X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=5a6d72db8cecc59b6f839881a5159e5853319763;hb=696235b60874be65fe029a39969f44a0133ec2f8;hp=198e99ead266eb59b836cb86c8e43fdf14733972;hpb=545666dba9cc33d16d0b8341e36facdb43c44913;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index 198e99e..5a6d72d 100644 --- a/regexec.c +++ b/regexec.c @@ -67,7 +67,7 @@ * **** Alterations to Henry's code are... **** - **** Copyright (c) 1991-2001, Larry Wall + **** Copyright (c) 1991-2002, Larry Wall **** **** You may distribute under the terms of either the GNU General Public **** License or the Artistic License, as specified in the README file. @@ -96,11 +96,13 @@ #define STATIC static #endif +#define REGINCLASS(p,c) (ANYOF_FLAGS(p) ? reginclass(p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c))) + /* * Forwards. */ -#define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) +#define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv)) #define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b) #define reghop_c(pos,off) ((char*)reghop((U8*)pos, off)) @@ -111,7 +113,7 @@ #define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off)) #define HOPBACK(pos, off) ( \ - (UTF && PL_reg_match_utf8) \ + (PL_reg_match_utf8) \ ? reghopmaybe((U8*)pos, -off) \ : (pos - off >= PL_bostr) \ ? (U8*)(pos - off) \ @@ -126,20 +128,32 @@ #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim)) #define HOPMAYBE3c(pos,off,lim) ((char*)HOPMAYBE3(pos,off,lim)) -#define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) (void)CAT2(is_utf8_, a)((U8*)b); } STMT_END +#define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) { ENTER; save_re_context(); (void)CAT2(is_utf8_, a)((U8*)b); LEAVE; } } STMT_END /* for use after a quantifier and before an EXACT-like node -- japhy */ #define JUMPABLE(rn) ( \ OP(rn) == OPEN || OP(rn) == CLOSE || OP(rn) == EVAL || \ - OP(rn) == SUSPEND || OP(rn) == IFMATCH \ + OP(rn) == SUSPEND || OP(rn) == IFMATCH || \ + OP(rn) == PLUS || OP(rn) == MINMOD || \ + (PL_regkind[(U8)OP(rn)] == CURLY && ARG1(rn) > 0) \ ) -#define NEAR_EXACT(rn) (PL_regkind[(U8)OP(rn)] == EXACT || JUMPABLE(rn)) +#define HAS_TEXT(rn) ( \ + PL_regkind[(U8)OP(rn)] == EXACT || PL_regkind[(U8)OP(rn)] == REF \ +) -#define NEXT_IMPT(rn) STMT_START { \ +/* + Search for mandatory following text node; for lookahead, the text must + follow but for lookbehind (rn->flags != 0) we skip to the next step. +*/ +#define FIND_NEXT_IMPT(rn) STMT_START { \ while (JUMPABLE(rn)) \ - if (OP(rn) == SUSPEND || OP(rn) == IFMATCH) \ + if (OP(rn) == SUSPEND || PL_regkind[(U8)OP(rn)] == CURLY) \ rn = NEXTOPER(NEXTOPER(rn)); \ + else if (OP(rn) == PLUS) \ + rn = NEXTOPER(rn); \ + else if (OP(rn) == IFMATCH) \ + rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \ else rn += NEXT_OFF(rn); \ } STMT_END @@ -378,19 +392,32 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, register SV *check; char *strbeg; char *t; + int do_utf8 = sv ? SvUTF8(sv) : 0; /* if no sv we have to assume bytes */ I32 ml_anch; register char *other_last = Nullch; /* other substr checked before this */ char *check_at = Nullch; /* check substr found at this pos */ #ifdef DEBUGGING char *i_strpos = strpos; - SV *dsv = sv_2mortal(newSVpvn("", 0)); + SV *dsv = PERL_DEBUG_PAD_ZERO(0); #endif + if (prog->reganch & ROPT_UTF8) { + DEBUG_r(PerlIO_printf(Perl_debug_log, + "UTF-8 regex...\n")); + PL_reg_flags |= RF_utf8; + } + DEBUG_r({ - char*s = UTF ? sv_uni_display(dsv, sv, 60, 0) : strpos; - int len = UTF ? strlen(s) : strend - strpos; + char *s = PL_reg_match_utf8 ? + sv_uni_display(dsv, sv, 60, UNI_DISPLAY_REGEX) : + strpos; + int len = PL_reg_match_utf8 ? + strlen(s) : strend - strpos; if (!PL_colorset) reginitcolors(); + if (PL_reg_match_utf8) + DEBUG_r(PerlIO_printf(Perl_debug_log, + "UTF-8 target...\n")); PerlIO_printf(Perl_debug_log, "%sGuessing start of match, REx%s `%s%.60s%s%s' against `%s%.*s%s%s'...\n", PL_colors[4],PL_colors[5],PL_colors[0], @@ -404,16 +431,27 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, ); }); - if (prog->reganch & ROPT_UTF8) - PL_reg_flags |= RF_utf8; - if (prog->minlen > CHR_DIST((U8*)strend, (U8*)strpos)) { - DEBUG_r(PerlIO_printf(Perl_debug_log, "String too short...\n")); + DEBUG_r(PerlIO_printf(Perl_debug_log, + "String too short... [re_intuit_start]\n")); goto fail; } strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos; PL_regeol = strend; - check = prog->check_substr; + if (do_utf8) { + if (!prog->check_utf8 && prog->check_substr) + to_utf8_substr(prog); + check = prog->check_utf8; + } else { + if (!prog->check_substr && prog->check_utf8) + to_byte_substr(prog); + check = prog->check_substr; + } + if (check == &PL_sv_undef) { + DEBUG_r(PerlIO_printf(Perl_debug_log, + "Non-utf string cannot match utf check string\n")); + goto fail; + } if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */ ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE) || ( (prog->reganch & ROPT_ANCH_BOL) @@ -519,7 +557,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, DEBUG_r(PerlIO_printf(Perl_debug_log, "%s %s substr `%s%.