X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=e358d63d5f2d5e18fc45aca59ec2d725d460ee8b;hb=d235852b65d51c442c4a84cfa07cd04d89233033;hp=0b65d110f210aff70fdc7091182ea241c1b01a27;hpb=fd400ab9029cac48f6054d57b59a99ec1c5d5e1e;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index 0b65d11..e358d63 100644 --- a/regexec.c +++ b/regexec.c @@ -38,11 +38,11 @@ # define Perl_re_intuit_start my_re_intuit_start /* *These* symbols are masked to allow static link. */ # define Perl_pregexec my_pregexec -# define Perl_reginitcolors my_reginitcolors +# define Perl_reginitcolors my_reginitcolors # define Perl_regclass_swash my_regclass_swash # define PERL_NO_GET_CONTEXT -#endif +#endif /*SUPPRESS 112*/ /* @@ -123,28 +123,37 @@ #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim)) #define HOPMAYBE3c(pos,off,lim) ((char*)HOPMAYBE3(pos,off,lim)) -static void restore_pos(pTHXo_ void *arg); +#define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) (void)CAT2(is_utf8_, a)((U8*)b); } STMT_END +static void restore_pos(pTHXo_ void *arg); STATIC CHECKPOINT S_regcppush(pTHX_ I32 parenfloor) { int retval = PL_savestack_ix; - int i = (PL_regsize - parenfloor) * 4; +#define REGCP_PAREN_ELEMS 4 + int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; int p; - SSCHECK(i + 5); +#define REGCP_OTHER_ELEMS 5 + SSCHECK(paren_elems_to_push + REGCP_OTHER_ELEMS); for (p = PL_regsize; p > parenfloor; p--) { +/* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ SSPUSHINT(PL_regendp[p]); SSPUSHINT(PL_regstartp[p]); SSPUSHPTR(PL_reg_start_tmp[p]); SSPUSHINT(p); } +/* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ SSPUSHINT(PL_regsize); SSPUSHINT(*PL_reglastparen); SSPUSHPTR(PL_reginput); - SSPUSHINT(i + 3); - SSPUSHINT(SAVEt_REGCONTEXT); +#define REGCP_FRAME_ELEMS 2 +/* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and + * are needed for the regexp context stack bookkeeping. */ + SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); + SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */ + return retval; } @@ -161,16 +170,22 @@ S_regcppush(pTHX_ I32 parenfloor) STATIC char * S_regcppop(pTHX) { - I32 i = SSPOPINT; + I32 i; U32 paren = 0; char *input; I32 tmps; - assert(i == SAVEt_REGCONTEXT); + + /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ i = SSPOPINT; + assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ + i = SSPOPINT; /* Parentheses elements to pop. */ input = (char *) SSPOPPTR; *PL_reglastparen = SSPOPINT; PL_regsize = SSPOPINT; - for (i -= 3; i > 0; i -= 4) { + + /* Now restore the parentheses context. */ + for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); + i > 0; i -= REGCP_PAREN_ELEMS) { paren = (U32)SSPOPINT; PL_reg_start_tmp[paren] = (char *) SSPOPPTR; PL_regstartp[paren] = SSPOPINT; @@ -180,9 +195,9 @@ S_regcppop(pTHX) DEBUG_r( PerlIO_printf(Perl_debug_log, " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", - (UV)paren, (IV)PL_regstartp[paren], + (UV)paren, (IV)PL_regstartp[paren], (IV)(PL_reg_start_tmp[paren] - PL_bostr), - (IV)PL_regendp[paren], + (IV)PL_regendp[paren], (paren > *PL_reglastparen ? "(no)" : "")); ); } @@ -267,7 +282,7 @@ Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *stren /* nosave: For optimizations. */ { return - regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, + regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, nosave ? 