X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/80aecb99acb05e810c6136645b97c6bc9f385ca3..492652be590915fcb2621eeceaf000a1c070956a:/regexec.c diff --git a/regexec.c b/regexec.c index e67774d..61d175a 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. @@ -111,7 +111,7 @@ #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) \ @@ -140,13 +140,18 @@ PL_regkind[(U8)OP(rn)] == EXACT || PL_regkind[(U8)OP(rn)] == REF \ ) +/* + 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 || \ - PL_regkind[(U8)OP(rn)] == CURLY) \ + 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 @@ -401,7 +406,8 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, DEBUG_r({ char *s = PL_reg_match_utf8 ? - sv_uni_display(dsv, sv, 60, 0) : strpos; + sv_uni_display(dsv, sv, 60, UNI_DISPLAY_REGEX) : + strpos; int len = PL_reg_match_utf8 ? strlen(s) : strend - strpos; if (!PL_colorset) @@ -915,15 +921,22 @@ 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 (reginclass(c, (U8*)s, do_utf8) || + (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 - tmp = 1; - s += do_utf8 ? UTF8SKIP(s) : 1; + else + tmp = 1; + s += skip; } break; case CANY: @@ -946,8 +959,8 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta 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; @@ -960,45 +973,47 @@ 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) { UV c, f; U8 tmpbuf [UTF8_MAXLEN+1]; U8 foldbuf[UTF8_MAXLEN_FOLD+1]; STRLEN len, foldlen; - - /* The ibcmp_utf8() uses to_uni_fold() which is more - * correct folding for Unicode than using lowercase. - * However, it doesn't work quite fully since the folding - * is a one-to-many mapping and the regex optimizer is - * unaware of this, so it may throw out good matches. - * Fortunately, not getting this right is allowed - * for Unicode Regular Expression Support level 1, - * only one-to-one matching is required. --jhi */ - + if (c1 == c2) { while (s <= e) { c = utf8_to_uvchr((U8*)s, &len); if ( c == c1 && (ln == len || - !ibcmp_utf8(s, do_utf8, strend - s, - m, UTF, ln)) + 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); - to_utf8_fold(tmpbuf, foldbuf, &foldlen); - f = utf8_to_uvchr(foldbuf, 0); + f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); if ( f != c && (f == c1 || f == c2) && (ln == foldlen || - !ibcmp_utf8((char *)foldbuf, - do_utf8, foldlen, - m, UTF, ln)) + !ibcmp_utf8((char *) foldbuf, + (char **)0, foldlen, do_utf8, + m, + (char **)0, ln, UTF)) && (norun || regtry(prog, s)) ) goto got_it; } @@ -1009,26 +1024,33 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta while (s <= e) { 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, do_utf8, strend - s, - m, UTF, ln)) + 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); - to_utf8_fold(tmpbuf, foldbuf, &foldlen); - f = utf8_to_uvchr(foldbuf, 0); + f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); if ( f != c && (f == c1 || f == c2) && (ln == foldlen || - !ibcmp_utf8((char *)foldbuf, - do_utf8, foldlen, - m, UTF, ln)) + !ibcmp_utf8((char *) foldbuf, + (char **)0, foldlen, do_utf8, + m, + (char **)0, ln, UTF)) && (norun || regtry(prog, s)) ) goto got_it; } @@ -1527,7 +1549,8 @@ 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 = PERL_DEBUG_PAD_ZERO(0); + SV *dsv0 = PERL_DEBUG_PAD_ZERO(0); + SV *dsv1 = PERL_DEBUG_PAD_ZERO(1); #endif PL_regcc = 0; @@ -1613,20 +1636,26 @@ 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 ? "..." : "") ); }); @@ -1797,8 +1826,24 @@ 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; @@ -2076,6 +2121,7 @@ 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 @@ -2164,17 +2210,17 @@ 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, @@ -2290,7 +2336,7 @@ S_regmatch(pTHX_ regnode *prog) if (l >= PL_regeol) sayNO; if (NATIVE_TO_UNI(*(U8*)s) != - utf8_to_uvchr((U8*)l, &ulen)) + utf8_to_uvuni((U8*)l, &ulen)) sayNO; l += ulen; s ++; @@ -2302,7 +2348,7 @@ S_regmatch(pTHX_ regnode *prog) if (l >= PL_regeol) sayNO; if (NATIVE_TO_UNI(*((U8*)l)) != - utf8_to_uvchr((U8*)s, &ulen)) + utf8_to_uvuni((U8*)s, &ulen)) sayNO; s += ulen; l ++; @@ -2330,92 +2376,29 @@ S_regmatch(pTHX_ regnode *prog) s = STRING(scan); ln = STR_LEN(scan); - { + if (do_utf8 || UTF) { + /* Either target or the pattern are utf8. */ char *l = locinput; - char *e = s + ln; - U8 tmpbuf[UTF8_MAXLEN_FOLD+1]; - - if (do_utf8 != (UTF!