X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=17ee6af0f722bc257686d2a46371ce39ab118466;hb=87a1ef3dbc71790d3dec7643306775e0d1e4768a;hp=5dbe166af7a0daaa23e95782351c9c0647fa8a89;hpb=d2212429110ea5c7ffe992aa2cb18bde7ce6e83d;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index 5dbe166..17ee6af 100644 --- a/regexec.c +++ b/regexec.c @@ -5,6 +5,17 @@ * "One Ring to rule them all, One Ring to find them..." */ +/* This file contains functions for executing a regular expression. See + * also regcomp.c which funnily enough, contains functions for compiling + * a regular expression. + * + * This file is also copied at build time to ext/re/re_exec.c, where + * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT. + * This causes the main functions to be compiled under new names and with + * debugging support added, which makes "use re 'debug'" work. + + */ + /* NOTE: this is derived from Henry Spencer's regexp code, and should not * confused with the original package (see point 3 below). Thanks, Henry! */ @@ -67,7 +78,8 @@ * **** Alterations to Henry's code are... **** - **** Copyright (c) 1991-2002, Larry Wall + **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, + **** 2000, 2001, 2002, 2003, 2004, 2005, by Larry Wall and others **** **** You may distribute under the terms of either the GNU General Public **** License or the Artistic License, as specified in the README file. @@ -87,7 +99,7 @@ #define RF_evaled 4 /* Did an EVAL with setting? */ #define RF_utf8 8 /* String contains multibyte chars? */ -#define UTF (PL_reg_flags & RF_utf8) +#define UTF ((PL_reg_flags & RF_utf8) != 0) #define RS_init 1 /* eval environment created */ #define RS_set 2 /* replsv value is set */ @@ -128,7 +140,7 @@ #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)) { ENTER; save_re_context(); (void)CAT2(is_utf8_, a)((U8*)b); LEAVE; } } STMT_END +#define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) { ENTER; save_re_context(); (void)CAT2(is_utf8_, a)((const U8*)b); LEAVE; } } STMT_END /* for use after a quantifier and before an EXACT-like node -- japhy */ #define JUMPABLE(rn) ( \ @@ -171,7 +183,7 @@ S_regcppush(pTHX_ I32 parenfloor) Perl_croak(aTHX_ "panic: paren_elems_to_push < 0"); #define REGCP_OTHER_ELEMS 6 - SSCHECK(paren_elems_to_push + REGCP_OTHER_ELEMS); + SSGROW(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]); @@ -194,11 +206,11 @@ S_regcppush(pTHX_ I32 parenfloor) } /* These are needed since we do not localize EVAL nodes: */ -# define REGCP_SET(cp) DEBUG_r(PerlIO_printf(Perl_debug_log, \ +# define REGCP_SET(cp) DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, \ " Setting an EVAL scope, savestack=%"IVdf"\n", \ (IV)PL_savestack_ix)); cp = PL_savestack_ix -# define REGCP_UNWIND(cp) DEBUG_r(cp != PL_savestack_ix ? \ +# define REGCP_UNWIND(cp) DEBUG_EXECUTE_r(cp != PL_savestack_ix ? \ PerlIO_printf(Perl_debug_log, \ " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \ (IV)(cp), (IV)PL_savestack_ix) : 0); regcpblow(cp) @@ -211,6 +223,8 @@ S_regcppop(pTHX) char *input; I32 tmps; + GET_RE_DEBUG_FLAGS_DECL; + /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ i = SSPOPINT; assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ @@ -229,7 +243,7 @@ S_regcppop(pTHX) tmps = SSPOPINT; if (paren <= *PL_reglastparen) PL_regendp[paren] = tmps; - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", (UV)paren, (IV)PL_regstartp[paren], @@ -238,8 +252,8 @@ S_regcppop(pTHX) (paren > *PL_reglastparen ? "(no)" : "")); ); } - DEBUG_r( - if (*PL_reglastparen + 1 <= PL_regnpar) { + DEBUG_EXECUTE_r( + if ((I32)(*PL_reglastparen + 1) <= PL_regnpar) { PerlIO_printf(Perl_debug_log, " restoring \\%"IVdf"..\\%"IVdf" to undef\n", (IV)(*PL_reglastparen + 1), (IV)PL_regnpar); @@ -256,8 +270,8 @@ S_regcppop(pTHX) * building DynaLoader will fail: * "Error: '*' not in typemap in DynaLoader.xs, line 164" * --jhi */ - for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) { - if (paren > PL_regsize) + for (paren = *PL_reglastparen + 1; (I32)paren <= PL_regnpar; paren++) { + if ((I32)paren > PL_regsize) PL_regstartp[paren] = -1; PL_regendp[paren] = -1; } @@ -396,18 +410,23 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, I32 ml_anch; register char *other_last = Nullch; /* other substr checked before this */ char *check_at = Nullch; /* check substr found at this pos */ + I32 multiline = prog->reganch & PMf_MULTILINE; #ifdef DEBUGGING char *i_strpos = strpos; SV *dsv = PERL_DEBUG_PAD_ZERO(0); #endif + GET_RE_DEBUG_FLAGS_DECL; + + RX_MATCH_UTF8_set(prog,do_utf8); + if (prog->reganch & ROPT_UTF8) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "UTF-8 regex...