0 : REXEC_COPY_STR); } @@ -279,11 +294,11 @@ S_cache_re(pTHX_ regexp *prog) PL_regprogram = prog->program; #endif PL_regnpar = prog->nparens; - PL_regdata = prog->data; - PL_reg_re = prog; + PL_regdata = prog->data; + PL_reg_re = prog; } -/* +/* * Need to implement the following flags for reg_anch: * * USE_INTUIT_NOML - Useful to call re_intuit_start() first @@ -377,14 +392,16 @@ 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)) { DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); goto fail; } - if (prog->check_offset_min == prog->check_offset_max) { + if (prog->check_offset_min == prog->check_offset_max && + !(prog->reganch & ROPT_SANY_SEEN)) { /* Substring at constant offset from beg-of-str... */ I32 slen; @@ -392,7 +409,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if (SvTAIL(check)) { slen = SvCUR(check); /* >= 1 */ - if ( strend - s > slen || strend - s < slen - 1 + if ( strend - s > slen || strend - s < slen - 1 || (strend - s == slen && strend[-1] != '\n')) { DEBUG_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); goto fail_finish; @@ -453,13 +470,17 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, || ( BmRARE(check) == '\n' && (BmPREVIOUS(check) == SvCUR(check) - 1) && SvTAIL(check) )) - s = screaminstr(sv, check, + s = screaminstr(sv, check, start_shift + (s - strbeg), end_shift, pp, 0); else goto fail_finish; if (data) *data->scream_olds = s; } + else if (prog->reganch & ROPT_SANY_SEEN) + s = fbm_instr((U8*)(s + start_shift), + (U8*)(strend - end_shift), + check, PL_multiline ? FBMrf_MULTILINE : 0); else s = fbm_instr(HOP3(s, start_shift, strend), HOP3(strend, -end_shift, strbeg), @@ -595,14 +616,14 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, DEBUG_r(PerlIO_printf(Perl_debug_log, ", trying anchored starting at offset %ld...\n", (long)(s1 + 1 - i_strpos))); - other_last = last + 1; + 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 + 1; + other_last = s; /* Fix this later. --Hugo */ s = s1; if (t == strpos) goto try_at_start; @@ -638,7 +659,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, Thus we can arrive here only if check substr is float. Redo checking for "other"=="fixed". */ - strpos = t + 1; + strpos = t + 1; DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", PL_colors[0],PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); goto do_other_anchored; @@ -779,7 +800,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if (t + start_shift >= check_at) /* Contradicts floating=check */ goto retry_floating_check; /* Recheck anchored substring, but not floating... */ - s = check_at; + s = check_at; if (!check) goto giveup; DEBUG_r( PerlIO_printf(Perl_debug_log, @@ -807,11 +828,11 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, goto hop_and_restart; } DEBUG_r( if (t != s) - PerlIO_printf(Perl_debug_log, + PerlIO_printf(Perl_debug_log, "By STCLASS: moving %ld --> %ld\n", (long)(t - i_strpos), (long)(s - i_strpos)); else - PerlIO_printf(Perl_debug_log, + PerlIO_printf(Perl_debug_log, "Does not contradict STCLASS...