, ARGUMENTS, 0, SLJIT_S0, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_S0, 0);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str));
OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, end));
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, stack));
OP1(SLJIT_MOV_U32, TMP1, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, limit_match));
OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(struct sljit_stack, end));
OP1(SLJIT_MOV, STACK_LIMIT, 0, SLJIT_MEM1(TMP2), SLJIT_OFFSETOF(struct sljit_stack, start));
OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LIMIT_MATCH, TMP1, 0);
if (common->early_fail_start_ptr < common->early_fail_end_ptr)
reset_early_fail(common);
if (mode == PCRE2_JIT_PARTIAL_SOFT)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, -1);
if (common->mark_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->mark_ptr, SLJIT_IMM, 0);
if (common->control_head_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->control_head_ptr, SLJIT_IMM, 0);
/* Main part of the matching */
if ((re->overall_options & PCRE2_ANCHORED) == 0)
{
mainloop_label = mainloop_entry(common);
continue_match_label = LABEL();
/* Forward search if possible. */
if ((re->optimization_flags & PCRE2_OPTIM_START_OPTIMIZE) != 0)
{
if (mode == PCRE2_JIT_COMPLETE && fast_forward_first_n_chars(common))
;
else if ((re->flags & PCRE2_FIRSTSET) != 0)
fast_forward_first_char(common);
else if ((re->flags & PCRE2_STARTLINE) != 0)
fast_forward_newline(common);
else if ((re->flags & PCRE2_FIRSTMAPSET) != 0)
fast_forward_start_bits(common);
}
}
else
continue_match_label = LABEL();
if (mode == PCRE2_JIT_COMPLETE && re->minlength > 0 &&
(re->optimization_flags & PCRE2_OPTIM_START_OPTIMIZE) != 0)
{
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE2_ERROR_NOMATCH);
OP2(SLJIT_ADD, TMP2, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(re->minlength));
minlength_check_failed = CMP(SLJIT_GREATER, TMP2, 0, STR_END, 0);
}
if (common->req_char_ptr != 0)
reqcu_not_found = search_requested_char(common, (PCRE2_UCHAR)(re->last_codeunit), (re->flags & PCRE2_LASTCASELESS) != 0, (re->flags & PCRE2_FIRSTSET) != 0);
/* Store the current STR_PTR in OVECTOR(0). */
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), OVECTOR(0), STR_PTR, 0);
/* Copy the limit of allowed recursions. */
OP1(SLJIT_MOV, COUNT_MATCH, 0, SLJIT_MEM1(SLJIT_SP), LIMIT_MATCH);
if (common->capture_last_ptr != 0)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->capture_last_ptr, SLJIT_IMM, 0);
if (common->fast_forward_bc_ptr != NULL)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), PRIVATE_DATA(common->fast_forward_bc_ptr + 1) >> 3, STR_PTR, 0);
if (common->start_ptr != OVECTOR(0))
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->start_ptr, STR_PTR, 0);
/* Copy the beginning of the string. */
if (mode == PCRE2_JIT_PARTIAL_SOFT)
{
jump = CMP(SLJIT_NOT_EQUAL, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, -1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0);
JUMPHERE(jump);
}
else if (mode == PCRE2_JIT_PARTIAL_HARD)
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, STR_PTR, 0);
compile_matchingpath(common, common->start, ccend, &rootbacktrack);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
{
sljit_free_compiler(compiler);
SLJIT_FREE(common->private_data_ptrs, allocator_data);
if (common->has_then)
SLJIT_FREE(common->then_offsets, allocator_data);
PRIV(jit_free_rodata)(common->read_only_data_head, allocator_data);
return PCRE2_ERROR_NOMEMORY;
}
if ((re->overall_options & PCRE2_ENDANCHORED) != 0)
end_anchor_failed = CMP(SLJIT_NOT_EQUAL, STR_PTR, 0, STR_END, 0);
if (common->might_be_empty)
{
empty_match = CMP(SLJIT_EQUAL, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), OVECTOR(0));
empty_match_found_label = LABEL();
}
common->accept_label = LABEL();
if (common->accept != NULL)
set_jumps(common->accept, common->accept_label);
/* This means we have a match. Update the ovector. */
copy_ovector(common, re->top_bracket + 1);
common->quit_label = common->abort_label = LABEL();
if (common->quit != NULL)
set_jumps(common->quit, common->quit_label);
if (common->abort != NULL)
set_jumps(common->abort, common->abort_label);
if (minlength_check_failed != NULL)
SET_LABEL(minlength_check_failed, common->abort_label);
sljit_emit_op0(compiler, SLJIT_SKIP_FRAMES_BEFORE_RETURN);
sljit_emit_return(compiler, SLJIT_MOV, SLJIT_RETURN_REG, 0);
if (common->failed_match != NULL)
{
SLJIT_ASSERT(common->mode == PCRE2_JIT_COMPLETE);
set_jumps(common->failed_match, LABEL());
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE2_ERROR_NOMATCH);
JUMPTO(SLJIT_JUMP, common->abort_label);
}
if ((re->overall_options & PCRE2_ENDANCHORED) != 0)
JUMPHERE(end_anchor_failed);
if (mode != PCRE2_JIT_COMPLETE)
{
common->partialmatchlabel = LABEL();
set_jumps(common->partialmatch, common->partialmatchlabel);
return_with_partial_match(common, common->quit_label);
}
if (common->might_be_empty)
empty_match_backtrack_label = LABEL();
compile_backtrackingpath(common, rootbacktrack.top);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
{
sljit_free_compiler(compiler);
SLJIT_FREE(common->private_data_ptrs, allocator_data);
if (common->has_then)
SLJIT_FREE(common->then_offsets, allocator_data);
PRIV(jit_free_rodata)(common->read_only_data_head, allocator_data);
return PCRE2_ERROR_NOMEMORY;
}
SLJIT_ASSERT(rootbacktrack.prev == NULL);
reset_match_label = LABEL();
if (mode == PCRE2_JIT_PARTIAL_SOFT)
{
/* Update hit_start only in the first time. */
jump = CMP(SLJIT_NOT_EQUAL, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->start_ptr);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->start_used_ptr, SLJIT_IMM, -1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), common->hit_start, TMP1, 0);
JUMPHERE(jump);
}
/* Check we have remaining characters. */
if ((re->overall_options & PCRE2_ANCHORED) == 0 && common->match_end_ptr != 0)
{
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), common->match_end_ptr);
}
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP),
(common->fast_forward_bc_ptr != NULL) ? (PRIVATE_DATA(common->fast_forward_bc_ptr + 1) >> 3) : common->start_ptr);
if ((re->overall_options & PCRE2_ANCHORED) == 0)
{
if (common->ff_newline_shortcut != NULL)
{
/* There cannot be more newlines if PCRE2_FIRSTLINE is set. */
if ((re->overall_options & PCRE2_FIRSTLINE) == 0)
{
if (common->match_end_ptr != 0)
{
OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);
OP1(SLJIT_MOV, STR_END, 0, TMP1, 0);
CMPTO(SLJIT_LESS, STR_PTR, 0, TMP1, 0, common->ff_newline_shortcut);
OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
}
else
CMPTO(SLJIT_LESS, STR_PTR, 0, STR_END, 0, common->ff_newline_shortcut);
}
}
else
CMPTO(SLJIT_LESS, STR_PTR, 0, (common->match_end_ptr == 0) ? STR_END : TMP1, 0, mainloop_label);
}
/* No more remaining characters. */
if (reqcu_not_found != NULL)
set_jumps(reqcu_not_found, LABEL());
if (mode == PCRE2_JIT_PARTIAL_SOFT)
CMPTO(SLJIT_NOT_EQUAL, SLJIT_MEM1(SLJIT_SP), common->hit_start, SLJIT_IMM, -1, common->partialmatchlabel);
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE2_ERROR_NOMATCH);
JUMPTO(SLJIT_JUMP, common->quit_label);
flush_stubs(common);
if (common->might_be_empty)
{
JUMPHERE(empty_match);
OP1(SLJIT_MOV, TMP1, 0, ARGUMENTS, 0);
OP1(SLJIT_MOV_U32, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, options));
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY);
JUMPTO(SLJIT_NOT_ZERO, empty_match_backtrack_label);
OP2U(SLJIT_AND | SLJIT_SET_Z, TMP2, 0, SLJIT_IMM, PCRE2_NOTEMPTY_ATSTART);
JUMPTO(SLJIT_ZERO, empty_match_found_label);
OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP1), SLJIT_OFFSETOF(jit_arguments, str));
CMPTO(SLJIT_NOT_EQUAL, TMP2, 0, STR_PTR, 0, empty_match_found_label);
JUMPTO(SLJIT_JUMP, empty_match_backtrack_label);
}
common->fast_forward_bc_ptr = NULL;
common->early_fail_start_ptr = 0;
common->early_fail_end_ptr = 0;
common->currententry = common->entries;
common->local_quit_available = TRUE;
quit_label = common->quit_label;
SLJIT_ASSERT(common->restore_end_ptr == 0);
if (common->currententry != NULL)
{
/* A free bit for each private data. */
common->recurse_bitset_size = ((private_data_size / SSIZE_OF(sw)) + 7) >> 3;
SLJIT_ASSERT(common->recurse_bitset_size > 0);
common->recurse_bitset = (sljit_u8*)SLJIT_MALLOC(common->recurse_bitset_size, allocator_data);;
if (common->recurse_bitset != NULL)
{
do
{
/* Might add new entries. */
compile_recurse(common);
if (SLJIT_UNLIKELY(sljit_get_compiler_error(compiler)))
break;
flush_stubs(common);
common->currententry = common->currententry->next;
}
while (common->currententry != NULL);
SLJIT_FREE(common->recurse_bitset, allocator_data);
}
if (common->currententry != NULL)
{
/* The common->recurse_bitset has been freed. */
SLJIT_ASSERT(sljit_get_compiler_error(compiler) || common->recurse_bitset == NULL);
sljit_free_compiler(compiler);
SLJIT_FREE(common->private_data_ptrs, allocator_data);
if (common->has_then)
SLJIT_FREE(common->then_offsets, allocator_data);
PRIV(jit_free_rodata)(common->read_only_data_head, allocator_data);
return PCRE2_ERROR_NOMEMORY;
}
}
common->local_quit_available = FALSE;
common->quit_label = quit_label;
SLJIT_ASSERT(common->restore_end_ptr == 0);
/* Allocating stack, returns with PCRE2_ERROR_JIT_STACKLIMIT if fails. */
/* This is a (really) rare case. */
set_jumps(common->stackalloc, LABEL());
/* RETURN_ADDR is not a saved register. */
SLJIT_ASSERT(common->locals_size >= 2 * SSIZE_OF(sw));
sljit_emit_op_dst(compiler, SLJIT_FAST_ENTER, SLJIT_MEM1(SLJIT_SP), LOCAL0);
SLJIT_ASSERT(TMP1 == SLJIT_R0 && STR_PTR == SLJIT_R1);
OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_SP), LOCAL1, STR_PTR, 0);
OP1(SLJIT_MOV, SLJIT_R0, 0, ARGUMENTS, 0);
OP2(SLJIT_SUB, SLJIT_R1, 0, STACK_LIMIT, 0, SLJIT_IMM, STACK_GROWTH_RATE);
OP1(SLJIT_MOV, SLJIT_R0, 0, SLJIT_MEM1(SLJIT_R0), SLJIT_OFFSETOF(jit_arguments, stack));
OP1(SLJIT_MOV, STACK_LIMIT, 0, TMP2, 0);
sljit_emit_icall(compiler, SLJIT_CALL, SLJIT_ARGS2(W, W, W), SLJIT_IMM, SLJIT_FUNC_ADDR(sljit_stack_resize));
jump = CMP(SLJIT_EQUAL, SLJIT_RETURN_REG, 0, SLJIT_IMM, 0);
OP1(SLJIT_MOV, TMP2, 0, STACK_LIMIT, 0);
OP1(SLJIT_MOV, STACK_LIMIT, 0, SLJIT_RETURN_REG, 0);
OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_SP), LOCAL0);
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), LOCAL1);
OP_SRC(SLJIT_FAST_RETURN, TMP1, 0);
/* Allocation failed. */
JUMPHERE(jump);
/* We break the return address cache here, but this is a really rare case. */
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE2_ERROR_JIT_STACKLIMIT);
JUMPTO(SLJIT_JUMP, common->quit_label);
/* Call limit reached. */
set_jumps(common->calllimit, LABEL());
OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE2_ERROR_MATCHLIMIT);
JUMPTO(SLJIT_JUMP, common->quit_label);
if (common->revertframes != NULL)
{
set_jumps(common->revertframes, LABEL());
do_revertframes(common);
}
if (common->wordboundary != NULL)
{
set_jumps(common->wordboundary, LABEL());
check_wordboundary(common, FALSE);
}
if (common->ucp_wordboundary != NULL)
{
set_jumps(common->ucp_wordboundary, LABEL());
check_wordboundary(common, TRUE);
}
if (common->anynewline != NULL)
{
set_jumps(common->anynewline, LABEL());
check_anynewline(common);
}
if (common->hspace != NULL)
{
set_jumps(common->hspace, LABEL());
check_hspace(common);
}
if (common->vspace != NULL)
{
set_jumps(common->vspace, LABEL());
check_vspace(common);
}
if (common->casefulcmp != NULL)
{
set_jumps(common->casefulcmp, LABEL());
do_casefulcmp(common);
}
if (common->caselesscmp != NULL)
{
set_jumps(common->caselesscmp, LABEL());
do_caselesscmp(common);
}
if (common->reset_match != NULL || common->restart_match != NULL)
{
if (common->restart_match != NULL)
{
set_jumps(common->restart_match, LABEL());
OP1(SLJIT_MOV, STR_PTR, 0, SLJIT_MEM1(SLJIT_SP), common->start_ptr);
}
set_jumps(common->reset_match, LABEL());
do_reset_match(common, (re->top_bracket + 1) * 2);
/* The value of restart_match is in TMP1. */
CMPTO(SLJIT_GREATER, STR_PTR, 0, TMP1, 0, continue_match_label);
OP1(SLJIT_MOV, STR_PTR, 0, TMP1, 0);
JUMPTO(SLJIT_JUMP, reset_match_label);
}
#ifdef SUPPORT_UNICODE
#if PCRE2_CODE_UNIT_WIDTH == 8
if (common->utfreadchar != NULL)
{
set_jumps(common->utfreadchar, LABEL());
do_utfreadchar(common);
}
if (common->utfreadtype8 != NULL)
{
set_jumps(common->utfreadtype8, LABEL());
do_utfreadtype8(common);
}
if (common->utfpeakcharback != NULL)
{
set_jumps(common->utfpeakcharback, LABEL());
do_utfpeakcharback(common);
}
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */
#if PCRE2_CODE_UNIT_WIDTH == 8 || PCRE2_CODE_UNIT_WIDTH == 16
if (common->utfreadchar_invalid != NULL)
{
set_jumps(common->utfreadchar_invalid, LABEL());
do_utfreadchar_invalid(common);
}
if (common->utfreadnewline_invalid != NULL)
{
set_jumps(common->utfreadnewline_invalid, LABEL());
do_utfreadnewline_invalid(common);
}
if (common->utfmoveback_invalid)
{
set_jumps(common->utfmoveback_invalid, LABEL());
do_utfmoveback_invalid(common);
}
if (common->utfpeakcharback_invalid)
{
set_jumps(common->utfpeakcharback_invalid, LABEL());
do_utfpeakcharback_invalid(common);
}
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 || PCRE2_CODE_UNIT_WIDTH == 16 */
if (common->getucd != NULL)
{
set_jumps(common->getucd, LABEL());
do_getucd(common);
}
if (common->getucdtype != NULL)
{
set_jumps(common->getucdtype, LABEL());
do_getucdtype(common);
}
#endif /* SUPPORT_UNICODE */
SLJIT_FREE(common->private_data_ptrs, allocator_data);
if (common->has_then)
SLJIT_FREE(common->then_offsets, allocator_data);
executable_func = sljit_generate_code(compiler, 0, NULL);
executable_size = sljit_get_generated_code_size(compiler);
sljit_free_compiler(compiler);
if (executable_func == NULL)
{
PRIV(jit_free_rodata)(common->read_only_data_head, allocator_data);
return PCRE2_ERROR_NOMEMORY;
}
/* Reuse the function descriptor if possible. */
if (re->executable_jit != NULL)
functions = (executable_functions *)re->executable_jit;
else
{
functions = SLJIT_MALLOC(sizeof(executable_functions), allocator_data);
if (functions == NULL)
{
/* This case is highly unlikely since we just recently
freed a lot of memory. Not impossible though. */
sljit_free_code(executable_func, NULL);
PRIV(jit_free_rodata)(common->read_only_data_head, allocator_data);
return PCRE2_ERROR_NOMEMORY;
}
memset(functions, 0, sizeof(executable_functions));
functions->top_bracket = re->top_bracket + 1;
functions->limit_match = re->limit_match;
re->executable_jit = functions;
}
/* Turn mode into an index. */
if (mode == PCRE2_JIT_COMPLETE)
mode = 0;
else
mode = (mode == PCRE2_JIT_PARTIAL_SOFT) ? 1 : 2;
SLJIT_ASSERT(mode < JIT_NUMBER_OF_COMPILE_MODES);
functions->executable_funcs[mode] = executable_func;
functions->read_only_data_heads[mode] = common->read_only_data_head;
functions->executable_sizes[mode] = executable_size;
return 0;
}
#endif
/*************************************************
* JIT compile a Regular Expression *
*************************************************/
/* This function used JIT to convert a previously-compiled pattern into machine
code.
Arguments:
code a compiled pattern
options JIT option bits
Returns: 0: success or (*NOJIT) was used
<0: an error code
*/
#define PUBLIC_JIT_COMPILE_OPTIONS \
(PCRE2_JIT_COMPLETE|PCRE2_JIT_PARTIAL_SOFT|PCRE2_JIT_PARTIAL_HARD|PCRE2_JIT_INVALID_UTF)
PCRE2_EXP_DEFN int PCRE2_CALL_CONVENTION
pcre2_jit_compile(pcre2_code *code, uint32_t options)
{
pcre2_real_code *re = (pcre2_real_code *)code;
#ifdef SUPPORT_JIT
void *exec_memory;
executable_functions *functions;
static int executable_allocator_is_working = -1;
if (executable_allocator_is_working == -1)
{
/* Checks whether the executable allocator is working. This check
might run multiple times in multi-threaded environments, but the
result should not be affected by it. */
exec_memory = SLJIT_MALLOC_EXEC(32, NULL);
if (exec_memory != NULL)
{
SLJIT_FREE_EXEC(((sljit_u8*)(exec_memory)) + SLJIT_EXEC_OFFSET(exec_memory), NULL);
executable_allocator_is_working = 1;
}
else executable_allocator_is_working = 0;
}
#endif
if (options & PCRE2_JIT_TEST_ALLOC)
{
if (options != PCRE2_JIT_TEST_ALLOC)
return PCRE2_ERROR_JIT_BADOPTION;
#ifdef SUPPORT_JIT
return executable_allocator_is_working ? 0 : PCRE2_ERROR_NOMEMORY;
#else
return PCRE2_ERROR_JIT_UNSUPPORTED;
#endif
}
if (code == NULL)
return PCRE2_ERROR_NULL;
if ((options & ~PUBLIC_JIT_COMPILE_OPTIONS) != 0)
return PCRE2_ERROR_JIT_BADOPTION;
/* Support for invalid UTF was first introduced in JIT, with the option
PCRE2_JIT_INVALID_UTF. Later, support was added to the interpreter, and the
compile-time option PCRE2_MATCH_INVALID_UTF was created. This is now the
preferred feature, with the earlier option deprecated. However, for backward
compatibility, if the earlier option is set, it forces the new option so that
if JIT matching falls back to the interpreter, there is still support for
invalid UTF. However, if this function has already been successfully called
without PCRE2_JIT_INVALID_UTF and without PCRE2_MATCH_INVALID_UTF (meaning that
non-invalid-supporting JIT code was compiled), give an error.
If in the future support for PCRE2_JIT_INVALID_UTF is withdrawn, the following
actions are needed:
1. Remove the definition from pcre2.h.in and from the list in
PUBLIC_JIT_COMPILE_OPTIONS above.
2. Replace PCRE2_JIT_INVALID_UTF with a local flag in this module.
3. Replace PCRE2_JIT_INVALID_UTF in pcre2_jit_test.c.
4. Delete the following short block of code. The setting of "re" and
"functions" can be moved into the JIT-only block below, but if that is
done, (void)re and (void)functions will be needed in the non-JIT case, to
avoid compiler warnings.
*/
#ifdef SUPPORT_JIT
functions = (executable_functions *)re->executable_jit;
#endif
if ((options & PCRE2_JIT_INVALID_UTF) != 0)
{
if ((re->overall_options & PCRE2_MATCH_INVALID_UTF) == 0)
{
#ifdef SUPPORT_JIT
if (functions != NULL) return PCRE2_ERROR_JIT_BADOPTION;
#endif
re->overall_options |= PCRE2_MATCH_INVALID_UTF;
}
}
/* The above tests are run with and without JIT support. This means that
PCRE2_JIT_INVALID_UTF propagates back into the regex options (ensuring
interpreter support) even in the absence of JIT. But now, if there is no JIT
support, give an error return. */
#ifndef SUPPORT_JIT
return PCRE2_ERROR_JIT_BADOPTION;
#else /* SUPPORT_JIT */
/* There is JIT support. Do the necessary. */
if ((re->flags & PCRE2_NOJIT) != 0) return 0;
if (!executable_allocator_is_working)
return PCRE2_ERROR_NOMEMORY;
if ((re->overall_options & PCRE2_MATCH_INVALID_UTF) != 0)
options |= PCRE2_JIT_INVALID_UTF;
if ((options & PCRE2_JIT_COMPLETE) != 0 && (functions == NULL
|| functions->executable_funcs[0] == NULL)) {
uint32_t excluded_options = (PCRE2_JIT_PARTIAL_SOFT | PCRE2_JIT_PARTIAL_HARD);
int result = jit_compile(code, options & ~excluded_options);
if (result != 0)
return result;
}
if ((options & PCRE2_JIT_PARTIAL_SOFT) != 0 && (functions == NULL
|| functions->executable_funcs[1] == NULL)) {
uint32_t excluded_options = (PCRE2_JIT_COMPLETE | PCRE2_JIT_PARTIAL_HARD);
int result = jit_compile(code, options & ~excluded_options);
if (result != 0)
return result;
}
if ((options & PCRE2_JIT_PARTIAL_HARD) != 0 && (functions == NULL
|| functions->executable_funcs[2] == NULL)) {
uint32_t excluded_options = (PCRE2_JIT_COMPLETE | PCRE2_JIT_PARTIAL_SOFT);
int result = jit_compile(code, options & ~excluded_options);
if (result != 0)
return result;
}
return 0;
#endif /* SUPPORT_JIT */
}
/* JIT compiler uses an all-in-one approach. This improves security,
since the code generator functions are not exported. */
#define INCLUDED_FROM_PCRE2_JIT_COMPILE
#include "pcre2_jit_match.c"
#include "pcre2_jit_misc.c"
/* End of pcre2_jit_compile.c */
pcre2-10.45/src/pcre2_compile.h 0000644 0001751 0000166 00000031623 14750415257 011722 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE2 is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
#ifndef PCRE2_COMPILE_H_IDEMPOTENT_GUARD
#define PCRE2_COMPILE_H_IDEMPOTENT_GUARD
#include "pcre2_internal.h"
/* Compile time error code numbers. They are given names so that they can more
easily be tracked. When a new number is added, the tables called eint1 and
eint2 in pcre2posix.c may need to be updated, and a new error text must be
added to compile_error_texts in pcre2_error.c. Also, the error codes in
pcre2.h.in must be updated - their values are exactly 100 greater than these
values. */
enum { ERR0 = COMPILE_ERROR_BASE,
ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9, ERR10,
ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19, ERR20,
ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29, ERR30,
ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39, ERR40,
ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49, ERR50,
ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59, ERR60,
ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69, ERR70,
ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERR78, ERR79, ERR80,
ERR81, ERR82, ERR83, ERR84, ERR85, ERR86, ERR87, ERR88, ERR89, ERR90,
ERR91, ERR92, ERR93, ERR94, ERR95, ERR96, ERR97, ERR98, ERR99, ERR100,
ERR101,ERR102,ERR103,ERR104,ERR105,ERR106,ERR107,ERR108,ERR109,ERR110,
ERR111,ERR112,ERR113,ERR114,ERR115,ERR116 };
/* Code values for parsed patterns, which are stored in a vector of 32-bit
unsigned ints. Values less than META_END are literal data values. The coding
for identifying the item is in the top 16-bits, leaving 16 bits for the
additional data that some of them need. The META_CODE, META_DATA, and META_DIFF
macros are used to manipulate parsed pattern elements.
NOTE: When these definitions are changed, the table of extra lengths for each
code (meta_extra_lengths) must be updated to remain in step. */
#define META_END 0x80000000u /* End of pattern */
#define META_ALT 0x80010000u /* alternation */
#define META_ATOMIC 0x80020000u /* atomic group */
#define META_BACKREF 0x80030000u /* Back ref */
#define META_BACKREF_BYNAME 0x80040000u /* \k'name' */
#define META_BIGVALUE 0x80050000u /* Next is a literal > META_END */
#define META_CALLOUT_NUMBER 0x80060000u /* (?C with numerical argument */
#define META_CALLOUT_STRING 0x80070000u /* (?C with string argument */
#define META_CAPTURE 0x80080000u /* Capturing parenthesis */
#define META_CIRCUMFLEX 0x80090000u /* ^ metacharacter */
#define META_CLASS 0x800a0000u /* start non-empty class */
#define META_CLASS_EMPTY 0x800b0000u /* empty class */
#define META_CLASS_EMPTY_NOT 0x800c0000u /* negative empty class */
#define META_CLASS_END 0x800d0000u /* end of non-empty class */
#define META_CLASS_NOT 0x800e0000u /* start non-empty negative class */
#define META_COND_ASSERT 0x800f0000u /* (?(?assertion)... */
#define META_COND_DEFINE 0x80100000u /* (?(DEFINE)... */
#define META_COND_NAME 0x80110000u /* (?()... */
#define META_COND_NUMBER 0x80120000u /* (?(digits)... */
#define META_COND_RNAME 0x80130000u /* (?(R&name)... */
#define META_COND_RNUMBER 0x80140000u /* (?(Rdigits)... */
#define META_COND_VERSION 0x80150000u /* (?(VERSIONx.y)... */
#define META_OFFSET 0x80160000u /* Setting offset for various
META codes (e.g. META_SCS_NAME) */
#define META_SCS 0x80170000u /* (*scan_substring:... */
#define META_SCS_NAME 0x80180000u /* Next of scan_substring */
#define META_SCS_NUMBER 0x80190000u /* Next digits of scan_substring */
#define META_DOLLAR 0x801a0000u /* $ metacharacter */
#define META_DOT 0x801b0000u /* . metacharacter */
#define META_ESCAPE 0x801c0000u /* \d and friends */
#define META_KET 0x801d0000u /* closing parenthesis */
#define META_NOCAPTURE 0x801e0000u /* no capture parens */
#define META_OPTIONS 0x801f0000u /* (?i) and friends */
#define META_POSIX 0x80200000u /* POSIX class item */
#define META_POSIX_NEG 0x80210000u /* negative POSIX class item */
#define META_RANGE_ESCAPED 0x80220000u /* range with at least one escape */
#define META_RANGE_LITERAL 0x80230000u /* range defined literally */
#define META_RECURSE 0x80240000u /* Recursion */
#define META_RECURSE_BYNAME 0x80250000u /* (?&name) */
#define META_SCRIPT_RUN 0x80260000u /* (*script_run:...) */
/* These must be kept together to make it easy to check that an assertion
is present where expected in a conditional group. */
#define META_LOOKAHEAD 0x80270000u /* (?= */
#define META_LOOKAHEADNOT 0x80280000u /* (?! */
#define META_LOOKBEHIND 0x80290000u /* (?<= */
#define META_LOOKBEHINDNOT 0x802a0000u /* (?>16)
/* Extended class management flags. */
#define CLASS_IS_ECLASS 0x1
/* Macro for the highest character value. */
#if PCRE2_CODE_UNIT_WIDTH == 8
#define MAX_UCHAR_VALUE 0xffu
#elif PCRE2_CODE_UNIT_WIDTH == 16
#define MAX_UCHAR_VALUE 0xffffu
#else
#define MAX_UCHAR_VALUE 0xffffffffu
#endif
#define GET_MAX_CHAR_VALUE(utf) \
((utf) ? MAX_UTF_CODE_POINT : MAX_UCHAR_VALUE)
/* Macro for setting individual bits in class bitmaps. */
#define SETBIT(a,b) a[(b) >> 3] |= (uint8_t)(1u << ((b) & 0x7))
/* Macro for 8 bit specific checks. */
#if PCRE2_CODE_UNIT_WIDTH == 8
#define SELECT_VALUE8(value8, value) (value8)
#else
#define SELECT_VALUE8(value8, value) (value)
#endif
/* Macro for aligning data. */
#define CLIST_ALIGN_TO(base, align) \
((base + ((size_t)(align) - 1)) & ~((size_t)(align) - 1))
/* Structure for holding information about an OP_ECLASS internal operand.
An "operand" here could be just a single OP_[X]CLASS, or it could be some
complex expression; but it's some sequence of ECL_* codes which pushes one
value to the stack. */
typedef struct {
/* The position of the operand - or NULL if (lengthptr != NULL). */
PCRE2_UCHAR *code_start;
PCRE2_SIZE length;
/* The operand's type if it is a single code (ECL_XCLASS, ECL_ANY, ECL_NONE);
otherwise zero if the operand is not atomic. */
uint8_t op_single_type;
/* Regardless of whether it's a single code or not, we fully constant-fold
the bitmap for code points < 256. */
class_bits_storage bits;
} eclass_op_info;
/* Macros for the definitions below, to prevent name collisions. */
#define _pcre2_posix_class_maps PCRE2_SUFFIX(_pcre2_posix_class_maps)
#define _pcre2_update_classbits PCRE2_SUFFIX(_pcre2_update_classbits_)
#define _pcre2_compile_class_nested PCRE2_SUFFIX(_pcre2_compile_class_nested_)
#define _pcre2_compile_class_not_nested PCRE2_SUFFIX(_pcre2_compile_class_not_nested_)
/* Indices of the POSIX classes in posix_names, posix_name_lengths,
posix_class_maps, and posix_substitutes. They must be kept in sync. */
#define PC_DIGIT 7
#define PC_GRAPH 8
#define PC_PRINT 9
#define PC_PUNCT 10
#define PC_XDIGIT 13
extern const int PRIV(posix_class_maps)[];
/* Set bits in classbits according to the property type */
void PRIV(update_classbits)(uint32_t ptype, uint32_t pdata, BOOL negated,
uint8_t *classbits);
/* Compile the META codes from start_ptr...end_ptr, writing a single OP_CLASS
OP_CLASS, OP_NCLASS, OP_XCLASS, or OP_ALLANY into pcode. */
uint32_t *PRIV(compile_class_not_nested)(uint32_t options, uint32_t xoptions,
uint32_t *start_ptr, PCRE2_UCHAR **pcode, BOOL negate_class, BOOL* has_bitmap,
int *errorcodeptr, compile_block *cb, PCRE2_SIZE *lengthptr);
/* Compile the META codes in pptr into opcodes written to pcode. The pptr must
start at a META_CLASS or META_CLASS_NOT.
The pptr will be left pointing at the matching META_CLASS_END. */
BOOL PRIV(compile_class_nested)(uint32_t options, uint32_t xoptions,
uint32_t **pptr, PCRE2_UCHAR **pcode, int *errorcodeptr,
compile_block *cb, PCRE2_SIZE *lengthptr);
#endif /* PCRE2_COMPILE_H_IDEMPOTENT_GUARD */
/* End of pcre2_compile.h */
pcre2-10.45/src/pcre2_error.c 0000644 0001751 0000166 00000035016 14750415257 011416 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "pcre2_internal.h"
#define STRING(a) # a
#define XSTRING(s) STRING(s)
/* The texts of compile-time error messages. Compile-time error numbers start
at COMPILE_ERROR_BASE (100).
This used to be a table of strings, but in order to reduce the number of
relocations needed when a shared library is loaded dynamically, it is now one
long string. We cannot use a table of offsets, because the lengths of inserts
such as XSTRING(MAX_NAME_SIZE) are not known. Instead,
pcre2_get_error_message() counts through to the one it wants - this isn't a
performance issue because these strings are used only when there is an error.
Each substring ends with \0 to insert a null character. This includes the final
substring, so that the whole string ends with \0\0, which can be detected when
counting through. */
static const unsigned char compile_error_texts[] =
"no error\0"
"\\ at end of pattern\0"
"\\c at end of pattern\0"
"unrecognized character follows \\\0"
"numbers out of order in {} quantifier\0"
/* 5 */
"number too big in {} quantifier\0"
"missing terminating ] for character class\0"
"escape sequence is invalid in character class\0"
"range out of order in character class\0"
"quantifier does not follow a repeatable item\0"
/* 10 */
"internal error: unexpected repeat\0"
"unrecognized character after (? or (?-\0"
"POSIX named classes are supported only within a class\0"
"POSIX collating elements are not supported\0"
"missing closing parenthesis\0"
/* 15 */
"reference to non-existent subpattern\0"
"pattern passed as NULL with non-zero length\0"
"unrecognised compile-time option bit(s)\0"
"missing ) after (?# comment\0"
"parentheses are too deeply nested\0"
/* 20 */
"regular expression is too large\0"
"failed to allocate heap memory\0"
"unmatched closing parenthesis\0"
"internal error: code overflow\0"
"missing closing parenthesis for condition\0"
/* 25 */
"length of lookbehind assertion is not limited\0"
"a relative value of zero is not allowed\0"
"conditional subpattern contains more than two branches\0"
"atomic assertion expected after (?( or (?(?C)\0"
"digit expected after (?+ or (?-\0"
/* 30 */
"unknown POSIX class name\0"
"internal error in pcre2_study(): should not occur\0"
"this version of PCRE2 does not have Unicode support\0"
"parentheses are too deeply nested (stack check)\0"
"character code point value in \\x{} or \\o{} is too large\0"
/* 35 */
"lookbehind is too complicated\0"
"\\C is not allowed in a lookbehind assertion in UTF-" XSTRING(PCRE2_CODE_UNIT_WIDTH) " mode\0"
"PCRE2 does not support \\F, \\L, \\l, \\N{name}, \\U, or \\u\0"
"number after (?C is greater than 255\0"
"closing parenthesis for (?C expected\0"
/* 40 */
"invalid escape sequence in (*VERB) name\0"
"unrecognized character after (?P\0"
"syntax error in subpattern name (missing terminator?)\0"
"two named subpatterns have the same name (PCRE2_DUPNAMES not set)\0"
"subpattern name must start with a non-digit\0"
/* 45 */
"this version of PCRE2 does not have support for \\P, \\p, or \\X\0"
"malformed \\P or \\p sequence\0"
"unknown property after \\P or \\p\0"
"subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " code units)\0"
"too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
/* 50 */
"invalid range in character class\0"
"octal value is greater than \\377 in 8-bit non-UTF-8 mode\0"
"internal error: overran compiling workspace\0"
"internal error: previously-checked referenced subpattern not found\0"
"DEFINE subpattern contains more than one branch\0"
/* 55 */
"missing opening brace after \\o\0"
"internal error: unknown newline setting\0"
"\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"
"(?R (recursive pattern call) must be followed by a closing parenthesis\0"
/* "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0" */
"obsolete error (should not occur)\0" /* Was the above */
/* 60 */
"(*VERB) not recognized or malformed\0"
"subpattern number is too big\0"
"subpattern name expected\0"
"internal error: parsed pattern overflow\0"
"non-octal character in \\o{} (closing brace missing?)\0"
/* 65 */
"different names for subpatterns of the same number are not allowed\0"
"(*MARK) must have an argument\0"
"non-hex character in \\x{} (closing brace missing?)\0"
#ifndef EBCDIC
"\\c must be followed by a printable ASCII character\0"
#else
"\\c must be followed by a letter or one of [\\]^_?\0"
#endif
"\\k is not followed by a braced, angle-bracketed, or quoted name\0"
/* 70 */
"internal error: unknown meta code in check_lookbehinds()\0"
"\\N is not supported in a class\0"
"callout string is too long\0"
"disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
"using UTF is disabled by the application\0"
/* 75 */
"using UCP is disabled by the application\0"
"name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
"character code point value in \\u.... sequence is too large\0"
"digits missing after \\x or in \\x{} or \\o{} or \\N{U+}\0"
"syntax error or number too big in (?(VERSION condition\0"
/* 80 */
"internal error: unknown opcode in auto_possessify()\0"
"missing terminating delimiter for callout with string argument\0"
"unrecognized string delimiter follows (?C\0"
"using \\C is disabled by the application\0"
"(?| and/or (?J: or (?x: parentheses are too deeply nested\0"
/* 85 */
"using \\C is disabled in this PCRE2 library\0"
"regular expression is too complicated\0"
"lookbehind assertion is too long\0"
"pattern string is longer than the limit set by the application\0"
"internal error: unknown code in parsed pattern\0"
/* 90 */
"internal error: bad code value in parsed_skip()\0"
"PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES is not allowed in UTF-16 mode\0"
"invalid option bits with PCRE2_LITERAL\0"
"\\N{U+dddd} is supported only in Unicode (UTF) mode\0"
"invalid hyphen in option setting\0"
/* 95 */
"(*alpha_assertion) not recognized\0"
"script runs require Unicode support, which this version of PCRE2 does not have\0"
"too many capturing groups (maximum 65535)\0"
"octal digit missing after \\0 (PCRE2_EXTRA_NO_BS0 is set)\0"
"\\K is not allowed in lookarounds (but see PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK)\0"
/* 100 */
"branch too long in variable-length lookbehind assertion\0"
"compiled pattern would be longer than the limit set by the application\0"
"octal value given by \\ddd is greater than \\377 (forbidden by PCRE2_EXTRA_PYTHON_OCTAL)\0"
"using callouts is disabled by the application\0"
"PCRE2_EXTRA_TURKISH_CASING require Unicode (UTF or UCP) mode\0"
/* 105 */
"PCRE2_EXTRA_TURKISH_CASING requires UTF in 8-bit mode\0"
"PCRE2_EXTRA_TURKISH_CASING and PCRE2_EXTRA_CASELESS_RESTRICT are not compatible\0"
"extended character class nesting is too deep\0"
"invalid operator in extended character class\0"
"unexpected operator in extended character class (no preceding operand)\0"
/* 110 */
"expected operand after operator in extended character class\0"
"square brackets needed to clarify operator precedence in extended character class\0"
"missing terminating ] for extended character class (note '[' must be escaped under PCRE2_ALT_EXTENDED_CLASS)\0"
"unexpected expression in extended character class (no preceding operator)\0"
"empty expression in extended character class\0"
/* 115 */
"terminating ] with no following closing parenthesis in (?[...]\0"
"unexpected character in (?[...]) extended character class\0"
;
/* Match-time and UTF error texts are in the same format. */
static const unsigned char match_error_texts[] =
"no error\0"
"no match\0"
"partial match\0"
"UTF-8 error: 1 byte missing at end\0"
"UTF-8 error: 2 bytes missing at end\0"
/* 5 */
"UTF-8 error: 3 bytes missing at end\0"
"UTF-8 error: 4 bytes missing at end\0"
"UTF-8 error: 5 bytes missing at end\0"
"UTF-8 error: byte 2 top bits not 0x80\0"
"UTF-8 error: byte 3 top bits not 0x80\0"
/* 10 */
"UTF-8 error: byte 4 top bits not 0x80\0"
"UTF-8 error: byte 5 top bits not 0x80\0"
"UTF-8 error: byte 6 top bits not 0x80\0"
"UTF-8 error: 5-byte character is not allowed (RFC 3629)\0"
"UTF-8 error: 6-byte character is not allowed (RFC 3629)\0"
/* 15 */
"UTF-8 error: code points greater than 0x10ffff are not defined\0"
"UTF-8 error: code points 0xd800-0xdfff are not defined\0"
"UTF-8 error: overlong 2-byte sequence\0"
"UTF-8 error: overlong 3-byte sequence\0"
"UTF-8 error: overlong 4-byte sequence\0"
/* 20 */
"UTF-8 error: overlong 5-byte sequence\0"
"UTF-8 error: overlong 6-byte sequence\0"
"UTF-8 error: isolated byte with 0x80 bit set\0"
"UTF-8 error: illegal byte (0xfe or 0xff)\0"
"UTF-16 error: missing low surrogate at end\0"
/* 25 */
"UTF-16 error: invalid low surrogate\0"
"UTF-16 error: isolated low surrogate\0"
"UTF-32 error: code points 0xd800-0xdfff are not defined\0"
"UTF-32 error: code points greater than 0x10ffff are not defined\0"
"bad data value\0"
/* 30 */
"patterns do not all use the same character tables\0"
"magic number missing\0"
"pattern compiled in wrong mode: 8/16/32-bit error\0"
"bad offset value\0"
"bad option value\0"
/* 35 */
"invalid replacement string\0"
"bad offset into UTF string\0"
"callout error code\0" /* Never returned by PCRE2 itself */
"invalid data in workspace for DFA restart\0"
"too much recursion for DFA matching\0"
/* 40 */
"backreference condition or recursion test is not supported for DFA matching\0"
"function is not supported for DFA matching\0"
"pattern contains an item that is not supported for DFA matching\0"
"workspace size exceeded in DFA matching\0"
"internal error - pattern overwritten?\0"
/* 45 */
"bad JIT option\0"
"JIT stack limit reached\0"
"match limit exceeded\0"
"no more memory\0"
"unknown substring\0"
/* 50 */
"non-unique substring name\0"
"NULL argument passed with non-zero length\0"
"nested recursion at the same subject position\0"
"matching depth limit exceeded\0"
"requested value is not available\0"
/* 55 */
"requested value is not set\0"
"offset limit set without PCRE2_USE_OFFSET_LIMIT\0"
"bad escape sequence in replacement string\0"
"expected closing curly bracket in replacement string\0"
"bad substitution in replacement string\0"
/* 60 */
"match with end before start or start moved backwards is not supported\0"
"too many replacements (more than INT_MAX)\0"
"bad serialized data\0"
"heap limit exceeded\0"
"invalid syntax\0"
/* 65 */
"internal error - duplicate substitution match\0"
"PCRE2_MATCH_INVALID_UTF is not supported for DFA matching\0"
"INTERNAL ERROR: invalid substring offset\0"
"feature is not supported by the JIT compiler\0"
"error performing replacement case transformation\0"
/* 70 */
"replacement too large (longer than PCRE2_SIZE)\0"
;
/*************************************************
* Return error message *
*************************************************/
/* This function copies an error message into a buffer whose units are of an
appropriate width. Error numbers are positive for compile-time errors, and
negative for match-time errors (except for UTF errors), but the numbers are all
distinct.
Arguments:
enumber error number
buffer where to put the message (zero terminated)
size size of the buffer in code units
Returns: length of message if all is well
negative on error
*/
PCRE2_EXP_DEFN int PCRE2_CALL_CONVENTION
pcre2_get_error_message(int enumber, PCRE2_UCHAR *buffer, PCRE2_SIZE size)
{
const unsigned char *message;
PCRE2_SIZE i;
int n;
if (size == 0) return PCRE2_ERROR_NOMEMORY;
if (enumber >= COMPILE_ERROR_BASE) /* Compile error */
{
message = compile_error_texts;
n = enumber - COMPILE_ERROR_BASE;
}
else if (enumber < 0) /* Match or UTF error */
{
message = match_error_texts;
n = -enumber;
}
else /* Invalid error number */
{
message = (const unsigned char *)"\0"; /* Empty message list */
n = 1;
}
for (; n > 0; n--)
{
while (*message++ != CHAR_NUL) {};
if (*message == CHAR_NUL) return PCRE2_ERROR_BADDATA;
}
for (i = 0; *message != 0; i++)
{
if (i >= size - 1)
{
buffer[i] = 0; /* Terminate partial message */
return PCRE2_ERROR_NOMEMORY;
}
buffer[i] = *message++;
}
buffer[i] = 0;
return (int)i;
}
/* End of pcre2_error.c */
pcre2-10.45/src/pcre2_auto_possess.c 0000644 0001751 0000166 00000121652 14750415257 013016 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This module contains functions that scan a compiled pattern and change
repeats into possessive repeats where possible. */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "pcre2_internal.h"
/* This macro represents the max size of list[] and that is used to keep
track of UCD info in several places, it should be kept on sync with the
value used by GenerateUcd.py */
#define MAX_LIST 8
/*************************************************
* Tables for auto-possessification *
*************************************************/
/* This table is used to check whether auto-possessification is possible
between adjacent character-type opcodes. The left-hand (repeated) opcode is
used to select the row, and the right-hand opcode is use to select the column.
A value of 1 means that auto-possessification is OK. For example, the second
value in the first row means that \D+\d can be turned into \D++\d.
The Unicode property types (\P and \p) have to be present to fill out the table
because of what their opcode values are, but the table values should always be
zero because property types are handled separately in the code. The last four
columns apply to items that cannot be repeated, so there is no need to have
rows for them. Note that OP_DIGIT etc. are generated only when PCRE2_UCP is
*not* set. When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
#define APTROWS (LAST_AUTOTAB_LEFT_OP - FIRST_AUTOTAB_OP + 1)
#define APTCOLS (LAST_AUTOTAB_RIGHT_OP - FIRST_AUTOTAB_OP + 1)
static const uint8_t autoposstab[APTROWS][APTCOLS] = {
/* \D \d \S \s \W \w . .+ \C \P \p \R \H \h \V \v \X \Z \z $ $M */
{ 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \D */
{ 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 }, /* \d */
{ 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 }, /* \S */
{ 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \s */
{ 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \W */
{ 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 }, /* \w */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* . */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* .+ */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }, /* \C */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* \P */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* \p */
{ 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0 }, /* \R */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0 }, /* \H */
{ 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0 }, /* \h */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0 }, /* \V */
{ 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0 }, /* \v */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 } /* \X */
};
#ifdef SUPPORT_UNICODE
/* This table is used to check whether auto-possessification is possible
between adjacent Unicode property opcodes (OP_PROP and OP_NOTPROP). The
left-hand (repeated) opcode is used to select the row, and the right-hand
opcode is used to select the column. The values are as follows:
0 Always return FALSE (never auto-possessify)
1 Character groups are distinct (possessify if both are OP_PROP)
2 Check character categories in the same group (general or particular)
3 TRUE if the two opcodes are not the same (PROP vs NOTPROP)
4 Check left general category vs right particular category
5 Check right general category vs left particular category
6 Left alphanum vs right general category
7 Left space vs right general category
8 Left word vs right general category
9 Right alphanum vs left general category
10 Right space vs left general category
11 Right word vs left general category
12 Left alphanum vs right particular category
13 Left space vs right particular category
14 Left word vs right particular category
15 Right alphanum vs left particular category
16 Right space vs left particular category
17 Right word vs left particular category
*/
static const uint8_t propposstab[PT_TABSIZE][PT_TABSIZE] = {
/* LAMP GC PC SC SCX ALNUM SPACE PXSPACE WORD CLIST UCNC BIDICL BOOL */
{ 3, 0, 0, 0, 0, 3, 1, 1, 0, 0, 0, 0, 0 }, /* PT_LAMP */
{ 0, 2, 4, 0, 0, 9, 10, 10, 11, 0, 0, 0, 0 }, /* PT_GC */
{ 0, 5, 2, 0, 0, 15, 16, 16, 17, 0, 0, 0, 0 }, /* PT_PC */
{ 0, 0, 0, 2, 2, 0, 0, 0, 0, 0, 0, 0, 0 }, /* PT_SC */
{ 0, 0, 0, 2, 2, 0, 0, 0, 0, 0, 0, 0, 0 }, /* PT_SCX */
{ 3, 6, 12, 0, 0, 3, 1, 1, 0, 0, 0, 0, 0 }, /* PT_ALNUM */
{ 1, 7, 13, 0, 0, 1, 3, 3, 1, 0, 0, 0, 0 }, /* PT_SPACE */
{ 1, 7, 13, 0, 0, 1, 3, 3, 1, 0, 0, 0, 0 }, /* PT_PXSPACE */
{ 0, 8, 14, 0, 0, 0, 1, 1, 3, 0, 0, 0, 0 }, /* PT_WORD */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* PT_CLIST */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0, 0 }, /* PT_UCNC */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* PT_BIDICL */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } /* PT_BOOL */
/* PT_ANY does not need a record. */
};
/* This table is used to check whether auto-possessification is possible
between adjacent Unicode property opcodes (OP_PROP and OP_NOTPROP) when one
specifies a general category and the other specifies a particular category. The
row is selected by the general category and the column by the particular
category. The value is 1 if the particular category is not part of the general
category. */
static const uint8_t catposstab[7][30] = {
/* Cc Cf Cn Co Cs Ll Lm Lo Lt Lu Mc Me Mn Nd Nl No Pc Pd Pe Pf Pi Po Ps Sc Sk Sm So Zl Zp Zs */
{ 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* C */
{ 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* L */
{ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* M */
{ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 }, /* N */
{ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1 }, /* P */
{ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1 }, /* S */
{ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0 } /* Z */
};
/* This table is used when checking ALNUM, (PX)SPACE, SPACE, and WORD against
a general or particular category. The properties in each row are those
that apply to the character set in question. Duplication means that a little
unnecessary work is done when checking, but this keeps things much simpler
because they can all use the same code. For more details see the comment where
this table is used.
Note: SPACE and PXSPACE used to be different because Perl excluded VT from
"space", but from Perl 5.18 it's included, so both categories are treated the
same here. */
static const uint8_t posspropstab[3][4] = {
{ ucp_L, ucp_N, ucp_N, ucp_Nl }, /* ALNUM, 3rd and 4th values redundant */
{ ucp_Z, ucp_Z, ucp_C, ucp_Cc }, /* SPACE and PXSPACE, 2nd value redundant */
{ ucp_L, ucp_N, ucp_P, ucp_Po } /* WORD */
};
#endif /* SUPPORT_UNICODE */
#ifdef SUPPORT_UNICODE
/*************************************************
* Check a character and a property *
*************************************************/
/* This function is called by compare_opcodes() when a property item is
adjacent to a fixed character.
Arguments:
c the character
ptype the property type
pdata the data for the type
negated TRUE if it's a negated property (\P or \p{^)
Returns: TRUE if auto-possessifying is OK
*/
static BOOL
check_char_prop(uint32_t c, unsigned int ptype, unsigned int pdata,
BOOL negated)
{
BOOL ok, rc;
const uint32_t *p;
const ucd_record *prop = GET_UCD(c);
switch(ptype)
{
case PT_LAMP:
return (prop->chartype == ucp_Lu ||
prop->chartype == ucp_Ll ||
prop->chartype == ucp_Lt) == negated;
case PT_GC:
return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
case PT_PC:
return (pdata == prop->chartype) == negated;
case PT_SC:
return (pdata == prop->script) == negated;
case PT_SCX:
ok = (pdata == prop->script
|| MAPBIT(PRIV(ucd_script_sets) + UCD_SCRIPTX_PROP(prop), pdata) != 0);
return ok == negated;
/* These are specials */
case PT_ALNUM:
return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
/* Perl space used to exclude VT, but from Perl 5.18 it is included, which
means that Perl space and POSIX space are now identical. PCRE was changed
at release 8.34. */
case PT_SPACE: /* Perl space */
case PT_PXSPACE: /* POSIX space */
switch(c)
{
HSPACE_CASES:
VSPACE_CASES:
rc = negated;
break;
default:
rc = (PRIV(ucp_gentype)[prop->chartype] == ucp_Z) == negated;
}
return rc;
case PT_WORD:
return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
c == CHAR_UNDERSCORE) == negated;
case PT_CLIST:
p = PRIV(ucd_caseless_sets) + prop->caseset;
for (;;)
{
if (c < *p) return !negated;
if (c == *p++) return negated;
}
PCRE2_DEBUG_UNREACHABLE(); /* Control should never reach here */
break;
/* Haven't yet thought these through. */
case PT_BIDICL:
return FALSE;
case PT_BOOL:
return FALSE;
}
return FALSE;
}
#endif /* SUPPORT_UNICODE */
/*************************************************
* Base opcode of repeated opcodes *
*************************************************/
/* Returns the base opcode for repeated single character type opcodes. If the
opcode is not a repeated character type, it returns with the original value.
Arguments: c opcode
Returns: base opcode for the type
*/
static PCRE2_UCHAR
get_repeat_base(PCRE2_UCHAR c)
{
return (c > OP_TYPEPOSUPTO)? c :
(c >= OP_TYPESTAR)? OP_TYPESTAR :
(c >= OP_NOTSTARI)? OP_NOTSTARI :
(c >= OP_NOTSTAR)? OP_NOTSTAR :
(c >= OP_STARI)? OP_STARI :
OP_STAR;
}
/*************************************************
* Fill the character property list *
*************************************************/
/* Checks whether the code points to an opcode that can take part in auto-
possessification, and if so, fills a list with its properties.
Arguments:
code points to start of expression
utf TRUE if in UTF mode
ucp TRUE if in UCP mode
fcc points to the case-flipping table
list points to output list
list[0] will be filled with the opcode
list[1] will be non-zero if this opcode
can match an empty character string
list[2..7] depends on the opcode
Returns: points to the start of the next opcode if *code is accepted
NULL if *code is not accepted
*/
static PCRE2_SPTR
get_chr_property_list(PCRE2_SPTR code, BOOL utf, BOOL ucp, const uint8_t *fcc,
uint32_t *list)
{
PCRE2_UCHAR c = *code;
PCRE2_UCHAR base;
PCRE2_SPTR end;
PCRE2_SPTR class_end;
uint32_t chr;
#ifdef SUPPORT_UNICODE
uint32_t *clist_dest;
const uint32_t *clist_src;
#else
(void)utf; /* Suppress "unused parameter" compiler warnings */
(void)ucp;
#endif
list[0] = c;
list[1] = FALSE;
code++;
if (c >= OP_STAR && c <= OP_TYPEPOSUPTO)
{
base = get_repeat_base(c);
c -= (base - OP_STAR);
if (c == OP_UPTO || c == OP_MINUPTO || c == OP_EXACT || c == OP_POSUPTO)
code += IMM2_SIZE;
list[1] = (c != OP_PLUS && c != OP_MINPLUS && c != OP_EXACT &&
c != OP_POSPLUS);
switch(base)
{
case OP_STAR:
list[0] = OP_CHAR;
break;
case OP_STARI:
list[0] = OP_CHARI;
break;
case OP_NOTSTAR:
list[0] = OP_NOT;
break;
case OP_NOTSTARI:
list[0] = OP_NOTI;
break;
case OP_TYPESTAR:
list[0] = *code;
code++;
break;
}
c = list[0];
}
switch(c)
{
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
case OP_ALLANY:
case OP_ANYNL:
case OP_NOT_HSPACE:
case OP_HSPACE:
case OP_NOT_VSPACE:
case OP_VSPACE:
case OP_EXTUNI:
case OP_EODN:
case OP_EOD:
case OP_DOLL:
case OP_DOLLM:
return code;
case OP_CHAR:
case OP_NOT:
GETCHARINCTEST(chr, code);
list[2] = chr;
list[3] = NOTACHAR;
return code;
case OP_CHARI:
case OP_NOTI:
list[0] = (c == OP_CHARI) ? OP_CHAR : OP_NOT;
GETCHARINCTEST(chr, code);
list[2] = chr;
#ifdef SUPPORT_UNICODE
if (chr < 128 || (chr < 256 && !utf && !ucp))
list[3] = fcc[chr];
else
list[3] = UCD_OTHERCASE(chr);
#elif defined SUPPORT_WIDE_CHARS
list[3] = (chr < 256) ? fcc[chr] : chr;
#else
list[3] = fcc[chr];
#endif
/* The othercase might be the same value. */
if (chr == list[3])
list[3] = NOTACHAR;
else
list[4] = NOTACHAR;
return code;
#ifdef SUPPORT_UNICODE
case OP_PROP:
case OP_NOTPROP:
if (code[0] != PT_CLIST)
{
list[2] = code[0];
list[3] = code[1];
return code + 2;
}
/* Convert only if we have enough space. */
clist_src = PRIV(ucd_caseless_sets) + code[1];
clist_dest = list + 2;
code += 2;
do {
if (clist_dest >= list + MAX_LIST)
{
/* Early return if there is not enough space. GenerateUcd.py
generated a list with more than 5 characters and something
must be done about that going forward. */
PCRE2_DEBUG_UNREACHABLE(); /* Remove if it ever triggers */
list[2] = code[0];
list[3] = code[1];
return code;
}
*clist_dest++ = *clist_src;
}
while(*clist_src++ != NOTACHAR);
/* All characters are stored. The terminating NOTACHAR is copied from the
clist itself. */
list[0] = (c == OP_PROP) ? OP_CHAR : OP_NOT;
return code;
#endif
case OP_NCLASS:
case OP_CLASS:
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
case OP_ECLASS:
if (c == OP_XCLASS || c == OP_ECLASS)
end = code + GET(code, 0) - 1;
else
#endif
end = code + 32 / sizeof(PCRE2_UCHAR);
class_end = end;
switch(*end)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSQUERY:
list[1] = TRUE;
end++;
break;
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRPOSPLUS:
end++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
list[1] = (GET2(end, 1) == 0);
end += 1 + 2 * IMM2_SIZE;
break;
}
list[2] = (uint32_t)(end - code);
list[3] = (uint32_t)(end - class_end);
return end;
}
return NULL; /* Opcode not accepted */
}
/*************************************************
* Scan further character sets for match *
*************************************************/
/* Checks whether the base and the current opcode have a common character, in
which case the base cannot be possessified.
Arguments:
code points to the byte code
utf TRUE in UTF mode
ucp TRUE in UCP mode
cb compile data block
base_list the data list of the base opcode
base_end the end of the base opcode
rec_limit points to recursion depth counter
Returns: TRUE if the auto-possessification is possible
*/
static BOOL
compare_opcodes(PCRE2_SPTR code, BOOL utf, BOOL ucp, const compile_block *cb,
const uint32_t *base_list, PCRE2_SPTR base_end, int *rec_limit)
{
PCRE2_UCHAR c;
uint32_t list[MAX_LIST];
const uint32_t *chr_ptr;
const uint32_t *ochr_ptr;
const uint32_t *list_ptr;
PCRE2_SPTR next_code;
#ifdef SUPPORT_WIDE_CHARS
PCRE2_SPTR xclass_flags;
#endif
const uint8_t *class_bitset;
const uint8_t *set1, *set2, *set_end;
uint32_t chr;
BOOL accepted, invert_bits;
BOOL entered_a_group = FALSE;
if (--(*rec_limit) <= 0) return FALSE; /* Recursion has gone too deep */
/* Note: the base_list[1] contains whether the current opcode has a greedy
(represented by a non-zero value) quantifier. This is a different from
other character type lists, which store here that the character iterator
matches to an empty string (also represented by a non-zero value). */
for(;;)
{
PCRE2_SPTR bracode;
/* All operations move the code pointer forward.
Therefore infinite recursions are not possible. */
c = *code;
/* Skip over callouts */
if (c == OP_CALLOUT)
{
code += PRIV(OP_lengths)[c];
continue;
}
if (c == OP_CALLOUT_STR)
{
code += GET(code, 1 + 2*LINK_SIZE);
continue;
}
/* At the end of a branch, skip to the end of the group and process it. */
if (c == OP_ALT)
{
do code += GET(code, 1); while (*code == OP_ALT);
c = *code;
}
/* Inspect the next opcode. */
switch(c)
{
/* We can always possessify a greedy iterator at the end of the pattern,
which is reached after skipping over the final OP_KET. A non-greedy
iterator must never be possessified. */
case OP_END:
return base_list[1] != 0;
/* When an iterator is at the end of certain kinds of group we can inspect
what follows the group by skipping over the closing ket. Note that this
does not apply to OP_KETRMAX or OP_KETRMIN because what follows any given
iteration is variable (could be another iteration or could be the next
item). As these two opcodes are not listed in the next switch, they will
end up as the next code to inspect, and return FALSE by virtue of being
unsupported. */
case OP_KET:
case OP_KETRPOS:
/* The non-greedy case cannot be converted to a possessive form. */
if (base_list[1] == 0) return FALSE;
/* If the bracket is capturing it might be referenced by an OP_RECURSE
so its last iterator can never be possessified if the pattern contains
recursions. (This could be improved by keeping a list of group numbers that
are called by recursion.) */
bracode = code - GET(code, 1);
switch(*bracode)
{
case OP_CBRA:
case OP_SCBRA:
case OP_CBRAPOS:
case OP_SCBRAPOS:
if (cb->had_recurse) return FALSE;
break;
/* A script run might have to backtrack if the iterated item can match
characters from more than one script. So give up unless repeating an
explicit character. */
case OP_SCRIPT_RUN:
if (base_list[0] != OP_CHAR && base_list[0] != OP_CHARI)
return FALSE;
break;
/* Atomic sub-patterns and forward assertions can always auto-possessify
their last iterator. However, if the group was entered as a result of
checking a previous iterator, this is not possible. */
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ONCE:
return !entered_a_group;
/* Fixed-length lookbehinds can be treated the same way, but variable
length lookbehinds must not auto-possessify their last iterator. Note
that in order to identify a variable length lookbehind we must check
through all branches, because some may be of fixed length. */
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
do
{
if (bracode[1+LINK_SIZE] == OP_VREVERSE) return FALSE; /* Variable */
bracode += GET(bracode, 1);
}
while (*bracode == OP_ALT);
return !entered_a_group; /* Not variable length */
/* Non-atomic assertions - don't possessify last iterator. This needs
more thought. */
case OP_ASSERT_NA:
case OP_ASSERTBACK_NA:
return FALSE;
}
/* Skip over the bracket and inspect what comes next. */
code += PRIV(OP_lengths)[c];
continue;
/* Handle cases where the next item is a group. */
case OP_ONCE:
case OP_BRA:
case OP_CBRA:
next_code = code + GET(code, 1);
code += PRIV(OP_lengths)[c];
/* Check each branch. We have to recurse a level for all but the last
branch. */
while (*next_code == OP_ALT)
{
if (!compare_opcodes(code, utf, ucp, cb, base_list, base_end, rec_limit))
return FALSE;
code = next_code + 1 + LINK_SIZE;
next_code += GET(next_code, 1);
}
entered_a_group = TRUE;
continue;
case OP_BRAZERO:
case OP_BRAMINZERO:
next_code = code + 1;
if (*next_code != OP_BRA && *next_code != OP_CBRA &&
*next_code != OP_ONCE) return FALSE;
do next_code += GET(next_code, 1); while (*next_code == OP_ALT);
/* The bracket content will be checked by the OP_BRA/OP_CBRA case above. */
next_code += 1 + LINK_SIZE;
if (!compare_opcodes(next_code, utf, ucp, cb, base_list, base_end,
rec_limit))
return FALSE;
code += PRIV(OP_lengths)[c];
continue;
/* The next opcode does not need special handling; fall through and use it
to see if the base can be possessified. */
default:
break;
}
/* We now have the next appropriate opcode to compare with the base. Check
for a supported opcode, and load its properties. */
code = get_chr_property_list(code, utf, ucp, cb->fcc, list);
if (code == NULL) return FALSE; /* Unsupported */
/* If either opcode is a small character list, set pointers for comparing
characters from that list with another list, or with a property. */
if (base_list[0] == OP_CHAR)
{
chr_ptr = base_list + 2;
list_ptr = list;
}
else if (list[0] == OP_CHAR)
{
chr_ptr = list + 2;
list_ptr = base_list;
}
/* Character bitsets can also be compared to certain opcodes. */
else if (base_list[0] == OP_CLASS || list[0] == OP_CLASS
#if PCRE2_CODE_UNIT_WIDTH == 8
/* In 8 bit, non-UTF mode, OP_CLASS and OP_NCLASS are the same. */
|| (!utf && (base_list[0] == OP_NCLASS || list[0] == OP_NCLASS))
#endif
)
{
#if PCRE2_CODE_UNIT_WIDTH == 8
if (base_list[0] == OP_CLASS || (!utf && base_list[0] == OP_NCLASS))
#else
if (base_list[0] == OP_CLASS)
#endif
{
set1 = (const uint8_t *)(base_end - base_list[2]);
list_ptr = list;
}
else
{
set1 = (const uint8_t *)(code - list[2]);
list_ptr = base_list;
}
invert_bits = FALSE;
switch(list_ptr[0])
{
case OP_CLASS:
case OP_NCLASS:
set2 = (const uint8_t *)
((list_ptr == list ? code : base_end) - list_ptr[2]);
break;
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
xclass_flags = (list_ptr == list ? code : base_end) -
list_ptr[2] + LINK_SIZE;
if ((*xclass_flags & XCL_HASPROP) != 0) return FALSE;
if ((*xclass_flags & XCL_MAP) == 0)
{
/* No bits are set for characters < 256. */
if (list[1] == 0) return (*xclass_flags & XCL_NOT) == 0;
/* Might be an empty repeat. */
continue;
}
set2 = (const uint8_t *)(xclass_flags + 1);
break;
#endif
case OP_NOT_DIGIT:
invert_bits = TRUE;
/* Fall through */
case OP_DIGIT:
set2 = (const uint8_t *)(cb->cbits + cbit_digit);
break;
case OP_NOT_WHITESPACE:
invert_bits = TRUE;
/* Fall through */
case OP_WHITESPACE:
set2 = (const uint8_t *)(cb->cbits + cbit_space);
break;
case OP_NOT_WORDCHAR:
invert_bits = TRUE;
/* Fall through */
case OP_WORDCHAR:
set2 = (const uint8_t *)(cb->cbits + cbit_word);
break;
default:
return FALSE;
}
/* Because the bit sets are unaligned bytes, we need to perform byte
comparison here. */
set_end = set1 + 32;
if (invert_bits)
{
do
{
if ((*set1++ & ~(*set2++)) != 0) return FALSE;
}
while (set1 < set_end);
}
else
{
do
{
if ((*set1++ & *set2++) != 0) return FALSE;
}
while (set1 < set_end);
}
if (list[1] == 0) return TRUE;
/* Might be an empty repeat. */
continue;
}
/* Some property combinations also acceptable. Unicode property opcodes are
processed specially; the rest can be handled with a lookup table. */
else
{
uint32_t leftop, rightop;
leftop = base_list[0];
rightop = list[0];
#ifdef SUPPORT_UNICODE
accepted = FALSE; /* Always set in non-unicode case. */
if (leftop == OP_PROP || leftop == OP_NOTPROP)
{
if (rightop == OP_EOD)
accepted = TRUE;
else if (rightop == OP_PROP || rightop == OP_NOTPROP)
{
int n;
const uint8_t *p;
BOOL same = leftop == rightop;
BOOL lisprop = leftop == OP_PROP;
BOOL risprop = rightop == OP_PROP;
BOOL bothprop = lisprop && risprop;
/* There's a table that specifies how each combination is to be
processed:
0 Always return FALSE (never auto-possessify)
1 Character groups are distinct (possessify if both are OP_PROP)
2 Check character categories in the same group (general or particular)
3 Return TRUE if the two opcodes are not the same
... see comments below
*/
n = propposstab[base_list[2]][list[2]];
switch(n)
{
case 0: break;
case 1: accepted = bothprop; break;
case 2: accepted = (base_list[3] == list[3]) != same; break;
case 3: accepted = !same; break;
case 4: /* Left general category, right particular category */
accepted = risprop && catposstab[base_list[3]][list[3]] == same;
break;
case 5: /* Right general category, left particular category */
accepted = lisprop && catposstab[list[3]][base_list[3]] == same;
break;
/* This code is logically tricky. Think hard before fiddling with it.
The posspropstab table has four entries per row. Each row relates to
one of PCRE's special properties such as ALNUM or SPACE or WORD.
Only WORD actually needs all four entries, but using repeats for the
others means they can all use the same code below.
The first two entries in each row are Unicode general categories, and
apply always, because all the characters they include are part of the
PCRE character set. The third and fourth entries are a general and a
particular category, respectively, that include one or more relevant
characters. One or the other is used, depending on whether the check
is for a general or a particular category. However, in both cases the
category contains more characters than the specials that are defined
for the property being tested against. Therefore, it cannot be used
in a NOTPROP case.
Example: the row for WORD contains ucp_L, ucp_N, ucp_P, ucp_Po.
Underscore is covered by ucp_P or ucp_Po. */
case 6: /* Left alphanum vs right general category */
case 7: /* Left space vs right general category */
case 8: /* Left word vs right general category */
p = posspropstab[n-6];
accepted = risprop && lisprop ==
(list[3] != p[0] &&
list[3] != p[1] &&
(list[3] != p[2] || !lisprop));
break;
case 9: /* Right alphanum vs left general category */
case 10: /* Right space vs left general category */
case 11: /* Right word vs left general category */
p = posspropstab[n-9];
accepted = lisprop && risprop ==
(base_list[3] != p[0] &&
base_list[3] != p[1] &&
(base_list[3] != p[2] || !risprop));
break;
case 12: /* Left alphanum vs right particular category */
case 13: /* Left space vs right particular category */
case 14: /* Left word vs right particular category */
p = posspropstab[n-12];
accepted = risprop && lisprop ==
(catposstab[p[0]][list[3]] &&
catposstab[p[1]][list[3]] &&
(list[3] != p[3] || !lisprop));
break;
case 15: /* Right alphanum vs left particular category */
case 16: /* Right space vs left particular category */
case 17: /* Right word vs left particular category */
p = posspropstab[n-15];
accepted = lisprop && risprop ==
(catposstab[p[0]][base_list[3]] &&
catposstab[p[1]][base_list[3]] &&
(base_list[3] != p[3] || !risprop));
break;
}
}
}
else
#endif /* SUPPORT_UNICODE */
accepted = leftop >= FIRST_AUTOTAB_OP && leftop <= LAST_AUTOTAB_LEFT_OP &&
rightop >= FIRST_AUTOTAB_OP && rightop <= LAST_AUTOTAB_RIGHT_OP &&
autoposstab[leftop - FIRST_AUTOTAB_OP][rightop - FIRST_AUTOTAB_OP];
if (!accepted) return FALSE;
if (list[1] == 0) return TRUE;
/* Might be an empty repeat. */
continue;
}
/* Control reaches here only if one of the items is a small character list.
All characters are checked against the other side. */
do
{
chr = *chr_ptr;
switch(list_ptr[0])
{
case OP_CHAR:
ochr_ptr = list_ptr + 2;
do
{
if (chr == *ochr_ptr) return FALSE;
ochr_ptr++;
}
while(*ochr_ptr != NOTACHAR);
break;
case OP_NOT:
ochr_ptr = list_ptr + 2;
do
{
if (chr == *ochr_ptr)
break;
ochr_ptr++;
}
while(*ochr_ptr != NOTACHAR);
if (*ochr_ptr == NOTACHAR) return FALSE; /* Not found */
break;
/* Note that OP_DIGIT etc. are generated only when PCRE2_UCP is *not*
set. When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
case OP_DIGIT:
if (chr < 256 && (cb->ctypes[chr] & ctype_digit) != 0) return FALSE;
break;
case OP_NOT_DIGIT:
if (chr > 255 || (cb->ctypes[chr] & ctype_digit) == 0) return FALSE;
break;
case OP_WHITESPACE:
if (chr < 256 && (cb->ctypes[chr] & ctype_space) != 0) return FALSE;
break;
case OP_NOT_WHITESPACE:
if (chr > 255 || (cb->ctypes[chr] & ctype_space) == 0) return FALSE;
break;
case OP_WORDCHAR:
if (chr < 255 && (cb->ctypes[chr] & ctype_word) != 0) return FALSE;
break;
case OP_NOT_WORDCHAR:
if (chr > 255 || (cb->ctypes[chr] & ctype_word) == 0) return FALSE;
break;
case OP_HSPACE:
switch(chr)
{
HSPACE_CASES: return FALSE;
default: break;
}
break;
case OP_NOT_HSPACE:
switch(chr)
{
HSPACE_CASES: break;
default: return FALSE;
}
break;
case OP_ANYNL:
case OP_VSPACE:
switch(chr)
{
VSPACE_CASES: return FALSE;
default: break;
}
break;
case OP_NOT_VSPACE:
switch(chr)
{
VSPACE_CASES: break;
default: return FALSE;
}
break;
case OP_DOLL:
case OP_EODN:
switch (chr)
{
case CHAR_CR:
case CHAR_LF:
case CHAR_VT:
case CHAR_FF:
case CHAR_NEL:
#ifndef EBCDIC
case 0x2028:
case 0x2029:
#endif /* Not EBCDIC */
return FALSE;
}
break;
case OP_EOD: /* Can always possessify before \z */
break;
#ifdef SUPPORT_UNICODE
case OP_PROP:
case OP_NOTPROP:
if (!check_char_prop(chr, list_ptr[2], list_ptr[3],
list_ptr[0] == OP_NOTPROP))
return FALSE;
break;
#endif
case OP_NCLASS:
if (chr > 255) return FALSE;
/* Fall through */
case OP_CLASS:
if (chr > 255) break;
class_bitset = (const uint8_t *)
((list_ptr == list ? code : base_end) - list_ptr[2]);
if ((class_bitset[chr >> 3] & (1u << (chr & 7))) != 0) return FALSE;
break;
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
if (PRIV(xclass)(chr, (list_ptr == list ? code : base_end) -
list_ptr[2] + LINK_SIZE, (const uint8_t*)cb->start_code, utf))
return FALSE;
break;
case OP_ECLASS:
if (PRIV(eclass)(chr,
(list_ptr == list ? code : base_end) - list_ptr[2] + LINK_SIZE,
(list_ptr == list ? code : base_end) - list_ptr[3],
(const uint8_t*)cb->start_code, utf))
return FALSE;
break;
#endif /* SUPPORT_WIDE_CHARS */
default:
return FALSE;
}
chr_ptr++;
}
while(*chr_ptr != NOTACHAR);
/* At least one character must be matched from this opcode. */
if (list[1] == 0) return TRUE;
}
PCRE2_DEBUG_UNREACHABLE(); /* Control should never reach here */
return FALSE; /* Avoid compiler warnings */
}
/*************************************************
* Scan compiled regex for auto-possession *
*************************************************/
/* Replaces single character iterations with their possessive alternatives
if appropriate. This function modifies the compiled opcode! Hitting a
non-existent opcode may indicate a bug in PCRE2, but it can also be caused if a
bad UTF string was compiled with PCRE2_NO_UTF_CHECK. The rec_limit catches
overly complicated or large patterns. In these cases, the check just stops,
leaving the remainder of the pattern unpossessified.
Arguments:
code points to start of the byte code
cb compile data block
Returns: 0 for success
-1 if a non-existant opcode is encountered
*/
int
PRIV(auto_possessify)(PCRE2_UCHAR *code, const compile_block *cb)
{
PCRE2_UCHAR c;
PCRE2_SPTR end;
PCRE2_UCHAR *repeat_opcode;
uint32_t list[MAX_LIST];
int rec_limit = 1000; /* Was 10,000 but clang+ASAN uses a lot of stack. */
BOOL utf = (cb->external_options & PCRE2_UTF) != 0;
BOOL ucp = (cb->external_options & PCRE2_UCP) != 0;
for (;;)
{
c = *code;
if (c >= OP_TABLE_LENGTH)
{
PCRE2_DEBUG_UNREACHABLE();
return -1; /* Something gone wrong */
}
if (c >= OP_STAR && c <= OP_TYPEPOSUPTO)
{
c -= get_repeat_base(c) - OP_STAR;
end = (c <= OP_MINUPTO) ?
get_chr_property_list(code, utf, ucp, cb->fcc, list) : NULL;
list[1] = c == OP_STAR || c == OP_PLUS || c == OP_QUERY || c == OP_UPTO;
if (end != NULL && compare_opcodes(end, utf, ucp, cb, list, end,
&rec_limit))
{
switch(c)
{
case OP_STAR:
*code += OP_POSSTAR - OP_STAR;
break;
case OP_MINSTAR:
*code += OP_POSSTAR - OP_MINSTAR;
break;
case OP_PLUS:
*code += OP_POSPLUS - OP_PLUS;
break;
case OP_MINPLUS:
*code += OP_POSPLUS - OP_MINPLUS;
break;
case OP_QUERY:
*code += OP_POSQUERY - OP_QUERY;
break;
case OP_MINQUERY:
*code += OP_POSQUERY - OP_MINQUERY;
break;
case OP_UPTO:
*code += OP_POSUPTO - OP_UPTO;
break;
case OP_MINUPTO:
*code += OP_POSUPTO - OP_MINUPTO;
break;
}
}
c = *code;
}
else if (c == OP_CLASS || c == OP_NCLASS
#ifdef SUPPORT_WIDE_CHARS
|| c == OP_XCLASS || c == OP_ECLASS
#endif
)
{
#ifdef SUPPORT_WIDE_CHARS
if (c == OP_XCLASS || c == OP_ECLASS)
repeat_opcode = code + GET(code, 1);
else
#endif
repeat_opcode = code + 1 + (32 / sizeof(PCRE2_UCHAR));
c = *repeat_opcode;
if (c >= OP_CRSTAR && c <= OP_CRMINRANGE)
{
/* The return from get_chr_property_list() will never be NULL when
*code (aka c) is one of the four class opcodes. However, gcc with
-fanalyzer notes that a NULL return is possible, and grumbles. Hence we
put in a check. */
end = get_chr_property_list(code, utf, ucp, cb->fcc, list);
list[1] = (c & 1) == 0;
if (end != NULL &&
compare_opcodes(end, utf, ucp, cb, list, end, &rec_limit))
{
switch (c)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
*repeat_opcode = OP_CRPOSSTAR;
break;
case OP_CRPLUS:
case OP_CRMINPLUS:
*repeat_opcode = OP_CRPOSPLUS;
break;
case OP_CRQUERY:
case OP_CRMINQUERY:
*repeat_opcode = OP_CRPOSQUERY;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
*repeat_opcode = OP_CRPOSRANGE;
break;
}
}
}
c = *code;
}
switch(c)
{
case OP_END:
return 0;
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSPLUS:
case OP_TYPEPOSQUERY:
if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
break;
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEEXACT:
case OP_TYPEPOSUPTO:
if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
code += 2;
break;
case OP_CALLOUT_STR:
code += GET(code, 1 + 2*LINK_SIZE);
break;
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
case OP_ECLASS:
code += GET(code, 1);
break;
#endif
case OP_MARK:
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_SKIP_ARG:
case OP_THEN_ARG:
code += code[1];
break;
}
/* Add in the fixed length from the table */
code += PRIV(OP_lengths)[c];
/* In UTF-8 and UTF-16 modes, opcodes that are followed by a character may be
followed by a multi-byte character. The length in the table is a minimum, so
we have to arrange to skip the extra code units. */
#ifdef MAYBE_UTF_MULTI
if (utf) switch(c)
{
case OP_CHAR:
case OP_CHARI:
case OP_NOT:
case OP_NOTI:
case OP_STAR:
case OP_MINSTAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_QUERY:
case OP_MINQUERY:
case OP_UPTO:
case OP_MINUPTO:
case OP_EXACT:
case OP_POSSTAR:
case OP_POSPLUS:
case OP_POSQUERY:
case OP_POSUPTO:
case OP_STARI:
case OP_MINSTARI:
case OP_PLUSI:
case OP_MINPLUSI:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_UPTOI:
case OP_MINUPTOI:
case OP_EXACTI:
case OP_POSSTARI:
case OP_POSPLUSI:
case OP_POSQUERYI:
case OP_POSUPTOI:
case OP_NOTSTAR:
case OP_NOTMINSTAR:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
case OP_NOTUPTO:
case OP_NOTMINUPTO:
case OP_NOTEXACT:
case OP_NOTPOSSTAR:
case OP_NOTPOSPLUS:
case OP_NOTPOSQUERY:
case OP_NOTPOSUPTO:
case OP_NOTSTARI:
case OP_NOTMINSTARI:
case OP_NOTPLUSI:
case OP_NOTMINPLUSI:
case OP_NOTQUERYI:
case OP_NOTMINQUERYI:
case OP_NOTUPTOI:
case OP_NOTMINUPTOI:
case OP_NOTEXACTI:
case OP_NOTPOSSTARI:
case OP_NOTPOSPLUSI:
case OP_NOTPOSQUERYI:
case OP_NOTPOSUPTOI:
if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
break;
}
#else
(void)(utf); /* Keep compiler happy by referencing function argument */
#endif /* SUPPORT_WIDE_CHARS */
}
}
/* End of pcre2_auto_possess.c */
pcre2-10.45/src/pcre2_config.c 0000644 0001751 0000166 00000017076 14750415257 011540 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2020 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
/* Save the configured link size, which is in bytes. In 16-bit and 32-bit modes
its value gets changed by pcre2_intmodedep.h (included by pcre2_internal.h) to
be in code units. */
static int configured_link_size = LINK_SIZE;
#include "pcre2_internal.h"
/* These macros are the standard way of turning unquoted text into C strings.
They allow macros like PCRE2_MAJOR to be defined without quotes, which is
convenient for user programs that want to test their values. */
#define STRING(a) # a
#define XSTRING(s) STRING(s)
/*************************************************
* Return info about what features are configured *
*************************************************/
/* If where is NULL, the length of memory required is returned.
Arguments:
what what information is required
where where to put the information
Returns: 0 if a numerical value is returned
>= 0 if a string value
PCRE2_ERROR_BADOPTION if "where" not recognized
or JIT target requested when JIT not enabled
*/
PCRE2_EXP_DEFN int PCRE2_CALL_CONVENTION
pcre2_config(uint32_t what, void *where)
{
if (where == NULL) /* Requests a length */
{
switch(what)
{
default:
return PCRE2_ERROR_BADOPTION;
case PCRE2_CONFIG_BSR:
case PCRE2_CONFIG_COMPILED_WIDTHS:
case PCRE2_CONFIG_DEPTHLIMIT:
case PCRE2_CONFIG_HEAPLIMIT:
case PCRE2_CONFIG_JIT:
case PCRE2_CONFIG_LINKSIZE:
case PCRE2_CONFIG_MATCHLIMIT:
case PCRE2_CONFIG_NEVER_BACKSLASH_C:
case PCRE2_CONFIG_NEWLINE:
case PCRE2_CONFIG_PARENSLIMIT:
case PCRE2_CONFIG_STACKRECURSE: /* Obsolete */
case PCRE2_CONFIG_TABLES_LENGTH:
case PCRE2_CONFIG_UNICODE:
return sizeof(uint32_t);
/* These are handled below */
case PCRE2_CONFIG_JITTARGET:
case PCRE2_CONFIG_UNICODE_VERSION:
case PCRE2_CONFIG_VERSION:
break;
}
}
switch (what)
{
default:
return PCRE2_ERROR_BADOPTION;
case PCRE2_CONFIG_BSR:
#ifdef BSR_ANYCRLF
*((uint32_t *)where) = PCRE2_BSR_ANYCRLF;
#else
*((uint32_t *)where) = PCRE2_BSR_UNICODE;
#endif
break;
case PCRE2_CONFIG_COMPILED_WIDTHS:
*((uint32_t *)where) = 0
#ifdef SUPPORT_PCRE2_8
+ 1
#endif
#ifdef SUPPORT_PCRE2_16
+ 2
#endif
#ifdef SUPPORT_PCRE2_32
+ 4
#endif
;
break;
case PCRE2_CONFIG_DEPTHLIMIT:
*((uint32_t *)where) = MATCH_LIMIT_DEPTH;
break;
case PCRE2_CONFIG_HEAPLIMIT:
*((uint32_t *)where) = HEAP_LIMIT;
break;
case PCRE2_CONFIG_JIT:
#ifdef SUPPORT_JIT
*((uint32_t *)where) = 1;
#else
*((uint32_t *)where) = 0;
#endif
break;
case PCRE2_CONFIG_JITTARGET:
#ifdef SUPPORT_JIT
{
const char *v = PRIV(jit_get_target)();
return (int)(1 + ((where == NULL)?
strlen(v) : PRIV(strcpy_c8)((PCRE2_UCHAR *)where, v)));
}
#else
return PCRE2_ERROR_BADOPTION;
#endif
case PCRE2_CONFIG_LINKSIZE:
*((uint32_t *)where) = (uint32_t)configured_link_size;
break;
case PCRE2_CONFIG_MATCHLIMIT:
*((uint32_t *)where) = MATCH_LIMIT;
break;
case PCRE2_CONFIG_NEWLINE:
*((uint32_t *)where) = NEWLINE_DEFAULT;
break;
case PCRE2_CONFIG_NEVER_BACKSLASH_C:
#ifdef NEVER_BACKSLASH_C
*((uint32_t *)where) = 1;
#else
*((uint32_t *)where) = 0;
#endif
break;
case PCRE2_CONFIG_PARENSLIMIT:
*((uint32_t *)where) = PARENS_NEST_LIMIT;
break;
/* This is now obsolete. The stack is no longer used via recursion for
handling backtracking in pcre2_match(). */
case PCRE2_CONFIG_STACKRECURSE:
*((uint32_t *)where) = 0;
break;
case PCRE2_CONFIG_TABLES_LENGTH:
*((uint32_t *)where) = TABLES_LENGTH;
break;
case PCRE2_CONFIG_UNICODE_VERSION:
{
#if defined SUPPORT_UNICODE
const char *v = PRIV(unicode_version);
#else
const char *v = "Unicode not supported";
#endif
return (int)(1 + ((where == NULL)?
strlen(v) : PRIV(strcpy_c8)((PCRE2_UCHAR *)where, v)));
}
break;
case PCRE2_CONFIG_UNICODE:
#if defined SUPPORT_UNICODE
*((uint32_t *)where) = 1;
#else
*((uint32_t *)where) = 0;
#endif
break;
/* The hackery in setting "v" below is to cope with the case when
PCRE2_PRERELEASE is set to an empty string (which it is for real releases).
If the second alternative is used in this case, it does not leave a space
before the date. On the other hand, if all four macros are put into a single
XSTRING when PCRE2_PRERELEASE is not empty, an unwanted space is inserted.
There are problems using an "obvious" approach like this:
XSTRING(PCRE2_MAJOR) "." XSTRING(PCRE2_MINOR)
XSTRING(PCRE2_PRERELEASE) " " XSTRING(PCRE2_DATE)
because, when PCRE2_PRERELEASE is empty, this leads to an attempted expansion
of STRING(). The C standard states: "If (before argument substitution) any
argument consists of no preprocessing tokens, the behavior is undefined." It
turns out the gcc treats this case as a single empty string - which is what
we really want - but Visual C grumbles about the lack of an argument for the
macro. Unfortunately, both are within their rights. As there seems to be no
way to test for a macro's value being empty at compile time, we have to
resort to a runtime test. */
case PCRE2_CONFIG_VERSION:
{
const char *v = (XSTRING(Z PCRE2_PRERELEASE)[1] == 0)?
XSTRING(PCRE2_MAJOR.PCRE2_MINOR PCRE2_DATE) :
XSTRING(PCRE2_MAJOR.PCRE2_MINOR) XSTRING(PCRE2_PRERELEASE PCRE2_DATE);
return (int)(1 + ((where == NULL)?
strlen(v) : PRIV(strcpy_c8)((PCRE2_UCHAR *)where, v)));
}
}
return 0;
}
/* End of pcre2_config.c */
pcre2-10.45/src/pcre2.h.generic 0000644 0001751 0000166 00000144161 14750415257 011627 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* This is the public header file for the PCRE library, second API, to be
#included by applications that call PCRE2 functions.
Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
#ifndef PCRE2_H_IDEMPOTENT_GUARD
#define PCRE2_H_IDEMPOTENT_GUARD
/* The current PCRE version information. */
#define PCRE2_MAJOR 10
#define PCRE2_MINOR 45
#define PCRE2_PRERELEASE
#define PCRE2_DATE 2025-02-05
/* When an application links to a PCRE DLL in Windows, the symbols that are
imported have to be identified as such. When building PCRE2, the appropriate
export setting is defined in pcre2_internal.h, which includes this file. So we
don't change existing definitions of PCRE2_EXP_DECL. */
#if defined(_WIN32) && !defined(PCRE2_STATIC)
# ifndef PCRE2_EXP_DECL
# define PCRE2_EXP_DECL extern __declspec(dllimport)
# endif
#endif
/* By default, we use the standard "extern" declarations. */
#ifndef PCRE2_EXP_DECL
# ifdef __cplusplus
# define PCRE2_EXP_DECL extern "C"
# else
# define PCRE2_EXP_DECL extern
# endif
#endif
/* When compiling with the MSVC compiler, it is sometimes necessary to include
a "calling convention" before exported function names. (This is secondhand
information; I know nothing about MSVC myself). For example, something like
void __cdecl function(....)
might be needed. In order so make this easy, all the exported functions have
PCRE2_CALL_CONVENTION just before their names. It is rarely needed; if not
set, we ensure here that it has no effect. */
#ifndef PCRE2_CALL_CONVENTION
#define PCRE2_CALL_CONVENTION
#endif
/* Have to include limits.h, stdlib.h, and inttypes.h to ensure that size_t and
uint8_t, UCHAR_MAX, etc are defined. Some systems that do have inttypes.h do
not have stdint.h, which is why we use inttypes.h, which according to the C
standard is a superset of stdint.h. If inttypes.h is not available the build
will break and the relevant values must be provided by some other means. */
#include
#include
#include
/* Allow for C++ users compiling this directly. */
#ifdef __cplusplus
extern "C" {
#endif
/* The following option bits can be passed to pcre2_compile(), pcre2_match(),
or pcre2_dfa_match(). PCRE2_NO_UTF_CHECK affects only the function to which it
is passed. Put these bits at the most significant end of the options word so
others can be added next to them */
#define PCRE2_ANCHORED 0x80000000u
#define PCRE2_NO_UTF_CHECK 0x40000000u
#define PCRE2_ENDANCHORED 0x20000000u
/* The following option bits can be passed only to pcre2_compile(). However,
they may affect compilation, JIT compilation, and/or interpretive execution.
The following tags indicate which:
C alters what is compiled by pcre2_compile()
J alters what is compiled by pcre2_jit_compile()
M is inspected during pcre2_match() execution
D is inspected during pcre2_dfa_match() execution
*/
#define PCRE2_ALLOW_EMPTY_CLASS 0x00000001u /* C */
#define PCRE2_ALT_BSUX 0x00000002u /* C */
#define PCRE2_AUTO_CALLOUT 0x00000004u /* C */
#define PCRE2_CASELESS 0x00000008u /* C */
#define PCRE2_DOLLAR_ENDONLY 0x00000010u /* J M D */
#define PCRE2_DOTALL 0x00000020u /* C */
#define PCRE2_DUPNAMES 0x00000040u /* C */
#define PCRE2_EXTENDED 0x00000080u /* C */
#define PCRE2_FIRSTLINE 0x00000100u /* J M D */
#define PCRE2_MATCH_UNSET_BACKREF 0x00000200u /* C J M */
#define PCRE2_MULTILINE 0x00000400u /* C */
#define PCRE2_NEVER_UCP 0x00000800u /* C */
#define PCRE2_NEVER_UTF 0x00001000u /* C */
#define PCRE2_NO_AUTO_CAPTURE 0x00002000u /* C */
#define PCRE2_NO_AUTO_POSSESS 0x00004000u /* C */
#define PCRE2_NO_DOTSTAR_ANCHOR 0x00008000u /* C */
#define PCRE2_NO_START_OPTIMIZE 0x00010000u /* J M D */
#define PCRE2_UCP 0x00020000u /* C J M D */
#define PCRE2_UNGREEDY 0x00040000u /* C */
#define PCRE2_UTF 0x00080000u /* C J M D */
#define PCRE2_NEVER_BACKSLASH_C 0x00100000u /* C */
#define PCRE2_ALT_CIRCUMFLEX 0x00200000u /* J M D */
#define PCRE2_ALT_VERBNAMES 0x00400000u /* C */
#define PCRE2_USE_OFFSET_LIMIT 0x00800000u /* J M D */
#define PCRE2_EXTENDED_MORE 0x01000000u /* C */
#define PCRE2_LITERAL 0x02000000u /* C */
#define PCRE2_MATCH_INVALID_UTF 0x04000000u /* J M D */
#define PCRE2_ALT_EXTENDED_CLASS 0x08000000u /* C */
/* An additional compile options word is available in the compile context. */
#define PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES 0x00000001u /* C */
#define PCRE2_EXTRA_BAD_ESCAPE_IS_LITERAL 0x00000002u /* C */
#define PCRE2_EXTRA_MATCH_WORD 0x00000004u /* C */
#define PCRE2_EXTRA_MATCH_LINE 0x00000008u /* C */
#define PCRE2_EXTRA_ESCAPED_CR_IS_LF 0x00000010u /* C */
#define PCRE2_EXTRA_ALT_BSUX 0x00000020u /* C */
#define PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK 0x00000040u /* C */
#define PCRE2_EXTRA_CASELESS_RESTRICT 0x00000080u /* C */
#define PCRE2_EXTRA_ASCII_BSD 0x00000100u /* C */
#define PCRE2_EXTRA_ASCII_BSS 0x00000200u /* C */
#define PCRE2_EXTRA_ASCII_BSW 0x00000400u /* C */
#define PCRE2_EXTRA_ASCII_POSIX 0x00000800u /* C */
#define PCRE2_EXTRA_ASCII_DIGIT 0x00001000u /* C */
#define PCRE2_EXTRA_PYTHON_OCTAL 0x00002000u /* C */
#define PCRE2_EXTRA_NO_BS0 0x00004000u /* C */
#define PCRE2_EXTRA_NEVER_CALLOUT 0x00008000u /* C */
#define PCRE2_EXTRA_TURKISH_CASING 0x00010000u /* C */
/* These are for pcre2_jit_compile(). */
#define PCRE2_JIT_COMPLETE 0x00000001u /* For full matching */
#define PCRE2_JIT_PARTIAL_SOFT 0x00000002u
#define PCRE2_JIT_PARTIAL_HARD 0x00000004u
#define PCRE2_JIT_INVALID_UTF 0x00000100u
#define PCRE2_JIT_TEST_ALLOC 0x00000200u
/* These are for pcre2_match(), pcre2_dfa_match(), pcre2_jit_match(), and
pcre2_substitute(). Some are allowed only for one of the functions, and in
these cases it is noted below. Note that PCRE2_ANCHORED, PCRE2_ENDANCHORED and
PCRE2_NO_UTF_CHECK can also be passed to these functions (though
pcre2_jit_match() ignores the latter since it bypasses all sanity checks). */
#define PCRE2_NOTBOL 0x00000001u
#define PCRE2_NOTEOL 0x00000002u
#define PCRE2_NOTEMPTY 0x00000004u /* ) These two must be kept */
#define PCRE2_NOTEMPTY_ATSTART 0x00000008u /* ) adjacent to each other. */
#define PCRE2_PARTIAL_SOFT 0x00000010u
#define PCRE2_PARTIAL_HARD 0x00000020u
#define PCRE2_DFA_RESTART 0x00000040u /* pcre2_dfa_match() only */
#define PCRE2_DFA_SHORTEST 0x00000080u /* pcre2_dfa_match() only */
#define PCRE2_SUBSTITUTE_GLOBAL 0x00000100u /* pcre2_substitute() only */
#define PCRE2_SUBSTITUTE_EXTENDED 0x00000200u /* pcre2_substitute() only */
#define PCRE2_SUBSTITUTE_UNSET_EMPTY 0x00000400u /* pcre2_substitute() only */
#define PCRE2_SUBSTITUTE_UNKNOWN_UNSET 0x00000800u /* pcre2_substitute() only */
#define PCRE2_SUBSTITUTE_OVERFLOW_LENGTH 0x00001000u /* pcre2_substitute() only */
#define PCRE2_NO_JIT 0x00002000u /* not for pcre2_dfa_match() */
#define PCRE2_COPY_MATCHED_SUBJECT 0x00004000u
#define PCRE2_SUBSTITUTE_LITERAL 0x00008000u /* pcre2_substitute() only */
#define PCRE2_SUBSTITUTE_MATCHED 0x00010000u /* pcre2_substitute() only */
#define PCRE2_SUBSTITUTE_REPLACEMENT_ONLY 0x00020000u /* pcre2_substitute() only */
#define PCRE2_DISABLE_RECURSELOOP_CHECK 0x00040000u /* not for pcre2_dfa_match() or pcre2_jit_match() */
/* Options for pcre2_pattern_convert(). */
#define PCRE2_CONVERT_UTF 0x00000001u
#define PCRE2_CONVERT_NO_UTF_CHECK 0x00000002u
#define PCRE2_CONVERT_POSIX_BASIC 0x00000004u
#define PCRE2_CONVERT_POSIX_EXTENDED 0x00000008u
#define PCRE2_CONVERT_GLOB 0x00000010u
#define PCRE2_CONVERT_GLOB_NO_WILD_SEPARATOR 0x00000030u
#define PCRE2_CONVERT_GLOB_NO_STARSTAR 0x00000050u
/* Newline and \R settings, for use in compile contexts. The newline values
must be kept in step with values set in config.h and both sets must all be
greater than zero. */
#define PCRE2_NEWLINE_CR 1
#define PCRE2_NEWLINE_LF 2
#define PCRE2_NEWLINE_CRLF 3
#define PCRE2_NEWLINE_ANY 4
#define PCRE2_NEWLINE_ANYCRLF 5
#define PCRE2_NEWLINE_NUL 6
#define PCRE2_BSR_UNICODE 1
#define PCRE2_BSR_ANYCRLF 2
/* Error codes for pcre2_compile(). Some of these are also used by
pcre2_pattern_convert(). */
#define PCRE2_ERROR_END_BACKSLASH 101
#define PCRE2_ERROR_END_BACKSLASH_C 102
#define PCRE2_ERROR_UNKNOWN_ESCAPE 103
#define PCRE2_ERROR_QUANTIFIER_OUT_OF_ORDER 104
#define PCRE2_ERROR_QUANTIFIER_TOO_BIG 105
#define PCRE2_ERROR_MISSING_SQUARE_BRACKET 106
#define PCRE2_ERROR_ESCAPE_INVALID_IN_CLASS 107
#define PCRE2_ERROR_CLASS_RANGE_ORDER 108
#define PCRE2_ERROR_QUANTIFIER_INVALID 109
#define PCRE2_ERROR_INTERNAL_UNEXPECTED_REPEAT 110
#define PCRE2_ERROR_INVALID_AFTER_PARENS_QUERY 111
#define PCRE2_ERROR_POSIX_CLASS_NOT_IN_CLASS 112
#define PCRE2_ERROR_POSIX_NO_SUPPORT_COLLATING 113
#define PCRE2_ERROR_MISSING_CLOSING_PARENTHESIS 114
#define PCRE2_ERROR_BAD_SUBPATTERN_REFERENCE 115
#define PCRE2_ERROR_NULL_PATTERN 116
#define PCRE2_ERROR_BAD_OPTIONS 117
#define PCRE2_ERROR_MISSING_COMMENT_CLOSING 118
#define PCRE2_ERROR_PARENTHESES_NEST_TOO_DEEP 119
#define PCRE2_ERROR_PATTERN_TOO_LARGE 120
#define PCRE2_ERROR_HEAP_FAILED 121
#define PCRE2_ERROR_UNMATCHED_CLOSING_PARENTHESIS 122
#define PCRE2_ERROR_INTERNAL_CODE_OVERFLOW 123
#define PCRE2_ERROR_MISSING_CONDITION_CLOSING 124
#define PCRE2_ERROR_LOOKBEHIND_NOT_FIXED_LENGTH 125
#define PCRE2_ERROR_ZERO_RELATIVE_REFERENCE 126
#define PCRE2_ERROR_TOO_MANY_CONDITION_BRANCHES 127
#define PCRE2_ERROR_CONDITION_ASSERTION_EXPECTED 128
#define PCRE2_ERROR_BAD_RELATIVE_REFERENCE 129
#define PCRE2_ERROR_UNKNOWN_POSIX_CLASS 130
#define PCRE2_ERROR_INTERNAL_STUDY_ERROR 131
#define PCRE2_ERROR_UNICODE_NOT_SUPPORTED 132
#define PCRE2_ERROR_PARENTHESES_STACK_CHECK 133
#define PCRE2_ERROR_CODE_POINT_TOO_BIG 134
#define PCRE2_ERROR_LOOKBEHIND_TOO_COMPLICATED 135
#define PCRE2_ERROR_LOOKBEHIND_INVALID_BACKSLASH_C 136
#define PCRE2_ERROR_UNSUPPORTED_ESCAPE_SEQUENCE 137
#define PCRE2_ERROR_CALLOUT_NUMBER_TOO_BIG 138
#define PCRE2_ERROR_MISSING_CALLOUT_CLOSING 139
#define PCRE2_ERROR_ESCAPE_INVALID_IN_VERB 140
#define PCRE2_ERROR_UNRECOGNIZED_AFTER_QUERY_P 141
#define PCRE2_ERROR_MISSING_NAME_TERMINATOR 142
#define PCRE2_ERROR_DUPLICATE_SUBPATTERN_NAME 143
#define PCRE2_ERROR_INVALID_SUBPATTERN_NAME 144
#define PCRE2_ERROR_UNICODE_PROPERTIES_UNAVAILABLE 145
#define PCRE2_ERROR_MALFORMED_UNICODE_PROPERTY 146
#define PCRE2_ERROR_UNKNOWN_UNICODE_PROPERTY 147
#define PCRE2_ERROR_SUBPATTERN_NAME_TOO_LONG 148
#define PCRE2_ERROR_TOO_MANY_NAMED_SUBPATTERNS 149
#define PCRE2_ERROR_CLASS_INVALID_RANGE 150
#define PCRE2_ERROR_OCTAL_BYTE_TOO_BIG 151
#define PCRE2_ERROR_INTERNAL_OVERRAN_WORKSPACE 152
#define PCRE2_ERROR_INTERNAL_MISSING_SUBPATTERN 153
#define PCRE2_ERROR_DEFINE_TOO_MANY_BRANCHES 154
#define PCRE2_ERROR_BACKSLASH_O_MISSING_BRACE 155
#define PCRE2_ERROR_INTERNAL_UNKNOWN_NEWLINE 156
#define PCRE2_ERROR_BACKSLASH_G_SYNTAX 157
#define PCRE2_ERROR_PARENS_QUERY_R_MISSING_CLOSING 158
/* Error 159 is obsolete and should now never occur */
#define PCRE2_ERROR_VERB_ARGUMENT_NOT_ALLOWED 159
#define PCRE2_ERROR_VERB_UNKNOWN 160
#define PCRE2_ERROR_SUBPATTERN_NUMBER_TOO_BIG 161
#define PCRE2_ERROR_SUBPATTERN_NAME_EXPECTED 162
#define PCRE2_ERROR_INTERNAL_PARSED_OVERFLOW 163
#define PCRE2_ERROR_INVALID_OCTAL 164
#define PCRE2_ERROR_SUBPATTERN_NAMES_MISMATCH 165
#define PCRE2_ERROR_MARK_MISSING_ARGUMENT 166
#define PCRE2_ERROR_INVALID_HEXADECIMAL 167
#define PCRE2_ERROR_BACKSLASH_C_SYNTAX 168
#define PCRE2_ERROR_BACKSLASH_K_SYNTAX 169
#define PCRE2_ERROR_INTERNAL_BAD_CODE_LOOKBEHINDS 170
#define PCRE2_ERROR_BACKSLASH_N_IN_CLASS 171
#define PCRE2_ERROR_CALLOUT_STRING_TOO_LONG 172
#define PCRE2_ERROR_UNICODE_DISALLOWED_CODE_POINT 173
#define PCRE2_ERROR_UTF_IS_DISABLED 174
#define PCRE2_ERROR_UCP_IS_DISABLED 175
#define PCRE2_ERROR_VERB_NAME_TOO_LONG 176
#define PCRE2_ERROR_BACKSLASH_U_CODE_POINT_TOO_BIG 177
#define PCRE2_ERROR_MISSING_OCTAL_OR_HEX_DIGITS 178
#define PCRE2_ERROR_VERSION_CONDITION_SYNTAX 179
#define PCRE2_ERROR_INTERNAL_BAD_CODE_AUTO_POSSESS 180
#define PCRE2_ERROR_CALLOUT_NO_STRING_DELIMITER 181
#define PCRE2_ERROR_CALLOUT_BAD_STRING_DELIMITER 182
#define PCRE2_ERROR_BACKSLASH_C_CALLER_DISABLED 183
#define PCRE2_ERROR_QUERY_BARJX_NEST_TOO_DEEP 184
#define PCRE2_ERROR_BACKSLASH_C_LIBRARY_DISABLED 185
#define PCRE2_ERROR_PATTERN_TOO_COMPLICATED 186
#define PCRE2_ERROR_LOOKBEHIND_TOO_LONG 187
#define PCRE2_ERROR_PATTERN_STRING_TOO_LONG 188
#define PCRE2_ERROR_INTERNAL_BAD_CODE 189
#define PCRE2_ERROR_INTERNAL_BAD_CODE_IN_SKIP 190
#define PCRE2_ERROR_NO_SURROGATES_IN_UTF16 191
#define PCRE2_ERROR_BAD_LITERAL_OPTIONS 192
#define PCRE2_ERROR_SUPPORTED_ONLY_IN_UNICODE 193
#define PCRE2_ERROR_INVALID_HYPHEN_IN_OPTIONS 194
#define PCRE2_ERROR_ALPHA_ASSERTION_UNKNOWN 195
#define PCRE2_ERROR_SCRIPT_RUN_NOT_AVAILABLE 196
#define PCRE2_ERROR_TOO_MANY_CAPTURES 197
#define PCRE2_ERROR_MISSING_OCTAL_DIGIT 198
#define PCRE2_ERROR_BACKSLASH_K_IN_LOOKAROUND 199
#define PCRE2_ERROR_MAX_VAR_LOOKBEHIND_EXCEEDED 200
#define PCRE2_ERROR_PATTERN_COMPILED_SIZE_TOO_BIG 201
#define PCRE2_ERROR_OVERSIZE_PYTHON_OCTAL 202
#define PCRE2_ERROR_CALLOUT_CALLER_DISABLED 203
#define PCRE2_ERROR_EXTRA_CASING_REQUIRES_UNICODE 204
#define PCRE2_ERROR_TURKISH_CASING_REQUIRES_UTF 205
#define PCRE2_ERROR_EXTRA_CASING_INCOMPATIBLE 206
#define PCRE2_ERROR_ECLASS_NEST_TOO_DEEP 207
#define PCRE2_ERROR_ECLASS_INVALID_OPERATOR 208
#define PCRE2_ERROR_ECLASS_UNEXPECTED_OPERATOR 209
#define PCRE2_ERROR_ECLASS_EXPECTED_OPERAND 210
#define PCRE2_ERROR_ECLASS_MIXED_OPERATORS 211
#define PCRE2_ERROR_ECLASS_HINT_SQUARE_BRACKET 212
#define PCRE2_ERROR_PERL_ECLASS_UNEXPECTED_EXPR 213
#define PCRE2_ERROR_PERL_ECLASS_EMPTY_EXPR 214
#define PCRE2_ERROR_PERL_ECLASS_MISSING_CLOSE 215
#define PCRE2_ERROR_PERL_ECLASS_UNEXPECTED_CHAR 216
/* "Expected" matching error codes: no match and partial match. */
#define PCRE2_ERROR_NOMATCH (-1)
#define PCRE2_ERROR_PARTIAL (-2)
/* Error codes for UTF-8 validity checks */
#define PCRE2_ERROR_UTF8_ERR1 (-3)
#define PCRE2_ERROR_UTF8_ERR2 (-4)
#define PCRE2_ERROR_UTF8_ERR3 (-5)
#define PCRE2_ERROR_UTF8_ERR4 (-6)
#define PCRE2_ERROR_UTF8_ERR5 (-7)
#define PCRE2_ERROR_UTF8_ERR6 (-8)
#define PCRE2_ERROR_UTF8_ERR7 (-9)
#define PCRE2_ERROR_UTF8_ERR8 (-10)
#define PCRE2_ERROR_UTF8_ERR9 (-11)
#define PCRE2_ERROR_UTF8_ERR10 (-12)
#define PCRE2_ERROR_UTF8_ERR11 (-13)
#define PCRE2_ERROR_UTF8_ERR12 (-14)
#define PCRE2_ERROR_UTF8_ERR13 (-15)
#define PCRE2_ERROR_UTF8_ERR14 (-16)
#define PCRE2_ERROR_UTF8_ERR15 (-17)
#define PCRE2_ERROR_UTF8_ERR16 (-18)
#define PCRE2_ERROR_UTF8_ERR17 (-19)
#define PCRE2_ERROR_UTF8_ERR18 (-20)
#define PCRE2_ERROR_UTF8_ERR19 (-21)
#define PCRE2_ERROR_UTF8_ERR20 (-22)
#define PCRE2_ERROR_UTF8_ERR21 (-23)
/* Error codes for UTF-16 validity checks */
#define PCRE2_ERROR_UTF16_ERR1 (-24)
#define PCRE2_ERROR_UTF16_ERR2 (-25)
#define PCRE2_ERROR_UTF16_ERR3 (-26)
/* Error codes for UTF-32 validity checks */
#define PCRE2_ERROR_UTF32_ERR1 (-27)
#define PCRE2_ERROR_UTF32_ERR2 (-28)
/* Miscellaneous error codes for pcre2[_dfa]_match(), substring extraction
functions, context functions, and serializing functions. They are in numerical
order. Originally they were in alphabetical order too, but now that PCRE2 is
released, the numbers must not be changed. */
#define PCRE2_ERROR_BADDATA (-29)
#define PCRE2_ERROR_MIXEDTABLES (-30) /* Name was changed */
#define PCRE2_ERROR_BADMAGIC (-31)
#define PCRE2_ERROR_BADMODE (-32)
#define PCRE2_ERROR_BADOFFSET (-33)
#define PCRE2_ERROR_BADOPTION (-34)
#define PCRE2_ERROR_BADREPLACEMENT (-35)
#define PCRE2_ERROR_BADUTFOFFSET (-36)
#define PCRE2_ERROR_CALLOUT (-37) /* Never used by PCRE2 itself */
#define PCRE2_ERROR_DFA_BADRESTART (-38)
#define PCRE2_ERROR_DFA_RECURSE (-39)
#define PCRE2_ERROR_DFA_UCOND (-40)
#define PCRE2_ERROR_DFA_UFUNC (-41)
#define PCRE2_ERROR_DFA_UITEM (-42)
#define PCRE2_ERROR_DFA_WSSIZE (-43)
#define PCRE2_ERROR_INTERNAL (-44)
#define PCRE2_ERROR_JIT_BADOPTION (-45)
#define PCRE2_ERROR_JIT_STACKLIMIT (-46)
#define PCRE2_ERROR_MATCHLIMIT (-47)
#define PCRE2_ERROR_NOMEMORY (-48)
#define PCRE2_ERROR_NOSUBSTRING (-49)
#define PCRE2_ERROR_NOUNIQUESUBSTRING (-50)
#define PCRE2_ERROR_NULL (-51)
#define PCRE2_ERROR_RECURSELOOP (-52)
#define PCRE2_ERROR_DEPTHLIMIT (-53)
#define PCRE2_ERROR_RECURSIONLIMIT (-53) /* Obsolete synonym */
#define PCRE2_ERROR_UNAVAILABLE (-54)
#define PCRE2_ERROR_UNSET (-55)
#define PCRE2_ERROR_BADOFFSETLIMIT (-56)
#define PCRE2_ERROR_BADREPESCAPE (-57)
#define PCRE2_ERROR_REPMISSINGBRACE (-58)
#define PCRE2_ERROR_BADSUBSTITUTION (-59)
#define PCRE2_ERROR_BADSUBSPATTERN (-60)
#define PCRE2_ERROR_TOOMANYREPLACE (-61)
#define PCRE2_ERROR_BADSERIALIZEDDATA (-62)
#define PCRE2_ERROR_HEAPLIMIT (-63)
#define PCRE2_ERROR_CONVERT_SYNTAX (-64)
#define PCRE2_ERROR_INTERNAL_DUPMATCH (-65)
#define PCRE2_ERROR_DFA_UINVALID_UTF (-66)
#define PCRE2_ERROR_INVALIDOFFSET (-67)
#define PCRE2_ERROR_JIT_UNSUPPORTED (-68)
#define PCRE2_ERROR_REPLACECASE (-69)
#define PCRE2_ERROR_TOOLARGEREPLACE (-70)
/* Request types for pcre2_pattern_info() */
#define PCRE2_INFO_ALLOPTIONS 0
#define PCRE2_INFO_ARGOPTIONS 1
#define PCRE2_INFO_BACKREFMAX 2
#define PCRE2_INFO_BSR 3
#define PCRE2_INFO_CAPTURECOUNT 4
#define PCRE2_INFO_FIRSTCODEUNIT 5
#define PCRE2_INFO_FIRSTCODETYPE 6
#define PCRE2_INFO_FIRSTBITMAP 7
#define PCRE2_INFO_HASCRORLF 8
#define PCRE2_INFO_JCHANGED 9
#define PCRE2_INFO_JITSIZE 10
#define PCRE2_INFO_LASTCODEUNIT 11
#define PCRE2_INFO_LASTCODETYPE 12
#define PCRE2_INFO_MATCHEMPTY 13
#define PCRE2_INFO_MATCHLIMIT 14
#define PCRE2_INFO_MAXLOOKBEHIND 15
#define PCRE2_INFO_MINLENGTH 16
#define PCRE2_INFO_NAMECOUNT 17
#define PCRE2_INFO_NAMEENTRYSIZE 18
#define PCRE2_INFO_NAMETABLE 19
#define PCRE2_INFO_NEWLINE 20
#define PCRE2_INFO_DEPTHLIMIT 21
#define PCRE2_INFO_RECURSIONLIMIT 21 /* Obsolete synonym */
#define PCRE2_INFO_SIZE 22
#define PCRE2_INFO_HASBACKSLASHC 23
#define PCRE2_INFO_FRAMESIZE 24
#define PCRE2_INFO_HEAPLIMIT 25
#define PCRE2_INFO_EXTRAOPTIONS 26
/* Request types for pcre2_config(). */
#define PCRE2_CONFIG_BSR 0
#define PCRE2_CONFIG_JIT 1
#define PCRE2_CONFIG_JITTARGET 2
#define PCRE2_CONFIG_LINKSIZE 3
#define PCRE2_CONFIG_MATCHLIMIT 4
#define PCRE2_CONFIG_NEWLINE 5
#define PCRE2_CONFIG_PARENSLIMIT 6
#define PCRE2_CONFIG_DEPTHLIMIT 7
#define PCRE2_CONFIG_RECURSIONLIMIT 7 /* Obsolete synonym */
#define PCRE2_CONFIG_STACKRECURSE 8 /* Obsolete */
#define PCRE2_CONFIG_UNICODE 9
#define PCRE2_CONFIG_UNICODE_VERSION 10
#define PCRE2_CONFIG_VERSION 11
#define PCRE2_CONFIG_HEAPLIMIT 12
#define PCRE2_CONFIG_NEVER_BACKSLASH_C 13
#define PCRE2_CONFIG_COMPILED_WIDTHS 14
#define PCRE2_CONFIG_TABLES_LENGTH 15
/* Optimization directives for pcre2_set_optimize().
For binary compatibility, only add to this list; do not renumber. */
#define PCRE2_OPTIMIZATION_NONE 0
#define PCRE2_OPTIMIZATION_FULL 1
#define PCRE2_AUTO_POSSESS 64
#define PCRE2_AUTO_POSSESS_OFF 65
#define PCRE2_DOTSTAR_ANCHOR 66
#define PCRE2_DOTSTAR_ANCHOR_OFF 67
#define PCRE2_START_OPTIMIZE 68
#define PCRE2_START_OPTIMIZE_OFF 69
/* Types used in pcre2_set_substitute_case_callout().
PCRE2_SUBSTITUTE_CASE_LOWER and PCRE2_SUBSTITUTE_CASE_UPPER are passed to the
callout to indicate that the case of the entire callout input should be
case-transformed. PCRE2_SUBSTITUTE_CASE_TITLE_FIRST is passed to indicate that
only the first character or glyph should be transformed to Unicode titlecase,
and the rest to lowercase. */
#define PCRE2_SUBSTITUTE_CASE_LOWER 1
#define PCRE2_SUBSTITUTE_CASE_UPPER 2
#define PCRE2_SUBSTITUTE_CASE_TITLE_FIRST 3
/* Types for code units in patterns and subject strings. */
typedef uint8_t PCRE2_UCHAR8;
typedef uint16_t PCRE2_UCHAR16;
typedef uint32_t PCRE2_UCHAR32;
typedef const PCRE2_UCHAR8 *PCRE2_SPTR8;
typedef const PCRE2_UCHAR16 *PCRE2_SPTR16;
typedef const PCRE2_UCHAR32 *PCRE2_SPTR32;
/* The PCRE2_SIZE type is used for all string lengths and offsets in PCRE2,
including pattern offsets for errors and subject offsets after a match. We
define special values to indicate zero-terminated strings and unset offsets in
the offset vector (ovector). */
#define PCRE2_SIZE size_t
#define PCRE2_SIZE_MAX SIZE_MAX
#define PCRE2_ZERO_TERMINATED (~(PCRE2_SIZE)0)
#define PCRE2_UNSET (~(PCRE2_SIZE)0)
/* Generic types for opaque structures and JIT callback functions. These
declarations are defined in a macro that is expanded for each width later. */
#define PCRE2_TYPES_LIST \
struct pcre2_real_general_context; \
typedef struct pcre2_real_general_context pcre2_general_context; \
\
struct pcre2_real_compile_context; \
typedef struct pcre2_real_compile_context pcre2_compile_context; \
\
struct pcre2_real_match_context; \
typedef struct pcre2_real_match_context pcre2_match_context; \
\
struct pcre2_real_convert_context; \
typedef struct pcre2_real_convert_context pcre2_convert_context; \
\
struct pcre2_real_code; \
typedef struct pcre2_real_code pcre2_code; \
\
struct pcre2_real_match_data; \
typedef struct pcre2_real_match_data pcre2_match_data; \
\
struct pcre2_real_jit_stack; \
typedef struct pcre2_real_jit_stack pcre2_jit_stack; \
\
typedef pcre2_jit_stack *(*pcre2_jit_callback)(void *);
/* The structures for passing out data via callout functions. We use structures
so that new fields can be added on the end in future versions, without changing
the API of the function, thereby allowing old clients to work without
modification. Define the generic versions in a macro; the width-specific
versions are generated from this macro below. */
/* Flags for the callout_flags field. These are cleared after a callout. */
#define PCRE2_CALLOUT_STARTMATCH 0x00000001u /* Set for each bumpalong */
#define PCRE2_CALLOUT_BACKTRACK 0x00000002u /* Set after a backtrack */
#define PCRE2_STRUCTURE_LIST \
typedef struct pcre2_callout_block { \
uint32_t version; /* Identifies version of block */ \
/* ------------------------ Version 0 ------------------------------- */ \
uint32_t callout_number; /* Number compiled into pattern */ \
uint32_t capture_top; /* Max current capture */ \
uint32_t capture_last; /* Most recently closed capture */ \
PCRE2_SIZE *offset_vector; /* The offset vector */ \
PCRE2_SPTR mark; /* Pointer to current mark or NULL */ \
PCRE2_SPTR subject; /* The subject being matched */ \
PCRE2_SIZE subject_length; /* The length of the subject */ \
PCRE2_SIZE start_match; /* Offset to start of this match attempt */ \
PCRE2_SIZE current_position; /* Where we currently are in the subject */ \
PCRE2_SIZE pattern_position; /* Offset to next item in the pattern */ \
PCRE2_SIZE next_item_length; /* Length of next item in the pattern */ \
/* ------------------- Added for Version 1 -------------------------- */ \
PCRE2_SIZE callout_string_offset; /* Offset to string within pattern */ \
PCRE2_SIZE callout_string_length; /* Length of string compiled into pattern */ \
PCRE2_SPTR callout_string; /* String compiled into pattern */ \
/* ------------------- Added for Version 2 -------------------------- */ \
uint32_t callout_flags; /* See above for list */ \
/* ------------------------------------------------------------------ */ \
} pcre2_callout_block; \
\
typedef struct pcre2_callout_enumerate_block { \
uint32_t version; /* Identifies version of block */ \
/* ------------------------ Version 0 ------------------------------- */ \
PCRE2_SIZE pattern_position; /* Offset to next item in the pattern */ \
PCRE2_SIZE next_item_length; /* Length of next item in the pattern */ \
uint32_t callout_number; /* Number compiled into pattern */ \
PCRE2_SIZE callout_string_offset; /* Offset to string within pattern */ \
PCRE2_SIZE callout_string_length; /* Length of string compiled into pattern */ \
PCRE2_SPTR callout_string; /* String compiled into pattern */ \
/* ------------------------------------------------------------------ */ \
} pcre2_callout_enumerate_block; \
\
typedef struct pcre2_substitute_callout_block { \
uint32_t version; /* Identifies version of block */ \
/* ------------------------ Version 0 ------------------------------- */ \
PCRE2_SPTR input; /* Pointer to input subject string */ \
PCRE2_SPTR output; /* Pointer to output buffer */ \
PCRE2_SIZE output_offsets[2]; /* Changed portion of the output */ \
PCRE2_SIZE *ovector; /* Pointer to current ovector */ \
uint32_t oveccount; /* Count of pairs set in ovector */ \
uint32_t subscount; /* Substitution number */ \
/* ------------------------------------------------------------------ */ \
} pcre2_substitute_callout_block;
/* List the generic forms of all other functions in macros, which will be
expanded for each width below. Start with functions that give general
information. */
#define PCRE2_GENERAL_INFO_FUNCTIONS \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION pcre2_config(uint32_t, void *);
/* Functions for manipulating contexts. */
#define PCRE2_GENERAL_CONTEXT_FUNCTIONS \
PCRE2_EXP_DECL pcre2_general_context *PCRE2_CALL_CONVENTION \
pcre2_general_context_copy(pcre2_general_context *); \
PCRE2_EXP_DECL pcre2_general_context *PCRE2_CALL_CONVENTION \
pcre2_general_context_create(void *(*)(size_t, void *), \
void (*)(void *, void *), void *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_general_context_free(pcre2_general_context *);
#define PCRE2_COMPILE_CONTEXT_FUNCTIONS \
PCRE2_EXP_DECL pcre2_compile_context *PCRE2_CALL_CONVENTION \
pcre2_compile_context_copy(pcre2_compile_context *); \
PCRE2_EXP_DECL pcre2_compile_context *PCRE2_CALL_CONVENTION \
pcre2_compile_context_create(pcre2_general_context *);\
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_compile_context_free(pcre2_compile_context *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_bsr(pcre2_compile_context *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_character_tables(pcre2_compile_context *, const uint8_t *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_compile_extra_options(pcre2_compile_context *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_max_pattern_length(pcre2_compile_context *, PCRE2_SIZE); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_max_pattern_compiled_length(pcre2_compile_context *, PCRE2_SIZE); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_max_varlookbehind(pcre2_compile_context *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_newline(pcre2_compile_context *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_parens_nest_limit(pcre2_compile_context *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_compile_recursion_guard(pcre2_compile_context *, \
int (*)(uint32_t, void *), void *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_optimize(pcre2_compile_context *, uint32_t);
#define PCRE2_MATCH_CONTEXT_FUNCTIONS \
PCRE2_EXP_DECL pcre2_match_context *PCRE2_CALL_CONVENTION \
pcre2_match_context_copy(pcre2_match_context *); \
PCRE2_EXP_DECL pcre2_match_context *PCRE2_CALL_CONVENTION \
pcre2_match_context_create(pcre2_general_context *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_match_context_free(pcre2_match_context *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_callout(pcre2_match_context *, \
int (*)(pcre2_callout_block *, void *), void *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_substitute_callout(pcre2_match_context *, \
int (*)(pcre2_substitute_callout_block *, void *), void *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_substitute_case_callout(pcre2_match_context *, \
PCRE2_SIZE (*)(PCRE2_SPTR, PCRE2_SIZE, PCRE2_UCHAR *, PCRE2_SIZE, int, \
void *), \
void *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_depth_limit(pcre2_match_context *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_heap_limit(pcre2_match_context *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_match_limit(pcre2_match_context *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_offset_limit(pcre2_match_context *, PCRE2_SIZE); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_recursion_limit(pcre2_match_context *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_recursion_memory_management(pcre2_match_context *, \
void *(*)(size_t, void *), void (*)(void *, void *), void *);
#define PCRE2_CONVERT_CONTEXT_FUNCTIONS \
PCRE2_EXP_DECL pcre2_convert_context *PCRE2_CALL_CONVENTION \
pcre2_convert_context_copy(pcre2_convert_context *); \
PCRE2_EXP_DECL pcre2_convert_context *PCRE2_CALL_CONVENTION \
pcre2_convert_context_create(pcre2_general_context *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_convert_context_free(pcre2_convert_context *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_glob_escape(pcre2_convert_context *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_set_glob_separator(pcre2_convert_context *, uint32_t);
/* Functions concerned with compiling a pattern to PCRE internal code. */
#define PCRE2_COMPILE_FUNCTIONS \
PCRE2_EXP_DECL pcre2_code *PCRE2_CALL_CONVENTION \
pcre2_compile(PCRE2_SPTR, PCRE2_SIZE, uint32_t, int *, PCRE2_SIZE *, \
pcre2_compile_context *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_code_free(pcre2_code *); \
PCRE2_EXP_DECL pcre2_code *PCRE2_CALL_CONVENTION \
pcre2_code_copy(const pcre2_code *); \
PCRE2_EXP_DECL pcre2_code *PCRE2_CALL_CONVENTION \
pcre2_code_copy_with_tables(const pcre2_code *);
/* Functions that give information about a compiled pattern. */
#define PCRE2_PATTERN_INFO_FUNCTIONS \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_pattern_info(const pcre2_code *, uint32_t, void *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_callout_enumerate(const pcre2_code *, \
int (*)(pcre2_callout_enumerate_block *, void *), void *);
/* Functions for running a match and inspecting the result. */
#define PCRE2_MATCH_FUNCTIONS \
PCRE2_EXP_DECL pcre2_match_data *PCRE2_CALL_CONVENTION \
pcre2_match_data_create(uint32_t, pcre2_general_context *); \
PCRE2_EXP_DECL pcre2_match_data *PCRE2_CALL_CONVENTION \
pcre2_match_data_create_from_pattern(const pcre2_code *, \
pcre2_general_context *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_dfa_match(const pcre2_code *, PCRE2_SPTR, PCRE2_SIZE, PCRE2_SIZE, \
uint32_t, pcre2_match_data *, pcre2_match_context *, int *, PCRE2_SIZE); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_match(const pcre2_code *, PCRE2_SPTR, PCRE2_SIZE, PCRE2_SIZE, \
uint32_t, pcre2_match_data *, pcre2_match_context *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_match_data_free(pcre2_match_data *); \
PCRE2_EXP_DECL PCRE2_SPTR PCRE2_CALL_CONVENTION \
pcre2_get_mark(pcre2_match_data *); \
PCRE2_EXP_DECL PCRE2_SIZE PCRE2_CALL_CONVENTION \
pcre2_get_match_data_size(pcre2_match_data *); \
PCRE2_EXP_DECL PCRE2_SIZE PCRE2_CALL_CONVENTION \
pcre2_get_match_data_heapframes_size(pcre2_match_data *); \
PCRE2_EXP_DECL uint32_t PCRE2_CALL_CONVENTION \
pcre2_get_ovector_count(pcre2_match_data *); \
PCRE2_EXP_DECL PCRE2_SIZE *PCRE2_CALL_CONVENTION \
pcre2_get_ovector_pointer(pcre2_match_data *); \
PCRE2_EXP_DECL PCRE2_SIZE PCRE2_CALL_CONVENTION \
pcre2_get_startchar(pcre2_match_data *);
/* Convenience functions for handling matched substrings. */
#define PCRE2_SUBSTRING_FUNCTIONS \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_substring_copy_byname(pcre2_match_data *, PCRE2_SPTR, PCRE2_UCHAR *, \
PCRE2_SIZE *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_substring_copy_bynumber(pcre2_match_data *, uint32_t, PCRE2_UCHAR *, \
PCRE2_SIZE *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_substring_free(PCRE2_UCHAR *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_substring_get_byname(pcre2_match_data *, PCRE2_SPTR, PCRE2_UCHAR **, \
PCRE2_SIZE *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_substring_get_bynumber(pcre2_match_data *, uint32_t, PCRE2_UCHAR **, \
PCRE2_SIZE *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_substring_length_byname(pcre2_match_data *, PCRE2_SPTR, PCRE2_SIZE *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_substring_length_bynumber(pcre2_match_data *, uint32_t, PCRE2_SIZE *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_substring_nametable_scan(const pcre2_code *, PCRE2_SPTR, PCRE2_SPTR *, \
PCRE2_SPTR *); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_substring_number_from_name(const pcre2_code *, PCRE2_SPTR); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_substring_list_free(PCRE2_UCHAR **); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_substring_list_get(pcre2_match_data *, PCRE2_UCHAR ***, PCRE2_SIZE **);
/* Functions for serializing / deserializing compiled patterns. */
#define PCRE2_SERIALIZE_FUNCTIONS \
PCRE2_EXP_DECL int32_t PCRE2_CALL_CONVENTION \
pcre2_serialize_encode(const pcre2_code **, int32_t, uint8_t **, \
PCRE2_SIZE *, pcre2_general_context *); \
PCRE2_EXP_DECL int32_t PCRE2_CALL_CONVENTION \
pcre2_serialize_decode(pcre2_code **, int32_t, const uint8_t *, \
pcre2_general_context *); \
PCRE2_EXP_DECL int32_t PCRE2_CALL_CONVENTION \
pcre2_serialize_get_number_of_codes(const uint8_t *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_serialize_free(uint8_t *);
/* Convenience function for match + substitute. */
#define PCRE2_SUBSTITUTE_FUNCTION \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_substitute(const pcre2_code *, PCRE2_SPTR, PCRE2_SIZE, PCRE2_SIZE, \
uint32_t, pcre2_match_data *, pcre2_match_context *, PCRE2_SPTR, \
PCRE2_SIZE, PCRE2_UCHAR *, PCRE2_SIZE *);
/* Functions for converting pattern source strings. */
#define PCRE2_CONVERT_FUNCTIONS \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_pattern_convert(PCRE2_SPTR, PCRE2_SIZE, uint32_t, PCRE2_UCHAR **, \
PCRE2_SIZE *, pcre2_convert_context *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_converted_pattern_free(PCRE2_UCHAR *);
/* Functions for JIT processing */
#define PCRE2_JIT_FUNCTIONS \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_jit_compile(pcre2_code *, uint32_t); \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_jit_match(const pcre2_code *, PCRE2_SPTR, PCRE2_SIZE, PCRE2_SIZE, \
uint32_t, pcre2_match_data *, pcre2_match_context *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_jit_free_unused_memory(pcre2_general_context *); \
PCRE2_EXP_DECL pcre2_jit_stack *PCRE2_CALL_CONVENTION \
pcre2_jit_stack_create(size_t, size_t, pcre2_general_context *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_jit_stack_assign(pcre2_match_context *, pcre2_jit_callback, void *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_jit_stack_free(pcre2_jit_stack *);
/* Other miscellaneous functions. */
#define PCRE2_OTHER_FUNCTIONS \
PCRE2_EXP_DECL int PCRE2_CALL_CONVENTION \
pcre2_get_error_message(int, PCRE2_UCHAR *, PCRE2_SIZE); \
PCRE2_EXP_DECL const uint8_t *PCRE2_CALL_CONVENTION \
pcre2_maketables(pcre2_general_context *); \
PCRE2_EXP_DECL void PCRE2_CALL_CONVENTION \
pcre2_maketables_free(pcre2_general_context *, const uint8_t *);
/* Define macros that generate width-specific names from generic versions. The
three-level macro scheme is necessary to get the macros expanded when we want
them to be. First we get the width from PCRE2_LOCAL_WIDTH, which is used for
generating three versions of everything below. After that, PCRE2_SUFFIX will be
re-defined to use PCRE2_CODE_UNIT_WIDTH, for use when macros such as
pcre2_compile are called by application code. */
#define PCRE2_JOIN(a,b) a ## b
#define PCRE2_GLUE(a,b) PCRE2_JOIN(a,b)
#define PCRE2_SUFFIX(a) PCRE2_GLUE(a,PCRE2_LOCAL_WIDTH)
/* Data types */
#define PCRE2_UCHAR PCRE2_SUFFIX(PCRE2_UCHAR)
#define PCRE2_SPTR PCRE2_SUFFIX(PCRE2_SPTR)
#define pcre2_code PCRE2_SUFFIX(pcre2_code_)
#define pcre2_jit_callback PCRE2_SUFFIX(pcre2_jit_callback_)
#define pcre2_jit_stack PCRE2_SUFFIX(pcre2_jit_stack_)
#define pcre2_real_code PCRE2_SUFFIX(pcre2_real_code_)
#define pcre2_real_general_context PCRE2_SUFFIX(pcre2_real_general_context_)
#define pcre2_real_compile_context PCRE2_SUFFIX(pcre2_real_compile_context_)
#define pcre2_real_convert_context PCRE2_SUFFIX(pcre2_real_convert_context_)
#define pcre2_real_match_context PCRE2_SUFFIX(pcre2_real_match_context_)
#define pcre2_real_jit_stack PCRE2_SUFFIX(pcre2_real_jit_stack_)
#define pcre2_real_match_data PCRE2_SUFFIX(pcre2_real_match_data_)
/* Data blocks */
#define pcre2_callout_block PCRE2_SUFFIX(pcre2_callout_block_)
#define pcre2_callout_enumerate_block PCRE2_SUFFIX(pcre2_callout_enumerate_block_)
#define pcre2_substitute_callout_block PCRE2_SUFFIX(pcre2_substitute_callout_block_)
#define pcre2_general_context PCRE2_SUFFIX(pcre2_general_context_)
#define pcre2_compile_context PCRE2_SUFFIX(pcre2_compile_context_)
#define pcre2_convert_context PCRE2_SUFFIX(pcre2_convert_context_)
#define pcre2_match_context PCRE2_SUFFIX(pcre2_match_context_)
#define pcre2_match_data PCRE2_SUFFIX(pcre2_match_data_)
/* Functions: the complete list in alphabetical order */
#define pcre2_callout_enumerate PCRE2_SUFFIX(pcre2_callout_enumerate_)
#define pcre2_code_copy PCRE2_SUFFIX(pcre2_code_copy_)
#define pcre2_code_copy_with_tables PCRE2_SUFFIX(pcre2_code_copy_with_tables_)
#define pcre2_code_free PCRE2_SUFFIX(pcre2_code_free_)
#define pcre2_compile PCRE2_SUFFIX(pcre2_compile_)
#define pcre2_compile_context_copy PCRE2_SUFFIX(pcre2_compile_context_copy_)
#define pcre2_compile_context_create PCRE2_SUFFIX(pcre2_compile_context_create_)
#define pcre2_compile_context_free PCRE2_SUFFIX(pcre2_compile_context_free_)
#define pcre2_config PCRE2_SUFFIX(pcre2_config_)
#define pcre2_convert_context_copy PCRE2_SUFFIX(pcre2_convert_context_copy_)
#define pcre2_convert_context_create PCRE2_SUFFIX(pcre2_convert_context_create_)
#define pcre2_convert_context_free PCRE2_SUFFIX(pcre2_convert_context_free_)
#define pcre2_converted_pattern_free PCRE2_SUFFIX(pcre2_converted_pattern_free_)
#define pcre2_dfa_match PCRE2_SUFFIX(pcre2_dfa_match_)
#define pcre2_general_context_copy PCRE2_SUFFIX(pcre2_general_context_copy_)
#define pcre2_general_context_create PCRE2_SUFFIX(pcre2_general_context_create_)
#define pcre2_general_context_free PCRE2_SUFFIX(pcre2_general_context_free_)
#define pcre2_get_error_message PCRE2_SUFFIX(pcre2_get_error_message_)
#define pcre2_get_mark PCRE2_SUFFIX(pcre2_get_mark_)
#define pcre2_get_match_data_heapframes_size PCRE2_SUFFIX(pcre2_get_match_data_heapframes_size_)
#define pcre2_get_match_data_size PCRE2_SUFFIX(pcre2_get_match_data_size_)
#define pcre2_get_ovector_pointer PCRE2_SUFFIX(pcre2_get_ovector_pointer_)
#define pcre2_get_ovector_count PCRE2_SUFFIX(pcre2_get_ovector_count_)
#define pcre2_get_startchar PCRE2_SUFFIX(pcre2_get_startchar_)
#define pcre2_jit_compile PCRE2_SUFFIX(pcre2_jit_compile_)
#define pcre2_jit_match PCRE2_SUFFIX(pcre2_jit_match_)
#define pcre2_jit_free_unused_memory PCRE2_SUFFIX(pcre2_jit_free_unused_memory_)
#define pcre2_jit_stack_assign PCRE2_SUFFIX(pcre2_jit_stack_assign_)
#define pcre2_jit_stack_create PCRE2_SUFFIX(pcre2_jit_stack_create_)
#define pcre2_jit_stack_free PCRE2_SUFFIX(pcre2_jit_stack_free_)
#define pcre2_maketables PCRE2_SUFFIX(pcre2_maketables_)
#define pcre2_maketables_free PCRE2_SUFFIX(pcre2_maketables_free_)
#define pcre2_match PCRE2_SUFFIX(pcre2_match_)
#define pcre2_match_context_copy PCRE2_SUFFIX(pcre2_match_context_copy_)
#define pcre2_match_context_create PCRE2_SUFFIX(pcre2_match_context_create_)
#define pcre2_match_context_free PCRE2_SUFFIX(pcre2_match_context_free_)
#define pcre2_match_data_create PCRE2_SUFFIX(pcre2_match_data_create_)
#define pcre2_match_data_create_from_pattern PCRE2_SUFFIX(pcre2_match_data_create_from_pattern_)
#define pcre2_match_data_free PCRE2_SUFFIX(pcre2_match_data_free_)
#define pcre2_pattern_convert PCRE2_SUFFIX(pcre2_pattern_convert_)
#define pcre2_pattern_info PCRE2_SUFFIX(pcre2_pattern_info_)
#define pcre2_serialize_decode PCRE2_SUFFIX(pcre2_serialize_decode_)
#define pcre2_serialize_encode PCRE2_SUFFIX(pcre2_serialize_encode_)
#define pcre2_serialize_free PCRE2_SUFFIX(pcre2_serialize_free_)
#define pcre2_serialize_get_number_of_codes PCRE2_SUFFIX(pcre2_serialize_get_number_of_codes_)
#define pcre2_set_bsr PCRE2_SUFFIX(pcre2_set_bsr_)
#define pcre2_set_callout PCRE2_SUFFIX(pcre2_set_callout_)
#define pcre2_set_character_tables PCRE2_SUFFIX(pcre2_set_character_tables_)
#define pcre2_set_compile_extra_options PCRE2_SUFFIX(pcre2_set_compile_extra_options_)
#define pcre2_set_compile_recursion_guard PCRE2_SUFFIX(pcre2_set_compile_recursion_guard_)
#define pcre2_set_depth_limit PCRE2_SUFFIX(pcre2_set_depth_limit_)
#define pcre2_set_glob_escape PCRE2_SUFFIX(pcre2_set_glob_escape_)
#define pcre2_set_glob_separator PCRE2_SUFFIX(pcre2_set_glob_separator_)
#define pcre2_set_heap_limit PCRE2_SUFFIX(pcre2_set_heap_limit_)
#define pcre2_set_match_limit PCRE2_SUFFIX(pcre2_set_match_limit_)
#define pcre2_set_max_varlookbehind PCRE2_SUFFIX(pcre2_set_max_varlookbehind_)
#define pcre2_set_max_pattern_length PCRE2_SUFFIX(pcre2_set_max_pattern_length_)
#define pcre2_set_max_pattern_compiled_length PCRE2_SUFFIX(pcre2_set_max_pattern_compiled_length_)
#define pcre2_set_newline PCRE2_SUFFIX(pcre2_set_newline_)
#define pcre2_set_parens_nest_limit PCRE2_SUFFIX(pcre2_set_parens_nest_limit_)
#define pcre2_set_offset_limit PCRE2_SUFFIX(pcre2_set_offset_limit_)
#define pcre2_set_optimize PCRE2_SUFFIX(pcre2_set_optimize_)
#define pcre2_set_substitute_callout PCRE2_SUFFIX(pcre2_set_substitute_callout_)
#define pcre2_set_substitute_case_callout PCRE2_SUFFIX(pcre2_set_substitute_case_callout_)
#define pcre2_substitute PCRE2_SUFFIX(pcre2_substitute_)
#define pcre2_substring_copy_byname PCRE2_SUFFIX(pcre2_substring_copy_byname_)
#define pcre2_substring_copy_bynumber PCRE2_SUFFIX(pcre2_substring_copy_bynumber_)
#define pcre2_substring_free PCRE2_SUFFIX(pcre2_substring_free_)
#define pcre2_substring_get_byname PCRE2_SUFFIX(pcre2_substring_get_byname_)
#define pcre2_substring_get_bynumber PCRE2_SUFFIX(pcre2_substring_get_bynumber_)
#define pcre2_substring_length_byname PCRE2_SUFFIX(pcre2_substring_length_byname_)
#define pcre2_substring_length_bynumber PCRE2_SUFFIX(pcre2_substring_length_bynumber_)
#define pcre2_substring_list_get PCRE2_SUFFIX(pcre2_substring_list_get_)
#define pcre2_substring_list_free PCRE2_SUFFIX(pcre2_substring_list_free_)
#define pcre2_substring_nametable_scan PCRE2_SUFFIX(pcre2_substring_nametable_scan_)
#define pcre2_substring_number_from_name PCRE2_SUFFIX(pcre2_substring_number_from_name_)
/* Keep this old function name for backwards compatibility */
#define pcre2_set_recursion_limit PCRE2_SUFFIX(pcre2_set_recursion_limit_)
/* Keep this obsolete function for backwards compatibility: it is now a noop. */
#define pcre2_set_recursion_memory_management PCRE2_SUFFIX(pcre2_set_recursion_memory_management_)
/* Now generate all three sets of width-specific structures and function
prototypes. */
#define PCRE2_TYPES_STRUCTURES_AND_FUNCTIONS \
PCRE2_TYPES_LIST \
PCRE2_STRUCTURE_LIST \
PCRE2_GENERAL_INFO_FUNCTIONS \
PCRE2_GENERAL_CONTEXT_FUNCTIONS \
PCRE2_COMPILE_CONTEXT_FUNCTIONS \
PCRE2_CONVERT_CONTEXT_FUNCTIONS \
PCRE2_CONVERT_FUNCTIONS \
PCRE2_MATCH_CONTEXT_FUNCTIONS \
PCRE2_COMPILE_FUNCTIONS \
PCRE2_PATTERN_INFO_FUNCTIONS \
PCRE2_MATCH_FUNCTIONS \
PCRE2_SUBSTRING_FUNCTIONS \
PCRE2_SERIALIZE_FUNCTIONS \
PCRE2_SUBSTITUTE_FUNCTION \
PCRE2_JIT_FUNCTIONS \
PCRE2_OTHER_FUNCTIONS
#define PCRE2_LOCAL_WIDTH 8
PCRE2_TYPES_STRUCTURES_AND_FUNCTIONS
#undef PCRE2_LOCAL_WIDTH
#define PCRE2_LOCAL_WIDTH 16
PCRE2_TYPES_STRUCTURES_AND_FUNCTIONS
#undef PCRE2_LOCAL_WIDTH
#define PCRE2_LOCAL_WIDTH 32
PCRE2_TYPES_STRUCTURES_AND_FUNCTIONS
#undef PCRE2_LOCAL_WIDTH
/* Undefine the list macros; they are no longer needed. */
#undef PCRE2_TYPES_LIST
#undef PCRE2_STRUCTURE_LIST
#undef PCRE2_GENERAL_INFO_FUNCTIONS
#undef PCRE2_GENERAL_CONTEXT_FUNCTIONS
#undef PCRE2_COMPILE_CONTEXT_FUNCTIONS
#undef PCRE2_CONVERT_CONTEXT_FUNCTIONS
#undef PCRE2_MATCH_CONTEXT_FUNCTIONS
#undef PCRE2_COMPILE_FUNCTIONS
#undef PCRE2_PATTERN_INFO_FUNCTIONS
#undef PCRE2_MATCH_FUNCTIONS
#undef PCRE2_SUBSTRING_FUNCTIONS
#undef PCRE2_SERIALIZE_FUNCTIONS
#undef PCRE2_SUBSTITUTE_FUNCTION
#undef PCRE2_JIT_FUNCTIONS
#undef PCRE2_OTHER_FUNCTIONS
#undef PCRE2_TYPES_STRUCTURES_AND_FUNCTIONS
/* PCRE2_CODE_UNIT_WIDTH must be defined. If it is 8, 16, or 32, redefine
PCRE2_SUFFIX to use it. If it is 0, undefine the other macros and make
PCRE2_SUFFIX a no-op. Otherwise, generate an error. */
#undef PCRE2_SUFFIX
#ifndef PCRE2_CODE_UNIT_WIDTH
#error PCRE2_CODE_UNIT_WIDTH must be defined before including pcre2.h.
#error Use 8, 16, or 32; or 0 for a multi-width application.
#else /* PCRE2_CODE_UNIT_WIDTH is defined */
#if PCRE2_CODE_UNIT_WIDTH == 8 || \
PCRE2_CODE_UNIT_WIDTH == 16 || \
PCRE2_CODE_UNIT_WIDTH == 32
#define PCRE2_SUFFIX(a) PCRE2_GLUE(a, PCRE2_CODE_UNIT_WIDTH)
#elif PCRE2_CODE_UNIT_WIDTH == 0
#undef PCRE2_JOIN
#undef PCRE2_GLUE
#define PCRE2_SUFFIX(a) a
#else
#error PCRE2_CODE_UNIT_WIDTH must be 0, 8, 16, or 32.
#endif
#endif /* PCRE2_CODE_UNIT_WIDTH is defined */
#ifdef __cplusplus
} /* extern "C" */
#endif
#endif /* PCRE2_H_IDEMPOTENT_GUARD */
/* End of pcre2.h */
pcre2-10.45/src/pcre2_pattern_info.c 0000644 0001751 0000166 00000026760 14750415257 012763 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "pcre2_internal.h"
/*************************************************
* Return info about compiled pattern *
*************************************************/
/*
Arguments:
code points to compiled code
what what information is required
where where to put the information; if NULL, return length
Returns: 0 when data returned
> 0 when length requested
< 0 on error or unset value
*/
PCRE2_EXP_DEFN int PCRE2_CALL_CONVENTION
pcre2_pattern_info(const pcre2_code *code, uint32_t what, void *where)
{
const pcre2_real_code *re = (const pcre2_real_code *)code;
if (where == NULL) /* Requests field length */
{
switch(what)
{
case PCRE2_INFO_ALLOPTIONS:
case PCRE2_INFO_ARGOPTIONS:
case PCRE2_INFO_BACKREFMAX:
case PCRE2_INFO_BSR:
case PCRE2_INFO_CAPTURECOUNT:
case PCRE2_INFO_DEPTHLIMIT:
case PCRE2_INFO_EXTRAOPTIONS:
case PCRE2_INFO_FIRSTCODETYPE:
case PCRE2_INFO_FIRSTCODEUNIT:
case PCRE2_INFO_HASBACKSLASHC:
case PCRE2_INFO_HASCRORLF:
case PCRE2_INFO_HEAPLIMIT:
case PCRE2_INFO_JCHANGED:
case PCRE2_INFO_LASTCODETYPE:
case PCRE2_INFO_LASTCODEUNIT:
case PCRE2_INFO_MATCHEMPTY:
case PCRE2_INFO_MATCHLIMIT:
case PCRE2_INFO_MAXLOOKBEHIND:
case PCRE2_INFO_MINLENGTH:
case PCRE2_INFO_NAMEENTRYSIZE:
case PCRE2_INFO_NAMECOUNT:
case PCRE2_INFO_NEWLINE:
return sizeof(uint32_t);
case PCRE2_INFO_FIRSTBITMAP:
return sizeof(const uint8_t *);
case PCRE2_INFO_JITSIZE:
case PCRE2_INFO_SIZE:
case PCRE2_INFO_FRAMESIZE:
return sizeof(size_t);
case PCRE2_INFO_NAMETABLE:
return sizeof(PCRE2_SPTR);
}
}
if (re == NULL) return PCRE2_ERROR_NULL;
/* Check that the first field in the block is the magic number. If it is not,
return with PCRE2_ERROR_BADMAGIC. */
if (re->magic_number != MAGIC_NUMBER) return PCRE2_ERROR_BADMAGIC;
/* Check that this pattern was compiled in the correct bit mode */
if ((re->flags & (PCRE2_CODE_UNIT_WIDTH/8)) == 0) return PCRE2_ERROR_BADMODE;
switch(what)
{
case PCRE2_INFO_ALLOPTIONS:
*((uint32_t *)where) = re->overall_options;
break;
case PCRE2_INFO_ARGOPTIONS:
*((uint32_t *)where) = re->compile_options;
break;
case PCRE2_INFO_BACKREFMAX:
*((uint32_t *)where) = re->top_backref;
break;
case PCRE2_INFO_BSR:
*((uint32_t *)where) = re->bsr_convention;
break;
case PCRE2_INFO_CAPTURECOUNT:
*((uint32_t *)where) = re->top_bracket;
break;
case PCRE2_INFO_DEPTHLIMIT:
*((uint32_t *)where) = re->limit_depth;
if (re->limit_depth == UINT32_MAX) return PCRE2_ERROR_UNSET;
break;
case PCRE2_INFO_EXTRAOPTIONS:
*((uint32_t *)where) = re->extra_options;
break;
case PCRE2_INFO_FIRSTCODETYPE:
*((uint32_t *)where) = ((re->flags & PCRE2_FIRSTSET) != 0)? 1 :
((re->flags & PCRE2_STARTLINE) != 0)? 2 : 0;
break;
case PCRE2_INFO_FIRSTCODEUNIT:
*((uint32_t *)where) = ((re->flags & PCRE2_FIRSTSET) != 0)?
re->first_codeunit : 0;
break;
case PCRE2_INFO_FIRSTBITMAP:
*((const uint8_t **)where) = ((re->flags & PCRE2_FIRSTMAPSET) != 0)?
&(re->start_bitmap[0]) : NULL;
break;
case PCRE2_INFO_FRAMESIZE:
*((size_t *)where) = offsetof(heapframe, ovector) +
re->top_bracket * 2 * sizeof(PCRE2_SIZE);
break;
case PCRE2_INFO_HASBACKSLASHC:
*((uint32_t *)where) = (re->flags & PCRE2_HASBKC) != 0;
break;
case PCRE2_INFO_HASCRORLF:
*((uint32_t *)where) = (re->flags & PCRE2_HASCRORLF) != 0;
break;
case PCRE2_INFO_HEAPLIMIT:
*((uint32_t *)where) = re->limit_heap;
if (re->limit_heap == UINT32_MAX) return PCRE2_ERROR_UNSET;
break;
case PCRE2_INFO_JCHANGED:
*((uint32_t *)where) = (re->flags & PCRE2_JCHANGED) != 0;
break;
case PCRE2_INFO_JITSIZE:
#ifdef SUPPORT_JIT
*((size_t *)where) = (re->executable_jit != NULL)?
PRIV(jit_get_size)(re->executable_jit) : 0;
#else
*((size_t *)where) = 0;
#endif
break;
case PCRE2_INFO_LASTCODETYPE:
*((uint32_t *)where) = ((re->flags & PCRE2_LASTSET) != 0)? 1 : 0;
break;
case PCRE2_INFO_LASTCODEUNIT:
*((uint32_t *)where) = ((re->flags & PCRE2_LASTSET) != 0)?
re->last_codeunit : 0;
break;
case PCRE2_INFO_MATCHEMPTY:
*((uint32_t *)where) = (re->flags & PCRE2_MATCH_EMPTY) != 0;
break;
case PCRE2_INFO_MATCHLIMIT:
*((uint32_t *)where) = re->limit_match;
if (re->limit_match == UINT32_MAX) return PCRE2_ERROR_UNSET;
break;
case PCRE2_INFO_MAXLOOKBEHIND:
*((uint32_t *)where) = re->max_lookbehind;
break;
case PCRE2_INFO_MINLENGTH:
*((uint32_t *)where) = re->minlength;
break;
case PCRE2_INFO_NAMEENTRYSIZE:
*((uint32_t *)where) = re->name_entry_size;
break;
case PCRE2_INFO_NAMECOUNT:
*((uint32_t *)where) = re->name_count;
break;
case PCRE2_INFO_NAMETABLE:
*((PCRE2_SPTR *)where) = (PCRE2_SPTR)((const char *)re +
sizeof(pcre2_real_code));
break;
case PCRE2_INFO_NEWLINE:
*((uint32_t *)where) = re->newline_convention;
break;
case PCRE2_INFO_SIZE:
*((size_t *)where) = re->blocksize;
break;
default: return PCRE2_ERROR_BADOPTION;
}
return 0;
}
/*************************************************
* Callout enumerator *
*************************************************/
/*
Arguments:
code points to compiled code
callback function called for each callout block
callout_data user data passed to the callback
Returns: 0 when successfully completed
< 0 on local error
!= 0 for callback error
*/
PCRE2_EXP_DEFN int PCRE2_CALL_CONVENTION
pcre2_callout_enumerate(const pcre2_code *code,
int (*callback)(pcre2_callout_enumerate_block *, void *), void *callout_data)
{
const pcre2_real_code *re = (const pcre2_real_code *)code;
pcre2_callout_enumerate_block cb;
PCRE2_SPTR cc;
#ifdef SUPPORT_UNICODE
BOOL utf;
#endif
if (re == NULL) return PCRE2_ERROR_NULL;
#ifdef SUPPORT_UNICODE
utf = (re->overall_options & PCRE2_UTF) != 0;
#endif
/* Check that the first field in the block is the magic number. If it is not,
return with PCRE2_ERROR_BADMAGIC. */
if (re->magic_number != MAGIC_NUMBER) return PCRE2_ERROR_BADMAGIC;
/* Check that this pattern was compiled in the correct bit mode */
if ((re->flags & (PCRE2_CODE_UNIT_WIDTH/8)) == 0) return PCRE2_ERROR_BADMODE;
cb.version = 0;
cc = (PCRE2_SPTR)((const uint8_t *)re + sizeof(pcre2_real_code))
+ re->name_count * re->name_entry_size;
while (TRUE)
{
int rc;
switch (*cc)
{
case OP_END:
return 0;
case OP_CHAR:
case OP_CHARI:
case OP_NOT:
case OP_NOTI:
case OP_STAR:
case OP_MINSTAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_QUERY:
case OP_MINQUERY:
case OP_UPTO:
case OP_MINUPTO:
case OP_EXACT:
case OP_POSSTAR:
case OP_POSPLUS:
case OP_POSQUERY:
case OP_POSUPTO:
case OP_STARI:
case OP_MINSTARI:
case OP_PLUSI:
case OP_MINPLUSI:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_UPTOI:
case OP_MINUPTOI:
case OP_EXACTI:
case OP_POSSTARI:
case OP_POSPLUSI:
case OP_POSQUERYI:
case OP_POSUPTOI:
case OP_NOTSTAR:
case OP_NOTMINSTAR:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
case OP_NOTUPTO:
case OP_NOTMINUPTO:
case OP_NOTEXACT:
case OP_NOTPOSSTAR:
case OP_NOTPOSPLUS:
case OP_NOTPOSQUERY:
case OP_NOTPOSUPTO:
case OP_NOTSTARI:
case OP_NOTMINSTARI:
case OP_NOTPLUSI:
case OP_NOTMINPLUSI:
case OP_NOTQUERYI:
case OP_NOTMINQUERYI:
case OP_NOTUPTOI:
case OP_NOTMINUPTOI:
case OP_NOTEXACTI:
case OP_NOTPOSSTARI:
case OP_NOTPOSPLUSI:
case OP_NOTPOSQUERYI:
case OP_NOTPOSUPTOI:
cc += PRIV(OP_lengths)[*cc];
#ifdef SUPPORT_UNICODE
if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEEXACT:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSPLUS:
case OP_TYPEPOSQUERY:
case OP_TYPEPOSUPTO:
cc += PRIV(OP_lengths)[*cc];
#ifdef SUPPORT_UNICODE
if (cc[-1] == OP_PROP || cc[-1] == OP_NOTPROP) cc += 2;
#endif
break;
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
case OP_ECLASS:
cc += GET(cc, 1);
break;
#endif
case OP_MARK:
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_SKIP_ARG:
case OP_THEN_ARG:
cc += PRIV(OP_lengths)[*cc] + cc[1];
break;
case OP_CALLOUT:
cb.pattern_position = GET(cc, 1);
cb.next_item_length = GET(cc, 1 + LINK_SIZE);
cb.callout_number = cc[1 + 2*LINK_SIZE];
cb.callout_string_offset = 0;
cb.callout_string_length = 0;
cb.callout_string = NULL;
rc = callback(&cb, callout_data);
if (rc != 0) return rc;
cc += PRIV(OP_lengths)[*cc];
break;
case OP_CALLOUT_STR:
cb.pattern_position = GET(cc, 1);
cb.next_item_length = GET(cc, 1 + LINK_SIZE);
cb.callout_number = 0;
cb.callout_string_offset = GET(cc, 1 + 3*LINK_SIZE);
cb.callout_string_length =
GET(cc, 1 + 2*LINK_SIZE) - (1 + 4*LINK_SIZE) - 2;
cb.callout_string = cc + (1 + 4*LINK_SIZE) + 1;
rc = callback(&cb, callout_data);
if (rc != 0) return rc;
cc += GET(cc, 1 + 2*LINK_SIZE);
break;
default:
cc += PRIV(OP_lengths)[*cc];
break;
}
}
}
/* End of pcre2_pattern_info.c */
pcre2-10.45/src/pcre2_ord2utf.c 0000644 0001751 0000166 00000007236 14750415257 011655 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This file contains a function that converts a Unicode character code point
into a UTF string. The behaviour is different for each code unit width. */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "pcre2_internal.h"
/* If SUPPORT_UNICODE is not defined, this function will never be called.
Supply a dummy function because some compilers do not like empty source
modules. */
#ifndef SUPPORT_UNICODE
unsigned int
PRIV(ord2utf)(uint32_t cvalue, PCRE2_UCHAR *buffer)
{
(void)(cvalue);
(void)(buffer);
return 0;
}
#else /* SUPPORT_UNICODE */
/*************************************************
* Convert code point to UTF *
*************************************************/
/*
Arguments:
cvalue the character value
buffer pointer to buffer for result
Returns: number of code units placed in the buffer
*/
unsigned int
PRIV(ord2utf)(uint32_t cvalue, PCRE2_UCHAR *buffer)
{
/* Convert to UTF-8 */
#if PCRE2_CODE_UNIT_WIDTH == 8
int i, j;
for (i = 0; i < PRIV(utf8_table1_size); i++)
if ((int)cvalue <= PRIV(utf8_table1)[i]) break;
buffer += i;
for (j = i; j > 0; j--)
{
*buffer-- = 0x80 | (cvalue & 0x3f);
cvalue >>= 6;
}
*buffer = PRIV(utf8_table2)[i] | cvalue;
return i + 1;
/* Convert to UTF-16 */
#elif PCRE2_CODE_UNIT_WIDTH == 16
if (cvalue <= 0xffff)
{
*buffer = (PCRE2_UCHAR)cvalue;
return 1;
}
cvalue -= 0x10000;
*buffer++ = 0xd800 | (cvalue >> 10);
*buffer = 0xdc00 | (cvalue & 0x3ff);
return 2;
/* Convert to UTF-32 */
#else
*buffer = (PCRE2_UCHAR)cvalue;
return 1;
#endif
}
#endif /* SUPPORT_UNICODE */
/* End of pcre2_ord2utf.c */
pcre2-10.45/src/pcre2_newline.c 0000644 0001751 0000166 00000014324 14750415257 011725 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This module contains internal functions for testing newlines when more than
one kind of newline is to be recognized. When a newline is found, its length is
returned. In principle, we could implement several newline "types", each
referring to a different set of newline characters. At present, PCRE2 supports
only NLTYPE_FIXED, which gets handled without these functions, NLTYPE_ANYCRLF,
and NLTYPE_ANY. The full list of Unicode newline characters is taken from
http://unicode.org/unicode/reports/tr18/. */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "pcre2_internal.h"
/*************************************************
* Check for newline at given position *
*************************************************/
/* This function is called only via the IS_NEWLINE macro, which does so only
when the newline type is NLTYPE_ANY or NLTYPE_ANYCRLF. The case of a fixed
newline (NLTYPE_FIXED) is handled inline. It is guaranteed that the code unit
pointed to by ptr is less than the end of the string.
Arguments:
ptr pointer to possible newline
type the newline type
endptr pointer to the end of the string
lenptr where to return the length
utf TRUE if in utf mode
Returns: TRUE or FALSE
*/
BOOL
PRIV(is_newline)(PCRE2_SPTR ptr, uint32_t type, PCRE2_SPTR endptr,
uint32_t *lenptr, BOOL utf)
{
uint32_t c;
#ifdef SUPPORT_UNICODE
if (utf) { GETCHAR(c, ptr); } else c = *ptr;
#else
(void)utf;
c = *ptr;
#endif /* SUPPORT_UNICODE */
if (type == NLTYPE_ANYCRLF) switch(c)
{
case CHAR_LF:
*lenptr = 1;
return TRUE;
case CHAR_CR:
*lenptr = (ptr < endptr - 1 && ptr[1] == CHAR_LF)? 2 : 1;
return TRUE;
default:
return FALSE;
}
/* NLTYPE_ANY */
else switch(c)
{
#ifdef EBCDIC
case CHAR_NEL:
#endif
case CHAR_LF:
case CHAR_VT:
case CHAR_FF:
*lenptr = 1;
return TRUE;
case CHAR_CR:
*lenptr = (ptr < endptr - 1 && ptr[1] == CHAR_LF)? 2 : 1;
return TRUE;
#ifndef EBCDIC
#if PCRE2_CODE_UNIT_WIDTH == 8
case CHAR_NEL:
*lenptr = utf? 2 : 1;
return TRUE;
case 0x2028: /* LS */
case 0x2029: /* PS */
*lenptr = 3;
return TRUE;
#else /* 16-bit or 32-bit code units */
case CHAR_NEL:
case 0x2028: /* LS */
case 0x2029: /* PS */
*lenptr = 1;
return TRUE;
#endif
#endif /* Not EBCDIC */
default:
return FALSE;
}
}
/*************************************************
* Check for newline at previous position *
*************************************************/
/* This function is called only via the WAS_NEWLINE macro, which does so only
when the newline type is NLTYPE_ANY or NLTYPE_ANYCRLF. The case of a fixed
newline (NLTYPE_FIXED) is handled inline. It is guaranteed that the initial
value of ptr is greater than the start of the string that is being processed.
Arguments:
ptr pointer to possible newline
type the newline type
startptr pointer to the start of the string
lenptr where to return the length
utf TRUE if in utf mode
Returns: TRUE or FALSE
*/
BOOL
PRIV(was_newline)(PCRE2_SPTR ptr, uint32_t type, PCRE2_SPTR startptr,
uint32_t *lenptr, BOOL utf)
{
uint32_t c;
ptr--;
#ifdef SUPPORT_UNICODE
if (utf)
{
BACKCHAR(ptr);
GETCHAR(c, ptr);
}
else c = *ptr;
#else
(void)utf;
c = *ptr;
#endif /* SUPPORT_UNICODE */
if (type == NLTYPE_ANYCRLF) switch(c)
{
case CHAR_LF:
*lenptr = (ptr > startptr && ptr[-1] == CHAR_CR)? 2 : 1;
return TRUE;
case CHAR_CR:
*lenptr = 1;
return TRUE;
default:
return FALSE;
}
/* NLTYPE_ANY */
else switch(c)
{
case CHAR_LF:
*lenptr = (ptr > startptr && ptr[-1] == CHAR_CR)? 2 : 1;
return TRUE;
#ifdef EBCDIC
case CHAR_NEL:
#endif
case CHAR_VT:
case CHAR_FF:
case CHAR_CR:
*lenptr = 1;
return TRUE;
#ifndef EBCDIC
#if PCRE2_CODE_UNIT_WIDTH == 8
case CHAR_NEL:
*lenptr = utf? 2 : 1;
return TRUE;
case 0x2028: /* LS */
case 0x2029: /* PS */
*lenptr = 3;
return TRUE;
#else /* 16-bit or 32-bit code units */
case CHAR_NEL:
case 0x2028: /* LS */
case 0x2029: /* PS */
*lenptr = 1;
return TRUE;
#endif
#endif /* Not EBCDIC */
default:
return FALSE;
}
}
/* End of pcre2_newline.c */
pcre2-10.45/src/pcre2_jit_match.c 0000644 0001751 0000166 00000015324 14750415257 012227 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2023 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
#ifndef INCLUDED_FROM_PCRE2_JIT_COMPILE
#error This file must be included from pcre2_jit_compile.c.
#endif
#if defined(__has_feature)
#if __has_feature(memory_sanitizer)
#include
#endif /* __has_feature(memory_sanitizer) */
#endif /* defined(__has_feature) */
#ifdef SUPPORT_JIT
static SLJIT_NOINLINE int jit_machine_stack_exec(jit_arguments *arguments, jit_function executable_func)
{
sljit_u8 local_space[MACHINE_STACK_SIZE];
struct sljit_stack local_stack;
local_stack.min_start = local_space;
local_stack.start = local_space;
local_stack.end = local_space + MACHINE_STACK_SIZE;
local_stack.top = local_space + MACHINE_STACK_SIZE;
arguments->stack = &local_stack;
return executable_func(arguments);
}
#endif
/*************************************************
* Do a JIT pattern match *
*************************************************/
/* This function runs a JIT pattern match.
Arguments:
code points to the compiled expression
subject points to the subject string
length length of subject string (may contain binary zeros)
start_offset where to start in the subject string
options option bits
match_data points to a match_data block
mcontext points to a match context
Returns: > 0 => success; value is the number of ovector pairs filled
= 0 => success, but ovector is not big enough
-1 => failed to match (PCRE2_ERROR_NOMATCH)
< -1 => some kind of unexpected problem
*/
PCRE2_EXP_DEFN int PCRE2_CALL_CONVENTION
pcre2_jit_match(const pcre2_code *code, PCRE2_SPTR subject, PCRE2_SIZE length,
PCRE2_SIZE start_offset, uint32_t options, pcre2_match_data *match_data,
pcre2_match_context *mcontext)
{
#ifndef SUPPORT_JIT
(void)code;
(void)subject;
(void)length;
(void)start_offset;
(void)options;
(void)match_data;
(void)mcontext;
return PCRE2_ERROR_JIT_BADOPTION;
#else /* SUPPORT_JIT */
pcre2_real_code *re = (pcre2_real_code *)code;
executable_functions *functions = (executable_functions *)re->executable_jit;
pcre2_jit_stack *jit_stack;
uint32_t oveccount = match_data->oveccount;
uint32_t max_oveccount;
union {
void *executable_func;
jit_function call_executable_func;
} convert_executable_func;
jit_arguments arguments;
int rc;
int index = 0;
if ((options & PCRE2_PARTIAL_HARD) != 0)
index = 2;
else if ((options & PCRE2_PARTIAL_SOFT) != 0)
index = 1;
if (functions == NULL || functions->executable_funcs[index] == NULL)
return PCRE2_ERROR_JIT_BADOPTION;
/* Sanity checks should be handled by pcre2_match. */
arguments.str = subject + start_offset;
arguments.begin = subject;
arguments.end = subject + length;
arguments.match_data = match_data;
arguments.startchar_ptr = subject;
arguments.mark_ptr = NULL;
arguments.options = options;
if (mcontext != NULL)
{
arguments.callout = mcontext->callout;
arguments.callout_data = mcontext->callout_data;
arguments.offset_limit = mcontext->offset_limit;
arguments.limit_match = (mcontext->match_limit < re->limit_match)?
mcontext->match_limit : re->limit_match;
if (mcontext->jit_callback != NULL)
jit_stack = mcontext->jit_callback(mcontext->jit_callback_data);
else
jit_stack = (pcre2_jit_stack *)mcontext->jit_callback_data;
}
else
{
arguments.callout = NULL;
arguments.callout_data = NULL;
arguments.offset_limit = PCRE2_UNSET;
arguments.limit_match = (MATCH_LIMIT < re->limit_match)?
MATCH_LIMIT : re->limit_match;
jit_stack = NULL;
}
max_oveccount = functions->top_bracket;
if (oveccount > max_oveccount)
oveccount = max_oveccount;
arguments.oveccount = oveccount << 1;
convert_executable_func.executable_func = functions->executable_funcs[index];
if (jit_stack != NULL)
{
arguments.stack = (struct sljit_stack *)(jit_stack->stack);
rc = convert_executable_func.call_executable_func(&arguments);
}
else
rc = jit_machine_stack_exec(&arguments, convert_executable_func.call_executable_func);
if (rc > (int)oveccount)
rc = 0;
match_data->code = re;
match_data->subject = (rc >= 0 || rc == PCRE2_ERROR_PARTIAL)? subject : NULL;
match_data->subject_length = length;
match_data->rc = rc;
match_data->startchar = arguments.startchar_ptr - subject;
match_data->leftchar = 0;
match_data->rightchar = 0;
match_data->mark = arguments.mark_ptr;
match_data->matchedby = PCRE2_MATCHEDBY_JIT;
#if defined(__has_feature)
#if __has_feature(memory_sanitizer)
if (rc > 0)
__msan_unpoison(match_data->ovector, 2 * rc * sizeof(match_data->ovector[0]));
#endif /* __has_feature(memory_sanitizer) */
#endif /* defined(__has_feature) */
return match_data->rc;
#endif /* SUPPORT_JIT */
}
/* End of pcre2_jit_match.c */
pcre2-10.45/src/pcre2_printint.c 0000644 0001751 0000166 00000072327 14750415257 012142 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This module contains a PCRE private debugging function for printing out the
internal form of a compiled regular expression, along with some supporting
local functions. This source file is #included in pcre2test.c at each supported
code unit width, with PCRE2_SUFFIX set appropriately, just like the functions
that comprise the library. It can also optionally be included in
pcre2_compile.c for detailed debugging in error situations. */
/* Tables of operator names. The same 8-bit table is used for all code unit
widths, so it must be defined only once. The list itself is defined in
pcre2_internal.h, which is #included by pcre2test before this file. */
#ifndef OP_LISTS_DEFINED
static const char *OP_names[] = { OP_NAME_LIST };
STATIC_ASSERT(sizeof(OP_names)/sizeof(*OP_names) == OP_TABLE_LENGTH, OP_names);
#define OP_LISTS_DEFINED
#endif
/* The functions and tables herein must all have mode-dependent names. */
#define OP_lengths PCRE2_SUFFIX(OP_lengths_)
#define get_ucpname PCRE2_SUFFIX(get_ucpname_)
#define pcre2_printint PCRE2_SUFFIX(pcre2_printint_)
#define print_char PCRE2_SUFFIX(print_char_)
#define print_custring PCRE2_SUFFIX(print_custring_)
#define print_custring_bylen PCRE2_SUFFIX(print_custring_bylen_)
#define print_prop PCRE2_SUFFIX(print_prop_)
#define print_char_list PCRE2_SUFFIX(print_char_list_)
#define print_map PCRE2_SUFFIX(print_map_)
#define print_class PCRE2_SUFFIX(print_class_)
/* Table of sizes for the fixed-length opcodes. It's defined in a macro so that
the definition is next to the definition of the opcodes in pcre2_internal.h.
The contents of the table are, however, mode-dependent. */
static const uint8_t OP_lengths[] = { OP_LENGTHS };
STATIC_ASSERT(sizeof(OP_lengths)/sizeof(*OP_lengths) == OP_TABLE_LENGTH,
PCRE2_SUFFIX(OP_lengths_));
/*************************************************
* Print one character from a string *
*************************************************/
/* In UTF mode the character may occupy more than one code unit.
Arguments:
f file to write to
ptr pointer to first code unit of the character
utf TRUE if string is UTF (will be FALSE if UTF is not supported)
Returns: number of additional code units used
*/
static unsigned int
print_char(FILE *f, PCRE2_SPTR ptr, BOOL utf)
{
uint32_t c = *ptr;
BOOL one_code_unit = !utf;
/* If UTF is supported and requested, check for a valid single code unit. */
#ifdef SUPPORT_UNICODE
if (utf)
{
#if PCRE2_CODE_UNIT_WIDTH == 8
one_code_unit = c < 0x80;
#elif PCRE2_CODE_UNIT_WIDTH == 16
one_code_unit = (c & 0xfc00) != 0xd800;
#else
one_code_unit = (c & 0xfffff800u) != 0xd800u;
#endif /* CODE_UNIT_WIDTH */
}
#endif /* SUPPORT_UNICODE */
/* Handle a valid one-code-unit character at any width. */
if (one_code_unit)
{
if (PRINTABLE(c)) fprintf(f, "%c", (char)c);
else if (c < 0x80) fprintf(f, "\\x%02x", c);
else fprintf(f, "\\x{%02x}", c);
return 0;
}
/* Code for invalid UTF code units and multi-unit UTF characters is different
for each width. If UTF is not supported, control should never get here, but we
need a return statement to keep the compiler happy. */
#ifndef SUPPORT_UNICODE
return 0;
#else
/* Malformed UTF-8 should occur only if the sanity check has been turned off.
Rather than swallow random bytes, just stop if we hit a bad one. Print it with
\X instead of \x as an indication. */
#if PCRE2_CODE_UNIT_WIDTH == 8
if ((c & 0xc0) != 0xc0)
{
fprintf(f, "\\X{%x}", c); /* Invalid starting byte */
return 0;
}
else
{
int i;
int a = PRIV(utf8_table4)[c & 0x3f]; /* Number of additional bytes */
int s = 6*a;
c = (c & PRIV(utf8_table3)[a]) << s;
for (i = 1; i <= a; i++)
{
if ((ptr[i] & 0xc0) != 0x80)
{
fprintf(f, "\\X{%x}", c); /* Invalid secondary byte */
return i - 1;
}
s -= 6;
c |= (ptr[i] & 0x3f) << s;
}
fprintf(f, "\\x{%x}", c);
return a;
}
#endif /* PCRE2_CODE_UNIT_WIDTH == 8 */
/* UTF-16: rather than swallow a low surrogate, just stop if we hit a bad one.