*s%s'%s%s", (s ? "Found" : "Did not find"), - ((check == prog->anchored_substr) ? "anchored" : "floating"), + (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) ? "anchored" : "floating"), PL_colors[0], (int)(SvCUR(check) - (SvTAIL(check)!=0)), SvPVX(check), @@ -542,16 +580,17 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, Probably it is right to do no SCREAM here... */ - if (prog->float_substr && prog->anchored_substr) { + if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) : (prog->float_substr && prog->anchored_substr)) { /* Take into account the "other" substring. */ /* XXXX May be hopelessly wrong for UTF... */ if (!other_last) other_last = strpos; - if (check == prog->float_substr) { + if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) { do_other_anchored: { char *last = HOP3c(s, -start_shift, strbeg), *last1, *last2; char *s1 = s; + SV* must; t = s - prog->check_offset_max; if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ @@ -569,20 +608,27 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, last1 = last; /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ /* On end-of-str: see comment below. */ - s = fbm_instr((unsigned char*)t, - HOP3(HOP3(last1, prog->anchored_offset, strend) - + SvCUR(prog->anchored_substr), - -(SvTAIL(prog->anchored_substr)!=0), strbeg), - prog->anchored_substr, - PL_multiline ? FBMrf_MULTILINE : 0); + must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; + if (must == &PL_sv_undef) { + s = (char*)NULL; + DEBUG_r(must = prog->anchored_utf8); /* for debug */ + } + else + s = fbm_instr( + (unsigned char*)t, + HOP3(HOP3(last1, prog->anchored_offset, strend) + + SvCUR(must), -(SvTAIL(must)!=0), strbeg), + must, + PL_multiline ? FBMrf_MULTILINE : 0 + ); DEBUG_r(PerlIO_printf(Perl_debug_log, "%s anchored substr `%s%.*s%s'%s", (s ? "Found" : "Contradicts"), PL_colors[0], - (int)(SvCUR(prog->anchored_substr) - - (SvTAIL(prog->anchored_substr)!=0)), - SvPVX(prog->anchored_substr), - PL_colors[1], (SvTAIL(prog->anchored_substr) ? "$" : ""))); + (int)(SvCUR(must) + - (SvTAIL(must)!=0)), + SvPVX(must), + PL_colors[1], (SvTAIL(must) ? "$" : ""))); if (!s) { if (last1 >= last2) { DEBUG_r(PerlIO_printf(Perl_debug_log, @@ -609,54 +655,60 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, } } else { /* Take into account the floating substring. */ - char *last, *last1; - char *s1 = s; - - t = HOP3c(s, -start_shift, strbeg); - last1 = last = - HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); - if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) - last = HOP3c(t, prog->float_max_offset, strend); - s = HOP3c(t, prog->float_min_offset, strend); - if (s < other_last) - s = other_last; + char *last, *last1; + char *s1 = s; + SV* must; + + t = HOP3c(s, -start_shift, strbeg); + last1 = last = + HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); + if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) + last = HOP3c(t, prog->float_max_offset, strend); + s = HOP3c(t, prog->float_min_offset, strend); + if (s < other_last) + s = other_last; /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ - /* fbm_instr() takes into account exact value of end-of-str - if the check is SvTAIL(ed). Since false positives are OK, - and end-of-str is not later than strend we are OK. */ + must = do_utf8 ? prog->float_utf8 : prog->float_substr; + /* fbm_instr() takes into account exact value of end-of-str + if the check is SvTAIL(ed). Since false positives are OK, + and end-of-str is not later than strend we are OK. */ + if (must == &PL_sv_undef) { + s = (char*)NULL; + DEBUG_r(must = prog->float_utf8); /* for debug message */ + } + else s = fbm_instr((unsigned char*)s, - (unsigned char*)last + SvCUR(prog->float_substr) - - (SvTAIL(prog->float_substr)!=0), - prog->float_substr, PL_multiline ? FBMrf_MULTILINE : 0); - DEBUG_r(PerlIO_printf(Perl_debug_log, "%s floating substr `%s%.*s%s'%s", - (s ? "Found" : "Contradicts"), - PL_colors[0], - (int)(SvCUR(prog->float_substr) - - (SvTAIL(prog->float_substr)!=0)), - SvPVX(prog->float_substr), - PL_colors[1], (SvTAIL(prog->float_substr) ? "$" : ""))); - if (!s) { - if (last1 == last) { - DEBUG_r(PerlIO_printf(Perl_debug_log, - ", giving up...\n")); - goto fail_finish; - } + (unsigned char*)last + SvCUR(must) + - (SvTAIL(must)!=0), + must, PL_multiline ? FBMrf_MULTILINE : 0); + DEBUG_r(PerlIO_printf(Perl_debug_log, "%s floating substr `%s%.*s%s'%s", + (s ? "Found" : "Contradicts"), + PL_colors[0], + (int)(SvCUR(must) - (SvTAIL(must)!=0)), + SvPVX(must), + PL_colors[1], (SvTAIL(must) ? "$" : ""))); + if (!s) { + if (last1 == last) { DEBUG_r(PerlIO_printf(Perl_debug_log, - ", trying anchored starting at offset %ld...\n", - (long)(s1 + 1 - i_strpos))); - other_last = last; - s = HOP3c(t, 1, strend); - goto restart; - } - else { - DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", - (long)(s - i_strpos))); - other_last = s; /* Fix this later. --Hugo */ - s = s1; - if (t == strpos) - goto try_at_start; - goto try_at_offset; + ", giving up...\n")); + goto fail_finish; } + DEBUG_r(PerlIO_printf(Perl_debug_log, + ", trying anchored starting at offset %ld...