\n")); PL_reg_flags |= RF_utf8; } - DEBUG_r({ + DEBUG_EXECUTE_r({ char *s = PL_reg_match_utf8 ? sv_uni_display(dsv, sv, 60, UNI_DISPLAY_REGEX) : strpos; @@ -416,11 +435,11 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, if (!PL_colorset) reginitcolors(); if (PL_reg_match_utf8) - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_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], + PL_colors[4], PL_colors[5], PL_colors[0], prog->precomp, PL_colors[1], (strlen(prog->precomp) > 60 ? "..." : ""), @@ -431,8 +450,9 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, ); }); - if (prog->minlen > CHR_DIST((U8*)strend, (U8*)strpos)) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + /* CHR_DIST() would be more correct here but it makes things slow. */ + if (prog->minlen > strend - strpos) { + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too short... [re_intuit_start]\n")); goto fail; } @@ -448,14 +468,14 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, check = prog->check_substr; } if (check == &PL_sv_undef) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_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) - && !PL_multiline ) ); /* Check after \n? */ + && !multiline ) ); /* Check after \n? */ if (!ml_anch) { if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */ @@ -463,7 +483,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, /* 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")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); goto fail; } if (prog->check_offset_min == prog->check_offset_max && @@ -477,7 +497,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, 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")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); goto fail_finish; } /* Now should match s[0..slen-2] */ @@ -486,7 +506,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, || (slen > 1 && memNE(SvPVX(check), s, slen)))) { report_neq: - DEBUG_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); goto fail_finish; } } @@ -540,22 +560,25 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, start_shift + (s - strbeg), end_shift, pp, 0); else goto fail_finish; + /* we may be pointing at the wrong string */ + if (s && RX_MATCH_COPIED(prog)) + s = strbeg + (s - SvPVX(sv)); if (data) *data->scream_olds = s; } else if (prog->reganch & ROPT_CANY_SEEN) s = fbm_instr((U8*)(s + start_shift), (U8*)(strend - end_shift), - check, PL_multiline ? FBMrf_MULTILINE : 0); + check, multiline ? FBMrf_MULTILINE : 0); else s = fbm_instr(HOP3(s, start_shift, strend), HOP3(strend, -end_shift, strbeg), - check, PL_multiline ? FBMrf_MULTILINE : 0); + check, multiline ? FBMrf_MULTILINE : 0); /* Update the count-of-usability, remove useless subpatterns, unshift s. */ - DEBUG_r(PerlIO_printf(Perl_debug_log, "%s %s substr `%s%.*s%s'%s%s", + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s %s substr `%s%.*s%s'%s%s", (s ? "Found" : "Did not find"), (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) ? "anchored" : "floating"), PL_colors[0], @@ -570,7 +593,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, check_at = s; /* Finish the diagnostic message */ - DEBUG_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); /* Got a candidate. Check MBOL anchoring, and the *other* substr. Start with the other substr. @@ -594,7 +617,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, t = s - prog->check_offset_max; if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ - && (!(prog->reganch & ROPT_UTF8) + && (!do_utf8 || ((t = reghopmaybe3_c(s, -(prog->check_offset_max), strpos)) && t > strpos))) /* EMPTY */; @@ -611,7 +634,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, 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 */ + DEBUG_EXECUTE_r(must = prog->anchored_utf8); /* for debug */ } else s = fbm_instr( @@ -619,9 +642,9 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, HOP3(HOP3(last1, prog->anchored_offset, strend) + SvCUR(must), -(SvTAIL(must)!