Print it with \X instead of \x as an indication. */
#if PCRE2_CODE_UNIT_WIDTH == 16
if ((ptr[1] & 0xfc00) != 0xdc00)
{
fprintf(f, "\\X{%x}", c);
return 0;
}
c = (((c & 0x3ff) << 10) | (ptr[1] & 0x3ff)) + 0x10000;
fprintf(f, "\\x{%x}", c);
return 1;
#endif /* PCRE2_CODE_UNIT_WIDTH == 16 */
/* For UTF-32 we get here only for a malformed code unit, which should only
occur if the sanity check has been turned off. Print it with \X instead of \x
as an indication. */
#if PCRE2_CODE_UNIT_WIDTH == 32
fprintf(f, "\\X{%x}", c);
return 0;
#endif /* PCRE2_CODE_UNIT_WIDTH == 32 */
#endif /* SUPPORT_UNICODE */
}
/*************************************************
* Print string as a list of code units *
*************************************************/
/* These take no account of UTF as they always print each individual code unit.
The string is zero-terminated for print_custring(); the length is given for
print_custring_bylen().
Arguments:
f file to write to
ptr point to the string
len length for print_custring_bylen()
Returns: nothing
*/
static void
print_custring(FILE *f, PCRE2_SPTR ptr)
{
while (*ptr != '\0')
{
uint32_t c = *ptr++;
if (PRINTABLE(c)) fprintf(f, "%c", c); else fprintf(f, "\\x{%x}", c);
}
}
static void
print_custring_bylen(FILE *f, PCRE2_SPTR ptr, PCRE2_UCHAR len)
{
for (; len > 0; len--)
{
uint32_t c = *ptr++;
if (PRINTABLE(c)) fprintf(f, "%c", c); else fprintf(f, "\\x{%x}", c);
}
}
/*************************************************
* Find Unicode property name *
*************************************************/
/* When there is no UTF/UCP support, the table of names does not exist. This
function should not be called in such configurations, because a pattern that
tries to use Unicode properties won't compile. Rather than put lots of #ifdefs
into the main code, however, we just put one into this function.
Now that the table contains both full names and their abbreviations, we do some
fiddling to try to get the full name, which is either the longer of two found
names, or a 3-character script name. */
static const char *
get_ucpname(unsigned int ptype, unsigned int pvalue)
{
#ifdef SUPPORT_UNICODE
int count = 0;
const char *yield = "??";
size_t len = 0;
unsigned int ptypex = (ptype == PT_SC)? PT_SCX : ptype;
for (ptrdiff_t i = PRIV(utt_size) - 1; i >= 0; i--)
{
const ucp_type_table *u = PRIV(utt) + i;
if ((ptype == u->type || ptypex == u->type) && pvalue == u->value)
{
const char *s = PRIV(utt_names) + u->name_offset;
size_t sl = strlen(s);
if (sl == 3 && (u->type == PT_SC || u->type == PT_SCX))
{
yield = s;
break;
}
if (sl > len)
{
yield = s;
len = sl;
}
if (++count >= 2) break;
}
}
return yield;
#else /* No UTF support */
(void)ptype;
(void)pvalue;
return "??";
#endif /* SUPPORT_UNICODE */
}
/*************************************************
* Print Unicode property value *
*************************************************/
/* "Normal" properties can be printed from tables. The PT_CLIST property is a
pseudo-property that contains a pointer to a list of case-equivalent
characters.
Arguments:
f file to write to
code pointer in the compiled code
before text to print before
after text to print after
Returns: nothing
*/
static void
print_prop(FILE *f, PCRE2_SPTR code, const char *before, const char *after)
{
if (code[1] != PT_CLIST)
{
const char *sc = (code[1] == PT_SC)? "script:" : "";
const char *s = get_ucpname(code[1], code[2]);
fprintf(f, "%s%s %s%c%s%s", before, OP_names[*code], sc, toupper(s[0]), s+1, after);
}
else
{
const uint32_t *p = PRIV(ucd_caseless_sets) + code[2];
fprintf (f, "%s%sclist", before, (*code == OP_PROP)? "" : "not ");
while (*p < NOTACHAR) fprintf(f, " %04x", *p++);
fprintf(f, "%s", after);
}
}
/*************************************************
* Print character list *
*************************************************/
/* Prints the characters and character ranges in a character list.
Arguments:
f file to write to
code pointer in the compiled code
*/
static PCRE2_SPTR
print_char_list(FILE *f, PCRE2_SPTR code, const uint8_t *char_lists_end)
{
uint32_t type, list_ind;
uint32_t char_list_add = XCL_CHAR_LIST_LOW_16_ADD;
uint32_t range_start = ~(uint32_t)0, range_end = 0;
const uint8_t *next_char;
#if PCRE2_CODE_UNIT_WIDTH == 8
type = (uint32_t)(code[0] << 8) | code[1];
code += 2;
#else
type = code[0];
code++;
#endif /* CODE_UNIT_WIDTH */
/* Align characters. */
next_char = char_lists_end - (GET(code, 0) << 1);
type &= XCL_TYPE_MASK;
list_ind = 0;
if ((type & XCL_BEGIN_WITH_RANGE) != 0)
range_start = XCL_CHAR_LIST_LOW_16_START;
while (type > 0)
{
uint32_t item_count = type & XCL_ITEM_COUNT_MASK;
if (item_count == XCL_ITEM_COUNT_MASK)
{
if (list_ind <= 1)
{
item_count = *(const uint16_t*)next_char;
next_char += 2;
}
else
{
item_count = *(const uint32_t*)next_char;
next_char += 4;
}
}
while (item_count > 0)
{
if (list_ind <= 1)
{
range_end = *(const uint16_t*)next_char;
next_char += 2;
}
else
{
range_end = *(const uint32_t*)next_char;
next_char += 4;
}
if ((range_end & XCL_CHAR_END) != 0)
{
range_end = char_list_add + (range_end >> XCL_CHAR_SHIFT);
if (range_start < range_end)
fprintf(f, "\\x{%x}-", range_start);
fprintf(f, "\\x{%x}", range_end);
range_start = ~(uint32_t)0;
}
else
range_start = char_list_add + (range_end >> XCL_CHAR_SHIFT);
item_count--;
}
list_ind++;
type >>= XCL_TYPE_BIT_LEN;
/* The following code could be optimized to 8/16/32 bit,
but it is not worth it for a debugging function. */
if (range_start == ~(uint32_t)0)
{
if ((type & XCL_BEGIN_WITH_RANGE) != 0)
{
if (list_ind == 1) range_start = XCL_CHAR_LIST_HIGH_16_START;
else if (list_ind == 2) range_start = XCL_CHAR_LIST_LOW_32_START;
else range_start = XCL_CHAR_LIST_HIGH_32_START;
}
}
else if ((type & XCL_BEGIN_WITH_RANGE) == 0)
{
fprintf(f, "\\x{%x}-", range_start);
if (list_ind == 1) range_end = XCL_CHAR_LIST_LOW_16_END;
else if (list_ind == 2) range_end = XCL_CHAR_LIST_HIGH_16_END;
else if (list_ind == 3) range_end = XCL_CHAR_LIST_LOW_32_END;
else range_end = XCL_CHAR_LIST_HIGH_32_END;
fprintf(f, "\\x{%x}", range_end);
range_start = ~(uint32_t)0;
}
if (list_ind == 1) char_list_add = XCL_CHAR_LIST_HIGH_16_ADD;
else if (list_ind == 2) char_list_add = XCL_CHAR_LIST_LOW_32_ADD;
else char_list_add = XCL_CHAR_LIST_HIGH_32_ADD;
}
return code + LINK_SIZE;
}
/*************************************************
* Print a character bitmap *
*************************************************/
/* Prints a 32-byte bitmap, which occurs within a character class opcode.
Arguments:
f file to write to
map pointer to the bitmap
negated TRUE if the bitmap will be printed as negated
Returns: nothing
*/
static void
print_map(FILE *f, const uint8_t *map, BOOL negated)
{
BOOL first = TRUE;
uint8_t inverted_map[32];
int i;
if (negated)
{
/* Using 255 ^ instead of ~ avoids clang sanitize warning. */
for (i = 0; i < 32; i++) inverted_map[i] = 255 ^ map[i];
map = inverted_map;
}
for (i = 0; i < 256; i++)
{
if ((map[i/8] & (1u << (i&7))) != 0)
{
int j;
for (j = i+1; j < 256; j++)
if ((map[j/8] & (1u << (j&7))) == 0) break;
if (i == '-' || i == '\\' || i == ']' || (first && i == '^'))
fprintf(f, "\\");
if (PRINTABLE(i)) fprintf(f, "%c", i);
else fprintf(f, "\\x%02x", i);
first = FALSE;
if (--j > i)
{
if (j != i + 1) fprintf(f, "-");
if (j == '-' || j == '\\' || j == ']') fprintf(f, "\\");
if (PRINTABLE(j)) fprintf(f, "%c", j);
else fprintf(f, "\\x%02x", j);
}
i = j;
}
}
}
/*************************************************
* Print character class *
*************************************************/
/* Prints a character class, which must be either an OP_CLASS, OP_NCLASS, or
OP_XCLASS.
Arguments:
f file to write to
type OP_CLASS, OP_NCLASS, or OP_XCLASS
code pointer in the compiled code (after the OP tag)
utf TRUE if re is UTF (will be FALSE if UTF is not supported)
before text to print before
after text to print after
Returns: nothing
*/
static void
print_class(FILE *f, int type, PCRE2_SPTR code, const uint8_t *char_lists_end,
BOOL utf, const char *before, const char *after)
{
BOOL printmap, negated;
PCRE2_SPTR ccode;
/* Negative XCLASS and NCLASS both have a bitmap indicating which characters
are accepted. For clarity we print this inverted and prefixed by "^". */
if (type == OP_XCLASS)
{
ccode = code + LINK_SIZE;
printmap = (*ccode & XCL_MAP) != 0;
negated = (*ccode & XCL_NOT) != 0;
ccode++;
}
else /* CLASS or NCLASS */
{
printmap = TRUE;
negated = type == OP_NCLASS;
ccode = code;
}
fprintf(f, "%s[%s", before, negated? "^" : "");
/* Print a bit map */
if (printmap)
{
print_map(f, (const uint8_t *)ccode, negated);
ccode += 32 / sizeof(PCRE2_UCHAR);
}
/* For an XCLASS there is always some additional data */
if (type == OP_XCLASS)
{
PCRE2_UCHAR ch;
while ((ch = *ccode++) != XCL_END)
{
const char *notch = "";
if (ch >= XCL_LIST)
{
ccode = print_char_list(f, ccode - 1, char_lists_end);
break;
}
switch(ch)
{
case XCL_NOTPROP:
notch = "^";
/* Fall through */
case XCL_PROP:
{
unsigned int ptype = *ccode++;
unsigned int pvalue = *ccode++;
const char *s;
switch(ptype)
{
case PT_PXGRAPH:
fprintf(f, "[:%sgraph:]", notch);
break;
case PT_PXPRINT:
fprintf(f, "[:%sprint:]", notch);
break;
case PT_PXPUNCT:
fprintf(f, "[:%spunct:]", notch);
break;
case PT_PXXDIGIT:
fprintf(f, "[:%sxdigit:]", notch);
break;
default:
s = get_ucpname(ptype, pvalue);
fprintf(f, "\\%c{%c%s}", ((notch[0] == '^')? 'P':'p'),
toupper(s[0]), s+1);
break;
}
}
break;
default:
ccode += 1 + print_char(f, ccode, utf);
if (ch == XCL_RANGE)
{
fprintf(f, "-");
ccode += 1 + print_char(f, ccode, utf);
}
break;
}
}
PCRE2_ASSERT(ccode == code + (GET(code, 0) - 1));
}
/* Indicate a non-UTF class which was created by negation */
fprintf(f, "]%s", after);
}
/*************************************************
* Print compiled pattern *
*************************************************/
/* The print_lengths flag controls whether offsets and lengths of items are
printed. Lenths can be turned off from pcre2test so that automatic tests on
bytecode can be written that do not depend on the value of LINK_SIZE.
Arguments:
re a compiled pattern
f the file to write to
print_lengths show various lengths
Returns: nothing
*/
static void
pcre2_printint(pcre2_code *re, FILE *f, BOOL print_lengths)
{
PCRE2_SPTR codestart, nametable, code;
uint32_t nesize = re->name_entry_size;
BOOL utf = (re->overall_options & PCRE2_UTF) != 0;
nametable = (PCRE2_SPTR)((uint8_t *)re + sizeof(pcre2_real_code));
code = codestart = (PCRE2_SPTR)((uint8_t *)re + re->code_start);
for(;;)
{
PCRE2_SPTR ccode;
uint32_t c;
int i;
const char *flag = " ";
unsigned int extra = 0;
if (print_lengths)
fprintf(f, "%3d ", (int)(code - codestart));
else
fprintf(f, " ");
switch(*code)
{
case OP_END:
fprintf(f, " %s\n", OP_names[*code]);
fprintf(f, "------------------------------------------------------------------\n");
return;
case OP_CHAR:
fprintf(f, " ");
do
{
code++;
code += 1 + print_char(f, code, utf);
}
while (*code == OP_CHAR);
fprintf(f, "\n");
continue;
case OP_CHARI:
fprintf(f, " /i ");
do
{
code++;
code += 1 + print_char(f, code, utf);
}
while (*code == OP_CHARI);
fprintf(f, "\n");
continue;
case OP_CBRA:
case OP_CBRAPOS:
case OP_SCBRA:
case OP_SCBRAPOS:
if (print_lengths) fprintf(f, "%3d ", GET(code, 1));
else fprintf(f, " ");
fprintf(f, "%s %d", OP_names[*code], GET2(code, 1+LINK_SIZE));
break;
case OP_BRA:
case OP_BRAPOS:
case OP_SBRA:
case OP_SBRAPOS:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_KETRPOS:
case OP_ALT:
case OP_KET:
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
case OP_ASSERT_NA:
case OP_ASSERTBACK_NA:
case OP_ASSERT_SCS:
case OP_ONCE:
case OP_SCRIPT_RUN:
case OP_COND:
case OP_SCOND:
if (print_lengths) fprintf(f, "%3d ", GET(code, 1));
else fprintf(f, " ");
fprintf(f, "%s", OP_names[*code]);
break;
case OP_REVERSE:
if (print_lengths) fprintf(f, "%3d ", GET2(code, 1));
else fprintf(f, " ");
fprintf(f, "%s", OP_names[*code]);
break;
case OP_VREVERSE:
if (print_lengths) fprintf(f, "%3d %d ", GET2(code, 1),
GET2(code, 1 + IMM2_SIZE));
else fprintf(f, " ");
fprintf(f, "%s", OP_names[*code]);
break;
case OP_CLOSE:
fprintf(f, " %s %d", OP_names[*code], GET2(code, 1));
break;
case OP_CREF:
fprintf(f, "%3d %s", GET2(code,1), OP_names[*code]);
break;
case OP_DNCREF:
{
PCRE2_SPTR entry = nametable + (GET2(code, 1) * nesize) + IMM2_SIZE;
fprintf(f, " %s Capture ref <", flag);
print_custring(f, entry);
fprintf(f, ">%d", GET2(code, 1 + IMM2_SIZE));
}
break;
case OP_RREF:
c = GET2(code, 1);
if (c == RREF_ANY)
fprintf(f, " Cond recurse any");
else
fprintf(f, " Cond recurse %d", c);
break;
case OP_DNRREF:
{
PCRE2_SPTR entry = nametable + (GET2(code, 1) * nesize) + IMM2_SIZE;
fprintf(f, " %s Cond recurse <", flag);
print_custring(f, entry);
fprintf(f, ">%d", GET2(code, 1 + IMM2_SIZE));
}
break;
case OP_FALSE:
fprintf(f, " Cond false");
break;
case OP_TRUE:
fprintf(f, " Cond true");
break;
case OP_STARI:
case OP_MINSTARI:
case OP_POSSTARI:
case OP_PLUSI:
case OP_MINPLUSI:
case OP_POSPLUSI:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_POSQUERYI:
flag = "/i";
/* Fall through */
case OP_STAR:
case OP_MINSTAR:
case OP_POSSTAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_POSPLUS:
case OP_QUERY:
case OP_MINQUERY:
case OP_POSQUERY:
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPOSSTAR:
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSQUERY:
fprintf(f, " %s ", flag);
if (*code >= OP_TYPESTAR)
{
if (code[1] == OP_PROP || code[1] == OP_NOTPROP)
{
print_prop(f, code + 1, "", " ");
extra = 2;
}
else fprintf(f, "%s", OP_names[code[1]]);
}
else extra = print_char(f, code+1, utf);
fprintf(f, "%s", OP_names[*code]);
break;
case OP_EXACTI:
case OP_UPTOI:
case OP_MINUPTOI:
case OP_POSUPTOI:
flag = "/i";
/* Fall through */
case OP_EXACT:
case OP_UPTO:
case OP_MINUPTO:
case OP_POSUPTO:
fprintf(f, " %s ", flag);
extra = print_char(f, code + 1 + IMM2_SIZE, utf);
fprintf(f, "{");
if (*code != OP_EXACT && *code != OP_EXACTI) fprintf(f, "0,");
fprintf(f, "%d}", GET2(code,1));
if (*code == OP_MINUPTO || *code == OP_MINUPTOI) fprintf(f, "?");
else if (*code == OP_POSUPTO || *code == OP_POSUPTOI) fprintf(f, "+");
break;
case OP_TYPEEXACT:
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
{
print_prop(f, code + IMM2_SIZE + 1, " ", " ");
extra = 2;
}
else fprintf(f, " %s", OP_names[code[1 + IMM2_SIZE]]);
fprintf(f, "{");
if (*code != OP_TYPEEXACT) fprintf(f, "0,");
fprintf(f, "%d}", GET2(code,1));
if (*code == OP_TYPEMINUPTO) fprintf(f, "?");
else if (*code == OP_TYPEPOSUPTO) fprintf(f, "+");
break;
case OP_NOTI:
flag = "/i";
/* Fall through */
case OP_NOT:
fprintf(f, " %s [^", flag);
extra = print_char(f, code + 1, utf);
fprintf(f, "] (not)");
break;
case OP_NOTSTARI:
case OP_NOTMINSTARI:
case OP_NOTPOSSTARI:
case OP_NOTPLUSI:
case OP_NOTMINPLUSI:
case OP_NOTPOSPLUSI:
case OP_NOTQUERYI:
case OP_NOTMINQUERYI:
case OP_NOTPOSQUERYI:
flag = "/i";
/* Fall through */
case OP_NOTSTAR:
case OP_NOTMINSTAR:
case OP_NOTPOSSTAR:
case OP_NOTPLUS:
case OP_NOTMINPLUS:
case OP_NOTPOSPLUS:
case OP_NOTQUERY:
case OP_NOTMINQUERY:
case OP_NOTPOSQUERY:
fprintf(f, " %s [^", flag);
extra = print_char(f, code + 1, utf);
fprintf(f, "]%s (not)", OP_names[*code]);
break;
case OP_NOTEXACTI:
case OP_NOTUPTOI:
case OP_NOTMINUPTOI:
case OP_NOTPOSUPTOI:
flag = "/i";
/* Fall through */
case OP_NOTEXACT:
case OP_NOTUPTO:
case OP_NOTMINUPTO:
case OP_NOTPOSUPTO:
fprintf(f, " %s [^", flag);
extra = print_char(f, code + 1 + IMM2_SIZE, utf);
fprintf(f, "]{");
if (*code != OP_NOTEXACT && *code != OP_NOTEXACTI) fprintf(f, "0,");
fprintf(f, "%d}", GET2(code,1));
if (*code == OP_NOTMINUPTO || *code == OP_NOTMINUPTOI) fprintf(f, "?");
else
if (*code == OP_NOTPOSUPTO || *code == OP_NOTPOSUPTOI) fprintf(f, "+");
fprintf(f, " (not)");
break;
case OP_RECURSE:
if (print_lengths) fprintf(f, "%3d ", GET(code, 1));
else fprintf(f, " ");
fprintf(f, "%s", OP_names[*code]);
break;
case OP_REFI:
flag = "/i";
extra = code[1 + IMM2_SIZE];
/* Fall through */
case OP_REF:
fprintf(f, " %s \\%d", flag, GET2(code,1));
if (extra != 0) fprintf(f, " 0x%02x", extra);
ccode = code + OP_lengths[*code];
goto CLASS_REF_REPEAT;
case OP_DNREFI:
flag = "/i";
extra = code[1 + 2*IMM2_SIZE];
/* Fall through */
case OP_DNREF:
{
PCRE2_SPTR entry = nametable + (GET2(code, 1) * nesize) + IMM2_SIZE;
fprintf(f, " %s \\k<", flag);
print_custring(f, entry);
fprintf(f, ">%d", GET2(code, 1 + IMM2_SIZE));
if (extra != 0) fprintf(f, " 0x%02x", extra);
}
ccode = code + OP_lengths[*code];
goto CLASS_REF_REPEAT;
case OP_CALLOUT:
fprintf(f, " %s %d %d %d", OP_names[*code], code[1 + 2*LINK_SIZE],
GET(code, 1), GET(code, 1 + LINK_SIZE));
break;
case OP_CALLOUT_STR:
c = code[1 + 4*LINK_SIZE];
fprintf(f, " %s %c", OP_names[*code], c);
extra = GET(code, 1 + 2*LINK_SIZE);
print_custring_bylen(f, code + 2 + 4*LINK_SIZE, extra - 3 - 4*LINK_SIZE);
for (i = 0; PRIV(callout_start_delims)[i] != 0; i++)
if (c == PRIV(callout_start_delims)[i])
{
c = PRIV(callout_end_delims)[i];
break;
}
fprintf(f, "%c %d %d %d", c, GET(code, 1 + 3*LINK_SIZE), GET(code, 1),
GET(code, 1 + LINK_SIZE));
break;
case OP_PROP:
case OP_NOTPROP:
print_prop(f, code, " ", "");
break;
#ifdef SUPPORT_WIDE_CHARS
case OP_ECLASS:
extra = GET(code, 1);
fprintf(f, " eclass[\n");
/* We print the opcodes contained inside as well. */
ccode = code + 1 + LINK_SIZE + 1;
if ((ccode[-1] & ECL_MAP) != 0)
{
const uint8_t *map = (const uint8_t *)ccode;
/* The first 6 ASCII characters (SOH...ACK) are totally, utterly useless.
If they're set in the bitmap, then it's clearly been formed by negation.*/
BOOL print_negated = (map[0] & 0x7e) == 0x7e;
fprintf(f, " bitmap: [%s", print_negated? "^" : "");
print_map(f, map, print_negated);
fprintf(f, "]\n");
ccode += 32 / sizeof(PCRE2_UCHAR);
}
else
fprintf(f, " no bitmap\n");
while (ccode < code + extra)
{
if (print_lengths)
fprintf(f, "%3d ", (int)(ccode - codestart));
else
fprintf(f, " ");
switch (*ccode)
{
case ECL_AND:
fprintf(f, " AND\n");
ccode += 1;
break;
case ECL_OR:
fprintf(f, " OR\n");
ccode += 1;
break;
case ECL_XOR:
fprintf(f, " XOR\n");
ccode += 1;
break;
case ECL_NOT:
fprintf(f, " NOT\n");
ccode += 1;
break;
case ECL_XCLASS:
print_class(f, OP_XCLASS, ccode+1, (uint8_t*)codestart, utf,
" xclass: ", "\n");
ccode += GET(ccode, 1);
break;
default:
fprintf(f, " UNEXPECTED\n");
ccode += 1;
break;
}
}
fprintf(f, " ]");
goto CLASS_REF_REPEAT;
#endif /* SUPPORT_WIDE_CHARS */
case OP_CLASS:
case OP_NCLASS:
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
if (*code == OP_XCLASS)
extra = GET(code, 1);
#endif
print_class(f, *code, code+1, (uint8_t*)codestart, utf, " ", "");
ccode = code + OP_lengths[*code] + extra;
/* Handle repeats after a class or a back reference */
CLASS_REF_REPEAT:
switch(*ccode)
{
unsigned int min, max;
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSPLUS:
case OP_CRPOSQUERY:
fprintf(f, "%s", OP_names[*ccode]);
extra += OP_lengths[*ccode];
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
min = GET2(ccode,1);
max = GET2(ccode,1 + IMM2_SIZE);
if (max == 0) fprintf(f, "{%u,}", min);
else fprintf(f, "{%u,%u}", min, max);
if (*ccode == OP_CRMINRANGE) fprintf(f, "?");
else if (*ccode == OP_CRPOSRANGE) fprintf(f, "+");
extra += OP_lengths[*ccode];
break;
/* Do nothing if it's not a repeat; this code stops picky compilers
warning about the lack of a default code path. */
default:
break;
}
break;
case OP_MARK:
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_SKIP_ARG:
case OP_THEN_ARG:
fprintf(f, " %s ", OP_names[*code]);
print_custring_bylen(f, code + 2, code[1]);
extra += code[1];
break;
case OP_THEN:
fprintf(f, " %s", OP_names[*code]);
break;
case OP_CIRCM:
case OP_DOLLM:
flag = "/m";
/* Fall through */
/* Anything else is just an item with no data, but possibly a flag. */
default:
fprintf(f, " %s %s", flag, OP_names[*code]);
break;
}
code += OP_lengths[*code] + extra;
fprintf(f, "\n");
}
}
/* End of pcre2_printint.c */
pcre2-10.45/src/pcre2_internal.h 0000644 0001751 0000166 00000311425 14750415257 012107 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE2 is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
#ifndef PCRE2_INTERNAL_H_IDEMPOTENT_GUARD
#define PCRE2_INTERNAL_H_IDEMPOTENT_GUARD
/* We do not support both EBCDIC and Unicode at the same time. The "configure"
script prevents both being selected, but not everybody uses "configure". EBCDIC
is only supported for the 8-bit library, but the check for this has to be later
in this file, because the first part is not width-dependent, and is included by
pcre2test.c with CODE_UNIT_WIDTH == 0. */
#if defined EBCDIC && defined SUPPORT_UNICODE
#error The use of both EBCDIC and SUPPORT_UNICODE is not supported.
#endif
/* When compiling one of the libraries, the value of PCRE2_CODE_UNIT_WIDTH must
be 8, 16, or 32. AutoTools and CMake ensure that this is always the case, but
other other building methods may not, so here is a check. It is cut out when
building pcre2test, bcause that sets the value to zero. No other source should
be including this file. There is no explicit way of forcing a compile to be
abandoned, but trying to include a non-existent file seems cleanest. Otherwise
there will be many irrelevant consequential errors. */
#if (!defined PCRE2_BUILDING_PCRE2TEST && !defined PCRE2_DFTABLES) && \
(!defined PCRE2_CODE_UNIT_WIDTH || \
(PCRE2_CODE_UNIT_WIDTH != 8 && \
PCRE2_CODE_UNIT_WIDTH != 16 && \
PCRE2_CODE_UNIT_WIDTH != 32))
#error PCRE2_CODE_UNIT_WIDTH must be defined as 8, 16, or 32.
#include
#endif
/* Standard C headers */
#include
#include
#include
#include
#include
#include
/* Macros to make boolean values more obvious. The #ifndef is to pacify
compiler warnings in environments where these macros are defined elsewhere.
Unfortunately, there is no way to do the same for the typedef. */
typedef int BOOL;
#ifndef FALSE
#define FALSE 0
#define TRUE 1
#endif
/* Helper macro for static (compile-time) assertions. Can be used inside
functions, or at the top-level of a file. */
#define STATIC_ASSERT_JOIN(a,b) a ## b
#define STATIC_ASSERT(cond, msg) \
typedef int STATIC_ASSERT_JOIN(static_assertion_,msg)[(cond)?1:-1]
/* Valgrind (memcheck) support */
#ifdef SUPPORT_VALGRIND
#include
#endif
/* -ftrivial-auto-var-init support supports initializing all local variables
to avoid some classes of bug, but this can cause an unacceptable slowdown
for large on-stack arrays in hot functions. This macro lets us annotate
such arrays. */
#ifdef HAVE_ATTRIBUTE_UNINITIALIZED
#define PCRE2_KEEP_UNINITIALIZED __attribute__((uninitialized))
#else
#define PCRE2_KEEP_UNINITIALIZED
#endif
/* Older versions of MSVC lack snprintf(). This define allows for
warning/error-free compilation and testing with MSVC compilers back to at least
MSVC 10/2010. Except for VC6 (which is missing some fundamentals and fails). */
#if defined(_MSC_VER) && (_MSC_VER < 1900)
#define snprintf _snprintf
#endif
/* When compiling a DLL for Windows, the exported symbols have to be declared
using some MS magic. I found some useful information on this web page:
http://msdn2.microsoft.com/en-us/library/y4h7bcy6(VS.80).aspx. According to the
information there, using __declspec(dllexport) without "extern" we have a
definition; with "extern" we have a declaration. The settings here override the
setting in pcre2.h (which is included below); it defines only PCRE2_EXP_DECL,
which is all that is needed for applications (they just import the symbols). We
use:
PCRE2_EXP_DECL for declarations
PCRE2_EXP_DEFN for definitions
The reason for wrapping this in #ifndef PCRE2_EXP_DECL is so that pcre2test,
which is an application, but needs to import this file in order to "peek" at
internals, can #include pcre2.h first to get an application's-eye view.
In principle, people compiling for non-Windows, non-Unix-like (i.e. uncommon,
special-purpose environments) might want to stick other stuff in front of
exported symbols. That's why, in the non-Windows case, we set PCRE2_EXP_DEFN
only if it is not already set. */
#ifndef PCRE2_EXP_DECL
# ifdef _WIN32
# ifndef PCRE2_STATIC
# define PCRE2_EXP_DECL extern __declspec(dllexport)
# define PCRE2_EXP_DEFN __declspec(dllexport)
# else
# define PCRE2_EXP_DECL extern PCRE2_EXPORT
# define PCRE2_EXP_DEFN
# endif
# else
# ifdef __cplusplus
# define PCRE2_EXP_DECL extern "C" PCRE2_EXPORT
# else
# define PCRE2_EXP_DECL extern PCRE2_EXPORT
# endif
# ifndef PCRE2_EXP_DEFN
# define PCRE2_EXP_DEFN PCRE2_EXP_DECL
# endif
# endif
#endif
/* Include the public PCRE2 header and the definitions of UCP character
property values. This must follow the setting of PCRE2_EXP_DECL above. */
#include "pcre2.h"
#include "pcre2_ucp.h"
/* When PCRE2 is compiled as a C++ library, the subject pointer can be replaced
with a custom type. This makes it possible, for example, to allow pcre2_match()
to process subject strings that are discontinuous by using a smart pointer
class. It must always be possible to inspect all of the subject string in
pcre2_match() because of the way it backtracks. */
/* WARNING: This is as yet untested for PCRE2. */
#ifdef CUSTOM_SUBJECT_PTR
#undef PCRE2_SPTR
#define PCRE2_SPTR CUSTOM_SUBJECT_PTR
#endif
/* When checking for integer overflow, we need to handle large integers.
If a 64-bit integer type is available, we can use that.
Otherwise we have to cast to double, which of course requires floating point
arithmetic. Handle this by defining a macro for the appropriate type. */
#if defined INT64_MAX || defined int64_t
#define INT64_OR_DOUBLE int64_t
#else
#define INT64_OR_DOUBLE double
#endif
/* External (in the C sense) functions and tables that are private to the
libraries are always referenced using the PRIV macro. This makes it possible
for pcre2test.c to include some of the source files from the libraries using a
different PRIV definition to avoid name clashes. It also makes it clear in the
code that a non-static object is being referenced. */
#ifndef PRIV
#define PRIV(name) _pcre2_##name
#endif
/* When compiling for use with the Virtual Pascal compiler, these functions
need to have their names changed. PCRE2 must be compiled with the -DVPCOMPAT
option on the command line. */
#ifdef VPCOMPAT
#define strlen(s) _strlen(s)
#define strncmp(s1,s2,m) _strncmp(s1,s2,m)
#define memcmp(s,c,n) _memcmp(s,c,n)
#define memcpy(d,s,n) _memcpy(d,s,n)
#define memmove(d,s,n) _memmove(d,s,n)
#define memset(s,c,n) _memset(s,c,n)
#else /* VPCOMPAT */
/* Otherwise, to cope with SunOS4 and other systems that lack memmove(), define
a macro that calls an emulating function. */
#ifndef HAVE_MEMMOVE
#undef memmove /* Some systems may have a macro */
#define memmove(a, b, c) PRIV(memmove)(a, b, c)
#endif /* not HAVE_MEMMOVE */
#endif /* not VPCOMPAT */
/* This is an unsigned int value that no UTF character can ever have, as
Unicode doesn't go beyond 0x0010ffff. */
#define NOTACHAR 0xffffffff
/* This is the largest valid UTF/Unicode code point. */
#define MAX_UTF_CODE_POINT 0x10ffff
/* Compile-time positive error numbers (all except UTF errors, which are
negative) start at this value. It should probably never be changed, in case
some application is checking for specific numbers. There is a copy of this
#define in pcre2posix.c (which now no longer includes this file). Ideally, a
way of having a single definition should be found, but as the number is
unlikely to change, this is not a pressing issue. The original reason for
having a base other than 0 was to keep the absolute values of compile-time and
run-time error numbers numerically different, but in the event the code does
not rely on this. */
#define COMPILE_ERROR_BASE 100
/* The initial frames vector for remembering pcre2_match() backtracking points
is allocated on the heap, of this size (bytes) or ten times the frame size if
larger, unless the heap limit is smaller. Typical frame sizes are a few hundred
bytes (it depends on the number of capturing parentheses) so 20KiB handles
quite a few frames. A larger vector on the heap is obtained for matches that
need more frames, subject to the heap limit. */
#define START_FRAMES_SIZE 20480
/* For DFA matching, an initial internal workspace vector is allocated on the
stack. The heap is used only if this turns out to be too small. */
#define DFA_START_RWS_SIZE 30720
/* Define the default BSR convention. */
#ifdef BSR_ANYCRLF
#define BSR_DEFAULT PCRE2_BSR_ANYCRLF
#else
#define BSR_DEFAULT PCRE2_BSR_UNICODE
#endif
/* ---------------- Basic UTF-8 macros ---------------- */
/* These UTF-8 macros are always defined because they are used in pcre2test for
handling wide characters in 16-bit and 32-bit modes, even if an 8-bit library
is not supported. */
/* Tests whether a UTF-8 code point needs extra bytes to decode. */
#define HASUTF8EXTRALEN(c) ((c) >= 0xc0)
/* The following macros were originally written in the form of loops that used
data from the tables whose names start with PRIV(utf8_table). They were
rewritten by a user so as not to use loops, because in some environments this
gives a significant performance advantage, and it seems never to do any harm.
*/
/* Base macro to pick up the remaining bytes of a UTF-8 character, not
advancing the pointer. */
#define GETUTF8(c, eptr) \
{ \
if ((c & 0x20u) == 0) \
c = ((c & 0x1fu) << 6) | (eptr[1] & 0x3fu); \
else if ((c & 0x10u) == 0) \
c = ((c & 0x0fu) << 12) | ((eptr[1] & 0x3fu) << 6) | (eptr[2] & 0x3fu); \
else if ((c & 0x08u) == 0) \
c = ((c & 0x07u) << 18) | ((eptr[1] & 0x3fu) << 12) | \
((eptr[2] & 0x3fu) << 6) | (eptr[3] & 0x3fu); \
else if ((c & 0x04u) == 0) \
c = ((c & 0x03u) << 24) | ((eptr[1] & 0x3fu) << 18) | \
((eptr[2] & 0x3fu) << 12) | ((eptr[3] & 0x3fu) << 6) | \
(eptr[4] & 0x3fu); \
else \
c = ((c & 0x01u) << 30) | ((eptr[1] & 0x3fu) << 24) | \
((eptr[2] & 0x3fu) << 18) | ((eptr[3] & 0x3fu) << 12) | \
((eptr[4] & 0x3fu) << 6) | (eptr[5] & 0x3fu); \
}
/* Base macro to pick up the remaining bytes of a UTF-8 character, advancing
the pointer. */
#define GETUTF8INC(c, eptr) \
{ \
if ((c & 0x20u) == 0) \
c = ((c & 0x1fu) << 6) | (*eptr++ & 0x3fu); \
else if ((c & 0x10u) == 0) \
{ \
c = ((c & 0x0fu) << 12) | ((*eptr & 0x3fu) << 6) | (eptr[1] & 0x3fu); \
eptr += 2; \
} \
else if ((c & 0x08u) == 0) \
{ \
c = ((c & 0x07u) << 18) | ((*eptr & 0x3fu) << 12) | \
((eptr[1] & 0x3fu) << 6) | (eptr[2] & 0x3fu); \
eptr += 3; \
} \
else if ((c & 0x04u) == 0) \
{ \
c = ((c & 0x03u) << 24) | ((*eptr & 0x3fu) << 18) | \
((eptr[1] & 0x3fu) << 12) | ((eptr[2] & 0x3fu) << 6) | \
(eptr[3] & 0x3fu); \
eptr += 4; \
} \
else \
{ \
c = ((c & 0x01u) << 30) | ((*eptr & 0x3fu) << 24) | \
((eptr[1] & 0x3fu) << 18) | ((eptr[2] & 0x3fu) << 12) | \
((eptr[3] & 0x3fu) << 6) | (eptr[4] & 0x3fu); \
eptr += 5; \
} \
}
/* Base macro to pick up the remaining bytes of a UTF-8 character, not
advancing the pointer, incrementing the length. */
#define GETUTF8LEN(c, eptr, len) \
{ \
if ((c & 0x20u) == 0) \
{ \
c = ((c & 0x1fu) << 6) | (eptr[1] & 0x3fu); \
len++; \
} \
else if ((c & 0x10u) == 0) \
{ \
c = ((c & 0x0fu) << 12) | ((eptr[1] & 0x3fu) << 6) | (eptr[2] & 0x3fu); \
len += 2; \
} \
else if ((c & 0x08u) == 0) \
{\
c = ((c & 0x07u) << 18) | ((eptr[1] & 0x3fu) << 12) | \
((eptr[2] & 0x3fu) << 6) | (eptr[3] & 0x3fu); \
len += 3; \
} \
else if ((c & 0x04u) == 0) \
{ \
c = ((c & 0x03u) << 24) | ((eptr[1] & 0x3fu) << 18) | \
((eptr[2] & 0x3fu) << 12) | ((eptr[3] & 0x3fu) << 6) | \
(eptr[4] & 0x3fu); \
len += 4; \
} \
else \
{\
c = ((c & 0x01u) << 30) | ((eptr[1] & 0x3fu) << 24) | \
((eptr[2] & 0x3fu) << 18) | ((eptr[3] & 0x3fu) << 12) | \
((eptr[4] & 0x3fu) << 6) | (eptr[5] & 0x3fu); \
len += 5; \
} \
}
/* --------------- Whitespace macros ---------------- */
/* Tests for Unicode horizontal and vertical whitespace characters must check a
number of different values. Using a switch statement for this generates the
fastest code (no loop, no memory access), and there are several places in the
interpreter code where this happens. In order to ensure that all the case lists
remain in step, we use macros so that there is only one place where the lists
are defined.
These values are also required as lists in pcre2_compile.c when processing \h,
\H, \v and \V in a character class. The lists are defined in pcre2_tables.c,
but macros that define the values are here so that all the definitions are
together. The lists must be in ascending character order, terminated by
NOTACHAR (which is 0xffffffff).
Any changes should ensure that the various macros are kept in step with each
other. NOTE: The values also appear in pcre2_jit_compile.c. */
/* -------------- ASCII/Unicode environments -------------- */
#ifndef EBCDIC
/* Character U+180E (Mongolian Vowel Separator) is not included in the list of
spaces in the Unicode file PropList.txt, and Perl does not recognize it as a
space. However, in many other sources it is listed as a space and has been in
PCRE (both APIs) for a long time. */
#define HSPACE_LIST \
CHAR_HT, CHAR_SPACE, CHAR_NBSP, \
0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \
0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \
NOTACHAR
#define HSPACE_MULTIBYTE_CASES \
case 0x1680: /* OGHAM SPACE MARK */ \
case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ \
case 0x2000: /* EN QUAD */ \
case 0x2001: /* EM QUAD */ \
case 0x2002: /* EN SPACE */ \
case 0x2003: /* EM SPACE */ \
case 0x2004: /* THREE-PER-EM SPACE */ \
case 0x2005: /* FOUR-PER-EM SPACE */ \
case 0x2006: /* SIX-PER-EM SPACE */ \
case 0x2007: /* FIGURE SPACE */ \
case 0x2008: /* PUNCTUATION SPACE */ \
case 0x2009: /* THIN SPACE */ \
case 0x200A: /* HAIR SPACE */ \
case 0x202f: /* NARROW NO-BREAK SPACE */ \
case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ \
case 0x3000 /* IDEOGRAPHIC SPACE */
#define HSPACE_BYTE_CASES \
case CHAR_HT: \
case CHAR_SPACE: \
case CHAR_NBSP
#define HSPACE_CASES \
HSPACE_BYTE_CASES: \
HSPACE_MULTIBYTE_CASES
#define VSPACE_LIST \
CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR
#define VSPACE_MULTIBYTE_CASES \
case 0x2028: /* LINE SEPARATOR */ \
case 0x2029 /* PARAGRAPH SEPARATOR */
#define VSPACE_BYTE_CASES \
case CHAR_LF: \
case CHAR_VT: \
case CHAR_FF: \
case CHAR_CR: \
case CHAR_NEL
#define VSPACE_CASES \
VSPACE_BYTE_CASES: \
VSPACE_MULTIBYTE_CASES
/* -------------- EBCDIC environments -------------- */
#else
#define HSPACE_LIST CHAR_HT, CHAR_SPACE, CHAR_NBSP, NOTACHAR
#define HSPACE_BYTE_CASES \
case CHAR_HT: \
case CHAR_SPACE: \
case CHAR_NBSP
#define HSPACE_CASES HSPACE_BYTE_CASES
#ifdef EBCDIC_NL25
#define VSPACE_LIST \
CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR
#else
#define VSPACE_LIST \
CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR
#endif
#define VSPACE_BYTE_CASES \
case CHAR_LF: \
case CHAR_VT: \
case CHAR_FF: \
case CHAR_CR: \
case CHAR_NEL
#define VSPACE_CASES VSPACE_BYTE_CASES
#endif /* EBCDIC */
/* -------------- End of whitespace macros -------------- */
/* PCRE2 is able to support several different kinds of newline (CR, LF, CRLF,
"any" and "anycrlf" at present). The following macros are used to package up
testing for newlines. NLBLOCK, PSSTART, and PSEND are defined in the various
modules to indicate in which datablock the parameters exist, and what the
start/end of string field names are. */
#define NLTYPE_FIXED 0 /* Newline is a fixed length string */
#define NLTYPE_ANY 1 /* Newline is any Unicode line ending */
#define NLTYPE_ANYCRLF 2 /* Newline is CR, LF, or CRLF */
/* This macro checks for a newline at the given position */
#define IS_NEWLINE(p) \
((NLBLOCK->nltype != NLTYPE_FIXED)? \
((p) < NLBLOCK->PSEND && \
PRIV(is_newline)((p), NLBLOCK->nltype, NLBLOCK->PSEND, \
&(NLBLOCK->nllen), utf)) \
: \
((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
UCHAR21TEST(p) == NLBLOCK->nl[0] && \
(NLBLOCK->nllen == 1 || UCHAR21TEST(p+1) == NLBLOCK->nl[1]) \
) \
)
/* This macro checks for a newline immediately preceding the given position */
#define WAS_NEWLINE(p) \
((NLBLOCK->nltype != NLTYPE_FIXED)? \
((p) > NLBLOCK->PSSTART && \
PRIV(was_newline)((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \
&(NLBLOCK->nllen), utf)) \
: \
((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
UCHAR21TEST(p - NLBLOCK->nllen) == NLBLOCK->nl[0] && \
(NLBLOCK->nllen == 1 || UCHAR21TEST(p - NLBLOCK->nllen + 1) == NLBLOCK->nl[1]) \
) \
)
/* Private flags containing information about the compiled pattern. The first
three must not be changed, because whichever is set is actually the number of
bytes in a code unit in that mode. */
#define PCRE2_MODE8 0x00000001u /* compiled in 8 bit mode */
#define PCRE2_MODE16 0x00000002u /* compiled in 16 bit mode */
#define PCRE2_MODE32 0x00000004u /* compiled in 32 bit mode */
#define PCRE2_FIRSTSET 0x00000010u /* first_code unit is set */
#define PCRE2_FIRSTCASELESS 0x00000020u /* caseless first code unit */
#define PCRE2_FIRSTMAPSET 0x00000040u /* bitmap of first code units is set */
#define PCRE2_LASTSET 0x00000080u /* last code unit is set */
#define PCRE2_LASTCASELESS 0x00000100u /* caseless last code unit */
#define PCRE2_STARTLINE 0x00000200u /* start after \n for multiline */
#define PCRE2_JCHANGED 0x00000400u /* j option used in pattern */
#define PCRE2_HASCRORLF 0x00000800u /* explicit \r or \n in pattern */
#define PCRE2_HASTHEN 0x00001000u /* pattern contains (*THEN) */
#define PCRE2_MATCH_EMPTY 0x00002000u /* pattern can match empty string */
#define PCRE2_BSR_SET 0x00004000u /* BSR was set in the pattern */
#define PCRE2_NL_SET 0x00008000u /* newline was set in the pattern */
#define PCRE2_NOTEMPTY_SET 0x00010000u /* (*NOTEMPTY) used ) keep */
#define PCRE2_NE_ATST_SET 0x00020000u /* (*NOTEMPTY_ATSTART) used) together */
#define PCRE2_DEREF_TABLES 0x00040000u /* release character tables */
#define PCRE2_NOJIT 0x00080000u /* (*NOJIT) used */
#define PCRE2_HASBKPORX 0x00100000u /* contains \P, \p, or \X */
#define PCRE2_DUPCAPUSED 0x00200000u /* contains (?| */
#define PCRE2_HASBKC 0x00400000u /* contains \C */
#define PCRE2_HASACCEPT 0x00800000u /* contains (*ACCEPT) */
#define PCRE2_MODE_MASK (PCRE2_MODE8 | PCRE2_MODE16 | PCRE2_MODE32)
/* Values for the matchedby field in a match data block. */
enum { PCRE2_MATCHEDBY_INTERPRETER, /* pcre2_match() */
PCRE2_MATCHEDBY_DFA_INTERPRETER, /* pcre2_dfa_match() */
PCRE2_MATCHEDBY_JIT }; /* pcre2_jit_match() */
/* Values for the flags field in a match data block. */
#define PCRE2_MD_COPIED_SUBJECT 0x01u
/* Magic number to provide a small check against being handed junk. */
#define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */
/* The maximum remaining length of subject we are prepared to search for a
req_unit match from an anchored pattern. In 8-bit mode, memchr() is used and is
much faster than the search loop that has to be used in 16-bit and 32-bit
modes. */
#if PCRE2_CODE_UNIT_WIDTH == 8
#define REQ_CU_MAX 5000
#else
#define REQ_CU_MAX 2000
#endif
/* The maximum nesting depth for Unicode character class sets.
Currently fixed. Warning: the interpreter relies on this so it can encode
the operand stack in a uint32_t. A nesting limit of 15 implies (15*2+1)=31
stack operands required, due to the fact that we have two (and only two)
levels of operator precedence. In the UTS#18 syntax, you can write 'x&&y[z]'
and in Perl syntax you can write '(?[ x - y & (z) ])', both of which imply
pushing the match results for x & y to the stack. */
#define ECLASS_NEST_LIMIT 15
/* Offsets for the bitmap tables in the cbits set of tables. Each table
contains a set of bits for a class map. Some classes are built by combining
these tables. */
#define cbit_space 0 /* [:space:] or \s */
#define cbit_xdigit 32 /* [:xdigit:] */
#define cbit_digit 64 /* [:digit:] or \d */
#define cbit_upper 96 /* [:upper:] */
#define cbit_lower 128 /* [:lower:] */
#define cbit_word 160 /* [:word:] or \w */
#define cbit_graph 192 /* [:graph:] */
#define cbit_print 224 /* [:print:] */
#define cbit_punct 256 /* [:punct:] */
#define cbit_cntrl 288 /* [:cntrl:] */
#define cbit_length 320 /* Length of the cbits table */
/* Bit definitions for entries in the ctypes table. Do not change these values
without checking pcre2_jit_compile.c, which has an assertion to ensure that
ctype_word has the value 16. */
#define ctype_space 0x01
#define ctype_letter 0x02
#define ctype_lcletter 0x04
#define ctype_digit 0x08
#define ctype_word 0x10 /* alphanumeric or '_' */
/* Offsets of the various tables from the base tables pointer, and
total length of the tables. */
#define lcc_offset 0 /* Lower case */
#define fcc_offset 256 /* Flip case */
#define cbits_offset 512 /* Character classes */
#define ctypes_offset (cbits_offset + cbit_length) /* Character types */
#define TABLES_LENGTH (ctypes_offset + 256)
/* Private flags used in compile_context.optimization_flags */
#define PCRE2_OPTIM_AUTO_POSSESS 0x00000001u
#define PCRE2_OPTIM_DOTSTAR_ANCHOR 0x00000002u
#define PCRE2_OPTIM_START_OPTIMIZE 0x00000004u
#define PCRE2_OPTIMIZATION_ALL 0x00000007u
/* -------------------- Character and string names ------------------------ */
/* If PCRE2 is to support UTF-8 on EBCDIC platforms, we cannot use normal
character constants like '*' because the compiler would emit their EBCDIC code,
which is different from their ASCII/UTF-8 code. Instead we define macros for
the characters so that they always use the ASCII/UTF-8 code when UTF-8 support
is enabled. When UTF-8 support is not enabled, the definitions use character
literals. Both character and string versions of each character are needed, and
there are some longer strings as well.
This means that, on EBCDIC platforms, the PCRE2 library can handle either
EBCDIC, or UTF-8, but not both. To support both in the same compiled library
would need different lookups depending on whether PCRE2_UTF was set or not.
This would make it impossible to use characters in switch/case statements,
which would reduce performance. For a theoretical use (which nobody has asked
for) in a minority area (EBCDIC platforms), this is not sensible. Any
application that did need both could compile two versions of the library, using
macros to give the functions distinct names. */
#ifndef SUPPORT_UNICODE
/* UTF-8 support is not enabled; use the platform-dependent character literals
so that PCRE2 works in both ASCII and EBCDIC environments, but only in non-UTF
mode. Newline characters are problematic in EBCDIC. Though it has CR and LF
characters, a common practice has been to use its NL (0x15) character as the
line terminator in C-like processing environments. However, sometimes the LF
(0x25) character is used instead, according to this Unicode document:
http://unicode.org/standard/reports/tr13/tr13-5.html
PCRE2 defaults EBCDIC NL to 0x15, but has a build-time option to select 0x25
instead. Whichever is *not* chosen is defined as NEL.
In both ASCII and EBCDIC environments, CHAR_NL and CHAR_LF are synonyms for the
same code point. */
#ifdef EBCDIC
#ifndef EBCDIC_NL25
#define CHAR_NL '\x15'
#define CHAR_NEL '\x25'
#define STR_NL "\x15"
#define STR_NEL "\x25"
#else
#define CHAR_NL '\x25'
#define CHAR_NEL '\x15'
#define STR_NL "\x25"
#define STR_NEL "\x15"
#endif
#define CHAR_LF CHAR_NL
#define STR_LF STR_NL
#define CHAR_ESC '\047'
#define CHAR_DEL '\007'
#define CHAR_NBSP ((unsigned char)'\x41')
#define STR_ESC "\047"
#define STR_DEL "\007"
#else /* Not EBCDIC */
/* In ASCII/Unicode, linefeed is '\n' and we equate this to NL for
compatibility. NEL is the Unicode newline character; make sure it is
a positive value. */
#define CHAR_LF '\n'
#define CHAR_NL CHAR_LF
#define CHAR_NEL ((unsigned char)'\x85')
#define CHAR_ESC '\033'
#define CHAR_DEL '\177'
#define CHAR_NBSP ((unsigned char)'\xa0')
#define STR_LF "\n"
#define STR_NL STR_LF
#define STR_NEL "\x85"
#define STR_ESC "\033"
#define STR_DEL "\177"
#endif /* EBCDIC */
/* The remaining definitions work in both environments. */
#define CHAR_NUL '\0'
#define CHAR_HT '\t'
#define CHAR_VT '\v'
#define CHAR_FF '\f'
#define CHAR_CR '\r'
#define CHAR_BS '\b'
#define CHAR_BEL '\a'
#define CHAR_SPACE ' '
#define CHAR_EXCLAMATION_MARK '!'
#define CHAR_QUOTATION_MARK '"'
#define CHAR_NUMBER_SIGN '#'
#define CHAR_DOLLAR_SIGN '$'
#define CHAR_PERCENT_SIGN '%'
#define CHAR_AMPERSAND '&'
#define CHAR_APOSTROPHE '\''
#define CHAR_LEFT_PARENTHESIS '('
#define CHAR_RIGHT_PARENTHESIS ')'
#define CHAR_ASTERISK '*'
#define CHAR_PLUS '+'
#define CHAR_COMMA ','
#define CHAR_MINUS '-'
#define CHAR_DOT '.'
#define CHAR_SLASH '/'
#define CHAR_0 '0'
#define CHAR_1 '1'
#define CHAR_2 '2'
#define CHAR_3 '3'
#define CHAR_4 '4'
#define CHAR_5 '5'
#define CHAR_6 '6'
#define CHAR_7 '7'
#define CHAR_8 '8'
#define CHAR_9 '9'
#define CHAR_COLON ':'
#define CHAR_SEMICOLON ';'
#define CHAR_LESS_THAN_SIGN '<'
#define CHAR_EQUALS_SIGN '='
#define CHAR_GREATER_THAN_SIGN '>'
#define CHAR_QUESTION_MARK '?'
#define CHAR_COMMERCIAL_AT '@'
#define CHAR_A 'A'
#define CHAR_B 'B'
#define CHAR_C 'C'
#define CHAR_D 'D'
#define CHAR_E 'E'
#define CHAR_F 'F'
#define CHAR_G 'G'
#define CHAR_H 'H'
#define CHAR_I 'I'
#define CHAR_J 'J'
#define CHAR_K 'K'
#define CHAR_L 'L'
#define CHAR_M 'M'
#define CHAR_N 'N'
#define CHAR_O 'O'
#define CHAR_P 'P'
#define CHAR_Q 'Q'
#define CHAR_R 'R'
#define CHAR_S 'S'
#define CHAR_T 'T'
#define CHAR_U 'U'
#define CHAR_V 'V'
#define CHAR_W 'W'
#define CHAR_X 'X'
#define CHAR_Y 'Y'
#define CHAR_Z 'Z'
#define CHAR_LEFT_SQUARE_BRACKET '['
#define CHAR_BACKSLASH '\\'
#define CHAR_RIGHT_SQUARE_BRACKET ']'
#define CHAR_CIRCUMFLEX_ACCENT '^'
#define CHAR_UNDERSCORE '_'
#define CHAR_GRAVE_ACCENT '`'
#define CHAR_a 'a'
#define CHAR_b 'b'
#define CHAR_c 'c'
#define CHAR_d 'd'
#define CHAR_e 'e'
#define CHAR_f 'f'
#define CHAR_g 'g'
#define CHAR_h 'h'
#define CHAR_i 'i'
#define CHAR_j 'j'
#define CHAR_k 'k'
#define CHAR_l 'l'
#define CHAR_m 'm'
#define CHAR_n 'n'
#define CHAR_o 'o'
#define CHAR_p 'p'
#define CHAR_q 'q'
#define CHAR_r 'r'
#define CHAR_s 's'
#define CHAR_t 't'
#define CHAR_u 'u'
#define CHAR_v 'v'
#define CHAR_w 'w'
#define CHAR_x 'x'
#define CHAR_y 'y'
#define CHAR_z 'z'
#define CHAR_LEFT_CURLY_BRACKET '{'
#define CHAR_VERTICAL_LINE '|'
#define CHAR_RIGHT_CURLY_BRACKET '}'
#define CHAR_TILDE '~'
#define STR_HT "\t"
#define STR_VT "\v"
#define STR_FF "\f"
#define STR_CR "\r"
#define STR_BS "\b"
#define STR_BEL "\a"
#define STR_SPACE " "
#define STR_EXCLAMATION_MARK "!"
#define STR_QUOTATION_MARK "\""
#define STR_NUMBER_SIGN "#"
#define STR_DOLLAR_SIGN "$"
#define STR_PERCENT_SIGN "%"
#define STR_AMPERSAND "&"
#define STR_APOSTROPHE "'"
#define STR_LEFT_PARENTHESIS "("
#define STR_RIGHT_PARENTHESIS ")"
#define STR_ASTERISK "*"
#define STR_PLUS "+"
#define STR_COMMA ","
#define STR_MINUS "-"
#define STR_DOT "."
#define STR_SLASH "/"
#define STR_0 "0"
#define STR_1 "1"
#define STR_2 "2"
#define STR_3 "3"
#define STR_4 "4"
#define STR_5 "5"
#define STR_6 "6"
#define STR_7 "7"
#define STR_8 "8"
#define STR_9 "9"
#define STR_COLON ":"
#define STR_SEMICOLON ";"
#define STR_LESS_THAN_SIGN "<"
#define STR_EQUALS_SIGN "="
#define STR_GREATER_THAN_SIGN ">"
#define STR_QUESTION_MARK "?"
#define STR_COMMERCIAL_AT "@"
#define STR_A "A"
#define STR_B "B"
#define STR_C "C"
#define STR_D "D"
#define STR_E "E"
#define STR_F "F"
#define STR_G "G"
#define STR_H "H"
#define STR_I "I"
#define STR_J "J"
#define STR_K "K"
#define STR_L "L"
#define STR_M "M"
#define STR_N "N"
#define STR_O "O"
#define STR_P "P"
#define STR_Q "Q"
#define STR_R "R"
#define STR_S "S"
#define STR_T "T"
#define STR_U "U"
#define STR_V "V"
#define STR_W "W"
#define STR_X "X"
#define STR_Y "Y"
#define STR_Z "Z"
#define STR_LEFT_SQUARE_BRACKET "["
#define STR_BACKSLASH "\\"
#define STR_RIGHT_SQUARE_BRACKET "]"
#define STR_CIRCUMFLEX_ACCENT "^"
#define STR_UNDERSCORE "_"
#define STR_GRAVE_ACCENT "`"
#define STR_a "a"
#define STR_b "b"
#define STR_c "c"
#define STR_d "d"
#define STR_e "e"
#define STR_f "f"
#define STR_g "g"
#define STR_h "h"
#define STR_i "i"
#define STR_j "j"
#define STR_k "k"
#define STR_l "l"
#define STR_m "m"
#define STR_n "n"
#define STR_o "o"
#define STR_p "p"
#define STR_q "q"
#define STR_r "r"
#define STR_s "s"
#define STR_t "t"
#define STR_u "u"
#define STR_v "v"
#define STR_w "w"
#define STR_x "x"
#define STR_y "y"
#define STR_z "z"
#define STR_LEFT_CURLY_BRACKET "{"
#define STR_VERTICAL_LINE "|"
#define STR_RIGHT_CURLY_BRACKET "}"
#define STR_TILDE "~"
#define STRING_ACCEPT0 "ACCEPT\0"
#define STRING_COMMIT0 "COMMIT\0"
#define STRING_F0 "F\0"
#define STRING_FAIL0 "FAIL\0"
#define STRING_MARK0 "MARK\0"
#define STRING_PRUNE0 "PRUNE\0"
#define STRING_SKIP0 "SKIP\0"
#define STRING_THEN "THEN"
#define STRING_atomic0 "atomic\0"
#define STRING_pla0 "pla\0"
#define STRING_plb0 "plb\0"
#define STRING_napla0 "napla\0"
#define STRING_naplb0 "naplb\0"
#define STRING_nla0 "nla\0"
#define STRING_nlb0 "nlb\0"
#define STRING_scs0 "scs\0"
#define STRING_sr0 "sr\0"
#define STRING_asr0 "asr\0"
#define STRING_positive_lookahead0 "positive_lookahead\0"
#define STRING_positive_lookbehind0 "positive_lookbehind\0"
#define STRING_non_atomic_positive_lookahead0 "non_atomic_positive_lookahead\0"
#define STRING_non_atomic_positive_lookbehind0 "non_atomic_positive_lookbehind\0"
#define STRING_negative_lookahead0 "negative_lookahead\0"
#define STRING_negative_lookbehind0 "negative_lookbehind\0"
#define STRING_script_run0 "script_run\0"
#define STRING_atomic_script_run "atomic_script_run"
#define STRING_scan_substring0 "scan_substring\0"
#define STRING_alpha0 "alpha\0"
#define STRING_lower0 "lower\0"
#define STRING_upper0 "upper\0"
#define STRING_alnum0 "alnum\0"
#define STRING_ascii0 "ascii\0"
#define STRING_blank0 "blank\0"
#define STRING_cntrl0 "cntrl\0"
#define STRING_digit0 "digit\0"
#define STRING_graph0 "graph\0"
#define STRING_print0 "print\0"
#define STRING_punct0 "punct\0"
#define STRING_space0 "space\0"
#define STRING_word0 "word\0"
#define STRING_xdigit "xdigit"
#define STRING_DEFINE "DEFINE"
#define STRING_VERSION "VERSION"
#define STRING_WEIRD_STARTWORD "[:<:]]"
#define STRING_WEIRD_ENDWORD "[:>:]]"
#define STRING_CR_RIGHTPAR "CR)"
#define STRING_LF_RIGHTPAR "LF)"
#define STRING_CRLF_RIGHTPAR "CRLF)"
#define STRING_ANY_RIGHTPAR "ANY)"
#define STRING_ANYCRLF_RIGHTPAR "ANYCRLF)"
#define STRING_NUL_RIGHTPAR "NUL)"
#define STRING_BSR_ANYCRLF_RIGHTPAR "BSR_ANYCRLF)"
#define STRING_BSR_UNICODE_RIGHTPAR "BSR_UNICODE)"
#define STRING_UTF8_RIGHTPAR "UTF8)"
#define STRING_UTF16_RIGHTPAR "UTF16)"
#define STRING_UTF32_RIGHTPAR "UTF32)"
#define STRING_UTF_RIGHTPAR "UTF)"
#define STRING_UCP_RIGHTPAR "UCP)"
#define STRING_NO_AUTO_POSSESS_RIGHTPAR "NO_AUTO_POSSESS)"
#define STRING_NO_DOTSTAR_ANCHOR_RIGHTPAR "NO_DOTSTAR_ANCHOR)"
#define STRING_NO_JIT_RIGHTPAR "NO_JIT)"
#define STRING_NO_START_OPT_RIGHTPAR "NO_START_OPT)"
#define STRING_NOTEMPTY_RIGHTPAR "NOTEMPTY)"
#define STRING_NOTEMPTY_ATSTART_RIGHTPAR "NOTEMPTY_ATSTART)"
#define STRING_CASELESS_RESTRICT_RIGHTPAR "CASELESS_RESTRICT)"
#define STRING_TURKISH_CASING_RIGHTPAR "TURKISH_CASING)"
#define STRING_LIMIT_HEAP_EQ "LIMIT_HEAP="
#define STRING_LIMIT_MATCH_EQ "LIMIT_MATCH="
#define STRING_LIMIT_DEPTH_EQ "LIMIT_DEPTH="
#define STRING_LIMIT_RECURSION_EQ "LIMIT_RECURSION="
#define STRING_MARK "MARK"
#define STRING_bc "bc"
#define STRING_bidiclass "bidiclass"
#define STRING_sc "sc"
#define STRING_script "script"
#define STRING_scriptextensions "scriptextensions"
#define STRING_scx "scx"
#else /* SUPPORT_UNICODE */
/* UTF-8 support is enabled; always use UTF-8 (=ASCII) character codes. This
works in both modes non-EBCDIC platforms, and on EBCDIC platforms in UTF-8 mode
only. */
#define CHAR_HT '\011'
#define CHAR_VT '\013'
#define CHAR_FF '\014'
#define CHAR_CR '\015'
#define CHAR_LF '\012'
#define CHAR_NL CHAR_LF
#define CHAR_NEL ((unsigned char)'\x85')
#define CHAR_BS '\010'
#define CHAR_BEL '\007'
#define CHAR_ESC '\033'
#define CHAR_DEL '\177'
#define CHAR_NUL '\0'
#define CHAR_SPACE '\040'
#define CHAR_EXCLAMATION_MARK '\041'
#define CHAR_QUOTATION_MARK '\042'
#define CHAR_NUMBER_SIGN '\043'
#define CHAR_DOLLAR_SIGN '\044'
#define CHAR_PERCENT_SIGN '\045'
#define CHAR_AMPERSAND '\046'
#define CHAR_APOSTROPHE '\047'
#define CHAR_LEFT_PARENTHESIS '\050'
#define CHAR_RIGHT_PARENTHESIS '\051'
#define CHAR_ASTERISK '\052'
#define CHAR_PLUS '\053'
#define CHAR_COMMA '\054'
#define CHAR_MINUS '\055'
#define CHAR_DOT '\056'
#define CHAR_SLASH '\057'
#define CHAR_0 '\060'
#define CHAR_1 '\061'
#define CHAR_2 '\062'
#define CHAR_3 '\063'
#define CHAR_4 '\064'
#define CHAR_5 '\065'
#define CHAR_6 '\066'
#define CHAR_7 '\067'
#define CHAR_8 '\070'
#define CHAR_9 '\071'
#define CHAR_COLON '\072'
#define CHAR_SEMICOLON '\073'
#define CHAR_LESS_THAN_SIGN '\074'
#define CHAR_EQUALS_SIGN '\075'
#define CHAR_GREATER_THAN_SIGN '\076'
#define CHAR_QUESTION_MARK '\077'
#define CHAR_COMMERCIAL_AT '\100'
#define CHAR_A '\101'
#define CHAR_B '\102'
#define CHAR_C '\103'
#define CHAR_D '\104'
#define CHAR_E '\105'
#define CHAR_F '\106'
#define CHAR_G '\107'
#define CHAR_H '\110'
#define CHAR_I '\111'
#define CHAR_J '\112'
#define CHAR_K '\113'
#define CHAR_L '\114'
#define CHAR_M '\115'
#define CHAR_N '\116'
#define CHAR_O '\117'
#define CHAR_P '\120'
#define CHAR_Q '\121'
#define CHAR_R '\122'
#define CHAR_S '\123'
#define CHAR_T '\124'
#define CHAR_U '\125'
#define CHAR_V '\126'
#define CHAR_W '\127'
#define CHAR_X '\130'
#define CHAR_Y '\131'
#define CHAR_Z '\132'
#define CHAR_LEFT_SQUARE_BRACKET '\133'
#define CHAR_BACKSLASH '\134'
#define CHAR_RIGHT_SQUARE_BRACKET '\135'
#define CHAR_CIRCUMFLEX_ACCENT '\136'
#define CHAR_UNDERSCORE '\137'
#define CHAR_GRAVE_ACCENT '\140'
#define CHAR_a '\141'
#define CHAR_b '\142'
#define CHAR_c '\143'
#define CHAR_d '\144'
#define CHAR_e '\145'
#define CHAR_f '\146'
#define CHAR_g '\147'
#define CHAR_h '\150'
#define CHAR_i '\151'
#define CHAR_j '\152'
#define CHAR_k '\153'
#define CHAR_l '\154'
#define CHAR_m '\155'
#define CHAR_n '\156'
#define CHAR_o '\157'
#define CHAR_p '\160'
#define CHAR_q '\161'
#define CHAR_r '\162'
#define CHAR_s '\163'
#define CHAR_t '\164'
#define CHAR_u '\165'
#define CHAR_v '\166'
#define CHAR_w '\167'
#define CHAR_x '\170'
#define CHAR_y '\171'
#define CHAR_z '\172'
#define CHAR_LEFT_CURLY_BRACKET '\173'
#define CHAR_VERTICAL_LINE '\174'
#define CHAR_RIGHT_CURLY_BRACKET '\175'
#define CHAR_TILDE '\176'
#define CHAR_NBSP ((unsigned char)'\xa0')
#define STR_HT "\011"
#define STR_VT "\013"
#define STR_FF "\014"
#define STR_CR "\015"
#define STR_NL "\012"
#define STR_BS "\010"
#define STR_BEL "\007"
#define STR_ESC "\033"
#define STR_DEL "\177"
#define STR_SPACE "\040"
#define STR_EXCLAMATION_MARK "\041"
#define STR_QUOTATION_MARK "\042"
#define STR_NUMBER_SIGN "\043"
#define STR_DOLLAR_SIGN "\044"
#define STR_PERCENT_SIGN "\045"
#define STR_AMPERSAND "\046"
#define STR_APOSTROPHE "\047"
#define STR_LEFT_PARENTHESIS "\050"
#define STR_RIGHT_PARENTHESIS "\051"
#define STR_ASTERISK "\052"
#define STR_PLUS "\053"
#define STR_COMMA "\054"
#define STR_MINUS "\055"
#define STR_DOT "\056"
#define STR_SLASH "\057"
#define STR_0 "\060"
#define STR_1 "\061"
#define STR_2 "\062"
#define STR_3 "\063"
#define STR_4 "\064"
#define STR_5 "\065"
#define STR_6 "\066"
#define STR_7 "\067"
#define STR_8 "\070"
#define STR_9 "\071"
#define STR_COLON "\072"
#define STR_SEMICOLON "\073"
#define STR_LESS_THAN_SIGN "\074"
#define STR_EQUALS_SIGN "\075"
#define STR_GREATER_THAN_SIGN "\076"
#define STR_QUESTION_MARK "\077"
#define STR_COMMERCIAL_AT "\100"
#define STR_A "\101"
#define STR_B "\102"
#define STR_C "\103"
#define STR_D "\104"
#define STR_E "\105"
#define STR_F "\106"
#define STR_G "\107"
#define STR_H "\110"
#define STR_I "\111"
#define STR_J "\112"
#define STR_K "\113"
#define STR_L "\114"
#define STR_M "\115"
#define STR_N "\116"
#define STR_O "\117"
#define STR_P "\120"
#define STR_Q "\121"
#define STR_R "\122"
#define STR_S "\123"
#define STR_T "\124"
#define STR_U "\125"
#define STR_V "\126"
#define STR_W "\127"
#define STR_X "\130"
#define STR_Y "\131"
#define STR_Z "\132"
#define STR_LEFT_SQUARE_BRACKET "\133"
#define STR_BACKSLASH "\134"
#define STR_RIGHT_SQUARE_BRACKET "\135"
#define STR_CIRCUMFLEX_ACCENT "\136"
#define STR_UNDERSCORE "\137"
#define STR_GRAVE_ACCENT "\140"
#define STR_a "\141"
#define STR_b "\142"
#define STR_c "\143"
#define STR_d "\144"
#define STR_e "\145"
#define STR_f "\146"
#define STR_g "\147"
#define STR_h "\150"
#define STR_i "\151"
#define STR_j "\152"
#define STR_k "\153"
#define STR_l "\154"
#define STR_m "\155"
#define STR_n "\156"
#define STR_o "\157"
#define STR_p "\160"
#define STR_q "\161"
#define STR_r "\162"
#define STR_s "\163"
#define STR_t "\164"
#define STR_u "\165"
#define STR_v "\166"
#define STR_w "\167"
#define STR_x "\170"
#define STR_y "\171"
#define STR_z "\172"
#define STR_LEFT_CURLY_BRACKET "\173"
#define STR_VERTICAL_LINE "\174"
#define STR_RIGHT_CURLY_BRACKET "\175"
#define STR_TILDE "\176"
#define STRING_ACCEPT0 STR_A STR_C STR_C STR_E STR_P STR_T "\0"
#define STRING_COMMIT0 STR_C STR_O STR_M STR_M STR_I STR_T "\0"
#define STRING_F0 STR_F "\0"
#define STRING_FAIL0 STR_F STR_A STR_I STR_L "\0"
#define STRING_MARK0 STR_M STR_A STR_R STR_K "\0"
#define STRING_PRUNE0 STR_P STR_R STR_U STR_N STR_E "\0"
#define STRING_SKIP0 STR_S STR_K STR_I STR_P "\0"
#define STRING_THEN STR_T STR_H STR_E STR_N
#define STRING_atomic0 STR_a STR_t STR_o STR_m STR_i STR_c "\0"
#define STRING_pla0 STR_p STR_l STR_a "\0"
#define STRING_plb0 STR_p STR_l STR_b "\0"
#define STRING_napla0 STR_n STR_a STR_p STR_l STR_a "\0"
#define STRING_naplb0 STR_n STR_a STR_p STR_l STR_b "\0"
#define STRING_nla0 STR_n STR_l STR_a "\0"
#define STRING_nlb0 STR_n STR_l STR_b "\0"
#define STRING_scs0 STR_s STR_c STR_s "\0"
#define STRING_sr0 STR_s STR_r "\0"
#define STRING_asr0 STR_a STR_s STR_r "\0"
#define STRING_positive_lookahead0 STR_p STR_o STR_s STR_i STR_t STR_i STR_v STR_e STR_UNDERSCORE STR_l STR_o STR_o STR_k STR_a STR_h STR_e STR_a STR_d "\0"
#define STRING_positive_lookbehind0 STR_p STR_o STR_s STR_i STR_t STR_i STR_v STR_e STR_UNDERSCORE STR_l STR_o STR_o STR_k STR_b STR_e STR_h STR_i STR_n STR_d "\0"
#define STRING_non_atomic_positive_lookahead0 STR_n STR_o STR_n STR_UNDERSCORE STR_a STR_t STR_o STR_m STR_i STR_c STR_UNDERSCORE STR_p STR_o STR_s STR_i STR_t STR_i STR_v STR_e STR_UNDERSCORE STR_l STR_o STR_o STR_k STR_a STR_h STR_e STR_a STR_d "\0"
#define STRING_non_atomic_positive_lookbehind0 STR_n STR_o STR_n STR_UNDERSCORE STR_a STR_t STR_o STR_m STR_i STR_c STR_UNDERSCORE STR_p STR_o STR_s STR_i STR_t STR_i STR_v STR_e STR_UNDERSCORE STR_l STR_o STR_o STR_k STR_b STR_e STR_h STR_i STR_n STR_d "\0"
#define STRING_negative_lookahead0 STR_n STR_e STR_g STR_a STR_t STR_i STR_v STR_e STR_UNDERSCORE STR_l STR_o STR_o STR_k STR_a STR_h STR_e STR_a STR_d "\0"
#define STRING_negative_lookbehind0 STR_n STR_e STR_g STR_a STR_t STR_i STR_v STR_e STR_UNDERSCORE STR_l STR_o STR_o STR_k STR_b STR_e STR_h STR_i STR_n STR_d "\0"
#define STRING_script_run0 STR_s STR_c STR_r STR_i STR_p STR_t STR_UNDERSCORE STR_r STR_u STR_n "\0"
#define STRING_atomic_script_run STR_a STR_t STR_o STR_m STR_i STR_c STR_UNDERSCORE STR_s STR_c STR_r STR_i STR_p STR_t STR_UNDERSCORE STR_r STR_u STR_n
#define STRING_scan_substring0 STR_s STR_c STR_a STR_n STR_UNDERSCORE STR_s STR_u STR_b STR_s STR_t STR_r STR_i STR_n STR_g "\0"
#define STRING_alpha0 STR_a STR_l STR_p STR_h STR_a "\0"
#define STRING_lower0 STR_l STR_o STR_w STR_e STR_r "\0"
#define STRING_upper0 STR_u STR_p STR_p STR_e STR_r "\0"
#define STRING_alnum0 STR_a STR_l STR_n STR_u STR_m "\0"
#define STRING_ascii0 STR_a STR_s STR_c STR_i STR_i "\0"
#define STRING_blank0 STR_b STR_l STR_a STR_n STR_k "\0"
#define STRING_cntrl0 STR_c STR_n STR_t STR_r STR_l "\0"
#define STRING_digit0 STR_d STR_i STR_g STR_i STR_t "\0"
#define STRING_graph0 STR_g STR_r STR_a STR_p STR_h "\0"
#define STRING_print0 STR_p STR_r STR_i STR_n STR_t "\0"
#define STRING_punct0 STR_p STR_u STR_n STR_c STR_t "\0"
#define STRING_space0 STR_s STR_p STR_a STR_c STR_e "\0"
#define STRING_word0 STR_w STR_o STR_r STR_d "\0"
#define STRING_xdigit STR_x STR_d STR_i STR_g STR_i STR_t
#define STRING_DEFINE STR_D STR_E STR_F STR_I STR_N STR_E
#define STRING_VERSION STR_V STR_E STR_R STR_S STR_I STR_O STR_N
#define STRING_WEIRD_STARTWORD STR_LEFT_SQUARE_BRACKET STR_COLON STR_LESS_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
#define STRING_WEIRD_ENDWORD STR_LEFT_SQUARE_BRACKET STR_COLON STR_GREATER_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
#define STRING_CR_RIGHTPAR STR_C STR_R STR_RIGHT_PARENTHESIS
#define STRING_LF_RIGHTPAR STR_L STR_F STR_RIGHT_PARENTHESIS
#define STRING_CRLF_RIGHTPAR STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
#define STRING_ANY_RIGHTPAR STR_A STR_N STR_Y STR_RIGHT_PARENTHESIS
#define STRING_ANYCRLF_RIGHTPAR STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
#define STRING_NUL_RIGHTPAR STR_N STR_U STR_L STR_RIGHT_PARENTHESIS
#define STRING_BSR_ANYCRLF_RIGHTPAR STR_B STR_S STR_R STR_UNDERSCORE STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
#define STRING_BSR_UNICODE_RIGHTPAR STR_B STR_S STR_R STR_UNDERSCORE STR_U STR_N STR_I STR_C STR_O STR_D STR_E STR_RIGHT_PARENTHESIS
#define STRING_UTF8_RIGHTPAR STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS
#define STRING_UTF16_RIGHTPAR STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS
#define STRING_UTF32_RIGHTPAR STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS
#define STRING_UTF_RIGHTPAR STR_U STR_T STR_F STR_RIGHT_PARENTHESIS
#define STRING_UCP_RIGHTPAR STR_U STR_C STR_P STR_RIGHT_PARENTHESIS
#define STRING_NO_AUTO_POSSESS_RIGHTPAR STR_N STR_O STR_UNDERSCORE STR_A STR_U STR_T STR_O STR_UNDERSCORE STR_P STR_O STR_S STR_S STR_E STR_S STR_S STR_RIGHT_PARENTHESIS
#define STRING_NO_DOTSTAR_ANCHOR_RIGHTPAR STR_N STR_O STR_UNDERSCORE STR_D STR_O STR_T STR_S STR_T STR_A STR_R STR_UNDERSCORE STR_A STR_N STR_C STR_H STR_O STR_R STR_RIGHT_PARENTHESIS
#define STRING_NO_JIT_RIGHTPAR STR_N STR_O STR_UNDERSCORE STR_J STR_I STR_T STR_RIGHT_PARENTHESIS
#define STRING_NO_START_OPT_RIGHTPAR STR_N STR_O STR_UNDERSCORE STR_S STR_T STR_A STR_R STR_T STR_UNDERSCORE STR_O STR_P STR_T STR_RIGHT_PARENTHESIS
#define STRING_NOTEMPTY_RIGHTPAR STR_N STR_O STR_T STR_E STR_M STR_P STR_T STR_Y STR_RIGHT_PARENTHESIS
#define STRING_NOTEMPTY_ATSTART_RIGHTPAR STR_N STR_O STR_T STR_E STR_M STR_P STR_T STR_Y STR_UNDERSCORE STR_A STR_T STR_S STR_T STR_A STR_R STR_T STR_RIGHT_PARENTHESIS
#define STRING_CASELESS_RESTRICT_RIGHTPAR STR_C STR_A STR_S STR_E STR_L STR_E STR_S STR_S STR_UNDERSCORE STR_R STR_E STR_S STR_T STR_R STR_I STR_C STR_T STR_RIGHT_PARENTHESIS
#define STRING_TURKISH_CASING_RIGHTPAR STR_T STR_U STR_R STR_K STR_I STR_S STR_H STR_UNDERSCORE STR_C STR_A STR_S STR_I STR_N STR_G STR_RIGHT_PARENTHESIS
#define STRING_LIMIT_HEAP_EQ STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_H STR_E STR_A STR_P STR_EQUALS_SIGN
#define STRING_LIMIT_MATCH_EQ STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_M STR_A STR_T STR_C STR_H STR_EQUALS_SIGN
#define STRING_LIMIT_DEPTH_EQ STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_D STR_E STR_P STR_T STR_H STR_EQUALS_SIGN
#define STRING_LIMIT_RECURSION_EQ STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_R STR_E STR_C STR_U STR_R STR_S STR_I STR_O STR_N STR_EQUALS_SIGN
#define STRING_MARK STR_M STR_A STR_R STR_K
#define STRING_bc STR_b STR_c
#define STRING_bidiclass STR_b STR_i STR_d STR_i STR_c STR_l STR_a STR_s STR_s
#define STRING_sc STR_s STR_c
#define STRING_script STR_s STR_c STR_r STR_i STR_p STR_t
#define STRING_scriptextensions STR_s STR_c STR_r STR_i STR_p STR_t STR_e STR_x STR_t STR_e STR_n STR_s STR_i STR_o STR_n STR_s
#define STRING_scx STR_s STR_c STR_x
#endif /* SUPPORT_UNICODE */
/* -------------------- End of character and string names -------------------*/
/* -------------------- Definitions for compiled patterns -------------------*/
/* Codes for different types of Unicode property. If these definitions are
changed, the autopossessifying table in pcre2_auto_possess.c must be updated to
match. */
#define PT_LAMP 0 /* L& - the union of Lu, Ll, Lt */
#define PT_GC 1 /* Specified general characteristic (e.g. L) */
#define PT_PC 2 /* Specified particular characteristic (e.g. Lu) */
#define PT_SC 3 /* Script only (e.g. Han) */
#define PT_SCX 4 /* Script extensions (includes SC) */
#define PT_ALNUM 5 /* Alphanumeric - the union of L and N */
#define PT_SPACE 6 /* Perl space - general category Z plus 9,10,12,13 */
#define PT_PXSPACE 7 /* POSIX space - Z plus 9,10,11,12,13 */
#define PT_WORD 8 /* Word - L, N, Mn, or Pc */
#define PT_CLIST 9 /* Pseudo-property: match character list */
#define PT_UCNC 10 /* Universal Character nameable character */
#define PT_BIDICL 11 /* Specified bidi class */
#define PT_BOOL 12 /* Boolean property */
#define PT_ANY 13 /* Must be the last entry!
Any property - matches all chars */
#define PT_TABSIZE PT_ANY /* Size of square table for autopossessify tests */
/* The following special properties are used only in XCLASS items, when POSIX
classes are specified and PCRE2_UCP is set - in other words, for Unicode
handling of these classes. They are not available via the \p or \P escapes like
those in the above list, and so they do not take part in the autopossessifying
table. */
#define PT_PXGRAPH 14 /* [:graph:] - characters that mark the paper */
#define PT_PXPRINT 15 /* [:print:] - [:graph:] plus non-control spaces */
#define PT_PXPUNCT 16 /* [:punct:] - punctuation characters */
#define PT_PXXDIGIT 17 /* [:xdigit:] - hex digits */
/* This value is used when parsing \p and \P escapes to indicate that neither
\p{script:...} nor \p{scx:...} has been encountered. */
#define PT_NOTSCRIPT 255
/* Flag bits and data types for the extended class (OP_XCLASS) for classes that
contain characters with values greater than 255. */
#define XCL_NOT 0x01 /* Flag: this is a negative class */
#define XCL_MAP 0x02 /* Flag: a 32-byte map is present */
#define XCL_HASPROP 0x04 /* Flag: property checks are present. */
#define XCL_END 0 /* Marks end of individual items */
#define XCL_SINGLE 1 /* Single item (one multibyte char) follows */
#define XCL_RANGE 2 /* A range (two multibyte chars) follows */
#define XCL_PROP 3 /* Unicode property (2-byte property code follows) */
#define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */
/* This value represents the beginning of character lists. The value
is 16 bit long, and stored as a high and low byte pair in 8 bit mode.
The lower 12 bit contains information about character lists (see later). */
#define XCL_LIST (sizeof(PCRE2_UCHAR) == 1 ? 0x10 : 0x1000)
/* When a character class contains many characters/ranges,
they are stored in character lists. There are four character
lists which contain characters/ranges within a given range.
The name, character range and item size for each list:
Low16 [0x100 - 0x7fff] 16 bit items
High16 [0x8000 - 0xffff] 16 bit items
Low32 [0x10000 - 0x7fffffff] 32 bit items
High32 [0x80000000 - 0xffffffff] 32 bit items
The Low32 character list is used only when utf encoding or 32 bit
character width is enabled, and the High32 character is used only
when 32 bit character width is enabled.
Each character list contain items. The lowest bit represents that
an item is the beginning of a range (bit is cleared), or not (bit
is set). The other bits represent the character shifted left by
one, so its highest bit is discarded. Due to the layout of character
lists, the highest bit of a character is always known:
Low16 and Low32: the highest bit is always zero
High16 and High32: the highest bit is always one
The items are ordered in increasing order, so binary search can be
used to find the lower bound of an input character. The lower bound
is the highest item, which value is less or equal than the input
character. If the lower bit of the item is cleard, or the character
stored in the item equals to the input character, the input
character is in the character list. */
/* Character list constants. */
#define XCL_CHAR_LIST_LOW_16_START 0x100
#define XCL_CHAR_LIST_LOW_16_END 0x7fff
#define XCL_CHAR_LIST_LOW_16_ADD 0x0
#define XCL_CHAR_LIST_HIGH_16_START 0x8000
#define XCL_CHAR_LIST_HIGH_16_END 0xffff
#define XCL_CHAR_LIST_HIGH_16_ADD 0x8000
#define XCL_CHAR_LIST_LOW_32_START 0x10000
#define XCL_CHAR_LIST_LOW_32_END 0x7fffffff
#define XCL_CHAR_LIST_LOW_32_ADD 0x0
#define XCL_CHAR_LIST_HIGH_32_START 0x80000000
#define XCL_CHAR_LIST_HIGH_32_END 0xffffffff
#define XCL_CHAR_LIST_HIGH_32_ADD 0x80000000
/* Mask for getting the descriptors of character list ranges.
Each descriptor has XCL_TYPE_BIT_LEN bits, and can be processed
by XCL_BEGIN_WITH_RANGE and XCL_ITEM_COUNT_MASK macros. */
#define XCL_TYPE_MASK 0xfff
#define XCL_TYPE_BIT_LEN 3
/* If this bit is set, the first item of the character list is the
end of a range, which started before the starting character of the
character list. */
#define XCL_BEGIN_WITH_RANGE 0x4
/* Number of items in the character list: 0, 1, or 2. The value 3
represents that the item count is stored at the begining of the
character list. The item count has the same width as the items
in the character list (e.g. 16 bit for Low16 and High16 lists). */
#define XCL_ITEM_COUNT_MASK 0x3
/* Shift and flag for constructing character list items. The XCL_CHAR_END
is set, when the item is not the beginning of a range. The XCL_CHAR_SHIFT
can be used to encode / decode the character value stored in an item. */
#define XCL_CHAR_END 0x1
#define XCL_CHAR_SHIFT 1
/* Flag bits for an extended class (OP_ECLASS), which is used for complex
character matches such as [\p{Greek} && \p{Ll}]. */
#define ECL_MAP 0x01 /* Flag: a 32-byte map is present */
/* Type tags for the items stored in an extended class (OP_ECLASS). These items
follow the OP_ECLASS's flag char and bitmap, and represent a Reverse Polish
Notation list of operands and operators manipulating a stack of bits. */
#define ECL_AND 1 /* Pop two from the stack, AND, and push result. */
#define ECL_OR 2 /* Pop two from the stack, OR, and push result. */
#define ECL_XOR 3 /* Pop two from the stack, XOR, and push result. */
#define ECL_NOT 4 /* Pop one from the stack, NOT, and push result. */
#define ECL_XCLASS 5 /* XCLASS nested within ECLASS; match and push result. */
#define ECL_ANY 6 /* Temporary, only used during compilation. */
#define ECL_NONE 7 /* Temporary, only used during compilation. */
/* These are escaped items that aren't just an encoding of a particular data
value such as \n. They must have non-zero values, as check_escape() returns 0
for a data character. In the escapes[] table in pcre2_compile.c their values
are negated in order to distinguish them from data values.