\n", + (long)(s1 + 1 - i_strpos))); + other_last = last; + s = HOP3c(t, 1, strend); + goto restart; + } + else { + DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", + (long)(s - i_strpos))); + other_last = s; /* Fix this later. --Hugo */ + s = s1; + if (t == strpos) + goto try_at_start; + goto try_at_offset; + } } } @@ -679,7 +731,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, while (t < strend - prog->minlen) { if (*t == '\n') { if (t < check_at - prog->check_offset_min) { - if (prog->anchored_substr) { + if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) { /* Since we moved from the found position, we definitely contradict the found anchored substr. Due to the above check we do not @@ -719,7 +771,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, } s = t; set_useful: - ++BmUSEFUL(prog->check_substr); /* hooray/5 */ + ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */ } else { /* The found string does not prohibit matching at strpos, @@ -743,15 +795,23 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, ); success_at_start: if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */ - && prog->check_substr /* Could be deleted already */ - && --BmUSEFUL(prog->check_substr) < 0 - && prog->check_substr == prog->float_substr) + && (do_utf8 ? ( + prog->check_utf8 /* Could be deleted already */ + && --BmUSEFUL(prog->check_utf8) < 0 + && (prog->check_utf8 == prog->float_utf8) + ) : ( + prog->check_substr /* Could be deleted already */ + && --BmUSEFUL(prog->check_substr) < 0 + && (prog->check_substr == prog->float_substr) + ))) { /* If flags & SOMETHING - do not do it many times on the same match */ DEBUG_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); - SvREFCNT_dec(prog->check_substr); - prog->check_substr = Nullsv; /* disable */ - prog->float_substr = Nullsv; /* clear */ + SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr); + if (do_utf8 ? prog->check_substr : prog->check_utf8) + SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8); + prog->check_substr = prog->check_utf8 = Nullsv; /* disable */ + prog->float_substr = prog->float_utf8 = Nullsv; /* clear */ check = Nullsv; /* abort */ s = strpos; /* XXXX This is a remnant of the old implementation. It @@ -778,9 +838,9 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT ? CHR_DIST(str+STR_LEN(prog->regstclass), str) : 1); - char *endpos = (prog->anchored_substr || ml_anch) + char *endpos = (prog->anchored_substr || prog->anchored_utf8 || ml_anch) ? HOP3c(s, (prog->minlen ? cl_l : 0), strend) - : (prog->float_substr + : (prog->float_substr || prog->float_utf8 ? HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend) : strend); @@ -806,8 +866,8 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if ((prog->reganch & ROPT_ANCH) && !ml_anch) goto fail; /* Contradict one of substrings */ - if (prog->anchored_substr) { - if (prog->anchored_substr == check) { + if (prog->anchored_substr || prog->anchored_utf8) { + if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { DEBUG_r( what = "anchored" ); hop_and_restart: s = HOP3c(t, 1, strend); @@ -847,7 +907,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, PL_colors[0],PL_colors[1], (long)(t - i_strpos)) ); goto try_at_offset; } - if (!prog->float_substr) /* Could have been deleted */ + if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ goto fail; /* Check is floating subtring. */ retry_floating_check: @@ -874,8 +934,8 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, return s; fail_finish: /* Substring not found */ - if (prog->check_substr) /* could be removed already */ - BmUSEFUL(prog->check_substr) += 5; /* hooray */ + if (prog->check_substr || prog->check_utf8) /* could be removed already */ + BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */ fail: DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", PL_colors[4],PL_colors[5])); @@ -899,15 +959,24 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta switch (OP(c)) { case ANYOF: while (s < strend) { - if (reginclass(c, (U8*)s, do_utf8)) { + STRLEN skip = do_utf8 ? UTF8SKIP(s) : 1; + + if (do_utf8 ? + reginclass(c, (U8*)s, 0, do_utf8) : + REGINCLASS(c, (U8*)s) || + (ANYOF_FOLD_SHARP_S(c, s, strend) && + /* The assignment of 2 is intentional: + * for the sharp s, the skip is 2. */ + (skip = SHARP_S_SKIP) + )) { if (tmp && (norun || regtry(prog, s))) goto got_it; else tmp = doevery; } - else + else tmp = 1; - s += do_utf8 ? UTF8SKIP(s) : 1; + s += skip; } break; case CANY: @@ -924,14 +993,14 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta ln = STR_LEN(c); if (UTF) { STRLEN ulen1, ulen2; - U8 tmpbuf1[UTF8_MAXLEN*2+1]; - U8 tmpbuf2[UTF8_MAXLEN*2+1]; + U8 tmpbuf1[UTF8_MAXLEN_UCLC+1]; + U8 tmpbuf2[UTF8_MAXLEN_UCLC+1]; to_utf8_lower((U8*)m, tmpbuf1, &ulen1); to_utf8_upper((U8*)m, tmpbuf2, &ulen2); - c1 = utf8_to_uvuni(tmpbuf1, 0); - c2 = utf8_to_uvuni(tmpbuf2, 0); + c1 = utf8_to_uvchr(tmpbuf1, 0); + c2 = utf8_to_uvchr(tmpbuf2, 0); } else { c1 = *(U8*)m; @@ -944,27 +1013,90 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta c1 = *(U8*)m; c2 = PL_fold_locale[c1]; do_exactf: - e = strend - ln; + e = do_utf8 ? s + ln : strend - ln; if (norun && e < s) e = s; /* Due to minlen logic of intuit() */ + /* The idea in the EXACTF* cases is to first find the + * first character of the EXACTF* node and then, if + * necessary, case-insensitively compare the full + * text of the node. The c1 and c2 are the first + * characters (though in Unicode it gets a bit + * more complicated because there are more cases + * than just upper and lower: one needs to use + * the so-called folding case for case-insensitive + * matching (called "loose matching" in Unicode). + * ibcmp_utf8() will do just that. */ + if (do_utf8) { - STRLEN len; - if (c1 == c2) + UV c, f; + U8 tmpbuf [UTF8_MAXLEN+1]; + U8 foldbuf[UTF8_MAXLEN_FOLD+1]; + STRLEN len, foldlen; + + if (c1 == c2) { while (s <= e) { - if ( utf8_to_uvchr((U8*)s, &len) == c1 - && regtry(prog, s) ) + c = utf8_to_uvchr((U8*)s, &len); + if ( c == c1 + && (ln == len || + ibcmp_utf8(s, (char **)0, 0, do_utf8, + m, (char **)0, ln, UTF)) + && (norun || regtry(prog, s)) ) goto got_it; + else { + uvchr_to_utf8(tmpbuf, c); + f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); + if ( f != c + && (f == c1 || f == c2) + && (ln == foldlen || + !ibcmp_utf8((char *) foldbuf, + (char **)0, foldlen, do_utf8, + m, + (char **)0, ln, UTF)) + && (norun || regtry(prog, s)) ) + goto got_it; + } s += len; } - else + } + else { while (s <= e) { - UV c = utf8_to_uvchr((U8*)s, &len); - if ( (c == c1 || c == c2) && regtry(prog, s) ) + c = utf8_to_uvchr((U8*)s, &len); + + /* Handle some of the three Greek sigmas cases. + * Note that not all the possible combinations + * are handled here: some of them are handled + * by the standard folding rules, and some of + * them (the character class or ANYOF cases) + * are handled during compiletime in + * regexec.c:S_regclass(). */ + if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA || + c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) + c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA; + + if ( (c == c1 || c == c2) + && (ln == len || + ibcmp_utf8(s, (char **)0, 0, do_utf8, + m, (char **)0, ln, UTF)) + && (norun || regtry(prog, s)) ) goto got_it; + else { + uvchr_to_utf8(tmpbuf, c); + f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); + if ( f != c + && (f == c1 || f == c2) + && (ln == foldlen || + !ibcmp_utf8((char *) foldbuf, + (char **)0, foldlen, do_utf8, + m, + (char **)0, ln, UTF)) + && (norun || regtry(prog, s)) ) + goto got_it; + } s += len; } + } } else { if (c1 == c2) @@ -1457,7 +1589,8 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * SV* oreplsv = GvSV(PL_replgv); bool do_utf8 = DO_UTF8(sv); #ifdef DEBUGGING - SV *dsv = sv_2mortal(newSVpvn("", 0)); + SV *dsv0 = PERL_DEBUG_PAD_ZERO(0); + SV *dsv1 = PERL_DEBUG_PAD_ZERO(1); #endif PL_regcc = 0; @@ -1474,19 +1607,10 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * } minlen = prog->minlen; - if (do_utf8 && !(prog->reganch & ROPT_CANY_SEEN)) { - if (utf8_distance((U8*)strend, (U8*)startpos) < minlen) { - DEBUG_r(PerlIO_printf(Perl_debug_log, - "Too short (in characters)...\n")); - goto phooey; - } - } - else { - if (strend - startpos < minlen) { - DEBUG_r(PerlIO_printf(Perl_debug_log, - "Too short (in bytes)...\n")); - goto phooey; - } + if (strend - startpos < minlen) { + DEBUG_r(PerlIO_printf(Perl_debug_log, + "String too short [regexec_flags]...\n")); + goto phooey; } /* Check validity of program. */ @@ -1538,8 +1662,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * PL_reg_ganch = strbeg; } - if (do_utf8 == (UTF!=0) && - !(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) { + if (!(flags & REXEC_CHECKED) && (prog->check_substr != Nullsv || prog->check_utf8 != Nullsv)) { re_scream_pos_data d; d.scream_olds = &scream_olds; @@ -1552,20 +1675,26 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * } DEBUG_r({ - char *s = do_utf8 ? sv_uni_display(dsv, sv, 60, 0) : startpos; - int len = do_utf8 ? strlen(s) : strend - startpos; + char *s0 = UTF ? + pv_uni_display(dsv0, (U8*)prog->precomp, prog->prelen, 60, + UNI_DISPLAY_REGEX) : + prog->precomp; + int len0 = UTF ? SvCUR(dsv0) : prog->prelen; + char *s1 = do_utf8 ? sv_uni_display(dsv1, sv, 60, + UNI_DISPLAY_REGEX) : startpos; + int len1 = do_utf8 ? SvCUR(dsv1) : strend - startpos; if (!PL_colorset) reginitcolors(); PerlIO_printf(Perl_debug_log, - "%sMatching REx%s `%s%.60s%s%s' against `%s%.*s%s%s'\n", + "%sMatching REx%s `%s%*.*s%s%s' against `%s%.*s%s%s'\n", PL_colors[4],PL_colors[5],PL_colors[0], - prog->precomp, + len0, len0, s0, PL_colors[1], - (strlen(prog->precomp) > 60 ? "..." : ""), + len0 > 60 ? "..." : "", PL_colors[0], - (int)(len > 60 ? 60 : len), - s, PL_colors[1], - (len > 60 ? "..." : "") + (int)(len1 > 60 ? 