=0), strbeg), must, - PL_multiline ? FBMrf_MULTILINE : 0 + multiline ? FBMrf_MULTILINE : 0 ); - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s anchored substr `%s%.*s%s'%s", (s ? "Found" : "Contradicts"), PL_colors[0], @@ -631,11 +654,11 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, PL_colors[1], (SvTAIL(must) ? "$" : ""))); if (!s) { if (last1 >= last2) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, ", giving up...\n")); goto fail_finish; } - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, ", trying floating at offset %ld...\n", (long)(HOP3c(s1, 1, strend) - i_strpos))); other_last = HOP3c(last1, prog->anchored_offset+1, strend); @@ -643,7 +666,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, goto restart; } else { - DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", (long)(s - i_strpos))); t = HOP3c(s, -prog->anchored_offset, strbeg); other_last = HOP3c(s, 1, strend); @@ -674,14 +697,14 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, and end-of-str is not later than strend we are OK. */ if (must == &PL_sv_undef) { s = (char*)NULL; - DEBUG_r(must = prog->float_utf8); /* for debug message */ + DEBUG_EXECUTE_r(must = prog->float_utf8); /* for debug message */ } else s = fbm_instr((unsigned char*)s, (unsigned char*)last + SvCUR(must) - (SvTAIL(must)!=0), - must, PL_multiline ? FBMrf_MULTILINE : 0); - DEBUG_r(PerlIO_printf(Perl_debug_log, "%s floating substr `%s%.*s%s'%s", + must, multiline ? FBMrf_MULTILINE : 0); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s floating substr `%s%.*s%s'%s", (s ? "Found" : "Contradicts"), PL_colors[0], (int)(SvCUR(must) - (SvTAIL(must)!=0)), @@ -689,11 +712,11 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, PL_colors[1], (SvTAIL(must) ? "$" : ""))); if (!s) { if (last1 == last) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, ", giving up...\n")); goto fail_finish; } - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, ", trying anchored starting at offset %ld...\n", (long)(s1 + 1 - i_strpos))); other_last = last; @@ -701,7 +724,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, goto restart; } else { - DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", (long)(s - i_strpos))); other_last = s; /* Fix this later. --Hugo */ s = s1; @@ -714,7 +737,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, t = s - prog->check_offset_max; if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ - && (!(prog->reganch & ROPT_UTF8) + && (!do_utf8 || ((t = reghopmaybe3_c(s, -prog->check_offset_max, strpos)) && t > strpos))) { /* Fixed substring is found far enough so that the match @@ -740,34 +763,34 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, is float. Redo checking for "other"=="fixed". */ 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))); + DEBUG_EXECUTE_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; } /* We don't contradict the found floating substring. */ /* XXXX Why not check for STCLASS? */ s = t + 1; - DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", - PL_colors[0],PL_colors[1], (long)(s - i_strpos))); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", + PL_colors[0], PL_colors[1], (long)(s - i_strpos))); goto set_useful; } /* Position contradicts check-string */ /* XXXX probably better to look for check-string than for "\n", so one should lower the limit for t? */ - DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", - PL_colors[0],PL_colors[1], (long)(t + 1 - i_strpos))); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", + PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos))); other_last = strpos = s = t + 1; goto restart; } t++; } - DEBUG_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", - PL_colors[0],PL_colors[1])); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", + PL_colors[0], PL_colors[1])); goto fail_finish; } else { - DEBUG_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", - PL_colors[0],PL_colors[1])); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", + PL_colors[0], PL_colors[1])); } s = t; set_useful: @@ -789,9 +812,9 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, t = strpos; goto find_anchor; } - DEBUG_r( if (ml_anch) + DEBUG_EXECUTE_r( if (ml_anch) PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n", - (long)(strpos - i_strpos), PL_colors[0],PL_colors[1]); + (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]); ); success_at_start: if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */ @@ -806,7 +829,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, ))) { /* If flags & SOMETHING - do not do it many times on the same match */ - DEBUG_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); 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); @@ -834,8 +857,8 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, regstclass does not come from lookahead... */ /* If regstclass takes bytelength more than 1: If charlength==1, OK. This leaves EXACTF only, which is dealt with in find_byclass(). */ - U8* str = (U8*)STRING(prog->regstclass); - int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT + const U8* str = (U8*)STRING(prog->regstclass); + const int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT ? CHR_DIST(str+STR_LEN(prog->regstclass), str) : 1); char *endpos = (prog->anchored_substr || prog->anchored_utf8 || ml_anch) @@ -844,42 +867,38 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, ? HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend) : strend); - char *startpos = strbeg; t = s; - if (prog->reganch & ROPT_UTF8) { - PL_regdata = prog->data; - PL_bostr = startpos; - } - s = find_byclass(prog, prog->regstclass, s, endpos, startpos, 1); + cache_re(prog); + s = find_byclass(prog, prog->regstclass, s, endpos, 1); if (!s) { #ifdef DEBUGGING - char *what = 0; + const char *what = 0; #endif if (endpos == strend) { - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Could not match STCLASS...\n") ); goto fail; } - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "This position contradicts STCLASS...\n") ); if ((prog->reganch & ROPT_ANCH) && !ml_anch) goto fail; /* Contradict one of substrings */ if (prog->anchored_substr || prog->anchored_utf8) { if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { - DEBUG_r( what = "anchored" ); + DEBUG_EXECUTE_r( what = "anchored" ); hop_and_restart: s = HOP3c(t, 1, strend); if (s + start_shift + end_shift > strend) { /* XXXX Should be taken into account earlier? */ - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Could not match STCLASS...\n") ); goto fail; } if (!check) goto giveup; - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Looking for %s substr starting at offset %ld...\n", what, (long)(s + start_shift - i_strpos)) ); goto restart; @@ -891,7 +910,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, s = check_at; if (!check) goto giveup; - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Looking for anchored substr starting at offset %ld...\n", (long)(other_last - i_strpos)) ); goto do_other_anchored; @@ -902,9 +921,9 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, s = t = t + 1; if (!check) goto giveup; - DEBUG_r( PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Looking for /%s^%s/m starting at offset %ld...\n", - PL_colors[0],PL_colors[1], (long)(t - i_strpos)) ); + PL_colors[0], PL_colors[1], (long)(t - i_strpos)) ); goto try_at_offset; } if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ @@ -912,23 +931,23 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, /* Check is floating subtring. */ retry_floating_check: t = check_at - start_shift; - DEBUG_r( what = "floating" ); + DEBUG_EXECUTE_r( what = "floating" ); goto hop_and_restart; } if (t != s) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "By STCLASS: moving %ld --> %ld\n", (long)(t - i_strpos), (long)(s - i_strpos)) ); } else { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Does not contradict STCLASS...\n"); ); } } giveup: - DEBUG_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", PL_colors[4], (check ? "Guessed" : "Giving up"), PL_colors[5], (long)(s - i_strpos)) ); return s; @@ -937,18 +956,20 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, 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])); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", + PL_colors[4], PL_colors[5])); return Nullch; } /* We know what class REx starts with. Try to find this position... */ STATIC char * -S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *startpos, I32 norun) +S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, I32 norun) { I32 doevery = (prog->reganch & ROPT_SKIP) == 0; char *m; STRLEN ln; + STRLEN lnc; + register STRLEN