They must appear here in the same order as in the opcode definitions below, up
to ESC_z. There's a dummy for OP_ALLANY because it corresponds to "." in DOTALL
mode rather than an escape sequence. It is also used for [^] in JavaScript
compatibility mode, and for \C in non-utf mode. In non-DOTALL mode, "." behaves
like \N.
ESC_ub is a special return from check_escape() when, in BSUX mode, \u{ is not
followed by hex digits and }, in which case it should mean a literal "u"
followed by a literal "{". This hack is necessary for cases like \u{ 12}
because without it, this is interpreted as u{12} now that spaces are allowed in
quantifiers.
Negative numbers are used to encode a backreference (\1, \2, \3, etc.) in
check_escape(). There are tests in the code for an escape greater than ESC_b
and less than ESC_Z to detect the types that may be repeated. These are the
types that consume characters. If any new escapes are put in between that don't
consume a character, that code will have to change. */
enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s,
ESC_W, ESC_w, ESC_N, ESC_dum, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H,
ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z,
ESC_E, ESC_Q, ESC_g, ESC_k, ESC_ub };
/********************** Opcode definitions ******************/
/****** NOTE NOTE NOTE ******
Starting from 1 (i.e. after OP_END), the values up to OP_EOD must correspond in
order to the list of escapes immediately above. Furthermore, values up to
OP_DOLLM must not be changed without adjusting the table called autoposstab in
pcre2_auto_possess.c.
Whenever this list is updated, the two macro definitions that follow must be
updated to match. The possessification table called "opcode_possessify" in
pcre2_compile.c must also be updated, and also the tables called "coptable"
and "poptable" in pcre2_dfa_match.c.
****** NOTE NOTE NOTE ******/
/* The values between FIRST_AUTOTAB_OP and LAST_AUTOTAB_RIGHT_OP, inclusive,
are used in a table for deciding whether a repeated character type can be
auto-possessified. */
#define FIRST_AUTOTAB_OP OP_NOT_DIGIT
#define LAST_AUTOTAB_LEFT_OP OP_EXTUNI
#define LAST_AUTOTAB_RIGHT_OP OP_DOLLM
enum {
OP_END, /* 0 End of pattern */
/* Values corresponding to backslashed metacharacters */
OP_SOD, /* 1 Start of data: \A */
OP_SOM, /* 2 Start of match (subject + offset): \G */
OP_SET_SOM, /* 3 Set start of match (\K) */
OP_NOT_WORD_BOUNDARY, /* 4 \B -- see also OP_NOT_UCP_WORD_BOUNDARY */
OP_WORD_BOUNDARY, /* 5 \b -- see also OP_UCP_WORD_BOUNDARY */
OP_NOT_DIGIT, /* 6 \D */
OP_DIGIT, /* 7 \d */
OP_NOT_WHITESPACE, /* 8 \S */
OP_WHITESPACE, /* 9 \s */
OP_NOT_WORDCHAR, /* 10 \W */
OP_WORDCHAR, /* 11 \w */
OP_ANY, /* 12 Match any character except newline (\N) */
OP_ALLANY, /* 13 Match any character */
OP_ANYBYTE, /* 14 Match any byte (\C); different to OP_ANY for UTF-8 */
OP_NOTPROP, /* 15 \P (not Unicode property) */
OP_PROP, /* 16 \p (Unicode property) */
OP_ANYNL, /* 17 \R (any newline sequence) */
OP_NOT_HSPACE, /* 18 \H (not horizontal whitespace) */
OP_HSPACE, /* 19 \h (horizontal whitespace) */
OP_NOT_VSPACE, /* 20 \V (not vertical whitespace) */
OP_VSPACE, /* 21 \v (vertical whitespace) */
OP_EXTUNI, /* 22 \X (extended Unicode sequence */
OP_EODN, /* 23 End of data or \n at end of data (\Z) */
OP_EOD, /* 24 End of data (\z) */
/* Line end assertions */
OP_DOLL, /* 25 End of line - not multiline */
OP_DOLLM, /* 26 End of line - multiline */
OP_CIRC, /* 27 Start of line - not multiline */
OP_CIRCM, /* 28 Start of line - multiline */
/* Single characters; caseful must precede the caseless ones, and these
must remain in this order, and adjacent. */
OP_CHAR, /* 29 Match one character, casefully */
OP_CHARI, /* 30 Match one character, caselessly */
OP_NOT, /* 31 Match one character, not the given one, casefully */
OP_NOTI, /* 32 Match one character, not the given one, caselessly */
/* The following sets of 13 opcodes must always be kept in step because
the offset from the first one is used to generate the others. */
/* Repeated characters; caseful must precede the caseless ones */
OP_STAR, /* 33 The maximizing and minimizing versions of */
OP_MINSTAR, /* 34 these six opcodes must come in pairs, with */
OP_PLUS, /* 35 the minimizing one second. */
OP_MINPLUS, /* 36 */
OP_QUERY, /* 37 */
OP_MINQUERY, /* 38 */
OP_UPTO, /* 39 From 0 to n matches of one character, caseful*/
OP_MINUPTO, /* 40 */
OP_EXACT, /* 41 Exactly n matches */
OP_POSSTAR, /* 42 Possessified star, caseful */
OP_POSPLUS, /* 43 Possessified plus, caseful */
OP_POSQUERY, /* 44 Posesssified query, caseful */
OP_POSUPTO, /* 45 Possessified upto, caseful */
/* Repeated characters; caseless must follow the caseful ones */
OP_STARI, /* 46 */
OP_MINSTARI, /* 47 */
OP_PLUSI, /* 48 */
OP_MINPLUSI, /* 49 */
OP_QUERYI, /* 50 */
OP_MINQUERYI, /* 51 */
OP_UPTOI, /* 52 From 0 to n matches of one character, caseless */
OP_MINUPTOI, /* 53 */
OP_EXACTI, /* 54 */
OP_POSSTARI, /* 55 Possessified star, caseless */
OP_POSPLUSI, /* 56 Possessified plus, caseless */
OP_POSQUERYI, /* 57 Posesssified query, caseless */
OP_POSUPTOI, /* 58 Possessified upto, caseless */
/* The negated ones must follow the non-negated ones, and match them */
/* Negated repeated character, caseful; must precede the caseless ones */
OP_NOTSTAR, /* 59 The maximizing and minimizing versions of */
OP_NOTMINSTAR, /* 60 these six opcodes must come in pairs, with */
OP_NOTPLUS, /* 61 the minimizing one second. They must be in */
OP_NOTMINPLUS, /* 62 exactly the same order as those above. */
OP_NOTQUERY, /* 63 */
OP_NOTMINQUERY, /* 64 */
OP_NOTUPTO, /* 65 From 0 to n matches, caseful */
OP_NOTMINUPTO, /* 66 */
OP_NOTEXACT, /* 67 Exactly n matches */
OP_NOTPOSSTAR, /* 68 Possessified versions, caseful */
OP_NOTPOSPLUS, /* 69 */
OP_NOTPOSQUERY, /* 70 */
OP_NOTPOSUPTO, /* 71 */
/* Negated repeated character, caseless; must follow the caseful ones */
OP_NOTSTARI, /* 72 */
OP_NOTMINSTARI, /* 73 */
OP_NOTPLUSI, /* 74 */
OP_NOTMINPLUSI, /* 75 */
OP_NOTQUERYI, /* 76 */
OP_NOTMINQUERYI, /* 77 */
OP_NOTUPTOI, /* 78 From 0 to n matches, caseless */
OP_NOTMINUPTOI, /* 79 */
OP_NOTEXACTI, /* 80 Exactly n matches */
OP_NOTPOSSTARI, /* 81 Possessified versions, caseless */
OP_NOTPOSPLUSI, /* 82 */
OP_NOTPOSQUERYI, /* 83 */
OP_NOTPOSUPTOI, /* 84 */
/* Character types */
OP_TYPESTAR, /* 85 The maximizing and minimizing versions of */
OP_TYPEMINSTAR, /* 86 these six opcodes must come in pairs, with */
OP_TYPEPLUS, /* 87 the minimizing one second. These codes must */
OP_TYPEMINPLUS, /* 88 be in exactly the same order as those above. */
OP_TYPEQUERY, /* 89 */
OP_TYPEMINQUERY, /* 90 */
OP_TYPEUPTO, /* 91 From 0 to n matches */
OP_TYPEMINUPTO, /* 92 */
OP_TYPEEXACT, /* 93 Exactly n matches */
OP_TYPEPOSSTAR, /* 94 Possessified versions */
OP_TYPEPOSPLUS, /* 95 */
OP_TYPEPOSQUERY, /* 96 */
OP_TYPEPOSUPTO, /* 97 */
/* These are used for character classes and back references; only the
first six are the same as the sets above. */
OP_CRSTAR, /* 98 The maximizing and minimizing versions of */
OP_CRMINSTAR, /* 99 all these opcodes must come in pairs, with */
OP_CRPLUS, /* 100 the minimizing one second. These codes must */
OP_CRMINPLUS, /* 101 be in exactly the same order as those above. */
OP_CRQUERY, /* 102 */
OP_CRMINQUERY, /* 103 */
OP_CRRANGE, /* 104 These are different to the three sets above. */
OP_CRMINRANGE, /* 105 */
OP_CRPOSSTAR, /* 106 Possessified versions */
OP_CRPOSPLUS, /* 107 */
OP_CRPOSQUERY, /* 108 */
OP_CRPOSRANGE, /* 109 */
/* End of quantifier opcodes */
OP_CLASS, /* 110 Match a character class, chars < 256 only */
OP_NCLASS, /* 111 Same, but the bitmap was created from a negative
class - the difference is relevant only when a
character > 255 is encountered. */
OP_XCLASS, /* 112 Extended class for handling > 255 chars within the
class. This does both positive and negative. */
OP_ECLASS, /* 113 Really-extended class, for handling logical
expressions computed over characters. */
OP_REF, /* 114 Match a back reference, casefully */
OP_REFI, /* 115 Match a back reference, caselessly */
OP_DNREF, /* 116 Match a duplicate name backref, casefully */
OP_DNREFI, /* 117 Match a duplicate name backref, caselessly */
OP_RECURSE, /* 118 Match a numbered subpattern (possibly recursive) */
OP_CALLOUT, /* 119 Call out to external function if provided */
OP_CALLOUT_STR, /* 120 Call out with string argument */
OP_ALT, /* 121 Start of alternation */
OP_KET, /* 122 End of group that doesn't have an unbounded repeat */
OP_KETRMAX, /* 123 These two must remain together and in this */
OP_KETRMIN, /* 124 order. They are for groups the repeat for ever. */
OP_KETRPOS, /* 125 Possessive unlimited repeat. */
/* The assertions must come before BRA, CBRA, ONCE, and COND. */
OP_REVERSE, /* 126 Move pointer back - used in lookbehind assertions */
OP_VREVERSE, /* 127 Move pointer back - variable */
OP_ASSERT, /* 128 Positive lookahead */
OP_ASSERT_NOT, /* 129 Negative lookahead */
OP_ASSERTBACK, /* 130 Positive lookbehind */
OP_ASSERTBACK_NOT, /* 131 Negative lookbehind */
OP_ASSERT_NA, /* 132 Positive non-atomic lookahead */
OP_ASSERTBACK_NA, /* 133 Positive non-atomic lookbehind */
OP_ASSERT_SCS, /* 134 Scan substring */
/* ONCE, SCRIPT_RUN, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come
immediately after the assertions, with ONCE first, as there's a test for >=
ONCE for a subpattern that isn't an assertion. The POS versions must
immediately follow the non-POS versions in each case. */
OP_ONCE, /* 135 Atomic group, contains captures */
OP_SCRIPT_RUN, /* 136 Non-capture, but check characters' scripts */
OP_BRA, /* 137 Start of non-capturing bracket */
OP_BRAPOS, /* 138 Ditto, with unlimited, possessive repeat */
OP_CBRA, /* 139 Start of capturing bracket */
OP_CBRAPOS, /* 140 Ditto, with unlimited, possessive repeat */
OP_COND, /* 141 Conditional group */
/* These five must follow the previous five, in the same order. There's a
check for >= SBRA to distinguish the two sets. */
OP_SBRA, /* 142 Start of non-capturing bracket, check empty */
OP_SBRAPOS, /* 143 Ditto, with unlimited, possessive repeat */
OP_SCBRA, /* 144 Start of capturing bracket, check empty */
OP_SCBRAPOS, /* 145 Ditto, with unlimited, possessive repeat */
OP_SCOND, /* 146 Conditional group, check empty */
/* The next two pairs must (respectively) be kept together. */
OP_CREF, /* 147 Used to hold a capture number as condition */
OP_DNCREF, /* 148 Used to point to duplicate names as a condition */
OP_RREF, /* 149 Used to hold a recursion number as condition */
OP_DNRREF, /* 150 Used to point to duplicate names as a condition */
OP_FALSE, /* 151 Always false (used by DEFINE and VERSION) */
OP_TRUE, /* 152 Always true (used by VERSION) */
OP_BRAZERO, /* 153 These two must remain together and in this */
OP_BRAMINZERO, /* 154 order. */
OP_BRAPOSZERO, /* 155 */
/* These are backtracking control verbs */
OP_MARK, /* 156 always has an argument */
OP_PRUNE, /* 157 */
OP_PRUNE_ARG, /* 158 same, but with argument */
OP_SKIP, /* 159 */
OP_SKIP_ARG, /* 160 same, but with argument */
OP_THEN, /* 161 */
OP_THEN_ARG, /* 162 same, but with argument */
OP_COMMIT, /* 163 */
OP_COMMIT_ARG, /* 164 same, but with argument */
/* These are forced failure and success verbs. FAIL and ACCEPT do accept an
argument, but these cases can be compiled as, for example, (*MARK:X)(*FAIL)
without the need for a special opcode. */
OP_FAIL, /* 165 */
OP_ACCEPT, /* 166 */
OP_ASSERT_ACCEPT, /* 167 Used inside assertions */
OP_CLOSE, /* 168 Used before OP_ACCEPT to close open captures */
/* This is used to skip a subpattern with a {0} quantifier */
OP_SKIPZERO, /* 169 */
/* This is used to identify a DEFINE group during compilation so that it can
be checked for having only one branch. It is changed to OP_FALSE before
compilation finishes. */
OP_DEFINE, /* 170 */
/* These opcodes replace their normal counterparts in UCP mode when
PCRE2_EXTRA_ASCII_BSW is not set. */
OP_NOT_UCP_WORD_BOUNDARY, /* 171 */
OP_UCP_WORD_BOUNDARY, /* 172 */
/* This is not an opcode, but is used to check that tables indexed by opcode
are the correct length, in order to catch updating errors - there have been
some in the past. */
OP_TABLE_LENGTH
};
/* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro
definitions that follow must also be updated to match. There are also tables
called "opcode_possessify" in pcre2_compile.c and "coptable" and "poptable" in
pcre2_dfa_match.c that must be updated. */
/* This macro defines textual names for all the opcodes. These are used only
for debugging, and some of them are only partial names. The macro is referenced
only in pcre2_printint.c, which fills out the full names in many cases (and in
some cases doesn't actually use these names at all). */
#define OP_NAME_LIST \
"End", "\\A", "\\G", "\\K", "\\B", "\\b", "\\D", "\\d", \
"\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte", \
"notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v", \
"extuni", "\\Z", "\\z", \
"$", "$", "^", "^", "char", "chari", "not", "noti", \
"*", "*?", "+", "+?", "?", "??", \
"{", "{", "{", \
"*+","++", "?+", "{", \
"*", "*?", "+", "+?", "?", "??", \
"{", "{", "{", \
"*+","++", "?+", "{", \
"*", "*?", "+", "+?", "?", "??", \
"{", "{", "{", \
"*+","++", "?+", "{", \
"*", "*?", "+", "+?", "?", "??", \
"{", "{", "{", \
"*+","++", "?+", "{", \
"*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
"*+","++", "?+", "{", \
"*", "*?", "+", "+?", "?", "??", "{", "{", \
"*+","++", "?+", "{", \
"class", "nclass", "xclass", "eclass", \
"Ref", "Refi", "DnRef", "DnRefi", \
"Recurse", "Callout", "CalloutStr", \
"Alt", "Ket", "KetRmax", "KetRmin", "KetRpos", \
"Reverse", "VReverse", "Assert", "Assert not", \
"Assert back", "Assert back not", \
"Non-atomic assert", "Non-atomic assert back", \
"Scan substring", \
"Once", \
"Script run", \
"Bra", "BraPos", "CBra", "CBraPos", \
"Cond", \
"SBra", "SBraPos", "SCBra", "SCBraPos", \
"SCond", \
"Capture ref", "Capture dnref", "Cond rec", "Cond dnrec", \
"Cond false", "Cond true", \
"Brazero", "Braminzero", "Braposzero", \
"*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP", \
"*THEN", "*THEN", "*COMMIT", "*COMMIT", "*FAIL", \
"*ACCEPT", "*ASSERT_ACCEPT", \
"Close", "Skip zero", "Define", "\\B (ucp)", "\\b (ucp)"
/* This macro defines the length of fixed length operations in the compiled
regex. The lengths are used when searching for specific things, and also in the
debugging printing of a compiled regex. We use a macro so that it can be
defined close to the definitions of the opcodes themselves.
As things have been extended, some of these are no longer fixed lenths, but are
minima instead. For example, the length of a single-character repeat may vary
in UTF-8 mode. The code that uses this table must know about such things. */
#define OP_LENGTHS \
1, /* End */ \
1, 1, 1, 1, 1, /* \A, \G, \K, \B, \b */ \
1, 1, 1, 1, 1, 1, /* \D, \d, \S, \s, \W, \w */ \
1, 1, 1, /* Any, AllAny, Anybyte */ \
3, 3, /* \P, \p */ \
1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */ \
1, /* \X */ \
1, 1, 1, 1, 1, 1, /* \Z, \z, $, $M ^, ^M */ \
2, /* Char - the minimum length */ \
2, /* Chari - the minimum length */ \
2, /* not */ \
2, /* noti */ \
/* Positive single-char repeats ** These are */ \
2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \
2+IMM2_SIZE, 2+IMM2_SIZE, /* upto, minupto ** mode */ \
2+IMM2_SIZE, /* exact */ \
2, 2, 2, 2+IMM2_SIZE, /* *+, ++, ?+, upto+ */ \
2, 2, 2, 2, 2, 2, /* *I, *?I, +I, +?I, ?I, ??I ** UTF-8 */ \
2+IMM2_SIZE, 2+IMM2_SIZE, /* upto I, minupto I */ \
2+IMM2_SIZE, /* exact I */ \
2, 2, 2, 2+IMM2_SIZE, /* *+I, ++I, ?+I, upto+I */ \
/* Negative single-char repeats - only for chars < 256 */ \
2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \
2+IMM2_SIZE, 2+IMM2_SIZE, /* NOT upto, minupto */ \
2+IMM2_SIZE, /* NOT exact */ \
2, 2, 2, 2+IMM2_SIZE, /* Possessive NOT *, +, ?, upto */ \
2, 2, 2, 2, 2, 2, /* NOT *I, *?I, +I, +?I, ?I, ??I */ \
2+IMM2_SIZE, 2+IMM2_SIZE, /* NOT upto I, minupto I */ \
2+IMM2_SIZE, /* NOT exact I */ \
2, 2, 2, 2+IMM2_SIZE, /* Possessive NOT *I, +I, ?I, upto I */ \
/* Positive type repeats */ \
2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \
2+IMM2_SIZE, 2+IMM2_SIZE, /* Type upto, minupto */ \
2+IMM2_SIZE, /* Type exact */ \
2, 2, 2, 2+IMM2_SIZE, /* Possessive *+, ++, ?+, upto+ */ \
/* Character class & ref repeats */ \
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \
1+2*IMM2_SIZE, 1+2*IMM2_SIZE, /* CRRANGE, CRMINRANGE */ \
1, 1, 1, 1+2*IMM2_SIZE, /* Possessive *+, ++, ?+, CRPOSRANGE */ \
1+(32/sizeof(PCRE2_UCHAR)), /* CLASS */ \
1+(32/sizeof(PCRE2_UCHAR)), /* NCLASS */ \
0, /* XCLASS - variable length */ \
0, /* ECLASS - variable length */ \
1+IMM2_SIZE, /* REF */ \
1+IMM2_SIZE+1, /* REFI */ \
1+2*IMM2_SIZE, /* DNREF */ \
1+2*IMM2_SIZE+1, /* DNREFI */ \
1+LINK_SIZE, /* RECURSE */ \
1+2*LINK_SIZE+1, /* CALLOUT */ \
0, /* CALLOUT_STR - variable length */ \
1+LINK_SIZE, /* Alt */ \
1+LINK_SIZE, /* Ket */ \
1+LINK_SIZE, /* KetRmax */ \
1+LINK_SIZE, /* KetRmin */ \
1+LINK_SIZE, /* KetRpos */ \
1+IMM2_SIZE, /* Reverse */ \
1+2*IMM2_SIZE, /* VReverse */ \
1+LINK_SIZE, /* Assert */ \
1+LINK_SIZE, /* Assert not */ \
1+LINK_SIZE, /* Assert behind */ \
1+LINK_SIZE, /* Assert behind not */ \
1+LINK_SIZE, /* NA Assert */ \
1+LINK_SIZE, /* NA Assert behind */ \
1+LINK_SIZE, /* Scan substring */ \
1+LINK_SIZE, /* ONCE */ \
1+LINK_SIZE, /* SCRIPT_RUN */ \
1+LINK_SIZE, /* BRA */ \
1+LINK_SIZE, /* BRAPOS */ \
1+LINK_SIZE+IMM2_SIZE, /* CBRA */ \
1+LINK_SIZE+IMM2_SIZE, /* CBRAPOS */ \
1+LINK_SIZE, /* COND */ \
1+LINK_SIZE, /* SBRA */ \
1+LINK_SIZE, /* SBRAPOS */ \
1+LINK_SIZE+IMM2_SIZE, /* SCBRA */ \
1+LINK_SIZE+IMM2_SIZE, /* SCBRAPOS */ \
1+LINK_SIZE, /* SCOND */ \
1+IMM2_SIZE, 1+2*IMM2_SIZE, /* CREF, DNCREF */ \
1+IMM2_SIZE, 1+2*IMM2_SIZE, /* RREF, DNRREF */ \
1, 1, /* FALSE, TRUE */ \
1, 1, 1, /* BRAZERO, BRAMINZERO, BRAPOSZERO */ \
3, 1, 3, /* MARK, PRUNE, PRUNE_ARG */ \
1, 3, /* SKIP, SKIP_ARG */ \
1, 3, /* THEN, THEN_ARG */ \
1, 3, /* COMMIT, COMMIT_ARG */ \
1, 1, 1, /* FAIL, ACCEPT, ASSERT_ACCEPT */ \
1+IMM2_SIZE, 1, /* CLOSE, SKIPZERO */ \
1, /* DEFINE */ \
1, 1 /* \B and \b in UCP mode */
/* A magic value for OP_RREF to indicate the "any recursion" condition. */
#define RREF_ANY 0xffff
/* Constants used by OP_REFI and OP_DNREFI to control matching behaviour. */
#define REFI_FLAG_CASELESS_RESTRICT 0x1
#define REFI_FLAG_TURKISH_CASING 0x2
/* ---------- Private structures that are mode-independent. ---------- */
/* Structure to hold data for custom memory management. */
typedef struct pcre2_memctl {
void * (*malloc)(size_t, void *);
void (*free)(void *, void *);
void *memory_data;
} pcre2_memctl;
/* Structure for building a chain of open capturing subpatterns during
compiling, so that instructions to close them can be compiled when (*ACCEPT) is
encountered. */
typedef struct open_capitem {
struct open_capitem *next; /* Chain link */
uint16_t number; /* Capture number */
uint16_t assert_depth; /* Assertion depth when opened */
} open_capitem;
/* Layout of the UCP type table that translates property names into types and
codes. Each entry used to point directly to a name, but to reduce the number of
relocations in shared libraries, it now has an offset into a single string
instead. */
typedef struct {
uint16_t name_offset;
uint16_t type;
uint16_t value;
} ucp_type_table;
/* Unicode character database (UCD) record format */
typedef struct {
uint8_t script; /* ucp_Arabic, etc. */
uint8_t chartype; /* ucp_Cc, etc. (general categories) */
uint8_t gbprop; /* ucp_gbControl, etc. (grapheme break property) */
uint8_t caseset; /* offset to multichar other cases or zero */
int32_t other_case; /* offset to other case, or zero if none */
uint16_t scriptx_bidiclass; /* script extension (11 bit) and bidi class (5 bit) values */
uint16_t bprops; /* binary properties offset */
} ucd_record;
/* UCD access macros */
#define UCD_BLOCK_SIZE 128
#define REAL_GET_UCD(ch) (PRIV(ucd_records) + \
PRIV(ucd_stage2)[PRIV(ucd_stage1)[(int)(ch) / UCD_BLOCK_SIZE] * \
UCD_BLOCK_SIZE + (int)(ch) % UCD_BLOCK_SIZE])
#if PCRE2_CODE_UNIT_WIDTH == 32
#define GET_UCD(ch) ((ch > MAX_UTF_CODE_POINT)? \
PRIV(dummy_ucd_record) : REAL_GET_UCD(ch))
#else
#define GET_UCD(ch) REAL_GET_UCD(ch)
#endif
#define UCD_SCRIPTX_MASK 0x3ff
#define UCD_BIDICLASS_SHIFT 11
#define UCD_BPROPS_MASK 0xfff
#define UCD_SCRIPTX_PROP(prop) ((prop)->scriptx_bidiclass & UCD_SCRIPTX_MASK)
#define UCD_BIDICLASS_PROP(prop) ((prop)->scriptx_bidiclass >> UCD_BIDICLASS_SHIFT)
#define UCD_BPROPS_PROP(prop) ((prop)->bprops & UCD_BPROPS_MASK)
#define UCD_CHARTYPE(ch) GET_UCD(ch)->chartype
#define UCD_SCRIPT(ch) GET_UCD(ch)->script
#define UCD_CATEGORY(ch) PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]
#define UCD_GRAPHBREAK(ch) GET_UCD(ch)->gbprop
#define UCD_CASESET(ch) GET_UCD(ch)->caseset
#define UCD_OTHERCASE(ch) ((uint32_t)((int)ch + (int)(GET_UCD(ch)->other_case)))
#define UCD_SCRIPTX(ch) UCD_SCRIPTX_PROP(GET_UCD(ch))
#define UCD_BPROPS(ch) UCD_BPROPS_PROP(GET_UCD(ch))
#define UCD_BIDICLASS(ch) UCD_BIDICLASS_PROP(GET_UCD(ch))
#define UCD_ANY_I(ch) \
/* match any of the four characters 'i', 'I', U+0130, U+0131 */ \
(((uint32_t)(ch) | 0x20u) == 0x69u || ((uint32_t)(ch) | 1u) == 0x0131u)
#define UCD_DOTTED_I(ch) \
((uint32_t)(ch) == 0x69u || (uint32_t)(ch) == 0x0130u)
#define UCD_FOLD_I_TURKISH(ch) \
((uint32_t)(ch) == 0x0130u ? 0x69u : \
(uint32_t)(ch) == 0x49u ? 0x0131u : (uint32_t)(ch))
/* The "scriptx" and bprops fields contain offsets into vectors of 32-bit words
that form a bitmap representing a list of scripts or boolean properties. These
macros test or set a bit in the map by number. */
#define MAPBIT(map,n) ((map)[(n)/32]&(1u<<((n)%32)))
#define MAPSET(map,n) ((map)[(n)/32]|=(1u<<((n)%32)))
/* Header for serialized pcre2 codes. */
typedef struct pcre2_serialized_data {
uint32_t magic;
uint32_t version;
uint32_t config;
int32_t number_of_codes;
} pcre2_serialized_data;
/* ----------------- Items that need PCRE2_CODE_UNIT_WIDTH ----------------- */
/* When this file is included by pcre2test, PCRE2_CODE_UNIT_WIDTH is defined as
0, so the following items are omitted. */
#if defined PCRE2_CODE_UNIT_WIDTH && PCRE2_CODE_UNIT_WIDTH != 0
/* EBCDIC is supported only for the 8-bit library. */
#if defined EBCDIC && PCRE2_CODE_UNIT_WIDTH != 8
#error EBCDIC is not supported for the 16-bit or 32-bit libraries
#endif
/* This is the largest non-UTF code point. */
#define MAX_NON_UTF_CHAR (0xffffffffU >> (32 - PCRE2_CODE_UNIT_WIDTH))
/* Internal shared data tables and variables. These are used by more than one
of the exported public functions. They have to be "external" in the C sense,
but are not part of the PCRE2 public API. Although the data for some of them is
identical in all libraries, they must have different names so that multiple
libraries can be simultaneously linked to a single application. However, UTF-8
tables are needed only when compiling the 8-bit library. */
#if PCRE2_CODE_UNIT_WIDTH == 8
extern const int PRIV(utf8_table1)[];
extern const int PRIV(utf8_table1_size);
extern const int PRIV(utf8_table2)[];
extern const int PRIV(utf8_table3)[];
extern const uint8_t PRIV(utf8_table4)[];
#endif
#define _pcre2_OP_lengths PCRE2_SUFFIX(_pcre2_OP_lengths_)
#define _pcre2_callout_end_delims PCRE2_SUFFIX(_pcre2_callout_end_delims_)
#define _pcre2_callout_start_delims PCRE2_SUFFIX(_pcre2_callout_start_delims_)
#define _pcre2_default_compile_context PCRE2_SUFFIX(_pcre2_default_compile_context_)
#define _pcre2_default_convert_context PCRE2_SUFFIX(_pcre2_default_convert_context_)
#define _pcre2_default_match_context PCRE2_SUFFIX(_pcre2_default_match_context_)
#define _pcre2_default_tables PCRE2_SUFFIX(_pcre2_default_tables_)
#if PCRE2_CODE_UNIT_WIDTH == 32
#define _pcre2_dummy_ucd_record PCRE2_SUFFIX(_pcre2_dummy_ucd_record_)
#endif
#define _pcre2_hspace_list PCRE2_SUFFIX(_pcre2_hspace_list_)
#define _pcre2_vspace_list PCRE2_SUFFIX(_pcre2_vspace_list_)
#define _pcre2_ucd_boolprop_sets PCRE2_SUFFIX(_pcre2_ucd_boolprop_sets_)
#define _pcre2_ucd_caseless_sets PCRE2_SUFFIX(_pcre2_ucd_caseless_sets_)
#define _pcre2_ucd_turkish_dotted_i_caseset PCRE2_SUFFIX(_pcre2_ucd_turkish_dotted_i_caseset_)
#define _pcre2_ucd_nocase_ranges PCRE2_SUFFIX(_pcre2_ucd_nocase_ranges_)
#define _pcre2_ucd_nocase_ranges_size PCRE2_SUFFIX(_pcre2_ucd_nocase_ranges_size_)
#define _pcre2_ucd_digit_sets PCRE2_SUFFIX(_pcre2_ucd_digit_sets_)
#define _pcre2_ucd_script_sets PCRE2_SUFFIX(_pcre2_ucd_script_sets_)
#define _pcre2_ucd_records PCRE2_SUFFIX(_pcre2_ucd_records_)
#define _pcre2_ucd_stage1 PCRE2_SUFFIX(_pcre2_ucd_stage1_)
#define _pcre2_ucd_stage2 PCRE2_SUFFIX(_pcre2_ucd_stage2_)
#define _pcre2_ucp_gbtable PCRE2_SUFFIX(_pcre2_ucp_gbtable_)
#define _pcre2_ucp_gentype PCRE2_SUFFIX(_pcre2_ucp_gentype_)
#define _pcre2_ucp_typerange PCRE2_SUFFIX(_pcre2_ucp_typerange_)
#define _pcre2_unicode_version PCRE2_SUFFIX(_pcre2_unicode_version_)
#define _pcre2_utt PCRE2_SUFFIX(_pcre2_utt_)
#define _pcre2_utt_names PCRE2_SUFFIX(_pcre2_utt_names_)
#define _pcre2_utt_size PCRE2_SUFFIX(_pcre2_utt_size_)
extern const uint8_t PRIV(OP_lengths)[];
extern const uint32_t PRIV(callout_end_delims)[];
extern const uint32_t PRIV(callout_start_delims)[];
extern pcre2_compile_context PRIV(default_compile_context);
extern pcre2_convert_context PRIV(default_convert_context);
extern pcre2_match_context PRIV(default_match_context);
extern const uint8_t PRIV(default_tables)[];
extern const uint32_t PRIV(hspace_list)[];
extern const uint32_t PRIV(vspace_list)[];
extern const uint32_t PRIV(ucd_boolprop_sets)[];
extern const uint32_t PRIV(ucd_caseless_sets)[];
extern const uint32_t PRIV(ucd_turkish_dotted_i_caseset);
extern const uint32_t PRIV(ucd_nocase_ranges)[];
extern const uint32_t PRIV(ucd_nocase_ranges_size);
extern const uint32_t PRIV(ucd_digit_sets)[];
extern const uint32_t PRIV(ucd_script_sets)[];
extern const ucd_record PRIV(ucd_records)[];
#if PCRE2_CODE_UNIT_WIDTH == 32
extern const ucd_record PRIV(dummy_ucd_record)[];
#endif
extern const uint16_t PRIV(ucd_stage1)[];
extern const uint16_t PRIV(ucd_stage2)[];
extern const uint32_t PRIV(ucp_gbtable)[];
extern const uint32_t PRIV(ucp_gentype)[];
#ifdef SUPPORT_JIT
extern const int PRIV(ucp_typerange)[];
#endif
extern const char *PRIV(unicode_version);
extern const ucp_type_table PRIV(utt)[];
extern const char PRIV(utt_names)[];
extern const size_t PRIV(utt_size);
/* Mode-dependent macros and hidden and private structures are defined in a
separate file so that pcre2test can include them at all supported widths. When
compiling the library, PCRE2_CODE_UNIT_WIDTH will be defined, and we can
include them at the appropriate width, after setting up suffix macros for the
private structures. */
#define branch_chain PCRE2_SUFFIX(branch_chain_)
#define compile_block PCRE2_SUFFIX(compile_block_)
#define dfa_match_block PCRE2_SUFFIX(dfa_match_block_)
#define match_block PCRE2_SUFFIX(match_block_)
#define named_group PCRE2_SUFFIX(named_group_)
#include "pcre2_intmodedep.h"
/* Private "external" functions. These are internal functions that are called
from modules other than the one in which they are defined. They have to be
"external" in the C sense, but are not part of the PCRE2 public API. They are
not referenced from pcre2test, and must not be defined when no code unit width
is available. */
#define _pcre2_auto_possessify PCRE2_SUFFIX(_pcre2_auto_possessify_)
#define _pcre2_check_escape PCRE2_SUFFIX(_pcre2_check_escape_)
#define _pcre2_extuni PCRE2_SUFFIX(_pcre2_extuni_)
#define _pcre2_find_bracket PCRE2_SUFFIX(_pcre2_find_bracket_)
#define _pcre2_is_newline PCRE2_SUFFIX(_pcre2_is_newline_)
#define _pcre2_jit_free_rodata PCRE2_SUFFIX(_pcre2_jit_free_rodata_)
#define _pcre2_jit_free PCRE2_SUFFIX(_pcre2_jit_free_)
#define _pcre2_jit_get_size PCRE2_SUFFIX(_pcre2_jit_get_size_)
#define _pcre2_jit_get_target PCRE2_SUFFIX(_pcre2_jit_get_target_)
#define _pcre2_memctl_malloc PCRE2_SUFFIX(_pcre2_memctl_malloc_)
#define _pcre2_ord2utf PCRE2_SUFFIX(_pcre2_ord2utf_)
#define _pcre2_script_run PCRE2_SUFFIX(_pcre2_script_run_)
#define _pcre2_strcmp PCRE2_SUFFIX(_pcre2_strcmp_)
#define _pcre2_strcmp_c8 PCRE2_SUFFIX(_pcre2_strcmp_c8_)
#define _pcre2_strcpy_c8 PCRE2_SUFFIX(_pcre2_strcpy_c8_)
#define _pcre2_strlen PCRE2_SUFFIX(_pcre2_strlen_)
#define _pcre2_strncmp PCRE2_SUFFIX(_pcre2_strncmp_)
#define _pcre2_strncmp_c8 PCRE2_SUFFIX(_pcre2_strncmp_c8_)
#define _pcre2_study PCRE2_SUFFIX(_pcre2_study_)
#define _pcre2_valid_utf PCRE2_SUFFIX(_pcre2_valid_utf_)
#define _pcre2_was_newline PCRE2_SUFFIX(_pcre2_was_newline_)
#define _pcre2_xclass PCRE2_SUFFIX(_pcre2_xclass_)
#define _pcre2_eclass PCRE2_SUFFIX(_pcre2_eclass_)
extern int _pcre2_auto_possessify(PCRE2_UCHAR *,
const compile_block *);
extern int _pcre2_check_escape(PCRE2_SPTR *, PCRE2_SPTR, uint32_t *,
int *, uint32_t, uint32_t, uint32_t, BOOL, compile_block *);
extern PCRE2_SPTR _pcre2_extuni(uint32_t, PCRE2_SPTR, PCRE2_SPTR, PCRE2_SPTR,
BOOL, int *);
extern PCRE2_SPTR _pcre2_find_bracket(PCRE2_SPTR, BOOL, int);
extern BOOL _pcre2_is_newline(PCRE2_SPTR, uint32_t, PCRE2_SPTR,
uint32_t *, BOOL);
extern void _pcre2_jit_free_rodata(void *, void *);
extern void _pcre2_jit_free(void *, pcre2_memctl *);
extern size_t _pcre2_jit_get_size(void *);
const char * _pcre2_jit_get_target(void);
extern void * _pcre2_memctl_malloc(size_t, pcre2_memctl *);
extern unsigned int _pcre2_ord2utf(uint32_t, PCRE2_UCHAR *);
extern BOOL _pcre2_script_run(PCRE2_SPTR, PCRE2_SPTR, BOOL);
extern int _pcre2_strcmp(PCRE2_SPTR, PCRE2_SPTR);
extern int _pcre2_strcmp_c8(PCRE2_SPTR, const char *);
extern PCRE2_SIZE _pcre2_strcpy_c8(PCRE2_UCHAR *, const char *);
extern PCRE2_SIZE _pcre2_strlen(PCRE2_SPTR);
extern int _pcre2_strncmp(PCRE2_SPTR, PCRE2_SPTR, size_t);
extern int _pcre2_strncmp_c8(PCRE2_SPTR, const char *, size_t);
extern int _pcre2_study(pcre2_real_code *);
extern int _pcre2_valid_utf(PCRE2_SPTR, PCRE2_SIZE, PCRE2_SIZE *);
extern BOOL _pcre2_was_newline(PCRE2_SPTR, uint32_t, PCRE2_SPTR,
uint32_t *, BOOL);
extern BOOL _pcre2_xclass(uint32_t, PCRE2_SPTR, const uint8_t *, BOOL);
extern BOOL _pcre2_eclass(uint32_t, PCRE2_SPTR, PCRE2_SPTR,
const uint8_t *, BOOL);
/* This function is needed only when memmove() is not available. */
#if !defined(VPCOMPAT) && !defined(HAVE_MEMMOVE)
#define _pcre2_memmove PCRE2_SUFFIX(_pcre2_memmove)
extern void * _pcre2_memmove(void *, const void *, size_t);
#endif
#endif /* PCRE2_CODE_UNIT_WIDTH */
extern BOOL PRIV(ckd_smul)(PCRE2_SIZE *, int, int);
#include "pcre2_util.h"
#endif /* PCRE2_INTERNAL_H_IDEMPOTENT_GUARD */
/* End of pcre2_internal.h */
pcre2-10.45/src/pcre2_study.c 0000644 0001751 0000166 00000165174 14750415257 011446 /*************************************************
* Perl-Compatible Regular Expressions *
*************************************************/
/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.
Written by Philip Hazel
Original API code Copyright (c) 1997-2012 University of Cambridge
New API code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This module contains functions for scanning a compiled pattern and
collecting data (e.g. minimum matching length). */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "pcre2_internal.h"
/* The maximum remembered capturing brackets minimum. */
#define MAX_CACHE_BACKREF 128
/* Set a bit in the starting code unit bit map. */
#define SET_BIT(c) re->start_bitmap[(c)/8] |= (1u << ((c)&7))
/* Returns from set_start_bits() */
enum { SSB_FAIL, SSB_DONE, SSB_CONTINUE, SSB_UNKNOWN, SSB_TOODEEP };
/*************************************************
* Find the minimum subject length for a group *
*************************************************/
/* Scan a parenthesized group and compute the minimum length of subject that
is needed to match it. This is a lower bound; it does not mean there is a
string of that length that matches. In UTF mode, the result is in characters
rather than code units. The field in a compiled pattern for storing the minimum
length is 16-bits long (on the grounds that anything longer than that is
pathological), so we give up when we reach that amount. This also means that
integer overflow for really crazy patterns cannot happen.
Backreference minimum lengths are cached to speed up multiple references. This
function is called only when the highest back reference in the pattern is less
than or equal to MAX_CACHE_BACKREF, which is one less than the size of the
caching vector. The zeroth element contains the number of the highest set
value.
Arguments:
re compiled pattern block
code pointer to start of group (the bracket)
startcode pointer to start of the whole pattern's code
utf UTF flag
recurses chain of recurse_check to catch mutual recursion
countptr pointer to call count (to catch over complexity)
backref_cache vector for caching back references.
This function is no longer called when the pattern contains (*ACCEPT); however,
the old code for returning -1 is retained, just in case.
Returns: the minimum length
-1 \C in UTF-8 mode
or (*ACCEPT)
or pattern too complicated
-2 internal error (missing capturing bracket)
-3 internal error (opcode not listed)
*/
static int
find_minlength(const pcre2_real_code *re, PCRE2_SPTR code,
PCRE2_SPTR startcode, BOOL utf, recurse_check *recurses, int *countptr,
int *backref_cache)
{
int length = -1;
int branchlength = 0;
int prev_cap_recno = -1;
int prev_cap_d = 0;
int prev_recurse_recno = -1;
int prev_recurse_d = 0;
uint32_t once_fudge = 0;
BOOL had_recurse = FALSE;
BOOL dupcapused = (re->flags & PCRE2_DUPCAPUSED) != 0;
PCRE2_SPTR nextbranch = code + GET(code, 1);
PCRE2_SPTR cc = code + 1 + LINK_SIZE;
recurse_check this_recurse;
/* If this is a "could be empty" group, its minimum length is 0. */
if (*code >= OP_SBRA && *code <= OP_SCOND) return 0;
/* Skip over capturing bracket number */
if (*code == OP_CBRA || *code == OP_CBRAPOS) cc += IMM2_SIZE;
/* A large and/or complex regex can take too long to process. */
if ((*countptr)++ > 1000) return -1;
/* Scan along the opcodes for this branch. If we get to the end of the branch,
check the length against that of the other branches. If the accumulated length
passes 16-bits, reset to that value and skip the rest of the branch. */
for (;;)
{
int d, min, recno;
PCRE2_UCHAR op;
PCRE2_SPTR cs, ce;
if (branchlength >= UINT16_MAX)
{
branchlength = UINT16_MAX;
cc = nextbranch;
}
op = *cc;
switch (op)
{
case OP_COND:
case OP_SCOND:
/* If there is only one branch in a condition, the implied branch has zero
length, so we don't add anything. This covers the DEFINE "condition"
automatically. If there are two branches we can treat it the same as any
other non-capturing subpattern. */
cs = cc + GET(cc, 1);
if (*cs != OP_ALT)
{
cc = cs + 1 + LINK_SIZE;
break;
}
goto PROCESS_NON_CAPTURE;
case OP_BRA:
/* There's a special case of OP_BRA, when it is wrapped round a repeated
OP_RECURSE. We'd like to process the latter at this level so that
remembering the value works for repeated cases. So we do nothing, but
set a fudge value to skip over the OP_KET after the recurse. */
if (cc[1+LINK_SIZE] == OP_RECURSE && cc[2*(1+LINK_SIZE)] == OP_KET)
{
once_fudge = 1 + LINK_SIZE;
cc += 1 + LINK_SIZE;
break;
}
/* Fall through */
case OP_ONCE:
case OP_SCRIPT_RUN:
case OP_SBRA:
case OP_BRAPOS:
case OP_SBRAPOS:
PROCESS_NON_CAPTURE:
d = find_minlength(re, cc, startcode, utf, recurses, countptr,
backref_cache);
if (d < 0) return d;
branchlength += d;
do cc += GET(cc, 1); while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* To save time for repeated capturing subpatterns, we remember the
length of the previous one. Unfortunately we can't do the same for
the unnumbered ones above. Nor can we do this if (?| is present in the
pattern because captures with the same number are not then identical. */
case OP_CBRA:
case OP_SCBRA:
case OP_CBRAPOS:
case OP_SCBRAPOS:
recno = (int)GET2(cc, 1+LINK_SIZE);
if (dupcapused || recno != prev_cap_recno)
{
prev_cap_recno = recno;
prev_cap_d = find_minlength(re, cc, startcode, utf, recurses, countptr,
backref_cache);
if (prev_cap_d < 0) return prev_cap_d;
}
branchlength += prev_cap_d;
do cc += GET(cc, 1); while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* ACCEPT makes things far too complicated; we have to give up. In fact,
from 10.34 onwards, if a pattern contains (*ACCEPT), this function is not
used. However, leave the code in place, just in case. */
case OP_ACCEPT:
case OP_ASSERT_ACCEPT:
return -1;
/* Reached end of a branch; if it's a ket it is the end of a nested
call. If it's ALT it is an alternation in a nested call. If it is END it's
the end of the outer call. All can be handled by the same code. If the
length of any branch is zero, there is no need to scan any subsequent
branches. */
case OP_ALT:
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_KETRPOS:
case OP_END:
if (length < 0 || (!had_recurse && branchlength < length))
length = branchlength;
if (op != OP_ALT || length == 0) return length;
nextbranch = cc + GET(cc, 1);
cc += 1 + LINK_SIZE;
branchlength = 0;
had_recurse = FALSE;
break;
/* Skip over assertive subpatterns */
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
case OP_ASSERT_NA:
case OP_ASSERT_SCS:
case OP_ASSERTBACK_NA:
do cc += GET(cc, 1); while (*cc == OP_ALT);
/* Fall through */
/* Skip over things that don't match chars */
case OP_REVERSE:
case OP_VREVERSE:
case OP_CREF:
case OP_DNCREF:
case OP_RREF:
case OP_DNRREF:
case OP_FALSE:
case OP_TRUE:
case OP_CALLOUT:
case OP_SOD:
case OP_SOM:
case OP_EOD:
case OP_EODN:
case OP_CIRC:
case OP_CIRCM:
case OP_DOLL:
case OP_DOLLM:
case OP_NOT_WORD_BOUNDARY:
case OP_WORD_BOUNDARY:
case OP_NOT_UCP_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
cc += PRIV(OP_lengths)[*cc];
break;
case OP_CALLOUT_STR:
cc += GET(cc, 1 + 2*LINK_SIZE);
break;
/* Skip over a subpattern that has a {0} or {0,x} quantifier */
case OP_BRAZERO:
case OP_BRAMINZERO:
case OP_BRAPOSZERO:
case OP_SKIPZERO:
cc += PRIV(OP_lengths)[*cc];
do cc += GET(cc, 1); while (*cc == OP_ALT);
cc += 1 + LINK_SIZE;
break;
/* Handle literal characters and + repetitions */
case OP_CHAR:
case OP_CHARI:
case OP_NOT:
case OP_NOTI:
case OP_PLUS:
case OP_PLUSI:
case OP_MINPLUS:
case OP_MINPLUSI:
case OP_POSPLUS:
case OP_POSPLUSI:
case OP_NOTPLUS:
case OP_NOTPLUSI:
case OP_NOTMINPLUS:
case OP_NOTMINPLUSI:
case OP_NOTPOSPLUS:
case OP_NOTPOSPLUSI:
branchlength++;
cc += 2;
#ifdef SUPPORT_UNICODE
if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
branchlength++;
cc += (cc[1] == OP_PROP || cc[1] == OP_NOTPROP)? 4 : 2;
break;
/* Handle exact repetitions. The count is already in characters, but we
may need to skip over a multibyte character in UTF mode. */
case OP_EXACT:
case OP_EXACTI:
case OP_NOTEXACT:
case OP_NOTEXACTI:
branchlength += GET2(cc,1);
cc += 2 + IMM2_SIZE;
#ifdef SUPPORT_UNICODE
if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
case OP_TYPEEXACT:
branchlength += GET2(cc,1);
cc += 2 + IMM2_SIZE + ((cc[1 + IMM2_SIZE] == OP_PROP
|| cc[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0);
break;
/* Handle single-char non-literal matchers */
case OP_PROP:
case OP_NOTPROP:
cc += 2;
/* Fall through */
case OP_NOT_DIGIT:
case OP_DIGIT:
case OP_NOT_WHITESPACE:
case OP_WHITESPACE:
case OP_NOT_WORDCHAR:
case OP_WORDCHAR:
case OP_ANY:
case OP_ALLANY:
case OP_EXTUNI:
case OP_HSPACE:
case OP_NOT_HSPACE:
case OP_VSPACE:
case OP_NOT_VSPACE:
branchlength++;
cc++;
break;
/* "Any newline" might match two characters, but it also might match just
one. */
case OP_ANYNL:
branchlength += 1;
cc++;
break;
/* The single-byte matcher means we can't proceed in UTF mode. (In
non-UTF mode \C will actually be turned into OP_ALLANY, so won't ever
appear, but leave the code, just in case.) */
case OP_ANYBYTE:
#ifdef SUPPORT_UNICODE
if (utf) return -1;
#endif
branchlength++;
cc++;
break;
/* For repeated character types, we have to test for \p and \P, which have
an extra two bytes of parameters. */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSSTAR:
case OP_TYPEPOSQUERY:
if (cc[1] == OP_PROP || cc[1] == OP_NOTPROP) cc += 2;
cc += PRIV(OP_lengths)[op];
break;
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
if (cc[1 + IMM2_SIZE] == OP_PROP
|| cc[1 + IMM2_SIZE] == OP_NOTPROP) cc += 2;
cc += PRIV(OP_lengths)[op];
break;
/* Check a class for variable quantification */
case OP_CLASS:
case OP_NCLASS:
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
case OP_ECLASS:
/* The original code caused an unsigned overflow in 64 bit systems,
so now we use a conditional statement. */
if (op == OP_XCLASS || op == OP_ECLASS)
cc += GET(cc, 1);
else
#endif
cc += PRIV(OP_lengths)[OP_CLASS];
switch (*cc)
{
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRPOSPLUS:
branchlength++;
/* Fall through */
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSQUERY:
cc++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
branchlength += GET2(cc,1);
cc += 1 + 2 * IMM2_SIZE;
break;
default:
branchlength++;
break;
}
break;
/* Backreferences and subroutine calls (OP_RECURSE) are treated in the same
way: we find the minimum length for the subpattern. A recursion
(backreference or subroutine) causes an a flag to be set that causes the
length of this branch to be ignored. The logic is that a recursion can only
make sense if there is another alternative that stops the recursing. That
will provide the minimum length (when no recursion happens).
If PCRE2_MATCH_UNSET_BACKREF is set, a backreference to an unset bracket
matches an empty string (by default it causes a matching failure), so in
that case we must set the minimum length to zero.
For backreferenes, if duplicate numbers are present in the pattern we check
for a reference to a duplicate. If it is, we don't know which version will
be referenced, so we have to set the minimum length to zero. */
/* Duplicate named pattern back reference. */
case OP_DNREF:
case OP_DNREFI:
if (!dupcapused && (re->overall_options & PCRE2_MATCH_UNSET_BACKREF) == 0)
{
int count = GET2(cc, 1+IMM2_SIZE);
PCRE2_SPTR slot =
(PCRE2_SPTR)((const uint8_t *)re + sizeof(pcre2_real_code)) +
GET2(cc, 1) * re->name_entry_size;
d = INT_MAX;
/* Scan all groups with the same name; find the shortest. */
while (count-- > 0)
{
int dd, i;
recno = GET2(slot, 0);
if (recno <= backref_cache[0] && backref_cache[recno] >= 0)
dd = backref_cache[recno];
else
{
ce = cs = PRIV(find_bracket)(startcode, utf, recno);
if (cs == NULL) return -2;
do ce += GET(ce, 1); while (*ce == OP_ALT);
dd = 0;
if (!dupcapused || PRIV(find_bracket)(ce, utf, recno) == NULL)
{
if (cc > cs && cc < ce) /* Simple recursion */
{
had_recurse = TRUE;
}
else
{
recurse_check *r = recurses;
for (r = recurses; r != NULL; r = r->prev)
if (r->group == cs) break;
if (r != NULL) /* Mutual recursion */
{
had_recurse = TRUE;
}
else
{
this_recurse.prev = recurses; /* No recursion */
this_recurse.group = cs;
dd = find_minlength(re, cs, startcode, utf, &this_recurse,
countptr, backref_cache);
if (dd < 0) return dd;
}
}
}
backref_cache[recno] = dd;
for (i = backref_cache[0] + 1; i < recno; i++) backref_cache[i] = -1;
backref_cache[0] = recno;
}
if (dd < d) d = dd;
if (d <= 0) break; /* No point looking at any more */
slot += re->name_entry_size;
}
}
else d = 0;
cc += PRIV(OP_lengths)[*cc];
goto REPEAT_BACK_REFERENCE;
/* Single back reference by number. References by name are converted to by
number when there is no duplication. */
case OP_REF:
case OP_REFI:
recno = GET2(cc, 1);
if (recno <= backref_cache[0] && backref_cache[recno] >= 0)
d = backref_cache[recno];
else
{
int i;
d = 0;
if ((re->overall_options & PCRE2_MATCH_UNSET_BACKREF) == 0)
{
ce = cs = PRIV(find_bracket)(startcode, utf, recno);
if (cs == NULL) return -2;
do ce += GET(ce, 1); while (*ce == OP_ALT);
if (!dupcapused || PRIV(find_bracket)(ce, utf, recno) == NULL)
{
if (cc > cs && cc < ce) /* Simple recursion */
{
had_recurse = TRUE;
}
else
{
recurse_check *r = recurses;
for (r = recurses; r != NULL; r = r->prev) if (r->group == cs) break;
if (r != NULL) /* Mutual recursion */
{
had_recurse = TRUE;
}
else /* No recursion */
{
this_recurse.prev = recurses;
this_recurse.group = cs;
d = find_minlength(re, cs, startcode, utf, &this_recurse, countptr,
backref_cache);
if (d < 0) return d;
}
}
}
}
backref_cache[recno] = d;
for (i = backref_cache[0] + 1; i < recno; i++) backref_cache[i] = -1;
backref_cache[0] = recno;
}
cc += PRIV(OP_lengths)[*cc];
/* Handle repeated back references */
REPEAT_BACK_REFERENCE:
switch (*cc)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSQUERY:
min = 0;
cc++;
break;
case OP_CRPLUS:
case OP_CRMINPLUS:
case OP_CRPOSPLUS:
min = 1;
cc++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
min = GET2(cc, 1);
cc += 1 + 2 * IMM2_SIZE;
break;
default:
min = 1;
break;
}
/* Take care not to overflow: (1) min and d are ints, so check that their
product is not greater than INT_MAX. (2) branchlength is limited to
UINT16_MAX (checked at the top of the loop). */
if ((d > 0 && (INT_MAX/d) < min) || UINT16_MAX - branchlength < min*d)
branchlength = UINT16_MAX;
else branchlength += min * d;
break;
/* Recursion always refers to the first occurrence of a subpattern with a
given number. Therefore, we can always make use of caching, even when the
pattern contains multiple subpatterns with the same number. */
case OP_RECURSE:
cs = ce = startcode + GET(cc, 1);
recno = GET2(cs, 1+LINK_SIZE);
if (recno == prev_recurse_recno)
{
branchlength += prev_recurse_d;
}
else
{
do ce += GET(ce, 1); while (*ce == OP_ALT);
if (cc > cs && cc < ce) /* Simple recursion */
had_recurse = TRUE;
else
{
recurse_check *r = recurses;
for (r = recurses; r != NULL; r = r->prev) if (r->group == cs) break;
if (r != NULL) /* Mutual recursion */
had_recurse = TRUE;
else
{
this_recurse.prev = recurses;
this_recurse.group = cs;
prev_recurse_d = find_minlength(re, cs, startcode, utf, &this_recurse,
countptr, backref_cache);
if (prev_recurse_d < 0) return prev_recurse_d;
prev_recurse_recno = recno;
branchlength += prev_recurse_d;
}
}
}
cc += 1 + LINK_SIZE + once_fudge;
once_fudge = 0;
break;
/* Anything else does not or need not match a character. We can get the
item's length from the table, but for those that can match zero occurrences
of a character, we must take special action for UTF-8 characters. As it
happens, the "NOT" versions of these opcodes are used at present only for
ASCII characters, so they could be omitted from this list. However, in
future that may change, so we include them here so as not to leave a
gotcha for a future maintainer. */
case OP_UPTO:
case OP_UPTOI:
case OP_NOTUPTO:
case OP_NOTUPTOI:
case OP_MINUPTO:
case OP_MINUPTOI:
case OP_NOTMINUPTO:
case OP_NOTMINUPTOI:
case OP_POSUPTO:
case OP_POSUPTOI:
case OP_NOTPOSUPTO:
case OP_NOTPOSUPTOI:
case OP_STAR:
case OP_STARI:
case OP_NOTSTAR:
case OP_NOTSTARI:
case OP_MINSTAR:
case OP_MINSTARI:
case OP_NOTMINSTAR:
case OP_NOTMINSTARI:
case OP_POSSTAR:
case OP_POSSTARI:
case OP_NOTPOSSTAR:
case OP_NOTPOSSTARI:
case OP_QUERY:
case OP_QUERYI:
case OP_NOTQUERY:
case OP_NOTQUERYI:
case OP_MINQUERY:
case OP_MINQUERYI:
case OP_NOTMINQUERY:
case OP_NOTMINQUERYI:
case OP_POSQUERY:
case OP_POSQUERYI:
case OP_NOTPOSQUERY:
case OP_NOTPOSQUERYI:
cc += PRIV(OP_lengths)[op];
#ifdef SUPPORT_UNICODE
if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
#endif
break;
/* Skip these, but we need to add in the name length. */
case OP_MARK:
case OP_COMMIT_ARG:
case OP_PRUNE_ARG:
case OP_SKIP_ARG:
case OP_THEN_ARG:
cc += PRIV(OP_lengths)[op] + cc[1];
break;
/* The remaining opcodes are just skipped over. */
case OP_CLOSE:
case OP_COMMIT:
case OP_FAIL:
case OP_PRUNE:
case OP_SET_SOM:
case OP_SKIP:
case OP_THEN:
cc += PRIV(OP_lengths)[op];
break;
/* This should not occur: we list all opcodes explicitly so that when
new ones get added they are properly considered. */
default:
PCRE2_DEBUG_UNREACHABLE();
return -3;
}
}
PCRE2_DEBUG_UNREACHABLE(); /* Control should never reach here */
return -3; /* Avoid compiler warnings */
}
/*************************************************
* Set a bit and maybe its alternate case *
*************************************************/
/* Given a character, set its first code unit's bit in the table, and also the
corresponding bit for the other version of a letter if we are caseless.
Arguments:
re points to the regex block
p points to the first code unit of the character
caseless TRUE if caseless
utf TRUE for UTF mode
ucp TRUE for UCP mode
Returns: pointer after the character
*/
static PCRE2_SPTR
set_table_bit(pcre2_real_code *re, PCRE2_SPTR p, BOOL caseless, BOOL utf,
BOOL ucp)
{
uint32_t c = *p++; /* First code unit */
(void)utf; /* Stop compiler warnings when UTF not supported */
(void)ucp;
/* In 16-bit and 32-bit modes, code units greater than 0xff set the bit for
0xff. */
#if PCRE2_CODE_UNIT_WIDTH != 8
if (c > 0xff) SET_BIT(0xff); else
#endif
SET_BIT(c);
/* In UTF-8 or UTF-16 mode, pick up the remaining code units in order to find
the end of the character, even when caseless. */
#ifdef SUPPORT_UNICODE
if (utf)
{
#if PCRE2_CODE_UNIT_WIDTH == 8
if (c >= 0xc0) GETUTF8INC(c, p);
#elif PCRE2_CODE_UNIT_WIDTH == 16
if ((c & 0xfc00) == 0xd800) GETUTF16INC(c, p);
#endif
}
#endif /* SUPPORT_UNICODE */
/* If caseless, handle the other case of the character. */
if (caseless)
{
#ifdef SUPPORT_UNICODE
if (utf || ucp)
{
c = UCD_OTHERCASE(c);
#if PCRE2_CODE_UNIT_WIDTH == 8
if (utf)
{
PCRE2_UCHAR buff[6];
(void)PRIV(ord2utf)(c, buff);
SET_BIT(buff[0]);
}
else if (c < 256) SET_BIT(c);
#else /* 16-bit or 32-bit mode */
if (c > 0xff) SET_BIT(0xff); else SET_BIT(c);
#endif
}
else
#endif /* SUPPORT_UNICODE */
/* Not UTF or UCP */
if (MAX_255(c)) SET_BIT(re->tables[fcc_offset + c]);
}
return p;
}
/*************************************************
* Set bits for a positive character type *
*************************************************/
/* This function sets starting bits for a character type. In UTF-8 mode, we can
only do a direct setting for bytes less than 128, as otherwise there can be
confusion with bytes in the middle of UTF-8 characters. In a "traditional"
environment, the tables will only recognize ASCII characters anyway, but in at
least one Windows environment, some higher bytes bits were set in the tables.
So we deal with that case by considering the UTF-8 encoding.
Arguments:
re the regex block
cbit type the type of character wanted
table_limit 32 for non-UTF-8; 16 for UTF-8
Returns: nothing
*/
static void
set_type_bits(pcre2_real_code *re, int cbit_type, unsigned int table_limit)
{
uint32_t c;
for (c = 0; c < table_limit; c++)
re->start_bitmap[c] |= re->tables[c+cbits_offset+cbit_type];
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (table_limit == 32) return;
for (c = 128; c < 256; c++)
{
if ((re->tables[cbits_offset + c/8] & (1u << (c&7))) != 0)
{
PCRE2_UCHAR buff[6];
(void)PRIV(ord2utf)(c, buff);
SET_BIT(buff[0]);
}
}
#endif /* UTF-8 */
}
/*************************************************
* Set bits for a negative character type *
*************************************************/
/* This function sets starting bits for a negative character type such as \D.
In UTF-8 mode, we can only do a direct setting for bytes less than 128, as
otherwise there can be confusion with bytes in the middle of UTF-8 characters.
Unlike in the positive case, where we can set appropriate starting bits for
specific high-valued UTF-8 characters, in this case we have to set the bits for
all high-valued characters. The lowest is 0xc2, but we overkill by starting at
0xc0 (192) for simplicity.
Arguments:
re the regex block
cbit type the type of character wanted
table_limit 32 for non-UTF-8; 16 for UTF-8
Returns: nothing
*/
static void
set_nottype_bits(pcre2_real_code *re, int cbit_type, unsigned int table_limit)
{
uint32_t c;
for (c = 0; c < table_limit; c++)
re->start_bitmap[c] |= (uint8_t)(~(re->tables[c+cbits_offset+cbit_type]));
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (table_limit != 32) for (c = 24; c < 32; c++) re->start_bitmap[c] = 0xff;
#endif
}
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
/*************************************************
* Set starting bits for a character list. *
*************************************************/
/* This function sets starting bits for a character list. It enumerates
all characters and character ranges in the character list, and sets
the starting bits accordingly.
Arguments:
code pointer to the code
start_bitmap pointer to the starting bitmap
Returns: nothing
*/
static void
study_char_list(PCRE2_SPTR code, uint8_t *start_bitmap,
const uint8_t *char_lists_end)
{
uint32_t type, list_ind;
uint32_t char_list_add = XCL_CHAR_LIST_LOW_16_ADD;
uint32_t range_start = ~(uint32_t)0, range_end = 0;
const uint8_t *next_char;
PCRE2_UCHAR start_buffer[6], end_buffer[6];
PCRE2_UCHAR start, end;
/* Only needed in 8-bit mode at the moment. */
type = (uint32_t)(code[0] << 8) | code[1];
code += 2;
/* Align characters. */
next_char = char_lists_end - (GET(code, 0) << 1);
type &= XCL_TYPE_MASK;
list_ind = 0;
if ((type & XCL_BEGIN_WITH_RANGE) != 0)
range_start = XCL_CHAR_LIST_LOW_16_START;
while (type > 0)
{
uint32_t item_count = type & XCL_ITEM_COUNT_MASK;
if (item_count == XCL_ITEM_COUNT_MASK)
{
if (list_ind <= 1)
{
item_count = *(const uint16_t*)next_char;
next_char += 2;
}
else
{
item_count = *(const uint32_t*)next_char;
next_char += 4;
}
}
while (item_count > 0)
{
if (list_ind <= 1)
{
range_end = *(const uint16_t*)next_char;
next_char += 2;
}
else
{
range_end = *(const uint32_t*)next_char;
next_char += 4;
}
if ((range_end & XCL_CHAR_END) != 0)
{
range_end = char_list_add + (range_end >> XCL_CHAR_SHIFT);
PRIV(ord2utf)(range_end, end_buffer);
end = end_buffer[0];
if (range_start < range_end)
{
PRIV(ord2utf)(range_start, start_buffer);
for (start = start_buffer[0]; start <= end; start++)
start_bitmap[start / 8] |= (1u << (start & 7));
}
else
start_bitmap[end / 8] |= (1u << (end & 7));
range_start = ~(uint32_t)0;
}
else
range_start = char_list_add + (range_end >> XCL_CHAR_SHIFT);
item_count--;
}
list_ind++;
type >>= XCL_TYPE_BIT_LEN;
if (range_start == ~(uint32_t)0)
{
if ((type & XCL_BEGIN_WITH_RANGE) != 0)
{
/* In 8 bit mode XCL_CHAR_LIST_HIGH_32_START is not possible. */
if (list_ind == 1) range_start = XCL_CHAR_LIST_HIGH_16_START;
else range_start = XCL_CHAR_LIST_LOW_32_START;
}
}
else if ((type & XCL_BEGIN_WITH_RANGE) == 0)
{
PRIV(ord2utf)(range_start, start_buffer);
/* In 8 bit mode XCL_CHAR_LIST_LOW_32_END and
XCL_CHAR_LIST_HIGH_32_END are not possible. */
if (list_ind == 1) range_end = XCL_CHAR_LIST_LOW_16_END;
else range_end = XCL_CHAR_LIST_HIGH_16_END;
PRIV(ord2utf)(range_end, end_buffer);
end = end_buffer[0];
for (start = start_buffer[0]; start <= end; start++)
start_bitmap[start / 8] |= (1u << (start & 7));
range_start = ~(uint32_t)0;
}
/* In 8 bit mode XCL_CHAR_LIST_HIGH_32_ADD is not possible. */
if (list_ind == 1) char_list_add = XCL_CHAR_LIST_HIGH_16_ADD;
else char_list_add = XCL_CHAR_LIST_LOW_32_ADD;
}
}
#endif
/*************************************************
* Create bitmap of starting code units *
*************************************************/
/* This function scans a compiled unanchored expression recursively and
attempts to build a bitmap of the set of possible starting code units whose
values are less than 256. In 16-bit and 32-bit mode, values above 255 all cause
the 255 bit to be set. When calling set[_not]_type_bits() in UTF-8 (sic) mode
we pass a value of 16 rather than 32 as the final argument. (See comments in
those functions for the reason.)
The SSB_CONTINUE return is useful for parenthesized groups in patterns such as
(a*)b where the group provides some optional starting code units but scanning
must continue at the outer level to find at least one mandatory code unit. At
the outermost level, this function fails unless the result is SSB_DONE.
We restrict recursion (for nested groups) to 1000 to avoid stack overflow
issues.
Arguments:
re points to the compiled regex block
code points to an expression
utf TRUE if in UTF mode
ucp TRUE if in UCP mode
depthptr pointer to recurse depth
Returns: SSB_FAIL => Failed to find any starting code units
SSB_DONE => Found mandatory starting code units
SSB_CONTINUE => Found optional starting code units
SSB_UNKNOWN => Hit an unrecognized opcode
SSB_TOODEEP => Recursion is too deep
*/
static int
set_start_bits(pcre2_real_code *re, PCRE2_SPTR code, BOOL utf, BOOL ucp,
int *depthptr)
{
uint32_t c;
int yield = SSB_DONE;
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
int table_limit = utf? 16:32;
#else
int table_limit = 32;
#endif
*depthptr += 1;
if (*depthptr > 1000) return SSB_TOODEEP;
do
{
BOOL try_next = TRUE;
PCRE2_SPTR tcode = code + 1 + LINK_SIZE;
if (*code == OP_CBRA || *code == OP_SCBRA ||
*code == OP_CBRAPOS || *code == OP_SCBRAPOS) tcode += IMM2_SIZE;
while (try_next) /* Loop for items in this branch */
{
int rc;
PCRE2_SPTR ncode;
const uint8_t *classmap = NULL;
#ifdef SUPPORT_WIDE_CHARS
PCRE2_UCHAR xclassflags;
#endif
switch(*tcode)
{
/* If we reach something we don't understand, it means a new opcode has
been created that hasn't been added to this function. Hopefully this
problem will be discovered during testing. */
default:
return SSB_UNKNOWN;
/* Fail for a valid opcode that implies no starting bits. */
case OP_ACCEPT:
case OP_ASSERT_ACCEPT:
case OP_ALLANY:
case OP_ANY:
case OP_ANYBYTE:
case OP_CIRCM:
case OP_CLOSE:
case OP_COMMIT:
case OP_COMMIT_ARG:
case OP_COND:
case OP_CREF:
case OP_FALSE:
case OP_TRUE:
case OP_DNCREF:
case OP_DNREF:
case OP_DNREFI:
case OP_DNRREF:
case OP_DOLL:
case OP_DOLLM:
case OP_END:
case OP_EOD:
case OP_EODN:
case OP_EXTUNI:
case OP_FAIL:
case OP_MARK:
case OP_NOT:
case OP_NOTEXACT:
case OP_NOTEXACTI:
case OP_NOTI:
case OP_NOTMINPLUS:
case OP_NOTMINPLUSI:
case OP_NOTMINQUERY:
case OP_NOTMINQUERYI:
case OP_NOTMINSTAR:
case OP_NOTMINSTARI:
case OP_NOTMINUPTO:
case OP_NOTMINUPTOI:
case OP_NOTPLUS:
case OP_NOTPLUSI:
case OP_NOTPOSPLUS:
case OP_NOTPOSPLUSI:
case OP_NOTPOSQUERY:
case OP_NOTPOSQUERYI:
case OP_NOTPOSSTAR:
case OP_NOTPOSSTARI:
case OP_NOTPOSUPTO:
case OP_NOTPOSUPTOI:
case OP_NOTPROP:
case OP_NOTQUERY:
case OP_NOTQUERYI:
case OP_NOTSTAR:
case OP_NOTSTARI:
case OP_NOTUPTO:
case OP_NOTUPTOI:
case OP_NOT_HSPACE:
case OP_NOT_VSPACE:
case OP_PRUNE:
case OP_PRUNE_ARG:
case OP_RECURSE:
case OP_REF:
case OP_REFI:
case OP_REVERSE:
case OP_VREVERSE:
case OP_RREF:
case OP_SCOND:
case OP_SET_SOM:
case OP_SKIP:
case OP_SKIP_ARG:
case OP_SOD:
case OP_SOM:
case OP_THEN:
case OP_THEN_ARG:
return SSB_FAIL;
/* OP_CIRC happens only at the start of an anchored branch (multiline ^
uses OP_CIRCM). Skip over it. */
case OP_CIRC:
tcode += PRIV(OP_lengths)[OP_CIRC];
break;
/* A "real" property test implies no starting bits, but the fake property
PT_CLIST identifies a list of characters. These lists are short, as they
are used for characters with more than one "other case", so there is no
point in recognizing them for OP_NOTPROP. */
case OP_PROP:
if (tcode[1] != PT_CLIST) return SSB_FAIL;
{
const uint32_t *p = PRIV(ucd_caseless_sets) + tcode[2];
while ((c = *p++) < NOTACHAR)
{
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (utf)
{
PCRE2_UCHAR buff[6];
(void)PRIV(ord2utf)(c, buff);
c = buff[0];
}
#endif
if (c > 0xff) SET_BIT(0xff); else SET_BIT(c);
}
}
try_next = FALSE;
break;
/* We can ignore word boundary tests. */
case OP_WORD_BOUNDARY:
case OP_NOT_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
case OP_NOT_UCP_WORD_BOUNDARY:
tcode++;
break;
/* For a positive lookahead assertion, inspect what immediately follows,
ignoring intermediate assertions and callouts. If the next item is one
that sets a mandatory character, skip this assertion. Otherwise, treat it
the same as other bracket groups. */
case OP_ASSERT:
case OP_ASSERT_NA:
ncode = tcode + GET(tcode, 1);
while (*ncode == OP_ALT) ncode += GET(ncode, 1);
ncode += 1 + LINK_SIZE;
/* Skip irrelevant items */
for (BOOL done = FALSE; !done;)
{
switch (*ncode)
{
case OP_ASSERT:
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
case OP_ASSERT_NA:
case OP_ASSERTBACK_NA:
case OP_ASSERT_SCS:
ncode += GET(ncode, 1);
while (*ncode == OP_ALT) ncode += GET(ncode, 1);
ncode += 1 + LINK_SIZE;
break;
case OP_WORD_BOUNDARY:
case OP_NOT_WORD_BOUNDARY:
case OP_UCP_WORD_BOUNDARY:
case OP_NOT_UCP_WORD_BOUNDARY:
ncode++;
break;
case OP_CALLOUT:
ncode += PRIV(OP_lengths)[OP_CALLOUT];
break;
case OP_CALLOUT_STR:
ncode += GET(ncode, 1 + 2*LINK_SIZE);
break;
default:
done = TRUE;
break;
}
}
/* Now check the next significant item. */
switch(*ncode)
{
default:
break;
case OP_PROP:
if (ncode[1] != PT_CLIST) break;
/* Fall through */
case OP_ANYNL:
case OP_CHAR:
case OP_CHARI:
case OP_EXACT:
case OP_EXACTI:
case OP_HSPACE:
case OP_MINPLUS:
case OP_MINPLUSI:
case OP_PLUS:
case OP_PLUSI:
case OP_POSPLUS:
case OP_POSPLUSI:
case OP_VSPACE:
/* Note that these types will only be present in non-UCP mode. */
case OP_DIGIT:
case OP_NOT_DIGIT:
case OP_WORDCHAR:
case OP_NOT_WORDCHAR:
case OP_WHITESPACE:
case OP_NOT_WHITESPACE:
tcode = ncode;
continue; /* With the following significant opcode */
}
/* Fall through */
/* For a group bracket or a positive assertion without an immediately
following mandatory setting, recurse to set bits from within the
subpattern. If it can't find anything, we have to give up. If it finds
some mandatory character(s), we are done for this branch. Otherwise,
carry on scanning after the subpattern. */
case OP_BRA:
case OP_SBRA:
case OP_CBRA:
case OP_SCBRA:
case OP_BRAPOS:
case OP_SBRAPOS:
case OP_CBRAPOS:
case OP_SCBRAPOS:
case OP_ONCE:
case OP_SCRIPT_RUN:
rc = set_start_bits(re, tcode, utf, ucp, depthptr);
if (rc == SSB_DONE)
{
try_next = FALSE;
}
else if (rc == SSB_CONTINUE)
{
do tcode += GET(tcode, 1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
}
else return rc; /* FAIL, UNKNOWN, or TOODEEP */
break;
/* If we hit ALT or KET, it means we haven't found anything mandatory in
this branch, though we might have found something optional. For ALT, we
continue with the next alternative, but we have to arrange that the final
result from subpattern is SSB_CONTINUE rather than SSB_DONE. For KET,
return SSB_CONTINUE: if this is the top level, that indicates failure,
but after a nested subpattern, it causes scanning to continue. */
case OP_ALT:
yield = SSB_CONTINUE;
try_next = FALSE;
break;
case OP_KET:
case OP_KETRMAX:
case OP_KETRMIN:
case OP_KETRPOS:
return SSB_CONTINUE;
/* Skip over callout */
case OP_CALLOUT:
tcode += PRIV(OP_lengths)[OP_CALLOUT];
break;
case OP_CALLOUT_STR:
tcode += GET(tcode, 1 + 2*LINK_SIZE);
break;
/* Skip over lookbehind, negative lookahead, and scan substring
assertions */
case OP_ASSERT_NOT:
case OP_ASSERTBACK:
case OP_ASSERTBACK_NOT:
case OP_ASSERTBACK_NA:
case OP_ASSERT_SCS:
do tcode += GET(tcode, 1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* BRAZERO does the bracket, but carries on. */
case OP_BRAZERO:
case OP_BRAMINZERO:
case OP_BRAPOSZERO:
rc = set_start_bits(re, ++tcode, utf, ucp, depthptr);
if (rc == SSB_FAIL || rc == SSB_UNKNOWN || rc == SSB_TOODEEP) return rc;
do tcode += GET(tcode,1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* SKIPZERO skips the bracket. */
case OP_SKIPZERO:
tcode++;
do tcode += GET(tcode,1); while (*tcode == OP_ALT);
tcode += 1 + LINK_SIZE;
break;
/* Single-char * or ? sets the bit and tries the next item */
case OP_STAR:
case OP_MINSTAR:
case OP_POSSTAR:
case OP_QUERY:
case OP_MINQUERY:
case OP_POSQUERY:
tcode = set_table_bit(re, tcode + 1, FALSE, utf, ucp);
break;
case OP_STARI:
case OP_MINSTARI:
case OP_POSSTARI:
case OP_QUERYI:
case OP_MINQUERYI:
case OP_POSQUERYI:
tcode = set_table_bit(re, tcode + 1, TRUE, utf, ucp);
break;
/* Single-char upto sets the bit and tries the next */
case OP_UPTO:
case OP_MINUPTO:
case OP_POSUPTO:
tcode = set_table_bit(re, tcode + 1 + IMM2_SIZE, FALSE, utf, ucp);
break;
case OP_UPTOI:
case OP_MINUPTOI:
case OP_POSUPTOI:
tcode = set_table_bit(re, tcode + 1 + IMM2_SIZE, TRUE, utf, ucp);
break;
/* At least one single char sets the bit and stops */
case OP_EXACT:
tcode += IMM2_SIZE;
/* Fall through */
case OP_CHAR:
case OP_PLUS:
case OP_MINPLUS:
case OP_POSPLUS:
(void)set_table_bit(re, tcode + 1, FALSE, utf, ucp);
try_next = FALSE;
break;
case OP_EXACTI:
tcode += IMM2_SIZE;
/* Fall through */
case OP_CHARI:
case OP_PLUSI:
case OP_MINPLUSI:
case OP_POSPLUSI:
(void)set_table_bit(re, tcode + 1, TRUE, utf, ucp);
try_next = FALSE;
break;
/* Special spacing and line-terminating items. These recognize specific
lists of characters. The difference between VSPACE and ANYNL is that the
latter can match the two-character CRLF sequence, but that is not
relevant for finding the first character, so their code here is
identical. */
case OP_HSPACE:
SET_BIT(CHAR_HT);
SET_BIT(CHAR_SPACE);
/* For the 16-bit and 32-bit libraries (which can never be EBCDIC), set
the bits for 0xA0 and for code units >= 255, independently of UTF. */
#if PCRE2_CODE_UNIT_WIDTH != 8
SET_BIT(0xA0);
SET_BIT(0xFF);
#else
/* For the 8-bit library in UTF-8 mode, set the bits for the first code
units of horizontal space characters. */
#ifdef SUPPORT_UNICODE
if (utf)
{
SET_BIT(0xC2); /* For U+00A0 */
SET_BIT(0xE1); /* For U+1680, U+180E */
SET_BIT(0xE2); /* For U+2000 - U+200A, U+202F, U+205F */
SET_BIT(0xE3); /* For U+3000 */
}
else
#endif
/* For the 8-bit library not in UTF-8 mode, set the bit for 0xA0, unless
the code is EBCDIC. */
{
#ifndef EBCDIC
SET_BIT(0xA0);
#endif /* Not EBCDIC */
}
#endif /* 8-bit support */
try_next = FALSE;
break;
case OP_ANYNL:
case OP_VSPACE:
SET_BIT(CHAR_LF);
SET_BIT(CHAR_VT);
SET_BIT(CHAR_FF);
SET_BIT(CHAR_CR);
/* For the 16-bit and 32-bit libraries (which can never be EBCDIC), set
the bits for NEL and for code units >= 255, independently of UTF. */
#if PCRE2_CODE_UNIT_WIDTH != 8
SET_BIT(CHAR_NEL);
SET_BIT(0xFF);
#else
/* For the 8-bit library in UTF-8 mode, set the bits for the first code
units of vertical space characters. */
#ifdef SUPPORT_UNICODE
if (utf)
{
SET_BIT(0xC2); /* For U+0085 (NEL) */
SET_BIT(0xE2); /* For U+2028, U+2029 */
}
else
#endif
/* For the 8-bit library not in UTF-8 mode, set the bit for NEL. */
{
SET_BIT(CHAR_NEL);
}
#endif /* 8-bit support */
try_next = FALSE;
break;
/* Single character types set the bits and stop. Note that if PCRE2_UCP
is set, we do not see these opcodes because \d etc are converted to
properties. Therefore, these apply in the case when only characters less
than 256 are recognized to match the types. */
case OP_NOT_DIGIT:
set_nottype_bits(re, cbit_digit, table_limit);
try_next = FALSE;
break;
case OP_DIGIT:
set_type_bits(re, cbit_digit, table_limit);
try_next = FALSE;
break;
case OP_NOT_WHITESPACE:
set_nottype_bits(re, cbit_space, table_limit);
try_next = FALSE;
break;
case OP_WHITESPACE:
set_type_bits(re, cbit_space, table_limit);
try_next = FALSE;
break;
case OP_NOT_WORDCHAR:
set_nottype_bits(re, cbit_word, table_limit);
try_next = FALSE;
break;
case OP_WORDCHAR:
set_type_bits(re, cbit_word, table_limit);
try_next = FALSE;
break;
/* One or more character type fudges the pointer and restarts, knowing
it will hit a single character type and stop there. */
case OP_TYPEPLUS:
case OP_TYPEMINPLUS:
case OP_TYPEPOSPLUS:
tcode++;
break;
case OP_TYPEEXACT:
tcode += 1 + IMM2_SIZE;
break;
/* Zero or more repeats of character types set the bits and then
try again. */
case OP_TYPEUPTO:
case OP_TYPEMINUPTO:
case OP_TYPEPOSUPTO:
tcode += IMM2_SIZE; /* Fall through */
case OP_TYPESTAR:
case OP_TYPEMINSTAR:
case OP_TYPEPOSSTAR:
case OP_TYPEQUERY:
case OP_TYPEMINQUERY:
case OP_TYPEPOSQUERY:
switch(tcode[1])
{
default:
case OP_ANY:
case OP_ALLANY:
return SSB_FAIL;
case OP_HSPACE:
SET_BIT(CHAR_HT);
SET_BIT(CHAR_SPACE);
/* For the 16-bit and 32-bit libraries (which can never be EBCDIC), set
the bits for 0xA0 and for code units >= 255, independently of UTF. */
#if PCRE2_CODE_UNIT_WIDTH != 8
SET_BIT(0xA0);
SET_BIT(0xFF);
#else
/* For the 8-bit library in UTF-8 mode, set the bits for the first code
units of horizontal space characters. */
#ifdef SUPPORT_UNICODE
if (utf)
{
SET_BIT(0xC2); /* For U+00A0 */
SET_BIT(0xE1); /* For U+1680, U+180E */
SET_BIT(0xE2); /* For U+2000 - U+200A, U+202F, U+205F */
SET_BIT(0xE3); /* For U+3000 */
}
else
#endif
/* For the 8-bit library not in UTF-8 mode, set the bit for 0xA0, unless
the code is EBCDIC. */
{
#ifndef EBCDIC
SET_BIT(0xA0);
#endif /* Not EBCDIC */
}
#endif /* 8-bit support */
break;
case OP_ANYNL:
case OP_VSPACE:
SET_BIT(CHAR_LF);
SET_BIT(CHAR_VT);
SET_BIT(CHAR_FF);
SET_BIT(CHAR_CR);
/* For the 16-bit and 32-bit libraries (which can never be EBCDIC), set
the bits for NEL and for code units >= 255, independently of UTF. */
#if PCRE2_CODE_UNIT_WIDTH != 8
SET_BIT(CHAR_NEL);
SET_BIT(0xFF);
#else
/* For the 8-bit library in UTF-8 mode, set the bits for the first code
units of vertical space characters. */
#ifdef SUPPORT_UNICODE
if (utf)
{
SET_BIT(0xC2); /* For U+0085 (NEL) */
SET_BIT(0xE2); /* For U+2028, U+2029 */
}
else
#endif
/* For the 8-bit library not in UTF-8 mode, set the bit for NEL. */
{
SET_BIT(CHAR_NEL);
}
#endif /* 8-bit support */
break;
case OP_NOT_DIGIT:
set_nottype_bits(re, cbit_digit, table_limit);
break;
case OP_DIGIT:
set_type_bits(re, cbit_digit, table_limit);
break;
case OP_NOT_WHITESPACE:
set_nottype_bits(re, cbit_space, table_limit);
break;
case OP_WHITESPACE:
set_type_bits(re, cbit_space, table_limit);
break;
case OP_NOT_WORDCHAR:
set_nottype_bits(re, cbit_word, table_limit);
break;
case OP_WORDCHAR:
set_type_bits(re, cbit_word, table_limit);
break;
}
tcode += 2;
break;
/* Set-based ECLASS: treat it the same as a "complex" XCLASS; give up. */
#ifdef SUPPORT_WIDE_CHARS
case OP_ECLASS:
return SSB_FAIL;
#endif
/* Extended class: if there are any property checks, or if this is a
negative XCLASS without a map, give up. If there are no property checks,
there must be wide characters on the XCLASS list, because otherwise an
XCLASS would not have been created. This means that code points >= 255
are potential starters. In the UTF-8 case we can scan them and set bits
for the relevant leading bytes. */
#ifdef SUPPORT_WIDE_CHARS
case OP_XCLASS:
xclassflags = tcode[1 + LINK_SIZE];
if ((xclassflags & XCL_HASPROP) != 0 ||
(xclassflags & (XCL_MAP|XCL_NOT)) == XCL_NOT)
return SSB_FAIL;
/* We have a positive XCLASS or a negative one without a map. Set up the
map pointer if there is one, and fall through. */
classmap = ((xclassflags & XCL_MAP) == 0)? NULL :
(const uint8_t *)(tcode + 1 + LINK_SIZE + 1);
/* In UTF-8 mode, scan the character list and set bits for leading bytes,
then jump to handle the map. */
#if PCRE2_CODE_UNIT_WIDTH == 8
if (utf && (xclassflags & XCL_NOT) == 0)
{
PCRE2_UCHAR b, e;
PCRE2_SPTR p = tcode + 1 + LINK_SIZE + 1 + ((classmap == NULL)? 0:32);
tcode += GET(tcode, 1);
if (*p >= XCL_LIST)
{
study_char_list(p, re->start_bitmap,
((const uint8_t *)re + re->code_start));
goto HANDLE_CLASSMAP;
}
for (;;) switch (*p++)
{
case XCL_SINGLE:
b = *p++;
while ((*p & 0xc0) == 0x80) p++;
re->start_bitmap[b/8] |= (1u << (b&7));
break;
case XCL_RANGE:
b = *p++;
while ((*p & 0xc0) == 0x80) p++;
e = *p++;
while ((*p & 0xc0) == 0x80) p++;
for (; b <= e; b++)
re->start_bitmap[b/8] |= (1u << (b&7));
break;
case XCL_END:
goto HANDLE_CLASSMAP;
default:
PCRE2_DEBUG_UNREACHABLE();
return SSB_UNKNOWN; /* Internal error, should not occur */
}
}
#endif /* SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8 */
#endif /* SUPPORT_WIDE_CHARS */
/* It seems that the fall through comment must be outside the #ifdef if
it is to avoid the gcc compiler warning. */
/* Fall through */
/* Enter here for a negative non-XCLASS. In the 8-bit library, if we are
in UTF mode, any byte with a value >= 0xc4 is a potentially valid starter
because it starts a character with a value > 255. In 8-bit non-UTF mode,
there is no difference between CLASS and NCLASS. In all other wide
character modes, set the 0xFF bit to indicate code units >= 255. */
case OP_NCLASS:
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (utf)
{
re->start_bitmap[24] |= 0xf0; /* Bits for 0xc4 - 0xc8 */
memset(re->start_bitmap+25, 0xff, 7); /* Bits for 0xc9 - 0xff */
}
#elif PCRE2_CODE_UNIT_WIDTH != 8
SET_BIT(0xFF); /* For characters >= 255 */
#endif
/* Fall through */
/* Enter here for a positive non-XCLASS. If we have fallen through from
an XCLASS, classmap will already be set; just advance the code pointer.