60 : len1), + s1, PL_colors[1], + (len1 > 60 ? "..." : "") ); }); @@ -1583,7 +1712,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * dontbother = minlen - 1; end = HOP3c(strend, -dontbother, strbeg) - 1; /* for multiline we only have to try after newlines */ - if (prog->check_substr) { + if (prog->check_substr || prog->check_utf8) { if (s == startpos) goto after_try; while (1) { @@ -1619,13 +1748,16 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * } /* Messy cases: unanchored match. */ - if (prog->anchored_substr && prog->reganch & ROPT_SKIP) { + if ((prog->anchored_substr || prog->anchored_utf8) && prog->reganch & ROPT_SKIP) { /* we have /x+whatever/ */ /* it must be a one character string (XXXX Except UTF?) */ - char ch = SvPVX(prog->anchored_substr)[0]; + char ch; #ifdef DEBUGGING int did_match = 0; #endif + if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) + do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); + ch = SvPVX(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; if (do_utf8) { while (s < strend) { @@ -1657,23 +1789,37 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * ); } /*SUPPRESS 560*/ - else if (do_utf8 == (UTF!=0) && - (prog->anchored_substr != Nullsv - || (prog->float_substr != Nullsv - && prog->float_max_offset < strend - s))) { - SV *must = prog->anchored_substr - ? prog->anchored_substr : prog->float_substr; - I32 back_max = - prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset; - I32 back_min = - prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset; - char *last = HOP3c(strend, /* Cannot start after this */ - -(I32)(CHR_SVLEN(must) - - (SvTAIL(must) != 0) + back_min), strbeg); + else if (prog->anchored_substr != Nullsv + || prog->anchored_utf8 != Nullsv + || ((prog->float_substr != Nullsv || prog->float_utf8 != Nullsv) + && prog->float_max_offset < strend - s)) { + SV *must; + I32 back_max; + I32 back_min; + char *last; char *last1; /* Last position checked before */ #ifdef DEBUGGING int did_match = 0; #endif + if (prog->anchored_substr || prog->anchored_utf8) { + if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) + do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); + must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; + back_max = back_min = prog->anchored_offset; + } else { + if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) + do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); + must = do_utf8 ? prog->float_utf8 : prog->float_substr; + back_max = prog->float_max_offset; + back_min = prog->float_min_offset; + } + if (must == &PL_sv_undef) + /* could not downgrade utf8 check substring, so must fail */ + goto phooey; + + last = HOP3c(strend, /* Cannot start after this */ + -(I32)(CHR_SVLEN(must) + - (SvTAIL(must) != 0) + back_min), strbeg); if (s > PL_bostr) last1 = HOPc(s, -1); @@ -1721,7 +1867,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * DEBUG_r(if (!did_match) PerlIO_printf(Perl_debug_log, "Did not find %s substr `%s%.*s%s'%s...\n", - ((must == prog->anchored_substr) + ((must == prog->anchored_substr || must == prog->anchored_utf8) ? "anchored" : "floating"), PL_colors[0], (int)(SvCUR(must) - (SvTAIL(must)!=0)), @@ -1736,8 +1882,24 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * strend = HOPc(strend, -(minlen - 1)); DEBUG_r({ SV *prop = sv_newmortal(); + char *s0; + char *s1; + int len0; + int len1; + regprop(prop, c); - PerlIO_printf(Perl_debug_log, "Matching stclass `%s' against `%s'\n", SvPVX(prop), UTF ? sv_uni_display(dsv, sv, 60, 0) : s); + s0 = UTF ? + pv_uni_display(dsv0, (U8*)SvPVX(prop), SvCUR(prop), 60, + UNI_DISPLAY_REGEX) : + SvPVX(prop); + len0 = UTF ? SvCUR(dsv0) : SvCUR(prop); + s1 = UTF ? + sv_uni_display(dsv1, sv, 60, UNI_DISPLAY_REGEX) : s; + len1 = UTF ? SvCUR(dsv1) : strend - s; + PerlIO_printf(Perl_debug_log, + "Matching stclass `%*.*s' against `%*.*s'\n", + len0, len0, s0, + len1, len1, s1); }); if (find_byclass(prog, c, s, strend, startpos, 0)) goto got_it; @@ -1745,20 +1907,26 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * } else { dontbother = 0; - if (prog->float_substr != Nullsv) { /* Trim the end. */ + if (prog->float_substr != Nullsv || prog->float_utf8 != Nullsv) { + /* Trim the end. */ char *last; + SV* float_real; + + if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) + do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); + float_real = do_utf8 ? prog->float_utf8 : prog->float_substr; if (flags & REXEC_SCREAM) { - last = screaminstr(sv, prog->float_substr, s - strbeg, + last = screaminstr(sv, float_real, s - strbeg, end_shift, &scream_pos, 1); /* last one */ if (!last) last = scream_olds; /* Only one occurrence. */ } else { STRLEN len; - char *little = SvPV(prog->float_substr, len); + char *little = SvPV(float_real, len); - if (SvTAIL(prog->float_substr)) { + if (SvTAIL(float_real)) { if (memEQ(strend - len + 1, little, len - 1)) last = strend - len + 1; else if (!PL_multiline) @@ -2009,6 +2177,7 @@ typedef union re_unwind_t { #define sayYES goto yes #define sayNO goto no +#define sayNO_ANYOF goto no_anyof #define sayYES_FINAL goto yes_final #define sayYES_LOUD goto yes_loud #define sayNO_FINAL goto no_final @@ -2051,9 +2220,9 @@ S_regmatch(pTHX_ regnode *prog) #endif register bool do_utf8 = PL_reg_match_utf8; #ifdef DEBUGGING - SV *dsv0 = sv_2mortal(newSVpvn("", 0)); - SV *dsv1 = sv_2mortal(newSVpvn("", 0)); - SV *dsv2 = sv_2mortal(newSVpvn("", 0)); + SV *dsv0 = PERL_DEBUG_PAD_ZERO(0); + SV *dsv1 = PERL_DEBUG_PAD_ZERO(1); + SV *dsv2 = PERL_DEBUG_PAD_ZERO(2); #endif #ifdef DEBUGGING @@ -2097,17 +2266,17 @@ S_regmatch(pTHX_ regnode *prog) char *s0 = do_utf8 ? pv_uni_display(dsv0, (U8*)(locinput - pref_len), - pref0_len, 60, 0) : + pref0_len, 60, UNI_DISPLAY_REGEX) : locinput - pref_len; int len0 = do_utf8 ? strlen(s0) : pref0_len; char *s1 = do_utf8 ? pv_uni_display(dsv1, (U8*)(locinput - pref_len + pref0_len), - pref_len - pref0_len, 60, 0) : + pref_len - pref0_len, 60, UNI_DISPLAY_REGEX) : locinput - pref_len + pref0_len; int len1 = do_utf8 ? strlen(s1) : pref_len - pref0_len; char *s2 = do_utf8 ? pv_uni_display(dsv2, (U8*)locinput, - PL_regeol - locinput, 60, 0) : + PL_regeol - locinput, 60, UNI_DISPLAY_REGEX) : locinput; int len2 = do_utf8 ? strlen(s2) : l; PerlIO_printf(Perl_debug_log, @@ -2212,31 +2381,40 @@ S_regmatch(pTHX_ regnode *prog) s = STRING(scan); ln = STR_LEN(scan); if (do_utf8 != (UTF!=0)) { + /* The target and the pattern have differing utf8ness. */ char *l = locinput; char *e = s + ln; - STRLEN len; - if (do_utf8) + STRLEN ulen; + + if (do_utf8) { + /* The target is utf8, the pattern is not utf8. */ while (s < e) { if (l >= PL_regeol) - sayNO; - if (*((U8*)s) != utf8_to_uvchr((U8*)l, &len)) - sayNO; - s += len; - l += len; + sayNO; + if (NATIVE_TO_UNI(*(U8*)s) != + utf8_to_uvuni((U8*)l, &ulen)) + sayNO; + l += ulen; + s ++; } - else + } + else { + /* The target is not utf8, the pattern is utf8. */ while (s < e) { if (l >= PL_regeol) sayNO; - if (*((U8*)l) != utf8_to_uvchr((U8*)s, &len)) + if (NATIVE_TO_UNI(*((U8*)l)) != + utf8_to_uvuni((U8*)s, &ulen)) sayNO; - s+=len; - l++; + s += ulen; + l ++; } + } locinput = l; nextchr = UCHARAT(locinput); break; } + /* The target and the pattern have the same utf8ness. */ /* Inline the first character, for speed. */ if (UCHARAT(s) != nextchr) sayNO; @@ -2254,26 +2432,33 @@ S_regmatch(pTHX_ regnode *prog) s = STRING(scan); ln = STR_LEN(scan); - if (do_utf8) { + if (do_utf8 || UTF) { + /* Either target or the pattern are utf8. */ char *l = locinput; - char *e; - STRLEN ulen; - U8 tmpbuf[UTF8_MAXLEN*2+1]; - e = s + ln; - while (s < e) { - if (l >= PL_regeol) - sayNO; - toLOWER_utf8((U8*)l, tmpbuf, &ulen); - if (memNE(s, (char*)tmpbuf, ulen)) - sayNO; - s += UTF8SKIP(s); - l += ulen; + char *e = PL_regeol; + + if (ibcmp_utf8(s, 0, ln, UTF, + l, &e, 0, do_utf8)) { + /* One more case for the sharp s: + * 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' && + ln >= 2 && + toLOWER(s[1]) == 's' && + (U8)l[0] == 0xC3 && + e - l >= 2 && + (U8)l[1] == 0x9F)) + sayNO; } - locinput = l; + locinput = e; nextchr = UCHARAT(locinput); break; } + /* Neither the target and the pattern are utf8. */ + /* Inline the first character, for speed. */ if (UCHARAT(s) != nextchr && UCHARAT(s) != ((OP(scan) == EXACTF) @@ -2290,22 +2475,36 @@ S_regmatch(pTHX_ regnode *prog) break; case ANYOF: if (do_utf8) { - if (!reginclass(scan, (U8*)locinput, do_utf8)) - sayNO; + STRLEN inclasslen = PL_regeol - locinput; + + if (!reginclass(scan, (U8*)locinput, &inclasslen, do_utf8)) + sayNO_ANYOF; if (locinput >= PL_regeol) sayNO; - locinput += PL_utf8skip[nextchr]; + locinput += inclasslen ? inclasslen : UTF8SKIP(locinput); nextchr = UCHARAT(locinput); + break; } else { if (nextchr < 0) nextchr = UCHARAT(locinput); - if (!reginclass(scan, (U8*)locinput, do_utf8)) - sayNO; + if (!REGINCLASS(scan, (U8*)locinput)) + sayNO_ANYOF; if (!nextchr && locinput >= PL_regeol) sayNO; nextchr = UCHARAT(++locinput); + break; } + no_anyof: + /* If we might have the case of the German sharp s + * in a casefolding Unicode character class. */ + + if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) { + locinput += SHARP_S_SKIP; + nextchr = UCHARAT(locinput); + } + else + sayNO; break; case ALNUML: PL_reg_flags |= RF_tainted; @@ -2495,16 +2694,21 @@ S_regmatch(pTHX_ regnode *prog) nextchr = UCHARAT(++locinput); break; case CLUMP: - LOAD_UTF8_CHARCLASS(mark,"~"); - 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, do_utf8)) - locinput += UTF8SKIP(locinput); - if (locinput > PL_regeol) + if (locinput >= PL_regeol) sayNO; + if (do_utf8) { + LOAD_UTF8_CHARCLASS(mark,"~"); + if (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, do_utf8)) + locinput += UTF8SKIP(locinput); + if (locinput > PL_regeol) + sayNO; + } + else + locinput++; nextchr = UCHARAT(locinput); break; case REFFL: @@ -2531,8 +2735,8 @@ S_regmatch(pTHX_ regnode *prog) */ if (OP(scan) == REFF) { STRLEN ulen1, ulen2; - U8 tmpbuf1[UTF8_MAXLEN*2+1]; - U8 tmpbuf2[UTF8_MAXLEN*2+1]; + U8 tmpbuf1[UTF8_MAXLEN_UCLC+1]; + U8 tmpbuf2[UTF8_MAXLEN_UCLC+1]; while (s < e) { if (l >= PL_regeol) sayNO; @@ -2962,7 +3166,7 @@ S_regmatch(pTHX_ regnode *prog) if (ckWARN(WARN_REGEXP) && n >= REG_INFTY && !(PL_reg_flags & RF_warned)) { PL_reg_flags |= RF_warned; - Perl_warner(aTHX_ WARN_REGEXP, "%s limit (%d) exceeded", + Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded", "Complex regular subexpression recursion", REG_INFTY - 1); } @@ -3014,7 +3218,7 @@ S_regmatch(pTHX_ regnode *prog) if (ckWARN(WARN_REGEXP) && n >= REG_INFTY && !