uskip; unsigned int c1; unsigned int c2; char *e; @@ -958,25 +979,40 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta /* We know what class it must start with. */ switch (OP(c)) { case ANYOF: - while (s < strend) { - 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 - tmp = 1; - s += skip; + if (do_utf8) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { + if ((ANYOF_FLAGS(c) & ANYOF_UNICODE) || + !UTF8_IS_INVARIANT((U8)s[0]) ? + reginclass(c, (U8*)s, 0, do_utf8) : + REGINCLASS(c, (U8*)s)) { + if (tmp && (norun || regtry(prog, s))) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += uskip; + } + } + else { + while (s < strend) { + STRLEN skip = 1; + + if (REGINCLASS(c, (U8*)s) || + (ANYOF_FOLD_SHARP_S(c, s, strend) && + /* The assignment of 2 is intentional: + * for the folded 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 += skip; + } } break; case CANY: @@ -989,18 +1025,27 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } break; case EXACTF: - m = STRING(c); - ln = STR_LEN(c); + m = STRING(c); + ln = STR_LEN(c); /* length to match in octets/bytes */ + lnc = (I32) ln; /* length to match in characters */ if (UTF) { STRLEN ulen1, ulen2; - U8 tmpbuf1[UTF8_MAXLEN_UCLC+1]; - U8 tmpbuf2[UTF8_MAXLEN_UCLC+1]; + U8 *sm = (U8 *) m; + U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; + U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; to_utf8_lower((U8*)m, tmpbuf1, &ulen1); to_utf8_upper((U8*)m, tmpbuf2, &ulen2); - c1 = utf8_to_uvchr(tmpbuf1, 0); - c2 = utf8_to_uvchr(tmpbuf2, 0); + c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, + 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); + c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE, + 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); + lnc = 0; + while (sm < ((U8 *) m + ln)) { + lnc++; + sm += UTF8SKIP(sm); + } } else { c1 = *(U8*)m; @@ -1008,12 +1053,13 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } goto do_exactf; case EXACTFL: - m = STRING(c); - ln = STR_LEN(c); + m = STRING(c); + ln = STR_LEN(c); + lnc = (I32) ln; c1 = *(U8*)m; c2 = PL_fold_locale[c1]; do_exactf: - e = do_utf8 ? s + ln : strend - ln; + e = HOP3c(strend, -((I32)lnc), s); if (norun && e < s) e = s; /* Due to minlen logic of intuit() */ @@ -1031,17 +1077,21 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (do_utf8) { UV c, f; - U8 tmpbuf [UTF8_MAXLEN+1]; - U8 foldbuf[UTF8_MAXLEN_FOLD+1]; + U8 tmpbuf [UTF8_MAXBYTES+1]; + U8 foldbuf[UTF8_MAXBYTES_CASE+1]; STRLEN len, foldlen; if (c1 == c2) { + /* Upper and lower of 1st char are equal - + * probably not a "letter". */ while (s <= e) { - c = utf8_to_uvchr((U8*)s, &len); + c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); if ( c == c1 && (ln == len || ibcmp_utf8(s, (char **)0, 0, do_utf8, - m, (char **)0, ln, UTF)) + m, (char **)0, ln, (bool)UTF)) && (norun || regtry(prog, s)) ) goto got_it; else { @@ -1053,7 +1103,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta !ibcmp_utf8((char *) foldbuf, (char **)0, foldlen, do_utf8, m, - (char **)0, ln, UTF)) + (char **)0, ln, (bool)UTF)) && (norun || regtry(prog, s)) ) goto got_it; } @@ -1062,7 +1112,9 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else { while (s <= e) { - c = utf8_to_uvchr((U8*)s, &len); + c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); /* Handle some of the three Greek sigmas cases. * Note that not all the possible combinations @@ -1078,7 +1130,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if ( (c == c1 || c == c2) && (ln == len || ibcmp_utf8(s, (char **)0, 0, do_utf8, - m, (char **)0, ln, UTF)) + m, (char **)0, ln, (bool)UTF)) && (norun || regtry(prog, s)) ) goto got_it; else { @@ -1090,7 +1142,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta !ibcmp_utf8((char *) foldbuf, (char **)0, foldlen, do_utf8, m, - (char **)0, ln, UTF)) + (char **)0, ln, (bool)UTF)) && (norun || regtry(prog, s)) ) goto got_it; } @@ -1129,16 +1181,14 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (s == PL_bostr) tmp = '\n'; else { - U8 *r = reghop3((U8*)s, -1, (U8*)startpos); + U8 *r = reghop3((U8*)s, -1, (U8*)PL_bostr); - tmp = s > (char*)r ? - utf8n_to_uvchr(r, s - (char*)r, 0, 0) : - utf8n_to_uvchr(s, UTF8SKIP(s), 0, 0); + tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0); } tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); LOAD_UTF8_CHARCLASS(alnum,"a"); - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (tmp == !