Otherwise, set up classmap for a a non-XCLASS and advance past it. */
case OP_CLASS:
if (*tcode == OP_XCLASS) tcode += GET(tcode, 1); else
{
classmap = (const uint8_t *)(++tcode);
tcode += 32 / sizeof(PCRE2_UCHAR);
}
/* When wide characters are supported, classmap may be NULL. In UTF-8
(sic) mode, the bits in a class bit map correspond to character values,
not to byte values. However, the bit map we are constructing is for byte
values. So we have to do a conversion for characters whose code point is
greater than 127. In fact, there are only two possible starting bytes for
characters in the range 128 - 255. */
#if defined SUPPORT_WIDE_CHARS && PCRE2_CODE_UNIT_WIDTH == 8
HANDLE_CLASSMAP:
#endif
if (classmap != NULL)
{
#if defined SUPPORT_UNICODE && PCRE2_CODE_UNIT_WIDTH == 8
if (utf)
{
for (c = 0; c < 16; c++) re->start_bitmap[c] |= classmap[c];
for (c = 128; c < 256; c++)
{
if ((classmap[c/8] & (1u << (c&7))) != 0)
{
int d = (c >> 6) | 0xc0; /* Set bit for this starter */
re->start_bitmap[d/8] |= (1u << (d&7)); /* and then skip on to the */
c = (c & 0xc0) + 0x40 - 1; /* next relevant character. */
}
}
}
else
#endif
/* In all modes except UTF-8, the two bit maps are compatible. */
{
for (c = 0; c < 32; c++) re->start_bitmap[c] |= classmap[c];
}
}
/* Act on what follows the class. For a zero minimum repeat, continue;
otherwise stop processing. */
switch (*tcode)
{
case OP_CRSTAR:
case OP_CRMINSTAR:
case OP_CRQUERY:
case OP_CRMINQUERY:
case OP_CRPOSSTAR:
case OP_CRPOSQUERY:
tcode++;
break;
case OP_CRRANGE:
case OP_CRMINRANGE:
case OP_CRPOSRANGE:
if (GET2(tcode, 1) == 0) tcode += 1 + 2 * IMM2_SIZE;
else try_next = FALSE;
break;
default:
try_next = FALSE;
break;
}
break; /* End of class handling case */
} /* End of switch for opcodes */
} /* End of try_next loop */
code += GET(code, 1); /* Advance to next branch */
}
while (*code == OP_ALT);
return yield;
}
/*************************************************
* Study a compiled expression *
*************************************************/
/* This function is handed a compiled expression that it must study to produce
information that will speed up the matching.
Argument:
re points to the compiled expression
Returns: 0 normally; non-zero should never normally occur
1 unknown opcode in set_start_bits
2 missing capturing bracket
3 unknown opcode in find_minlength
*/
int
PRIV(study)(pcre2_real_code *re)
{
int count = 0;
PCRE2_UCHAR *code;
BOOL utf = (re->overall_options & PCRE2_UTF) != 0;
BOOL ucp = (re->overall_options & PCRE2_UCP) != 0;
/* Find start of compiled code */
code = (PCRE2_UCHAR *)((uint8_t *)re + re->code_start);
/* For a pattern that has a first code unit, or a multiline pattern that
matches only at "line start", there is no point in seeking a list of starting
code units. */
if ((re->flags & (PCRE2_FIRSTSET|PCRE2_STARTLINE)) == 0)
{
int depth = 0;
int rc = set_start_bits(re, code, utf, ucp, &depth);
if (rc == SSB_UNKNOWN)
{
PCRE2_DEBUG_UNREACHABLE();
return 1;
}
/* If a list of starting code units was set up, scan the list to see if only
one or two were listed. Having only one listed is rare because usually a
single starting code unit will have been recognized and PCRE2_FIRSTSET set.
If two are listed, see if they are caseless versions of the same character;
if so we can replace the list with a caseless first code unit. This gives
better performance and is plausibly worth doing for patterns such as [Ww]ord
or (word|WORD). */
if (rc == SSB_DONE)
{
int i;
int a = -1;
int b = -1;
uint8_t *p = re->start_bitmap;
uint32_t flags = PCRE2_FIRSTMAPSET;
for (i = 0; i < 256; p++, i += 8)
{
uint8_t x = *p;
if (x != 0)
{
int c;
uint8_t y = x & (~x + 1); /* Least significant bit */
if (y != x) goto DONE; /* More than one bit set */
/* In the 16-bit and 32-bit libraries, the bit for 0xff means "0xff and
all wide characters", so we cannot use it here. */
#if PCRE2_CODE_UNIT_WIDTH != 8
if (i == 248 && x == 0x80) goto DONE;
#endif
/* Compute the character value */
c = i;
switch (x)
{
case 1: break;
case 2: c += 1; break; case 4: c += 2; break;
case 8: c += 3; break; case 16: c += 4; break;
case 32: c += 5; break; case 64: c += 6; break;
case 128: c += 7; break;
}
/* c contains the code unit value, in the range 0-255. In 8-bit UTF
mode, only values < 128 can be used. In all the other cases, c is a
character value. */
#if PCRE2_CODE_UNIT_WIDTH == 8
if (utf && c > 127) goto DONE;
#endif
if (a < 0) a = c; /* First one found, save in a */
else if (b < 0) /* Second one found */
{
int d = TABLE_GET((unsigned int)c, re->tables + fcc_offset, c);
#ifdef SUPPORT_UNICODE
if (utf || ucp)
{
if (UCD_CASESET(c) != 0) goto DONE; /* Multiple case set */
if (c > 127) d = UCD_OTHERCASE(c);
}
#endif /* SUPPORT_UNICODE */
if (d != a) goto DONE; /* Not the other case of a */
b = c; /* Save second in b */
}
else goto DONE; /* More than two characters found */
}
}
/* Replace the start code unit bits with a first code unit. If it is the
same as a required later code unit, then clear the required later code
unit. This is because a search for a required code unit starts after an
explicit first code unit, but at a code unit found from the bitmap.
Patterns such as /a*a/ don't work if both the start unit and required
unit are the same. */
if (a >= 0) {
if ((re->flags & PCRE2_LASTSET) && (re->last_codeunit == (uint32_t)a || (b >= 0 && re->last_codeunit == (uint32_t)b))) {
re->flags &= ~(PCRE2_LASTSET | PCRE2_LASTCASELESS);
re->last_codeunit = 0;
}
re->first_codeunit = a;
flags = PCRE2_FIRSTSET;
if (b >= 0) flags |= PCRE2_FIRSTCASELESS;
}
DONE:
re->flags |= flags;
}
}
/* Find the minimum length of subject string. If the pattern can match an empty
string, the minimum length is already known. If the pattern contains (*ACCEPT)
all bets are off, and we don't even try to find a minimum length. If there are
more back references than the size of the vector we are going to cache them in,
do nothing. A pattern that complicated will probably take a long time to
analyze and may in any case turn out to be too complicated. Note that back
reference minima are held as 16-bit numbers. */
if ((re->flags & (PCRE2_MATCH_EMPTY|PCRE2_HASACCEPT)) == 0 &&
re->top_backref <= MAX_CACHE_BACKREF)
{
int min;
int backref_cache[MAX_CACHE_BACKREF+1];
backref_cache[0] = 0; /* Highest one that is set */
min = find_minlength(re, code, code, utf, NULL, &count, backref_cache);
switch(min)
{
case -1: /* \C in UTF mode or over-complex regex */
break; /* Leave minlength unchanged (will be zero) */
case -2:
PCRE2_DEBUG_UNREACHABLE();
return 2; /* missing capturing bracket */
case -3:
PCRE2_DEBUG_UNREACHABLE();
return 3; /* unrecognized opcode */
default:
re->minlength = (min > UINT16_MAX)? UINT16_MAX : min;
break;
}
}
return 0;
}
/* End of pcre2_study.c */
pcre2-10.45/src/pcre2test.c 0000644 0001751 0000166 00001221446 14750415257 011112 /*************************************************
* PCRE2 testing program *
*************************************************/
/* PCRE2 is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language. In 2014
the API was completely revised and '2' was added to the name, because the old
API, which had lasted for 16 years, could not accommodate new requirements. At
the same time, this testing program was re-designed because its original
hacked-up (non-) design had also run out of steam.
Written by Philip Hazel
Original code Copyright (c) 1997-2012 University of Cambridge
Rewritten code Copyright (c) 2016-2024 University of Cambridge
-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the University of Cambridge nor the names of its
contributors may be used to endorse or promote products derived from
this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/
/* This program supports testing of the 8-bit, 16-bit, and 32-bit PCRE2
libraries in a single program, though its input and output are always 8-bit.
It is different from modules such as pcre2_compile.c in the library itself,
which are compiled separately for each code unit width. If two widths are
enabled, for example, pcre2_compile.c is compiled twice. In contrast,
pcre2test.c is compiled only once, and linked with all the enabled libraries.
Therefore, it must not make use of any of the macros from pcre2.h or
pcre2_internal.h that depend on PCRE2_CODE_UNIT_WIDTH. It does, however, make
use of SUPPORT_PCRE2_8, SUPPORT_PCRE2_16, and SUPPORT_PCRE2_32, to ensure that
it references only the enabled library functions. */
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include
#include
#include
#include
#include
#include
#include
#if defined NATIVE_ZOS
#include "pcrzoscs.h"
/* That header is not included in the main PCRE2 distribution because other
apparatus is needed to compile pcre2test for z/OS. The header can be found in
the special z/OS distribution, which is available from www.zaconsultants.net or
from www.cbttape.org. */
#endif
#ifdef HAVE_UNISTD_H
#include
#endif
/* Debugging code enabler */
/* #define DEBUG_SHOW_MALLOC_ADDRESSES */
/* Both libreadline and libedit are optionally supported */
#if defined(SUPPORT_LIBREADLINE) || defined(SUPPORT_LIBEDIT)
#if defined(SUPPORT_LIBREADLINE)
#include
#include
#else
#if defined(HAVE_EDITLINE_READLINE_H)
#include
#elif defined(HAVE_EDIT_READLINE_READLINE_H)
#include
#else
#include
/* GNU readline defines this macro but libedit doesn't, if that ever changes
this needs to be updated or the build could break */
#ifdef RL_VERSION_MAJOR
#include
#endif
#endif
#endif
#endif
/* Put the test for interactive input into a macro so that it can be changed if
required for different environments. */
#define INTERACTIVE(f) isatty(fileno(f))
/* ---------------------- System-specific definitions ---------------------- */
/* A number of things vary for Windows builds. Originally, pcretest opened its
input and output without "b"; then I was told that "b" was needed in some
environments, so it was added for release 5.0 to both the input and output. (It
makes no difference on Unix-like systems.) Later I was told that it is wrong
for the input on Windows. I've now abstracted the modes into macros that are
set here, to make it easier to fiddle with them, and removed "b" from the input
mode under Windows. The BINARY versions are used when saving/restoring compiled
patterns. */
#if defined(_WIN32) || defined(WIN32)
#include /* For _setmode() */
#include /* For _O_BINARY */
#define INPUT_MODE "r"
#define OUTPUT_MODE "wb"
#define BINARY_INPUT_MODE "rb"
#define BINARY_OUTPUT_MODE "wb"
#ifndef isatty
#define isatty _isatty /* This is what Windows calls them, I'm told, */
#endif /* though in some environments they seem to */
/* be already defined, hence the #ifndefs. */
#ifndef fileno
#define fileno _fileno
#endif
/* A user sent this fix for Borland Builder 5 under Windows. */
#ifdef __BORLANDC__
#define _setmode(handle, mode) setmode(handle, mode)
#endif
/* Not Windows */
#else
#include /* These two includes are needed */
#include /* for setrlimit(). */
#if defined NATIVE_ZOS /* z/OS uses non-binary I/O */
#define INPUT_MODE "r"
#define OUTPUT_MODE "w"
#define BINARY_INPUT_MODE "rb"
#define BINARY_OUTPUT_MODE "wb"
#else
#define INPUT_MODE "rb"
#define OUTPUT_MODE "wb"
#define BINARY_INPUT_MODE "rb"
#define BINARY_OUTPUT_MODE "wb"
#endif
#endif
/* VMS-specific code was included as suggested by a VMS user [1]. Another VMS
user [2] provided alternative code which worked better for him. I have
commented out the original, but kept it around just in case. */
#ifdef __VMS
#include
/* These two includes came from [2]. */
#include descrip
#include lib$routines
/* void vms_setsymbol( char *, char *, int ); Original code from [1]. */
#endif
/* old VC and older compilers don't support %td or %zu, and even some that
claim to be C99 don't support it (hence DISABLE_PERCENT_ZT). */
#if defined(DISABLE_PERCENT_ZT) || (defined(_MSC_VER) && (_MSC_VER < 1800)) || \
(!defined(_MSC_VER) && (!defined(__STDC_VERSION__) || (__STDC_VERSION__ < 199901L)))
#ifdef _WIN64
#define PTR_FORM "lld"
#define SIZ_FORM "llu"
#else
#define PTR_FORM "ld"
#define SIZ_FORM "lu"
#endif
#else
#define PTR_FORM "td"
#define SIZ_FORM "zu"
#endif
/* ------------------End of system-specific definitions -------------------- */
/* Glueing macros that are used in several places below. */
#define glue(a,b) a##b
#define G(a,b) glue(a,b)
/* Miscellaneous parameters and manifests */
#ifndef CLOCKS_PER_SEC
#ifdef CLK_TCK
#define CLOCKS_PER_SEC CLK_TCK
#else
#define CLOCKS_PER_SEC 100
#endif
#endif
#define CFORE_UNSET UINT32_MAX /* Unset value for startend/cfail/cerror fields */
#define CONVERT_UNSET UINT32_MAX /* Unset value for convert_type field */
#define DFA_WS_DIMENSION 1000 /* Size of DFA workspace */
#define DEFAULT_OVECCOUNT 15 /* Default ovector count */
#define JUNK_OFFSET 0xdeadbeef /* For initializing ovector */
#define LOCALESIZE 32 /* Size of locale name */
#define LOOPREPEAT 500000 /* Default loop count for timing */
#define MALLOCLISTSIZE 20 /* For remembering mallocs */
#define PARENS_NEST_DEFAULT 220 /* Default parentheses nest limit */
#define PATSTACKSIZE 20 /* Pattern stack for save/restore testing */
#define REPLACE_MODSIZE 100 /* Field for reading 8-bit replacement */
#define VERSION_SIZE 64 /* Size of buffer for the version strings */
/* Default JIT compile options */
#define JIT_DEFAULT (PCRE2_JIT_COMPLETE|\
PCRE2_JIT_PARTIAL_SOFT|\
PCRE2_JIT_PARTIAL_HARD)
/* Make sure the buffer into which replacement strings are copied is big enough
to hold them as 32-bit code units. */
#define REPLACE_BUFFSIZE 1024 /* This is a byte value */
/* Execution modes */
#define PCRE8_MODE 8
#define PCRE16_MODE 16
#define PCRE32_MODE 32
/* Processing returns */
enum { PR_OK, PR_SKIP, PR_ABEND };
/* The macro PRINTABLE determines whether to print an output character as-is or
as a hex value when showing compiled patterns. We use it in cases when the
locale has not been explicitly changed, so as to get consistent output from
systems that differ in their output from isprint() even in the "C" locale. */
#ifdef EBCDIC
#define PRINTABLE(c) ((c) >= 64 && (c) < 255)
#else
#define PRINTABLE(c) ((c) >= 32 && (c) < 127)
#endif
#define PRINTOK(c) ((use_tables != NULL && c < 256)? isprint(c) : PRINTABLE(c))
/* We have to include some of the library source files because we need
to use some of the macros, internal structure definitions, and other internal
values - pcre2test has "inside information" compared to an application program
that strictly follows the PCRE2 API.
Before including pcre2_internal.h we define PRIV so that it does not get
defined therein. This ensures that PRIV names in the included files do not
clash with those in the libraries. Also, although pcre2_internal.h does itself
include pcre2.h, we explicitly include it beforehand, along with pcre2posix.h,
so that the PCRE2_EXP_xxx macros get set appropriately for an application, not
for building the library.
Setting PCRE2_CODE_UNIT_WIDTH to zero cuts out all the width-specific settings
in pcre2.h and pcre2_internal.h. Defining PCRE2_BUILDING_PCRE2TEST cuts out the
check in pcre2_internal.h that ensures PCRE2_CODE_UNIT_WIDTH is 8, 16, or 32
(which it needs to be when compiling one of the libraries). */
#define PRIV(name) name
#define PCRE2_CODE_UNIT_WIDTH 0
#define PCRE2_BUILDING_PCRE2TEST
#include "pcre2.h"
#include "pcre2posix.h"
#include "pcre2_internal.h"
/* We need access to some of the data tables that PCRE2 uses. Defining
PCRE2_PCRE2TEST makes some minor changes in the files. The previous definition
of PRIV avoids name clashes. */
#define PCRE2_PCRE2TEST
#include "pcre2_tables.c"
#include "pcre2_ucd.c"
/* 32-bit integer values in the input are read by strtoul() or strtol(). The
check needed for overflow depends on whether long ints are in fact longer than
ints. They are defined not to be shorter. */
#if ULONG_MAX > UINT32_MAX
#define U32OVERFLOW(x) (x > UINT32_MAX)
#else
#define U32OVERFLOW(x) (x == UINT32_MAX)
#endif
#if LONG_MAX > INT32_MAX
#define S32OVERFLOW(x) (x > INT32_MAX || x < INT32_MIN)
#else
#define S32OVERFLOW(x) (x == INT32_MAX || x == INT32_MIN)
#endif
/* When PCRE2_CODE_UNIT_WIDTH is zero, pcre2_internal.h does not include
pcre2_intmodedep.h, which is where mode-dependent macros and structures are
defined. We can now include it for each supported code unit width. Because
PCRE2_CODE_UNIT_WIDTH was defined as zero before including pcre2.h, it will
have left PCRE2_SUFFIX defined as a no-op. We must re-define it appropriately
while including these files, and then restore it to a no-op. Because LINK_SIZE
may be changed in 16-bit mode and forced to 1 in 32-bit mode, the order of
these inclusions should not be changed. */
#undef PCRE2_SUFFIX
#undef PCRE2_CODE_UNIT_WIDTH
#ifdef SUPPORT_PCRE2_8
#define PCRE2_CODE_UNIT_WIDTH 8
#define PCRE2_SUFFIX(a) G(a,8)
#include "pcre2_intmodedep.h"
#include "pcre2_printint.c"
#undef PCRE2_CODE_UNIT_WIDTH
#undef PCRE2_SUFFIX
#endif /* SUPPORT_PCRE2_8 */
#ifdef SUPPORT_PCRE2_16
#define PCRE2_CODE_UNIT_WIDTH 16
#define PCRE2_SUFFIX(a) G(a,16)
#include "pcre2_intmodedep.h"
#include "pcre2_printint.c"
#undef PCRE2_CODE_UNIT_WIDTH
#undef PCRE2_SUFFIX
#endif /* SUPPORT_PCRE2_16 */
#ifdef SUPPORT_PCRE2_32
#define PCRE2_CODE_UNIT_WIDTH 32
#define PCRE2_SUFFIX(a) G(a,32)
#include "pcre2_intmodedep.h"
#include "pcre2_printint.c"
#undef PCRE2_CODE_UNIT_WIDTH
#undef PCRE2_SUFFIX
#endif /* SUPPORT_PCRE2_32 */
#define PCRE2_SUFFIX(a) a
#include "pcre2_chkdint.c"
/* We need to be able to check input text for UTF-8 validity, whatever code
widths are actually available, because the input to pcre2test is always in
8-bit code units. So we include the UTF validity checking function for 8-bit
code units. */
extern int valid_utf(PCRE2_SPTR8, PCRE2_SIZE, PCRE2_SIZE *);
#define PCRE2_CODE_UNIT_WIDTH 8
#undef PCRE2_SPTR
#define PCRE2_SPTR PCRE2_SPTR8
#include "pcre2_valid_utf.c"
#undef PCRE2_CODE_UNIT_WIDTH
#undef PCRE2_SPTR
/* If we have 8-bit support, default to it; if there is also 16-or 32-bit
support, it can be selected by a command-line option. If there is no 8-bit
support, there must be 16-bit or 32-bit support, so default to one of them. The
config function, JIT stack, contexts, and version string are the same in all
modes, so use the form of the first that is available. */
#if defined SUPPORT_PCRE2_8
#define DEFAULT_TEST_MODE PCRE8_MODE
#define VERSION_TYPE PCRE2_UCHAR8
#define PCRE2_CONFIG pcre2_config_8
#define PCRE2_JIT_STACK pcre2_jit_stack_8
#define PCRE2_REAL_GENERAL_CONTEXT pcre2_real_general_context_8
#define PCRE2_REAL_COMPILE_CONTEXT pcre2_real_compile_context_8
#define PCRE2_REAL_CONVERT_CONTEXT pcre2_real_convert_context_8
#define PCRE2_REAL_MATCH_CONTEXT pcre2_real_match_context_8
#elif defined SUPPORT_PCRE2_16
#define DEFAULT_TEST_MODE PCRE16_MODE
#define VERSION_TYPE PCRE2_UCHAR16
#define PCRE2_CONFIG pcre2_config_16
#define PCRE2_JIT_STACK pcre2_jit_stack_16
#define PCRE2_REAL_GENERAL_CONTEXT pcre2_real_general_context_16
#define PCRE2_REAL_COMPILE_CONTEXT pcre2_real_compile_context_16
#define PCRE2_REAL_CONVERT_CONTEXT pcre2_real_convert_context_16
#define PCRE2_REAL_MATCH_CONTEXT pcre2_real_match_context_16
#elif defined SUPPORT_PCRE2_32
#define DEFAULT_TEST_MODE PCRE32_MODE
#define VERSION_TYPE PCRE2_UCHAR32
#define PCRE2_CONFIG pcre2_config_32
#define PCRE2_JIT_STACK pcre2_jit_stack_32
#define PCRE2_REAL_GENERAL_CONTEXT pcre2_real_general_context_32
#define PCRE2_REAL_COMPILE_CONTEXT pcre2_real_compile_context_32
#define PCRE2_REAL_CONVERT_CONTEXT pcre2_real_convert_context_32
#define PCRE2_REAL_MATCH_CONTEXT pcre2_real_match_context_32
#endif
/* ------------- Structure and table for handling #-commands ------------- */
typedef struct cmdstruct {
const char *name;
int value;
} cmdstruct;
enum { CMD_FORBID_UTF, CMD_LOAD, CMD_LOADTABLES, CMD_NEWLINE_DEFAULT,
CMD_PATTERN, CMD_PERLTEST, CMD_POP, CMD_POPCOPY, CMD_SAVE, CMD_SUBJECT,
CMD_UNKNOWN };
static cmdstruct cmdlist[] = {
{ "forbid_utf", CMD_FORBID_UTF },
{ "load", CMD_LOAD },
{ "loadtables", CMD_LOADTABLES },
{ "newline_default", CMD_NEWLINE_DEFAULT },
{ "pattern", CMD_PATTERN },
{ "perltest", CMD_PERLTEST },
{ "pop", CMD_POP },
{ "popcopy", CMD_POPCOPY },
{ "save", CMD_SAVE },
{ "subject", CMD_SUBJECT }};
#define cmdlistcount (sizeof(cmdlist)/sizeof(cmdstruct))
/* ------------- Structures and tables for handling modifiers -------------- */
/* Table of names for newline types. Must be kept in step with the definitions
of PCRE2_NEWLINE_xx in pcre2.h. */
static const char *newlines[] = {
"DEFAULT", "CR", "LF", "CRLF", "ANY", "ANYCRLF", "NUL" };
/* Structure and table for handling pattern conversion types. */
typedef struct convertstruct {
const char *name;
uint32_t option;
} convertstruct;
static convertstruct convertlist[] = {
{ "glob", PCRE2_CONVERT_GLOB },
{ "glob_no_starstar", PCRE2_CONVERT_GLOB_NO_STARSTAR },
{ "glob_no_wild_separator", PCRE2_CONVERT_GLOB_NO_WILD_SEPARATOR },
{ "posix_basic", PCRE2_CONVERT_POSIX_BASIC },
{ "posix_extended", PCRE2_CONVERT_POSIX_EXTENDED },
{ "unset", CONVERT_UNSET }};
#define convertlistcount (sizeof(convertlist)/sizeof(convertstruct))
/* Modifier types and applicability */
enum { MOD_CTC, /* Applies to a compile context */
MOD_CTM, /* Applies to a match context */
MOD_PAT, /* Applies to a pattern */
MOD_PATP, /* Ditto, OK for Perl test */
MOD_DAT, /* Applies to a data line */
MOD_DATP, /* Ditto, OK for Perl test */
MOD_PD, /* Applies to a pattern or a data line */
MOD_PDP, /* As MOD_PD, OK for Perl test */
MOD_PND, /* As MOD_PD, but not for a default pattern */
MOD_PNDP, /* As MOD_PND, OK for Perl test */
MOD_CHR, /* Is a single character */
MOD_CON, /* Is a "convert" type/options list */
MOD_CTL, /* Is a control bit */
MOD_BSR, /* Is a BSR value */
MOD_IN2, /* Is one or two unsigned integers */
MOD_INS, /* Is a signed integer */
MOD_INT, /* Is an unsigned integer */
MOD_IND, /* Is an unsigned integer, but no value => default */
MOD_NL, /* Is a newline value */
MOD_NN, /* Is a number or a name; more than one may occur */
MOD_OPT, /* Is an option bit */
MOD_OPTMZ, /* Is an optimization directive */
MOD_SIZ, /* Is a PCRE2_SIZE value */
MOD_STR }; /* Is a string */
/* Control bits. Some apply to compiling, some to matching, but some can be set
either on a pattern or a data line, so they must all be distinct. There are now
so many of them that they are split into two fields. */
#define CTL_AFTERTEXT 0x00000001u
#define CTL_ALLAFTERTEXT 0x00000002u
#define CTL_ALLCAPTURES 0x00000004u
#define CTL_ALLUSEDTEXT 0x00000008u
#define CTL_ALTGLOBAL 0x00000010u
#define CTL_BINCODE 0x00000020u
#define CTL_CALLOUT_CAPTURE 0x00000040u
#define CTL_CALLOUT_INFO 0x00000080u
#define CTL_CALLOUT_NONE 0x00000100u
#define CTL_DFA 0x00000200u
#define CTL_EXPAND 0x00000400u
#define CTL_FINDLIMITS 0x00000800u
#define CTL_FINDLIMITS_NOHEAP 0x00001000u
#define CTL_FULLBINCODE 0x00002000u
#define CTL_GETALL 0x00004000u
#define CTL_GLOBAL 0x00008000u
#define CTL_HEXPAT 0x00010000u /* Same word as USE_LENGTH */
#define CTL_INFO 0x00020000u
#define CTL_JITFAST 0x00040000u
#define CTL_JITVERIFY 0x00080000u
#define CTL_MARK 0x00100000u
#define CTL_MEMORY 0x00200000u
#define CTL_NULLCONTEXT 0x00400000u
#define CTL_POSIX 0x00800000u
#define CTL_POSIX_NOSUB 0x01000000u
#define CTL_PUSH 0x02000000u /* These three must be */
#define CTL_PUSHCOPY 0x04000000u /* all in the same */
#define CTL_PUSHTABLESCOPY 0x08000000u /* word. */
#define CTL_STARTCHAR 0x10000000u
#define CTL_USE_LENGTH 0x20000000u /* Same word as HEXPAT */
#define CTL_UTF8_INPUT 0x40000000u
#define CTL_ZERO_TERMINATE 0x80000000u
/* Combinations */
#define CTL_DEBUG (CTL_FULLBINCODE|CTL_INFO) /* For setting */
#define CTL_ANYINFO (CTL_DEBUG|CTL_BINCODE|CTL_CALLOUT_INFO)
#define CTL_ANYGLOB (CTL_ALTGLOBAL|CTL_GLOBAL)
/* Second control word */
#define CTL2_SUBSTITUTE_CALLOUT 0x00000001u
#define CTL2_SUBSTITUTE_EXTENDED 0x00000002u
#define CTL2_SUBSTITUTE_LITERAL 0x00000004u
#define CTL2_SUBSTITUTE_MATCHED 0x00000008u
#define CTL2_SUBSTITUTE_OVERFLOW_LENGTH 0x00000010u
#define CTL2_SUBSTITUTE_REPLACEMENT_ONLY 0x00000020u
#define CTL2_SUBSTITUTE_UNKNOWN_UNSET 0x00000040u
#define CTL2_SUBSTITUTE_UNSET_EMPTY 0x00000080u
#define CTL2_SUBJECT_LITERAL 0x00000100u
#define CTL2_CALLOUT_NO_WHERE 0x00000200u
#define CTL2_CALLOUT_EXTRA 0x00000400u
#define CTL2_ALLVECTOR 0x00000800u
#define CTL2_NULL_PATTERN 0x00001000u
#define CTL2_NULL_SUBJECT 0x00002000u
#define CTL2_NULL_REPLACEMENT 0x00004000u
#define CTL2_FRAMESIZE 0x00008000u
#define CTL2_SUBSTITUTE_CASE_CALLOUT 0x00010000u
#define CTL2_HEAPFRAMES_SIZE 0x20000000u /* Informational */
#define CTL2_NL_SET 0x40000000u /* Informational */
#define CTL2_BSR_SET 0x80000000u /* Informational */
/* These are the matching controls that may be set either on a pattern or on a
data line. They are copied from the pattern controls as initial settings for
data line controls. Note that CTL_MEMORY is not included here, because it does
different things in the two cases. */
#define CTL_ALLPD (CTL_AFTERTEXT|\
CTL_ALLAFTERTEXT|\
CTL_ALLCAPTURES|\
CTL_ALLUSEDTEXT|\
CTL_ALTGLOBAL|\
CTL_GLOBAL|\
CTL_MARK|\
CTL_STARTCHAR|\
CTL_UTF8_INPUT)
#define CTL2_ALLPD (CTL2_SUBSTITUTE_CALLOUT|\
CTL2_SUBSTITUTE_EXTENDED|\
CTL2_SUBSTITUTE_LITERAL|\
CTL2_SUBSTITUTE_MATCHED|\
CTL2_SUBSTITUTE_OVERFLOW_LENGTH|\
CTL2_SUBSTITUTE_REPLACEMENT_ONLY|\
CTL2_SUBSTITUTE_UNKNOWN_UNSET|\
CTL2_SUBSTITUTE_UNSET_EMPTY|\
CTL2_ALLVECTOR|\
CTL2_SUBSTITUTE_CASE_CALLOUT|\
CTL2_HEAPFRAMES_SIZE)
/* Structures for holding modifier information for patterns and subject strings
(data). Fields containing modifiers that can be set either for a pattern or a
subject must be at the start and in the same order in both cases so that the
same offset in the big table below works for both. */
typedef struct patctl { /* Structure for pattern modifiers. */
uint32_t options; /* Must be in same position as datctl */
uint32_t control; /* Must be in same position as datctl */
uint32_t control2; /* Must be in same position as datctl */
uint32_t jitstack; /* Must be in same position as datctl */
uint8_t replacement[REPLACE_MODSIZE]; /* So must this */
uint32_t substitute_skip; /* Must be in same position as datctl */
uint32_t substitute_stop; /* Must be in same position as datctl */
uint32_t jit;
uint32_t stackguard_test;
uint32_t tables_id;
uint32_t convert_type;
uint32_t convert_length;
uint32_t convert_glob_escape;
uint32_t convert_glob_separator;
uint32_t regerror_buffsize;
uint8_t locale[LOCALESIZE];
} patctl;
#define MAXCPYGET 10
#define LENCPYGET 64
typedef struct datctl { /* Structure for data line modifiers. */
uint32_t options; /* Must be in same position as patctl */
uint32_t control; /* Must be in same position as patctl */
uint32_t control2; /* Must be in same position as patctl */
uint32_t jitstack; /* Must be in same position as patctl */
uint8_t replacement[REPLACE_MODSIZE]; /* So must this */
uint32_t substitute_skip; /* Must be in same position as patctl */
uint32_t substitute_stop; /* Must be in same position as patctl */
uint32_t startend[2];
uint32_t cerror[2];
uint32_t cfail[2];
int32_t callout_data;
int32_t copy_numbers[MAXCPYGET];
int32_t get_numbers[MAXCPYGET];
uint32_t oveccount;
uint32_t offset;
uint8_t copy_names[LENCPYGET];
uint8_t get_names[LENCPYGET];
} datctl;
/* Ids for which context to modify. */
enum { CTX_PAT, /* Active pattern context */
CTX_POPPAT, /* Ditto, for a popped pattern */
CTX_DEFPAT, /* Default pattern context */
CTX_DAT, /* Active data (match) context */
CTX_DEFDAT }; /* Default data (match) context */
/* Macros to simplify the big table below. */
#define CO(name) offsetof(PCRE2_REAL_COMPILE_CONTEXT, name)
#define MO(name) offsetof(PCRE2_REAL_MATCH_CONTEXT, name)
#define PO(name) offsetof(patctl, name)
#define PD(name) PO(name)
#define DO(name) offsetof(datctl, name)
/* Table of all long-form modifiers. Must be in collating sequence of modifier
name because it is searched by binary chop. */
typedef struct modstruct {
const char *name;
uint16_t which;
uint16_t type;
uint32_t value;
PCRE2_SIZE offset;
} modstruct;
#define PCRE2_EXTRA_ASCII_ALL (PCRE2_EXTRA_ASCII_BSD|PCRE2_EXTRA_ASCII_BSS| \
PCRE2_EXTRA_ASCII_BSW|PCRE2_EXTRA_ASCII_POSIX)
static modstruct modlist[] = {
{ "aftertext", MOD_PNDP, MOD_CTL, CTL_AFTERTEXT, PO(control) },
{ "allaftertext", MOD_PNDP, MOD_CTL, CTL_ALLAFTERTEXT, PO(control) },
{ "allcaptures", MOD_PND, MOD_CTL, CTL_ALLCAPTURES, PO(control) },
{ "allow_empty_class", MOD_PAT, MOD_OPT, PCRE2_ALLOW_EMPTY_CLASS, PO(options) },
{ "allow_lookaround_bsk", MOD_CTC, MOD_OPT, PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK, CO(extra_options) },
{ "allow_surrogate_escapes", MOD_CTC, MOD_OPT, PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES, CO(extra_options) },
{ "allusedtext", MOD_PNDP, MOD_CTL, CTL_ALLUSEDTEXT, PO(control) },
{ "allvector", MOD_PND, MOD_CTL, CTL2_ALLVECTOR, PO(control2) },
{ "alt_bsux", MOD_PAT, MOD_OPT, PCRE2_ALT_BSUX, PO(options) },
{ "alt_circumflex", MOD_PAT, MOD_OPT, PCRE2_ALT_CIRCUMFLEX, PO(options) },
{ "alt_extended_class", MOD_PAT, MOD_OPT, PCRE2_ALT_EXTENDED_CLASS, PO(options) },
{ "alt_verbnames", MOD_PAT, MOD_OPT, PCRE2_ALT_VERBNAMES, PO(options) },
{ "altglobal", MOD_PND, MOD_CTL, CTL_ALTGLOBAL, PO(control) },
{ "anchored", MOD_PD, MOD_OPT, PCRE2_ANCHORED, PD(options) },
{ "ascii_all", MOD_CTC, MOD_OPT, PCRE2_EXTRA_ASCII_ALL, CO(extra_options) },
{ "ascii_bsd", MOD_CTC, MOD_OPT, PCRE2_EXTRA_ASCII_BSD, CO(extra_options) },
{ "ascii_bss", MOD_CTC, MOD_OPT, PCRE2_EXTRA_ASCII_BSS, CO(extra_options) },
{ "ascii_bsw", MOD_CTC, MOD_OPT, PCRE2_EXTRA_ASCII_BSW, CO(extra_options) },
{ "ascii_digit", MOD_CTC, MOD_OPT, PCRE2_EXTRA_ASCII_DIGIT, CO(extra_options) },
{ "ascii_posix", MOD_CTC, MOD_OPT, PCRE2_EXTRA_ASCII_POSIX, CO(extra_options) },
{ "auto_callout", MOD_PAT, MOD_OPT, PCRE2_AUTO_CALLOUT, PO(options) },
{ "auto_possess", MOD_CTC, MOD_OPTMZ, PCRE2_AUTO_POSSESS, 0 },
{ "auto_possess_off", MOD_CTC, MOD_OPTMZ, PCRE2_AUTO_POSSESS_OFF, 0 },
{ "bad_escape_is_literal", MOD_CTC, MOD_OPT, PCRE2_EXTRA_BAD_ESCAPE_IS_LITERAL, CO(extra_options) },
{ "bincode", MOD_PAT, MOD_CTL, CTL_BINCODE, PO(control) },
{ "bsr", MOD_CTC, MOD_BSR, 0, CO(bsr_convention) },
{ "callout_capture", MOD_DAT, MOD_CTL, CTL_CALLOUT_CAPTURE, DO(control) },
{ "callout_data", MOD_DAT, MOD_INS, 0, DO(callout_data) },
{ "callout_error", MOD_DAT, MOD_IN2, 0, DO(cerror) },
{ "callout_extra", MOD_DAT, MOD_CTL, CTL2_CALLOUT_EXTRA, DO(control2) },
{ "callout_fail", MOD_DAT, MOD_IN2, 0, DO(cfail) },
{ "callout_info", MOD_PAT, MOD_CTL, CTL_CALLOUT_INFO, PO(control) },
{ "callout_no_where", MOD_DAT, MOD_CTL, CTL2_CALLOUT_NO_WHERE, DO(control2) },
{ "callout_none", MOD_DAT, MOD_CTL, CTL_CALLOUT_NONE, DO(control) },
{ "caseless", MOD_PATP, MOD_OPT, PCRE2_CASELESS, PO(options) },
{ "caseless_restrict", MOD_CTC, MOD_OPT, PCRE2_EXTRA_CASELESS_RESTRICT, CO(extra_options) },
{ "convert", MOD_PAT, MOD_CON, 0, PO(convert_type) },
{ "convert_glob_escape", MOD_PAT, MOD_CHR, 0, PO(convert_glob_escape) },
{ "convert_glob_separator", MOD_PAT, MOD_CHR, 0, PO(convert_glob_separator) },
{ "convert_length", MOD_PAT, MOD_INT, 0, PO(convert_length) },
{ "copy", MOD_DAT, MOD_NN, DO(copy_numbers), DO(copy_names) },
{ "copy_matched_subject", MOD_DAT, MOD_OPT, PCRE2_COPY_MATCHED_SUBJECT, DO(options) },
{ "debug", MOD_PAT, MOD_CTL, CTL_DEBUG, PO(control) },
{ "depth_limit", MOD_CTM, MOD_INT, 0, MO(depth_limit) },
{ "dfa", MOD_DAT, MOD_CTL, CTL_DFA, DO(control) },
{ "dfa_restart", MOD_DAT, MOD_OPT, PCRE2_DFA_RESTART, DO(options) },
{ "dfa_shortest", MOD_DAT, MOD_OPT, PCRE2_DFA_SHORTEST, DO(options) },
{ "disable_recurseloop_check", MOD_DAT, MOD_OPT, PCRE2_DISABLE_RECURSELOOP_CHECK, DO(options) },
{ "dollar_endonly", MOD_PAT, MOD_OPT, PCRE2_DOLLAR_ENDONLY, PO(options) },
{ "dotall", MOD_PATP, MOD_OPT, PCRE2_DOTALL, PO(options) },
{ "dotstar_anchor", MOD_CTC, MOD_OPTMZ, PCRE2_DOTSTAR_ANCHOR, 0 },
{ "dotstar_anchor_off", MOD_CTC, MOD_OPTMZ, PCRE2_DOTSTAR_ANCHOR_OFF, 0 },
{ "dupnames", MOD_PATP, MOD_OPT, PCRE2_DUPNAMES, PO(options) },
{ "endanchored", MOD_PD, MOD_OPT, PCRE2_ENDANCHORED, PD(options) },
{ "escaped_cr_is_lf", MOD_CTC, MOD_OPT, PCRE2_EXTRA_ESCAPED_CR_IS_LF, CO(extra_options) },
{ "expand", MOD_PAT, MOD_CTL, CTL_EXPAND, PO(control) },
{ "extended", MOD_PATP, MOD_OPT, PCRE2_EXTENDED, PO(options) },
{ "extended_more", MOD_PATP, MOD_OPT, PCRE2_EXTENDED_MORE, PO(options) },
{ "extra_alt_bsux", MOD_CTC, MOD_OPT, PCRE2_EXTRA_ALT_BSUX, CO(extra_options) },
{ "find_limits", MOD_DAT, MOD_CTL, CTL_FINDLIMITS, DO(control) },
{ "find_limits_noheap", MOD_DAT, MOD_CTL, CTL_FINDLIMITS_NOHEAP, DO(control) },
{ "firstline", MOD_PAT, MOD_OPT, PCRE2_FIRSTLINE, PO(options) },
{ "framesize", MOD_PAT, MOD_CTL, CTL2_FRAMESIZE, PO(control2) },
{ "fullbincode", MOD_PAT, MOD_CTL, CTL_FULLBINCODE, PO(control) },
{ "get", MOD_DAT, MOD_NN, DO(get_numbers), DO(get_names) },
{ "getall", MOD_DAT, MOD_CTL, CTL_GETALL, DO(control) },
{ "global", MOD_PNDP, MOD_CTL, CTL_GLOBAL, PO(control) },
{ "heap_limit", MOD_CTM, MOD_INT, 0, MO(heap_limit) },
{ "heapframes_size", MOD_PND, MOD_CTL, CTL2_HEAPFRAMES_SIZE, PO(control2) },
{ "hex", MOD_PATP, MOD_CTL, CTL_HEXPAT, PO(control) },
{ "info", MOD_PAT, MOD_CTL, CTL_INFO, PO(control) },
{ "jit", MOD_PAT, MOD_IND, 7, PO(jit) },
{ "jitfast", MOD_PAT, MOD_CTL, CTL_JITFAST, PO(control) },
{ "jitstack", MOD_PNDP, MOD_INT, 0, PO(jitstack) },
{ "jitverify", MOD_PAT, MOD_CTL, CTL_JITVERIFY, PO(control) },
{ "literal", MOD_PAT, MOD_OPT, PCRE2_LITERAL, PO(options) },
{ "locale", MOD_PATP, MOD_STR, LOCALESIZE, PO(locale) },
{ "mark", MOD_PNDP, MOD_CTL, CTL_MARK, PO(control) },
{ "match_invalid_utf", MOD_PAT, MOD_OPT, PCRE2_MATCH_INVALID_UTF, PO(options) },
{ "match_limit", MOD_CTM, MOD_INT, 0, MO(match_limit) },
{ "match_line", MOD_CTC, MOD_OPT, PCRE2_EXTRA_MATCH_LINE, CO(extra_options) },
{ "match_unset_backref", MOD_PAT, MOD_OPT, PCRE2_MATCH_UNSET_BACKREF, PO(options) },
{ "match_word", MOD_CTC, MOD_OPT, PCRE2_EXTRA_MATCH_WORD, CO(extra_options) },
{ "max_pattern_compiled_length", MOD_CTC, MOD_SIZ, 0, CO(max_pattern_compiled_length) },
{ "max_pattern_length", MOD_CTC, MOD_SIZ, 0, CO(max_pattern_length) },
{ "max_varlookbehind", MOD_CTC, MOD_INT, 0, CO(max_varlookbehind) },
{ "memory", MOD_PD, MOD_CTL, CTL_MEMORY, PD(control) },
{ "multiline", MOD_PATP, MOD_OPT, PCRE2_MULTILINE, PO(options) },
{ "never_backslash_c", MOD_PAT, MOD_OPT, PCRE2_NEVER_BACKSLASH_C, PO(options) },
{ "never_callout", MOD_CTC, MOD_OPT, PCRE2_EXTRA_NEVER_CALLOUT, CO(extra_options) },
{ "never_ucp", MOD_PAT, MOD_OPT, PCRE2_NEVER_UCP, PO(options) },
{ "never_utf", MOD_PAT, MOD_OPT, PCRE2_NEVER_UTF, PO(options) },
{ "newline", MOD_CTC, MOD_NL, 0, CO(newline_convention) },
{ "no_auto_capture", MOD_PAT, MOD_OPT, PCRE2_NO_AUTO_CAPTURE, PO(options) },
{ "no_auto_possess", MOD_PATP, MOD_OPT, PCRE2_NO_AUTO_POSSESS, PO(options) },
{ "no_bs0", MOD_CTC, MOD_OPT, PCRE2_EXTRA_NO_BS0, CO(extra_options) },
{ "no_dotstar_anchor", MOD_PAT, MOD_OPT, PCRE2_NO_DOTSTAR_ANCHOR, PO(options) },
{ "no_jit", MOD_DATP, MOD_OPT, PCRE2_NO_JIT, DO(options) },
{ "no_start_optimize", MOD_PATP, MOD_OPT, PCRE2_NO_START_OPTIMIZE, PO(options) },
{ "no_utf_check", MOD_PD, MOD_OPT, PCRE2_NO_UTF_CHECK, PD(options) },
{ "notbol", MOD_DAT, MOD_OPT, PCRE2_NOTBOL, DO(options) },
{ "notempty", MOD_DAT, MOD_OPT, PCRE2_NOTEMPTY, DO(options) },
{ "notempty_atstart", MOD_DAT, MOD_OPT, PCRE2_NOTEMPTY_ATSTART, DO(options) },
{ "noteol", MOD_DAT, MOD_OPT, PCRE2_NOTEOL, DO(options) },
{ "null_context", MOD_PD, MOD_CTL, CTL_NULLCONTEXT, PO(control) },
{ "null_pattern", MOD_PAT, MOD_CTL, CTL2_NULL_PATTERN, PO(control2) },
{ "null_replacement", MOD_DAT, MOD_CTL, CTL2_NULL_REPLACEMENT, DO(control2) },
{ "null_subject", MOD_DAT, MOD_CTL, CTL2_NULL_SUBJECT, DO(control2) },
{ "offset", MOD_DAT, MOD_INT, 0, DO(offset) },
{ "offset_limit", MOD_CTM, MOD_SIZ, 0, MO(offset_limit)},
{ "optimization_full", MOD_CTC, MOD_OPTMZ, PCRE2_OPTIMIZATION_FULL, 0 },
{ "optimization_none", MOD_CTC, MOD_OPTMZ, PCRE2_OPTIMIZATION_NONE, 0 },
{ "ovector", MOD_DAT, MOD_INT, 0, DO(oveccount) },
{ "parens_nest_limit", MOD_CTC, MOD_INT, 0, CO(parens_nest_limit) },
{ "partial_hard", MOD_DAT, MOD_OPT, PCRE2_PARTIAL_HARD, DO(options) },
{ "partial_soft", MOD_DAT, MOD_OPT, PCRE2_PARTIAL_SOFT, DO(options) },
{ "ph", MOD_DAT, MOD_OPT, PCRE2_PARTIAL_HARD, DO(options) },
{ "posix", MOD_PAT, MOD_CTL, CTL_POSIX, PO(control) },
{ "posix_nosub", MOD_PAT, MOD_CTL, CTL_POSIX|CTL_POSIX_NOSUB, PO(control) },
{ "posix_startend", MOD_DAT, MOD_IN2, 0, DO(startend) },
{ "ps", MOD_DAT, MOD_OPT, PCRE2_PARTIAL_SOFT, DO(options) },
{ "push", MOD_PAT, MOD_CTL, CTL_PUSH, PO(control) },
{ "pushcopy", MOD_PAT, MOD_CTL, CTL_PUSHCOPY, PO(control) },
{ "pushtablescopy", MOD_PAT, MOD_CTL, CTL_PUSHTABLESCOPY, PO(control) },
{ "python_octal", MOD_CTC, MOD_OPT, PCRE2_EXTRA_PYTHON_OCTAL, CO(extra_options) },
{ "recursion_limit", MOD_CTM, MOD_INT, 0, MO(depth_limit) }, /* Obsolete synonym */
{ "regerror_buffsize", MOD_PAT, MOD_INT, 0, PO(regerror_buffsize) },
{ "replace", MOD_PND, MOD_STR, REPLACE_MODSIZE, PO(replacement) },
{ "stackguard", MOD_PAT, MOD_INT, 0, PO(stackguard_test) },
{ "start_optimize", MOD_CTC, MOD_OPTMZ, PCRE2_START_OPTIMIZE, 0 },
{ "start_optimize_off", MOD_CTC, MOD_OPTMZ, PCRE2_START_OPTIMIZE_OFF, 0 },
{ "startchar", MOD_PND, MOD_CTL, CTL_STARTCHAR, PO(control) },
{ "startoffset", MOD_DAT, MOD_INT, 0, DO(offset) },
{ "subject_literal", MOD_PATP, MOD_CTL, CTL2_SUBJECT_LITERAL, PO(control2) },
{ "substitute_callout", MOD_PND, MOD_CTL, CTL2_SUBSTITUTE_CALLOUT, PO(control2) },
{ "substitute_case_callout", MOD_PND, MOD_CTL, CTL2_SUBSTITUTE_CASE_CALLOUT, PO(control2) },
{ "substitute_extended", MOD_PND, MOD_CTL, CTL2_SUBSTITUTE_EXTENDED, PO(control2) },
{ "substitute_literal", MOD_PND, MOD_CTL, CTL2_SUBSTITUTE_LITERAL, PO(control2) },
{ "substitute_matched", MOD_PND, MOD_CTL, CTL2_SUBSTITUTE_MATCHED, PO(control2) },
{ "substitute_overflow_length", MOD_PND, MOD_CTL, CTL2_SUBSTITUTE_OVERFLOW_LENGTH, PO(control2) },
{ "substitute_replacement_only", MOD_PND, MOD_CTL, CTL2_SUBSTITUTE_REPLACEMENT_ONLY, PO(control2) },
{ "substitute_skip", MOD_PND, MOD_INT, 0, PO(substitute_skip) },
{ "substitute_stop", MOD_PND, MOD_INT, 0, PO(substitute_stop) },
{ "substitute_unknown_unset", MOD_PND, MOD_CTL, CTL2_SUBSTITUTE_UNKNOWN_UNSET, PO(control2) },
{ "substitute_unset_empty", MOD_PND, MOD_CTL, CTL2_SUBSTITUTE_UNSET_EMPTY, PO(control2) },
{ "tables", MOD_PAT, MOD_INT, 0, PO(tables_id) },
{ "turkish_casing", MOD_CTC, MOD_OPT, PCRE2_EXTRA_TURKISH_CASING, CO(extra_options) },
{ "ucp", MOD_PATP, MOD_OPT, PCRE2_UCP, PO(options) },
{ "ungreedy", MOD_PAT, MOD_OPT, PCRE2_UNGREEDY, PO(options) },
{ "use_length", MOD_PAT, MOD_CTL, CTL_USE_LENGTH, PO(control) },
{ "use_offset_limit", MOD_PAT, MOD_OPT, PCRE2_USE_OFFSET_LIMIT, PO(options) },
{ "utf", MOD_PATP, MOD_OPT, PCRE2_UTF, PO(options) },
{ "utf8_input", MOD_PAT, MOD_CTL, CTL_UTF8_INPUT, PO(control) },
{ "zero_terminate", MOD_DAT, MOD_CTL, CTL_ZERO_TERMINATE, DO(control) }
};
#define MODLISTCOUNT sizeof(modlist)/sizeof(modstruct)
/* Controls and options that are supported for use with the POSIX interface. */
#define POSIX_SUPPORTED_COMPILE_OPTIONS ( \
PCRE2_CASELESS|PCRE2_DOTALL|PCRE2_LITERAL|PCRE2_MULTILINE|PCRE2_UCP| \
PCRE2_UTF|PCRE2_UNGREEDY)
#define POSIX_SUPPORTED_COMPILE_EXTRA_OPTIONS (0)
#define POSIX_SUPPORTED_COMPILE_CONTROLS ( \
CTL_AFTERTEXT|CTL_ALLAFTERTEXT|CTL_EXPAND|CTL_HEXPAT|CTL_POSIX| \
CTL_POSIX_NOSUB|CTL_USE_LENGTH)
#define POSIX_SUPPORTED_COMPILE_CONTROLS2 (0)
#define POSIX_SUPPORTED_MATCH_OPTIONS ( \
PCRE2_NOTBOL|PCRE2_NOTEMPTY|PCRE2_NOTEOL)
#define POSIX_SUPPORTED_MATCH_CONTROLS (CTL_AFTERTEXT|CTL_ALLAFTERTEXT)
#define POSIX_SUPPORTED_MATCH_CONTROLS2 (CTL2_NULL_SUBJECT)
/* Control bits that are not ignored with 'push'. */
#define PUSH_SUPPORTED_COMPILE_CONTROLS ( \
CTL_BINCODE|CTL_CALLOUT_INFO|CTL_FULLBINCODE|CTL_HEXPAT|CTL_INFO| \
CTL_JITVERIFY|CTL_MEMORY|CTL_PUSH|CTL_PUSHCOPY| \
CTL_PUSHTABLESCOPY|CTL_USE_LENGTH)
#define PUSH_SUPPORTED_COMPILE_CONTROLS2 (CTL2_BSR_SET| \
CTL2_HEAPFRAMES_SIZE|CTL2_FRAMESIZE|CTL2_NL_SET)
/* Controls that apply only at compile time with 'push'. */
#define PUSH_COMPILE_ONLY_CONTROLS CTL_JITVERIFY
#define PUSH_COMPILE_ONLY_CONTROLS2 (0)
/* Controls that are forbidden with #pop or #popcopy. */
#define NOTPOP_CONTROLS (CTL_HEXPAT|CTL_POSIX|CTL_POSIX_NOSUB|CTL_PUSH| \
CTL_PUSHCOPY|CTL_PUSHTABLESCOPY|CTL_USE_LENGTH)
/* Pattern controls that are mutually exclusive. At present these are all in
the first control word. Note that CTL_POSIX_NOSUB is always accompanied by
CTL_POSIX, so it doesn't need its own entries. */
static uint32_t exclusive_pat_controls[] = {
CTL_POSIX | CTL_PUSH,
CTL_POSIX | CTL_PUSHCOPY,
CTL_POSIX | CTL_PUSHTABLESCOPY,
CTL_PUSH | CTL_PUSHCOPY,
CTL_PUSH | CTL_PUSHTABLESCOPY,
CTL_PUSHCOPY | CTL_PUSHTABLESCOPY,
CTL_EXPAND | CTL_HEXPAT };
/* Data controls that are mutually exclusive. At present these are all in the
first control word. */
static uint32_t exclusive_dat_controls[] = {
CTL_ALLUSEDTEXT | CTL_STARTCHAR,
CTL_FINDLIMITS | CTL_NULLCONTEXT,
CTL_FINDLIMITS_NOHEAP | CTL_NULLCONTEXT };
/* Table of single-character abbreviated modifiers. The index field is
initialized to -1, but the first time the modifier is encountered, it is filled
in with the index of the full entry in modlist, to save repeated searching when
processing multiple test items. This short list is searched serially, so its
order does not matter. */
typedef struct c1modstruct {
const char *fullname;
uint32_t onechar;
int index;
} c1modstruct;
static c1modstruct c1modlist[] = {
{ "bincode", 'B', -1 },
{ "info", 'I', -1 },
{ "ascii_all", 'a', -1 },
{ "global", 'g', -1 },
{ "caseless", 'i', -1 },
{ "multiline", 'm', -1 },
{ "no_auto_capture", 'n', -1 },
{ "caseless_restrict", 'r', -1 },
{ "dotall", 's', -1 },
{ "extended", 'x', -1 }
};
#define C1MODLISTCOUNT sizeof(c1modlist)/sizeof(c1modstruct)
/* Table of arguments for the -C command line option. Use macros to make the
table itself easier to read. */
#if defined SUPPORT_PCRE2_8
#define SUPPORT_8 1
#endif
#if defined SUPPORT_PCRE2_16
#define SUPPORT_16 1
#endif
#if defined SUPPORT_PCRE2_32
#define SUPPORT_32 1
#endif
#ifndef SUPPORT_8
#define SUPPORT_8 0
#endif
#ifndef SUPPORT_16
#define SUPPORT_16 0
#endif
#ifndef SUPPORT_32
#define SUPPORT_32 0
#endif
#ifdef EBCDIC
#define SUPPORT_EBCDIC 1
#define EBCDIC_NL CHAR_LF
#else
#define SUPPORT_EBCDIC 0
#define EBCDIC_NL 0
#endif
#ifdef NEVER_BACKSLASH_C
#define BACKSLASH_C 0
#else
#define BACKSLASH_C 1
#endif
typedef struct coptstruct {
const char *name;
uint32_t type;
uint32_t value;
} coptstruct;
enum { CONF_BSR,
CONF_FIX,
CONF_FIZ,
CONF_INT,
CONF_NL,
CONF_JU
};
static coptstruct coptlist[] = {
{ "backslash-C", CONF_FIX, BACKSLASH_C },
{ "bsr", CONF_BSR, PCRE2_CONFIG_BSR },
{ "ebcdic", CONF_FIX, SUPPORT_EBCDIC },
{ "ebcdic-nl", CONF_FIZ, EBCDIC_NL },
{ "jit", CONF_INT, PCRE2_CONFIG_JIT },
{ "jitusable", CONF_JU, 0 },
{ "linksize", CONF_INT, PCRE2_CONFIG_LINKSIZE },
{ "newline", CONF_NL, PCRE2_CONFIG_NEWLINE },
{ "pcre2-16", CONF_FIX, SUPPORT_16 },
{ "pcre2-32", CONF_FIX, SUPPORT_32 },
{ "pcre2-8", CONF_FIX, SUPPORT_8 },
{ "unicode", CONF_INT, PCRE2_CONFIG_UNICODE }
};
#define COPTLISTCOUNT sizeof(coptlist)/sizeof(coptstruct)
#undef SUPPORT_8
#undef SUPPORT_16
#undef SUPPORT_32
#undef SUPPORT_EBCDIC
/* Types for the parser, to be used in process_data() */
enum force_encoding {
FORCE_NONE, /* No preference, follow utf modifier */
FORCE_RAW, /* Encode as a code point or error if too wide */
FORCE_UTF /* Encode as a character or error if too wide */
};
/* ----------------------- Static variables ------------------------ */
static FILE *infile;
static FILE *outfile;
static const void *last_callout_mark;
static PCRE2_JIT_STACK *jit_stack = NULL;
static size_t jit_stack_size = 0;
static BOOL first_callout;
static BOOL jit_was_used;
static BOOL restrict_for_perl_test = FALSE;
static BOOL show_memory = FALSE;
static int jitrc; /* Return from JIT compile */
static int test_mode = DEFAULT_TEST_MODE;
static int timeit = 0;
static int timeitm = 0;
clock_t total_compile_time = 0;
clock_t total_jit_compile_time = 0;
clock_t total_match_time = 0;
static uint32_t code_unit_size; /* Bytes */
static uint32_t dfa_matched;
static uint32_t forbid_utf = 0;
static uint32_t maxlookbehind;
static uint32_t max_oveccount;
static uint32_t callout_count;
static uint32_t maxcapcount;
static uint16_t local_newline_default = 0;
static VERSION_TYPE jittarget[VERSION_SIZE];
static VERSION_TYPE version[VERSION_SIZE];
static VERSION_TYPE uversion[VERSION_SIZE];
static patctl def_patctl;
static patctl pat_patctl;
static datctl def_datctl;
static datctl dat_datctl;
static void *patstack[PATSTACKSIZE];
static int patstacknext = 0;
static void *malloclist[MALLOCLISTSIZE];
static PCRE2_SIZE malloclistlength[MALLOCLISTSIZE];
static uint32_t malloclistptr = 0;
#ifdef SUPPORT_PCRE2_8
static regex_t preg = { NULL, NULL, 0, 0, 0, 0 };
#endif
static int *dfa_workspace = NULL;
static const uint8_t *locale_tables = NULL;
static const uint8_t *use_tables = NULL;
static uint8_t locale_name[32];
static uint8_t *tables3 = NULL; /* For binary-loaded tables */
static uint32_t loadtables_length = 0;
/* We need buffers for building 16/32-bit strings; 8-bit strings don't need
rebuilding, but set up the same naming scheme for use in macros. The "buffer"
buffer is where all input lines are read. Its size is the same as pbuffer8.