\n") ); } giveup: @@ -884,14 +905,14 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta STRLEN len; if (c1 == c2) while (s <= e) { - if ( utf8_to_uv_simple((U8*)s, &len) == c1 + if ( utf8_to_uvchr((U8*)s, &len) == c1 && regtry(prog, s) ) goto got_it; s += len; } else while (s <= e) { - UV c = utf8_to_uv_simple((U8*)s, &len); + UV c = utf8_to_uvchr((U8*)s, &len); if ( (c == c1 || c == c2) && regtry(prog, s) ) goto got_it; s += len; @@ -929,11 +950,12 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta tmp = '\n'; else { U8 *r = reghop3((U8*)s, -1, (U8*)startpos); - - tmp = (I32)utf8_to_uv(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_uni(tmp)) != 0); + 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) : @@ -971,11 +993,12 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta tmp = '\n'; else { U8 *r = reghop3((U8*)s, -1, (U8*)startpos); - - tmp = (I32)utf8_to_uv(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_uni(tmp)) != 0); + 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) : @@ -1004,6 +1027,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case ALNUM: if (do_utf8) { + LOAD_UTF8_CHARCLASS(alnum,"a"); while (s < strend) { if (swash_fetch(PL_utf8_alnum, (U8*)s)) { if (tmp && (norun || regtry(prog, s))) @@ -1061,6 +1085,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case NALNUM: if (do_utf8) { + LOAD_UTF8_CHARCLASS(alnum,"a"); while (s < strend) { if (!swash_fetch(PL_utf8_alnum, (U8*)s)) { if (tmp && (norun || regtry(prog, s))) @@ -1118,6 +1143,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case SPACE: if (do_utf8) { + LOAD_UTF8_CHARCLASS(space," "); while (s < strend) { if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) { if (tmp && (norun || regtry(prog, s))) @@ -1175,6 +1201,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case NSPACE: if (do_utf8) { + LOAD_UTF8_CHARCLASS(space," "); while (s < strend) { if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) { if (tmp && (norun || regtry(prog, s))) @@ -1232,6 +1259,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case DIGIT: if (do_utf8) { + LOAD_UTF8_CHARCLASS(digit,"0"); while (s < strend) { if (swash_fetch(PL_utf8_digit,(U8*)s)) { if (tmp && (norun || regtry(prog, s))) @@ -1289,6 +1317,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta break; case NDIGIT: if (do_utf8) { + LOAD_UTF8_CHARCLASS(digit,"0"); while (s < strend) { if (!swash_fetch(PL_utf8_digit,(U8*)s)) { if (tmp && (norun || regtry(prog, s))) @@ -1382,7 +1411,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * cache_re(prog); #ifdef DEBUGGING - PL_regnarrate = PL_debug & 512; + PL_regnarrate = DEBUG_r_TEST; #endif /* Be paranoid... */ @@ -1393,7 +1422,8 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * minlen = prog->minlen; if (do_utf8) { - if (utf8_distance((U8*)strend, (U8*)startpos) < minlen) goto phooey; + if (!(prog->reganch & ROPT_SANY_SEEN)) + if (utf8_distance((U8*)strend, (U8*)startpos) < minlen) goto phooey; } else { if (strend - startpos < minlen) goto phooey; @@ -1404,7 +1434,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * else { if (prog->reganch & ROPT_UTF8 && do_utf8) { U8 *s = reghop3((U8*)stringarg, -1, (U8*)strbeg); - PL_regprev = utf8_to_uv(s, (U8*)stringarg - s, NULL, 0); + PL_regprev = utf8n_to_uvchr(s, (U8*)stringarg - s, NULL, 0); } else PL_regprev = (U32)stringarg[-1]; @@ -1460,7 +1490,8 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * PL_reg_ganch = strbeg; } - if (!(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) { + if (do_utf8 == (UTF!=0) && + !