(OP(c) == BOUND ? swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : isALNUM_LC_utf8((U8*)s))) @@ -1147,7 +1197,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if ((norun || regtry(prog, s))) goto got_it; } - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1174,23 +1224,21 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (s == PL_bostr) tmp = '\n'; else { - U8 *r = reghop3((U8*)s, -1, (U8*)startpos); + U8 *r = reghop3((U8*)s, -1, (U8*)PL_bostr); - tmp = s > (char*)r ? - utf8n_to_uvchr(r, s - (char*)r, 0, 0) : - utf8n_to_uvchr(s, UTF8SKIP(s), 0, 0); + tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0); } tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); LOAD_UTF8_CHARCLASS(alnum,"a"); - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (tmp == !(OP(c) == NBOUND ? swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : isALNUM_LC_utf8((U8*)s))) tmp = !tmp; else if ((norun || regtry(prog, s))) goto got_it; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1212,7 +1260,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case ALNUM: if (do_utf8) { LOAD_UTF8_CHARCLASS(alnum,"a"); - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1221,7 +1269,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1241,7 +1289,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case ALNUML: PL_reg_flags |= RF_tainted; if (do_utf8) { - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (isALNUM_LC_utf8((U8*)s)) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1250,7 +1298,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1270,7 +1318,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case NALNUM: if (do_utf8) { LOAD_UTF8_CHARCLASS(alnum,"a"); - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (!swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1279,7 +1327,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1299,7 +1347,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case NALNUML: PL_reg_flags |= RF_tainted; if (do_utf8) { - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (!isALNUM_LC_utf8((U8*)s)) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1308,7 +1356,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1328,7 +1376,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case SPACE: if (do_utf8) { LOAD_UTF8_CHARCLASS(space," "); - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1337,7 +1385,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1357,7 +1405,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case SPACEL: PL_reg_flags |= RF_tainted; if (do_utf8) { - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1366,7 +1414,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1386,7 +1434,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case NSPACE: if (do_utf8) { LOAD_UTF8_CHARCLASS(space," "); - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8))) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1395,7 +1443,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1415,7 +1463,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case NSPACEL: PL_reg_flags |= RF_tainted; if (do_utf8) { - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1424,7 +1472,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1444,7 +1492,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case DIGIT: if (do_utf8) { LOAD_UTF8_CHARCLASS(digit,"0"); - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1453,7 +1501,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1473,7 +1521,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case DIGITL: PL_reg_flags |= RF_tainted; if (do_utf8) { - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (isDIGIT_LC_utf8((U8*)s)) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1482,7 +1530,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1502,7 +1550,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case NDIGIT: if (do_utf8) { LOAD_UTF8_CHARCLASS(digit,"0"); - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (!swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1511,7 +1559,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1531,7 +1579,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta case NDIGITL: PL_reg_flags |= RF_tainted; if (do_utf8) { - while (s < strend) { + while (s + (uskip = UTF8SKIP(s)) <= strend) { if (!isDIGIT_LC_utf8((U8*)s)) { if (tmp && (norun || regtry(prog, s))) goto got_it; @@ -1540,7 +1588,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta } else tmp = 1; - s += UTF8SKIP(s); + s += uskip; } } else { @@ -1590,11 +1638,16 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * char *scream_olds; SV* oreplsv = GvSV(PL_replgv); bool do_utf8 = DO_UTF8(sv); + I32 multiline = prog->reganch & PMf_MULTILINE; #ifdef DEBUGGING SV *dsv0 = PERL_DEBUG_PAD_ZERO(0); SV *dsv1 = PERL_DEBUG_PAD_ZERO(1); #endif + GET_RE_DEBUG_FLAGS_DECL; + + RX_MATCH_UTF8_set(prog,do_utf8); + PL_regcc = 0; cache_re(prog); @@ -1610,7 +1663,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * minlen = prog->minlen; if (strend - startpos < minlen) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too short [regexec_flags]...\n")); goto phooey; } @@ -1671,12 +1724,12 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * d.scream_pos = &scream_pos; s = re_intuit_start(prog, sv, s, strend, flags, &d); if (!s) { - DEBUG_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); goto phooey; /* not present */ } } - DEBUG_r({ + DEBUG_EXECUTE_r({ char *s0 = UTF ? pv_uni_display(dsv0, (U8*)prog->precomp, prog->prelen, 60, UNI_DISPLAY_REGEX) : @@ -1689,7 +1742,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * reginitcolors(); PerlIO_printf(Perl_debug_log, "%sMatching REx%s `%s%*.*s%s%s' against `%s%.*s%s%s'\n", - PL_colors[4],PL_colors[5],PL_colors[0], + PL_colors[4], PL_colors[5], PL_colors[0], len0, len0, s0, PL_colors[1], len0 > 60 ? "..." : "", @@ -1705,7 +1758,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { if (s == startpos && regtry(prog, startpos)) goto got_it; - else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT) + else if (multiline || (prog->reganch & ROPT_IMPLICIT) || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */ { char *end; @@ -1764,7 +1817,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * if (do_utf8) { while (s < strend) { if (*s == ch) { - DEBUG_r( did_match = 1 ); + DEBUG_EXECUTE_r( did_match = 1 ); if (regtry(prog, s)) goto got_it; s += UTF8SKIP(s); while (s < strend && *s == ch) @@ -1776,7 +1829,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * else { while (s < strend) { if (*s == ch) { - DEBUG_r( did_match = 1 ); + DEBUG_EXECUTE_r( did_match = 1 ); if (regtry(prog, s)) goto got_it; s++; while (s < strend && *s == ch) @@ -1785,7 +1838,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * s++; } } - DEBUG_r(if (!did_match) + DEBUG_EXECUTE_r(if (!did_match) PerlIO_printf(Perl_debug_log, "Did not find anchored character...\n") ); @@ -1839,8 +1892,11 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * end_shift, &scream_pos, 0)) : (s = fbm_instr((unsigned char*)HOP3(s, back_min, strend), (unsigned char*)strend, must, - PL_multiline ? FBMrf_MULTILINE : 0))) ) { - DEBUG_r( did_match = 1 ); + multiline ? FBMrf_MULTILINE : 0))) ) { + /* we may be pointing at the wrong string */ + if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog)) + s = strbeg + (s - SvPVX(sv)); + DEBUG_EXECUTE_r( did_match = 1 ); if (HOPc(s, -back_max) > last1) { last1 = HOPc(s, -back_min); s = HOPc(s, -back_max); @@ -1866,7 +1922,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * } } } - DEBUG_r(if (!did_match) + DEBUG_EXECUTE_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_utf8) @@ -1879,10 +1935,13 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * goto phooey; } else if ((c = prog->regstclass)) { - if (minlen && PL_regkind[(U8)OP(prog->regstclass)] != EXACT) + if (minlen) { + I32 op = (U8)OP(prog->regstclass); /* don't bother with what can't match */ - strend = HOPc(strend, -(minlen - 1)); - DEBUG_r({ + if (PL_regkind[op] != EXACT && op != CANY) + strend = HOPc(strend, -(minlen - 1)); + } + DEBUG_EXECUTE_r({ SV *prop = sv_newmortal(); char *s0; char *s1; @@ -1903,9 +1962,9 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * len0, len0, s0, len1, len1, s1); }); - if (find_byclass(prog, c, s, strend, startpos, 0)) + if (find_byclass(prog, c, s, strend, 0)) goto got_it; - DEBUG_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n")); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n")); } else { dontbother = 0; @@ -1923,15 +1982,18 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * end_shift, &scream_pos, 