Pattern lines are always copied to pbuffer8 for use in callouts, even if they
are actually compiled from pbuffer16 or pbuffer32. */
static size_t pbuffer8_size = 50000; /* Initial size, bytes */
static uint8_t *pbuffer8 = NULL;
static uint8_t *buffer = NULL;
/* The dbuffer is where all processed data lines are put. In non-8-bit modes it
is cast as needed. For long data lines it grows as necessary. */
static size_t dbuffer_size = 1u << 14; /* Initial size, bytes */
static uint8_t *dbuffer = NULL;
/* ---------------- Mode-dependent variables -------------------*/
#ifdef SUPPORT_PCRE2_8
static pcre2_code_8 *compiled_code8;
static pcre2_general_context_8 *general_context8, *general_context_copy8;
static pcre2_compile_context_8 *pat_context8, *default_pat_context8;
static pcre2_convert_context_8 *con_context8, *default_con_context8;
static pcre2_match_context_8 *dat_context8, *default_dat_context8;
static pcre2_match_data_8 *match_data8;
#endif
#ifdef SUPPORT_PCRE2_16
static pcre2_code_16 *compiled_code16;
static pcre2_general_context_16 *general_context16, *general_context_copy16;
static pcre2_compile_context_16 *pat_context16, *default_pat_context16;
static pcre2_convert_context_16 *con_context16, *default_con_context16;
static pcre2_match_context_16 *dat_context16, *default_dat_context16;
static pcre2_match_data_16 *match_data16;
static PCRE2_SIZE pbuffer16_size = 0; /* Set only when needed */
static uint16_t *pbuffer16 = NULL;
#endif
#ifdef SUPPORT_PCRE2_32
static pcre2_code_32 *compiled_code32;
static pcre2_general_context_32 *general_context32, *general_context_copy32;
static pcre2_compile_context_32 *pat_context32, *default_pat_context32;
static pcre2_convert_context_32 *con_context32, *default_con_context32;
static pcre2_match_context_32 *dat_context32, *default_dat_context32;
static pcre2_match_data_32 *match_data32;
static PCRE2_SIZE pbuffer32_size = 0; /* Set only when needed */
static uint32_t *pbuffer32 = NULL;
#endif
/* ---------------- Macros that work in all modes ----------------- */
#define CAST8VAR(x) CASTVAR(uint8_t *, x)
#define SET(x,y) SETOP(x,y,=)
#define SETPLUS(x,y) SETOP(x,y,+=)
#define strlen8(x) strlen((char *)x)
/* ---------------- Mode-dependent, runtime-testing macros ------------------*/
/* Define macros for variables and functions that must be selected dynamically
depending on the mode setting (8, 16, 32). These are dependent on which modes
are supported. */
#if (defined (SUPPORT_PCRE2_8) + defined (SUPPORT_PCRE2_16) + \
defined (SUPPORT_PCRE2_32)) >= 2
/* ----- All three modes supported ----- */
#if defined(SUPPORT_PCRE2_8) && defined(SUPPORT_PCRE2_16) && defined(SUPPORT_PCRE2_32)
#define CASTFLD(t,a,b) ((test_mode == PCRE8_MODE)? (t)(G(a,8)->b) : \
(test_mode == PCRE16_MODE)? (t)(G(a,16)->b) : (t)(G(a,32)->b))
#define CASTVAR(t,x) ( \
(test_mode == PCRE8_MODE)? (t)G(x,8) : \
(test_mode == PCRE16_MODE)? (t)G(x,16) : (t)G(x,32))
#define CODE_UNIT(a,b) ( \
(test_mode == PCRE8_MODE)? (uint32_t)(((PCRE2_SPTR8)(a))[b]) : \
(test_mode == PCRE16_MODE)? (uint32_t)(((PCRE2_SPTR16)(a))[b]) : \
(uint32_t)(((PCRE2_SPTR32)(a))[b]))
#define CONCTXCPY(a,b) \
if (test_mode == PCRE8_MODE) \
memcpy(G(a,8),G(b,8),sizeof(pcre2_convert_context_8)); \
else if (test_mode == PCRE16_MODE) \
memcpy(G(a,16),G(b,16),sizeof(pcre2_convert_context_16)); \
else memcpy(G(a,32),G(b,32),sizeof(pcre2_convert_context_32))
#define CONVERT_COPY(a,b,c) \
if (test_mode == PCRE8_MODE) \
memcpy(G(a,8),(char *)b,c); \
else if (test_mode == PCRE16_MODE) \
memcpy(G(a,16),(char *)b,(c)*2); \
else if (test_mode == PCRE32_MODE) \
memcpy(G(a,32),(char *)b,(c)*4)
#define DATCTXCPY(a,b) \
if (test_mode == PCRE8_MODE) \
memcpy(G(a,8),G(b,8),sizeof(pcre2_match_context_8)); \
else if (test_mode == PCRE16_MODE) \
memcpy(G(a,16),G(b,16),sizeof(pcre2_match_context_16)); \
else memcpy(G(a,32),G(b,32),sizeof(pcre2_match_context_32))
#define FLD(a,b) ((test_mode == PCRE8_MODE)? G(a,8)->b : \
(test_mode == PCRE16_MODE)? G(a,16)->b : G(a,32)->b)
#define PATCTXCPY(a,b) \
if (test_mode == PCRE8_MODE) \
memcpy(G(a,8),G(b,8),sizeof(pcre2_compile_context_8)); \
else if (test_mode == PCRE16_MODE) \
memcpy(G(a,16),G(b,16),sizeof(pcre2_compile_context_16)); \
else memcpy(G(a,32),G(b,32),sizeof(pcre2_compile_context_32))
#define PCHARS(lv, p, offset, len, utf, f) \
if (test_mode == PCRE32_MODE) \
lv = pchars32((PCRE2_SPTR32)(p)+offset, len, utf, f); \
else if (test_mode == PCRE16_MODE) \
lv = pchars16((PCRE2_SPTR16)(p)+offset, len, utf, f); \
else \
lv = pchars8((PCRE2_SPTR8)(p)+offset, len, utf, f)
#define PCHARSV(p, offset, len, utf, f) \
if (test_mode == PCRE32_MODE) \
(void)pchars32((PCRE2_SPTR32)(p)+offset, len, utf, f); \
else if (test_mode == PCRE16_MODE) \
(void)pchars16((PCRE2_SPTR16)(p)+offset, len, utf, f); \
else \
(void)pchars8((PCRE2_SPTR8)(p)+offset, len, utf, f)
#define PCRE2_CALLOUT_ENUMERATE(a,b,c) \
if (test_mode == PCRE8_MODE) \
a = pcre2_callout_enumerate_8(compiled_code8, \
(int (*)(struct pcre2_callout_enumerate_block_8 *, void *))b,c); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_callout_enumerate_16(compiled_code16, \
(int(*)(struct pcre2_callout_enumerate_block_16 *, void *))b,c); \
else \
a = pcre2_callout_enumerate_32(compiled_code32, \
(int (*)(struct pcre2_callout_enumerate_block_32 *, void *))b,c)
#define PCRE2_CODE_COPY_FROM_VOID(a,b) \
if (test_mode == PCRE8_MODE) \
G(a,8) = pcre2_code_copy_8(b); \
else if (test_mode == PCRE16_MODE) \
G(a,16) = pcre2_code_copy_16(b); \
else \
G(a,32) = pcre2_code_copy_32(b)
#define PCRE2_CODE_COPY_TO_VOID(a,b) \
if (test_mode == PCRE8_MODE) \
a = (void *)pcre2_code_copy_8(G(b,8)); \
else if (test_mode == PCRE16_MODE) \
a = (void *)pcre2_code_copy_16(G(b,16)); \
else \
a = (void *)pcre2_code_copy_32(G(b,32))
#define PCRE2_CODE_COPY_WITH_TABLES_TO_VOID(a,b) \
if (test_mode == PCRE8_MODE) \
a = (void *)pcre2_code_copy_with_tables_8(G(b,8)); \
else if (test_mode == PCRE16_MODE) \
a = (void *)pcre2_code_copy_with_tables_16(G(b,16)); \
else \
a = (void *)pcre2_code_copy_with_tables_32(G(b,32))
#define PCRE2_COMPILE(a,b,c,d,e,f,g) \
if (test_mode == PCRE8_MODE) \
G(a,8) = pcre2_compile_8(b,c,d,e,f,g); \
else if (test_mode == PCRE16_MODE) \
G(a,16) = pcre2_compile_16(b,c,d,e,f,g); \
else \
G(a,32) = pcre2_compile_32(b,c,d,e,f,g)
#define PCRE2_CONVERTED_PATTERN_FREE(a) \
if (test_mode == PCRE8_MODE) pcre2_converted_pattern_free_8((PCRE2_UCHAR8 *)a); \
else if (test_mode == PCRE16_MODE) pcre2_converted_pattern_free_16((PCRE2_UCHAR16 *)a); \
else pcre2_converted_pattern_free_32((PCRE2_UCHAR32 *)a)
#define PCRE2_DFA_MATCH(a,b,c,d,e,f,g,h,i,j) \
if (test_mode == PCRE8_MODE) \
a = pcre2_dfa_match_8(G(b,8),(PCRE2_SPTR8)c,d,e,f,G(g,8),h,i,j); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_dfa_match_16(G(b,16),(PCRE2_SPTR16)c,d,e,f,G(g,16),h,i,j); \
else \
a = pcre2_dfa_match_32(G(b,32),(PCRE2_SPTR32)c,d,e,f,G(g,32),h,i,j)
#define PCRE2_GET_ERROR_MESSAGE(r,a,b) \
if (test_mode == PCRE8_MODE) \
r = pcre2_get_error_message_8(a,G(b,8),G(G(b,8),_size)); \
else if (test_mode == PCRE16_MODE) \
r = pcre2_get_error_message_16(a,G(b,16),G(G(b,16),_size/2)); \
else \
r = pcre2_get_error_message_32(a,G(b,32),G(G(b,32),_size/4))
#define PCRE2_GET_MATCH_DATA_HEAPFRAMES_SIZE(r,a) \
if (test_mode == PCRE8_MODE) \
r = pcre2_get_match_data_heapframes_size_8(G(a,8)); \
else if (test_mode == PCRE16_MODE) \
r = pcre2_get_match_data_heapframes_size_16(G(a,16)); \
else \
r = pcre2_get_match_data_heapframes_size_32(G(a,32))
#define PCRE2_GET_OVECTOR_COUNT(a,b) \
if (test_mode == PCRE8_MODE) \
a = pcre2_get_ovector_count_8(G(b,8)); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_get_ovector_count_16(G(b,16)); \
else \
a = pcre2_get_ovector_count_32(G(b,32))
#define PCRE2_GET_STARTCHAR(a,b) \
if (test_mode == PCRE8_MODE) \
a = pcre2_get_startchar_8(G(b,8)); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_get_startchar_16(G(b,16)); \
else \
a = pcre2_get_startchar_32(G(b,32))
#define PCRE2_JIT_COMPILE(r,a,b) \
if (test_mode == PCRE8_MODE) r = pcre2_jit_compile_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) r = pcre2_jit_compile_16(G(a,16),b); \
else r = pcre2_jit_compile_32(G(a,32),b)
#define PCRE2_JIT_FREE_UNUSED_MEMORY(a) \
if (test_mode == PCRE8_MODE) pcre2_jit_free_unused_memory_8(G(a,8)); \
else if (test_mode == PCRE16_MODE) pcre2_jit_free_unused_memory_16(G(a,16)); \
else pcre2_jit_free_unused_memory_32(G(a,32))
#define PCRE2_JIT_MATCH(a,b,c,d,e,f,g,h) \
if (test_mode == PCRE8_MODE) \
a = pcre2_jit_match_8(G(b,8),(PCRE2_SPTR8)c,d,e,f,G(g,8),h); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_jit_match_16(G(b,16),(PCRE2_SPTR16)c,d,e,f,G(g,16),h); \
else \
a = pcre2_jit_match_32(G(b,32),(PCRE2_SPTR32)c,d,e,f,G(g,32),h)
#define PCRE2_JIT_STACK_CREATE(a,b,c,d) \
if (test_mode == PCRE8_MODE) \
a = (PCRE2_JIT_STACK *)pcre2_jit_stack_create_8(b,c,d); \
else if (test_mode == PCRE16_MODE) \
a = (PCRE2_JIT_STACK *)pcre2_jit_stack_create_16(b,c,d); \
else \
a = (PCRE2_JIT_STACK *)pcre2_jit_stack_create_32(b,c,d);
#define PCRE2_JIT_STACK_ASSIGN(a,b,c) \
if (test_mode == PCRE8_MODE) \
pcre2_jit_stack_assign_8(G(a,8),(pcre2_jit_callback_8)b,c); \
else if (test_mode == PCRE16_MODE) \
pcre2_jit_stack_assign_16(G(a,16),(pcre2_jit_callback_16)b,c); \
else \
pcre2_jit_stack_assign_32(G(a,32),(pcre2_jit_callback_32)b,c);
#define PCRE2_JIT_STACK_FREE(a) \
if (test_mode == PCRE8_MODE) \
pcre2_jit_stack_free_8((pcre2_jit_stack_8 *)a); \
else if (test_mode == PCRE16_MODE) \
pcre2_jit_stack_free_16((pcre2_jit_stack_16 *)a); \
else \
pcre2_jit_stack_free_32((pcre2_jit_stack_32 *)a);
#define PCRE2_MAKETABLES(a,c) \
if (test_mode == PCRE8_MODE) a = pcre2_maketables_8(G(c,8)); \
else if (test_mode == PCRE16_MODE) a = pcre2_maketables_16(G(c,16)); \
else a = pcre2_maketables_32(G(c,32))
#define PCRE2_MAKETABLES_FREE(c,a) \
if (test_mode == PCRE8_MODE) pcre2_maketables_free_8(G(c,8),a); \
else if (test_mode == PCRE16_MODE) pcre2_maketables_free_16(G(c,16),a); \
else pcre2_maketables_free_32(G(c,32),a)
#define PCRE2_MATCH(a,b,c,d,e,f,g,h) \
if (test_mode == PCRE8_MODE) \
a = pcre2_match_8(G(b,8),(PCRE2_SPTR8)c,d,e,f,G(g,8),h); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_match_16(G(b,16),(PCRE2_SPTR16)c,d,e,f,G(g,16),h); \
else \
a = pcre2_match_32(G(b,32),(PCRE2_SPTR32)c,d,e,f,G(g,32),h)
#define PCRE2_MATCH_DATA_CREATE(a,b,c) \
if (test_mode == PCRE8_MODE) \
G(a,8) = pcre2_match_data_create_8(b,G(c,8)); \
else if (test_mode == PCRE16_MODE) \
G(a,16) = pcre2_match_data_create_16(b,G(c,16)); \
else \
G(a,32) = pcre2_match_data_create_32(b,G(c,32))
#define PCRE2_MATCH_DATA_CREATE_FROM_PATTERN(a,b,c) \
if (test_mode == PCRE8_MODE) \
G(a,8) = pcre2_match_data_create_from_pattern_8(G(b,8),G(c,8)); \
else if (test_mode == PCRE16_MODE) \
G(a,16) = pcre2_match_data_create_from_pattern_16(G(b,16),G(c,16)); \
else \
G(a,32) = pcre2_match_data_create_from_pattern_32(G(b,32),G(c,32))
#define PCRE2_MATCH_DATA_FREE(a) \
if (test_mode == PCRE8_MODE) \
pcre2_match_data_free_8(G(a,8)); \
else if (test_mode == PCRE16_MODE) \
pcre2_match_data_free_16(G(a,16)); \
else \
pcre2_match_data_free_32(G(a,32))
#define PCRE2_PATTERN_CONVERT(a,b,c,d,e,f,g) \
if (test_mode == PCRE8_MODE) \
a = pcre2_pattern_convert_8(G(b,8),c,d,(PCRE2_UCHAR8 **)e,f,G(g,8)); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_pattern_convert_16(G(b,16),c,d,(PCRE2_UCHAR16 **)e,f,G(g,16)); \
else \
a = pcre2_pattern_convert_32(G(b,32),c,d,(PCRE2_UCHAR32 **)e,f,G(g,32))
#define PCRE2_PATTERN_INFO(a,b,c,d) \
if (test_mode == PCRE8_MODE) \
a = pcre2_pattern_info_8(G(b,8),c,d); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_pattern_info_16(G(b,16),c,d); \
else \
a = pcre2_pattern_info_32(G(b,32),c,d)
#define PCRE2_PRINTINT(a) \
if (test_mode == PCRE8_MODE) \
pcre2_printint_8(compiled_code8,outfile,a); \
else if (test_mode == PCRE16_MODE) \
pcre2_printint_16(compiled_code16,outfile,a); \
else \
pcre2_printint_32(compiled_code32,outfile,a)
#define PCRE2_SERIALIZE_DECODE(r,a,b,c,d) \
if (test_mode == PCRE8_MODE) \
r = pcre2_serialize_decode_8((pcre2_code_8 **)a,b,c,G(d,8)); \
else if (test_mode == PCRE16_MODE) \
r = pcre2_serialize_decode_16((pcre2_code_16 **)a,b,c,G(d,16)); \
else \
r = pcre2_serialize_decode_32((pcre2_code_32 **)a,b,c,G(d,32))
#define PCRE2_SERIALIZE_ENCODE(r,a,b,c,d,e) \
if (test_mode == PCRE8_MODE) \
r = pcre2_serialize_encode_8((const pcre2_code_8 **)a,b,c,d,G(e,8)); \
else if (test_mode == PCRE16_MODE) \
r = pcre2_serialize_encode_16((const pcre2_code_16 **)a,b,c,d,G(e,16)); \
else \
r = pcre2_serialize_encode_32((const pcre2_code_32 **)a,b,c,d,G(e,32))
#define PCRE2_SERIALIZE_FREE(a) \
if (test_mode == PCRE8_MODE) \
pcre2_serialize_free_8(a); \
else if (test_mode == PCRE16_MODE) \
pcre2_serialize_free_16(a); \
else \
pcre2_serialize_free_32(a)
#define PCRE2_SERIALIZE_GET_NUMBER_OF_CODES(r,a) \
if (test_mode == PCRE8_MODE) \
r = pcre2_serialize_get_number_of_codes_8(a); \
else if (test_mode == PCRE16_MODE) \
r = pcre2_serialize_get_number_of_codes_16(a); \
else \
r = pcre2_serialize_get_number_of_codes_32(a); \
#define PCRE2_SET_CALLOUT(a,b,c) \
if (test_mode == PCRE8_MODE) \
pcre2_set_callout_8(G(a,8),(int (*)(pcre2_callout_block_8 *, void *))b,c); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_callout_16(G(a,16),(int (*)(pcre2_callout_block_16 *, void *))b,c); \
else \
pcre2_set_callout_32(G(a,32),(int (*)(pcre2_callout_block_32 *, void *))b,c);
#define PCRE2_SET_CHARACTER_TABLES(a,b) \
if (test_mode == PCRE8_MODE) \
pcre2_set_character_tables_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_character_tables_16(G(a,16),b); \
else \
pcre2_set_character_tables_32(G(a,32),b)
#define PCRE2_SET_COMPILE_RECURSION_GUARD(a,b,c) \
if (test_mode == PCRE8_MODE) \
pcre2_set_compile_recursion_guard_8(G(a,8),b,c); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_compile_recursion_guard_16(G(a,16),b,c); \
else \
pcre2_set_compile_recursion_guard_32(G(a,32),b,c)
#define PCRE2_SET_DEPTH_LIMIT(a,b) \
if (test_mode == PCRE8_MODE) \
pcre2_set_depth_limit_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_depth_limit_16(G(a,16),b); \
else \
pcre2_set_depth_limit_32(G(a,32),b)
#define PCRE2_SET_GLOB_SEPARATOR(r,a,b) \
if (test_mode == PCRE8_MODE) \
r = pcre2_set_glob_separator_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
r = pcre2_set_glob_separator_16(G(a,16),b); \
else \
r = pcre2_set_glob_separator_32(G(a,32),b)
#define PCRE2_SET_GLOB_ESCAPE(r,a,b) \
if (test_mode == PCRE8_MODE) \
r = pcre2_set_glob_escape_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
r = pcre2_set_glob_escape_16(G(a,16),b); \
else \
r = pcre2_set_glob_escape_32(G(a,32),b)
#define PCRE2_SET_HEAP_LIMIT(a,b) \
if (test_mode == PCRE8_MODE) \
pcre2_set_heap_limit_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_heap_limit_16(G(a,16),b); \
else \
pcre2_set_heap_limit_32(G(a,32),b)
#define PCRE2_SET_MATCH_LIMIT(a,b) \
if (test_mode == PCRE8_MODE) \
pcre2_set_match_limit_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_match_limit_16(G(a,16),b); \
else \
pcre2_set_match_limit_32(G(a,32),b)
#define PCRE2_SET_MAX_PATTERN_COMPILED_LENGTH(a,b) \
if (test_mode == PCRE8_MODE) \
pcre2_set_max_pattern_compiled_length_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_max_pattern_compiled_length_16(G(a,16),b); \
else \
pcre2_set_max_pattern_compiled_length_32(G(a,32),b)
#define PCRE2_SET_MAX_PATTERN_LENGTH(a,b) \
if (test_mode == PCRE8_MODE) \
pcre2_set_max_pattern_length_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_max_pattern_length_16(G(a,16),b); \
else \
pcre2_set_max_pattern_length_32(G(a,32),b)
#define PCRE2_SET_MAX_VARLOOKBEHIND(a,b) \
if (test_mode == PCRE8_MODE) \
pcre2_set_max_varlookbehind_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_max_varlookbehind_16(G(a,16),b); \
else \
pcre2_set_max_varlookbehind_32(G(a,32),b)
#define PCRE2_SET_OFFSET_LIMIT(a,b) \
if (test_mode == PCRE8_MODE) \
pcre2_set_offset_limit_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_offset_limit_16(G(a,16),b); \
else \
pcre2_set_offset_limit_32(G(a,32),b)
#define PCRE2_SET_PARENS_NEST_LIMIT(a,b) \
if (test_mode == PCRE8_MODE) \
pcre2_set_parens_nest_limit_8(G(a,8),b); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_parens_nest_limit_16(G(a,16),b); \
else \
pcre2_set_parens_nest_limit_32(G(a,32),b)
#define PCRE2_SET_SUBSTITUTE_CALLOUT(a,b,c) \
if (test_mode == PCRE8_MODE) \
pcre2_set_substitute_callout_8(G(a,8), \
(int (*)(pcre2_substitute_callout_block_8 *, void *))b,c); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_substitute_callout_16(G(a,16), \
(int (*)(pcre2_substitute_callout_block_16 *, void *))b,c); \
else \
pcre2_set_substitute_callout_32(G(a,32), \
(int (*)(pcre2_substitute_callout_block_32 *, void *))b,c)
#define PCRE2_SET_SUBSTITUTE_CASE_CALLOUT(a,b,c) \
if (test_mode == PCRE8_MODE) \
pcre2_set_substitute_case_callout_8(G(a,8),G(b,8),c); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_substitute_case_callout_16(G(a,16),G(b,16),c); \
else \
pcre2_set_substitute_case_callout_32(G(a,32),G(b,32),c)
#define PCRE2_SET_SUBSTITUTE_CASE_CALLOUT_NULL(a) \
if (test_mode == PCRE8_MODE) \
pcre2_set_substitute_case_callout_8(G(a,8),NULL,NULL); \
else if (test_mode == PCRE16_MODE) \
pcre2_set_substitute_case_callout_16(G(a,16),NULL,NULL); \
else \
pcre2_set_substitute_case_callout_32(G(a,32),NULL,NULL)
#define PCRE2_SUBSTITUTE(a,b,c,d,e,f,g,h,i,j,k,l) \
if (test_mode == PCRE8_MODE) \
a = pcre2_substitute_8(G(b,8),(PCRE2_SPTR8)c,d,e,f,G(g,8),h, \
(PCRE2_SPTR8)i,j,(PCRE2_UCHAR8 *)k,l); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_substitute_16(G(b,16),(PCRE2_SPTR16)c,d,e,f,G(g,16),h, \
(PCRE2_SPTR16)i,j,(PCRE2_UCHAR16 *)k,l); \
else \
a = pcre2_substitute_32(G(b,32),(PCRE2_SPTR32)c,d,e,f,G(g,32),h, \
(PCRE2_SPTR32)i,j,(PCRE2_UCHAR32 *)k,l)
#define PCRE2_SUBSTRING_COPY_BYNAME(a,b,c,d,e) \
if (test_mode == PCRE8_MODE) \
a = pcre2_substring_copy_byname_8(G(b,8),G(c,8),(PCRE2_UCHAR8 *)d,e); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_substring_copy_byname_16(G(b,16),G(c,16),(PCRE2_UCHAR16 *)d,e); \
else \
a = pcre2_substring_copy_byname_32(G(b,32),G(c,32),(PCRE2_UCHAR32 *)d,e)
#define PCRE2_SUBSTRING_COPY_BYNUMBER(a,b,c,d,e) \
if (test_mode == PCRE8_MODE) \
a = pcre2_substring_copy_bynumber_8(G(b,8),c,(PCRE2_UCHAR8 *)d,e); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_substring_copy_bynumber_16(G(b,16),c,(PCRE2_UCHAR16 *)d,e); \
else \
a = pcre2_substring_copy_bynumber_32(G(b,32),c,(PCRE2_UCHAR32 *)d,e)
#define PCRE2_SUBSTRING_FREE(a) \
if (test_mode == PCRE8_MODE) pcre2_substring_free_8((PCRE2_UCHAR8 *)a); \
else if (test_mode == PCRE16_MODE) \
pcre2_substring_free_16((PCRE2_UCHAR16 *)a); \
else pcre2_substring_free_32((PCRE2_UCHAR32 *)a)
#define PCRE2_SUBSTRING_GET_BYNAME(a,b,c,d,e) \
if (test_mode == PCRE8_MODE) \
a = pcre2_substring_get_byname_8(G(b,8),G(c,8),(PCRE2_UCHAR8 **)d,e); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_substring_get_byname_16(G(b,16),G(c,16),(PCRE2_UCHAR16 **)d,e); \
else \
a = pcre2_substring_get_byname_32(G(b,32),G(c,32),(PCRE2_UCHAR32 **)d,e)
#define PCRE2_SUBSTRING_GET_BYNUMBER(a,b,c,d,e) \
if (test_mode == PCRE8_MODE) \
a = pcre2_substring_get_bynumber_8(G(b,8),c,(PCRE2_UCHAR8 **)d,e); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_substring_get_bynumber_16(G(b,16),c,(PCRE2_UCHAR16 **)d,e); \
else \
a = pcre2_substring_get_bynumber_32(G(b,32),c,(PCRE2_UCHAR32 **)d,e)
#define PCRE2_SUBSTRING_LENGTH_BYNAME(a,b,c,d) \
if (test_mode == PCRE8_MODE) \
a = pcre2_substring_length_byname_8(G(b,8),G(c,8),d); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_substring_length_byname_16(G(b,16),G(c,16),d); \
else \
a = pcre2_substring_length_byname_32(G(b,32),G(c,32),d)
#define PCRE2_SUBSTRING_LENGTH_BYNUMBER(a,b,c,d) \
if (test_mode == PCRE8_MODE) \
a = pcre2_substring_length_bynumber_8(G(b,8),c,d); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_substring_length_bynumber_16(G(b,16),c,d); \
else \
a = pcre2_substring_length_bynumber_32(G(b,32),c,d)
#define PCRE2_SUBSTRING_LIST_GET(a,b,c,d) \
if (test_mode == PCRE8_MODE) \
a = pcre2_substring_list_get_8(G(b,8),(PCRE2_UCHAR8 ***)c,d); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_substring_list_get_16(G(b,16),(PCRE2_UCHAR16 ***)c,d); \
else \
a = pcre2_substring_list_get_32(G(b,32),(PCRE2_UCHAR32 ***)c,d)
#define PCRE2_SUBSTRING_LIST_FREE(a) \
if (test_mode == PCRE8_MODE) \
pcre2_substring_list_free_8((PCRE2_UCHAR8 **)a); \
else if (test_mode == PCRE16_MODE) \
pcre2_substring_list_free_16((PCRE2_UCHAR16 **)a); \
else \
pcre2_substring_list_free_32((PCRE2_UCHAR32 **)a)
#define PCRE2_SUBSTRING_NUMBER_FROM_NAME(a,b,c) \
if (test_mode == PCRE8_MODE) \
a = pcre2_substring_number_from_name_8(G(b,8),G(c,8)); \
else if (test_mode == PCRE16_MODE) \
a = pcre2_substring_number_from_name_16(G(b,16),G(c,16)); \
else \
a = pcre2_substring_number_from_name_32(G(b,32),G(c,32))
#define PTR(x) ( \
(test_mode == PCRE8_MODE)? (void *)G(x,8) : \
(test_mode == PCRE16_MODE)? (void *)G(x,16) : \
(void *)G(x,32))
#define SETFLD(x,y,z) \
if (test_mode == PCRE8_MODE) G(x,8)->y = z; \
else if (test_mode == PCRE16_MODE) G(x,16)->y = z; \
else G(x,32)->y = z
#define SETFLDVEC(x,y,v,z) \
if (test_mode == PCRE8_MODE) G(x,8)->y[v] = z; \
else if (test_mode == PCRE16_MODE) G(x,16)->y[v] = z; \
else G(x,32)->y[v] = z
#define SETOP(x,y,z) \
if (test_mode == PCRE8_MODE) G(x,8) z y; \
else if (test_mode == PCRE16_MODE) G(x,16) z y; \
else G(x,32) z y
#define SETCASTPTR(x,y) \
if (test_mode == PCRE8_MODE) \
G(x,8) = (uint8_t *)(y); \
else if (test_mode == PCRE16_MODE) \
G(x,16) = (uint16_t *)(y); \
else \
G(x,32) = (uint32_t *)(y)
#define STRLEN(p) ((test_mode == PCRE8_MODE)? ((int)strlen((char *)p)) : \
(test_mode == PCRE16_MODE)? ((int)strlen16((PCRE2_SPTR16)p)) : \
((int)strlen32((PCRE2_SPTR32)p)))
#define SUB1(a,b) \
if (test_mode == PCRE8_MODE) G(a,8)(G(b,8)); \
else if (test_mode == PCRE16_MODE) G(a,16)(G(b,16)); \
else G(a,32)(G(b,32))
#define SUB2(a,b,c) \
if (test_mode == PCRE8_MODE) G(a,8)(G(b,8),G(c,8)); \
else if (test_mode == PCRE16_MODE) G(a,16)(G(b,16),G(c,16)); \
else G(a,32)(G(b,32),G(c,32))
#define TEST(x,r,y) ( \
(test_mode == PCRE8_MODE && G(x,8) r (y)) || \
(test_mode == PCRE16_MODE && G(x,16) r (y)) || \
(test_mode == PCRE32_MODE && G(x,32) r (y)))
#define TESTFLD(x,f,r,y) ( \
(test_mode == PCRE8_MODE && G(x,8)->f r (y)) || \
(test_mode == PCRE16_MODE && G(x,16)->f r (y)) || \
(test_mode == PCRE32_MODE && G(x,32)->f r (y)))
/* ----- Two out of three modes are supported ----- */
#else
/* We can use some macro trickery to make a single set of definitions work in
the three different cases. */
/* ----- 32-bit and 16-bit but not 8-bit supported ----- */
#if defined(SUPPORT_PCRE2_32) && defined(SUPPORT_PCRE2_16)
#define BITONE 32
#define BITTWO 16
/* ----- 32-bit and 8-bit but not 16-bit supported ----- */
#elif defined(SUPPORT_PCRE2_32) && defined(SUPPORT_PCRE2_8)
#define BITONE 32
#define BITTWO 8
/* ----- 16-bit and 8-bit but not 32-bit supported ----- */
#else
#define BITONE 16
#define BITTWO 8
#endif
/* ----- Common macros for two-mode cases ----- */
#define BYTEONE (BITONE/8)
#define BYTETWO (BITTWO/8)
#define CASTFLD(t,a,b) \
((test_mode == G(G(PCRE,BITONE),_MODE))? (t)(G(a,BITONE)->b) : \
(t)(G(a,BITTWO)->b))
#define CASTVAR(t,x) ( \
(test_mode == G(G(PCRE,BITONE),_MODE))? \
(t)G(x,BITONE) : (t)G(x,BITTWO))
#define CODE_UNIT(a,b) ( \
(test_mode == G(G(PCRE,BITONE),_MODE))? \
(uint32_t)(((G(PCRE2_SPTR,BITONE))(a))[b]) : \
(uint32_t)(((G(PCRE2_SPTR,BITTWO))(a))[b]))
#define CONCTXCPY(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
memcpy(G(a,BITONE),G(b,BITONE),sizeof(G(pcre2_convert_context_,BITONE))); \
else \
memcpy(G(a,BITTWO),G(b,BITTWO),sizeof(G(pcre2_convert_context_,BITTWO)))
#define CONVERT_COPY(a,b,c) \
(test_mode == G(G(PCRE,BITONE),_MODE))? \
memcpy(G(a,BITONE),(char *)b,(c)*BYTEONE) : \
memcpy(G(a,BITTWO),(char *)b,(c)*BYTETWO)
#define DATCTXCPY(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
memcpy(G(a,BITONE),G(b,BITONE),sizeof(G(pcre2_match_context_,BITONE))); \
else \
memcpy(G(a,BITTWO),G(b,BITTWO),sizeof(G(pcre2_match_context_,BITTWO)))
#define FLD(a,b) \
((test_mode == G(G(PCRE,BITONE),_MODE))? G(a,BITONE)->b : G(a,BITTWO)->b)
#define PATCTXCPY(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
memcpy(G(a,BITONE),G(b,BITONE),sizeof(G(pcre2_compile_context_,BITONE))); \
else \
memcpy(G(a,BITTWO),G(b,BITTWO),sizeof(G(pcre2_compile_context_,BITTWO)))
#define PCHARS(lv, p, offset, len, utf, f) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
lv = G(pchars,BITONE)((G(PCRE2_SPTR,BITONE))(p)+offset, len, utf, f); \
else \
lv = G(pchars,BITTWO)((G(PCRE2_SPTR,BITTWO))(p)+offset, len, utf, f)
#define PCHARSV(p, offset, len, utf, f) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
(void)G(pchars,BITONE)((G(PCRE2_SPTR,BITONE))(p)+offset, len, utf, f); \
else \
(void)G(pchars,BITTWO)((G(PCRE2_SPTR,BITTWO))(p)+offset, len, utf, f)
#define PCRE2_CALLOUT_ENUMERATE(a,b,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_callout_enumerate,BITONE)(G(compiled_code,BITONE), \
(int (*)(struct G(pcre2_callout_enumerate_block_,BITONE) *, void *))b,c); \
else \
a = G(pcre2_callout_enumerate,BITTWO)(G(compiled_code,BITTWO), \
(int (*)(struct G(pcre2_callout_enumerate_block_,BITTWO) *, void *))b,c)
#define PCRE2_CODE_COPY_FROM_VOID(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(a,BITONE) = G(pcre2_code_copy_,BITONE)(b); \
else \
G(a,BITTWO) = G(pcre2_code_copy_,BITTWO)(b)
#define PCRE2_CODE_COPY_TO_VOID(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = (void *)G(pcre2_code_copy_,BITONE)(G(b,BITONE)); \
else \
a = (void *)G(pcre2_code_copy_,BITTWO)(G(b,BITTWO))
#define PCRE2_CODE_COPY_WITH_TABLES_TO_VOID(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = (void *)G(pcre2_code_copy_with_tables_,BITONE)(G(b,BITONE)); \
else \
a = (void *)G(pcre2_code_copy_with_tables_,BITTWO)(G(b,BITTWO))
#define PCRE2_COMPILE(a,b,c,d,e,f,g) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(a,BITONE) = G(pcre2_compile_,BITONE)(b,c,d,e,f,g); \
else \
G(a,BITTWO) = G(pcre2_compile_,BITTWO)(b,c,d,e,f,g)
#define PCRE2_CONVERTED_PATTERN_FREE(a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_converted_pattern_free_,BITONE)((G(PCRE2_UCHAR,BITONE) *)a); \
else \
G(pcre2_converted_pattern_free_,BITTWO)((G(PCRE2_UCHAR,BITTWO) *)a)
#define PCRE2_DFA_MATCH(a,b,c,d,e,f,g,h,i,j) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_dfa_match_,BITONE)(G(b,BITONE),(G(PCRE2_SPTR,BITONE))c,d,e,f, \
G(g,BITONE),h,i,j); \
else \
a = G(pcre2_dfa_match_,BITTWO)(G(b,BITTWO),(G(PCRE2_SPTR,BITTWO))c,d,e,f, \
G(g,BITTWO),h,i,j)
#define PCRE2_GET_ERROR_MESSAGE(r,a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
r = G(pcre2_get_error_message_,BITONE)(a,G(b,BITONE),G(G(b,BITONE),_size/BYTEONE)); \
else \
r = G(pcre2_get_error_message_,BITTWO)(a,G(b,BITTWO),G(G(b,BITTWO),_size/BYTETWO))
#define PCRE2_GET_MATCH_DATA_HEAPFRAMES_SIZE(r,a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
r = G(pcre2_get_match_data_heapframes_size_,BITONE)(G(a,BITONE)); \
else \
r = G(pcre2_get_match_data_heapframes_size_,BITTWO)(G(a,BITTWO))
#define PCRE2_GET_OVECTOR_COUNT(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_get_ovector_count_,BITONE)(G(b,BITONE)); \
else \
a = G(pcre2_get_ovector_count_,BITTWO)(G(b,BITTWO))
#define PCRE2_GET_STARTCHAR(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_get_startchar_,BITONE)(G(b,BITONE)); \
else \
a = G(pcre2_get_startchar_,BITTWO)(G(b,BITTWO))
#define PCRE2_JIT_COMPILE(r,a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
r = G(pcre2_jit_compile_,BITONE)(G(a,BITONE),b); \
else \
r = G(pcre2_jit_compile_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_JIT_FREE_UNUSED_MEMORY(a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_jit_free_unused_memory_,BITONE)(G(a,BITONE)); \
else \
G(pcre2_jit_free_unused_memory_,BITTWO)(G(a,BITTWO))
#define PCRE2_JIT_MATCH(a,b,c,d,e,f,g,h) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_jit_match_,BITONE)(G(b,BITONE),(G(PCRE2_SPTR,BITONE))c,d,e,f, \
G(g,BITONE),h); \
else \
a = G(pcre2_jit_match_,BITTWO)(G(b,BITTWO),(G(PCRE2_SPTR,BITTWO))c,d,e,f, \
G(g,BITTWO),h)
#define PCRE2_JIT_STACK_CREATE(a,b,c,d) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = (PCRE2_JIT_STACK *)G(pcre2_jit_stack_create_,BITONE)(b,c,d); \
else \
a = (PCRE2_JIT_STACK *)G(pcre2_jit_stack_create_,BITTWO)(b,c,d); \
#define PCRE2_JIT_STACK_ASSIGN(a,b,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_jit_stack_assign_,BITONE)(G(a,BITONE),(G(pcre2_jit_callback_,BITONE))b,c); \
else \
G(pcre2_jit_stack_assign_,BITTWO)(G(a,BITTWO),(G(pcre2_jit_callback_,BITTWO))b,c);
#define PCRE2_JIT_STACK_FREE(a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_jit_stack_free_,BITONE)((G(pcre2_jit_stack_,BITONE) *)a); \
else \
G(pcre2_jit_stack_free_,BITTWO)((G(pcre2_jit_stack_,BITTWO) *)a);
#define PCRE2_MAKETABLES(a,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_maketables_,BITONE)(G(c,BITONE)); \
else \
a = G(pcre2_maketables_,BITTWO)(G(c,BITTWO))
#define PCRE2_MAKETABLES_FREE(c,a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_maketables_free_,BITONE)(G(c,BITONE),a); \
else \
G(pcre2_maketables_free_,BITTWO)(G(c,BITTWO),a)
#define PCRE2_MATCH(a,b,c,d,e,f,g,h) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_match_,BITONE)(G(b,BITONE),(G(PCRE2_SPTR,BITONE))c,d,e,f, \
G(g,BITONE),h); \
else \
a = G(pcre2_match_,BITTWO)(G(b,BITTWO),(G(PCRE2_SPTR,BITTWO))c,d,e,f, \
G(g,BITTWO),h)
#define PCRE2_MATCH_DATA_CREATE(a,b,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(a,BITONE) = G(pcre2_match_data_create_,BITONE)(b,G(c,BITONE)); \
else \
G(a,BITTWO) = G(pcre2_match_data_create_,BITTWO)(b,G(c,BITTWO))
#define PCRE2_MATCH_DATA_CREATE_FROM_PATTERN(a,b,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(a,BITONE) = G(pcre2_match_data_create_from_pattern_,BITONE)(G(b,BITONE),G(c,BITONE)); \
else \
G(a,BITTWO) = G(pcre2_match_data_create_from_pattern_,BITTWO)(G(b,BITTWO),G(c,BITTWO))
#define PCRE2_MATCH_DATA_FREE(a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_match_data_free_,BITONE)(G(a,BITONE)); \
else \
G(pcre2_match_data_free_,BITTWO)(G(a,BITTWO))
#define PCRE2_PATTERN_CONVERT(a,b,c,d,e,f,g) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_pattern_convert_,BITONE)(G(b,BITONE),c,d,(G(PCRE2_UCHAR,BITONE) **)e,f,G(g,BITONE)); \
else \
a = G(pcre2_pattern_convert_,BITTWO)(G(b,BITTWO),c,d,(G(PCRE2_UCHAR,BITTWO) **)e,f,G(g,BITTWO))
#define PCRE2_PATTERN_INFO(a,b,c,d) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_pattern_info_,BITONE)(G(b,BITONE),c,d); \
else \
a = G(pcre2_pattern_info_,BITTWO)(G(b,BITTWO),c,d)
#define PCRE2_PRINTINT(a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_printint_,BITONE)(G(compiled_code,BITONE),outfile,a); \
else \
G(pcre2_printint_,BITTWO)(G(compiled_code,BITTWO),outfile,a)
#define PCRE2_SERIALIZE_DECODE(r,a,b,c,d) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
r = G(pcre2_serialize_decode_,BITONE)((G(pcre2_code_,BITONE) **)a,b,c,G(d,BITONE)); \
else \
r = G(pcre2_serialize_decode_,BITTWO)((G(pcre2_code_,BITTWO) **)a,b,c,G(d,BITTWO))
#define PCRE2_SERIALIZE_ENCODE(r,a,b,c,d,e) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
r = G(pcre2_serialize_encode_,BITONE)((G(const pcre2_code_,BITONE) **)a,b,c,d,G(e,BITONE)); \
else \
r = G(pcre2_serialize_encode_,BITTWO)((G(const pcre2_code_,BITTWO) **)a,b,c,d,G(e,BITTWO))
#define PCRE2_SERIALIZE_FREE(a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_serialize_free_,BITONE)(a); \
else \
G(pcre2_serialize_free_,BITTWO)(a)
#define PCRE2_SERIALIZE_GET_NUMBER_OF_CODES(r,a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
r = G(pcre2_serialize_get_number_of_codes_,BITONE)(a); \
else \
r = G(pcre2_serialize_get_number_of_codes_,BITTWO)(a)
#define PCRE2_SET_CALLOUT(a,b,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_callout_,BITONE)(G(a,BITONE), \
(int (*)(G(pcre2_callout_block_,BITONE) *, void *))b,c); \
else \
G(pcre2_set_callout_,BITTWO)(G(a,BITTWO), \
(int (*)(G(pcre2_callout_block_,BITTWO) *, void *))b,c);
#define PCRE2_SET_CHARACTER_TABLES(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_character_tables_,BITONE)(G(a,BITONE),b); \
else \
G(pcre2_set_character_tables_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_COMPILE_RECURSION_GUARD(a,b,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_compile_recursion_guard_,BITONE)(G(a,BITONE),b,c); \
else \
G(pcre2_set_compile_recursion_guard_,BITTWO)(G(a,BITTWO),b,c)
#define PCRE2_SET_DEPTH_LIMIT(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_depth_limit_,BITONE)(G(a,BITONE),b); \
else \
G(pcre2_set_depth_limit_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_GLOB_ESCAPE(r,a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
r = G(pcre2_set_glob_escape_,BITONE)(G(a,BITONE),b); \
else \
r = G(pcre2_set_glob_escape_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_GLOB_SEPARATOR(r,a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
r = G(pcre2_set_glob_separator_,BITONE)(G(a,BITONE),b); \
else \
r = G(pcre2_set_glob_separator_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_HEAP_LIMIT(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_heap_limit_,BITONE)(G(a,BITONE),b); \
else \
G(pcre2_set_heap_limit_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_MATCH_LIMIT(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_match_limit_,BITONE)(G(a,BITONE),b); \
else \
G(pcre2_set_match_limit_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_MAX_PATTERN_COMPILED_LENGTH(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_max_pattern_compiled_length_,BITONE)(G(a,BITONE),b); \
else \
G(pcre2_set_max_pattern_compiled_length_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_MAX_PATTERN_LENGTH(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_max_pattern_length_,BITONE)(G(a,BITONE),b); \
else \
G(pcre2_set_max_pattern_length_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_MAX_VARLOOKBEHIND(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_max_varlookbehind_,BITONE)(G(a,BITONE),b); \
else \
G(pcre2_set_max_varlookbehind_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_OFFSET_LIMIT(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_offset_limit_,BITONE)(G(a,BITONE),b); \
else \
G(pcre2_set_offset_limit_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_PARENS_NEST_LIMIT(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_parens_nest_limit_,BITONE)(G(a,BITONE),b); \
else \
G(pcre2_set_parens_nest_limit_,BITTWO)(G(a,BITTWO),b)
#define PCRE2_SET_SUBSTITUTE_CALLOUT(a,b,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_substitute_callout_,BITONE)(G(a,BITONE), \
(int (*)(G(pcre2_substitute_callout_block_,BITONE) *, void *))b,c); \
else \
G(pcre2_set_substitute_callout_,BITTWO)(G(a,BITTWO), \
(int (*)(G(pcre2_substitute_callout_block_,BITTWO) *, void *))b,c)
#define PCRE2_SET_SUBSTITUTE_CASE_CALLOUT(a,b,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_substitute_case_callout_,BITONE)(G(a,BITONE),G(b,BITONE),c); \
else \
G(pcre2_set_substitute_case_callout_,BITTWO)(G(a,BITTWO),G(b,BITTWO),c)
#define PCRE2_SET_SUBSTITUTE_CASE_CALLOUT_NULL(a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_set_substitute_case_callout_,BITONE)(G(a,BITONE),NULL,NULL); \
else \
G(pcre2_set_substitute_case_callout_,BITTWO)(G(a,BITTWO),NULL,NULL)
#define PCRE2_SUBSTITUTE(a,b,c,d,e,f,g,h,i,j,k,l) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_substitute_,BITONE)(G(b,BITONE),(G(PCRE2_SPTR,BITONE))c,d,e,f, \
G(g,BITONE),h,(G(PCRE2_SPTR,BITONE))i,j, \
(G(PCRE2_UCHAR,BITONE) *)k,l); \
else \
a = G(pcre2_substitute_,BITTWO)(G(b,BITTWO),(G(PCRE2_SPTR,BITTWO))c,d,e,f, \
G(g,BITTWO),h,(G(PCRE2_SPTR,BITTWO))i,j, \
(G(PCRE2_UCHAR,BITTWO) *)k,l)
#define PCRE2_SUBSTRING_COPY_BYNAME(a,b,c,d,e) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_substring_copy_byname_,BITONE)(G(b,BITONE),G(c,BITONE),\
(G(PCRE2_UCHAR,BITONE) *)d,e); \
else \
a = G(pcre2_substring_copy_byname_,BITTWO)(G(b,BITTWO),G(c,BITTWO),\
(G(PCRE2_UCHAR,BITTWO) *)d,e)
#define PCRE2_SUBSTRING_COPY_BYNUMBER(a,b,c,d,e) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_substring_copy_bynumber_,BITONE)(G(b,BITONE),c,\
(G(PCRE2_UCHAR,BITONE) *)d,e); \
else \
a = G(pcre2_substring_copy_bynumber_,BITTWO)(G(b,BITTWO),c,\
(G(PCRE2_UCHAR,BITTWO) *)d,e)
#define PCRE2_SUBSTRING_FREE(a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_substring_free_,BITONE)((G(PCRE2_UCHAR,BITONE) *)a); \
else G(pcre2_substring_free_,BITTWO)((G(PCRE2_UCHAR,BITTWO) *)a)
#define PCRE2_SUBSTRING_GET_BYNAME(a,b,c,d,e) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_substring_get_byname_,BITONE)(G(b,BITONE),G(c,BITONE),\
(G(PCRE2_UCHAR,BITONE) **)d,e); \
else \
a = G(pcre2_substring_get_byname_,BITTWO)(G(b,BITTWO),G(c,BITTWO),\
(G(PCRE2_UCHAR,BITTWO) **)d,e)
#define PCRE2_SUBSTRING_GET_BYNUMBER(a,b,c,d,e) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_substring_get_bynumber_,BITONE)(G(b,BITONE),c,\
(G(PCRE2_UCHAR,BITONE) **)d,e); \
else \
a = G(pcre2_substring_get_bynumber_,BITTWO)(G(b,BITTWO),c,\
(G(PCRE2_UCHAR,BITTWO) **)d,e)
#define PCRE2_SUBSTRING_LENGTH_BYNAME(a,b,c,d) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_substring_length_byname_,BITONE)(G(b,BITONE),G(c,BITONE),d); \
else \
a = G(pcre2_substring_length_byname_,BITTWO)(G(b,BITTWO),G(c,BITTWO),d)
#define PCRE2_SUBSTRING_LENGTH_BYNUMBER(a,b,c,d) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_substring_length_bynumber_,BITONE)(G(b,BITONE),c,d); \
else \
a = G(pcre2_substring_length_bynumber_,BITTWO)(G(b,BITTWO),c,d)
#define PCRE2_SUBSTRING_LIST_GET(a,b,c,d) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_substring_list_get_,BITONE)(G(b,BITONE), \
(G(PCRE2_UCHAR,BITONE) ***)c,d); \
else \
a = G(pcre2_substring_list_get_,BITTWO)(G(b,BITTWO), \
(G(PCRE2_UCHAR,BITTWO) ***)c,d)
#define PCRE2_SUBSTRING_LIST_FREE(a) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(pcre2_substring_list_free_,BITONE)((G(PCRE2_UCHAR,BITONE) **)a); \
else \
G(pcre2_substring_list_free_,BITTWO)((G(PCRE2_UCHAR,BITTWO) **)a)
#define PCRE2_SUBSTRING_NUMBER_FROM_NAME(a,b,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
a = G(pcre2_substring_number_from_name_,BITONE)(G(b,BITONE),G(c,BITONE)); \
else \
a = G(pcre2_substring_number_from_name_,BITTWO)(G(b,BITTWO),G(c,BITTWO))
#define PTR(x) ( \
(test_mode == G(G(PCRE,BITONE),_MODE))? (void *)G(x,BITONE) : \
(void *)G(x,BITTWO))
#define SETFLD(x,y,z) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) G(x,BITONE)->y = z; \
else G(x,BITTWO)->y = z
#define SETFLDVEC(x,y,v,z) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) G(x,BITONE)->y[v] = z; \
else G(x,BITTWO)->y[v] = z
#define SETOP(x,y,z) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) G(x,BITONE) z y; \
else G(x,BITTWO) z y
#define SETCASTPTR(x,y) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(x,BITONE) = (G(G(uint,BITONE),_t) *)(y); \
else \
G(x,BITTWO) = (G(G(uint,BITTWO),_t) *)(y)
#define STRLEN(p) ((test_mode == G(G(PCRE,BITONE),_MODE))? \
G(strlen,BITONE)((G(PCRE2_SPTR,BITONE))p) : \
G(strlen,BITTWO)((G(PCRE2_SPTR,BITTWO))p))
#define SUB1(a,b) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(a,BITONE)(G(b,BITONE)); \
else \
G(a,BITTWO)(G(b,BITTWO))
#define SUB2(a,b,c) \
if (test_mode == G(G(PCRE,BITONE),_MODE)) \
G(a,BITONE))(G(b,BITONE),G(c,BITONE)); \
else \
G(a,BITTWO))(G(b,BITTWO),G(c,BITTWO))
#define TEST(x,r,y) ( \
(test_mode == G(G(PCRE,BITONE),_MODE) && G(x,BITONE) r (y)) || \
(test_mode == G(G(PCRE,BITTWO),_MODE) && G(x,BITTWO) r (y)))
#define TESTFLD(x,f,r,y) ( \
(test_mode == G(G(PCRE,BITONE),_MODE) && G(x,BITONE)->f r (y)) || \
(test_mode == G(G(PCRE,BITTWO),_MODE) && G(x,BITTWO)->f r (y)))
#endif /* Two out of three modes */
/* ----- End of cases where more than one mode is supported ----- */
/* ----- Only 8-bit mode is supported ----- */
#elif defined SUPPORT_PCRE2_8
#define CASTFLD(t,a,b) (t)(G(a,8)->b)
#define CASTVAR(t,x) (t)G(x,8)
#define CODE_UNIT(a,b) (uint32_t)(((PCRE2_SPTR8)(a))[b])
#define CONCTXCPY(a,b) memcpy(G(a,8),G(b,8),sizeof(pcre2_convert_context_8))
#define CONVERT_COPY(a,b,c) memcpy(G(a,8),(char *)b, c)
#define DATCTXCPY(a,b) memcpy(G(a,8),G(b,8),sizeof(pcre2_match_context_8))
#define FLD(a,b) G(a,8)->b
#define PATCTXCPY(a,b) memcpy(G(a,8),G(b,8),sizeof(pcre2_compile_context_8))
#define PCHARS(lv, p, offset, len, utf, f) \
lv = pchars8((PCRE2_SPTR8)(p)+offset, len, utf, f)
#define PCHARSV(p, offset, len, utf, f) \
(void)pchars8((PCRE2_SPTR8)(p)+offset, len, utf, f)
#define PCRE2_CALLOUT_ENUMERATE(a,b,c) \
a = pcre2_callout_enumerate_8(compiled_code8, \
(int (*)(struct pcre2_callout_enumerate_block_8 *, void *))b,c)
#define PCRE2_CODE_COPY_FROM_VOID(a,b) G(a,8) = pcre2_code_copy_8(b)
#define PCRE2_CODE_COPY_TO_VOID(a,b) a = (void *)pcre2_code_copy_8(G(b,8))
#define PCRE2_CODE_COPY_WITH_TABLES_TO_VOID(a,b) a = (void *)pcre2_code_copy_with_tables_8(G(b,8))
#define PCRE2_COMPILE(a,b,c,d,e,f,g) G(a,8) = pcre2_compile_8(b,c,d,e,f,g)
#define PCRE2_CONVERTED_PATTERN_FREE(a) \
pcre2_converted_pattern_free_8((PCRE2_UCHAR8 *)a)
#define PCRE2_DFA_MATCH(a,b,c,d,e,f,g,h,i,j) \
a = pcre2_dfa_match_8(G(b,8),(PCRE2_SPTR8)c,d,e,f,G(g,8),h,i,j)
#define PCRE2_GET_ERROR_MESSAGE(r,a,b) \
r = pcre2_get_error_message_8(a,G(b,8),G(G(b,8),_size))
#define PCRE2_GET_MATCH_DATA_HEAPFRAMES_SIZE(r,a) \
r = pcre2_get_match_data_heapframes_size_8(G(a,8))
#define PCRE2_GET_OVECTOR_COUNT(a,b) a = pcre2_get_ovector_count_8(G(b,8))
#define PCRE2_GET_STARTCHAR(a,b) a = pcre2_get_startchar_8(G(b,8))
#define PCRE2_JIT_COMPILE(r,a,b) r = pcre2_jit_compile_8(G(a,8),b)
#define PCRE2_JIT_FREE_UNUSED_MEMORY(a) pcre2_jit_free_unused_memory_8(G(a,8))
#define PCRE2_JIT_MATCH(a,b,c,d,e,f,g,h) \
a = pcre2_jit_match_8(G(b,8),(PCRE2_SPTR8)c,d,e,f,G(g,8),h)
#define PCRE2_JIT_STACK_CREATE(a,b,c,d) \
a = (PCRE2_JIT_STACK *)pcre2_jit_stack_create_8(b,c,d);
#define PCRE2_JIT_STACK_ASSIGN(a,b,c) \
pcre2_jit_stack_assign_8(G(a,8),(pcre2_jit_callback_8)b,c);
#define PCRE2_JIT_STACK_FREE(a) pcre2_jit_stack_free_8((pcre2_jit_stack_8 *)a);
#define PCRE2_MAKETABLES(a,c) a = pcre2_maketables_8(G(c,8))
#define PCRE2_MAKETABLES_FREE(c,a) pcre2_maketables_free_8(G(c,8),a)
#define PCRE2_MATCH(a,b,c,d,e,f,g,h) \
a = pcre2_match_8(G(b,8),(PCRE2_SPTR8)c,d,e,f,G(g,8),h)
#define PCRE2_MATCH_DATA_CREATE(a,b,c) G(a,8) = pcre2_match_data_create_8(b,G(c,8))
#define PCRE2_MATCH_DATA_CREATE_FROM_PATTERN(a,b,c) \
G(a,8) = pcre2_match_data_create_from_pattern_8(G(b,8),G(c,8))
#define PCRE2_MATCH_DATA_FREE(a) pcre2_match_data_free_8(G(a,8))
#define PCRE2_PATTERN_CONVERT(a,b,c,d,e,f,g) a = pcre2_pattern_convert_8(G(b,8),c,d,(PCRE2_UCHAR8 **)e,f,G(g,8))
#define PCRE2_PATTERN_INFO(a,b,c,d) a = pcre2_pattern_info_8(G(b,8),c,d)
#define PCRE2_PRINTINT(a) pcre2_printint_8(compiled_code8,outfile,a)
#define PCRE2_SERIALIZE_DECODE(r,a,b,c,d) \
r = pcre2_serialize_decode_8((pcre2_code_8 **)a,b,c,G(d,8))
#define PCRE2_SERIALIZE_ENCODE(r,a,b,c,d,e) \
r = pcre2_serialize_encode_8((const pcre2_code_8 **)a,b,c,d,G(e,8))
#define PCRE2_SERIALIZE_FREE(a) pcre2_serialize_free_8(a)
#define PCRE2_SERIALIZE_GET_NUMBER_OF_CODES(r,a) \
r = pcre2_serialize_get_number_of_codes_8(a)
#define PCRE2_SET_CALLOUT(a,b,c) \
pcre2_set_callout_8(G(a,8),(int (*)(pcre2_callout_block_8 *, void *))b,c)
#define PCRE2_SET_CHARACTER_TABLES(a,b) pcre2_set_character_tables_8(G(a,8),b)
#define PCRE2_SET_COMPILE_RECURSION_GUARD(a,b,c) \
pcre2_set_compile_recursion_guard_8(G(a,8),b,c)
#define PCRE2_SET_DEPTH_LIMIT(a,b) pcre2_set_depth_limit_8(G(a,8),b)
#define PCRE2_SET_GLOB_ESCAPE(r,a,b) r = pcre2_set_glob_escape_8(G(a,8),b)
#define PCRE2_SET_GLOB_SEPARATOR(r,a,b) r = pcre2_set_glob_separator_8(G(a,8),b)
#define PCRE2_SET_HEAP_LIMIT(a,b) pcre2_set_heap_limit_8(G(a,8),b)
#define PCRE2_SET_MATCH_LIMIT(a,b) pcre2_set_match_limit_8(G(a,8),b)
#define PCRE2_SET_MAX_PATTERN_COMPILED_LENGTH(a,b) pcre2_set_max_pattern_compiled_length_8(G(a,8),b)
#define PCRE2_SET_MAX_PATTERN_LENGTH(a,b) pcre2_set_max_pattern_length_8(G(a,8),b)
#define PCRE2_SET_MAX_VARLOOKBEHIND(a,b) pcre2_set_max_varlookbehind_8(G(a,8),b)
#define PCRE2_SET_OFFSET_LIMIT(a,b) pcre2_set_offset_limit_8(G(a,8),b)
#define PCRE2_SET_PARENS_NEST_LIMIT(a,b) pcre2_set_parens_nest_limit_8(G(a,8),b)
#define PCRE2_SET_SUBSTITUTE_CALLOUT(a,b,c) \
pcre2_set_substitute_callout_8(G(a,8), \
(int (*)(pcre2_substitute_callout_block_8 *, void *))b,c)
#define PCRE2_SET_SUBSTITUTE_CASE_CALLOUT(a,b,c) \
pcre2_set_substitute_case_callout_8(G(a,8),G(b,8),c)
#define PCRE2_SET_SUBSTITUTE_CASE_CALLOUT_NULL(a) \
pcre2_set_substitute_case_callout_8(G(a,8),NULL,NULL)
#define PCRE2_SUBSTITUTE(a,b,c,d,e,f,g,h,i,j,k,l) \
a = pcre2_substitute_8(G(b,8),(PCRE2_SPTR8)c,d,e,f,G(g,8),h, \
(PCRE2_SPTR8)i,j,(PCRE2_UCHAR8 *)k,l)
#define PCRE2_SUBSTRING_COPY_BYNAME(a,b,c,d,e) \
a = pcre2_substring_copy_byname_8(G(b,8),G(c,8),(PCRE2_UCHAR8 *)d,e)
#define PCRE2_SUBSTRING_COPY_BYNUMBER(a,b,c,d,e) \
a = pcre2_substring_copy_bynumber_8(G(b,8),c,(PCRE2_UCHAR8 *)d,e)
#define PCRE2_SUBSTRING_FREE(a) pcre2_substring_free_8((PCRE2_UCHAR8 *)a)
#define PCRE2_SUBSTRING_GET_BYNAME(a,b,c,d,e) \
a = pcre2_substring_get_byname_8(G(b,8),G(c,8),(PCRE2_UCHAR8 **)d,e)
#define PCRE2_SUBSTRING_GET_BYNUMBER(a,b,c,d,e) \
a = pcre2_substring_get_bynumber_8(G(b,8),c,(PCRE2_UCHAR8 **)d,e)
#define PCRE2_SUBSTRING_LENGTH_BYNAME(a,b,c,d) \
a = pcre2_substring_length_byname_8(G(b,8),G(c,8),d)
#define PCRE2_SUBSTRING_LENGTH_BYNUMBER(a,b,c,d) \
a = pcre2_substring_length_bynumber_8(G(b,8),c,d)
#define PCRE2_SUBSTRING_LIST_GET(a,b,c,d) \
a = pcre2_substring_list_get_8(G(b,8),(PCRE2_UCHAR8 ***)c,d)
#define PCRE2_SUBSTRING_LIST_FREE(a) \
pcre2_substring_list_free_8((PCRE2_UCHAR8 **)a)
#define PCRE2_SUBSTRING_NUMBER_FROM_NAME(a,b,c) \
a = pcre2_substring_number_from_name_8(G(b,8),G(c,8));
#define PTR(x) (void *)G(x,8)
#define SETFLD(x,y,z) G(x,8)->y = z
#define SETFLDVEC(x,y,v,z) G(x,8)->y[v] = z
#define SETOP(x,y,z) G(x,8) z y
#define SETCASTPTR(x,y) G(x,8) = (uint8_t *)(y)
#define STRLEN(p) (int)strlen((char *)p)
#define SUB1(a,b) G(a,8)(G(b,8))
#define SUB2(a,b,c) G(a,8)(G(b,8),G(c,8))
#define TEST(x,r,y) (G(x,8) r (y))
#define TESTFLD(x,f,r,y) (G(x,8)->f r (y))
/* ----- Only 16-bit mode is supported ----- */
#elif defined SUPPORT_PCRE2_16
#define CASTFLD(t,a,b) (t)(G(a,16)->b)
#define CASTVAR(t,x) (t)G(x,16)
#define CODE_UNIT(a,b) (uint32_t)(((PCRE2_SPTR16)(a))[b])
#define CONCTXCPY(a,b) memcpy(G(a,16),G(b,16),sizeof(pcre2_convert_context_16))
#define CONVERT_COPY(a,b,c) memcpy(G(a,16),(char *)b, (c)*2)
#define DATCTXCPY(a,b) memcpy(G(a,16),G(b,16),sizeof(pcre2_match_context_16))
#define FLD(a,b) G(a,16)->b
#define PATCTXCPY(a,b) memcpy(G(a,16),G(b,16),sizeof(pcre2_compile_context_16))
#define PCHARS(lv, p, offset, len, utf, f) \
lv = pchars16((PCRE2_SPTR16)(p)+offset, len, utf, f)
#define PCHARSV(p, offset, len, utf, f) \
(void)pchars16((PCRE2_SPTR16)(p)+offset, len, utf, f)
#define PCRE2_CALLOUT_ENUMERATE(a,b,c) \
a = pcre2_callout_enumerate_16(compiled_code16, \
(int (*)(struct pcre2_callout_enumerate_block_16 *, void *))b,c)
#define PCRE2_CODE_COPY_FROM_VOID(a,b) G(a,16) = pcre2_code_copy_16(b)
#define PCRE2_CODE_COPY_TO_VOID(a,b) a = (void *)pcre2_code_copy_16(G(b,16))
#define PCRE2_CODE_COPY_WITH_TABLES_TO_VOID(a,b) a = (void *)pcre2_code_copy_with_tables_16(G(b,16))
#define PCRE2_COMPILE(a,b,c,d,e,f,g) G(a,16) = pcre2_compile_16(b,c,d,e,f,g)
#define PCRE2_CONVERTED_PATTERN_FREE(a) \
pcre2_converted_pattern_free_16((PCRE2_UCHAR16 *)a)
#define PCRE2_DFA_MATCH(a,b,c,d,e,f,g,h,i,j) \
a = pcre2_dfa_match_16(G(b,16),(PCRE2_SPTR16)c,d,e,f,G(g,16),h,i,j)
#define PCRE2_GET_ERROR_MESSAGE(r,a,b) \
r = pcre2_get_error_message_16(a,G(b,16),G(G(b,16),_size/2))
#define PCRE2_GET_OVECTOR_COUNT(a,b) a = pcre2_get_ovector_count_16(G(b,16))
#define PCRE2_GET_MATCH_DATA_HEAPFRAMES_SIZE(r,a) \
r = pcre2_get_match_data_heapframes_size_16(G(a,16))
#define PCRE2_GET_STARTCHAR(a,b) a = pcre2_get_startchar_16(G(b,16))
#define PCRE2_JIT_COMPILE(r,a,b) r = pcre2_jit_compile_16(G(a,16),b)
#define PCRE2_JIT_FREE_UNUSED_MEMORY(a) pcre2_jit_free_unused_memory_16(G(a,16))
#define PCRE2_JIT_MATCH(a,b,c,d,e,f,g,h) \
a = pcre2_jit_match_16(G(b,16),(PCRE2_SPTR16)c,d,e,f,G(g,16),h)
#define PCRE2_JIT_STACK_CREATE(a,b,c,d) \
a = (PCRE2_JIT_STACK *)pcre2_jit_stack_create_16(b,c,d);
#define PCRE2_JIT_STACK_ASSIGN(a,b,c) \
pcre2_jit_stack_assign_16(G(a,16),(pcre2_jit_callback_16)b,c);
#define PCRE2_JIT_STACK_FREE(a) pcre2_jit_stack_free_16((pcre2_jit_stack_16 *)a);
#define PCRE2_MAKETABLES(a,c) a = pcre2_maketables_16(G(c,16))
#define PCRE2_MAKETABLES_FREE(c,a) pcre2_maketables_free_16(G(c,16),a)
#define PCRE2_MATCH(a,b,c,d,e,f,g,h) \
a = pcre2_match_16(G(b,16),(PCRE2_SPTR16)c,d,e,f,G(g,16),h)
#define PCRE2_MATCH_DATA_CREATE(a,b,c) G(a,16) = pcre2_match_data_create_16(b,G(c,16))
#define PCRE2_MATCH_DATA_CREATE_FROM_PATTERN(a,b,c) \
G(a,16) = pcre2_match_data_create_from_pattern_16(G(b,16),G(c,16))
#define PCRE2_MATCH_DATA_FREE(a) pcre2_match_data_free_16(G(a,16))
#define PCRE2_PATTERN_CONVERT(a,b,c,d,e,f,g) a = pcre2_pattern_convert_16(G(b,16),c,d,(PCRE2_UCHAR16 **)e,f,G(g,16))
#define PCRE2_PATTERN_INFO(a,b,c,d) a = pcre2_pattern_info_16(G(b,16),c,d)
#define PCRE2_PRINTINT(a) pcre2_printint_16(compiled_code16,outfile,a)
#define PCRE2_SERIALIZE_DECODE(r,a,b,c,d) \
r = pcre2_serialize_decode_16((pcre2_code_16 **)a,b,c,G(d,16))
#define PCRE2_SERIALIZE_ENCODE(r,a,b,c,d,e) \
r = pcre2_serialize_encode_16((const pcre2_code_16 **)a,b,c,d,G(e,16))
#define PCRE2_SERIALIZE_FREE(a) pcre2_serialize_free_16(a)
#define PCRE2_SERIALIZE_GET_NUMBER_OF_CODES(r,a) \
r = pcre2_serialize_get_number_of_codes_16(a)
#define PCRE2_SET_CALLOUT(a,b,c) \
pcre2_set_callout_16(G(a,16),(int (*)(pcre2_callout_block_16 *, void *))b,c);
#define PCRE2_SET_CHARACTER_TABLES(a,b) pcre2_set_character_tables_16(G(a,16),b)
#define PCRE2_SET_COMPILE_RECURSION_GUARD(a,b,c) \
pcre2_set_compile_recursion_guard_16(G(a,16),b,c)
#define PCRE2_SET_DEPTH_LIMIT(a,b) pcre2_set_depth_limit_16(G(a,16),b)
#define PCRE2_SET_GLOB_ESCAPE(r,a,b) r = pcre2_set_glob_escape_16(G(a,16),b)
#define PCRE2_SET_GLOB_SEPARATOR(r,a,b) r = pcre2_set_glob_separator_16(G(a,16),b)
#define PCRE2_SET_HEAP_LIMIT(a,b) pcre2_set_heap_limit_16(G(a,16),b)
#define PCRE2_SET_MATCH_LIMIT(a,b) pcre2_set_match_limit_16(G(a,16),b)
#define PCRE2_SET_MAX_VARLOOKBEHIND(a,b) pcre2_set_max_varlookbehind_16(G(a,16),b)
#define PCRE2_SET_OFFSET_LIMIT(a,b) pcre2_set_offset_limit_16(G(a,16),b)
#define PCRE2_SET_PARENS_NEST_LIMIT(a,b) pcre2_set_parens_nest_limit_16(G(a,16),b)
#define PCRE2_SET_SUBSTITUTE_CALLOUT(a,b,c) \
pcre2_set_substitute_callout_16(G(a,16), \
(int (*)(pcre2_substitute_callout_block_16 *, void *))b,c)
#define PCRE2_SET_SUBSTITUTE_CASE_CALLOUT(a,b,c) \
pcre2_set_substitute_case_callout_16(G(a,16),G(b,16),c)
#define PCRE2_SET_SUBSTITUTE_CASE_CALLOUT_NULL(a) \
pcre2_set_substitute_case_callout_16(G(a,16),NULL,NULL)
#define PCRE2_SUBSTITUTE(a,b,c,d,e,f,g,h,i,j,k,l) \
a = pcre2_substitute_16(G(b,16),(PCRE2_SPTR16)c,d,e,f,G(g,16),h, \
(PCRE2_SPTR16)i,j,(PCRE2_UCHAR16 *)k,l)
#define PCRE2_SUBSTRING_COPY_BYNAME(a,b,c,d,e) \
a = pcre2_substring_copy_byname_16(G(b,16),G(c,16),(PCRE2_UCHAR16 *)d,e)
#define PCRE2_SUBSTRING_COPY_BYNUMBER(a,b,c,d,e) \
a = pcre2_substring_copy_bynumber_16(G(b,16),c,(PCRE2_UCHAR16 *)d,e)
#define PCRE2_SUBSTRING_FREE(a) pcre2_substring_free_16((PCRE2_UCHAR16 *)a)
#define PCRE2_SUBSTRING_GET_BYNAME(a,b,c,d,e) \
a = pcre2_substring_get_byname_16(G(b,16),G(c,16),(PCRE2_UCHAR16 **)d,e)
#define PCRE2_SUBSTRING_GET_BYNUMBER(a,b,c,d,e) \
a = pcre2_substring_get_bynumber_16(G(b,16),c,(PCRE2_UCHAR16 **)d,e)
#define PCRE2_SUBSTRING_LENGTH_BYNAME(a,b,c,d) \
a = pcre2_substring_length_byname_16(G(b,16),G(c,16),d)
#define PCRE2_SUBSTRING_LENGTH_BYNUMBER(a,b,c,d) \
a = pcre2_substring_length_bynumber_16(G(b,16),c,d)
#define PCRE2_SUBSTRING_LIST_GET(a,b,c,d) \
a = pcre2_substring_list_get_16(G(b,16),(PCRE2_UCHAR16 ***)c,d)
#define PCRE2_SUBSTRING_LIST_FREE(a) \
pcre2_substring_list_free_16((PCRE2_UCHAR16 **)a)
#define PCRE2_SUBSTRING_NUMBER_FROM_NAME(a,b,c) \
a = pcre2_substring_number_from_name_16(G(b,16),G(c,16));
#define PTR(x) (void *)G(x,16)
#define SETFLD(x,y,z) G(x,16)->y = z
#define SETFLDVEC(x,y,v,z) G(x,16)->y[v] = z
#define SETOP(x,y,z) G(x,16) z y
#define SETCASTPTR(x,y) G(x,16) = (uint16_t *)(y)
#define STRLEN(p) (int)strlen16((PCRE2_SPTR16)p)
#define SUB1(a,b) G(a,16)(G(b,16))
#define SUB2(a,b,c) G(a,16)(G(b,16),G(c,16))
#define TEST(x,r,y) (G(x,16) r (y))
#define TESTFLD(x,f,r,y) (G(x,16)->f r (y))
/* ----- Only 32-bit mode is supported ----- */
#elif defined SUPPORT_PCRE2_32
#define CASTFLD(t,a,b) (t)(G(a,32)->b)
#define CASTVAR(t,x) (t)G(x,32)
#define CODE_UNIT(a,b) (uint32_t)(((PCRE2_SPTR32)(a))[b])
#define CONCTXCPY(a,b) memcpy(G(a,32),G(b,32),sizeof(pcre2_convert_context_32))
#define CONVERT_COPY(a,b,c) memcpy(G(a,32),(char *)b, (c)*4)
#define DATCTXCPY(a,b) memcpy(G(a,32),G(b,32),sizeof(pcre2_match_context_32))
#define FLD(a,b) G(a,32)->b
#define PATCTXCPY(a,b) memcpy(G(a,32),G(b,32),sizeof(pcre2_compile_context_32))
#define PCHARS(lv, p, offset, len, utf, f) \
lv = pchars32((PCRE2_SPTR32)(p)+offset, len, utf, f)
#define PCHARSV(p, offset, len, utf, f) \
(void)pchars32((PCRE2_SPTR32)(p)+offset, len, utf, f)
#define PCRE2_CALLOUT_ENUMERATE(a,b,c) \
a = pcre2_callout_enumerate_32(compiled_code32, \
(int (*)(struct pcre2_callout_enumerate_block_32 *, void *))b,c)
#define PCRE2_CODE_COPY_FROM_VOID(a,b) G(a,32) = pcre2_code_copy_32(b)
#define PCRE2_CODE_COPY_TO_VOID(a,b) a = (void *)pcre2_code_copy_32(G(b,32))
#define PCRE2_CODE_COPY_WITH_TABLES_TO_VOID(a,b) a = (void *)pcre2_code_copy_with_tables_32(G(b,32))
#define PCRE2_COMPILE(a,b,c,d,e,f,g) G(a,32) = pcre2_compile_32(b,c,d,e,f,g)
#define PCRE2_CONVERTED_PATTERN_FREE(a) \
pcre2_converted_pattern_free_32((PCRE2_UCHAR32 *)a)
#define PCRE2_DFA_MATCH(a,b,c,d,e,f,g,h,i,j) \
a = pcre2_dfa_match_32(G(b,32),(PCRE2_SPTR32)c,d,e,f,G(g,32),h,i,j)
#define PCRE2_GET_ERROR_MESSAGE(r,a,b) \
r = pcre2_get_error_message_32(a,G(b,32),G(G(b,32),_size/4))
#define PCRE2_GET_OVECTOR_COUNT(a,b) a = pcre2_get_ovector_count_32(G(b,32))
#define PCRE2_GET_MATCH_DATA_HEAPFRAMES_SIZE(r,a) \
r = pcre2_get_match_data_heapframes_size_32(G(a,32))
#define PCRE2_GET_STARTCHAR(a,b) a = pcre2_get_startchar_32(G(b,32))
#define PCRE2_JIT_COMPILE(r,a,b) r = pcre2_jit_compile_32(G(a,32),b)
#define PCRE2_JIT_FREE_UNUSED_MEMORY(a) pcre2_jit_free_unused_memory_32(G(a,32))
#define PCRE2_JIT_MATCH(a,b,c,d,e,f,g,h) \
a = pcre2_jit_match_32(G(b,32),(PCRE2_SPTR32)c,d,e,f,G(g,32),h)
#define PCRE2_JIT_STACK_CREATE(a,b,c,d) \
a = (PCRE2_JIT_STACK *)pcre2_jit_stack_create_32(b,c,d);
#define PCRE2_JIT_STACK_ASSIGN(a,b,c) \
pcre2_jit_stack_assign_32(G(a,32),(pcre2_jit_callback_32)b,c);
#define PCRE2_JIT_STACK_FREE(a) pcre2_jit_stack_free_32((pcre2_jit_stack_32 *)a);
#define PCRE2_MAKETABLES(a,c) a = pcre2_maketables_32(G(c,32))
#define PCRE2_MAKETABLES_FREE(c,a) pcre2_maketables_free_32(G(c,32),a)
#define PCRE2_MATCH(a,b,c,d,e,f,g,h) \
a = pcre2_match_32(G(b,32),(PCRE2_SPTR32)c,d,e,f,G(g,32),h)
#define PCRE2_MATCH_DATA_CREATE(a,b,c) G(a,32) = pcre2_match_data_create_32(b,G(c,32))
#define PCRE2_MATCH_DATA_CREATE_FROM_PATTERN(a,b,c) \
G(a,32) = pcre2_match_data_create_from_pattern_32(G(b,32),G(c,32))
#define PCRE2_MATCH_DATA_FREE(a) pcre2_match_data_free_32(G(a,32))
#define PCRE2_PATTERN_CONVERT(a,b,c,d,e,f,g) a = pcre2_pattern_convert_32(G(b,32),c,d,(PCRE2_UCHAR32 **)e,f,G(g,32))
#define PCRE2_PATTERN_INFO(a,b,c,d) a = pcre2_pattern_info_32(G(b,32),c,d)
#define PCRE2_PRINTINT(a) pcre2_printint_32(compiled_code32,outfile,a)
#define PCRE2_SERIALIZE_DECODE(r,a,b,c,d) \
r = pcre2_serialize_decode_32((pcre2_code_32 **)a,b,c,G(d,32))
#define PCRE2_SERIALIZE_ENCODE(r,a,b,c,d,e) \
r = pcre2_serialize_encode_32((const pcre2_code_32 **)a,b,c,d,G(e,32))
#define PCRE2_SERIALIZE_FREE(a) pcre2_serialize_free_32(a)
#define PCRE2_SERIALIZE_GET_NUMBER_OF_CODES(r,a) \
r = pcre2_serialize_get_number_of_codes_32(a)
#define PCRE2_SET_CALLOUT(a,b,c) \
pcre2_set_callout_32(G(a,32),(int (*)(pcre2_callout_block_32 *, void *))b,c)
#define PCRE2_SET_CHARACTER_TABLES(a,b) pcre2_set_character_tables_32(G(a,32),b)
#define PCRE2_SET_COMPILE_RECURSION_GUARD(a,b,c) \
pcre2_set_compile_recursion_guard_32(G(a,32),b,c)
#define PCRE2_SET_DEPTH_LIMIT(a,b) pcre2_set_depth_limit_32(G(a,32),b)
#define PCRE2_SET_GLOB_ESCAPE(r,a,b) r = pcre2_set_glob_escape_32(G(a,32),b)
#define PCRE2_SET_GLOB_SEPARATOR(r,a,b) r = pcre2_set_glob_separator_32(G(a,32),b)
#define PCRE2_SET_HEAP_LIMIT(a,b) pcre2_set_heap_limit_32(G(a,32),b)
#define PCRE2_SET_MATCH_LIMIT(a,b) pcre2_set_match_limit_32(G(a,32),b)
#define PCRE2_SET_MAX_VARLOOKBEHIND(a,b) pcre2_set_max_varlookbehind_32(G(a,32),b)
#define PCRE2_SET_OFFSET_LIMIT(a,b) pcre2_set_offset_limit_32(G(a,32),b)
#define PCRE2_SET_PARENS_NEST_LIMIT(a,b) pcre2_set_parens_nest_limit_32(G(a,32),b)
#define PCRE2_SET_SUBSTITUTE_CALLOUT(a,b,c) \
pcre2_set_substitute_callout_32(G(a,32), \
(int (*)(pcre2_substitute_callout_block_32 *, void *))b,c)
#define PCRE2_SET_SUBSTITUTE_CASE_CALLOUT(a,b,c) \
pcre2_set_substitute_case_callout_32(G(a,32),G(b,32),c)
#define PCRE2_SET_SUBSTITUTE_CASE_CALLOUT_NULL(a) \
pcre2_set_substitute_case_callout_32(G(a,32),NULL,NULL)
#define PCRE2_SUBSTITUTE(a,b,c,d,e,f,g,h,i,j,k,l) \
a = pcre2_substitute_32(G(b,32),(PCRE2_SPTR32)c,d,e,f,G(g,32),h, \
(PCRE2_SPTR32)i,j,(PCRE2_UCHAR32 *)k,l)
#define PCRE2_SUBSTRING_COPY_BYNAME(a,b,c,d,e) \
a = pcre2_substring_copy_byname_32(G(b,32),G(c,32),(PCRE2_UCHAR32 *)d,e)
#define PCRE2_SUBSTRING_COPY_BYNUMBER(a,b,c,d,e) \
a = pcre2_substring_copy_bynumber_32(G(b,32),c,(PCRE2_UCHAR32 *)d,e);
#define PCRE2_SUBSTRING_FREE(a) pcre2_substring_free_32((PCRE2_UCHAR32 *)a)
#define PCRE2_SUBSTRING_GET_BYNAME(a,b,c,d,e) \
a = pcre2_substring_get_byname_32(G(b,32),G(c,32),(PCRE2_UCHAR32 **)d,e)
#define PCRE2_SUBSTRING_GET_BYNUMBER(a,b,c,d,e) \
a = pcre2_substring_get_bynumber_32(G(b,32),c,(PCRE2_UCHAR32 **)d,e)
#define PCRE2_SUBSTRING_LENGTH_BYNAME(a,b,c,d) \
a = pcre2_substring_length_byname_32(G(b,32),G(c,32),d)
#define PCRE2_SUBSTRING_LENGTH_BYNUMBER(a,b,c,d) \
a = pcre2_substring_length_bynumber_32(G(b,32),c,d)
#define PCRE2_SUBSTRING_LIST_GET(a,b,c,d) \
a = pcre2_substring_list_get_32(G(b,32),(PCRE2_UCHAR32 ***)c,d)
#define PCRE2_SUBSTRING_LIST_FREE(a) \
pcre2_substring_list_free_32((PCRE2_UCHAR32 **)a)
#define PCRE2_SUBSTRING_NUMBER_FROM_NAME(a,b,c) \
a = pcre2_substring_number_from_name_32(G(b,32),G(c,32));
#define PTR(x) (void *)G(x,32)
#define SETFLD(x,y,z) G(x,32)->y = z
#define SETFLDVEC(x,y,v,z) G(x,32)->y[v] = z
#define SETOP(x,y,z) G(x,32) z y
#define SETCASTPTR(x,y) G(x,32) = (uint32_t *)(y)
#define STRLEN(p) (int)strlen32((PCRE2_SPTR32)p)
#define SUB1(a,b) G(a,32)(G(b,32))
#define SUB2(a,b,c) G(a,32)(G(b,32),G(c,32))
#define TEST(x,r,y) (G(x,32) r (y))
#define TESTFLD(x,f,r,y) (G(x,32)->f r (y))
#endif
/* ----- End of mode-specific function call macros ----- */
/*************************************************
* Alternate character tables *
*************************************************/
/* By default, the "tables" pointer in the compile context when calling
pcre2_compile() is not set (= NULL), thereby using the default tables of the
library. However, the tables modifier can be used to select alternate sets of
tables, for different kinds of testing. Note that the locale modifier also
adjusts the tables. */
/* This is the set of tables distributed as default with PCRE2. It recognizes
only ASCII characters. */
static const uint8_t tables1[] = {
/* This table is a lower casing table. */
0, 1, 2, 3, 4, 5, 6, 7,
8, 9, 10, 11, 12, 13, 14, 15,
16, 17, 18, 19, 20, 21, 22, 23,
24, 25, 26, 27, 28, 29, 30, 31,
32, 33, 34, 35, 36, 37, 38, 39,
40, 41, 42, 43, 44, 45, 46, 47,
48, 49, 50, 51, 52, 53, 54, 55,
56, 57, 58, 59, 60, 61, 62, 63,
64, 97, 98, 99,100,101,102,103,
104,105,106,107,108,109,110,111,
112,113,114,115,116,117,118,119,
120,121,122, 91, 92, 93, 94, 95,
96, 97, 98, 99,100,101,102,103,
104,105,106,107,108,109,110,111,
112,113,114,115,116,117,118,119,
120,121,122,123,124,125,126,127,
128,129,130,131,132,133,134,135,
136,137,138,139,140,141,142,143,
144,145,146,147,148,149,150,151,
152,153,154,155,156,157,158,159,
160,161,162,163,164,165,166,167,
168,169,170,171,172,173,174,175,
176,177,178,179,180,181,182,183,
184,185,186,187,188,189,190,191,
192,193,194,195,196,197,198,199,
200,201,202,203,204,205,206,207,
208,209,210,211,212,213,214,215,
216,217,218,219,220,221,222,223,
224,225,226,227,228,229,230,231,
232,233,234,235,236,237,238,239,
240,241,242,243,244,245,246,247,
248,249,250,251,252,253,254,255,
/* This table is a case flipping table. */
0, 1, 2, 3, 4, 5, 6, 7,
8, 9, 10, 11, 12, 13, 14, 15,
16, 17, 18, 19, 20, 21, 22, 23,
24, 25, 26, 27, 28, 29, 30, 31,
32, 33, 34, 35, 36, 37, 38, 39,
40, 41, 42, 43, 44, 45, 46, 47,
48, 49, 50, 51, 52, 53, 54, 55,
56, 57, 58, 59, 60, 61, 62, 63,
64, 97, 98, 99,100,101,102,103,
104,105,106,107,108,109,110,111,
112,113,114,115,116,117,118,119,
120,121,122, 91, 92, 93, 94, 95,
96, 65, 66, 67, 68, 69, 70, 71,
72, 73, 74, 75, 76, 77, 78, 79,
80, 81, 82, 83, 84, 85, 86, 87,
88, 89, 90,123,124,125,126,127,
128,129,130,131,132,133,134,135,
136,137,138,139,140,141,142,143,
144,145,146,147,148,149,150,151,
152,153,154,155,156,157,158,159,
160,161,162,163,164,165,166,167,
168,169,170,171,172,173,174,175,
176,177,178,179,180,181,182,183,
184,185,186,187,188,189,190,191,
192,193,194,195,196,197,198,199,
200,201,202,203,204,205,206,207,
208,209,210,211,212,213,214,215,
216,217,218,219,220,221,222,223,
224,225,226,227,228,229,230,231,
232,233,234,235,236,237,238,239,
240,241,242,243,244,245,246,247,
248,249,250,251,252,253,254,255,
/* This table contains bit maps for various character classes. Each map is 32
bytes long and the bits run from the least significant end of each byte. The
classes that have their own maps are: space, xdigit, digit, upper, lower, word,
graph, print, punct, and cntrl. Other classes are built from combinations. */
0x00,0x3e,0x00,0x00,0x01,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0xff,0x03,
0x7e,0x00,0x00,0x00,0x7e,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0xff,0x03,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0xfe,0xff,0xff,0x07,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0xfe,0xff,0xff,0x07,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0xff,0x03,
0xfe,0xff,0xff,0x87,0xfe,0xff,0xff,0x07,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0xfe,0xff,0xff,0xff,
0xff,0xff,0xff,0xff,0xff,0xff,0xff,0x7f,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0xff,0xff,0xff,0xff,
0xff,0xff,0xff,0xff,0xff,0xff,0xff,0x7f,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0xfe,0xff,0x00,0xfc,
0x01,0x00,0x00,0xf8,0x01,0x00,0x00,0x78,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0xff,0xff,0xff,0xff,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x80,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
/* This table identifies various classes of character by individual bits:
0x01 white space character
0x02 letter
0x04 decimal digit
0x08 hexadecimal digit
0x10 alphanumeric or '_'
0x80 regular expression metacharacter or binary zero
*/
0x80,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 0- 7 */
0x00,0x01,0x01,0x01,0x01,0x01,0x00,0x00, /* 8- 15 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 16- 23 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 24- 31 */
0x01,0x00,0x00,0x00,0x80,0x00,0x00,0x00, /* - ' */
0x80,0x80,0x80,0x80,0x00,0x00,0x80,0x00, /* ( - / */
0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c,0x1c, /* 0 - 7 */
0x1c,0x1c,0x00,0x00,0x00,0x00,0x00,0x80, /* 8 - ? */
0x00,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /* @ - G */
0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* H - O */
0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* P - W */
0x12,0x12,0x12,0x80,0x80,0x00,0x80,0x10, /* X - _ */
0x00,0x1a,0x1a,0x1a,0x1a,0x1a,0x1a,0x12, /* ` - g */
0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* h - o */
0x12,0x12,0x12,0x12,0x12,0x12,0x12,0x12, /* p - w */
0x12,0x12,0x12,0x80,0x80,0x00,0x00,0x00, /* x -127 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 128-135 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 136-143 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 144-151 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 152-159 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 160-167 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 168-175 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 176-183 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 184-191 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 192-199 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 200-207 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 208-215 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 216-223 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 224-231 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 232-239 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */
/* This is a set of tables that came originally from a Windows user. It seems
to be at least an approximation of ISO 8859. In particular, there are
characters greater than 128 that are marked as spaces, letters, etc. */
static const uint8_t tables2[] = {
0,1,2,3,4,5,6,7,
8,9,10,11,12,13,14,15,
16,17,18,19,20,21,22,23,
24,25,26,27,28,29,30,31,
32,33,34,35,36,37,38,39,
40,41,42,43,44,45,46,47,
48,49,50,51,52,53,54,55,
56,57,58,59,60,61,62,63,
64,97,98,99,100,101,102,103,
104,105,106,107,108,109,110,111,
112,113,114,115,116,117,118,119,
120,121,122,91,92,93,94,95,
96,97,98,99,100,101,102,103,
104,105,106,107,108,109,110,111,
112,113,114,115,116,117,118,119,
120,121,122,123,124,125,126,127,
128,129,130,131,132,133,134,135,
136,137,138,139,140,141,142,143,
144,145,146,147,148,149,150,151,
152,153,154,155,156,157,158,159,
160,161,162,163,164,165,166,167,
168,169,170,171,172,173,174,175,
176,177,178,179,180,181,182,183,
184,185,186,187,188,189,190,191,
224,225,226,227,228,229,230,231,
232,233,234,235,236,237,238,239,
240,241,242,243,244,245,246,215,
248,249,250,251,252,253,254,223,
224,225,226,227,228,229,230,231,
232,233,234,235,236,237,238,239,
240,241,242,243,244,245,246,247,
248,249,250,251,252,253,254,255,
0,1,2,3,4,5,6,7,
8,9,10,11,12,13,14,15,
16,17,18,19,20,21,22,23,
24,25,26,27,28,29,30,31,
32,33,34,35,36,37,38,39,
40,41,42,43,44,45,46,47,
48,49,50,51,52,53,54,55,
56,57,58,59,60,61,62,63,
64,97,98,99,100,101,102,103,
104,105,106,107,108,109,110,111,
112,113,114,115,116,117,118,119,
120,121,122,91,92,93,94,95,
96,65,66,67,68,69,70,71,
72,73,74,75,76,77,78,79,
80,81,82,83,84,85,86,87,
88,89,90,123,124,125,126,127,
128,129,130,131,132,133,134,135,
136,137,138,139,140,141,142,143,
144,145,146,147,148,149,150,151,
152,153,154,155,156,157,158,159,
160,161,162,163,164,165,166,167,
168,169,170,171,172,173,174,175,
176,177,178,179,180,181,182,183,
184,185,186,187,188,189,190,191,
224,225,226,227,228,229,230,231,
232,233,234,235,236,237,238,239,
240,241,242,243,244,245,246,215,
248,249,250,251,252,253,254,223,
192,193,194,195,196,197,198,199,
200,201,202,203,204,205,206,207,
208,209,210,211,212,213,214,247,
216,217,218,219,220,221,222,255,
0,62,0,0,1,0,0,0,
0,0,0,0,0,0,0,0,
32,0,0,0,1,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,255,3,
126,0,0,0,126,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,255,3,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,12,2,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
254,255,255,7,0,0,0,0,
0,0,0,0,0,0,0,0,
255,255,127,127,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,254,255,255,7,
0,0,0,0,0,4,32,4,
0,0,0,128,255,255,127,255,
0,0,0,0,0,0,255,3,
254,255,255,135,254,255,255,7,
0,0,0,0,0,4,44,6,
255,255,127,255,255,255,127,255,
0,0,0,0,254,255,255,255,
255,255,255,255,255,255,255,127,
0,0,0,0,254,255,255,255,
255,255,255,255,255,255,255,255,
0,2,0,0,255,255,255,255,
255,255,255,255,255,255,255,127,
0,0,0,0,255,255,255,255,
255,255,255,255,255,255,255,255,
0,0,0,0,254,255,0,252,
1,0,0,248,1,0,0,120,
0,0,0,0,254,255,255,255,
0,0,128,0,0,0,128,0,
255,255,255,255,0,0,0,0,
0,0,0,0,0,0,0,128,
255,255,255,255,0,0,0,0,
0,0,0,0,0,0,0,0,
128,0,0,0,0,0,0,0,
0,1,1,0,1,1,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
1,0,0,0,128,0,0,0,
128,128,128,128,0,0,128,0,
28,28,28,28,28,28,28,28,
28,28,0,0,0,0,0,128,
0,26,26,26,26,26,26,18,
18,18,18,18,18,18,18,18,
18,18,18,18,18,18,18,18,
18,18,18,128,128,0,128,16,
0,26,26,26,26,26,26,18,
18,18,18,18,18,18,18,18,
18,18,18,18,18,18,18,18,
18,18,18,128,128,0,0,0,
0,0,0,0,0,1,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,
1,0,0,0,0,0,0,0,
0,0,18,0,0,0,0,0,
0,0,20,20,0,18,0,0,
0,20,18,0,0,0,0,0,
18,18,18,18,18,18,18,18,
18,18,18,18,18,18,18,18,
18,18,18,18,18,18,18,0,
18,18,18,18,18,18,18,18,
18,18,18,18,18,18,18,18,
18,18,18,18,18,18,18,18,
18,18,18,18,18,18,18,0,
18,18,18,18,18,18,18,18
};
#if !defined(VPCOMPAT) && !defined(HAVE_MEMMOVE)
/*************************************************
* Emulated memmove() for systems without it *
*************************************************/
/* This function can make use of bcopy() if it is available. Otherwise do it by
steam, as there are some non-Unix environments that lack both memmove() and
bcopy(). */
static void *
emulated_memmove(void *d, const void *s, size_t n)
{
#ifdef HAVE_BCOPY
bcopy(s, d, n);
return d;
#else
size_t i;
unsigned char *dest = (unsigned char *)d;
const unsigned char *src = (const unsigned char *)s;
if (dest > src)
{
dest += n;
src += n;
for (i = 0; i < n; ++i) *(--dest) = *(--src);
return (void *)dest;
}
else
{
for (i = 0; i < n; ++i) *dest++ = *src++;
return (void *)(dest - n);
}
#endif /* not HAVE_BCOPY */
}
#undef memmove
#define memmove(d,s,n) emulated_memmove(d,s,n)
#endif /* not VPCOMPAT && not HAVE_MEMMOVE */
#ifndef HAVE_STRERROR
/*************************************************
* Provide strerror() for non-ANSI libraries *
*************************************************/
/* Some old-fashioned systems (e.g. SunOS4) didn't have strerror() in their
libraries. They may no longer be around, but just in case, we can try to
provide the same facility by this simple alternative function. */
extern int sys_nerr;
extern char *sys_errlist[];
char *
strerror(int n)
{
if (n < 0 || n >= sys_nerr) return "unknown error number";
return sys_errlist[n];
}
#endif /* HAVE_STRERROR */
/*************************************************
* Local memory functions *
*************************************************/
/* Alternative memory functions, to test functionality. */
static void *my_malloc(size_t size, void *data)
{
void *block = malloc(size);
(void)data;
if (show_memory)
{
if (block == NULL)
{
fprintf(outfile, "** malloc() failed for %" SIZ_FORM "\n", size);
}
else
{
fprintf(outfile, "malloc %5" SIZ_FORM, size);
#ifdef DEBUG_SHOW_MALLOC_ADDRESSES
fprintf(outfile, " %p", block); /* Not portable */
#endif
if (malloclistptr < MALLOCLISTSIZE)
{
malloclist[malloclistptr] = block;
malloclistlength[malloclistptr++] = size;
}
else
fprintf(outfile, " (not remembered)");
fprintf(outfile, "\n");
}
}
return block;
}
static void my_free(void *block, void *data)
{
(void)data;
if (show_memory && block != NULL)
{
uint32_t i, j;
BOOL found = FALSE;
fprintf(outfile, "free");
for (i = 0; i < malloclistptr; i++)
{
if (block == malloclist[i])
{
fprintf(outfile, " %5" SIZ_FORM, malloclistlength[i]);
malloclistptr--;
for (j = i; j < malloclistptr; j++)
{
malloclist[j] = malloclist[j+1];
malloclistlength[j] = malloclistlength[j+1];
}
found = TRUE;
break;
}
}
if (!found) fprintf(outfile, " unremembered block");
#ifdef DEBUG_SHOW_MALLOC_ADDRESSES
fprintf(outfile, " %p", block); /* Not portable */
#endif
fprintf(outfile, "\n");
}
free(block);
}
/*************************************************
* Callback function for stack guard *
*************************************************/
/* This is set up to be called from pcre2_compile() when the stackguard=n
modifier sets a value greater than zero. The test we do is whether the
parenthesis nesting depth is greater than the value set by the modifier.
Argument: the current parenthesis nesting depth
Returns: non-zero to kill the compilation
*/
static int
stack_guard(uint32_t depth, void *user_data)
{
(void)user_data;
return depth > pat_patctl.stackguard_test;
}
/*************************************************
* JIT memory callback *
*************************************************/
static PCRE2_JIT_STACK*
jit_callback(void *arg)
{
jit_was_used = TRUE;
return (PCRE2_JIT_STACK *)arg;
}
/*************************************************
* Convert UTF-8 character to code point *
*************************************************/
/* This function reads one or more bytes that represent a UTF-8 character,
and returns the codepoint of that character. Note that the function supports
the original UTF-8 definition of RFC 2279, allowing for values in the range 0
to 0x7fffffff, up to 6 bytes long. This makes it possible to generate
codepoints greater than 0x10ffff which are useful for testing PCRE2's error
checking, and also for generating 32-bit non-UTF data values above the UTF
limit.
Argument:
utf8bytes a pointer to the byte vector
end a pointer to the end of the byte vector
vptr a pointer to an int to receive the value
Returns: > 0 => the number of bytes consumed
-6 to 0 => malformed UTF-8 character at offset = (-return)
*/
static int
utf82ord(PCRE2_SPTR8 utf8bytes, PCRE2_SPTR8 end, uint32_t *vptr)
{
uint32_t c = *utf8bytes++;
uint32_t d = c;
int i, j, s;
for (i = -1; i < 6; i++) /* i is number of additional bytes */
{
if ((d & 0x80) == 0) break;
d <<= 1;
}
if (i == -1) { *vptr = c; return 1; } /* ascii character */
if (i == 0 || i == 6) return 0; /* invalid UTF-8 */
/* i now has a value in the range 1-5 */
s = 6*i;
d = (c & utf8_table3[i]) << s;
for (j = 0; j < i; j++)
{
if (utf8bytes >= end) return 0;
c = *utf8bytes++;
if ((c & 0xc0) != 0x80) return -(j+1);
s -= 6;
d |= (c & 0x3f) << s;
}
/* Check that encoding was the correct unique one */
for (j = 0; j < utf8_table1_size; j++)
if (d <= (uint32_t)utf8_table1[j]) break;
if (j != i) return -(i+1);
/* Valid value */
*vptr = d;
return i+1;
}
/*************************************************
* Print one character *
*************************************************/
/* Print a single character either literally, or as a hex escape, and count how
many printed characters are used.
Arguments:
c the character
utf TRUE in UTF mode
f the FILE to print to, or NULL just to count characters
Returns: number of characters written
*/
static int
pchar(uint32_t c, BOOL utf, FILE *f)
{
int n = 0;
char tempbuffer[16];
if (PRINTOK(c))
{
if (f != NULL) fprintf(f, "%c", c);
return 1;
}
if (c < 0x100)
{
if (utf)
{
if (f != NULL) fprintf(f, "\\x{%02x}", c);
return 6;
}
else
{
if (f != NULL) fprintf(f, "\\x%02x", c);
return 4;
}
}
if (f != NULL) n = fprintf(f, "\\x{%02x}", c);
else n = sprintf(tempbuffer, "\\x{%02x}", c);
return n >= 0 ? n : 0;
}
#ifdef SUPPORT_PCRE2_16
/*************************************************
* Find length of 0-terminated 16-bit string *
*************************************************/
static size_t strlen16(PCRE2_SPTR16 p)
{
PCRE2_SPTR16 pp = p;
while (*pp != 0) pp++;
return (int)(pp - p);
}
#endif /* SUPPORT_PCRE2_16 */
#ifdef SUPPORT_PCRE2_32
/*************************************************
* Find length of 0-terminated 32-bit string *
*************************************************/
static size_t strlen32(PCRE2_SPTR32 p)
{
PCRE2_SPTR32 pp = p;
while (*pp != 0) pp++;
return (int)(pp - p);
}
#endif /* SUPPORT_PCRE2_32 */
#ifdef SUPPORT_PCRE2_8
/*************************************************
* Print 8-bit character string *
*************************************************/
/* Must handle UTF-8 strings in utf8 mode. Yields number of characters printed.
For printing *MARK strings, a negative length is given, indicating that the
length is in the first code unit. If handed a NULL file, this function just
counts chars without printing (because pchar() does that). */
static int pchars8(PCRE2_SPTR8 p, int length, BOOL utf, FILE *f)
{
PCRE2_SPTR8 end;
uint32_t c = 0;
int yield = 0;
if (length < 0) length = *p++;
end = p + length;
while (length-- > 0)
{
if (utf)
{
int rc = utf82ord(p, end, &c);
if (rc > 0 && rc <= length + 1) /* Mustn't run over the end */
{
length -= rc - 1;
p += rc;
yield += pchar(c, utf, f);
continue;
}
}
c = *p++;
yield += pchar(c, utf, f);
}
return yield;
}
#endif
#ifdef SUPPORT_PCRE2_16
/*************************************************
* Print 16-bit character string *
*************************************************/
/* Must handle UTF-16 strings in utf mode. Yields number of characters printed.
For printing *MARK strings, a negative length is given, indicating that the
length is in the first code unit. If handed a NULL file, just counts chars
without printing. */
static int pchars16(PCRE2_SPTR16 p, int length, BOOL utf, FILE *f)
{
int yield = 0;
if (length < 0) length = *p++;
while (length-- > 0)
{
uint32_t c = *p++ & 0xffff;
if (utf && c >= 0xD800 && c < 0xDC00 && length > 0)
{
int d = *p & 0xffff;
if (d >= 0xDC00 && d <= 0xDFFF)
{
c = ((c & 0x3ff) << 10) + (d & 0x3ff) + 0x10000;
length--;
p++;
}
}
yield += pchar(c, utf, f);
}
return yield;
}
#endif /* SUPPORT_PCRE2_16 */
#ifdef SUPPORT_PCRE2_32
/*************************************************
* Print 32-bit character string *
*************************************************/
/* Must handle UTF-32 strings in utf mode. Yields number of characters printed.
For printing *MARK strings, a negative length is given, indicating that the
length is in the first code unit. If handed a NULL file, just counts chars
without printing. */
static int pchars32(PCRE2_SPTR32 p, int length, BOOL utf, FILE *f)
{
int yield = 0;
(void)(utf); /* Avoid compiler warning */
if (length < 0) length = *p++;
while (length-- > 0)
{
uint32_t c = *p++;
yield += pchar(c, utf, f);
}
return yield;
}
#endif /* SUPPORT_PCRE2_32 */
/*************************************************
* Convert character value to UTF-8 *
*************************************************/
/* This function takes an integer value in the range 0 - 0x7fffffff
and encodes it as a UTF-8 character in 0 to 6 bytes. It is needed even when the
8-bit library is not supported, to generate UTF-8 output for non-ASCII
characters.
Arguments:
cvalue the character value
utf8bytes pointer to buffer for result - at least 6 bytes long
Returns: number of characters placed in the buffer
*/
static int
ord2utf8(uint32_t cvalue, uint8_t *utf8bytes)
{
int i, j;
if (cvalue > 0x7fffffffu)
return -1;
for (i = 0; i < utf8_table1_size; i++)
if (cvalue <= (uint32_t)utf8_table1[i]) break;
utf8bytes += i;
for (j = i; j > 0; j--)
{
*utf8bytes-- = 0x80 | (cvalue & 0x3f);
cvalue >>= 6;
}
*utf8bytes = utf8_table2[i] | cvalue;
return i + 1;
}
#ifdef SUPPORT_PCRE2_16
/*************************************************
* Convert string to 16-bit *
*************************************************/
/* In UTF mode the input is always interpreted as a string of UTF-8 bytes using
the original UTF-8 definition of RFC 2279, which allows for up to 6 bytes, and
code values from 0 to 0x7fffffff. However, values greater than the later UTF
limit of 0x10ffff cause an error. In non-UTF mode the input is interpreted as
UTF-8 if the utf8_input modifier is set, but an error is generated for values
greater than 0xffff.
If all the input bytes are ASCII, the space needed for a 16-bit string is
exactly double the 8-bit size. Otherwise, the size needed for a 16-bit string
is no more than double, because up to 0xffff uses no more than 3 bytes in UTF-8
but possibly 4 in UTF-16. Higher values use 4 bytes in UTF-8 and up to 4 bytes
in UTF-16. The result is always left in pbuffer16. Impose a minimum size to
save repeated re-sizing.
Note that this function does not object to surrogate values. This is
deliberate; it makes it possible to construct UTF-16 strings that are invalid,
for the purpose of testing that they are correctly faulted.
Arguments:
p points to a byte string
utf true in UTF mode
lenptr points to number of bytes in the string (excluding trailing zero)
Returns: 0 on success, with the length updated to the number of 16-bit
data items used (excluding the trailing zero)
OR -1 if a UTF-8 string is malformed
OR -2 if a value > 0x10ffff is encountered in UTF mode
OR -3 if a value > 0xffff is encountered when not in UTF mode
*/
static int
to16(uint8_t *p, int utf, PCRE2_SIZE *lenptr)
{
uint16_t *pp;
PCRE2_SIZE len = *lenptr;
if (pbuffer16_size < 2*len + 2)
{
if (pbuffer16 != NULL) free(pbuffer16);
pbuffer16_size = 2*len + 2;
if (pbuffer16_size < 4096) pbuffer16_size = 4096;
pbuffer16 = (uint16_t *)malloc(pbuffer16_size);
if (pbuffer16 == NULL)
{
fprintf(stderr, "pcre2test: malloc(%" SIZ_FORM ") failed for pbuffer16\n",
pbuffer16_size);
exit(1);
}
}
pp = pbuffer16;
if (!utf && (pat_patctl.control & CTL_UTF8_INPUT) == 0)
{
for (; len > 0; len--) *pp++ = *p++;
}
else while (len > 0)
{
uint32_t c;
const uint8_t *end = p + len;
int chlen = utf82ord(p, end, &c);
if (chlen <= 0) return -1;
if (!utf && c > 0xffff) return -3;
if (c > 0x10ffff) return -2;
p += chlen;
len -= chlen;
if (c < 0x10000) *pp++ = c; else
{
c -= 0x10000;
*pp++ = 0xD800 | (c >> 10);
*pp++ = 0xDC00 | (c & 0x3ff);
}
}
*pp = 0;
*lenptr = pp - pbuffer16;
return 0;
}
#endif
#ifdef SUPPORT_PCRE2_32
/*************************************************
* Convert string to 32-bit *
*************************************************/
/* In UTF mode the input is always interpreted as a string of UTF-8 bytes using
the original UTF-8 definition of RFC 2279, which allows for up to 6 bytes, and
code values from 0 to 0x7fffffff. However, values greater than the later UTF
limit of 0x10ffff cause an error.
In non-UTF mode the input is interpreted as UTF-8 if the utf8_input modifier
is set, and no limit is imposed. There is special interpretation of the 0xff
byte (which is illegal in UTF-8) in this case: it causes the top bit of the
next character to be set. This provides a way of generating 32-bit characters
greater than 0x7fffffff.
If all the input bytes are ASCII, the space needed for a 32-bit string is
exactly four times the 8-bit size. Otherwise, the size needed for a 32-bit
string is no more than four times, because the number of characters must be
less than the number of bytes. The result is always left in pbuffer32. Impose a
minimum size to save repeated re-sizing.
Note that this function does not object to surrogate values. This is
deliberate; it makes it possible to construct UTF-32 strings that are invalid,
for the purpose of testing that they are correctly faulted.
Arguments:
p points to a byte string
utf true in UTF mode
lenptr points to number of bytes in the string (excluding trailing zero)
Returns: 0 on success, with the length updated to the number of 32-bit
data items used (excluding the trailing zero)
OR -1 if a UTF-8 string is malformed
OR -2 if a value > 0x10ffff is encountered in UTF mode
*/
static int
to32(uint8_t *p, int utf, PCRE2_SIZE *lenptr)
{
uint32_t *pp;
PCRE2_SIZE len = *lenptr;
if (pbuffer32_size < 4*len + 4)
{
if (pbuffer32 != NULL) free(pbuffer32);
pbuffer32_size = 4*len + 4;
if (pbuffer32_size < 8192) pbuffer32_size = 8192;
pbuffer32 = (uint32_t *)malloc(pbuffer32_size);
if (pbuffer32 == NULL)
{
fprintf(stderr, "pcre2test: malloc(%" SIZ_FORM ") failed for pbuffer32\n",
pbuffer32_size);
exit(1);
}
}
pp = pbuffer32;
if (!utf && (pat_patctl.control & CTL_UTF8_INPUT) == 0)
{
for (; len > 0; len--) *pp++ = *p++;
}
else while (len > 0)
{
int chlen;
uint32_t c;
uint32_t topbit = 0;
const uint8_t *end = p + len;
if (!utf && *p == 0xff && len > 1)
{
topbit = 0x80000000u;
p++;
len--;
}
chlen = utf82ord(p, end, &c);
if (chlen <= 0) return -1;
if (utf && c > 0x10ffff) return -2;
p += chlen;
len -= chlen;
*pp++ = c | topbit;
}
*pp = 0;
*lenptr = pp - pbuffer32;
return 0;
}
#endif /* SUPPORT_PCRE2_32 */
/* This function is no longer used. Keep it around for a while, just in case it
needs to be re-instated. */
#ifdef NEVERNEVERNEVER
/*************************************************
* Move back by so many characters *
*************************************************/
/* Given a code unit offset in a subject string, move backwards by a number of
characters, and return the resulting offset.
