X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=709eef238f4b53e23f4bd37713476d63dd30d09d;hb=a365f2ce4defc0d7fecd4e9484f8f958454c9192;hp=1eb7ff28593342f769fdf67d1b31317cdb125bab;hpb=c737faaf63999e5a68ef536d362cea408ab990b7;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index 1eb7ff2..709eef2 100644 --- a/regexec.c +++ b/regexec.c @@ -285,9 +285,8 @@ S_regcppop(pTHX_ const regexp *rex) * requiring null fields (pat.t#187 and split.t#{13,14} * (as of patchlevel 7877) will fail. Then again, * this code seems to be necessary or otherwise - * building DynaLoader will fail: - * "Error: '*' not in typemap in DynaLoader.xs, line 164" - * --jhi */ + * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/ + * --jhi updated by dapm */ for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) { if (i > PL_regsize) PL_regoffs[i].start = -1; @@ -308,7 +307,7 @@ S_regcppop(pTHX_ const regexp *rex) - pregexec - match a regexp against a string */ I32 -Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, +Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, register char *strend, char *strbeg, I32 minend, SV *screamer, U32 nosave) /* strend: pointer to null at end of string */ /* strbeg: real beginning of string */ @@ -372,8 +371,8 @@ Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *stren deleted from the finite automaton. */ char * -Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, - char *strend, U32 flags, re_scream_pos_data *data) +Perl_re_intuit_start(pTHX_ REGEXP * const prog, SV *sv, char *strpos, + char *strend, const U32 flags, re_scream_pos_data *data) { dVAR; register I32 start_shift = 0; @@ -1110,6 +1109,15 @@ REXEC_FBC_SCAN( \ if ((!reginfo || regtry(reginfo, &s))) \ goto got_it +#define REXEC_FBC_CSCAN(CoNdUtF8,CoNd) \ + if (do_utf8) { \ + REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ + } \ + else { \ + REXEC_FBC_CLASS_SCAN(CoNd); \ + } \ + break + #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \ if (do_utf8) { \ UtFpReLoAd; \ @@ -1425,6 +1433,31 @@ S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, !isDIGIT_LC_utf8((U8*)s), !isDIGIT_LC(*s) ); + case LNBREAK: + REXEC_FBC_CSCAN( + is_LNBREAK_utf8(s), + is_LNBREAK_latin1(s) + ); + case VERTWS: + REXEC_FBC_CSCAN( + is_VERTWS_utf8(s), + is_VERTWS_latin1(s) + ); + case NVERTWS: + REXEC_FBC_CSCAN( + !is_VERTWS_utf8(s), + !is_VERTWS_latin1(s) + ); + case HORIZWS: + REXEC_FBC_CSCAN( + is_HORIZWS_utf8(s), + is_HORIZWS_latin1(s) + ); + case NHORIZWS: + REXEC_FBC_CSCAN( + !is_HORIZWS_utf8(s), + !is_HORIZWS_latin1(s) + ); case AHOCORASICKC: case AHOCORASICK: { @@ -1448,8 +1481,8 @@ S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, U8 **points; /* map of where we were in the input string when reading a given char. For ASCII this is unnecessary overhead as the relationship - is always 1:1, but for unicode, especially - case folded unicode this is not true. */ + is always 1:1, but for Unicode, especially + case folded Unicode this is not true. */ U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; U8 *bitmap=NULL; @@ -1671,7 +1704,7 @@ S_swap_match_buff (pTHX_ regexp *prog) { - regexec_flags - match a regexp against a string */ I32 -Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, +Perl_regexec_flags(pTHX_ REGEXP * const prog, char *stringarg, register char *strend, char *strbeg, I32 minend, SV *sv, void *data, U32 flags) /* strend: pointer to null at end of string */ /* strbeg: real beginning of string */ @@ -1811,7 +1844,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * if (regtry(®info, &s)) goto got_it; after_try: - if (s >= end) + if (s > end) goto phooey; if (prog->extflags & RXf_USE_INTUIT) { s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL); @@ -2233,13 +2266,12 @@ S_regtry(pTHX_ regmatch_info *reginfo, char **startpos) /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code. * Actually, the code in regcppop() (which Ilya may be meaning by * PL_reglastparen), is not needed at all by the test suite - * (op/regexp, op/pat, op/split), but that code is needed, oddly - * enough, for building DynaLoader, or otherwise this - * "Error: '*' not in typemap in DynaLoader.xs, line 164" - * will happen. Meanwhile, this code *is* needed for the + * (op/regexp, op/pat, op/split), but that code is needed otherwise + * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/ + * Meanwhile, this code *is* needed for the * above-mentioned test suite tests to succeed. The common theme * on those tests seems to be returning null fields from matches. - * --jhi */ + * --jhi updated by dapm */ #if 1 if (prog->nparens) { regexp_paren_pair *pp = PL_regoffs; @@ -3207,8 +3239,9 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) * pack("U0U*", 0xDF) =~ /ss/i, * the 0xC3 0x9F are the UTF-8 * byte sequence for the U+00DF. */ + if (!