X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=d3e347e1df00a040cf10f910aa1def30bf38cd55;hb=0ecd3ba2f9cc10124e1b1548816d989cfc59a802;hp=2420f8d3c98f85bf028d0f031705efd6f0daf68e;hpb=5fcd1c1b5ea5ccb322c2ff87d283f032058dbafe;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index 2420f8d..d3e347e 100644 --- a/regexec.c +++ b/regexec.c @@ -392,7 +392,8 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, && !PL_multiline ) ); /* Check after \n? */ if (!ml_anch) { - if ( !(prog->reganch & ROPT_ANCH_GPOS) /* Checked by the caller */ + if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */ + | ROPT_IMPLICIT)) /* not a real BOL */ /* SvCUR is not set on references: SvRV and SvPVX overlap */ && sv && !SvROK(sv) && (strpos != strbeg)) { @@ -945,19 +946,19 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta /* FALL THROUGH */ case BOUND: if (do_utf8) { - if (s == startpos) + if (s == PL_bostr) tmp = '\n'; else { U8 *r = reghop3((U8*)s, -1, (U8*)startpos); - tmp = (I32)utf8n_to_uvuni(r, s - (char*)r, 0, 0); + tmp = (I32)utf8n_to_uvchr(r, s - (char*)r, 0, 0); } tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); LOAD_UTF8_CHARCLASS(alnum,"a"); while (s < strend) { if (tmp == !(OP(c) == BOUND ? - swash_fetch(PL_utf8_alnum, (U8*)s) : + swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : isALNUM_LC_utf8((U8*)s))) { tmp = !tmp; @@ -968,7 +969,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } } else { - tmp = (s != startpos) ? UCHARAT(s - 1) : '\n'; + tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); while (s < strend) { if (tmp == @@ -988,19 +989,19 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta /* FALL THROUGH */ case NBOUND: if (do_utf8) { - if (s == startpos) + if (s == PL_bostr) tmp = '\n'; else { U8 *r = reghop3((U8*)s, -1, (U8*)startpos); - tmp = (I32)utf8n_to_uvuni(r, s - (char*)r, 0, 0); + tmp = (I32)utf8n_to_uvchr(r, s - (char*)r, 0, 0); } tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); LOAD_UTF8_CHARCLASS(alnum,"a"); while (s < strend) { if (tmp == !(OP(c) == NBOUND ? - swash_fetch(PL_utf8_alnum, (U8*)s) : + swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : isALNUM_LC_utf8((U8*)s))) tmp = !tmp; else if ((norun || regtry(prog, s))) @@ -1009,7 +1010,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } } else { - tmp = (s != startpos) ? UCHARAT(s - 1) : '\n'; + tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); while (s < strend) { @@ -1028,7 +1029,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (do_utf8) { LOAD_UTF8_CHARCLASS(alnum,"a"); while (s < strend) { - if (swash_fetch(PL_utf8_alnum, (U8*)s)) { + if (swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) goto got_it; else @@ -1086,7 +1087,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (do_utf8) { LOAD_UTF8_CHARCLASS(alnum,"a"); while (s < strend) { - if (!swash_fetch(PL_utf8_alnum, (U8*)s)) { + if (!swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) goto got_it; else @@ -1144,7 +1145,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (do_utf8) { LOAD_UTF8_CHARCLASS(space," "); while (s < strend) { - if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) { + if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) goto got_it; else @@ -1202,7 +1203,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (do_utf8) { LOAD_UTF8_CHARCLASS(space," "); while (s < strend) { - if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) { + if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8))) { if (tmp && (norun || regtry(prog, s))) goto got_it; else @@ -1260,7 +1261,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (do_utf8) { LOAD_UTF8_CHARCLASS(digit,"0"); while (s < strend) { - if (swash_fetch(PL_utf8_digit,(U8*)s)) { + if (swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) goto got_it; else @@ -1318,7 +1319,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (do_utf8) { LOAD_UTF8_CHARCLASS(digit,"0"); while (s < strend) { - if (!swash_fetch(PL_utf8_digit,(U8*)s)) { + if (!swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) goto got_it; else @@ -1428,19 +1429,6 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * if (strend - startpos < minlen) goto phooey; } - if (startpos == strbeg) /* is ^ valid at stringarg? */ - PL_regprev = '\n'; - else { - if (prog->reganch & ROPT_UTF8 && do_utf8) { - U8 *s = reghop3((U8*)stringarg, -1, (U8*)strbeg); - PL_regprev = utf8n_to_uvuni(s, (U8*)stringarg - s, NULL, 0); - } - else - PL_regprev = (U32)stringarg[-1]; - if (!PL_multiline && PL_regprev == '\n') - PL_regprev = '\0'; /* force ^ to NOT match */ - } - /* Check validity of program. */ if (UCHARAT(prog->program) != REG_MAGIC) { Perl_croak(aTHX_ "corrupted regexp program"); @@ -2043,19 +2031,16 @@ S_regmatch(pTHX_ regnode *prog) switch (OP(scan)) { case BOL: - if (locinput == PL_bostr - ? PL_regprev == '\n' - : (PL_multiline && - (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) + if (locinput == PL_bostr || (PL_multiline && + (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) { /* regtill = regbol; */ break; } sayNO; case MBOL: - if (locinput == PL_bostr - ? PL_regprev == '\n' - : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) + if (locinput == PL_bostr || + ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n')) { break; } @@ -2161,7 +2146,7 @@ S_regmatch(pTHX_ regnode *prog) if (l >= PL_regeol) { sayNO; } - if ((UTF ? utf8n_to_uvuni((U8*)s, e - s, 0, 0) : *((U8*)s)) != + if ((UTF ? utf8n_to_uvchr((U8*)s, e - s, 0, 0) : *((U8*)s)) != (c1 ? toLOWER_utf8((U8*)l) : toLOWER_LC_utf8((U8*)l))) sayNO; s += UTF ? UTF8SKIP(s) : 1; @@ -2213,7 +2198,7 @@ S_regmatch(pTHX_ regnode *prog) sayNO; if (do_utf8) { if (!(OP(scan) == ALNUM - ? swash_fetch(PL_utf8_alnum, (U8*)locinput) + ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) : isALNUM_LC_utf8((U8*)locinput))) { sayNO; @@ -2236,7 +2221,7 @@ S_regmatch(pTHX_ regnode *prog) if (do_utf8) { LOAD_UTF8_CHARCLASS(alnum,"a"); if (OP(scan) == NALNUM - ? swash_fetch(PL_utf8_alnum, (U8*)locinput) + ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) : isALNUM_LC_utf8((U8*)locinput)) { sayNO; @@ -2258,17 +2243,17 @@ S_regmatch(pTHX_ regnode *prog) case NBOUND: /* was last char in word? */ if (do_utf8) { - if (locinput == PL_regbol) - ln = PL_regprev; + if (locinput == PL_bostr) + ln = '\n'; else { U8 *r = reghop((U8*)locinput, -1); - ln = utf8n_to_uvuni(r, s - (char*)r, 0, 0); + ln = utf8n_to_uvchr(r, s - (char*)r, 0, 0); } if (OP(scan) == BOUND || OP(scan) == NBOUND) { ln = isALNUM_uni(ln); LOAD_UTF8_CHARCLASS(alnum,"a"); - n = swash_fetch(PL_utf8_alnum, (U8*)locinput); + n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8); } else { ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln)); @@ -2276,8 +2261,8 @@ S_regmatch(pTHX_ regnode *prog) } } else { - ln = (locinput != PL_regbol) ? - UCHARAT(locinput - 1) : PL_regprev; + ln = (locinput != PL_bostr) ? + UCHARAT(locinput - 1) : '\n'; if (OP(scan) == BOUND || OP(scan) == NBOUND) { ln = isALNUM(ln); n = isALNUM(nextchr); @@ -2301,7 +2286,7 @@ S_regmatch(pTHX_ regnode *prog) if (UTF8_IS_CONTINUED(nextchr)) { LOAD_UTF8_CHARCLASS(space," "); if (!(OP(scan) == SPACE - ? swash_fetch(PL_utf8_space, (U8*)locinput) + ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) : isSPACE_LC_utf8((U8*)locinput))) { sayNO; @@ -2331,7 +2316,7 @@ S_regmatch(pTHX_ regnode *prog) if (do_utf8) { LOAD_UTF8_CHARCLASS(space," "); if (OP(scan) == NSPACE - ? swash_fetch(PL_utf8_space, (U8*)locinput) + ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) : isSPACE_LC_utf8((U8*)locinput)) { sayNO; @@ -2354,7 +2339,7 @@ S_regmatch(pTHX_ regnode *prog) if (do_utf8) { LOAD_UTF8_CHARCLASS(digit,"0"); if (!