(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) { re_scream_pos_data d; d.scream_olds = &scream_olds; @@ -1533,7 +1564,7 @@ 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->reganch & ROPT_SKIP) { /* we have /x+whatever/ */ /* it must be a one character string (XXXX Except UTF?) */ char ch = SvPVX(prog->anchored_substr)[0]; @@ -1572,13 +1603,13 @@ 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_substr != Nullsv && prog->float_max_offset < strend - s))) { - SV *must = prog->anchored_substr + SV *must = prog->anchored_substr ? prog->anchored_substr : prog->float_substr; - I32 back_max = + I32 back_max = prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset; - I32 back_min = + 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) @@ -1599,11 +1630,11 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * dontbother = end_shift; strend = HOPc(strend, -dontbother); while ( (s <= last) && - ((flags & REXEC_SCREAM) + ((flags & REXEC_SCREAM) ? (s = screaminstr(sv, must, HOP3c(s, back_min, strend) - strbeg, end_shift, &scream_pos, 0)) : (s = fbm_instr((unsigned char*)HOP3(s, back_min, strend), - (unsigned char*)strend, must, + (unsigned char*)strend, must, PL_multiline ? FBMrf_MULTILINE : 0))) ) { DEBUG_r( did_match = 1 ); if (HOPc(s, -back_max) > last1) { @@ -1673,13 +1704,13 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * if (memEQ(strend - len + 1, little, len - 1)) last = strend - len + 1; else if (!PL_multiline) - last = memEQ(strend - len, little, len) + last = memEQ(strend - len, little, len) ? strend - len : Nullch; else goto find_last; } else { find_last: - if (len) + if (len) last = rninstr(s, strend, little, little + len); else last = strend; /* matching `$' */ @@ -1748,7 +1779,7 @@ got_it: prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ } } - + return 1; phooey: @@ -1797,7 +1828,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) DEFSV = PL_reg_sv; } - if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv) + if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv) && (mg = mg_find(PL_reg_sv, 'g')))) { /* prepare for quick setting of pos */ sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0); @@ -1975,7 +2006,7 @@ S_regmatch(pTHX_ regnode *prog) after the current position the third one. We assume that pref0_len <= pref_len, otherwise we decrease pref0_len. */ - int pref_len = (locinput - PL_bostr) > (5 + taill) - l + int pref_len = (locinput - PL_bostr) > (5 + taill) - l ? (5 + taill) - l : locinput - PL_bostr; int pref0_len; @@ -1983,7 +2014,7 @@ S_regmatch(pTHX_ regnode *prog) pref_len++; pref0_len = pref_len - (locinput - PL_reg_starttry); if (l + pref_len < (5 + taill) && l < PL_regeol - locinput) - l = ( PL_regeol - locinput > (5 + taill) - pref_len + l = ( PL_regeol - locinput > (5 + taill) - pref_len ? (5 + taill) - pref_len : PL_regeol - locinput); while (UTF8_IS_CONTINUATION(*(U8*)(locinput + l))) l--; @@ -1992,12 +2023,12 @@ S_regmatch(pTHX_ regnode *prog) if (pref0_len > pref_len) pref0_len = pref_len; regprop(prop, scan); - PerlIO_printf(Perl_debug_log, + PerlIO_printf(Perl_debug_log, "%4"IVdf" <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3"IVdf":%*s%s\n", - (IV)(locinput - PL_bostr), - PL_colors[4], pref0_len, + (IV)(locinput - PL_bostr), + PL_colors[4], pref0_len, locinput - pref_len, PL_colors[5], - PL_colors[2], pref_len - pref0_len, + PL_colors[2], pref_len - pref0_len, locinput - pref_len + pref0_len, PL_colors[3], (docolor ? "" : "> <"), PL_colors[0], l, locinput, PL_colors[1], @@ -2015,7 +2046,7 @@ S_regmatch(pTHX_ regnode *prog) case BOL: if (locinput == PL_bostr ? PL_regprev == '\n' - : (PL_multiline && + : (PL_multiline && (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) { /* regtill = regbol; */ @@ -2060,13 +2091,6 @@ S_regmatch(pTHX_ regnode *prog) sayNO; break; case SANY: - if (do_utf8) { - locinput += PL_utf8skip[nextchr]; - if (locinput > PL_regeol) - sayNO; - nextchr = UCHARAT(locinput); - break; - } if (!nextchr && locinput >= PL_regeol) sayNO; nextchr = UCHARAT(++locinput); @@ -2094,7 +2118,7 @@ S_regmatch(pTHX_ regnode *prog) while (s < e) { if (l >= PL_regeol) sayNO; - if (*((U8*)s) != utf8_to_uv_simple((U8*)l, &len)) + if (*((U8*)s) != utf8_to_uvchr((U8*)l, &len)) sayNO; s++; l += len; @@ -2103,7 +2127,7 @@ S_regmatch(pTHX_ regnode *prog) while (s < e) { if (l >= PL_regeol) sayNO; - if (*((U8*)l) != utf8_to_uv_simple((U8*)s, &len)) + if (*((U8*)l) != utf8_to_uvchr((U8*)s, &len)) sayNO; s += len; l++; @@ -2138,7 +2162,7 @@ S_regmatch(pTHX_ regnode *prog) if (l >= PL_regeol) { sayNO; } - if ((UTF ? utf8_to_uv((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; @@ -2211,6 +2235,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr && locinput >= PL_regeol) sayNO; if (do_utf8) { + LOAD_UTF8_CHARCLASS(alnum,"a"); if (OP(scan) == NALNUM ? swash_fetch(PL_utf8_alnum, (U8*)locinput) : isALNUM_LC_utf8((U8*)locinput)) @@ -2238,15 +2263,16 @@ S_regmatch(pTHX_ regnode *prog) ln = PL_regprev; else { U8 *r = reghop((U8*)locinput, -1); - - ln = utf8_to_uv(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); } else { - ln = isALNUM_LC_uni(ln); + ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln)); n = isALNUM_LC_utf8((U8*)locinput); } } @@ -2274,6 +2300,7 @@ S_regmatch(pTHX_ regnode *prog) sayNO; if (do_utf8) { if (UTF8_IS_CONTINUED(nextchr)) { + LOAD_UTF8_CHARCLASS(space," "); if (!(OP(scan) == SPACE ? swash_fetch(PL_utf8_space, (U8*)locinput) : isSPACE_LC_utf8((U8*)locinput))) @@ -2303,6 +2330,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr && locinput >= PL_regeol) sayNO; if (do_utf8) { + LOAD_UTF8_CHARCLASS(space," "); if (OP(scan) == NSPACE ? swash_fetch(PL_utf8_space, (U8*)locinput) : isSPACE_LC_utf8((U8*)locinput)) @@ -2325,6 +2353,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr) sayNO; if (do_utf8) { + LOAD_UTF8_CHARCLASS(digit,"0"); if (!(OP(scan) == DIGIT ? swash_fetch(PL_utf8_digit, (U8*)locinput) : isDIGIT_LC_utf8((U8*)locinput))) @@ -2347,6 +2376,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr && locinput >= PL_regeol) sayNO; if (do_utf8) { + LOAD_UTF8_CHARCLASS(digit,"0"); if (OP(scan) == NDIGIT ? swash_fetch(PL_utf8_digit, (U8*)locinput) : isDIGIT_LC_utf8((U8*)locinput)) @@ -2363,6 +2393,7 @@ 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)) sayNO; locinput += PL_utf8skip[nextchr]; @@ -2450,7 +2481,7 @@ S_regmatch(pTHX_ regnode *prog) COP *ocurcop = PL_curcop; SV **ocurpad = PL_curpad; SV *ret; - + n = ARG(scan); PL_op = (OP_4tree*)PL_regdata->data[n]; DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); @@ -2461,7 +2492,7 @@ S_regmatch(pTHX_ regnode *prog) SPAGAIN; ret = POPs; PUTBACK; - + PL_op = oop; PL_curpad = ocurpad; PL_curcop = ocurcop; @@ -2490,9 +2521,9 @@ S_regmatch(pTHX_ regnode *prog) I32 osize = PL_regsize; I32 onpar = PL_regnpar; - pm.op_pmflags = 0; + Zero(&pm, 1, PMOP); re = CALLREGCOMP(aTHX_ t, t + len, &pm); - if (!