Arguments:
subject pointer to the string
offset start offset
count count to move back by
utf TRUE if in UTF mode
Returns: a possibly changed offset
*/
static PCRE2_SIZE
backchars(uint8_t *subject, PCRE2_SIZE offset, uint32_t count, BOOL utf)
{
if (!utf || test_mode == PCRE32_MODE)
return (count >= offset)? 0 : (offset - count);
else if (test_mode == PCRE8_MODE)
{
PCRE2_SPTR8 pp = (PCRE2_SPTR8)subject + offset;
for (; count > 0 && pp > (PCRE2_SPTR8)subject; count--)
{
pp--;
while ((*pp & 0xc0) == 0x80) pp--;
}
return pp - (PCRE2_SPTR8)subject;
}
else /* 16-bit mode */
{
PCRE2_SPTR16 pp = (PCRE2_SPTR16)subject + offset;
for (; count > 0 && pp > (PCRE2_SPTR16)subject; count--)
{
pp--;
if ((*pp & 0xfc00) == 0xdc00) pp--;
}
return pp - (PCRE2_SPTR16)subject;
}
}
#endif /* NEVERNEVERNEVER */
/*************************************************
* Expand input buffers *
*************************************************/
/* This function doubles the size of the input buffer and the buffer for
keeping an 8-bit copy of patterns (pbuffer8), and copies the current buffers to
the new ones.
Arguments: none
Returns: nothing (aborts if malloc() fails)
*/
static void
expand_input_buffers(void)
{
int new_pbuffer8_size = 2*pbuffer8_size;
uint8_t *new_buffer = (uint8_t *)malloc(new_pbuffer8_size);
uint8_t *new_pbuffer8 = (uint8_t *)malloc(new_pbuffer8_size);
if (new_buffer == NULL || new_pbuffer8 == NULL)
{
fprintf(stderr, "pcre2test: malloc(%d) failed\n", new_pbuffer8_size);
exit(1);
}
memcpy(new_buffer, buffer, pbuffer8_size);
memcpy(new_pbuffer8, pbuffer8, pbuffer8_size);
pbuffer8_size = new_pbuffer8_size;
free(buffer);
free(pbuffer8);
buffer = new_buffer;
pbuffer8 = new_pbuffer8;
}
/*************************************************
* Read or extend an input line *
*************************************************/
/* Input lines are read into buffer, but both patterns and data lines can be
continued over multiple input lines. In addition, if the buffer fills up, we
want to automatically expand it so as to be able to handle extremely large
lines that are needed for certain stress tests, although this is less likely
now that there are repetition features for both patterns and data. When the
input buffer is expanded, the other two buffers must also be expanded likewise,
and the contents of pbuffer, which are a copy of the input for callouts, must
be preserved (for when expansion happens for a data line). This is not the most
optimal way of handling this, but hey, this is just a test program!
Arguments:
f the file to read
start where in buffer to start (this *must* be within buffer)
prompt for stdin or readline()
Returns: pointer to the start of new data
could be a copy of start, or could be moved
NULL if no data read and EOF reached
*/
static uint8_t *
extend_inputline(FILE *f, uint8_t *start, const char *prompt)
{
uint8_t *here = start;
for (;;)
{
size_t rlen = (size_t)(pbuffer8_size - (here - buffer));
if (rlen > 1000)
{
size_t dlen;
/* If libreadline or libedit support is required, use readline() to read a
line if the input is a terminal. Note that readline() removes the trailing
newline, so we must put it back again, to be compatible with fgets(). */
#if defined(SUPPORT_LIBREADLINE) || defined(SUPPORT_LIBEDIT)
if (INTERACTIVE(f))
{
size_t len;
char *s = readline(prompt);
if (s == NULL) return (here == start)? NULL : start;
len = strlen(s);
if (len > 0) add_history(s);
if (len > rlen - 1) len = rlen - 1;
memcpy(here, s, len);
here[len] = '\n';
here[len+1] = 0;
free(s);
}
else
#endif
/* Read the next line by normal means, prompting if the file is a tty. */
{
if (INTERACTIVE(f)) printf("%s", prompt);
if (fgets((char *)here, rlen, f) == NULL)
return (here == start)? NULL : start;
}
dlen = strlen((char *)here);
here += dlen;
/* Check for end of line reached. Take care not to read data from before
start (dlen will be zero for a file starting with a binary zero). */
if (here > start && here[-1] == '\n') return start;
/* If we have not read a newline when reading a file, we have either filled
the buffer or reached the end of the file. We can detect the former by
checking that the string fills the buffer, and the latter by feof(). If
neither of these is true, it means we read a binary zero which has caused
strlen() to give a short length. This is a hard error because pcre2test
expects to work with C strings. */
if (!INTERACTIVE(f) && dlen < rlen - 1 && !feof(f))
{
fprintf(outfile, "** Binary zero encountered in input\n");
fprintf(outfile, "** pcre2test run abandoned\n");
exit(1);
}
}
else
{
size_t start_offset = start - buffer;
size_t here_offset = here - buffer;
expand_input_buffers();
start = buffer + start_offset;
here = buffer + here_offset;
}
}
PCRE2_UNREACHABLE(); /* Control never reaches here */
}
/*************************************************
* Case-independent strncmp() function *
*************************************************/
/*
Arguments:
s first string
t second string
n number of characters to compare
Returns: < 0, = 0, or > 0, according to the comparison
*/
static int
strncmpic(const uint8_t *s, const uint8_t *t, int n)
{
while (n--)
{
int c = tolower(*s++) - tolower(*t++);
if (c != 0) return c;
}
return 0;
}
/*************************************************
* Scan the main modifier list *
*************************************************/
/* This function searches the modifier list for a long modifier name.
Argument:
p start of the name
lenp length of the name
Returns: an index in the modifier list, or -1 on failure
*/
static int
scan_modifiers(const uint8_t *p, unsigned int len)
{
int bot = 0;
int top = MODLISTCOUNT;
while (top > bot)
{
int mid = (bot + top)/2;
unsigned int mlen = strlen(modlist[mid].name);
int c = strncmp((const char *)p, modlist[mid].name, (len < mlen)? len : mlen);
if (c == 0)
{
if (len == mlen) return mid;
c = (int)len - (int)mlen;
}
if (c > 0) bot = mid + 1; else top = mid;
}
return -1;
}
/*************************************************
* Check a modifer and find its field *
*************************************************/
/* This function is called when a modifier has been identified. We check that
it is allowed here and find the field that is to be changed.
Arguments:
m the modifier list entry
ctx CTX_PAT => pattern context
CTX_POPPAT => pattern context for popped pattern
CTX_DEFPAT => default pattern context
CTX_DAT => data context
CTX_DEFDAT => default data context
pctl point to pattern control block
dctl point to data control block
c a single character or 0
Returns: a field pointer or NULL
*/
static void *
check_modifier(modstruct *m, int ctx, patctl *pctl, datctl *dctl, uint32_t c)
{
void *field = NULL;
PCRE2_SIZE offset = m->offset;
if (restrict_for_perl_test) switch(m->which)
{
case MOD_PNDP:
case MOD_PATP:
case MOD_DATP:
case MOD_PDP:
break;
default:
fprintf(outfile, "** \"%s\" is not allowed in a Perl-compatible test\n",
m->name);
return NULL;
}
switch (m->which)
{
case MOD_CTC: /* Compile context modifier */
if (ctx == CTX_DEFPAT) field = PTR(default_pat_context);
else if (ctx == CTX_PAT) field = PTR(pat_context);
break;
case MOD_CTM: /* Match context modifier */
if (ctx == CTX_DEFDAT) field = PTR(default_dat_context);
else if (ctx == CTX_DAT) field = PTR(dat_context);
break;
case MOD_DAT: /* Data line modifier */
case MOD_DATP: /* Allowed for Perl test */
if (dctl != NULL) field = dctl;
break;
case MOD_PAT: /* Pattern modifier */
case MOD_PATP: /* Allowed for Perl test */
if (pctl != NULL) field = pctl;
break;
case MOD_PD: /* Pattern or data line modifier */
case MOD_PDP: /* Ditto, allowed for Perl test */
case MOD_PND: /* Ditto, but not default pattern */
case MOD_PNDP: /* Ditto, allowed for Perl test */
if (dctl != NULL) field = dctl;
else if (pctl != NULL && (m->which == MOD_PD || m->which == MOD_PDP ||
ctx != CTX_DEFPAT))
field = pctl;
break;
}
if (field == NULL)
{
if (c == 0)
fprintf(outfile, "** \"%s\" is not valid here\n", m->name);
else
fprintf(outfile, "** /%c is not valid here\n", c);
return NULL;
}
return (char *)field + offset;
}
/*************************************************
* Decode a modifier list *
*************************************************/
/* A pointer to a control block is NULL when called in cases when that block is
not relevant. They are never all relevant in one call. At least one of patctl
and datctl is NULL. The second argument specifies which context to use for
modifiers that apply to contexts.
Arguments:
p point to modifier string
ctx CTX_PAT => pattern context
CTX_POPPAT => pattern context for popped pattern
CTX_DEFPAT => default pattern context
CTX_DAT => data context
CTX_DEFDAT => default data context
pctl point to pattern control block
dctl point to data control block
Returns: TRUE if successful decode, FALSE otherwise
*/
static BOOL
decode_modifiers(uint8_t *p, int ctx, patctl *pctl, datctl *dctl)
{
uint8_t *ep, *pp;
long li;
unsigned long uli;
BOOL first = TRUE;
for (;;)
{
void *field;
modstruct *m;
BOOL off = FALSE;
unsigned int i, len;
int index;
char *endptr;
/* Skip white space and commas. */
while (isspace(*p) || *p == ',') p++;
if (*p == 0) break;
/* Find the end of the item; lose trailing whitespace at end of line. */
for (ep = p; *ep != 0 && *ep != ','; ep++);
if (*ep == 0)
{
while (ep > p && isspace(ep[-1])) ep--;
*ep = 0;
}
/* Remember if the first character is '-'. */
if (*p == '-')
{
off = TRUE;
p++;
}
/* Find the length of a full-length modifier name, and scan for it. */
pp = p;
while (pp < ep && *pp != '=') pp++;
index = scan_modifiers(p, pp - p);
/* If the first modifier is unrecognized, try to interpret it as a sequence
of single-character abbreviated modifiers. None of these modifiers have any
associated data. They just set options or control bits. */
if (index < 0)
{
uint32_t cc;
uint8_t *mp = p;
if (!first)
{
fprintf(outfile, "** Unrecognized modifier \"%.*s\"\n", (int)(ep-p), p);
if (ep - p == 1)
fprintf(outfile, "** Single-character modifiers must come first\n");
return FALSE;
}
first = FALSE;
for (cc = *p; cc != ',' && cc != '\n' && cc != 0; cc = *(++p))
{
for (i = 0; i < C1MODLISTCOUNT; i++)
if (cc == c1modlist[i].onechar) break;
if (i >= C1MODLISTCOUNT)
{
fprintf(outfile, "** Unrecognized modifier '%c' in modifier string "
"\"%.*s\"\n", *p, (int)(ep-mp), mp);
return FALSE;
}
if (c1modlist[i].index >= 0)
{
index = c1modlist[i].index;
}
else
{
index = scan_modifiers((const uint8_t *)(c1modlist[i].fullname),
strlen(c1modlist[i].fullname));
if (index < 0)
{
fprintf(outfile, "** Internal error: single-character equivalent "
"modifier \"%s\" not found\n", c1modlist[i].fullname);
return FALSE;
}
c1modlist[i].index = index; /* Cache for next time */
}
field = check_modifier(modlist + index, ctx, pctl, dctl, *p);
if (field == NULL) return FALSE;
/* /x is a special case; a second appearance changes PCRE2_EXTENDED to
PCRE2_EXTENDED_MORE. */
if (cc == 'x' && (*((uint32_t *)field) & PCRE2_EXTENDED) != 0)
{
*((uint32_t *)field) &= ~PCRE2_EXTENDED;
*((uint32_t *)field) |= PCRE2_EXTENDED_MORE;
}
else
*((uint32_t *)field) |= modlist[index].value;
}
continue; /* With tne next (fullname) modifier */
}
/* We have a match on a full-name modifier. Check for the existence of data
when needed. */
m = modlist + index; /* Save typing */
if (m->type != MOD_CTL && m->type != MOD_OPT && m->type != MOD_OPTMZ &&
(m->type != MOD_IND || *pp == '='))
{
if (*pp++ != '=')
{
fprintf(outfile, "** '=' expected after \"%s\"\n", m->name);
return FALSE;
}
if (off)
{
fprintf(outfile, "** '-' is not valid for \"%s\"\n", m->name);
return FALSE;
}
}
/* These on/off types have no data. */
else if (*pp != ',' && *pp != '\n' && *pp != ' ' && *pp != 0)
{
fprintf(outfile, "** Unrecognized modifier '%.*s'\n", (int)(ep-p), p);
return FALSE;
}
/* Set the data length for those types that have data. Then find the field
that is to be set. If check_modifier() returns NULL, it has already output an
error message. */
len = ep - pp;
field = check_modifier(m, ctx, pctl, dctl, 0);
if (field == NULL) return FALSE;
/* Process according to data type. */
switch (m->type)
{
case MOD_CTL:
case MOD_OPT:
if (off) *((uint32_t *)field) &= ~m->value;
else *((uint32_t *)field) |= m->value;
break;
case MOD_OPTMZ:
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE)
pcre2_set_optimize_8((pcre2_compile_context_8*)field, m->value);
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE)
pcre2_set_optimize_16((pcre2_compile_context_16*)field, m->value);
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE)
pcre2_set_optimize_32((pcre2_compile_context_32*)field, m->value);
#endif
break;
case MOD_BSR:
if (len == 7 && strncmpic(pp, (const uint8_t *)"default", 7) == 0)
{
#ifdef BSR_ANYCRLF
*((uint16_t *)field) = PCRE2_BSR_ANYCRLF;
#else
*((uint16_t *)field) = PCRE2_BSR_UNICODE;
#endif
if (ctx == CTX_PAT || ctx == CTX_DEFPAT) pctl->control2 &= ~CTL2_BSR_SET;
else dctl->control2 &= ~CTL2_BSR_SET;
}
else
{
if (len == 7 && strncmpic(pp, (const uint8_t *)"anycrlf", 7) == 0)
*((uint16_t *)field) = PCRE2_BSR_ANYCRLF;
else if (len == 7 && strncmpic(pp, (const uint8_t *)"unicode", 7) == 0)
*((uint16_t *)field) = PCRE2_BSR_UNICODE;
else goto INVALID_VALUE;
if (ctx == CTX_PAT || ctx == CTX_DEFPAT) pctl->control2 |= CTL2_BSR_SET;
else dctl->control2 |= CTL2_BSR_SET;
}
pp = ep;
break;
case MOD_CHR: /* A single character */
*((uint32_t *)field) = *pp++;
break;
case MOD_CON: /* A convert type/options list */
for (;; pp++)
{
uint8_t *colon = (uint8_t *)strchr((const char *)pp, ':');
len = ((colon != NULL && colon < ep)? colon:ep) - pp;
for (i = 0; i < convertlistcount; i++)
{
if (strncmpic(pp, (const uint8_t *)convertlist[i].name, len) == 0)
{
if (*((uint32_t *)field) == CONVERT_UNSET)
*((uint32_t *)field) = convertlist[i].option;
else
*((uint32_t *)field) |= convertlist[i].option;
break;
}
}
if (i >= convertlistcount) goto INVALID_VALUE;
pp += len;
if (*pp != ':') break;
}
break;
case MOD_IN2: /* One or two unsigned integers */
if (!isdigit(*pp)) goto INVALID_VALUE;
uli = strtoul((const char *)pp, &endptr, 10);
if (U32OVERFLOW(uli)) goto INVALID_VALUE;
((uint32_t *)field)[0] = (uint32_t)uli;
if (*endptr == ':')
{
uli = strtoul((const char *)endptr+1, &endptr, 10);
if (U32OVERFLOW(uli)) goto INVALID_VALUE;
((uint32_t *)field)[1] = (uint32_t)uli;
}
else ((uint32_t *)field)[1] = 0;
pp = (uint8_t *)endptr;
break;
/* PCRE2_SIZE_MAX is usually SIZE_MAX, which may be greater, equal to, or
less than ULONG_MAX. So first test for overflowing the long int, and then
test for overflowing PCRE2_SIZE_MAX if it is smaller than ULONG_MAX. */
case MOD_SIZ: /* PCRE2_SIZE value */
if (!isdigit(*pp)) goto INVALID_VALUE;
uli = strtoul((const char *)pp, &endptr, 10);
if (uli == ULONG_MAX) goto INVALID_VALUE;
#if ULONG_MAX > PCRE2_SIZE_MAX
if (uli > PCRE2_SIZE_MAX) goto INVALID_VALUE;
#endif
*((PCRE2_SIZE *)field) = (PCRE2_SIZE)uli;
pp = (uint8_t *)endptr;
break;
case MOD_IND: /* Unsigned integer with default */
if (len == 0)
{
*((uint32_t *)field) = (uint32_t)(m->value);
break;
}
/* Fall through */
case MOD_INT: /* Unsigned integer */
if (!isdigit(*pp)) goto INVALID_VALUE;
uli = strtoul((const char *)pp, &endptr, 10);
if (U32OVERFLOW(uli)) goto INVALID_VALUE;
*((uint32_t *)field) = (uint32_t)uli;
pp = (uint8_t *)endptr;
break;
case MOD_INS: /* Signed integer */
if (!isdigit(*pp) && *pp != '-') goto INVALID_VALUE;
li = strtol((const char *)pp, &endptr, 10);
if (S32OVERFLOW(li)) goto INVALID_VALUE;
*((int32_t *)field) = (int32_t)li;
pp = (uint8_t *)endptr;
break;
case MOD_NL:
for (i = 0; i < sizeof(newlines)/sizeof(char *); i++)
if (len == strlen(newlines[i]) &&
strncmpic(pp, (const uint8_t *)newlines[i], len) == 0) break;
if (i >= sizeof(newlines)/sizeof(char *)) goto INVALID_VALUE;
if (i == 0)
{
*((uint16_t *)field) = NEWLINE_DEFAULT;
if (ctx == CTX_PAT || ctx == CTX_DEFPAT) pctl->control2 &= ~CTL2_NL_SET;
else dctl->control2 &= ~CTL2_NL_SET;
}
else
{
*((uint16_t *)field) = i;
if (ctx == CTX_PAT || ctx == CTX_DEFPAT) pctl->control2 |= CTL2_NL_SET;
else dctl->control2 |= CTL2_NL_SET;
}
pp = ep;
break;
case MOD_NN: /* Name or (signed) number; may be several */
if (isdigit(*pp) || *pp == '-')
{
int ct = MAXCPYGET - 1;
int32_t value;
li = strtol((const char *)pp, &endptr, 10);
if (S32OVERFLOW(li)) goto INVALID_VALUE;
value = (int32_t)li;
field = (char *)field - m->offset + m->value; /* Adjust field ptr */
if (value >= 0) /* Add new number */
{
while (*((int32_t *)field) >= 0 && ct-- > 0) /* Skip previous */
field = (char *)field + sizeof(int32_t);
if (ct <= 0)
{
fprintf(outfile, "** Too many numeric \"%s\" modifiers\n", m->name);
return FALSE;
}
}
*((int32_t *)field) = value;
if (ct > 0) ((int32_t *)field)[1] = -1;
pp = (uint8_t *)endptr;
}
/* Multiple strings are put end to end. */
else
{
char *nn = (char *)field;
if (len > 0) /* Add new name */
{
if (len > MAX_NAME_SIZE)
{
fprintf(outfile, "** Group name in \"%s\" is too long\n", m->name);
return FALSE;
}
while (*nn != 0) nn += strlen(nn) + 1;
if (nn + len + 2 - (char *)field > LENCPYGET)
{
fprintf(outfile, "** Too many characters in named \"%s\" modifiers\n",
m->name);
return FALSE;
}
memcpy(nn, pp, len);
}
nn[len] = 0 ;
nn[len+1] = 0;
pp = ep;
}
break;
case MOD_STR:
if (len + 1 > m->value)
{
fprintf(outfile, "** Overlong value for \"%s\" (max %d code units)\n",
m->name, m->value - 1);
return FALSE;
}
memcpy(field, pp, len);
((uint8_t *)field)[len] = 0;
pp = ep;
break;
}
if (*pp != ',' && *pp != '\n' && *pp != ' ' && *pp != 0)
{
fprintf(outfile, "** Comma expected after modifier item \"%s\"\n", m->name);
return FALSE;
}
p = pp;
if (ctx == CTX_POPPAT &&
(pctl->options != 0 ||
pctl->tables_id != 0 ||
pctl->locale[0] != 0 ||
(pctl->control & NOTPOP_CONTROLS) != 0))
{
fprintf(outfile, "** \"%s\" is not valid here\n", m->name);
return FALSE;
}
}
return TRUE;
INVALID_VALUE:
fprintf(outfile, "** Invalid value in \"%.*s\"\n", (int)(ep-p), p);
return FALSE;
}
/*************************************************
* Get info from a pattern *
*************************************************/
/* A wrapped call to pcre2_pattern_info(), applied to the current compiled
pattern.
Arguments:
what code for the required information
where where to put the answer
unsetok PCRE2_ERROR_UNSET is an "expected" result
Returns: the return from pcre2_pattern_info()
*/
static int
pattern_info(int what, void *where, BOOL unsetok)
{
int rc;
PCRE2_PATTERN_INFO(rc, compiled_code, what, NULL); /* Exercise the code */
PCRE2_PATTERN_INFO(rc, compiled_code, what, where);
if (rc >= 0) return 0;
if (rc != PCRE2_ERROR_UNSET || !unsetok)
{
fprintf(outfile, "Error %d from pcre2_pattern_info_%d(%d)\n", rc, test_mode,
what);
if (rc == PCRE2_ERROR_BADMODE)
fprintf(outfile, "Running in %d-bit mode but pattern was compiled in "
"%d-bit mode\n", test_mode,
8 * (FLD(compiled_code, flags) & PCRE2_MODE_MASK));
}
return rc;
}
#ifdef SUPPORT_PCRE2_8
/*************************************************
* Show something in a list *
*************************************************/
/* This function just helps to keep the code that uses it tidier. It's used for
various lists of things where there needs to be introductory text before the
first item. As these calls are all in the POSIX-support code, they happen only
when 8-bit mode is supported. */
static void
prmsg(const char **msg, const char *s)
{
fprintf(outfile, "%s %s", *msg, s);
*msg = "";
}
#endif /* SUPPORT_PCRE2_8 */
/*************************************************
* Show control bits *
*************************************************/
/* Called for mutually exclusive controls and for unsupported POSIX controls.
Because the bits are unique, this can be used for both pattern and data control
words.
Arguments:
controls control bits
controls2 more control bits
before text to print before
Returns: nothing
*/
static void
show_controls(uint32_t controls, uint32_t controls2, const char *before)
{
fprintf(outfile, "%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s",
before,
((controls & CTL_AFTERTEXT) != 0)? " aftertext" : "",
((controls & CTL_ALLAFTERTEXT) != 0)? " allaftertext" : "",
((controls & CTL_ALLCAPTURES) != 0)? " allcaptures" : "",
((controls & CTL_ALLUSEDTEXT) != 0)? " allusedtext" : "",
((controls2 & CTL2_ALLVECTOR) != 0)? " allvector" : "",
((controls & CTL_ALTGLOBAL) != 0)? " altglobal" : "",
((controls & CTL_BINCODE) != 0)? " bincode" : "",
((controls2 & CTL2_BSR_SET) != 0)? " bsr" : "",
((controls & CTL_CALLOUT_CAPTURE) != 0)? " callout_capture" : "",
((controls2 & CTL2_CALLOUT_EXTRA) != 0)? " callout_extra" : "",
((controls & CTL_CALLOUT_INFO) != 0)? " callout_info" : "",
((controls & CTL_CALLOUT_NONE) != 0)? " callout_none" : "",
((controls2 & CTL2_CALLOUT_NO_WHERE) != 0)? " callout_no_where" : "",
((controls & CTL_DFA) != 0)? " dfa" : "",
((controls & CTL_EXPAND) != 0)? " expand" : "",
((controls & CTL_FINDLIMITS) != 0)? " find_limits" : "",
((controls & CTL_FINDLIMITS_NOHEAP) != 0)? " find_limits_noheap" : "",
((controls2 & CTL2_FRAMESIZE) != 0)? " framesize" : "",
((controls & CTL_FULLBINCODE) != 0)? " fullbincode" : "",
((controls & CTL_GETALL) != 0)? " getall" : "",
((controls & CTL_GLOBAL) != 0)? " global" : "",
((controls2 & CTL2_HEAPFRAMES_SIZE) != 0)? " heapframes_size" : "",
((controls & CTL_HEXPAT) != 0)? " hex" : "",
((controls & CTL_INFO) != 0)? " info" : "",
((controls & CTL_JITFAST) != 0)? " jitfast" : "",
((controls & CTL_JITVERIFY) != 0)? " jitverify" : "",
((controls & CTL_MARK) != 0)? " mark" : "",
((controls & CTL_MEMORY) != 0)? " memory" : "",
((controls2 & CTL2_NL_SET) != 0)? " newline" : "",
((controls & CTL_NULLCONTEXT) != 0)? " null_context" : "",
((controls2 & CTL2_NULL_REPLACEMENT) != 0)? " null_replacement" : "",
((controls2 & CTL2_NULL_SUBJECT) != 0)? " null_subject" : "",
((controls & CTL_POSIX) != 0)? " posix" : "",
((controls & CTL_POSIX_NOSUB) != 0)? " posix_nosub" : "",
((controls & CTL_PUSH) != 0)? " push" : "",
((controls & CTL_PUSHCOPY) != 0)? " pushcopy" : "",
((controls & CTL_PUSHTABLESCOPY) != 0)? " pushtablescopy" : "",
((controls & CTL_STARTCHAR) != 0)? " startchar" : "",
((controls2 & CTL2_SUBSTITUTE_CALLOUT) != 0)? " substitute_callout" : "",
((controls2 & CTL2_SUBSTITUTE_CASE_CALLOUT) != 0)? " substitute_case_callout" : "",
((controls2 & CTL2_SUBSTITUTE_EXTENDED) != 0)? " substitute_extended" : "",
((controls2 & CTL2_SUBSTITUTE_LITERAL) != 0)? " substitute_literal" : "",
((controls2 & CTL2_SUBSTITUTE_MATCHED) != 0)? " substitute_matched" : "",
((controls2 & CTL2_SUBSTITUTE_OVERFLOW_LENGTH) != 0)? " substitute_overflow_length" : "",
((controls2 & CTL2_SUBSTITUTE_REPLACEMENT_ONLY) != 0)? " substitute_replacement_only" : "",
((controls2 & CTL2_SUBSTITUTE_UNKNOWN_UNSET) != 0)? " substitute_unknown_unset" : "",
((controls2 & CTL2_SUBSTITUTE_UNSET_EMPTY) != 0)? " substitute_unset_empty" : "",
((controls & CTL_USE_LENGTH) != 0)? " use_length" : "",
((controls & CTL_UTF8_INPUT) != 0)? " utf8_input" : "",
((controls & CTL_ZERO_TERMINATE) != 0)? " zero_terminate" : "");
}
/*************************************************
* Show compile options *
*************************************************/
/* Called from show_pattern_info() and for unsupported POSIX options.
Arguments:
options an options word
before text to print before
after text to print after
Returns: nothing
*/
static void
show_compile_options(uint32_t options, const char *before, const char *after)
{
if (options == 0) fprintf(outfile, "%s %s", before, after);
else fprintf(outfile, "%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s",
before,
((options & PCRE2_ALT_BSUX) != 0)? " alt_bsux" : "",
((options & PCRE2_ALT_CIRCUMFLEX) != 0)? " alt_circumflex" : "",
((options & PCRE2_ALT_EXTENDED_CLASS) != 0)? " alt_extended_class" : "",
((options & PCRE2_ALT_VERBNAMES) != 0)? " alt_verbnames" : "",
((options & PCRE2_ALLOW_EMPTY_CLASS) != 0)? " allow_empty_class" : "",
((options & PCRE2_ANCHORED) != 0)? " anchored" : "",
((options & PCRE2_AUTO_CALLOUT) != 0)? " auto_callout" : "",
((options & PCRE2_CASELESS) != 0)? " caseless" : "",
((options & PCRE2_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "",
((options & PCRE2_DOTALL) != 0)? " dotall" : "",
((options & PCRE2_DUPNAMES) != 0)? " dupnames" : "",
((options & PCRE2_ENDANCHORED) != 0)? " endanchored" : "",
((options & PCRE2_EXTENDED) != 0)? " extended" : "",
((options & PCRE2_EXTENDED_MORE) != 0)? " extended_more" : "",
((options & PCRE2_FIRSTLINE) != 0)? " firstline" : "",
((options & PCRE2_LITERAL) != 0)? " literal" : "",
((options & PCRE2_MATCH_INVALID_UTF) != 0)? " match_invalid_utf" : "",
((options & PCRE2_MATCH_UNSET_BACKREF) != 0)? " match_unset_backref" : "",
((options & PCRE2_MULTILINE) != 0)? " multiline" : "",
((options & PCRE2_NEVER_BACKSLASH_C) != 0)? " never_backslash_c" : "",
((options & PCRE2_NEVER_UCP) != 0)? " never_ucp" : "",
((options & PCRE2_NEVER_UTF) != 0)? " never_utf" : "",
((options & PCRE2_NO_AUTO_CAPTURE) != 0)? " no_auto_capture" : "",
((options & PCRE2_NO_AUTO_POSSESS) != 0)? " no_auto_possess" : "",
((options & PCRE2_NO_DOTSTAR_ANCHOR) != 0)? " no_dotstar_anchor" : "",
((options & PCRE2_NO_UTF_CHECK) != 0)? " no_utf_check" : "",
((options & PCRE2_NO_START_OPTIMIZE) != 0)? " no_start_optimize" : "",
((options & PCRE2_UCP) != 0)? " ucp" : "",
((options & PCRE2_UNGREEDY) != 0)? " ungreedy" : "",
((options & PCRE2_USE_OFFSET_LIMIT) != 0)? " use_offset_limit" : "",
((options & PCRE2_UTF) != 0)? " utf" : "",
after);
}
/*************************************************
* Show compile extra options *
*************************************************/
/* Called from show_pattern_info() and for unsupported POSIX options.
Arguments:
options an options word
before text to print before
after text to print after
Returns: nothing
*/
static void
show_compile_extra_options(uint32_t options, const char *before,
const char *after)
{
if (options == 0) fprintf(outfile, "%s %s", before, after);
else fprintf(outfile, "%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s",
before,
((options & PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK) != 0) ? " allow_lookaround_bsk" : "",
((options & PCRE2_EXTRA_ALLOW_SURROGATE_ESCAPES) != 0)? " allow_surrogate_escapes" : "",
((options & PCRE2_EXTRA_ALT_BSUX) != 0)? " alt_bsux" : "",
((options & PCRE2_EXTRA_ASCII_BSD) != 0)? " ascii_bsd" : "",
((options & PCRE2_EXTRA_ASCII_BSS) != 0)? " ascii_bss" : "",
((options & PCRE2_EXTRA_ASCII_BSW) != 0)? " ascii_bsw" : "",
((options & PCRE2_EXTRA_ASCII_DIGIT) != 0)? " ascii_digit" : "",
((options & PCRE2_EXTRA_ASCII_POSIX) != 0)? " ascii_posix" : "",
((options & PCRE2_EXTRA_BAD_ESCAPE_IS_LITERAL) != 0)? " bad_escape_is_literal" : "",
((options & PCRE2_EXTRA_CASELESS_RESTRICT) != 0)? " caseless_restrict" : "",
((options & PCRE2_EXTRA_ESCAPED_CR_IS_LF) != 0)? " escaped_cr_is_lf" : "",
((options & PCRE2_EXTRA_MATCH_WORD) != 0)? " match_word" : "",
((options & PCRE2_EXTRA_MATCH_LINE) != 0)? " match_line" : "",
((options & PCRE2_EXTRA_NEVER_CALLOUT) != 0)? " never_callout" : "",
((options & PCRE2_EXTRA_NO_BS0) != 0)? " no_bs0" : "",
((options & PCRE2_EXTRA_PYTHON_OCTAL) != 0)? " python_octal" : "",
((options & PCRE2_EXTRA_TURKISH_CASING) != 0)? " turkish_casing" : "",
after);
}
/*************************************************
* Show optimization flags *
*************************************************/
/*
Arguments:
flags an options word
before text to print before
after text to print after
Returns: nothing
*/
static void
show_optimize_flags(uint32_t flags, const char *before, const char *after)
{
if (flags == 0) fprintf(outfile, "%s%s", before, after);
else fprintf(outfile, "%s%s%s%s%s%s%s",
before,
((flags & PCRE2_OPTIM_AUTO_POSSESS) != 0) ? "auto_possess" : "",
((flags & PCRE2_OPTIM_AUTO_POSSESS) != 0 && (flags >> 1) != 0) ? "," : "",
((flags & PCRE2_OPTIM_DOTSTAR_ANCHOR) != 0) ? "dotstar_anchor" : "",
((flags & PCRE2_OPTIM_DOTSTAR_ANCHOR) != 0 && (flags >> 2) != 0) ? "," : "",
((flags & PCRE2_OPTIM_START_OPTIMIZE) != 0) ? "start_optimize" : "",
after);
}
#ifdef SUPPORT_PCRE2_8
/*************************************************
* Show match options *
*************************************************/
/* Called for unsupported POSIX options. */
static void
show_match_options(uint32_t options)
{
fprintf(outfile, "%s%s%s%s%s%s%s%s%s%s%s%s%s%s",
((options & PCRE2_ANCHORED) != 0)? " anchored" : "",
((options & PCRE2_COPY_MATCHED_SUBJECT) != 0)? " copy_matched_subject" : "",
((options & PCRE2_DFA_RESTART) != 0)? " dfa_restart" : "",
((options & PCRE2_DFA_SHORTEST) != 0)? " dfa_shortest" : "",
((options & PCRE2_DISABLE_RECURSELOOP_CHECK) != 0)? " disable_recurseloop_check" : "",
((options & PCRE2_ENDANCHORED) != 0)? " endanchored" : "",
((options & PCRE2_NO_JIT) != 0)? " no_jit" : "",
((options & PCRE2_NO_UTF_CHECK) != 0)? " no_utf_check" : "",
((options & PCRE2_NOTBOL) != 0)? " notbol" : "",
((options & PCRE2_NOTEMPTY) != 0)? " notempty" : "",
((options & PCRE2_NOTEMPTY_ATSTART) != 0)? " notempty_atstart" : "",
((options & PCRE2_NOTEOL) != 0)? " noteol" : "",
((options & PCRE2_PARTIAL_HARD) != 0)? " partial_hard" : "",
((options & PCRE2_PARTIAL_SOFT) != 0)? " partial_soft" : "");
}
#endif /* SUPPORT_PCRE2_8 */
/*************************************************
* Show memory usage info for a pattern *
*************************************************/
static void
show_memory_info(void)
{
uint32_t name_count, name_entry_size;
PCRE2_SIZE size, cblock_size, data_size;
/* One of the test_mode values will always be true, but to stop a compiler
warning we must initialize cblock_size. */
cblock_size = 0;
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE) cblock_size = sizeof(pcre2_real_code_8);
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE) cblock_size = sizeof(pcre2_real_code_16);
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE) cblock_size = sizeof(pcre2_real_code_32);
#endif
(void)pattern_info(PCRE2_INFO_SIZE, &size, FALSE);
(void)pattern_info(PCRE2_INFO_NAMECOUNT, &name_count, FALSE);
(void)pattern_info(PCRE2_INFO_NAMEENTRYSIZE, &name_entry_size, FALSE);
/* The uint32_t variables are cast before multiplying to avoid potential
integer overflow. */
data_size = (PCRE2_SIZE)name_count * (PCRE2_SIZE)name_entry_size * (PCRE2_SIZE)code_unit_size;
fprintf(outfile, "Memory allocation - code size : %" SIZ_FORM "\n", size -
cblock_size - data_size);
if (data_size != 0)
fprintf(outfile, "Memory allocation - data size : %" SIZ_FORM "\n", data_size);
if (pat_patctl.jit != 0)
{
(void)pattern_info(PCRE2_INFO_JITSIZE, &size, FALSE);
fprintf(outfile, "Memory allocation - JIT code : %" SIZ_FORM "\n", size);
}
}
/*************************************************
* Show frame size info for a pattern *
*************************************************/
static void
show_framesize(void)
{
PCRE2_SIZE frame_size;
(void)pattern_info(PCRE2_INFO_FRAMESIZE, &frame_size, FALSE);
fprintf(outfile, "Frame size for pcre2_match(): %" SIZ_FORM "\n", frame_size);
}
/*************************************************
* Show heapframes size info for a match_data *
*************************************************/
static void
show_heapframes_size(void)
{
PCRE2_SIZE heapframes_size;
PCRE2_GET_MATCH_DATA_HEAPFRAMES_SIZE(heapframes_size, match_data);
fprintf(outfile, "Heapframes size in match_data: %" SIZ_FORM "\n",
heapframes_size);
}
/*************************************************
* Get and output an error message *
*************************************************/
static BOOL
print_error_message(int errorcode, const char *before, const char *after)
{
int len;
PCRE2_GET_ERROR_MESSAGE(len, errorcode, pbuffer);
if (len < 0)
{
fprintf(outfile, "\n** pcre2test internal error: cannot interpret error "
"number\n** Unexpected return (%d) from pcre2_get_error_message()\n", len);
}
else
{
fprintf(outfile, "%s", before);
PCHARSV(CASTVAR(void *, pbuffer), 0, len, FALSE, outfile);
fprintf(outfile, "%s", after);
}
return len >= 0;
}
/*************************************************
* Callback function for callout enumeration *
*************************************************/
/* The only differences in the callout emumeration block for different code
unit widths are that the pointers to the subject, the most recent MARK, and a
callout argument string point to strings of the appropriate width. Casts can be
used to deal with this.
Argument:
cb pointer to enumerate block
callout_data user data
Returns: 0
*/
static int callout_callback(pcre2_callout_enumerate_block_8 *cb,
void *callout_data)
{
uint32_t i;
BOOL utf = (FLD(compiled_code, overall_options) & PCRE2_UTF) != 0;
(void)callout_data; /* Not currently displayed */
fprintf(outfile, "Callout ");
if (cb->callout_string != NULL)
{
uint32_t delimiter = CODE_UNIT(cb->callout_string, -1);
fprintf(outfile, "%c", delimiter);
PCHARSV(cb->callout_string, 0,
cb->callout_string_length, utf, outfile);
for (i = 0; callout_start_delims[i] != 0; i++)
if (delimiter == callout_start_delims[i])
{
delimiter = callout_end_delims[i];
break;
}
fprintf(outfile, "%c ", delimiter);
}
else fprintf(outfile, "%d ", cb->callout_number);
fprintf(outfile, "%.*s\n",
(int)((cb->next_item_length == 0)? 1 : cb->next_item_length),
pbuffer8 + cb->pattern_position);
return 0;
}
/*************************************************
* Show information about a pattern *
*************************************************/
/* This function is called after a pattern has been compiled if any of the
information-requesting controls have been set.
Arguments: none
Returns: PR_OK continue processing next line
PR_SKIP skip to a blank line
PR_ABEND abort the pcre2test run
*/
static int
show_pattern_info(void)
{
uint32_t compile_options, overall_options, extra_options;
BOOL utf = (FLD(compiled_code, overall_options) & PCRE2_UTF) != 0;
if ((pat_patctl.control & (CTL_BINCODE|CTL_FULLBINCODE)) != 0)
{
fprintf(outfile, "------------------------------------------------------------------\n");
PCRE2_PRINTINT((pat_patctl.control & CTL_FULLBINCODE) != 0);
}
if ((pat_patctl.control & CTL_INFO) != 0)
{
int rc;
void *nametable;
uint8_t *start_bits;
BOOL heap_limit_set, match_limit_set, depth_limit_set;
uint32_t backrefmax, bsr_convention, capture_count, first_ctype, first_cunit,
hasbackslashc, hascrorlf, jchanged, last_ctype, last_cunit, match_empty,
depth_limit, heap_limit, match_limit, minlength, nameentrysize, namecount,
newline_convention;
/* Exercise the error route. */
PCRE2_PATTERN_INFO(rc, compiled_code, 999, NULL);
(void)rc;
/* These info requests may return PCRE2_ERROR_UNSET. */
switch(pattern_info(PCRE2_INFO_HEAPLIMIT, &heap_limit, TRUE))
{
case 0:
heap_limit_set = TRUE;
break;
case PCRE2_ERROR_UNSET:
heap_limit_set = FALSE;
break;
default:
return PR_ABEND;
}
switch(pattern_info(PCRE2_INFO_MATCHLIMIT, &match_limit, TRUE))
{
case 0:
match_limit_set = TRUE;
break;
case PCRE2_ERROR_UNSET:
match_limit_set = FALSE;
break;
default:
return PR_ABEND;
}
switch(pattern_info(PCRE2_INFO_DEPTHLIMIT, &depth_limit, TRUE))
{
case 0:
depth_limit_set = TRUE;
break;
case PCRE2_ERROR_UNSET:
depth_limit_set = FALSE;
break;
default:
return PR_ABEND;
}
/* These info requests should always succeed. */
if (pattern_info(PCRE2_INFO_BACKREFMAX, &backrefmax, FALSE) +
pattern_info(PCRE2_INFO_BSR, &bsr_convention, FALSE) +
pattern_info(PCRE2_INFO_CAPTURECOUNT, &capture_count, FALSE) +
pattern_info(PCRE2_INFO_FIRSTBITMAP, &start_bits, FALSE) +
pattern_info(PCRE2_INFO_FIRSTCODEUNIT, &first_cunit, FALSE) +
pattern_info(PCRE2_INFO_FIRSTCODETYPE, &first_ctype, FALSE) +
pattern_info(PCRE2_INFO_HASBACKSLASHC, &hasbackslashc, FALSE) +
pattern_info(PCRE2_INFO_HASCRORLF, &hascrorlf, FALSE) +
pattern_info(PCRE2_INFO_JCHANGED, &jchanged, FALSE) +
pattern_info(PCRE2_INFO_LASTCODEUNIT, &last_cunit, FALSE) +
pattern_info(PCRE2_INFO_LASTCODETYPE, &last_ctype, FALSE) +
pattern_info(PCRE2_INFO_MATCHEMPTY, &match_empty, FALSE) +
pattern_info(PCRE2_INFO_MINLENGTH, &minlength, FALSE) +
pattern_info(PCRE2_INFO_NAMECOUNT, &namecount, FALSE) +
pattern_info(PCRE2_INFO_NAMEENTRYSIZE, &nameentrysize, FALSE) +
pattern_info(PCRE2_INFO_NAMETABLE, &nametable, FALSE) +
pattern_info(PCRE2_INFO_NEWLINE, &newline_convention, FALSE)
!= 0)
return PR_ABEND;
fprintf(outfile, "Capture group count = %d\n", capture_count);
if (backrefmax > 0)
fprintf(outfile, "Max back reference = %d\n", backrefmax);
if (maxlookbehind > 0)
fprintf(outfile, "Max lookbehind = %d\n", maxlookbehind);
if (heap_limit_set)
fprintf(outfile, "Heap limit = %u\n", heap_limit);
if (match_limit_set)
fprintf(outfile, "Match limit = %u\n", match_limit);
if (depth_limit_set)
fprintf(outfile, "Depth limit = %u\n", depth_limit);
if (namecount > 0)
{
fprintf(outfile, "Named capture groups:\n");
for (; namecount > 0; namecount--)
{
int imm2_size = test_mode == PCRE8_MODE ? 2 : 1;
uint32_t length = (uint32_t)STRLEN(nametable + imm2_size);
fprintf(outfile, " ");
/* In UTF mode the name may be a UTF string containing non-ASCII
letters and digits. We must output it as a UTF-8 string. In non-UTF mode,
use the normal string printing functions, which use escapes for all
non-ASCII characters. */
if (utf)
{
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE)
{
PCRE2_SPTR32 nameptr = (PCRE2_SPTR32)nametable + imm2_size;
while (*nameptr != 0)
{
uint8_t u8buff[6];
int len = ord2utf8(*nameptr++, u8buff);
fprintf(outfile, "%.*s", len, u8buff);
}
}
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE)
{
PCRE2_SPTR16 nameptr = (PCRE2_SPTR16)nametable + imm2_size;
while (*nameptr != 0)
{
int len;
uint8_t u8buff[6];
uint32_t c = *nameptr++ & 0xffff;
if (c >= 0xD800 && c < 0xDC00)
c = ((c & 0x3ff) << 10) + (*nameptr++ & 0x3ff) + 0x10000;
len = ord2utf8(c, u8buff);
fprintf(outfile, "%.*s", len, u8buff);
}
}
#endif
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE)
fprintf(outfile, "%s", (PCRE2_SPTR8)nametable + imm2_size);
#endif
}
else /* Not UTF mode */
{
PCHARSV(nametable, imm2_size, length, FALSE, outfile);
}
while (length++ < nameentrysize - imm2_size) putc(' ', outfile);
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE)
fprintf(outfile, "%3d\n", (int)(((PCRE2_SPTR32)nametable)[0]));
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE)
fprintf(outfile, "%3d\n", (int)(((PCRE2_SPTR16)nametable)[0]));
#endif
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE)
fprintf(outfile, "%3d\n", (int)(
((((PCRE2_SPTR8)nametable)[0]) << 8) | ((PCRE2_SPTR8)nametable)[1]));
#endif
nametable = (void *)((PCRE2_UCHAR8 *)nametable + nameentrysize * code_unit_size);
}
}
if (hascrorlf) fprintf(outfile, "Contains explicit CR or LF match\n");
if (hasbackslashc) fprintf(outfile, "Contains \\C\n");
if (match_empty) fprintf(outfile, "May match empty string\n");
pattern_info(PCRE2_INFO_ARGOPTIONS, &compile_options, FALSE);
pattern_info(PCRE2_INFO_ALLOPTIONS, &overall_options, FALSE);
pattern_info(PCRE2_INFO_EXTRAOPTIONS, &extra_options, FALSE);
/* Remove UTF/UCP if they were there only because of forbid_utf. This saves
cluttering up the verification output of non-UTF test files. */
if ((pat_patctl.options & PCRE2_NEVER_UTF) == 0)
{
compile_options &= ~PCRE2_NEVER_UTF;
overall_options &= ~PCRE2_NEVER_UTF;
}
if ((pat_patctl.options & PCRE2_NEVER_UCP) == 0)
{
compile_options &= ~PCRE2_NEVER_UCP;
overall_options &= ~PCRE2_NEVER_UCP;
}
if ((compile_options|overall_options) != 0)
{
if (compile_options == overall_options)
show_compile_options(compile_options, "Options:", "\n");
else
{
show_compile_options(compile_options, "Compile options:", "\n");
show_compile_options(overall_options, "Overall options:", "\n");
}
}
if (extra_options != 0)
show_compile_extra_options(extra_options, "Extra options:", "\n");
if (FLD(compiled_code, optimization_flags) != PCRE2_OPTIMIZATION_ALL)
show_optimize_flags(FLD(compiled_code, optimization_flags), "Optimizations: ", "\n");
if (jchanged) fprintf(outfile, "Duplicate name status changes\n");
if ((pat_patctl.control2 & CTL2_BSR_SET) != 0 ||
(FLD(compiled_code, flags) & PCRE2_BSR_SET) != 0)
fprintf(outfile, "\\R matches %s\n", (bsr_convention == PCRE2_BSR_UNICODE)?
"any Unicode newline" : "CR, LF, or CRLF");
if ((FLD(compiled_code, flags) & PCRE2_NL_SET) != 0)
{
switch (newline_convention)
{
case PCRE2_NEWLINE_CR:
fprintf(outfile, "Forced newline is CR\n");
break;
case PCRE2_NEWLINE_LF:
fprintf(outfile, "Forced newline is LF\n");
break;
case PCRE2_NEWLINE_CRLF:
fprintf(outfile, "Forced newline is CRLF\n");
break;
case PCRE2_NEWLINE_ANYCRLF:
fprintf(outfile, "Forced newline is CR, LF, or CRLF\n");
break;
case PCRE2_NEWLINE_ANY:
fprintf(outfile, "Forced newline is any Unicode newline\n");
break;
case PCRE2_NEWLINE_NUL:
fprintf(outfile, "Forced newline is NUL\n");
break;
default:
break;
}
}
if (first_ctype == 2)
{
fprintf(outfile, "First code unit at start or follows newline\n");
}
else if (first_ctype == 1)
{
const char *caseless =
((FLD(compiled_code, flags) & PCRE2_FIRSTCASELESS) == 0)?
"" : " (caseless)";
if (PRINTOK(first_cunit))
fprintf(outfile, "First code unit = \'%c\'%s\n", first_cunit, caseless);
else
{
fprintf(outfile, "First code unit = ");
pchar(first_cunit, FALSE, outfile);
fprintf(outfile, "%s\n", caseless);
}
}
else if (start_bits != NULL)
{
int i;
int c = 24;
fprintf(outfile, "Starting code units:");
for (i = 0; i < 256; i++)
{
if ((start_bits[i/8] & (1u << (i&7))) != 0)
{
if (c > 75)
{
fprintf(outfile, "\n ");
c = 2;
}
if (PRINTOK(i) && i != ' ')
{
fprintf(outfile, " %c", i);
c += 2;
}
else
{
fprintf(outfile, " \\x%02x", i);
c += 5;
}
}
}
fprintf(outfile, "\n");
}
if (last_ctype != 0)
{
const char *caseless =
((FLD(compiled_code, flags) & PCRE2_LASTCASELESS) == 0)?
"" : " (caseless)";
if (PRINTOK(last_cunit))
fprintf(outfile, "Last code unit = \'%c\'%s\n", last_cunit, caseless);
else
{
fprintf(outfile, "Last code unit = ");
pchar(last_cunit, FALSE, outfile);
fprintf(outfile, "%s\n", caseless);
}
}
if ((FLD(compiled_code, optimization_flags) & PCRE2_OPTIM_START_OPTIMIZE) != 0)
fprintf(outfile, "Subject length lower bound = %d\n", minlength);
if (pat_patctl.jit != 0 && (pat_patctl.control & CTL_JITVERIFY) != 0)
{
#ifdef SUPPORT_JIT
if (FLD(compiled_code, executable_jit) != NULL)
fprintf(outfile, "JIT compilation was successful\n");
else
{
fprintf(outfile, "JIT compilation was not successful");
if (jitrc != 0 && !print_error_message(jitrc, " (", ")"))
return PR_ABEND;
fprintf(outfile, "\n");
}
#else
fprintf(outfile, "JIT support is not available in this version of PCRE2\n");
#endif
}
}
if ((pat_patctl.control & CTL_CALLOUT_INFO) != 0)
{
int errorcode;
PCRE2_CALLOUT_ENUMERATE(errorcode, callout_callback, 0);
if (errorcode != 0)
{
fprintf(outfile, "Callout enumerate failed: error %d: ", errorcode);
if (errorcode < 0 && !print_error_message(errorcode, "", "\n"))
return PR_ABEND;
return PR_SKIP;
}
}
return PR_OK;
}
/*************************************************
* Handle serialization error *
*************************************************/
/* Print an error message after a serialization failure.
Arguments:
rc the error code
msg an initial message for what failed
Returns: FALSE if print_error_message() fails
*/
static BOOL
serial_error(int rc, const char *msg)
{
fprintf(outfile, "%s failed: error %d: ", msg, rc);
return print_error_message(rc, "", "\n");
}
/*************************************************
* Open file for save/load commands *
*************************************************/
/* This function decodes the file name and opens the file.
Arguments:
buffptr point after the #command
mode open mode
fptr points to the FILE variable
name name of # command
Returns: PR_OK or PR_ABEND
*/
static int
open_file(uint8_t *buffptr, const char *mode, FILE **fptr, const char *name)
{
char *endf;
char *filename = (char *)buffptr;
while (isspace(*filename)) filename++;
endf = filename + strlen8(filename);
while (endf > filename && isspace(endf[-1])) endf--;
if (endf == filename)
{
fprintf(outfile, "** File name expected after %s\n", name);
return PR_ABEND;
}
*endf = 0;
*fptr = fopen((const char *)filename, mode);
if (*fptr == NULL)
{
fprintf(outfile, "** Failed to open \"%s\": %s\n", filename, strerror(errno));
return PR_ABEND;
}
return PR_OK;
}
/*************************************************
* Process command line *
*************************************************/
/* This function is called for lines beginning with # and a character that is
not ! or whitespace, when encountered between tests, which means that there is
no compiled pattern (compiled_code is NULL). The line is in buffer.
Arguments: none
Returns: PR_OK continue processing next line
PR_SKIP skip to a blank line
PR_ABEND abort the pcre2test run
*/
static int
process_command(void)
{
FILE *f;
PCRE2_SIZE serial_size;
size_t i;
int rc, cmd, cmdlen, yield;
uint16_t first_listed_newline;
const char *cmdname;
uint8_t *argptr, *serial;
yield = PR_OK;
cmd = CMD_UNKNOWN;
cmdlen = 0;
for (i = 0; i < cmdlistcount; i++)
{
cmdname = cmdlist[i].name;
cmdlen = strlen(cmdname);
if (strncmp((char *)(buffer+1), cmdname, cmdlen) == 0 &&
isspace(buffer[cmdlen+1]))
{
cmd = cmdlist[i].value;
break;
}
}
argptr = buffer + cmdlen + 1;
if (restrict_for_perl_test && cmd != CMD_PATTERN && cmd != CMD_SUBJECT)
{
fprintf(outfile, "** #%s is not allowed after #perltest\n", cmdname);
return PR_ABEND;
}
switch(cmd)
{
case CMD_UNKNOWN:
fprintf(outfile, "** Unknown command: %s", buffer);
break;
case CMD_FORBID_UTF:
forbid_utf = PCRE2_NEVER_UTF|PCRE2_NEVER_UCP;
break;
case CMD_PERLTEST:
restrict_for_perl_test = TRUE;
break;
/* Set default pattern modifiers */
case CMD_PATTERN:
(void)decode_modifiers(argptr, CTX_DEFPAT, &def_patctl, NULL);
if (def_patctl.jit == 0 && (def_patctl.control & CTL_JITVERIFY) != 0)
def_patctl.jit = JIT_DEFAULT;
break;
/* Set default subject modifiers */
case CMD_SUBJECT:
(void)decode_modifiers(argptr, CTX_DEFDAT, NULL, &def_datctl);
break;
/* Check the default newline, and if not one of those listed, set up the
first one to be forced. An empty list unsets. */
case CMD_NEWLINE_DEFAULT:
local_newline_default = 0; /* Unset */
first_listed_newline = 0;
for (;;)
{
while (isspace(*argptr)) argptr++;
if (*argptr == 0) break;
for (uint16_t j = 1; j < sizeof(newlines)/sizeof(char *); j++)
{
size_t nlen = strlen(newlines[j]);
if (strncmpic(argptr, (const uint8_t *)newlines[j], nlen) == 0 &&
isspace(argptr[nlen]))
{
if (j == NEWLINE_DEFAULT) return PR_OK; /* Default is valid */
if (first_listed_newline == 0) first_listed_newline = j;
}
}
while (*argptr != 0 && !isspace(*argptr)) argptr++;
}
local_newline_default = first_listed_newline;
break;
/* Pop or copy a compiled pattern off the stack. Modifiers that do not affect
the compiled pattern (e.g. to give information) are permitted. The default
pattern modifiers are ignored. */
case CMD_POP:
case CMD_POPCOPY:
if (patstacknext <= 0)
{
fprintf(outfile, "** Can't pop off an empty stack\n");
return PR_SKIP;
}
memset(&pat_patctl, 0, sizeof(patctl)); /* Completely unset */
if (!decode_modifiers(argptr, CTX_POPPAT, &pat_patctl, NULL))
return PR_SKIP;
if (cmd == CMD_POP)
{
SET(compiled_code, patstack[--patstacknext]);
}
else
{
PCRE2_CODE_COPY_FROM_VOID(compiled_code, patstack[patstacknext - 1]);
}
if (pat_patctl.jit != 0)
{
PCRE2_JIT_COMPILE(jitrc, compiled_code, pat_patctl.jit);
}
if ((pat_patctl.control & CTL_MEMORY) != 0) show_memory_info();
if ((pat_patctl.control2 & CTL2_FRAMESIZE) != 0) show_framesize();
if ((pat_patctl.control & CTL_ANYINFO) != 0)
{
rc = show_pattern_info();
if (rc != PR_OK) return rc;
}
break;
/* Save the stack of compiled patterns to a file, then empty the stack. */
case CMD_SAVE:
if (patstacknext <= 0)
{
fprintf(outfile, "** No stacked patterns to save\n");
return PR_OK;
}
rc = open_file(argptr+1, BINARY_OUTPUT_MODE, &f, "#save");
if (rc != PR_OK) return rc;
PCRE2_SERIALIZE_ENCODE(rc, patstack, patstacknext, &serial, &serial_size,
general_context);
if (rc < 0)
{
fclose(f);
if (!serial_error(rc, "Serialization")) return PR_ABEND;
break;
}
/* Write the length at the start of the file to make it straightforward to
get the right memory when re-loading. This saves having to read the file size
in different operating systems. To allow for different endianness (even
though reloading with the opposite endianness does not work), write the
length byte-by-byte. */
for (i = 0; i < 4; i++) fputc((serial_size >> (i*8)) & 255, f);
if (fwrite(serial, 1, serial_size, f) != serial_size)
{
fprintf(outfile, "** Wrong return from fwrite()\n");
fclose(f);
return PR_ABEND;
}
fclose(f);
PCRE2_SERIALIZE_FREE(serial);
while(patstacknext > 0)
{
SET(compiled_code, patstack[--patstacknext]);
SUB1(pcre2_code_free, compiled_code);
}
SET(compiled_code, NULL);
break;
/* Load a set of compiled patterns from a file onto the stack */
case CMD_LOAD:
rc = open_file(argptr+1, BINARY_INPUT_MODE, &f, "#load");
if (rc != PR_OK) return rc;
serial_size = 0;
for (i = 0; i < 4; i++) serial_size |= fgetc(f) << (i*8);
serial = malloc(serial_size);
if (serial == NULL)
{
fprintf(outfile, "** Failed to get memory (size %" SIZ_FORM ") for #load\n",
serial_size);
fclose(f);
return PR_ABEND;
}
i = fread(serial, 1, serial_size, f);
fclose(f);
if (i != serial_size)
{
fprintf(outfile, "** Wrong return from fread()\n");
yield = PR_ABEND;
}
else
{
PCRE2_SERIALIZE_GET_NUMBER_OF_CODES(rc, serial);
if (rc < 0)
{
if (!serial_error(rc, "Get number of codes")) yield = PR_ABEND;
}
else
{
if (rc + patstacknext > PATSTACKSIZE)
{
fprintf(outfile, "** Not enough space on pattern stack for %d pattern%s\n",
rc, (rc == 1)? "" : "s");
rc = PATSTACKSIZE - patstacknext;
fprintf(outfile, "** Decoding %d pattern%s\n", rc,
(rc == 1)? "" : "s");
}
PCRE2_SERIALIZE_DECODE(rc, patstack + patstacknext, rc, serial,
general_context);
if (rc < 0)
{
if (!serial_error(rc, "Deserialization")) yield = PR_ABEND;
}
else patstacknext += rc;
}
}
free(serial);
break;
/* Load a set of binary tables into tables3. */
case CMD_LOADTABLES:
rc = open_file(argptr+1, BINARY_INPUT_MODE, &f, "#loadtables");
if (rc != PR_OK) return rc;
if (tables3 == NULL)
{
(void)PCRE2_CONFIG(PCRE2_CONFIG_TABLES_LENGTH, &loadtables_length);
tables3 = malloc(loadtables_length);
}
if (tables3 == NULL)
{
fprintf(outfile, "** Failed: malloc failed for #loadtables\n");
yield = PR_ABEND;
}
else if (fread(tables3, 1, loadtables_length, f) != loadtables_length)
{
fprintf(outfile, "** Wrong return from fread()\n");
yield = PR_ABEND;
}
fclose(f);
break;
}
return yield;
}
/*************************************************
* Process pattern line *
*************************************************/
/* This function is called when the input buffer contains the start of a
pattern. The first character is known to be a valid delimiter. The pattern is
read, modifiers are interpreted, and a suitable local context is set up for
this test. The pattern is then compiled.
Arguments: none
Returns: PR_OK continue processing next line
PR_SKIP skip to a blank line
PR_ABEND abort the pcre2test run
*/
static int
process_pattern(void)
{
BOOL utf;
uint32_t k;
uint8_t *p = buffer;
unsigned int delimiter = *p++;
int errorcode;
void *use_pat_context;
void *use_pbuffer = NULL;
uint32_t use_forbid_utf = forbid_utf;
PCRE2_SIZE patlen;
PCRE2_SIZE valgrind_access_length;
PCRE2_SIZE erroroffset;
/* The perltest.sh script supports only / as a delimiter. */
if (restrict_for_perl_test && delimiter != '/')
{
fprintf(outfile, "** The only allowed delimiter after #perltest is '/'\n");
return PR_ABEND;
}
/* Initialize the context and pattern/data controls for this test from the
defaults. */
PATCTXCPY(pat_context, default_pat_context);
memcpy(&pat_patctl, &def_patctl, sizeof(patctl));
/* Find the end of the pattern, reading more lines if necessary. */
for(;;)
{
while (*p != 0)
{
if (*p == '\\' && p[1] != 0) p++;
else if (*p == delimiter) break;
p++;
}
if (*p != 0) break;
if ((p = extend_inputline(infile, p, " > ")) == NULL)
{
fprintf(outfile, "** Unexpected EOF\n");
return PR_ABEND;
}
if (!INTERACTIVE(infile)) fprintf(outfile, "%s", (char *)p);
}
/* If the first character after the delimiter is backslash, make the pattern
end with backslash. This is purely to provide a way of testing for the error
message when a pattern ends with backslash. */
if (p[1] == '\\') *p++ = '\\';
/* Terminate the pattern at the delimiter, and compute the length. */
*p++ = 0;
patlen = p - buffer - 2;
/* Look for modifiers and options after the final delimiter. */
if (!decode_modifiers(p, CTX_PAT, &pat_patctl, NULL)) return PR_SKIP;
/* Note that the match_invalid_utf option also sets utf when passed to
pcre2_compile(). */
utf = (pat_patctl.options & (PCRE2_UTF|PCRE2_MATCH_INVALID_UTF)) != 0;
/* The utf8_input modifier is not allowed in 8-bit mode, and is mutually
exclusive with the utf modifier. */
if ((pat_patctl.control & CTL_UTF8_INPUT) != 0)
{
if (test_mode == PCRE8_MODE)
{
fprintf(outfile, "** The utf8_input modifier is not allowed in 8-bit mode\n");
return PR_SKIP;
}
if (utf)
{
fprintf(outfile, "** The utf and utf8_input modifiers are mutually exclusive\n");
return PR_SKIP;
}
}
/* The convert and posix modifiers are mutually exclusive. */
if (pat_patctl.convert_type != CONVERT_UNSET &&
(pat_patctl.control & CTL_POSIX) != 0)
{
fprintf(outfile, "** The convert and posix modifiers are mutually exclusive\n");
return PR_SKIP;
}
/* Check for mutually exclusive control modifiers. At present, these are all in
the first control word. */
for (k = 0; k < sizeof(exclusive_pat_controls)/sizeof(uint32_t); k++)
{
uint32_t c = pat_patctl.control & exclusive_pat_controls[k];
if (c != 0 && c != (c & (~c+1)))
{
show_controls(c, 0, "** Not allowed together:");
fprintf(outfile, "\n");
return PR_SKIP;
}
}
/* Assume full JIT compile for jitverify and/or jitfast if nothing else was
specified. */
if (pat_patctl.jit == 0 &&
(pat_patctl.control & (CTL_JITVERIFY|CTL_JITFAST)) != 0)
pat_patctl.jit = JIT_DEFAULT;
/* Now copy the pattern to pbuffer8 for use in 8-bit testing and for reflecting
in callouts. Convert from hex if requested (literal strings in quotes may be
present within the hexadecimal pairs). The result must necessarily be fewer
characters so will always fit in pbuffer8. */
if ((pat_patctl.control & CTL_HEXPAT) != 0)
{
uint8_t *pp, *pt;
uint32_t c, d;
pt = pbuffer8;
for (pp = buffer + 1; *pp != 0; pp++)
{
if (isspace(*pp)) continue;
c = *pp++;
/* Handle a literal substring */
if (c == '\'' || c == '"')
{
uint8_t *pq = pp;
for (;; pp++)
{
d = *pp;
if (d == 0)
{
fprintf(outfile, "** Missing closing quote in hex pattern: "
"opening quote is at offset %" PTR_FORM ".\n", pq - buffer - 2);
return PR_SKIP;
}
if (d == c) break;
*pt++ = d;
}
}
/* Expect a hex pair */
else
{
if (!isxdigit(c))
{
fprintf(outfile, "** Unexpected non-hex-digit '%c' at offset %"
PTR_FORM " in hex pattern: quote missing?\n", c, pp - buffer - 2);
return PR_SKIP;
}
if (*pp == 0)
{
fprintf(outfile, "** Odd number of digits in hex pattern\n");
return PR_SKIP;
}
d = *pp;
if (!isxdigit(d))
{
fprintf(outfile, "** Unexpected non-hex-digit '%c' at offset %"
PTR_FORM " in hex pattern: quote missing?\n", d, pp - buffer - 1);
return PR_SKIP;
}
c = toupper(c);
d = toupper(d);
*pt++ = ((isdigit(c)? (c - '0') : (c - 'A' + 10)) << 4) +
(isdigit(d)? (d - '0') : (d - 'A' + 10));
}
}
*pt = 0;
patlen = pt - pbuffer8;
}
/* If not a hex string, process for repetition expansion if requested. */
else if ((pat_patctl.control & CTL_EXPAND) != 0)
{
uint8_t *pp, *pt;
pt = pbuffer8;
for (pp = buffer + 1; *pp != 0; pp++)
{
uint8_t *pc = pp;
uint32_t count = 1;
size_t length = 1;
/* Check for replication syntax; if not found, the defaults just set will
prevail and one character will be copied. */
if (pp[0] == '\\' && pp[1] == '[')
{
uint8_t *pe;
for (pe = pp + 2; *pe != 0; pe++)
{
if (pe[0] == ']' && pe[1] == '{')
{
uint32_t clen = pe - pc - 2;
uint32_t i = 0;
unsigned long uli;
char *endptr;
pe += 2;
uli = strtoul((const char *)pe, &endptr, 10);
if (U32OVERFLOW(uli))
{
fprintf(outfile, "** Pattern repeat count too large\n");
return PR_SKIP;
}
i = (uint32_t)uli;
pe = (uint8_t *)endptr;
if (*pe == '}')
{
if (i == 0)
{
fprintf(outfile, "** Zero repeat not allowed\n");
return PR_SKIP;
}
pc += 2;
count = i;
length = clen;
pp = pe;
break;
}
}
}
}
/* Add to output. If the buffer is too small expand it. The function for
expanding buffers always keeps buffer and pbuffer8 in step as far as their
size goes. */
while (pt + count * length > pbuffer8 + pbuffer8_size)
{
size_t pc_offset = pc - buffer;
size_t pp_offset = pp - buffer;
size_t pt_offset = pt - pbuffer8;
expand_input_buffers();
pc = buffer + pc_offset;
pp = buffer + pp_offset;
pt = pbuffer8 + pt_offset;
}
for (; count > 0; count--)
{
memcpy(pt, pc, length);
pt += length;
}
}
*pt = 0;
patlen = pt - pbuffer8;
if ((pat_patctl.control & CTL_INFO) != 0)
fprintf(outfile, "Expanded: %s\n", pbuffer8);
}
/* Neither hex nor expanded, just copy the input verbatim. */
else
{
strncpy((char *)pbuffer8, (char *)(buffer+1), patlen + 1);
}
/* Sort out character tables */
if (pat_patctl.locale[0] != 0)
{
if (pat_patctl.tables_id != 0)
{
fprintf(outfile, "** 'Locale' and 'tables' must not both be set\n");
return PR_SKIP;
}
if (setlocale(LC_CTYPE, (const char *)pat_patctl.locale) == NULL)
{
fprintf(outfile, "** Failed to set locale \"%s\"\n", pat_patctl.locale);
return PR_SKIP;
}
if (strcmp((const char *)pat_patctl.locale, (const char *)locale_name) != 0)
{
strcpy((char *)locale_name, (char *)pat_patctl.locale);
if (locale_tables != NULL)
{
PCRE2_MAKETABLES_FREE(general_context, (const void *)locale_tables);
}
PCRE2_MAKETABLES(locale_tables, general_context);
}
use_tables = locale_tables;
}
else switch (pat_patctl.tables_id)
{
case 0: use_tables = NULL; break;
case 1: use_tables = tables1; break;
case 2: use_tables = tables2; break;
case 3:
if (tables3 == NULL)
{
fprintf(outfile, "** 'Tables = 3' is invalid: binary tables have not "
"been loaded\n");
return PR_SKIP;
}
use_tables = tables3;
break;
default:
fprintf(outfile, "** 'Tables' must specify 0, 1, 2, or 3.\n");
return PR_SKIP;
}
PCRE2_SET_CHARACTER_TABLES(pat_context, use_tables);
/* Set up for the stackguard test. */
if (pat_patctl.stackguard_test != 0)
{
PCRE2_SET_COMPILE_RECURSION_GUARD(pat_context, stack_guard, NULL);
}
/* Handle compiling via the POSIX interface, which doesn't support the
timing, showing, or debugging options, nor the ability to pass over
local character tables. Neither does it have 16-bit or 32-bit support. */
if ((pat_patctl.control & CTL_POSIX) != 0)
{
#ifdef SUPPORT_PCRE2_8
int rc;
int cflags = 0;
const char *msg = "** Ignored with POSIX interface:";
#endif
if (test_mode != PCRE8_MODE)
{
fprintf(outfile, "** The POSIX interface is available only in 8-bit mode\n");
return PR_SKIP;
}
#ifdef SUPPORT_PCRE2_8
/* Check for features that the POSIX interface does not support. */
if (pat_patctl.locale[0] != 0) prmsg(&msg, "locale");
if (pat_patctl.replacement[0] != 0) prmsg(&msg, "replace");
if (pat_patctl.tables_id != 0) prmsg(&msg, "tables");
if (pat_patctl.stackguard_test != 0) prmsg(&msg, "stackguard");
if (timeit > 0) prmsg(&msg, "timing");
if (pat_patctl.jit != 0) prmsg(&msg, "JIT");
if ((pat_patctl.options & ~POSIX_SUPPORTED_COMPILE_OPTIONS) != 0)
{
show_compile_options(
pat_patctl.options & (uint32_t)(~POSIX_SUPPORTED_COMPILE_OPTIONS),
msg, "");
msg = "";
}
if ((FLD(pat_context, extra_options) &
(uint32_t)(~POSIX_SUPPORTED_COMPILE_EXTRA_OPTIONS)) != 0)
{
show_compile_extra_options(
FLD(pat_context, extra_options) &
(uint32_t)(~POSIX_SUPPORTED_COMPILE_EXTRA_OPTIONS), msg, "");
msg = "";
}
if ((pat_patctl.control & (uint32_t)(~POSIX_SUPPORTED_COMPILE_CONTROLS)) != 0 ||
(pat_patctl.control2 & (uint32_t)(~POSIX_SUPPORTED_COMPILE_CONTROLS2)) != 0)
{
show_controls(
pat_patctl.control & (uint32_t)(~POSIX_SUPPORTED_COMPILE_CONTROLS),
pat_patctl.control2 & (uint32_t)(~POSIX_SUPPORTED_COMPILE_CONTROLS2),
msg);
msg = "";
/* Remove ignored options so as not to get a repeated message for those
that are actually subject controls. */
pat_patctl.control &= (uint32_t)(POSIX_SUPPORTED_COMPILE_CONTROLS);
pat_patctl.control2 &= (uint32_t)(POSIX_SUPPORTED_COMPILE_CONTROLS2);
}
if (local_newline_default != 0) prmsg(&msg, "#newline_default");
if (FLD(pat_context, max_pattern_length) != PCRE2_UNSET)
prmsg(&msg, "max_pattern_length");
if (FLD(pat_context, max_pattern_compiled_length) != PCRE2_UNSET)
prmsg(&msg, "max_pattern_compiled_length");
if (FLD(pat_context, parens_nest_limit) != PARENS_NEST_DEFAULT)
prmsg(&msg, "parens_nest_limit");
if (msg[0] == 0) fprintf(outfile, "\n");
/* Translate PCRE2 options to POSIX options and then compile. */
if (utf) cflags |= REG_UTF;
if ((pat_patctl.control & CTL_POSIX_NOSUB) != 0) cflags |= REG_NOSUB;
if ((pat_patctl.options & PCRE2_UCP) != 0) cflags |= REG_UCP;
if ((pat_patctl.options & PCRE2_CASELESS) != 0) cflags |= REG_ICASE;
if ((pat_patctl.options & PCRE2_LITERAL) != 0) cflags |= REG_NOSPEC;
if ((pat_patctl.options & PCRE2_MULTILINE) != 0) cflags |= REG_NEWLINE;
if ((pat_patctl.options & PCRE2_DOTALL) != 0) cflags |= REG_DOTALL;
if ((pat_patctl.options & PCRE2_UNGREEDY) != 0) cflags |= REG_UNGREEDY;
if ((pat_patctl.control & (CTL_HEXPAT|CTL_USE_LENGTH)) != 0)
{
preg.re_endp = (char *)pbuffer8 + patlen;
cflags |= REG_PEND;
}
rc = regcomp(&preg, (char *)pbuffer8, cflags);
/* Compiling failed */
if (rc != 0)
{
size_t bsize, usize;
int psize;
preg.re_pcre2_code = NULL; /* In case something was left in there */
preg.re_match_data = NULL;
bsize = (pat_patctl.regerror_buffsize != 0)?
pat_patctl.regerror_buffsize : pbuffer8_size;
if (bsize + 8 < pbuffer8_size)
memcpy(pbuffer8 + bsize, "DEADBEEF", 8);
usize = regerror(rc, &preg, (char *)pbuffer8, bsize);
/* Inside regerror(), snprintf() is used. If the buffer is too small, some
versions of snprintf() put a zero byte at the end, but others do not.
Therefore, we print a maximum of one less than the size of the buffer. */
psize = (int)bsize - 1;
fprintf(outfile, "Failed: POSIX code %d: %.*s\n", rc, psize, pbuffer8);
if (usize > bsize)
{
fprintf(outfile, "** regerror() message truncated\n");
if (memcmp(pbuffer8 + bsize, "DEADBEEF", 8) != 0)
fprintf(outfile, "** regerror() buffer overflow\n");
}
return PR_SKIP;
}
/* Compiling succeeded. Check that the values in the preg block are sensible.
It can happen that pcre2test is accidentally linked with a different POSIX
library which succeeds, but of course puts different things into preg. In
this situation, calling regfree() may cause a segfault (or invalid free() in
valgrind), so ensure that preg.re_pcre2_code is NULL, which suppresses the
calling of regfree() on exit. */
if (preg.re_pcre2_code == NULL ||
((pcre2_real_code_8 *)preg.re_pcre2_code)->magic_number != MAGIC_NUMBER ||
((pcre2_real_code_8 *)preg.re_pcre2_code)->top_bracket != preg.re_nsub ||
preg.re_match_data == NULL ||
preg.re_cflags != cflags)
{
fprintf(outfile,
"** The regcomp() function returned zero (success), but the values set\n"
"** in the preg block are not valid for PCRE2. Check that pcre2test is\n"
"** linked with PCRE2's pcre2posix module (-lpcre2-posix) and not with\n"
"** some other POSIX regex library.\n**\n");
preg.re_pcre2_code = NULL;
return PR_ABEND;
}
return PR_OK;
#endif /* SUPPORT_PCRE2_8 */
}
/* Handle compiling via the native interface. Controls that act later are
ignored with "push". Replacements are locked out. */
if ((pat_patctl.control & (CTL_PUSH|CTL_PUSHCOPY|CTL_PUSHTABLESCOPY)) != 0)
{
if (pat_patctl.replacement[0] != 0)
{
fprintf(outfile, "** Replacement text is not supported with 'push'.\n");
return PR_OK;
}
if ((pat_patctl.control & ~PUSH_SUPPORTED_COMPILE_CONTROLS) != 0 ||
(pat_patctl.control2 & ~PUSH_SUPPORTED_COMPILE_CONTROLS2) != 0)
{
show_controls(pat_patctl.control & ~PUSH_SUPPORTED_COMPILE_CONTROLS,
pat_patctl.control2 & ~PUSH_SUPPORTED_COMPILE_CONTROLS2,
"** Ignored when compiled pattern is stacked with 'push':");
fprintf(outfile, "\n");
}
if ((pat_patctl.control & PUSH_COMPILE_ONLY_CONTROLS) != 0 ||
(pat_patctl.control2 & PUSH_COMPILE_ONLY_CONTROLS2) != 0)
{
show_controls(pat_patctl.control & PUSH_COMPILE_ONLY_CONTROLS,
pat_patctl.control2 & PUSH_COMPILE_ONLY_CONTROLS2,
"** Applies only to compile when pattern is stacked with 'push':");
fprintf(outfile, "\n");
}
}
/* Convert the input in non-8-bit modes. */
errorcode = 0;
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE) errorcode = to16(pbuffer8, utf, &patlen);
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE) errorcode = to32(pbuffer8, utf, &patlen);
#endif
switch(errorcode)
{
case -1:
fprintf(outfile, "** Failed: invalid UTF-8 string cannot be "
"converted to %d-bit string\n", (test_mode == PCRE16_MODE)? 16:32);
return PR_SKIP;
case -2:
fprintf(outfile, "** Failed: character value greater than 0x10ffff "
"cannot be converted to UTF\n");
return PR_SKIP;
case -3:
fprintf(outfile, "** Failed: character value greater than 0xffff "
"cannot be converted to 16-bit in non-UTF mode\n");
return PR_SKIP;
default:
break;
}
/* The pattern is now in pbuffer[8|16|32], with the length in code units in
patlen. If it is to be converted, copy the result back afterwards so that it
ends up back in the usual place. */
if (pat_patctl.convert_type != CONVERT_UNSET)
{
int rc;
int convert_return = PR_OK;
uint32_t convert_options = pat_patctl.convert_type;
void *converted_pattern;
PCRE2_SIZE converted_length;
if (pat_patctl.convert_length != 0)
{
converted_length = pat_patctl.convert_length;
converted_pattern = malloc(converted_length * code_unit_size);
if (converted_pattern == NULL)
{
fprintf(outfile, "** Failed: malloc failed for converted pattern\n");
return PR_SKIP;
}
}
else converted_pattern = NULL; /* Let the library allocate */
if (utf) convert_options |= PCRE2_CONVERT_UTF;
if ((pat_patctl.options & PCRE2_NO_UTF_CHECK) != 0)
convert_options |= PCRE2_CONVERT_NO_UTF_CHECK;
CONCTXCPY(con_context, default_con_context);
if (pat_patctl.convert_glob_escape != 0)
{
uint32_t escape = (pat_patctl.convert_glob_escape == '0')? 0 :
pat_patctl.convert_glob_escape;
PCRE2_SET_GLOB_ESCAPE(rc, con_context, escape);
if (rc != 0)
{
fprintf(outfile, "** Invalid glob escape '%c'\n",
pat_patctl.convert_glob_escape);
convert_return = PR_SKIP;
goto CONVERT_FINISH;
}
}
if (pat_patctl.convert_glob_separator != 0)
{
PCRE2_SET_GLOB_SEPARATOR(rc, con_context, pat_patctl.convert_glob_separator);
if (rc != 0)
{
fprintf(outfile, "** Invalid glob separator '%c'\n",
pat_patctl.convert_glob_separator);
convert_return = PR_SKIP;
goto CONVERT_FINISH;
}
}
PCRE2_PATTERN_CONVERT(rc, pbuffer, patlen, convert_options,
&converted_pattern, &converted_length, con_context);
if (rc != 0)
{
fprintf(outfile, "** Pattern conversion error at offset %" SIZ_FORM ": ",
converted_length);
convert_return = print_error_message(rc, "", "\n")? PR_SKIP:PR_ABEND;
}
/* Output the converted pattern, then copy it. */
else
{
BOOL toolong;
PCHARSV(converted_pattern, 0, converted_length, utf, outfile);
fprintf(outfile, "\n");
if (test_mode == PCRE8_MODE)
toolong = (converted_length + 1 > pbuffer8_size);
else if (test_mode == PCRE16_MODE)
toolong = (2*(converted_length + 1) > pbuffer8_size);
else /* 32-bit */
toolong = (4*(converted_length + 1) > pbuffer8_size);
if (toolong)
{
fprintf(outfile, "** Pattern conversion is too long for the buffer\n");
convert_return = PR_SKIP;
}
else
{
CONVERT_COPY(pbuffer, converted_pattern, converted_length + 1);
patlen = converted_length;
}
}
/* Free the converted pattern. */
CONVERT_FINISH:
if (pat_patctl.convert_length != 0)
free(converted_pattern);
else
PCRE2_CONVERTED_PATTERN_FREE(converted_pattern);
/* Return if conversion was unsuccessful. */
if (convert_return != PR_OK) return convert_return;
}
/* By default we pass a zero-terminated pattern, but a length is passed if
"use_length" was specified or this is a hex pattern (which might contain binary
zeros). When valgrind is supported, arrange for the unused part of the buffer
to be marked as no access. */
valgrind_access_length = patlen;
if ((pat_patctl.control & (CTL_HEXPAT|CTL_USE_LENGTH)) == 0)
{
patlen = PCRE2_ZERO_TERMINATED;
valgrind_access_length += 1; /* For the terminating zero */
}
#ifdef SUPPORT_VALGRIND
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE && pbuffer8 != NULL)
{
VALGRIND_MAKE_MEM_NOACCESS(pbuffer8 + valgrind_access_length,
pbuffer8_size - valgrind_access_length);
}
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE && pbuffer16 != NULL)
{
VALGRIND_MAKE_MEM_NOACCESS(pbuffer16 + valgrind_access_length,
pbuffer16_size - valgrind_access_length*sizeof(uint16_t));
}
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE && pbuffer32 != NULL)
{
VALGRIND_MAKE_MEM_NOACCESS(pbuffer32 + valgrind_access_length,
pbuffer32_size - valgrind_access_length*sizeof(uint32_t));
}
#endif
#else /* Valgrind not supported */
(void)valgrind_access_length; /* Avoid compiler warning */
#endif
/* If #newline_default has been used and the library was not compiled with an
appropriate default newline setting, local_newline_default will be non-zero. We
use this if there is no explicit newline modifier. */
if ((pat_patctl.control2 & CTL2_NL_SET) == 0 && local_newline_default != 0)
{
SETFLD(pat_context, newline_convention, local_newline_default);
}
/* The null_context modifier is used to test calling pcre2_compile() with a
NULL context. */
use_pat_context = ((pat_patctl.control & CTL_NULLCONTEXT) != 0)?
NULL : PTR(pat_context);
/* If PCRE2_LITERAL is set, set use_forbid_utf zero because PCRE2_NEVER_UTF
and PCRE2_NEVER_UCP are invalid with it. */
if ((pat_patctl.options & PCRE2_LITERAL) != 0) use_forbid_utf = 0;
/* Set use_pbuffer to the input buffer, or leave it as NULL if requested. */
if ((pat_patctl.control2 & CTL2_NULL_PATTERN) == 0)
{
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE) use_pbuffer = pbuffer8;
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE) use_pbuffer = pbuffer16;
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE) use_pbuffer = pbuffer32;
#endif
}
/* Compile many times when timing. */
if (timeit > 0)
{
int i;
clock_t time_taken = 0;
for (i = 0; i < timeit; i++)
{
clock_t start_time = clock();
PCRE2_COMPILE(compiled_code, use_pbuffer, patlen,
pat_patctl.options|use_forbid_utf, &errorcode, &erroroffset,
use_pat_context);
time_taken += clock() - start_time;
if (TEST(compiled_code, !=, NULL))
{ SUB1(pcre2_code_free, compiled_code); }
}
total_compile_time += time_taken;
fprintf(outfile, "Compile time %8.4f microseconds\n",
((1000000 / CLOCKS_PER_SEC) * (double)time_taken) / timeit);
}
/* A final compile that is used "for real". */
PCRE2_COMPILE(compiled_code, use_pbuffer, patlen,
pat_patctl.options|use_forbid_utf, &errorcode, &erroroffset, use_pat_context);
/* If valgrind is supported, mark the pbuffer as accessible again. The 16-bit
and 32-bit buffers can be marked completely undefined, but we must leave the
pattern in the 8-bit buffer defined because it may be read from a callout
during matching. */
#ifdef SUPPORT_VALGRIND
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE)
{
VALGRIND_MAKE_MEM_UNDEFINED(pbuffer8 + valgrind_access_length,
pbuffer8_size - valgrind_access_length);
}
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE)
{
VALGRIND_MAKE_MEM_UNDEFINED(pbuffer16, pbuffer16_size);
}
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE)
{
VALGRIND_MAKE_MEM_UNDEFINED(pbuffer32, pbuffer32_size);
}
#endif
#endif
/* Call the JIT compiler if requested. When timing, we must free and recompile
the pattern each time because that is the only way to free the JIT compiled
code. We know that compilation will always succeed. */
if (TEST(compiled_code, !=, NULL) && pat_patctl.jit != 0)
{
if (timeit > 0)
{
int i;
clock_t time_taken = 0;
for (i = 0; i < timeit; i++)
{
clock_t start_time;
SUB1(pcre2_code_free, compiled_code);
PCRE2_COMPILE(compiled_code, use_pbuffer, patlen,
pat_patctl.options|use_forbid_utf, &errorcode, &erroroffset,
use_pat_context);
start_time = clock();
PCRE2_JIT_COMPILE(jitrc, compiled_code, pat_patctl.jit);
time_taken += clock() - start_time;
if (jitrc != 0)
{
fprintf(outfile, "JIT compilation was not successful");
if (!print_error_message(jitrc, " (", ")\n")) return PR_ABEND;
break;
}
}
total_jit_compile_time += time_taken;
if (jitrc == 0)
fprintf(outfile, "JIT compile %8.4f microseconds\n",
((1000000 / CLOCKS_PER_SEC) * (double)time_taken) / timeit);
}
else
{
PCRE2_JIT_COMPILE(jitrc, compiled_code, pat_patctl.jit);
if (jitrc != 0 && (pat_patctl.control & CTL_JITVERIFY) != 0)
{
fprintf(outfile, "JIT compilation was not successful");
if (!print_error_message(jitrc, " (", ")\n")) return PR_ABEND;
}
}
}
/* Compilation failed; go back for another re, skipping to blank line
if non-interactive. */
if (TEST(compiled_code, ==, NULL))
{
fprintf(outfile, "Failed: error %d at offset %d: ", errorcode,
(int)erroroffset);
if (!print_error_message(errorcode, "", "\n")) return PR_ABEND;
return PR_SKIP;
}
/* If forbid_utf is non-zero, we are running a non-UTF test. UTF and UCP are
locked out at compile time, but we must also check for occurrences of \P, \p,
and \X, which are only supported when Unicode is supported. */
if (forbid_utf != 0)
{
if ((FLD(compiled_code, flags) & PCRE2_HASBKPORX) != 0)
{
fprintf(outfile, "** \\P, \\p, and \\X are not allowed after the "
"#forbid_utf command\n");
return PR_SKIP;
}
}
/* Remember the maximum lookbehind, for partial matching. */
if (pattern_info(PCRE2_INFO_MAXLOOKBEHIND, &maxlookbehind, FALSE) != 0)
return PR_ABEND;
/* Remember the number of captures. */
if (pattern_info(PCRE2_INFO_CAPTURECOUNT, &maxcapcount, FALSE) < 0)
return PR_ABEND;
/* If an explicit newline modifier was given, set the information flag in the
pattern so that it is preserved over push/pop. */
if ((pat_patctl.control2 & CTL2_NL_SET) != 0)
{
SETFLD(compiled_code, flags, FLD(compiled_code, flags) | PCRE2_NL_SET);
}
/* Output code size and other information if requested. */
if ((pat_patctl.control & CTL_MEMORY) != 0) show_memory_info();
if ((pat_patctl.control2 & CTL2_FRAMESIZE) != 0) show_framesize();
if ((pat_patctl.control & CTL_ANYINFO) != 0)
{
int rc = show_pattern_info();
if (rc != PR_OK) return rc;
}
/* The "push" control requests that the compiled pattern be remembered on a
stack. This is mainly for testing the serialization functionality. */
if ((pat_patctl.control & CTL_PUSH) != 0)
{
if (patstacknext >= PATSTACKSIZE)
{
fprintf(outfile, "** Too many pushed patterns (max %d)\n", PATSTACKSIZE);
return PR_ABEND;
}
patstack[patstacknext++] = PTR(compiled_code);
SET(compiled_code, NULL);
}
/* The "pushcopy" and "pushtablescopy" controls are similar, but push a
copy of the pattern, the latter with a copy of its character tables. This tests
the pcre2_code_copy() and pcre2_code_copy_with_tables() functions. */
if ((pat_patctl.control & (CTL_PUSHCOPY|CTL_PUSHTABLESCOPY)) != 0)
{
if (patstacknext >= PATSTACKSIZE)
{
fprintf(outfile, "** Too many pushed patterns (max %d)\n", PATSTACKSIZE);
return PR_ABEND;
}
if ((pat_patctl.control & CTL_PUSHCOPY) != 0)
{
PCRE2_CODE_COPY_TO_VOID(patstack[patstacknext++], compiled_code);
}
else
{
PCRE2_CODE_COPY_WITH_TABLES_TO_VOID(patstack[patstacknext++],
compiled_code); }
}
return PR_OK;
}
/*************************************************
* Check heap, match or depth limit *
*************************************************/
/* This is used for DFA, normal, and JIT fast matching. For DFA matching it
should only be called with the third argument set to PCRE2_ERROR_DEPTHLIMIT.