(OP(scan) == DIGIT - ? swash_fetch(PL_utf8_digit, (U8*)locinput) + ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) : isDIGIT_LC_utf8((U8*)locinput))) { sayNO; @@ -2377,7 +2362,7 @@ S_regmatch(pTHX_ regnode *prog) if (do_utf8) { LOAD_UTF8_CHARCLASS(digit,"0"); if (OP(scan) == NDIGIT - ? swash_fetch(PL_utf8_digit, (U8*)locinput) + ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) : isDIGIT_LC_utf8((U8*)locinput)) { sayNO; @@ -2393,10 +2378,12 @@ S_regmatch(pTHX_ regnode *prog) break; case CLUMP: LOAD_UTF8_CHARCLASS(mark,"~"); - if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput)) + if (locinput >= PL_regeol || + swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) sayNO; locinput += PL_utf8skip[nextchr]; - while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput)) + while (locinput < PL_regeol && + swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) locinput += UTF8SKIP(locinput); if (locinput > PL_regeol) sayNO; @@ -3245,9 +3232,15 @@ S_regmatch(pTHX_ regnode *prog) c = utf8_to_uvchr((U8*)PL_reginput, NULL); else c = UCHARAT(PL_reginput); + /* If it could work, try it. */ + if (c == c1 || c == c2) + { + TRYPAREN(paren, n, PL_reginput); + REGCP_UNWIND(lastcp); + } } /* If it could work, try it. */ - if (c1 == -1000 || c == c1 || c == c2) + else if (c1 == -1000) { TRYPAREN(paren, n, PL_reginput); REGCP_UNWIND(lastcp); @@ -3616,7 +3609,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) loceol = PL_regeol; LOAD_UTF8_CHARCLASS(alnum,"a"); while (hardcount < max && scan < loceol && - swash_fetch(PL_utf8_alnum, (U8*)scan)) { + swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3644,7 +3637,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) loceol = PL_regeol; LOAD_UTF8_CHARCLASS(alnum,"a"); while (hardcount < max && scan < loceol && - !swash_fetch(PL_utf8_alnum, (U8*)scan)) { + !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3672,7 +3665,8 @@ S_regrepeat(pTHX_ regnode *p, I32 max) loceol = PL_regeol; LOAD_UTF8_CHARCLASS(space," "); while (hardcount < max && scan < loceol && - (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { + (*scan == ' ' || + swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { scan += UTF8SKIP(scan); hardcount++; } @@ -3700,7 +3694,8 @@ S_regrepeat(pTHX_ regnode *p, I32 max) loceol = PL_regeol; LOAD_UTF8_CHARCLASS(space," "); while (hardcount < max && scan < loceol && - !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { + !(*scan == ' ' || + swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { scan += UTF8SKIP(scan); hardcount++; } @@ -3728,7 +3723,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) loceol = PL_regeol; LOAD_UTF8_CHARCLASS(digit,"0"); while (hardcount < max && scan < loceol && - swash_fetch(PL_utf8_digit,(U8*)scan)) { + swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3742,7 +3737,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) loceol = PL_regeol; LOAD_UTF8_CHARCLASS(digit,"0"); while (hardcount < max && scan < loceol && - !swash_fetch(PL_utf8_digit,(U8*)scan)) { + !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { scan += UTF8SKIP(scan); hardcount++; } @@ -3872,25 +3867,22 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) char flags = ANYOF_FLAGS(n); bool match = FALSE; UV c; - STRLEN len; + STRLEN len = 0; - if (do_utf8) - c = utf8_to_uvchr(p, &len); - else - c = *p; + c = do_utf8 ? utf8_to_uvchr(p, &len) : *p; if (do_utf8 || (flags & ANYOF_UNICODE)) { if (do_utf8 && !ANYOF_RUNTIME(n)) { if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c)) match = TRUE; } - if (do_utf8 && flags & ANYOF_UNICODE_ALL && c >= 256) + if (!match && do_utf8 && (flags & ANYOF_UNICODE_ALL) && c >= 256) match = TRUE; if (!match) { SV *sw = regclass_swash(n, TRUE, 0); if (sw) { - if (swash_fetch(sw, p)) + if (swash_fetch(sw, p, do_utf8)) match = TRUE; else if (flags & ANYOF_FOLD) { U8 tmpbuf[UTF8_MAXLEN+1]; @@ -3901,7 +3893,7 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) } else uvchr_to_utf8(tmpbuf, toLOWER_utf8(p)); - if (swash_fetch(sw, tmpbuf)) + if (swash_fetch(sw, tmpbuf, do_utf8)) match = TRUE; } } @@ -3911,7 +3903,7 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) if (ANYOF_BITMAP_TEST(n, c)) match = TRUE; else if (flags & ANYOF_FOLD) { - I32 f; + I32 f; if (flags & ANYOF_LOCALE) { PL_reg_flags |= RF_tainted;