=0)) { - /* The target and the pattern have differing utf8ness. */ - STRLEN ulen1, ulen2; - UV cs, cl; - - if (do_utf8) { - /* The target is utf8, the pattern is not utf8. */ - while (s < e) { - if (l >= PL_regeol) - sayNO; - - cs = to_uni_fold(NATIVE_TO_UNI(*(U8*)s), - (U8*)s, &ulen1); - cl = utf8_to_uvchr((U8*)l, &ulen2); - - if (cs != cl) { - cl = to_uni_fold(cl, (U8*)l, &ulen2); - if (ulen1 != ulen2 || cs != cl) - sayNO; - } - l += ulen1; - s ++; - } - } - else { - /* The target is not utf8, the pattern is utf8. */ - while (s < e) { - if (l >= PL_regeol) - sayNO; - - cs = utf8_to_uvchr((U8*)s, &ulen1); - - cl = to_uni_fold(NATIVE_TO_UNI(*(U8*)l), - (U8*)l, &ulen2); - - if (cs != cl) { - cs = to_uni_fold(cs, (U8*)s, &ulen1); - if (ulen1 != ulen2 || cs != cl) - sayNO; - } - l ++; - s += ulen1; - } - } - locinput = l; - nextchr = UCHARAT(locinput); - break; - } - - if (do_utf8 && UTF) { - /* Both the target and the pattern are utf8. */ - STRLEN ulen; - - while (s < e) { - if (l >= PL_regeol) - sayNO; - if (UTF8SKIP(s) != UTF8SKIP(l) || - memNE(s, (char*)l, UTF8SKIP(s))) { - U8 lfoldbuf[UTF8_MAXLEN_FOLD+1]; - STRLEN lfoldlen; - - to_utf8_fold((U8*)l, lfoldbuf, &lfoldlen); - if (UTF8SKIP(s) != lfoldlen || - memNE(s, (char*)lfoldbuf, lfoldlen)) { - U8 sfoldbuf[UTF8_MAXLEN_FOLD+1]; - STRLEN sfoldlen; - - to_utf8_fold((U8*)s, sfoldbuf, &sfoldlen); - if (sfoldlen != lfoldlen || - memNE((char*)sfoldbuf, - (char*)lfoldbuf, lfoldlen)) - sayNO; - } - } - l += UTF8SKIP(l); - s += UTF8SKIP(s); - } - locinput = l; - nextchr = UCHARAT(locinput); - break; + 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 = e; + nextchr = UCHARAT(locinput); + break; } /* Neither the target and the pattern are utf8. */ @@ -2436,22 +2419,36 @@ 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 (!reginclasslen(scan, (U8*)locinput, &inclasslen, do_utf8)) + sayNO_ANYOF; if (locinput >= PL_regeol) sayNO; - locinput += PL_utf8skip[nextchr]; + locinput += inclasslen; nextchr = UCHARAT(locinput); + break; } else { if (nextchr < 0) nextchr = UCHARAT(locinput); if (!reginclass(scan, (U8*)locinput, do_utf8)) - sayNO; + 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; @@ -3625,7 +3622,9 @@ 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/. @@ -3916,7 +3915,15 @@ 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; @@ -4174,10 +4181,11 @@ 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); @@ -4185,10 +4193,14 @@ 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; @@ -4196,30 +4208,44 @@ 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 + - reginclasslen - 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_reginclasslen(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(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; @@ -4227,24 +4253,41 @@ 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_FOLD+1]; - - to_utf8_fold(p, tmpbuf, &ulen); - if (swash_fetch(sw, tmpbuf, do_utf8)) - match = TRUE; - to_utf8_upper(p, tmpbuf, &ulen); - if (swash_fetch(sw, tmpbuf, do_utf8)) - match = TRUE; + 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, p, len)) { + *lenp = len; + match = TRUE; + break; + } + } + } + if (!match) { + U8 tmpbuf[UTF8_MAXLEN_FOLD+1]; + STRLEN tmplen; + + to_utf8_fold(p, tmpbuf, &tmplen); + if (swash_fetch(sw, tmpbuf, do_utf8)) + match = TRUE; + } } } } + if (match && lenp && *lenp == 0) + *lenp = UNISKIP(c); } if (!match && c < 256) { if (ANYOF_BITMAP_TEST(n, c)) @@ -4305,6 +4348,20 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) return (flags & ANYOF_INVERT) ? !match : match; } +/* + - reginclass - determine if a character falls into a character class + + The n is the ANYOF regnode, the p is the target string, 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) +{ + return S_reginclasslen(aTHX_ n, p, 0, do_utf8); +} + STATIC U8 * S_reghop(pTHX_ U8 *s, I32 off) {