Arguments:
pp the subject string
ulen length of subject or PCRE2_ZERO_TERMINATED
errnumber defines which limit to test
msg string to include in final message
Returns: the return from the final match function call
*/
static int
check_match_limit(uint8_t *pp, PCRE2_SIZE ulen, int errnumber, const char *msg)
{
int capcount;
uint32_t min = 0;
uint32_t mid = 64;
uint32_t max = UINT32_MAX;
PCRE2_SET_MATCH_LIMIT(dat_context, max);
PCRE2_SET_DEPTH_LIMIT(dat_context, max);
PCRE2_SET_HEAP_LIMIT(dat_context, max);
for (;;)
{
uint32_t stack_start = 0;
/* If we are checking the heap limit, free any frames vector that is cached
in the match_data so we always start without one. */
if (errnumber == PCRE2_ERROR_HEAPLIMIT)
{
PCRE2_SET_HEAP_LIMIT(dat_context, mid);
#ifdef SUPPORT_PCRE2_8
if (code_unit_size == 1)
{
match_data8->memctl.free(match_data8->heapframes,
match_data8->memctl.memory_data);
match_data8->heapframes = NULL;
match_data8->heapframes_size = 0;
}
#endif
#ifdef SUPPORT_PCRE2_16
if (code_unit_size == 2)
{
match_data16->memctl.free(match_data16->heapframes,
match_data16->memctl.memory_data);
match_data16->heapframes = NULL;
match_data16->heapframes_size = 0;
}
#endif
#ifdef SUPPORT_PCRE2_32
if (code_unit_size == 4)
{
match_data32->memctl.free(match_data32->heapframes,
match_data32->memctl.memory_data);
match_data32->heapframes = NULL;
match_data32->heapframes_size = 0;
}
#endif
}
/* No need to mess with the frames vector for match or depth limits. */
else if (errnumber == PCRE2_ERROR_MATCHLIMIT)
{
PCRE2_SET_MATCH_LIMIT(dat_context, mid);
}
else
{
PCRE2_SET_DEPTH_LIMIT(dat_context, mid);
}
/* Do the appropriate match */
if ((dat_datctl.control & CTL_DFA) != 0)
{
stack_start = DFA_START_RWS_SIZE/1024;
if (dfa_workspace == NULL)
dfa_workspace = (int *)malloc(DFA_WS_DIMENSION*sizeof(int));
if (dfa_matched++ == 0)
dfa_workspace[0] = -1; /* To catch bad restart */
PCRE2_DFA_MATCH(capcount, compiled_code, pp, ulen, dat_datctl.offset,
dat_datctl.options, match_data,
PTR(dat_context), dfa_workspace, DFA_WS_DIMENSION);
}
else if ((pat_patctl.control & CTL_JITFAST) != 0)
PCRE2_JIT_MATCH(capcount, compiled_code, pp, ulen, dat_datctl.offset,
dat_datctl.options, match_data, PTR(dat_context));
else
{
PCRE2_MATCH(capcount, compiled_code, pp, ulen, dat_datctl.offset,
dat_datctl.options, match_data, PTR(dat_context));
}
if (capcount == errnumber)
{
if ((mid & 0x80000000u) != 0)
{
fprintf(outfile, "Can't find minimum %s limit: check pattern for "
"restriction\n", msg);
break;
}
min = mid;
mid = (mid == max - 1)? max : (max != UINT32_MAX)? (min + max)/2 : mid*2;
}
else if (capcount >= 0 ||
capcount == PCRE2_ERROR_NOMATCH ||
capcount == PCRE2_ERROR_PARTIAL)
{
/* If we've not hit the error with a heap limit less than the size of the
initial stack frame vector (for pcre2_match()) or the initial stack
workspace vector (for pcre2_dfa_match()), the heap is not being used, so
the minimum limit is zero; there's no need to go on. The other limits are
always greater than zero. */
if (errnumber == PCRE2_ERROR_HEAPLIMIT && mid < stack_start)
{
fprintf(outfile, "Minimum %s limit = 0\n", msg);
break;
}
if (mid == min + 1)
{
fprintf(outfile, "Minimum %s limit = %d\n", msg, mid);
break;
}
max = mid;
mid = (min + max)/2;
}
else break; /* Some other error */
}
return capcount;
}
/*************************************************
* Substitute callout function *
*************************************************/
/* Called from pcre2_substitute() when the substitute_callout modifier is set.
Print out the data that is passed back. The substitute callout block is
identical for all code unit widths, so we just pick one.
Arguments:
scb pointer to substitute callout block
data_ptr callout data
Returns: nothing
*/
static int
substitute_callout_function(pcre2_substitute_callout_block_8 *scb,
void *data_ptr)
{
int yield = 0;
BOOL utf = (FLD(compiled_code, overall_options) & PCRE2_UTF) != 0;
(void)data_ptr; /* Not used */
fprintf(outfile, "%2d(%d) Old %" SIZ_FORM " %" SIZ_FORM " \"",
scb->subscount, scb->oveccount,
scb->ovector[0], scb->ovector[1]);
PCHARSV(scb->input, scb->ovector[0], scb->ovector[1] - scb->ovector[0],
utf, outfile);
fprintf(outfile, "\" New %" SIZ_FORM " %" SIZ_FORM " \"",
scb->output_offsets[0], scb->output_offsets[1]);
PCHARSV(scb->output, scb->output_offsets[0],
scb->output_offsets[1] - scb->output_offsets[0], utf, outfile);
if (scb->subscount == dat_datctl.substitute_stop)
{
yield = -1;
fprintf(outfile, " STOPPED");
}
else if (scb->subscount == dat_datctl.substitute_skip)
{
yield = +1;
fprintf(outfile, " SKIPPED");
}
fprintf(outfile, "\"\n");
return yield;
}
/*************************************************
* Substitute case callout function *
*************************************************/
/* Function to implement our test-only custom case mappings.
To ease implementation, we only work in the ASCII range (so that we don't need
to read & write UTF sequences).
However, we aim to implement case mappings which fairly well represent the range
of interesting behaviours that exist for Unicode codepoints. */
static BOOL
case_transform(int to_case, int num_in, int *num_read, int *num_write,
uint32_t *c1, uint32_t *c2)
{
/* Let's have one character which aborts the substitution. */
if (*c1 == '!') return FALSE;
/* Default behaviour is to read one character, and write back that same one
character (treating all characters as "uncased"). */
*num_read = *num_write = 1;
/* Add a normal case pair 'a' (l) <-> 'B' (t,u). Standard ASCII letter
behaviour, but with switched letters for testing. */
if (*c1 == 'a' && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'B';
else if (*c1 == 'B' && to_case == PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'a';
/* Add a titlecased triplet 'd' (l) <-> 'D' (t) <-> 'Z' (u). Example: the
'dz'/'Dz'/'DZ' ligature character ("Latin Small Letter DZ" <-> "Latin Capital
Letter D with Small Letter Z" <-> "Latin Capital Letter DZ"). */
else if (*c1 == 'd' && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = (to_case == PCRE2_SUBSTITUTE_CASE_TITLE_FIRST)? 'D' : 'Z';
else if (*c1 == 'D' && to_case != PCRE2_SUBSTITUTE_CASE_TITLE_FIRST)
*c1 = (to_case == PCRE2_SUBSTITUTE_CASE_LOWER)? 'd' : 'Z';
else if (*c1 == 'Z' && to_case != PCRE2_SUBSTITUTE_CASE_UPPER)
*c1 = (to_case == PCRE2_SUBSTITUTE_CASE_LOWER)? 'd' : 'D';
/* Expands when uppercased. Example: Esszet 'f' <-> 'SS'. */
else if (*c1 == 'f' && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
{
*c1 = 'S';
*c2 = 'S';
*num_write = 2;
}
else if (*c1 == 's' && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'S';
else if (*c1 == 'S' && to_case == PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 's';
/* Expanding and contracting characters, 'o' <-> 'OO'. You can get this purely
due to UTF-8 encoding length, for example uppercase Omega (3 bytes in UTF-8)
lowercases to 2 bytes in UTF-8. */
else if (num_in == 2 && *c1 == 'O' && *c2 == 'O' &&
to_case == PCRE2_SUBSTITUTE_CASE_LOWER)
{
*c1 = 'o';
*num_read = 2;
}
else if (*c1 == 'o' && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
{
*c1 = 'O';
*c2 = 'O';
*num_write = 2;
}
else if (num_in == 2 && *c1 == 'p' && *c2 == 'p' &&
to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
{
*c1 = 'P';
*num_read = 2;
}
else if (*c1 == 'P' && to_case == PCRE2_SUBSTITUTE_CASE_LOWER)
{
*c1 = 'p';
*c2 = 'p';
*num_write = 2;
}
/* Use 'l' -> 'Mn' or 'MN' as an expanding ligature, like 'fi' -> 'Fi' ->
'FI'. */
else if (*c1 == 'l' && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
{
*c1 = 'M';
*c2 = (to_case == PCRE2_SUBSTITUTE_CASE_TITLE_FIRST)? 'n' : 'N';
*num_write = 2;
}
else if (*c1 == 'M' && to_case == PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'm';
else if (*c1 == 'm' && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'M';
else if (*c1 == 'N' && to_case == PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'n';
else if (*c1 == 'n' && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'N';
/* An example of a context-dependent mapping, the Greek Sigma. It lowercases
depending on the following character. Use 'c'/'k' -> 'K'. */
else if ((*c1 == 'c' || *c1 == 'k') && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'K';
else if (*c1 == 'K' && to_case == PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = (num_in == 1 || *c2 == ' ')? 'c' : 'k';
/* An example of a context-dependent multi mapping, the Dutch IJ. When those
letters appear together, they titlecase 'ij' (l) <-> 'IJ' (t) <-> 'IJ' (u).
Namely, English titlecasing of 'ijnssel' would be 'Ijnssel' (just uppercase the
first letter), but the Dutch rule is 'IJnssel'. */
else if (num_in == 2 && (*c1 == 'i' || *c1 == 'I') &&
(*c2 == 'j' || *c2 == 'J') &&
to_case == PCRE2_SUBSTITUTE_CASE_TITLE_FIRST)
{
*c1 = 'I';
*c2 = 'J';
*num_read = 2;
*num_write = 2;
}
else if (*c1 == 'i' && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'I';
else if (*c1 == 'I' && to_case == PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'i';
else if (*c1 == 'j' && to_case != PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'J';
else if (*c1 == 'J' && to_case == PCRE2_SUBSTITUTE_CASE_LOWER)
*c1 = 'j';
return TRUE;
}
/* Called from pcre2_substitute() when the substitute_case_callout
modifier is set. The substitute callout block is not identical for all code unit
widths, so we have to duplicate the function for each supported width.
Arguments:
input the input character
input_len the length of the input
output the output buffer
output_cap the output buffer capacity
to_case the case conversion type
data_ptr callout data (unused)
Returns: the number of code units of the output
*/
#define substitute_case_callout_function(BITS) \
static PCRE2_SIZE \
G(substitute_case_callout_function,BITS)( \
G(PCRE2_SPTR,BITS) input, PCRE2_SIZE input_len, \
G(PCRE2_UCHAR,BITS) *output, PCRE2_SIZE output_cap, \
int to_case, void *data_ptr) \
{ \
G(PCRE2_UCHAR,BITS) buf[16]; \
G(PCRE2_SPTR,BITS) input_copy; \
PCRE2_SIZE written = 0; \
\
(void)data_ptr; /* Not used */ \
\
if (input_len > sizeof(buf)/sizeof(*buf)) \
{ \
G(PCRE2_UCHAR,BITS) *input_buf = malloc( \
input_len * sizeof(G(PCRE2_UCHAR,BITS))); \
if (input_buf == NULL) return ~(PCRE2_SIZE)0; \
memcpy(input_buf, input, input_len * sizeof(G(PCRE2_UCHAR,BITS))); \
input_copy = input_buf; \
} \
else \
{ \
memcpy(buf, input, input_len * sizeof(G(PCRE2_UCHAR,BITS))); \
input_copy = buf; \
} \
\
for (PCRE2_SIZE i = 0; i < input_len; ) \
{ \
int num_in = i + 1 < input_len ? 2 : 1; \
uint32_t c1 = input_copy[i]; \
uint32_t c2 = i + 1 < input_len ? input_copy[i + 1] : 0; \
int num_read; \
int num_write; \
\
if (!case_transform(to_case, num_in, &num_read, &num_write, &c1, &c2)) \
{ \
written = ~(PCRE2_SIZE)0; \
goto END; \
} \
\
i += num_read; \
if (to_case == PCRE2_SUBSTITUTE_CASE_TITLE_FIRST) \
to_case = PCRE2_SUBSTITUTE_CASE_LOWER; \
\
if (written + num_write > output_cap) \
{ \
written += num_write; \
} \
else \
{ \
if (num_write > 0) output[written++] = c1; \
if (num_write > 1) output[written++] = c2; \
} \
} \
\
END: \
if (input_copy != buf) free((G(PCRE2_UCHAR,BITS) *)input_copy); \
\
/* Let's be maximally cruel. The case callout is allowed to leave the output
buffer in any state at all if it overflows, so let's use random garbage. */ \
if (written > output_cap) \
memset(output, time(NULL) & 1 ? 0xcd : 0xdc, \
output_cap * sizeof(G(PCRE2_UCHAR,BITS))); \
\
return written; \
}
#if defined SUPPORT_PCRE2_8
substitute_case_callout_function(8)
#endif
#if defined SUPPORT_PCRE2_16
substitute_case_callout_function(16)
#endif
#if defined SUPPORT_PCRE2_32
substitute_case_callout_function(32)
#endif
/*************************************************
* Callout function *
*************************************************/
/* Called from a PCRE2 library as a result of the (?C) item. We print out where
we are in the match (unless suppressed). Yield zero unless more callouts than
the fail count, or the callout data is not zero. The only differences in the
callout block for different code unit widths are that the pointers to the
subject, the most recent MARK, and a callout argument string point to strings
of the appropriate width. Casts can be used to deal with this.
Arguments:
cb a pointer to a callout block
callout_data_ptr the provided callout data
Returns: 0 or 1 or an error, as determined by settings
*/
static int
callout_function(pcre2_callout_block_8 *cb, void *callout_data_ptr)
{
FILE *f, *fdefault;
uint32_t i, pre_start, post_start, subject_length;
PCRE2_SIZE current_position;
BOOL utf = (FLD(compiled_code, overall_options) & PCRE2_UTF) != 0;
BOOL callout_capture = (dat_datctl.control & CTL_CALLOUT_CAPTURE) != 0;
BOOL callout_where = (dat_datctl.control2 & CTL2_CALLOUT_NO_WHERE) == 0;
/* The FILE f is used for echoing the subject string if it is non-NULL. This
happens only once in simple cases, but we want to repeat after any additional
output caused by CALLOUT_EXTRA. */
fdefault = (!first_callout && !callout_capture && cb->callout_string == NULL)?
NULL : outfile;
if ((dat_datctl.control2 & CTL2_CALLOUT_EXTRA) != 0)
{
f = outfile;
switch (cb->callout_flags)
{
case PCRE2_CALLOUT_BACKTRACK:
fprintf(f, "Backtrack\n");
break;
case PCRE2_CALLOUT_STARTMATCH|PCRE2_CALLOUT_BACKTRACK:
fprintf(f, "Backtrack\nNo other matching paths\n");
/* Fall through */
case PCRE2_CALLOUT_STARTMATCH:
fprintf(f, "New match attempt\n");
break;
default:
f = fdefault;
break;
}
}
else f = fdefault;
/* For a callout with a string argument, show the string first because there
isn't a tidy way to fit it in the rest of the data. */
if (cb->callout_string != NULL)
{
uint32_t delimiter = CODE_UNIT(cb->callout_string, -1);
fprintf(outfile, "Callout (%" SIZ_FORM "): %c",
cb->callout_string_offset, delimiter);
PCHARSV(cb->callout_string, 0,
cb->callout_string_length, utf, outfile);
for (i = 0; callout_start_delims[i] != 0; i++)
if (delimiter == callout_start_delims[i])
{
delimiter = callout_end_delims[i];
break;
}
fprintf(outfile, "%c", delimiter);
if (!callout_capture) fprintf(outfile, "\n");
}
/* Show captured strings if required */
if (callout_capture)
{
if (cb->callout_string == NULL)
fprintf(outfile, "Callout %d:", cb->callout_number);
fprintf(outfile, " last capture = %d\n", cb->capture_last);
for (i = 2; i < cb->capture_top * 2; i += 2)
{
fprintf(outfile, "%2d: ", i/2);
if (cb->offset_vector[i] == PCRE2_UNSET)
fprintf(outfile, "");
else
{
PCHARSV(cb->subject, cb->offset_vector[i],
cb->offset_vector[i+1] - cb->offset_vector[i], utf, f);
}
fprintf(outfile, "\n");
}
}
/* Unless suppressed, re-print the subject in canonical form (with escapes for
non-printing characters), the first time, or if giving full details. On
subsequent calls in the same match, we use PCHARS() just to find the printed
lengths of the substrings. */
if (callout_where)
{
if (f != NULL) fprintf(f, "--->");
/* The subject before the match start. */
PCHARS(pre_start, cb->subject, 0, cb->start_match, utf, f);
/* If a lookbehind is involved, the current position may be earlier than the
match start. If so, use the match start instead. */
current_position = (cb->current_position >= cb->start_match)?
cb->current_position : cb->start_match;
/* The subject between the match start and the current position. */
PCHARS(post_start, cb->subject, cb->start_match,
current_position - cb->start_match, utf, f);
/* Print from the current position to the end. */
PCHARSV(cb->subject, current_position, cb->subject_length - current_position,
utf, f);
/* Calculate the total subject printed length (no print). */
PCHARS(subject_length, cb->subject, 0, cb->subject_length, utf, NULL);
if (f != NULL) fprintf(f, "\n");
/* For automatic callouts, show the pattern offset. Otherwise, for a
numerical callout whose number has not already been shown with captured
strings, show the number here. A callout with a string argument has been
displayed above. */
if (cb->callout_number == 255)
{
fprintf(outfile, "%+3d ", (int)cb->pattern_position);
if (cb->pattern_position > 99) fprintf(outfile, "\n ");
}
else
{
if (callout_capture || cb->callout_string != NULL) fprintf(outfile, " ");
else fprintf(outfile, "%3d ", cb->callout_number);
}
/* Now show position indicators */
for (i = 0; i < pre_start; i++) fprintf(outfile, " ");
fprintf(outfile, "^");
if (post_start > 0)
{
for (i = 0; i < post_start - 1; i++) fprintf(outfile, " ");
fprintf(outfile, "^");
}
for (i = 0; i < subject_length - pre_start - post_start + 4; i++)
fprintf(outfile, " ");
if (cb->next_item_length != 0)
fprintf(outfile, "%.*s", (int)(cb->next_item_length),
pbuffer8 + cb->pattern_position);
else
fprintf(outfile, "End of pattern");
fprintf(outfile, "\n");
}
first_callout = FALSE;
/* Show any mark info */
if (cb->mark != last_callout_mark)
{
if (cb->mark == NULL)
fprintf(outfile, "Latest Mark: \n");
else
{
fprintf(outfile, "Latest Mark: ");
PCHARSV(cb->mark, -1, -1, utf, outfile);
putc('\n', outfile);
}
last_callout_mark = cb->mark;
}
/* Show callout data */
if (callout_data_ptr != NULL)
{
int callout_data = *((int32_t *)callout_data_ptr);
if (callout_data != 0)
{
fprintf(outfile, "Callout data = %d\n", callout_data);
return callout_data;
}
}
/* Keep count and give the appropriate return code */
callout_count++;
if (cb->callout_number == dat_datctl.cerror[0] &&
callout_count >= dat_datctl.cerror[1])
return PCRE2_ERROR_CALLOUT;
if (cb->callout_number == dat_datctl.cfail[0] &&
callout_count >= dat_datctl.cfail[1])
return 1;
return 0;
}
/*************************************************
* Handle *MARK and copy/get tests *
*************************************************/
/* This function is called after complete and partial matches. It runs the
tests for substring extraction.
Arguments:
utf TRUE for utf
capcount return from pcre2_match()
Returns: FALSE if print_error_message() fails
*/
static BOOL
copy_and_get(BOOL utf, int capcount)
{
int i;
uint8_t *nptr;
/* Test copy strings by number */
for (i = 0; i < MAXCPYGET && dat_datctl.copy_numbers[i] >= 0; i++)
{
int rc;
PCRE2_SIZE length, length2;
uint32_t copybuffer[256];
uint32_t n = (uint32_t)(dat_datctl.copy_numbers[i]);
length = sizeof(copybuffer)/code_unit_size;
PCRE2_SUBSTRING_COPY_BYNUMBER(rc, match_data, n, copybuffer, &length);
if (rc < 0)
{
fprintf(outfile, "Copy substring %d failed (%d): ", n, rc);
if (!print_error_message(rc, "", "\n")) return FALSE;
}
else
{
PCRE2_SUBSTRING_LENGTH_BYNUMBER(rc, match_data, n, &length2);
if (rc < 0)
{
fprintf(outfile, "Get substring %d length failed (%d): ", n, rc);
if (!print_error_message(rc, "", "\n")) return FALSE;
}
else if (length2 != length)
{
fprintf(outfile, "Mismatched substring lengths: %"
SIZ_FORM " %" SIZ_FORM "\n", length, length2);
}
fprintf(outfile, "%2dC ", n);
PCHARSV(copybuffer, 0, length, utf, outfile);
fprintf(outfile, " (%" SIZ_FORM ")\n", length);
}
}
/* Test copy strings by name */
nptr = dat_datctl.copy_names;
for (;;)
{
int rc;
int groupnumber;
PCRE2_SIZE length, length2;
uint32_t copybuffer[256];
int namelen = strlen((const char *)nptr);
#if defined SUPPORT_PCRE2_16 || defined SUPPORT_PCRE2_32
PCRE2_SIZE cnl = namelen;
#endif
if (namelen == 0) break;
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE) strcpy((char *)pbuffer8, (char *)nptr);
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE)(void)to16(nptr, utf, &cnl);
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE)(void)to32(nptr, utf, &cnl);
#endif
PCRE2_SUBSTRING_NUMBER_FROM_NAME(groupnumber, compiled_code, pbuffer);
if (groupnumber < 0 && groupnumber != PCRE2_ERROR_NOUNIQUESUBSTRING)
fprintf(outfile, "Number not found for group \"%s\"\n", nptr);
length = sizeof(copybuffer)/code_unit_size;
PCRE2_SUBSTRING_COPY_BYNAME(rc, match_data, pbuffer, copybuffer, &length);
if (rc < 0)
{
fprintf(outfile, "Copy substring \"%s\" failed (%d): ", nptr, rc);
if (!print_error_message(rc, "", "\n")) return FALSE;
}
else
{
PCRE2_SUBSTRING_LENGTH_BYNAME(rc, match_data, pbuffer, &length2);
if (rc < 0)
{
fprintf(outfile, "Get substring \"%s\" length failed (%d): ", nptr, rc);
if (!print_error_message(rc, "", "\n")) return FALSE;
}
else if (length2 != length)
{
fprintf(outfile, "Mismatched substring lengths: %"
SIZ_FORM " %" SIZ_FORM "\n", length, length2);
}
fprintf(outfile, " C ");
PCHARSV(copybuffer, 0, length, utf, outfile);
fprintf(outfile, " (%" SIZ_FORM ") %s", length, nptr);
if (groupnumber >= 0) fprintf(outfile, " (group %d)\n", groupnumber);
else fprintf(outfile, " (non-unique)\n");
}
nptr += namelen + 1;
}
/* Test get strings by number */
for (i = 0; i < MAXCPYGET && dat_datctl.get_numbers[i] >= 0; i++)
{
int rc;
PCRE2_SIZE length;
void *gotbuffer;
uint32_t n = (uint32_t)(dat_datctl.get_numbers[i]);
PCRE2_SUBSTRING_GET_BYNUMBER(rc, match_data, n, &gotbuffer, &length);
if (rc < 0)
{
fprintf(outfile, "Get substring %d failed (%d): ", n, rc);
if (!print_error_message(rc, "", "\n")) return FALSE;
}
else
{
fprintf(outfile, "%2dG ", n);
PCHARSV(gotbuffer, 0, length, utf, outfile);
fprintf(outfile, " (%" SIZ_FORM ")\n", length);
PCRE2_SUBSTRING_FREE(gotbuffer);
}
}
/* Test get strings by name */
nptr = dat_datctl.get_names;
for (;;)
{
PCRE2_SIZE length;
void *gotbuffer;
int rc;
int groupnumber;
int namelen = strlen((const char *)nptr);
#if defined SUPPORT_PCRE2_16 || defined SUPPORT_PCRE2_32
PCRE2_SIZE cnl = namelen;
#endif
if (namelen == 0) break;
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE) strcpy((char *)pbuffer8, (char *)nptr);
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE)(void)to16(nptr, utf, &cnl);
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE)(void)to32(nptr, utf, &cnl);
#endif
PCRE2_SUBSTRING_NUMBER_FROM_NAME(groupnumber, compiled_code, pbuffer);
if (groupnumber < 0 && groupnumber != PCRE2_ERROR_NOUNIQUESUBSTRING)
fprintf(outfile, "Number not found for group \"%s\"\n", nptr);
PCRE2_SUBSTRING_GET_BYNAME(rc, match_data, pbuffer, &gotbuffer, &length);
if (rc < 0)
{
fprintf(outfile, "Get substring \"%s\" failed (%d): ", nptr, rc);
if (!print_error_message(rc, "", "\n")) return FALSE;
}
else
{
fprintf(outfile, " G ");
PCHARSV(gotbuffer, 0, length, utf, outfile);
fprintf(outfile, " (%" SIZ_FORM ") %s", length, nptr);
if (groupnumber >= 0) fprintf(outfile, " (group %d)\n", groupnumber);
else fprintf(outfile, " (non-unique)\n");
PCRE2_SUBSTRING_FREE(gotbuffer);
}
nptr += namelen + 1;
}
/* Test getting the complete list of captured strings. */
if ((dat_datctl.control & CTL_GETALL) != 0)
{
int rc;
void **stringlist;
PCRE2_SIZE *lengths;
PCRE2_SUBSTRING_LIST_GET(rc, match_data, &stringlist, &lengths);
if (rc < 0)
{
fprintf(outfile, "get substring list failed (%d): ", rc);
if (!print_error_message(rc, "", "\n")) return FALSE;
}
else
{
for (i = 0; i < capcount; i++)
{
fprintf(outfile, "%2dL ", i);
PCHARSV(stringlist[i], 0, lengths[i], utf, outfile);
putc('\n', outfile);
}
if (stringlist[i] != NULL)
fprintf(outfile, "string list not terminated by NULL\n");
PCRE2_SUBSTRING_LIST_FREE(stringlist);
}
}
return TRUE;
}
/*************************************************
* Show an entire ovector *
*************************************************/
/* This function is called after partial matching or match failure, when the
"allvector" modifier is set. It is a means of checking the contents of the
entire ovector, to ensure no modification of fields that should be unchanged.
Arguments:
ovector points to the ovector
oveccount number of pairs
Returns: nothing
*/
static void
show_ovector(PCRE2_SIZE *ovector, uint32_t oveccount)
{
uint32_t i;
for (i = 0; i < 2*oveccount; i += 2)
{
PCRE2_SIZE start = ovector[i];
PCRE2_SIZE end = ovector[i+1];
fprintf(outfile, "%2d: ", i/2);
if (start == PCRE2_UNSET && end == PCRE2_UNSET)
fprintf(outfile, "\n");
else if (start == JUNK_OFFSET && end == JUNK_OFFSET)
fprintf(outfile, "\n");
else
fprintf(outfile, "%ld %ld\n", (unsigned long int)start,
(unsigned long int)end);
}
}
/*************************************************
* Process a data line *
*************************************************/
/* The line is in buffer; it will not be empty.
Arguments: none
Returns: PR_OK continue processing next line
PR_SKIP skip to a blank line
PR_ABEND abort the pcre2test run
*/
static int
process_data(void)
{
PCRE2_SIZE len, ulen, arg_ulen;
uint32_t gmatched;
uint32_t c, k;
uint32_t g_notempty = 0;
uint8_t *p, *pp, *start_rep;
size_t needlen;
void *use_dat_context;
BOOL utf;
BOOL subject_literal;
PCRE2_SIZE *ovector;
PCRE2_SIZE ovecsave[3];
uint32_t oveccount;
#ifdef SUPPORT_PCRE2_8
uint8_t *q8 = NULL;
#endif
#ifdef SUPPORT_PCRE2_16
uint16_t *q16 = NULL;
#endif
#ifdef SUPPORT_PCRE2_32
uint32_t *q32 = NULL;
#endif
subject_literal = (pat_patctl.control2 & CTL2_SUBJECT_LITERAL) != 0;
/* Copy the default context and data control blocks to the active ones. Then
copy from the pattern the controls that can be set in either the pattern or the
data. This allows them to be overridden in the data line. We do not do this for
options because those that are common apply separately to compiling and
matching. */
DATCTXCPY(dat_context, default_dat_context);
memcpy(&dat_datctl, &def_datctl, sizeof(datctl));
dat_datctl.control |= (pat_patctl.control & CTL_ALLPD);
dat_datctl.control2 |= (pat_patctl.control2 & CTL2_ALLPD);
strcpy((char *)dat_datctl.replacement, (char *)pat_patctl.replacement);
if (dat_datctl.jitstack == 0) dat_datctl.jitstack = pat_patctl.jitstack;
if (dat_datctl.substitute_skip == 0)
dat_datctl.substitute_skip = pat_patctl.substitute_skip;
if (dat_datctl.substitute_stop == 0)
dat_datctl.substitute_stop = pat_patctl.substitute_stop;
/* Initialize for scanning the data line. */
#ifdef SUPPORT_PCRE2_8
utf = ((((pat_patctl.control & CTL_POSIX) != 0)?
((pcre2_real_code_8 *)preg.re_pcre2_code)->overall_options :
FLD(compiled_code, overall_options)) & PCRE2_UTF) != 0;
#else
utf = (FLD(compiled_code, overall_options) & PCRE2_UTF) != 0;
#endif
start_rep = NULL;
len = strlen((const char *)buffer);
while (len > 0 && isspace(buffer[len-1])) len--;
buffer[len] = 0;
p = buffer;
while (isspace(*p))
{
p++;
len--;
}
/* Check that the data is well-formed UTF-8 if we're in UTF mode. To create
invalid input to pcre2_match(), you must use \x?? or \x{} sequences. */
if (utf)
{
uint8_t *q;
uint32_t cc;
int n = 1;
uint8_t *q_end = p + len;
for (q = p; n > 0 && *q; q += n) n = utf82ord(q, q_end, &cc);
if (n <= 0)
{
fprintf(outfile, "** Failed: invalid UTF-8 string cannot be used as input "
"in UTF mode\n");
return PR_OK;
}
}
#ifdef SUPPORT_VALGRIND
/* Mark the dbuffer as addressable but undefined again. */
if (dbuffer != NULL)
{
VALGRIND_MAKE_MEM_UNDEFINED(dbuffer, dbuffer_size);
}
#endif
/* Allocate a buffer to hold the data line; len+1 is an upper bound on
the number of code units that will be needed (though the buffer may have to be
extended if replication is involved). */
needlen = (len+1) * code_unit_size;
if (dbuffer == NULL || needlen >= dbuffer_size)
{
while (needlen >= dbuffer_size)
{
if (dbuffer_size < SIZE_MAX/2) dbuffer_size *= 2;
else dbuffer_size = needlen + 1;
}
dbuffer = (uint8_t *)realloc(dbuffer, dbuffer_size);
if (dbuffer == NULL)
{
fprintf(stderr, "pcre2test: realloc(%" SIZ_FORM ") failed\n", dbuffer_size);
exit(1);
}
}
SETCASTPTR(q, dbuffer); /* Sets q8, q16, or q32, as appropriate. */
/* Scan the data line, interpreting data escapes, and put the result into a
buffer of the appropriate width. In UTF mode, input is always UTF-8; otherwise,
in 16- and 32-bit modes, it can be forced to UTF-8 by the utf8_input modifier.
*/
while ((c = *p++) != 0)
{
int i = 0;
size_t replen;
enum force_encoding encoding = FORCE_NONE;
/* ] may mark the end of a replicated sequence */
if (c == ']' && start_rep != NULL)
{
PCRE2_SIZE d;
long li;
char *endptr;
if (*p++ != '{')
{
fprintf(outfile, "** Expected '{' after \\[....]\n");
return PR_OK;
}
li = strtol((const char *)p, &endptr, 10);
if (S32OVERFLOW(li))
{
fprintf(outfile, "** Repeat count too large\n");
return PR_OK;
}
i = (int)li;
p = (uint8_t *)endptr;
if (*p++ != '}')
{
fprintf(outfile, "** Expected '}' after \\[...]{...\n");
return PR_OK;
}
if (i-- <= 0)
{
fprintf(outfile, "** Zero or negative repeat not allowed\n");
return PR_OK;
}
replen = CAST8VAR(q) - start_rep;
if (PRIV(ckd_smul)(&d, replen, i))
{
fprintf(outfile, "** Expanded content too large\n");
return PR_OK;
}
needlen += d;
if (needlen >= dbuffer_size)
{
size_t qoffset = CAST8VAR(q) - dbuffer;
size_t rep_offset = start_rep - dbuffer;
while (needlen >= dbuffer_size)
{
if (dbuffer_size < SIZE_MAX/2) dbuffer_size *= 2;
else dbuffer_size = needlen + 1;
}
dbuffer = (uint8_t *)realloc(dbuffer, dbuffer_size);
if (dbuffer == NULL)
{
fprintf(stderr, "pcre2test: realloc(%" SIZ_FORM ") failed\n",
dbuffer_size);
exit(1);
}
SETCASTPTR(q, dbuffer + qoffset);
start_rep = dbuffer + rep_offset;
}
while (i-- > 0)
{
memcpy(CAST8VAR(q), start_rep, replen);
SETPLUS(q, replen/code_unit_size);
}
start_rep = NULL;
continue;
}
/* Handle a non-escaped character. In non-UTF 32-bit mode with utf8_input
set, do the fudge for setting the top bit. */
if (c != '\\' || subject_literal)
{
uint32_t topbit = 0;
if (test_mode == PCRE32_MODE && c == 0xff && *p != 0)
{
topbit = 0x80000000;
c = *p++;
}
if ((utf || (pat_patctl.control & CTL_UTF8_INPUT) != 0) &&
HASUTF8EXTRALEN(c)) { GETUTF8INC(c, p); }
c |= topbit;
}
/* Handle backslash escapes */
else switch ((c = *p++))
{
case '\\': break;
case 'a': c = CHAR_BEL; break;
case 'b': c = '\b'; break;
case 'e': c = CHAR_ESC; break;
case 'f': c = '\f'; break;
case 'n': c = '\n'; break;
case 'r': c = '\r'; break;
case 't': c = '\t'; break;
case 'v': c = '\v'; break;
case '0': case '1': case '2': case '3':
case '4': case '5': case '6': case '7':
c -= '0';
while (i++ < 2 && isdigit(*p) && *p < '8')
c = c * 8 + (*p++ - '0');
encoding = (utf && c > 255)? FORCE_UTF : FORCE_RAW;
break;
case 'o':
if (*p == '{')
{
uint8_t *pt = p;
c = 0;
for (pt++; isdigit(*pt) && *pt < '8'; ++i, pt++)
{
if (c >= 0x20000000u)
{
fprintf(outfile, "** \\o{ escape too large\n");
return PR_OK;
}
else c = c * 8 + (*pt - '0');
}
if (i == 0 || *pt != '}')
{
fprintf(outfile, "** Malformed \\o{ escape\n");
return PR_OK;
}
else p = pt + 1;
}
break;
case 'x':
c = 0;
if (*p == '{')
{
uint8_t *pt = p;
/* We used to have "while (isxdigit(*(++pt)))" here, but it fails
when isxdigit() is a macro that refers to its argument more than
once. This is banned by the C Standard, but apparently happens in at
least one macOS environment. */
for (pt++; isxdigit(*pt); pt++)
{
if (++i == 9)
{
fprintf(outfile, "** Too many hex digits in \\x{...} item; "
"using only the first eight.\n");
while (isxdigit(*pt)) pt++;
break;
}
else c = c * 16 + (tolower(*pt) - (isdigit(*pt)? '0' : 'a' - 10));
}
if (i == 0 || *pt != '}')
{
fprintf(outfile, "** Malformed \\x{ escape\n");
return PR_OK;
}
else p = pt + 1;
}
else
{
/* \x without {} always defines just one byte in 8-bit mode. This
allows UTF-8 characters to be constructed byte by byte, and also allows
invalid UTF-8 sequences to be made. Just copy the byte in UTF-8 mode.
Otherwise, pass it down as data. */
while (i++ < 2 && isxdigit(*p))
{
c = c * 16 + (tolower(*p) - (isdigit(*p)? '0' : 'a' - 10));
p++;
}
#if defined SUPPORT_PCRE2_8
if (utf && (test_mode == PCRE8_MODE)) encoding = FORCE_RAW;
#endif
}
break;
case 'N':
if (memcmp(p, "{U+", 3) == 0 && isxdigit(p[3]))
{
char *endptr;
unsigned long uli;
p += 3;
errno = 0;
uli = strtoul((const char *)p, &endptr, 16);
if (errno == 0 && *endptr == '}' && uli <= UINT32_MAX)
{
c = (uint32_t)uli;
p = (uint8_t *)endptr + 1;
encoding = FORCE_UTF;
break;
}
}
fprintf(outfile, "** Malformed \\N{U+ escape\n");
return PR_OK;
case 0: /* \ followed by EOF allows for an empty line */
p--;
continue;
case '=': /* \= terminates the data, starts modifiers */
goto ENDSTRING;
case '[': /* \[ introduces a replicated character sequence */
if (start_rep != NULL)
{
fprintf(outfile, "** Nested replication is not supported\n");
return PR_OK;
}
start_rep = CAST8VAR(q);
continue;
default:
if (isalnum(c))
{
fprintf(outfile, "** Unrecognized escape sequence \"\\%c\"\n", c);
return PR_OK;
}
}
/* We now have a character value in c that may be greater than 255.
Depending of how we got it, the encoding enum could be set to tell
us how to encode it, otherwise follow the utf modifier. */
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE)
{
if (encoding == FORCE_RAW || !(utf || encoding == FORCE_UTF))
{
if (c > 0xffu)
{
fprintf(outfile, "** Character \\x{%x} is greater than 255 "
"and UTF-8 mode is not enabled.\n", c);
fprintf(outfile, "** Truncation will probably give the wrong "
"result.\n");
}
*q8++ = (uint8_t)c;
}
else
{
if (c > 0x7fffffff)
{
fprintf(outfile, "** Character \\N{U+%x} is greater than 0x7fffffff "
"and therefore cannot be encoded as UTF-8\n", c);
return PR_OK;
}
else if (encoding == FORCE_UTF && c > MAX_UTF_CODE_POINT)
fprintf(outfile, "** Warning: character \\N{U+%x} is greater than "
"0x%x and should not be encoded as UTF-8\n",
c, MAX_UTF_CODE_POINT);
q8 += ord2utf8(c, q8);
}
}
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE)
{
/* Unlike the 8-bit code, there are no forced raw suggestions for the
16-bit mode, so assume raw unless utf is preferred */
if (!(encoding == FORCE_UTF || utf))
{
if (c > 0xffffu)
{
fprintf(outfile, "** Character \\x{%x} is greater than 0xffff "
"and UTF-16 mode is not enabled.\n", c);
fprintf(outfile, "** Truncation will probably give the wrong "
"result.\n");
}
*q16++ = (uint16_t)c;
}
else
{
if (c > MAX_UTF_CODE_POINT)
{
fprintf(outfile, "** Failed: character \\N{U+%x} is greater than "
"0x%x and therefore cannot be encoded as UTF-16\n",
c, MAX_UTF_CODE_POINT);
return PR_OK;
}
else if (c >= 0x10000u)
{
c -= 0x10000u;
*q16++ = 0xD800 | (c >> 10);
*q16++ = 0xDC00 | (c & 0x3ff);
}
else
{
if (encoding == FORCE_UTF && 0xe000u > c && c >= 0xd800u)
fprintf(outfile, "** Warning: character \\N{U+%x} is a surrogate "
"and should not be encoded as UTF-16\n", c);
*q16++ = c;
}
}
}
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE)
{
if (encoding == FORCE_UTF && c > MAX_UTF_CODE_POINT)
fprintf(outfile, "** Warning: character \\N{U+%x} is greater than "
"0x%x and should not be encoded as UTF-32\n",
c, MAX_UTF_CODE_POINT);
*q32++ = c;
}
#endif
}
ENDSTRING:
SET(*q, 0);
len = CASTVAR(uint8_t *, q) - dbuffer; /* Length in bytes */
ulen = len/code_unit_size; /* Length in code units */
arg_ulen = ulen; /* Value to use in match arg */
/* If the string was terminated by \= we must now interpret modifiers. */
if (p[-1] != 0 && !decode_modifiers(p, CTX_DAT, NULL, &dat_datctl))
return PR_OK;
/* Setting substitute_{skip,fail} implies a substitute callout. */
if (dat_datctl.substitute_skip != 0 || dat_datctl.substitute_stop != 0)
dat_datctl.control2 |= CTL2_SUBSTITUTE_CALLOUT;
/* Check for mutually exclusive modifiers. At present, these are all in the
first control word. */
for (k = 0; k < sizeof(exclusive_dat_controls)/sizeof(uint32_t); k++)
{
c = dat_datctl.control & exclusive_dat_controls[k];
if (c != 0 && c != (c & (~c+1)))
{
show_controls(c, 0, "** Not allowed together:");
fprintf(outfile, "\n");
return PR_OK;
}
}
if (dat_datctl.replacement[0] != 0)
{
if ((dat_datctl.control2 & CTL2_SUBSTITUTE_CALLOUT) != 0 &&
(dat_datctl.control & CTL_NULLCONTEXT) != 0)
{
fprintf(outfile, "** Replacement callouts are not supported with null_context.\n");
return PR_OK;
}
if ((dat_datctl.control2 & CTL2_SUBSTITUTE_CASE_CALLOUT) != 0 &&
(dat_datctl.control & CTL_NULLCONTEXT) != 0)
{
fprintf(outfile, "** Replacement case callouts are not supported with null_context.\n");
return PR_OK;
}
if ((dat_datctl.control & CTL_ALLCAPTURES) != 0)
fprintf(outfile, "** Ignored with replacement text: allcaptures\n");
}
/* Warn for modifiers that are ignored for DFA. */
if ((dat_datctl.control & CTL_DFA) != 0)
{
if ((dat_datctl.control & CTL_ALLCAPTURES) != 0)
fprintf(outfile, "** Ignored for DFA matching: allcaptures\n");
if ((dat_datctl.control2 & CTL2_HEAPFRAMES_SIZE) != 0)
fprintf(outfile, "** Ignored for DFA matching: heapframes_size\n");
}
/* We now have the subject in dbuffer, with len containing the byte length, and
ulen containing the code unit length, with a copy in arg_ulen for use in match
function arguments (this gets changed to PCRE2_ZERO_TERMINATED when the
zero_terminate modifier is present).
Move the data to the end of the buffer so that a read over the end can be
caught by valgrind or other means. If we have explicit valgrind support, mark
the unused start of the buffer unaddressable. If we are using the POSIX
interface, or testing zero-termination, we must include the terminating zero in
the usable data. */
c = code_unit_size * (((pat_patctl.control & CTL_POSIX) +
(dat_datctl.control & CTL_ZERO_TERMINATE) != 0)? 1:0);
pp = memmove(dbuffer + dbuffer_size - len - c, dbuffer, len + c);
#ifdef SUPPORT_VALGRIND
VALGRIND_MAKE_MEM_NOACCESS(dbuffer, dbuffer_size - (len + c));
#endif
/* Now pp points to the subject string, but if null_subject was specified, set
it to NULL to test PCRE2's behaviour. */
if ((dat_datctl.control2 & CTL2_NULL_SUBJECT) != 0) pp = NULL;
/* POSIX matching is only possible in 8-bit mode, and it does not support
timing or other fancy features. Some were checked at compile time, but we need
to check the match-time settings here. */
#ifdef SUPPORT_PCRE2_8
if ((pat_patctl.control & CTL_POSIX) != 0)
{
int rc;
int eflags = 0;
regmatch_t *pmatch = NULL;
const char *msg = "** Ignored with POSIX interface:";
if (dat_datctl.cerror[0] != CFORE_UNSET || dat_datctl.cerror[1] != CFORE_UNSET)
prmsg(&msg, "callout_error");
if (dat_datctl.cfail[0] != CFORE_UNSET || dat_datctl.cfail[1] != CFORE_UNSET)
prmsg(&msg, "callout_fail");
if (dat_datctl.copy_numbers[0] >= 0 || dat_datctl.copy_names[0] != 0)
prmsg(&msg, "copy");
if (dat_datctl.get_numbers[0] >= 0 || dat_datctl.get_names[0] != 0)
prmsg(&msg, "get");
if (dat_datctl.jitstack != 0) prmsg(&msg, "jitstack");
if (dat_datctl.offset != 0) prmsg(&msg, "offset");
if ((dat_datctl.options & ~POSIX_SUPPORTED_MATCH_OPTIONS) != 0)
{
fprintf(outfile, "%s", msg);
show_match_options(dat_datctl.options & ~POSIX_SUPPORTED_MATCH_OPTIONS);
msg = "";
}
if ((dat_datctl.control & ~POSIX_SUPPORTED_MATCH_CONTROLS) != 0 ||
(dat_datctl.control2 & ~POSIX_SUPPORTED_MATCH_CONTROLS2) != 0)
{
show_controls(dat_datctl.control & ~POSIX_SUPPORTED_MATCH_CONTROLS,
dat_datctl.control2 & ~POSIX_SUPPORTED_MATCH_CONTROLS2, msg);
msg = "";
}
if (msg[0] == 0) fprintf(outfile, "\n");
if (dat_datctl.oveccount > 0)
{
pmatch = (regmatch_t *)malloc(sizeof(regmatch_t) * dat_datctl.oveccount);
if (pmatch == NULL)
{
fprintf(outfile, "** Failed to get memory for recording matching "
"information (size set = %du)\n", dat_datctl.oveccount);
return PR_OK;
}
}
if (dat_datctl.startend[0] != CFORE_UNSET)
{
pmatch[0].rm_so = dat_datctl.startend[0];
pmatch[0].rm_eo = (dat_datctl.startend[1] != 0)?
dat_datctl.startend[1] : len;
eflags |= REG_STARTEND;
}
if ((dat_datctl.options & PCRE2_NOTBOL) != 0) eflags |= REG_NOTBOL;
if ((dat_datctl.options & PCRE2_NOTEOL) != 0) eflags |= REG_NOTEOL;
if ((dat_datctl.options & PCRE2_NOTEMPTY) != 0) eflags |= REG_NOTEMPTY;
rc = regexec(&preg, (const char *)pp, dat_datctl.oveccount, pmatch, eflags);
if (rc != 0)
{
(void)regerror(rc, &preg, (char *)pbuffer8, pbuffer8_size);
fprintf(outfile, "No match: POSIX code %d: %s\n", rc, pbuffer8);
}
else if ((pat_patctl.control & CTL_POSIX_NOSUB) != 0)
fprintf(outfile, "Matched with REG_NOSUB\n");
else if (dat_datctl.oveccount == 0)
fprintf(outfile, "Matched without capture\n");
else
{
size_t i, j;
size_t last_printed = (size_t)dat_datctl.oveccount;
for (i = 0; i < (size_t)dat_datctl.oveccount; i++)
{
if (pmatch[i].rm_so >= 0)
{
PCRE2_SIZE start = pmatch[i].rm_so;
PCRE2_SIZE end = pmatch[i].rm_eo;
for (j = last_printed + 1; j < i; j++)
fprintf(outfile, "%2d: \n", (int)j);
last_printed = i;
if (start > end)
{
start = pmatch[i].rm_eo;
end = pmatch[i].rm_so;
fprintf(outfile, "Start of matched string is beyond its end - "
"displaying from end to start.\n");
}
fprintf(outfile, "%2d: ", (int)i);
PCHARSV(pp, start, end - start, utf, outfile);
fprintf(outfile, "\n");
if ((i == 0 && (dat_datctl.control & CTL_AFTERTEXT) != 0) ||
(dat_datctl.control & CTL_ALLAFTERTEXT) != 0)
{
fprintf(outfile, "%2d+ ", (int)i);
/* Note: don't use the start/end variables here because we want to
show the text from what is reported as the end. */
PCHARSV(pp, pmatch[i].rm_eo, len - pmatch[i].rm_eo, utf, outfile);
fprintf(outfile, "\n"); }
}
}
}
free(pmatch);
return PR_OK;
}
#endif /* SUPPORT_PCRE2_8 */
/* Handle matching via the native interface. Check for consistency of
modifiers. */
if (dat_datctl.startend[0] != CFORE_UNSET)
fprintf(outfile, "** \\=posix_startend ignored for non-POSIX matching\n");
/* ALLUSEDTEXT is not supported with JIT, but JIT is not used with DFA
matching, even if the JIT compiler was used. */
if ((dat_datctl.control & (CTL_ALLUSEDTEXT|CTL_DFA)) == CTL_ALLUSEDTEXT &&
FLD(compiled_code, executable_jit) != NULL)
{
fprintf(outfile, "** Showing all consulted text is not supported by JIT: ignored\n");
dat_datctl.control &= ~CTL_ALLUSEDTEXT;
}
/* Handle passing the subject as zero-terminated. */
if ((dat_datctl.control & CTL_ZERO_TERMINATE) != 0)
arg_ulen = PCRE2_ZERO_TERMINATED;
/* The nullcontext modifier is used to test calling pcre2_[jit_]match() with a
NULL context. */
use_dat_context = ((dat_datctl.control & CTL_NULLCONTEXT) != 0)?
NULL : PTR(dat_context);
/* Enable display of malloc/free if wanted. We can do this only if either the
pattern or the subject is processed with a context. */
show_memory = (dat_datctl.control & CTL_MEMORY) != 0;
if (show_memory &&
(pat_patctl.control & dat_datctl.control & CTL_NULLCONTEXT) != 0)
fprintf(outfile, "** \\=memory requires either a pattern or a subject "
"context: ignored\n");
/* Create and assign a JIT stack if requested. */
if (dat_datctl.jitstack != 0)
{
if (dat_datctl.jitstack != jit_stack_size)
{
PCRE2_JIT_STACK_FREE(jit_stack);
PCRE2_JIT_STACK_CREATE(jit_stack, 1, dat_datctl.jitstack * 1024, NULL);
jit_stack_size = dat_datctl.jitstack;
}
PCRE2_JIT_STACK_ASSIGN(dat_context, jit_callback, jit_stack);
}
/* Or de-assign */
else if (jit_stack != NULL)
{
PCRE2_JIT_STACK_ASSIGN(dat_context, NULL, NULL);
PCRE2_JIT_STACK_FREE(jit_stack);
jit_stack = NULL;
jit_stack_size = 0;
}
/* When no JIT stack is assigned, we must ensure that there is a JIT callback
if we want to verify that JIT was actually used. */
if ((pat_patctl.control & CTL_JITVERIFY) != 0 && jit_stack == NULL)
{
PCRE2_JIT_STACK_ASSIGN(dat_context, jit_callback, NULL);
}
/* Adjust match_data according to size of offsets required. A size of zero
causes a new match data block to be obtained that exactly fits the pattern. */
if (dat_datctl.oveccount == 0)
{
PCRE2_MATCH_DATA_FREE(match_data);
PCRE2_MATCH_DATA_CREATE_FROM_PATTERN(match_data, compiled_code,
general_context);
PCRE2_GET_OVECTOR_COUNT(max_oveccount, match_data);
}
else if (dat_datctl.oveccount <= max_oveccount)
{
SETFLD(match_data, oveccount, dat_datctl.oveccount);
}
else
{
max_oveccount = dat_datctl.oveccount;
PCRE2_MATCH_DATA_FREE(match_data);
PCRE2_MATCH_DATA_CREATE(match_data, max_oveccount, general_context);
}
if (CASTVAR(void *, match_data) == NULL)
{
fprintf(outfile, "** Failed to get memory for recording matching "
"information (size requested: %d)\n", dat_datctl.oveccount);
max_oveccount = 0;
return PR_OK;
}
ovector = FLD(match_data, ovector);
PCRE2_GET_OVECTOR_COUNT(oveccount, match_data);
/* Replacement processing is ignored for DFA matching. */
if (dat_datctl.replacement[0] != 0 && (dat_datctl.control & CTL_DFA) != 0)
{
fprintf(outfile, "** Ignored for DFA matching: replace\n");
dat_datctl.replacement[0] = 0;
}
/* If a replacement string is provided, call pcre2_substitute() instead of or
after one of the matching functions. First we have to convert the replacement
string to the appropriate width. */
if (dat_datctl.replacement[0] != 0)
{
int rc;
uint8_t *pr;
uint8_t rbuffer[REPLACE_BUFFSIZE];
uint8_t nbuffer[REPLACE_BUFFSIZE];
uint8_t *rbptr;
uint32_t xoptions;
uint32_t emoption; /* External match option */
PCRE2_SIZE j, rlen, nsize, erroroffset;
BOOL badutf = FALSE;
#ifdef SUPPORT_PCRE2_8
uint8_t *r8 = NULL;
#endif
#ifdef SUPPORT_PCRE2_16
uint16_t *r16 = NULL;
#endif
#ifdef SUPPORT_PCRE2_32
uint32_t *r32 = NULL;
#endif
/* Fill the ovector with junk to detect elements that do not get set
when they should be (relevant only when "allvector" is specified). */
for (j = 0; j < 2*oveccount; j++) ovector[j] = JUNK_OFFSET;
if (timeitm)
fprintf(outfile, "** Timing is not supported with replace: ignored\n");
if ((dat_datctl.control & CTL_ALTGLOBAL) != 0)
fprintf(outfile, "** Altglobal is not supported with replace: ignored\n");
/* Check for a test that does substitution after an initial external match.
If this is set, we run the external match, but leave the interpretation of
its output to pcre2_substitute(). */
emoption = ((dat_datctl.control2 & CTL2_SUBSTITUTE_MATCHED) == 0)? 0 :
PCRE2_SUBSTITUTE_MATCHED;
if (emoption != 0)
{
if ((pat_patctl.control & CTL_JITFAST) != 0)
{
PCRE2_JIT_MATCH(rc, compiled_code, pp, arg_ulen, dat_datctl.offset,
dat_datctl.options, match_data, use_dat_context);
}
else
{
PCRE2_MATCH(rc, compiled_code, pp, arg_ulen, dat_datctl.offset,
dat_datctl.options, match_data, use_dat_context);
}
}
xoptions = emoption |
(((dat_datctl.control & CTL_GLOBAL) == 0)? 0 :
PCRE2_SUBSTITUTE_GLOBAL) |
(((dat_datctl.control2 & CTL2_SUBSTITUTE_EXTENDED) == 0)? 0 :
PCRE2_SUBSTITUTE_EXTENDED) |
(((dat_datctl.control2 & CTL2_SUBSTITUTE_LITERAL) == 0)? 0 :
PCRE2_SUBSTITUTE_LITERAL) |
(((dat_datctl.control2 & CTL2_SUBSTITUTE_OVERFLOW_LENGTH) == 0)? 0 :
PCRE2_SUBSTITUTE_OVERFLOW_LENGTH) |
(((dat_datctl.control2 & CTL2_SUBSTITUTE_REPLACEMENT_ONLY) == 0)? 0 :
PCRE2_SUBSTITUTE_REPLACEMENT_ONLY) |
(((dat_datctl.control2 & CTL2_SUBSTITUTE_UNKNOWN_UNSET) == 0)? 0 :
PCRE2_SUBSTITUTE_UNKNOWN_UNSET) |
(((dat_datctl.control2 & CTL2_SUBSTITUTE_UNSET_EMPTY) == 0)? 0 :
PCRE2_SUBSTITUTE_UNSET_EMPTY);
SETCASTPTR(r, rbuffer); /* Sets r8, r16, or r32, as appropriate. */
pr = dat_datctl.replacement;
/* If the replacement starts with '[]' we interpret that as length
value for the replacement buffer. */
nsize = REPLACE_BUFFSIZE/code_unit_size;
if (*pr == '[')
{
PCRE2_SIZE n = 0;
while ((c = *(++pr)) >= CHAR_0 && c <= CHAR_9) n = n * 10 + (c - CHAR_0);
if (*pr++ != ']')
{
fprintf(outfile, "Bad buffer size in replacement string\n");
return PR_OK;
}
if (n > nsize)
{
fprintf(outfile, "Replacement buffer setting (%" SIZ_FORM ") is too "
"large (max %" SIZ_FORM ")\n", n, nsize);
return PR_OK;
}
nsize = n;
}
/* Now copy the replacement string to a buffer of the appropriate width. No
escape processing is done for replacements. In UTF mode, check for an invalid
UTF-8 input string, and if it is invalid, just copy its code units without
UTF interpretation. This provides a means of checking that an invalid string
is detected. Otherwise, UTF-8 can be used to include wide characters in a
replacement. */
if (utf) badutf = valid_utf(pr, strlen((const char *)pr), &erroroffset);
/* Not UTF or invalid UTF-8: just copy the code units. */
if (!utf || badutf)
{
while ((c = *pr++) != 0)
{
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE) *r8++ = c;
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE) *r16++ = c;
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE) *r32++ = c;
#endif
}
}
/* Valid UTF-8 replacement string */
else while ((c = *pr++) != 0)
{
if (HASUTF8EXTRALEN(c)) { GETUTF8INC(c, pr); }
#ifdef SUPPORT_PCRE2_8
if (test_mode == PCRE8_MODE) r8 += ord2utf8(c, r8);
#endif
#ifdef SUPPORT_PCRE2_16
if (test_mode == PCRE16_MODE)
{
if (c >= 0x10000u)
{
c-= 0x10000u;
*r16++ = 0xD800 | (c >> 10);
*r16++ = 0xDC00 | (c & 0x3ff);
}
else *r16++ = c;
}
#endif
#ifdef SUPPORT_PCRE2_32
if (test_mode == PCRE32_MODE) *r32++ = c;
#endif
}
SET(*r, 0);
if ((dat_datctl.control & CTL_ZERO_TERMINATE) != 0)
rlen = PCRE2_ZERO_TERMINATED;
else
rlen = (CASTVAR(uint8_t *, r) - rbuffer)/code_unit_size;
if ((dat_datctl.control2 & CTL2_SUBSTITUTE_CALLOUT) != 0)
{
PCRE2_SET_SUBSTITUTE_CALLOUT(dat_context, substitute_callout_function, NULL);
}
else
{
PCRE2_SET_SUBSTITUTE_CALLOUT(dat_context, NULL, NULL); /* No callout */
}
if ((dat_datctl.control2 & CTL2_SUBSTITUTE_CASE_CALLOUT) != 0)
{
PCRE2_SET_SUBSTITUTE_CASE_CALLOUT(dat_context, substitute_case_callout_function, NULL);
}
else
{
PCRE2_SET_SUBSTITUTE_CASE_CALLOUT_NULL(dat_context); /* No callout */
}
/* There is a special option to set the replacement to NULL in order to test
that case. */
rbptr = ((dat_datctl.control2 & CTL2_NULL_REPLACEMENT) == 0)? rbuffer : NULL;
PCRE2_SUBSTITUTE(rc, compiled_code, pp, arg_ulen, dat_datctl.offset,
dat_datctl.options|xoptions, match_data, use_dat_context,
rbptr, rlen, nbuffer, &nsize);
if (rc < 0)
{
fprintf(outfile, "Failed: error %d", rc);
if (rc != PCRE2_ERROR_NOMEMORY && nsize != PCRE2_UNSET)
fprintf(outfile, " at offset %ld in replacement", (long int)nsize);
fprintf(outfile, ": ");
if (!print_error_message(rc, "", "")) return PR_ABEND;
if (rc == PCRE2_ERROR_NOMEMORY &&
(xoptions & PCRE2_SUBSTITUTE_OVERFLOW_LENGTH) != 0)
fprintf(outfile, ": %ld code units are needed", (long int)nsize);
}
else
{
fprintf(outfile, "%2d: ", rc);
PCHARSV(nbuffer, 0, nsize, utf, outfile);
}
fprintf(outfile, "\n");
show_memory = FALSE;
/* Show final ovector contents and resulting heapframe size if requested. */
if ((dat_datctl.control2 & CTL2_ALLVECTOR) != 0)
show_ovector(ovector, oveccount);
if ((dat_datctl.control2 & CTL2_HEAPFRAMES_SIZE) != 0 &&
(dat_datctl.control & CTL_DFA) == 0)
show_heapframes_size();
return PR_OK;
} /* End of substitution handling */
/* When a replacement string is not provided, run a loop for global matching
with one of the basic matching functions. For altglobal (or first time round
the loop), set an "unset" value for the previous match info. */
ovecsave[0] = ovecsave[1] = ovecsave[2] = PCRE2_UNSET;
for (gmatched = 0;; gmatched++)
{
PCRE2_SIZE j;
int capcount;
/* Fill the ovector with junk to detect elements that do not get set
when they should be. */
for (j = 0; j < 2*oveccount; j++) ovector[j] = JUNK_OFFSET;
/* When matching is via pcre2_match(), we will detect the use of JIT via the
stack callback function. */
jit_was_used = (pat_patctl.control & CTL_JITFAST) != 0;
/* Do timing if required. */
if (timeitm > 0)
{
int i;
clock_t start_time, time_taken;
if ((dat_datctl.control & CTL_DFA) != 0)
{
if ((dat_datctl.options & PCRE2_DFA_RESTART) != 0)
{
fprintf(outfile, "Timing DFA restarts is not supported\n");
return PR_OK;
}
if (dfa_workspace == NULL)
dfa_workspace = (int *)malloc(DFA_WS_DIMENSION*sizeof(int));
start_time = clock();
for (i = 0; i < timeitm; i++)
{
PCRE2_DFA_MATCH(capcount, compiled_code, pp, arg_ulen,
dat_datctl.offset, dat_datctl.options | g_notempty, match_data,
use_dat_context, dfa_workspace, DFA_WS_DIMENSION);
}
}
else if ((pat_patctl.control & CTL_JITFAST) != 0)
{
start_time = clock();
for (i = 0; i < timeitm; i++)
{
PCRE2_JIT_MATCH(capcount, compiled_code, pp, arg_ulen,
dat_datctl.offset, dat_datctl.options | g_notempty, match_data,
use_dat_context);
}
}
else
{
start_time = clock();
for (i = 0; i < timeitm; i++)
{
PCRE2_MATCH(capcount, compiled_code, pp, arg_ulen,
dat_datctl.offset, dat_datctl.options | g_notempty, match_data,
use_dat_context);
}
}
total_match_time += (time_taken = clock() - start_time);
fprintf(outfile, "Match time %7.4f microseconds\n",
((1000000 / CLOCKS_PER_SEC) * (double)time_taken) / timeitm);
}
/* Find the heap, match and depth limits if requested. The depth and heap
limits are not relevant for JIT. The return from check_match_limit() is the
return from the final call to pcre2_match() or pcre2_dfa_match(). */
if ((dat_datctl.control & (CTL_FINDLIMITS|CTL_FINDLIMITS_NOHEAP)) != 0)
{
capcount = 0; /* This stops compiler warnings */
(void)capcount;
if ((dat_datctl.control & CTL_FINDLIMITS_NOHEAP) == 0 &&
(FLD(compiled_code, executable_jit) == NULL ||
(dat_datctl.options & PCRE2_NO_JIT) != 0))
{
(void)check_match_limit(pp, arg_ulen, PCRE2_ERROR_HEAPLIMIT, "heap");
}
capcount = check_match_limit(pp, arg_ulen, PCRE2_ERROR_MATCHLIMIT,
"match");
if (FLD(compiled_code, executable_jit) == NULL ||
(dat_datctl.options & PCRE2_NO_JIT) != 0 ||
(dat_datctl.control & CTL_DFA) != 0)
{
capcount = check_match_limit(pp, arg_ulen, PCRE2_ERROR_DEPTHLIMIT,
"depth");
}
if (capcount == 0)
{
fprintf(outfile, "Matched, but offsets vector is too small to show all matches\n");
capcount = dat_datctl.oveccount;
}
}
/* Otherwise just run a single match, setting up a callout if required (the
default). There is a copy of the pattern in pbuffer8 for use by callouts. */
else
{
if ((dat_datctl.control & CTL_CALLOUT_NONE) == 0)
{
PCRE2_SET_CALLOUT(dat_context, callout_function,
(void *)(&dat_datctl.callout_data));
first_callout = TRUE;
last_callout_mark = NULL;
callout_count = 0;
}
else
{
PCRE2_SET_CALLOUT(dat_context, NULL, NULL); /* No callout */
}
/* Run a single DFA or NFA match. */
if ((dat_datctl.control & CTL_DFA) != 0)
{
if (dfa_workspace == NULL)
dfa_workspace = (int *)malloc(DFA_WS_DIMENSION*sizeof(int));
if (dfa_matched++ == 0)
dfa_workspace[0] = -1; /* To catch bad restart */
PCRE2_DFA_MATCH(capcount, compiled_code, pp, arg_ulen,
dat_datctl.offset, dat_datctl.options | g_notempty, match_data,
use_dat_context, dfa_workspace, DFA_WS_DIMENSION);
if (capcount == 0)
{
fprintf(outfile, "Matched, but offsets vector is too small to show all matches\n");
capcount = dat_datctl.oveccount;
}
}
else
{
if ((pat_patctl.control & CTL_JITFAST) != 0)
PCRE2_JIT_MATCH(capcount, compiled_code, pp, arg_ulen, dat_datctl.offset,
dat_datctl.options | g_notempty, match_data, use_dat_context);
else
PCRE2_MATCH(capcount, compiled_code, pp, arg_ulen, dat_datctl.offset,
dat_datctl.options | g_notempty, match_data, use_dat_context);
if (capcount == 0)
{
fprintf(outfile, "Matched, but too many substrings\n");
capcount = dat_datctl.oveccount;
}
}
}
/* The result of the match is now in capcount. First handle a successful
match. If pp was forced to be NULL (to test NULL handling) it will have been
treated as an empty string if the length was zero. So re-create that for
outputting. Don't just point to "" because that leads to a "loss of const"
warning. */
if (capcount >= 0)
{
if (pp == NULL)
{
#ifdef SUPPORT_VALGRIND
/* Mark the start of dbuffer addressable again. */
VALGRIND_MAKE_MEM_UNDEFINED(dbuffer, 1);
#endif
pp = dbuffer;
pp[0] = 0;
}
if (capcount > (int)oveccount) /* Check for lunatic return value */
{
fprintf(outfile,
"** PCRE2 error: returned count %d is too big for ovector count %d\n",
capcount, oveccount);
capcount = oveccount;
if ((dat_datctl.control & CTL_ANYGLOB) != 0)
{
fprintf(outfile, "** Global loop abandoned\n");
dat_datctl.control &= ~CTL_ANYGLOB; /* Break g/G loop */
}
}
/* If PCRE2_COPY_MATCHED_SUBJECT was set, check that things are as they
should be, but not for fast JIT, where it isn't supported. */
if ((dat_datctl.options & PCRE2_COPY_MATCHED_SUBJECT) != 0 &&
(pat_patctl.control & CTL_JITFAST) == 0)
{
if ((FLD(match_data, flags) & PCRE2_MD_COPIED_SUBJECT) == 0)
fprintf(outfile,
"** PCRE2 error: flag not set after copy_matched_subject\n");
if (CASTFLD(const void *, match_data, subject) == pp)
fprintf(outfile,
"** PCRE2 error: copy_matched_subject has not copied\n");
if (memcmp(CASTFLD(const void *, match_data, subject), pp, ulen) != 0)
fprintf(outfile,
"** PCRE2 error: copy_matched_subject mismatch\n");
}
/* If this is not the first time round a global loop, check that the
returned string has changed. If it has not, check for an empty string match
at different starting offset from the previous match. This is a failed test
retry for null-matching patterns that don't match at their starting offset,
for example /(?<=\G.)/. A repeated match at the same point is not such a
pattern, and must be discarded, and we then proceed to seek a non-null
match at the current point. For any other repeated match, there is a bug
somewhere and we must break the loop because it will go on for ever. We
know that there are always at least two elements in the ovector. */
if (gmatched > 0 && ovecsave[0] == ovector[0] && ovecsave[1] == ovector[1])
{
if (ovector[0] == ovector[1] && ovecsave[2] != dat_datctl.offset)
{
g_notempty = PCRE2_NOTEMPTY_ATSTART | PCRE2_ANCHORED;
ovecsave[2] = dat_datctl.offset;
continue; /* Back to the top of the loop */
}
fprintf(outfile,
"** PCRE2 error: global repeat returned the same string as previous\n");
fprintf(outfile, "** Global loop abandoned\n");
dat_datctl.control &= ~CTL_ANYGLOB; /* Break g/G loop */
}
/* "allcaptures" requests showing of all captures in the pattern, to check
unset ones at the end. It may be set on the pattern or the data. Implement
by setting capcount to the maximum. This is not relevant for DFA matching,
so ignore it (warning given above). */
if ((dat_datctl.control & (CTL_ALLCAPTURES|CTL_DFA)) == CTL_ALLCAPTURES)
{
capcount = maxcapcount + 1; /* Allow for full match */
if (capcount > (int)oveccount) capcount = oveccount;
}
/* "allvector" request showing the entire ovector. */
if ((dat_datctl.control2 & CTL2_ALLVECTOR) != 0) capcount = oveccount;
/* Output the captured substrings. Note that, for the matched string,
the use of \K in an assertion can make the start later than the end. */
for (int i = 0; i < 2*capcount; i += 2)
{
PCRE2_SIZE lleft, lmiddle, lright;
PCRE2_SIZE start = ovector[i];
PCRE2_SIZE end = ovector[i+1];
if (start > end)
{
start = ovector[i+1];
end = ovector[i];
fprintf(outfile, "Start of matched string is beyond its end - "
"displaying from end to start.\n");
}
fprintf(outfile, "%2d: ", i/2);
/* Check for an unset group */
if (start == PCRE2_UNSET && end == PCRE2_UNSET)
{
fprintf(outfile, "\n");
continue;
}
/* Check for silly offsets, in particular, values that have not been
set when they should have been. However, if we are past the end of the
captures for this pattern ("allvector" causes this), or if we are DFA
matching, it isn't an error if the entry is unchanged. */
if (start > ulen || end > ulen)
{
if (((dat_datctl.control & CTL_DFA) != 0 ||
i >= (int)(2*maxcapcount + 2)) &&
start == JUNK_OFFSET && end == JUNK_OFFSET)
fprintf(outfile, "\n");
else
fprintf(outfile, "ERROR: bad value(s) for offset(s): 0x%lx 0x%lx\n",
(unsigned long int)start, (unsigned long int)end);
continue;
}
/* When JIT is not being used, ALLUSEDTEXT may be set. (It if is set with
JIT, it is disabled above, with a comment.) When the match is done by the
interpreter, leftchar and rightchar are available, and if ALLUSEDTEXT is
set, and if the leftmost consulted character is before the start of the
match or the rightmost consulted character is past the end of the match,
we want to show all consulted characters for the main matched string, and
indicate which were lookarounds. */
if (i == 0)
{
BOOL showallused;
PCRE2_SIZE leftchar, rightchar;
if ((dat_datctl.control & CTL_ALLUSEDTEXT) != 0)
{
leftchar = FLD(match_data, leftchar);
rightchar = FLD(match_data, rightchar);
showallused = i == 0 && (leftchar < start || rightchar > end);
}
else showallused = FALSE;
if (showallused)
{
PCHARS(lleft, pp, leftchar, start - leftchar, utf, outfile);
PCHARS(lmiddle, pp, start, end - start, utf, outfile);
PCHARS(lright, pp, end, rightchar - end, utf, outfile);
if ((pat_patctl.control & CTL_JITVERIFY) != 0 && jit_was_used)
fprintf(outfile, " (JIT)");
fprintf(outfile, "\n ");
for (j = 0; j < lleft; j++) fprintf(outfile, "<");
for (j = 0; j < lmiddle; j++) fprintf(outfile, " ");
for (j = 0; j < lright; j++) fprintf(outfile, ">");
}
/* When a pattern contains \K, the start of match position may be
different to the start of the matched string. When this is the case,
show it when requested. */
else if ((dat_datctl.control & CTL_STARTCHAR) != 0)
{
PCRE2_SIZE startchar;
PCRE2_GET_STARTCHAR(startchar, match_data);
PCHARS(lleft, pp, startchar, start - startchar, utf, outfile);
PCHARSV(pp, start, end - start, utf, outfile);
if ((pat_patctl.control & CTL_JITVERIFY) != 0 && jit_was_used)
fprintf(outfile, " (JIT)");
if (startchar != start)
{
fprintf(outfile, "\n ");
for (j = 0; j < lleft; j++) fprintf(outfile, "^");
}
}
/* Otherwise, just show the matched string. */
else
{
PCHARSV(pp, start, end - start, utf, outfile);
if ((pat_patctl.control & CTL_JITVERIFY) != 0 && jit_was_used)
fprintf(outfile, " (JIT)");
}
}
/* Not the main matched string. Just show it unadorned. */
else
{
PCHARSV(pp, start, end - start, utf, outfile);
}
fprintf(outfile, "\n");
/* Note: don't use the start/end variables here because we want to
show the text from what is reported as the end. */
if ((dat_datctl.control & CTL_ALLAFTERTEXT) != 0 ||
(i == 0 && (dat_datctl.control & CTL_AFTERTEXT) != 0))
{
fprintf(outfile, "%2d+ ", i/2);
PCHARSV(pp, ovector[i+1], ulen - ovector[i+1], utf, outfile);
fprintf(outfile, "\n");
}
}
/* Output (*MARK) data if requested */
if ((dat_datctl.control & CTL_MARK) != 0 &&
TESTFLD(match_data, mark, !=, NULL))
{
fprintf(outfile, "MK: ");
PCHARSV(CASTFLD(const void *, match_data, mark), -1, -1, utf, outfile);
fprintf(outfile, "\n");
}
/* Process copy/get strings */
if (!copy_and_get(utf, capcount)) return PR_ABEND;
} /* End of handling a successful match */
/* There was a partial match. The value of ovector[0] is the bumpalong point,
that is, startchar, not any \K point that might have been passed. When JIT is
not in use, "allusedtext" may be set, in which case we indicate the leftmost
consulted character. */
else if (capcount == PCRE2_ERROR_PARTIAL)
{
PCRE2_SIZE leftchar;
int backlength;
int rubriclength = 0;
if ((dat_datctl.control & CTL_ALLUSEDTEXT) != 0)
{
leftchar = FLD(match_data, leftchar);
}
else leftchar = ovector[0];
fprintf(outfile, "Partial match");
if ((dat_datctl.control & CTL_MARK) != 0 &&
TESTFLD(match_data, mark, !=, NULL))
{
fprintf(outfile, ", mark=");
PCHARS(rubriclength, CASTFLD(const void *, match_data, mark), -1, -1, utf,
outfile);
rubriclength += 7;
}
fprintf(outfile, ": ");
rubriclength += 15;
PCHARS(backlength, pp, leftchar, ovector[0] - leftchar, utf, outfile);
PCHARSV(pp, ovector[0], ovector[1] - ovector[0], utf, outfile);
if ((pat_patctl.control & CTL_JITVERIFY) != 0 && jit_was_used)
fprintf(outfile, " (JIT)");
fprintf(outfile, "\n");
if (backlength != 0)
{
for (int i = 0; i < rubriclength; i++) fprintf(outfile, " ");
for (int i = 0; i < backlength; i++) fprintf(outfile, "<");
fprintf(outfile, "\n");
}
if (ulen != ovector[1])
fprintf(outfile, "** ovector[1] is not equal to the subject length: "
"%ld != %ld\n", (unsigned long int)ovector[1], (unsigned long int)ulen);
/* Process copy/get strings */
if (!copy_and_get(utf, 1)) return PR_ABEND;
/* "allvector" outputs the entire vector */
if ((dat_datctl.control2 & CTL2_ALLVECTOR) != 0)
show_ovector(ovector, oveccount);
break; /* Out of the /g loop */
} /* End of handling partial match */
/* Failed to match. If this is a /g or /G loop, we might previously have
set g_notempty (to PCRE2_NOTEMPTY_ATSTART|PCRE2_ANCHORED) after a null match.
If that is the case, this is not necessarily the end. We want to advance the
start offset, and continue. We won't be at the end of the string - that was
checked before setting g_notempty. We achieve the effect by pretending that a
single character was matched.
Complication arises in the case when the newline convention is "any", "crlf",
or "anycrlf". If the previous match was at the end of a line terminated by
CRLF, an advance of one character just passes the CR, whereas we should
prefer the longer newline sequence, as does the code in pcre2_match().
Otherwise, in the case of UTF-8 or UTF-16 matching, the advance must be one
character, not one byte. */
else if (g_notempty != 0) /* There was a previous null match */
{
uint16_t nl = FLD(compiled_code, newline_convention);
PCRE2_SIZE start_offset = dat_datctl.offset; /* Where the match was */
PCRE2_SIZE end_offset = start_offset + 1;
if ((nl == PCRE2_NEWLINE_CRLF || nl == PCRE2_NEWLINE_ANY ||
nl == PCRE2_NEWLINE_ANYCRLF) &&
start_offset < ulen - 1 &&
CODE_UNIT(pp, start_offset) == '\r' &&
CODE_UNIT(pp, end_offset) == '\n')
end_offset++;
else if (utf && test_mode != PCRE32_MODE)
{
if (test_mode == PCRE8_MODE)
{
for (; end_offset < ulen; end_offset++)
if ((((PCRE2_SPTR8)pp)[end_offset] & 0xc0) != 0x80) break;
}
else /* 16-bit mode */
{
for (; end_offset < ulen; end_offset++)
if ((((PCRE2_SPTR16)pp)[end_offset] & 0xfc00) != 0xdc00) break;
}
}
SETFLDVEC(match_data, ovector, 0, start_offset);
SETFLDVEC(match_data, ovector, 1, end_offset);
} /* End of handling null match in a global loop */
/* A "normal" match failure. There will be a negative error number in
capcount. */
else
{
switch(capcount)
{
case PCRE2_ERROR_NOMATCH:
if (gmatched == 0)
{
fprintf(outfile, "No match");
if ((dat_datctl.control & CTL_MARK) != 0 &&
TESTFLD(match_data, mark, !=, NULL))
{
fprintf(outfile, ", mark = ");
PCHARSV(CASTFLD(const void *, match_data, mark), -1, -1, utf, outfile);
}
if ((pat_patctl.control & CTL_JITVERIFY) != 0 && jit_was_used)
fprintf(outfile, " (JIT)");
fprintf(outfile, "\n");
/* "allvector" outputs the entire vector */
if ((dat_datctl.control2 & CTL2_ALLVECTOR) != 0)
show_ovector(ovector, oveccount);
}
break;
case PCRE2_ERROR_BADUTFOFFSET:
fprintf(outfile, "Error %d (bad UTF-%d offset)\n", capcount, test_mode);
break;
default:
fprintf(outfile, "Failed: error %d: ", capcount);
if (!print_error_message(capcount, "", "")) return PR_ABEND;
if (capcount <= PCRE2_ERROR_UTF8_ERR1 &&
capcount >= PCRE2_ERROR_UTF32_ERR2)
{
PCRE2_SIZE startchar;
PCRE2_GET_STARTCHAR(startchar, match_data);
fprintf(outfile, " at offset %" SIZ_FORM, startchar);
}
fprintf(outfile, "\n");
break;
}
break; /* Out of the /g loop */
} /* End of failed match handling */
/* Control reaches here in two circumstances: (a) after a match, and (b)
after a non-match that immediately followed a match on an empty string when
doing a global search. Such a match is done with PCRE2_NOTEMPTY_ATSTART and
PCRE2_ANCHORED set in g_notempty. The code above turns it into a fake match
of one character. So effectively we get here only after a match. If we
are not doing a global search, we are done. */
if ((dat_datctl.control & CTL_ANYGLOB) == 0) break; else
{
PCRE2_SIZE match_offset = FLD(match_data, ovector)[0];
PCRE2_SIZE end_offset = FLD(match_data, ovector)[1];
/* We must now set up for the next iteration of a global search. If we have
matched an empty string, first check to see if we are at the end of the
subject. If so, the loop is over. Otherwise, mimic what Perl's /g option
does. Set PCRE2_NOTEMPTY_ATSTART and PCRE2_ANCHORED and try the match again
at the same point. If this fails it will be picked up above, where a fake
match is set up so that at this point we advance to the next character.
However, in order to cope with patterns that never match at their starting
offset (e.g. /(?<=\G.)/) we don't do this when the match offset is greater
than the starting offset. This means there will be a retry with the
starting offset at the match offset. If this returns the same match again,
it is picked up above and ignored, and the special action is then taken. */
if (match_offset == end_offset)
{
if (end_offset == ulen) break; /* End of subject */
if (match_offset <= dat_datctl.offset)
g_notempty = PCRE2_NOTEMPTY_ATSTART | PCRE2_ANCHORED;
}
/* However, even after matching a non-empty string, there is still one
tricky case. If a pattern contains \K within a lookbehind assertion at the
start, the end of the matched string can be at the offset where the match
started. In the case of a normal /g iteration without special action, this
leads to a loop that keeps on returning the same substring. The loop would
be caught above, but we really want to move on to the next match. */
else
{
g_notempty = 0; /* Set for a "normal" repeat */
if ((dat_datctl.control & CTL_GLOBAL) != 0)
{
PCRE2_SIZE startchar;
PCRE2_GET_STARTCHAR(startchar, match_data);
if (end_offset <= startchar)
{
if (startchar >= ulen) break; /* End of subject */
end_offset = startchar + 1;
if (utf && test_mode != PCRE32_MODE)
{
if (test_mode == PCRE8_MODE)
{
for (; end_offset < ulen; end_offset++)
if ((((PCRE2_SPTR8)pp)[end_offset] & 0xc0) != 0x80) break;
}
else /* 16-bit mode */
{
for (; end_offset < ulen; end_offset++)
if ((((PCRE2_SPTR16)pp)[end_offset] & 0xfc00) != 0xdc00) break;
}
}
}
}
}
/* For a normal global (/g) iteration, save the current ovector[0,1] and
the starting offset so that we can check that they do change each time.
Otherwise a matching bug that returns the same string causes an infinite
loop. It has happened! Then update the start offset, leaving other
parameters alone. */
if ((dat_datctl.control & CTL_GLOBAL) != 0)
{
ovecsave[0] = ovector[0];
ovecsave[1] = ovector[1];
ovecsave[2] = dat_datctl.offset;
dat_datctl.offset = end_offset;
}
/* For altglobal, just update the pointer and length. */
else
{
pp += end_offset * code_unit_size;
len -= end_offset * code_unit_size;
ulen -= end_offset;
if (arg_ulen != PCRE2_ZERO_TERMINATED) arg_ulen -= end_offset;
}
}
} /* End of global loop */
/* All matching is done; show the resulting heapframe size if requested. */
if ((dat_datctl.control2 & CTL2_HEAPFRAMES_SIZE) != 0 &&
(dat_datctl.control & CTL_DFA) == 0)
show_heapframes_size();
show_memory = FALSE;
return PR_OK;
}
/*************************************************
* Print PCRE2 version *
*************************************************/
static void
print_version(FILE *f, BOOL include_mode)
{
char buf[16];
VERSION_TYPE *vp;
fprintf(f, "PCRE2 version ");
for (vp = version; *vp != 0; vp++) fprintf(f, "%c", *vp);
if (include_mode)
{
sprintf(buf, "%d-bit", test_mode);
fprintf(f, " (%s)", buf);
}
fprintf(f, "\n");
}
/*************************************************
* Print Unicode version *
*************************************************/
static void
print_unicode_version(FILE *f)
{
VERSION_TYPE *vp;
fprintf(f, "Unicode version ");
for (vp = uversion; *vp != 0; vp++) fprintf(f, "%c", *vp);
}
/*************************************************
* Print JIT target *
*************************************************/
static void
print_jit_target(FILE *f)
{
VERSION_TYPE *vp;
for (vp = jittarget; *vp != 0; vp++) fprintf(f, "%c", *vp);
}
/*************************************************
* Print newline configuration *
*************************************************/
/* Output is always to stdout.
Arguments:
rc the return code from PCRE2_CONFIG_NEWLINE
isc TRUE if called from "-C newline"
Returns: nothing
*/
static void
print_newline_config(uint32_t optval, BOOL isc)
{
if (!isc) printf(" Default newline sequence is ");
if (optval < sizeof(newlines)/sizeof(char *))
printf("%s\n", newlines[optval]);
else
printf("a non-standard value: %d\n", optval);
}
/*************************************************
* Usage function *
*************************************************/
static void
usage(void)
{
printf("Usage: pcre2test [options] [ [