(do_utf8 && - toLOWER(s[0]) == 's' && + toLOWER(s[0]) == 's' && ln >= 2 && toLOWER(s[1]) == 's' && (U8)l[0] == 0xC3 && @@ -3663,13 +3696,11 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) re = reg_temp_copy((regexp *)mg->mg_obj); /*XXX:dmq*/ } else { - STRLEN len; - const char * const t = SvPV_const(ret, len); U32 pm_flags = 0; const I32 osize = PL_regsize; if (DO_UTF8(ret)) pm_flags |= RXf_UTF8; - re = CALLREGCOMP((char*)t, (char*)t + len, pm_flags); + re = CALLREGCOMP(ret, pm_flags); if (!(SvFLAGS(ret) & (SVs_TEMP | SVs_PADTMP | SVf_READONLY | SVs_GMG))) @@ -4188,12 +4219,6 @@ NULL case BRANCH: /* /(...|A|...)/ */ scan = NEXTOPER(scan); /* scan now points to inner node */ - if ((!next || (OP(next) != BRANCH && OP(next) != BRANCHJ)) - && !has_cutgroup) - { - /* last branch; skip state push and jump direct to node */ - continue; - } ST.lastparen = *PL_reglastparen; ST.next_branch = next; REGCP_SET(ST.cp); @@ -4971,6 +4996,55 @@ NULL sayNO; /* NOTREACHED */ #undef ST + case FOLDCHAR: + n = ARG(scan); + if ( n == (U32)what_len_TRICKYFOLD(locinput,do_utf8,ln) ) { + locinput += ln; + } else if ( 0xDF == n && !do_utf8 && !UTF ) { + sayNO; + } else { + U8 folded[UTF8_MAXBYTES_CASE+1]; + STRLEN foldlen; + const char * const l = locinput; + char *e = PL_regeol; + to_uni_fold(n, folded, &foldlen); + + if (ibcmp_utf8((const char*) folded, 0, foldlen, 1, + l, &e, 0, do_utf8)) { + sayNO; + } + locinput = e; + } + nextchr = UCHARAT(locinput); + break; + case LNBREAK: + if ((n=is_LNBREAK(locinput,do_utf8))) { + locinput += n; + nextchr = UCHARAT(locinput); + } else + sayNO; + break; + +#define CASE_CLASS(nAmE) \ + case nAmE: \ + if ((n=is_##nAmE(locinput,do_utf8))) { \ + locinput += n; \ + nextchr = UCHARAT(locinput); \ + } else \ + sayNO; \ + break; \ + case N##nAmE: \ + if ((n=is_##nAmE(locinput,do_utf8))) { \ + sayNO; \ + } else { \ + locinput += UTF8SKIP(locinput); \ + nextchr = UCHARAT(locinput); \ + } \ + break + + CASE_CLASS(VERTWS); + CASE_CLASS(HORIZWS); +#undef CASE_CLASS default: PerlIO_printf(Perl_error_log, "%"UVxf" %d\n", @@ -5382,7 +5456,77 @@ S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max, int depth) while (scan < loceol && !isDIGIT(*scan)) scan++; } + case LNBREAK: + if (do_utf8) { + loceol = PL_regeol; + while (hardcount < max && scan < loceol && (c=is_LNBREAK_utf8(scan))) { + scan += c; + hardcount++; + } + } else { + /* + LNBREAK can match two latin chars, which is ok, + because we have a null terminated string, but we + have to use hardcount in this situation + */ + while (scan < loceol && (c=is_LNBREAK_latin1(scan))) { + scan+=c; + hardcount++; + } + } break; + case HORIZWS: + if (do_utf8) { + loceol = PL_regeol; + while (hardcount < max && scan < loceol && (c=is_HORIZWS_utf8(scan))) { + scan += c; + hardcount++; + } + } else { + while (scan < loceol && is_HORIZWS_latin1(scan)) + scan++; + } + break; + case NHORIZWS: + if (do_utf8) { + loceol = PL_regeol; + while (hardcount < max && scan < loceol && !is_HORIZWS_utf8(scan)) { + scan += UTF8SKIP(scan); + hardcount++; + } + } else { + while (scan < loceol && !is_HORIZWS_latin1(scan)) + scan++; + + } + break; + case VERTWS: + if (do_utf8) { + loceol = PL_regeol; + while (hardcount < max && scan < loceol && (c=is_VERTWS_utf8(scan))) { + scan += c; + hardcount++; + } + } else { + while (scan < loceol && is_VERTWS_latin1(scan)) + scan++; + + } + break; + case NVERTWS: + if (do_utf8) { + loceol = PL_regeol; + while (hardcount < max && scan < loceol && !is_VERTWS_utf8(scan)) { + scan += UTF8SKIP(scan); + hardcount++; + } + } else { + while (scan < loceol && !is_VERTWS_latin1(scan)) + scan++; + + } + break; + default: /* Called on something of 0 width. */ break; /* So match right here or not at all. */ }