(SvFLAGS(ret) + if (!(SvFLAGS(ret) & (SVs_TEMP | SVs_PADTMP | SVf_READONLY))) sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0); PL_regprecomp = oprecomp; @@ -2500,7 +2531,7 @@ S_regmatch(pTHX_ regnode *prog) PL_regnpar = onpar; } DEBUG_r( - PerlIO_printf(Perl_debug_log, + PerlIO_printf(Perl_debug_log, "Entering embedded `%s%.60s%s%s'\n", PL_colors[0], re->precomp, @@ -2513,7 +2544,7 @@ S_regmatch(pTHX_ regnode *prog) state.re = PL_reg_re; PL_regcc = 0; - + cp = regcppush(0); /* Save *all* the positions. */ REGCP_SET(lastcp); cache_re(re); @@ -2602,7 +2633,7 @@ S_regmatch(pTHX_ regnode *prog) 1) After matching X, regnode for CURLYX is processed; - 2) This regnode creates infoblock on the stack, and calls + 2) This regnode creates infoblock on the stack, and calls regmatch() recursively with the starting point at WHILEM node; 3) Each hit of WHILEM node tries to match A and Z (in the order @@ -2623,7 +2654,7 @@ S_regmatch(pTHX_ regnode *prog) and whatever it mentions via ->next, and additional attached trees corresponding to temporarily unset infoblocks as in "5" above. - In the following picture infoblocks for outer loop of + In the following picture infoblocks for outer loop of (Y(A)*?Z)*?T are denoted O, for inner I. NULL starting block is denoted by x. The matched string is YAAZYAZT. Temporarily postponed infoblocks are drawn below the "reset" infoblock. @@ -2716,10 +2747,10 @@ S_regmatch(pTHX_ regnode *prog) PL_reginput = locinput; DEBUG_r( - PerlIO_printf(Perl_debug_log, - "%*s %ld out of %ld..%ld cc=%lx\n", + PerlIO_printf(Perl_debug_log, + "%*s %ld out of %ld..%ld cc=%lx\n", REPORT_CODE_OFF+PL_regindent*2, "", - (long)n, (long)cc->min, + (long)n, (long)cc->min, (long)cc->max, (long)cc) ); @@ -2818,7 +2849,7 @@ S_regmatch(pTHX_ regnode *prog) PL_regcc = cc; if (n >= cc->max) { /* Maximum greed exceeded? */ - if (ckWARN(WARN_REGEXP) && n >= REG_INFTY + 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", @@ -2870,7 +2901,7 @@ S_regmatch(pTHX_ regnode *prog) REPORT_CODE_OFF+PL_regindent*2, "") ); } - if (ckWARN(WARN_REGEXP) && n >= REG_INFTY + 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", @@ -2892,13 +2923,13 @@ S_regmatch(pTHX_ regnode *prog) sayNO; } /* NOT REACHED */ - case BRANCHJ: + case BRANCHJ: next = scan + ARG(scan); if (next == scan) next = NULL; inner = NEXTOPER(NEXTOPER(scan)); goto do_branch; - case BRANCH: + case BRANCH: inner = NEXTOPER(scan); do_branch: { @@ -2941,7 +2972,7 @@ S_regmatch(pTHX_ regnode *prog) { I32 l = 0; CHECKPOINT lastcp; - + /* We suppose that the next guy does not need backtracking: in particular, it is of constant length, and has no parenths to influence future backrefs. */ @@ -3119,7 +3150,7 @@ S_regmatch(pTHX_ regnode *prog) c2 = to_utf8_upper(s); } else { - c2 = c1 = utf8_to_uv_simple(s, NULL); + c2 = c1 = utf8_to_uvchr(s, NULL); } } } @@ -3162,7 +3193,7 @@ S_regmatch(pTHX_ regnode *prog) while (locinput <= e && *locinput != c1) locinput++; } else { - while (locinput <= e + while (locinput <= e && *locinput != c1 && *locinput != c2) locinput++; @@ -3174,20 +3205,20 @@ S_regmatch(pTHX_ regnode *prog) if (c1 == c2) { for (count = 0; locinput <= e && - utf8_to_uv_simple((U8*)locinput, &len) != c1; + utf8_to_uvchr((U8*)locinput, &len) != c1; count++) locinput += len; } else { for (count = 0; locinput <= e; count++) { - UV c = utf8_to_uv_simple((U8*)locinput, &len); + UV c = utf8_to_uvchr((U8*)locinput, &len); if (c == c1 || c == c2) break; - locinput += len; + locinput += len; } } } - if (locinput > e) + if (locinput > e) sayNO; /* PL_reginput == old now */ if (locinput != old) { @@ -3212,12 +3243,18 @@ S_regmatch(pTHX_ regnode *prog) UV c; if (c1 != -1000) { if (do_utf8) - c = utf8_to_uv_simple((U8*)PL_reginput, NULL); + c = utf8_to_uvchr((U8*)PL_reginput, NULL); else - c = UCHARAT(PL_reginput); + 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); @@ -3251,9 +3288,9 @@ S_regmatch(pTHX_ regnode *prog) while (n >= ln) { if (c1 != -1000) { if (do_utf8) - c = utf8_to_uv_simple((U8*)PL_reginput, NULL); + c = utf8_to_uvchr((U8*)PL_reginput, NULL); else - c = UCHARAT(PL_reginput); + c = UCHARAT(PL_reginput); } /* If it could work, try it. */ if (c1 == -1000 || c == c1 || c == c2) @@ -3271,9 +3308,9 @@ S_regmatch(pTHX_ regnode *prog) while (n >= ln) { if (c1 != -1000) { if (do_utf8) - c = utf8_to_uv_simple((U8*)PL_reginput, NULL); + c = utf8_to_uvchr((U8*)PL_reginput, NULL); else - c = UCHARAT(PL_reginput); + c = UCHARAT(PL_reginput); } /* If it could work, try it. */ if (c1 == -1000 || c == c1 || c == c2) @@ -3341,7 +3378,7 @@ S_regmatch(pTHX_ regnode *prog) case SUSPEND: n = 1; PL_reginput = locinput; - goto do_ifmatch; + goto do_ifmatch; case UNLESSM: n = 0; if (scan->flags) { @@ -3354,7 +3391,7 @@ S_regmatch(pTHX_ regnode *prog) PL_reginput = s; } else { - if (locinput < PL_bostr + scan->flags) + if (locinput < PL_bostr + scan->flags) goto say_yes; PL_reginput = locinput - scan->flags; goto do_ifmatch; @@ -3375,7 +3412,7 @@ S_regmatch(pTHX_ regnode *prog) PL_reginput = s; } else { - if (locinput < PL_bostr + scan->flags) + if (locinput < PL_bostr + scan->flags) goto say_no; PL_reginput = locinput - scan->flags; goto do_ifmatch; @@ -3468,14 +3505,14 @@ do_no: { re_unwind_branch_t *uwb = &(uw->branch); I32 lastparen = uwb->lastparen; - + REGCP_UNWIND(uwb->lastcp); for (n = *PL_reglastparen; n > lastparen; n--) PL_regendp[n] = -1; *PL_reglastparen = n; scan = next = uwb->next; - if ( !scan || - OP(scan) != (uwb->type == RE_UNWIND_BRANCH + if ( !scan || + OP(scan) != (uwb->type == RE_UNWIND_BRANCH ? BRANCH : BRANCHJ) ) { /* Failure */ unwind = uwb->prev; #ifdef DEBUGGING @@ -3548,15 +3585,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) } break; case SANY: - if (do_utf8) { - loceol = PL_regeol; - while (hardcount < max && scan < loceol) { - scan += UTF8SKIP(scan); - hardcount++; - } - } else { - scan = loceol; - } + scan = loceol; break; case EXACT: /* length of string is 1 */ c = (U8)*STRING(p); @@ -3592,6 +3621,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case ALNUM: if (do_utf8) { loceol = PL_regeol; + LOAD_UTF8_CHARCLASS(alnum,"a"); while (hardcount < max && scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan)) { scan += UTF8SKIP(scan); @@ -3619,6 +3649,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case NALNUM: if (do_utf8) { loceol = PL_regeol; + LOAD_UTF8_CHARCLASS(alnum,"a"); while (hardcount < max && scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan)) { scan += UTF8SKIP(scan); @@ -3646,6 +3677,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case SPACE: if (do_utf8) { loceol = PL_regeol; + LOAD_UTF8_CHARCLASS(space," "); while (hardcount < max && scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { scan += UTF8SKIP(scan); @@ -3673,6 +3705,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case NSPACE: if (do_utf8) { loceol = PL_regeol; + LOAD_UTF8_CHARCLASS(space," "); while (hardcount < max && scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { scan += UTF8SKIP(scan); @@ -3700,6 +3733,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case DIGIT: if (do_utf8) { loceol = PL_regeol; + LOAD_UTF8_CHARCLASS(digit,"0"); while (hardcount < max && scan < loceol && swash_fetch(PL_utf8_digit,(U8*)scan)) { scan += UTF8SKIP(scan); @@ -3713,6 +3747,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) case NDIGIT: if (do_utf8) { loceol = PL_regeol; + LOAD_UTF8_CHARCLASS(digit,"0"); while (hardcount < max && scan < loceol && !swash_fetch(PL_utf8_digit,(U8*)scan)) { scan += UTF8SKIP(scan); @@ -3733,22 +3768,22 @@ S_regrepeat(pTHX_ regnode *p, I32 max) c = scan - PL_reginput; PL_reginput = scan; - DEBUG_r( + DEBUG_r( { SV *prop = sv_newmortal(); regprop(prop, p); - PerlIO_printf(Perl_debug_log, - "%*s %s can match %"IVdf" times out of %"IVdf"...\n", + PerlIO_printf(Perl_debug_log, + "%*s %s can match %"IVdf" times out of %"IVdf"...\n", REPORT_CODE_OFF+1, "", SvPVX(prop),(IV)c,(IV)max); }); - + return(c); } /* - regrepeat_hard - repeatedly match something, report total lenth and length - * + * * The repeater is supposed to have constant length. */ @@ -3794,7 +3829,7 @@ S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp) } if (!res) PL_reginput = scan; - + return count; } @@ -3815,10 +3850,10 @@ Perl_regclass_swash(pTHX_ register regnode* node, bool doinit, SV** initsvp) SV *rv = (SV*)PL_regdata->data[n]; AV *av = (AV*)SvRV((SV*)rv); SV **a; - + si = *av_fetch(av, 0, FALSE); a = av_fetch(av, 1, FALSE); - + if (a) sw = *a; else if (si && doinit) { @@ -3847,7 +3882,7 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) STRLEN len; if (do_utf8) - c = utf8_to_uv_simple(p, &len); + c = utf8_to_uvchr(p, &len); else c = *p; @@ -3866,13 +3901,13 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) match = TRUE; else if (flags & ANYOF_FOLD) { U8 tmpbuf[UTF8_MAXLEN+1]; - + if (flags & ANYOF_LOCALE) { PL_reg_flags |= RF_tainted; - uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p)); + uvchr_to_utf8(tmpbuf, toLOWER_LC_utf8(p)); } else - uv_to_utf8(tmpbuf, toLOWER_utf8(p)); + uvchr_to_utf8(tmpbuf, toLOWER_utf8(p)); if (swash_fetch(sw, tmpbuf)) match = TRUE; } @@ -3940,13 +3975,13 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, register bool do_utf8) STATIC U8 * S_reghop(pTHX_ U8 *s, I32 off) -{ +{ return S_reghop3(aTHX_ s, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)); } STATIC U8 * S_reghop3(pTHX_ U8 *s, I32 off, U8* lim) -{ +{ if (off >= 0) { while (off-- && s < lim) { /* XXX could check well-formedness here */ @@ -3970,7 +4005,7 @@ S_reghop3(pTHX_ U8 *s, I32 off, U8* lim) STATIC U8 * S_reghopmaybe(pTHX_ U8 *s, I32 off) -{ +{ return S_reghopmaybe3(aTHX_ s, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)); }