(PL_reg_flags & RF_warned)) { PL_reg_flags |= RF_warned; - Perl_warner(aTHX_ WARN_REGEXP, "%s limit (%d) exceeded", + Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded", "Complex regular subexpression recursion", REG_INFTY - 1); } @@ -3112,20 +3316,32 @@ S_regmatch(pTHX_ regnode *prog) if (ln && l == 0) n = ln; /* don't backtrack */ locinput = PL_reginput; - if (NEAR_EXACT(next)) { + if (HAS_TEXT(next) || JUMPABLE(next)) { regnode *text_node = next; - if (PL_regkind[(U8)OP(next)] != EXACT) - NEXT_IMPT(text_node); + if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node); - if (PL_regkind[(U8)OP(text_node)] != EXACT) { - c1 = c2 = -1000; - } + if (! HAS_TEXT(text_node)) c1 = c2 = -1000; else { - c1 = (U8)*STRING(text_node); - if (OP(next) == EXACTF) + if (PL_regkind[(U8)OP(text_node)] == REF) { + I32 n, ln; + n = ARG(text_node); /* which paren pair */ + ln = PL_regstartp[n]; + /* assume yes if we haven't seen CLOSEn */ + if ( + *PL_reglastparen < n || + ln == -1 || + ln == PL_regendp[n] + ) { + c1 = c2 = -1000; + goto assume_ok_MM; + } + c1 = *(PL_bostr + ln); + } + else { c1 = (U8)*STRING(text_node); } + if (OP(text_node) == EXACTF || OP(text_node) == REFF) c2 = PL_fold[c1]; - else if (OP(text_node) == EXACTFL) + else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL) c2 = PL_fold_locale[c1]; else c2 = c1; @@ -3133,6 +3349,7 @@ S_regmatch(pTHX_ regnode *prog) } else c1 = c2 = -1000; + assume_ok_MM: REGCP_SET(lastcp); /* This may be improved if l == 0. */ while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */ @@ -3181,20 +3398,33 @@ S_regmatch(pTHX_ regnode *prog) (IV) n, (IV)l) ); if (n >= ln) { - if (NEAR_EXACT(next)) { + if (HAS_TEXT(next) || JUMPABLE(next)) { regnode *text_node = next; - if (PL_regkind[(U8)OP(next)] != EXACT) - NEXT_IMPT(text_node); + if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node); - if (PL_regkind[(U8)OP(text_node)] != EXACT) { - c1 = c2 = -1000; - } + if (! HAS_TEXT(text_node)) c1 = c2 = -1000; else { - c1 = (U8)*STRING(text_node); - if (OP(text_node) == EXACTF) + if (PL_regkind[(U8)OP(text_node)] == REF) { + I32 n, ln; + n = ARG(text_node); /* which paren pair */ + ln = PL_regstartp[n]; + /* assume yes if we haven't seen CLOSEn */ + if ( + *PL_reglastparen < n || + ln == -1 || + ln == PL_regendp[n] + ) { + c1 = c2 = -1000; + goto assume_ok_REG; + } + c1 = *(PL_bostr + ln); + } + else { c1 = (U8)*STRING(text_node); } + + if (OP(text_node) == EXACTF || OP(text_node) == REFF) c2 = PL_fold[c1]; - else if (OP(text_node) == EXACTFL) + else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL) c2 = PL_fold_locale[c1]; else c2 = c1; @@ -3203,6 +3433,7 @@ S_regmatch(pTHX_ regnode *prog) else c1 = c2 = -1000; } + assume_ok_REG: REGCP_SET(lastcp); while (n >= ln) { /* If it could work, try it. */ @@ -3275,31 +3506,43 @@ S_regmatch(pTHX_ regnode *prog) * of the quantifier and the EXACT-like node. -- japhy */ - if (NEAR_EXACT(next)) { + if (HAS_TEXT(next) || JUMPABLE(next)) { U8 *s; regnode *text_node = next; - if (PL_regkind[(U8)OP(next)] != EXACT) - NEXT_IMPT(text_node); + if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node); - if (PL_regkind[(U8)OP(text_node)] != EXACT) { - c1 = c2 = -1000; - } + if (! HAS_TEXT(text_node)) c1 = c2 = -1000; else { - s = (U8*)STRING(text_node); + if (PL_regkind[(U8)OP(text_node)] == REF) { + I32 n, ln; + n = ARG(text_node); /* which paren pair */ + ln = PL_regstartp[n]; + /* assume yes if we haven't seen CLOSEn */ + if ( + *PL_reglastparen < n || + ln == -1 || + ln == PL_regendp[n] + ) { + c1 = c2 = -1000; + goto assume_ok_easy; + } + s = (U8*)PL_bostr + ln; + } + else { s = (U8*)STRING(text_node); } if (!UTF) { c2 = c1 = *s; - if (OP(text_node) == EXACTF) + if (OP(text_node) == EXACTF || OP(text_node) == REFF) c2 = PL_fold[c1]; - else if (OP(text_node) == EXACTFL) + else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL) c2 = PL_fold_locale[c1]; } else { /* UTF */ - if (OP(text_node) == EXACTF) { + if (OP(text_node) == EXACTF || OP(text_node) == REFF) { STRLEN ulen1, ulen2; - U8 tmpbuf1[UTF8_MAXLEN*2+1]; - U8 tmpbuf2[UTF8_MAXLEN*2+1]; + U8 tmpbuf1[UTF8_MAXLEN_UCLC+1]; + U8 tmpbuf2[UTF8_MAXLEN_UCLC+1]; to_utf8_lower((U8*)s, tmpbuf1, &ulen1); to_utf8_upper((U8*)s, tmpbuf2, &ulen2); @@ -3315,6 +3558,7 @@ S_regmatch(pTHX_ regnode *prog) } else c1 = c2 = -1000; + assume_ok_easy: PL_reginput = locinput; if (minmod) { CHECKPOINT lastcp; @@ -3434,7 +3678,9 @@ S_regmatch(pTHX_ regnode *prog) n = regrepeat(scan, n); locinput = PL_reginput; if (ln < n && PL_regkind[(U8)OP(next)] == EOL && - (!PL_multiline || OP(next) == SEOL || OP(next) == EOS)) { + ((!PL_multiline && OP(next) != MEOL) || + OP(next) == SEOL || OP(next) == EOS)) + { ln = n; /* why back off? */ /* ...because $ and \Z can match before *and* after newline at the end. Consider "\n\n" =~ /\n+\Z\n/. @@ -3725,7 +3971,15 @@ S_regrepeat(pTHX_ regnode *p, I32 max) } break; case SANY: - scan = loceol; + if (do_utf8) { + loceol = PL_regeol; + while (scan < loceol && hardcount < max) { + scan += UTF8SKIP(scan); + hardcount++; + } + } + else + scan = loceol; break; case CANY: scan = loceol; @@ -3752,12 +4006,12 @@ S_regrepeat(pTHX_ regnode *p, I32 max) if (do_utf8) { loceol = PL_regeol; while (hardcount < max && scan < loceol && - reginclass(p, (U8*)scan, do_utf8)) { + reginclass(p, (U8*)scan, 0, do_utf8)) { scan += UTF8SKIP(scan); hardcount++; } } else { - while (scan < loceol && reginclass(p, (U8*)scan, do_utf8)) + while (scan < loceol && REGINCLASS(p, (U8*)scan)) scan++; } break; @@ -3983,10 +4237,11 @@ S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp) */ SV * -Perl_regclass_swash(pTHX_ register regnode* node, bool doinit, SV** initsvp) +Perl_regclass_swash(pTHX_ register regnode* node, bool doinit, SV** listsvp, SV **altsvp) { - SV *sw = NULL; - SV *si = NULL; + SV *sw = NULL; + SV *si = NULL; + SV *alt = NULL; if (PL_regdata && PL_regdata->count) { U32 n = ARG(node); @@ -3994,10 +4249,14 @@ Perl_regclass_swash(pTHX_ register regnode* node, bool doinit, SV** initsvp) if (PL_regdata->what[n] == 's') { SV *rv = (SV*)PL_regdata->data[n]; AV *av = (AV*)SvRV((SV*)rv); - SV **a; + SV **a, **b; - si = *av_fetch(av, 0, FALSE); - a = av_fetch(av, 1, FALSE); + /* See the end of regcomp.c:S_reglass() for + * documentation of these array elements. */ + + si = *av_fetch(av, 0, FALSE); + a = av_fetch(av, 1, FALSE); + b = av_fetch(av, 2, FALSE); if (a) sw = *a; @@ -4005,30 +4264,44 @@ Perl_regclass_swash(pTHX_ register regnode* node, bool doinit, SV** initsvp) sw = swash_init("utf8", "", si, 1, 0); (void)av_store(av, 1, sw); } + if (b) + alt = *b; } } - if (initsvp) - *initsvp = si; + if (listsvp) + *listsvp = si; + if (altsvp) + *altsvp = alt; return sw; } /* - reginclass - determine if a character falls into a character class + + The n is the ANYOF regnode, the p is the target string, lenp + is pointer to the maximum length of how far to go in the p + (if the lenp is zero, UTF8SKIP(p) is used), + do_utf8 tells whether the target string is in UTF-8. + */ STATIC bool -S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) +S_reginclass(pTHX_ register regnode *n, register U8* p, STRLEN* lenp, register bool do_utf8) { char flags = ANYOF_FLAGS(n); bool match = FALSE; UV c; STRLEN len = 0; + STRLEN plen; c = do_utf8 ? utf8_to_uvchr(p, &len) : *p; + plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c)); if (do_utf8 || (flags & ANYOF_UNICODE)) { + if (lenp) + *lenp = 0; if (do_utf8 && !ANYOF_RUNTIME(n)) { if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c)) match = TRUE; @@ -4036,21 +4309,41 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) if (!match && do_utf8 && (flags & ANYOF_UNICODE_ALL) && c >= 256) match = TRUE; if (!match) { - SV *sw = regclass_swash(n, TRUE, 0); + AV *av; + SV *sw = regclass_swash(n, TRUE, 0, (SV**)&av); if (sw) { if (swash_fetch(sw, p, do_utf8)) match = TRUE; else if (flags & ANYOF_FOLD) { - STRLEN ulen; - U8 tmpbuf[UTF8_MAXLEN*2+1]; + if (!match && lenp && av) { + I32 i; + + for (i = 0; i <= av_len(av); i++) { + SV* sv = *av_fetch(av, i, FALSE); + STRLEN len; + char *s = SvPV(sv, len); + + if (len <= plen && memEQ(s, (char*)p, len)) { + *lenp = len; + match = TRUE; + break; + } + } + } + if (!match) { + U8 tmpbuf[UTF8_MAXLEN_FOLD+1]; + STRLEN tmplen; - toLOWER_utf8(p, tmpbuf, &ulen); - if (swash_fetch(sw, tmpbuf, do_utf8)) - match = TRUE; + to_utf8_fold(p, tmpbuf, &tmplen); + if (swash_fetch(sw, tmpbuf, do_utf8)) + match = TRUE; + } } } } + if (match && lenp && *lenp == 0) + *lenp = UNISKIP(NATIVE_TO_UNI(c)); } if (!match && c < 256) { if (ANYOF_BITMAP_TEST(n, c)) @@ -4191,3 +4484,59 @@ restore_pos(pTHX_ void *arg) PL_curpm = PL_reg_oldcurpm; } } + +STATIC void +S_to_utf8_substr(pTHX_ register regexp *prog) +{ + SV* sv; + if (prog->float_substr && !prog->float_utf8) { + prog->float_utf8 = sv = NEWSV(117, 0); + SvSetMagicSV(sv, prog->float_substr); + sv_utf8_upgrade(sv); + if (SvTAIL(prog->float_substr)) + SvTAIL_on(sv); + if (prog->float_substr == prog->check_substr) + prog->check_utf8 = sv; + } + if (prog->anchored_substr && !prog->anchored_utf8) { + prog->anchored_utf8 = sv = NEWSV(118, 0); + SvSetMagicSV(sv, prog->anchored_substr); + sv_utf8_upgrade(sv); + if (SvTAIL(prog->anchored_substr)) + SvTAIL_on(sv); + if (prog->anchored_substr == prog->check_substr) + prog->check_utf8 = sv; + } +} + +STATIC void +S_to_byte_substr(pTHX_ register regexp *prog) +{ + SV* sv; + if (prog->float_utf8 && !prog->float_substr) { + prog->float_substr = sv = NEWSV(117, 0); + SvSetMagicSV(sv, prog->float_utf8); + if (sv_utf8_downgrade(sv, TRUE)) { + if (SvTAIL(prog->float_utf8)) + SvTAIL_on(sv); + } else { + SvREFCNT_dec(sv); + prog->float_substr = sv = &PL_sv_undef; + } + if (prog->float_utf8 == prog->check_utf8) + prog->check_substr = sv; + } + if (prog->anchored_utf8 && !prog->anchored_substr) { + prog->anchored_substr = sv = NEWSV(118, 0); + SvSetMagicSV(sv, prog->anchored_utf8); + if (sv_utf8_downgrade(sv, TRUE)) { + if (SvTAIL(prog->anchored_utf8)) + SvTAIL_on(sv); + } else { + SvREFCNT_dec(sv); + prog->anchored_substr = sv = &PL_sv_undef; + } + if (prog->anchored_utf8 == prog->check_utf8) + prog->check_substr = sv; + } +}