1); /* last one */ if (!last) last = scream_olds; /* Only one occurrence. */ + /* we may be pointing at the wrong string */ + else if (RX_MATCH_COPIED(prog)) + s = strbeg + (s - SvPVX(sv)); } else { STRLEN len; - char *little = SvPV(float_real, len); + const char * const little = SvPV(float_real, len); if (SvTAIL(float_real)) { if (memEQ(strend - len + 1, little, len - 1)) last = strend - len + 1; - else if (!PL_multiline) + else if (!multiline) last = memEQ(strend - len, little, len) ? strend - len : Nullch; else @@ -1945,9 +2007,9 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * } } if (last == NULL) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sCan't trim the tail, match fails (should not happen)%s\n", - PL_colors[4],PL_colors[5])); + PL_colors[4], PL_colors[5])); goto phooey; /* Should not happen! */ } dontbother = strend - last + prog->float_min_offset; @@ -1990,17 +2052,28 @@ got_it: /* make sure $`, $&, $', and $digit will work later */ if ( !(flags & REXEC_NOT_FIRST) ) { - if (RX_MATCH_COPIED(prog)) { - Safefree(prog->subbeg); - RX_MATCH_COPIED_off(prog); - } + RX_MATCH_COPY_FREE(prog); if (flags & REXEC_COPY_STR) { I32 i = PL_regeol - startpos + (stringarg - strbeg); - - s = savepvn(strbeg, i); - prog->subbeg = s; +#ifdef PERL_COPY_ON_WRITE + if ((SvIsCOW(sv) + || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) { + if (DEBUG_C_TEST) { + PerlIO_printf(Perl_debug_log, + "Copy on write: regexp capture, type %d\n", + (int) SvTYPE(sv)); + } + prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv); + prog->subbeg = SvPVX(prog->saved_copy); + assert (SvPOKp(prog->saved_copy)); + } else +#endif + { + RX_MATCH_COPIED_on(prog); + s = savepvn(strbeg, i); + prog->subbeg = s; + } prog->sublen = i; - RX_MATCH_COPIED_on(prog); } else { prog->subbeg = strbeg; @@ -2011,8 +2084,8 @@ got_it: return 1; phooey: - DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", - PL_colors[4],PL_colors[5])); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", + PL_colors[4], PL_colors[5])); if (PL_reg_eval_set) restore_pos(aTHX_ 0); return 0; @@ -2028,6 +2101,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) register I32 *sp; register I32 *ep; CHECKPOINT lastcp; + GET_RE_DEBUG_FLAGS_DECL; #ifdef DEBUGGING PL_regindent = 0; /* XXXX Not good when matches are reenterable... */ @@ -2036,7 +2110,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) MAGIC *mg; PL_reg_eval_set = RS_init; - DEBUG_r(DEBUG_s( + DEBUG_EXECUTE_r(DEBUG_s( PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n", (IV)(PL_stack_sp - PL_stack_base)); )); @@ -2051,8 +2125,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) if (PL_reg_sv) { /* Make $_ available to executed code. */ if (PL_reg_sv != DEFSV) { - /* SAVE_DEFSV does *not* suffice here for USE_5005THREADS */ - SAVESPTR(DEFSV); + SAVE_DEFSV; DEFSV = PL_reg_sv; } @@ -2069,7 +2142,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) SAVEDESTRUCTOR_X(restore_pos, 0); } if (!PL_reg_curpm) { - Newz(22,PL_reg_curpm, 1, PMOP); + Newz(22, PL_reg_curpm, 1, PMOP); #ifdef USE_ITHREADS { SV* repointer = newSViv(0); @@ -2090,6 +2163,9 @@ S_regtry(pTHX_ regexp *prog, char *startpos) $` inside (?{}) could fail... */ PL_reg_oldsaved = prog->subbeg; PL_reg_oldsavedlen = prog->sublen; +#ifdef PERL_COPY_ON_WRITE + PL_nrs = prog->saved_copy; +#endif RX_MATCH_COPIED_off(prog); } else @@ -2104,14 +2180,15 @@ S_regtry(pTHX_ regexp *prog, char *startpos) PL_reglastparen = &prog->lastparen; PL_reglastcloseparen = &prog->lastcloseparen; prog->lastparen = 0; + prog->lastcloseparen = 0; PL_regsize = 0; - DEBUG_r(PL_reg_starttry = startpos); + DEBUG_EXECUTE_r(PL_reg_starttry = startpos); if (PL_reg_start_tmpl <= prog->nparens) { PL_reg_start_tmpl = prog->nparens*3/2 + 3; if(PL_reg_start_tmp) Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); else - New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*); + New(22, PL_reg_start_tmp, PL_reg_start_tmpl, char*); } /* XXXX What this code is doing here?!!! There should be no need @@ -2132,7 +2209,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) sp = prog->startp; ep = prog->endp; if (prog->nparens) { - for (i = prog->nparens; i > *PL_reglastparen; i--) { + for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) { *++sp = -1; *++ep = -1; } @@ -2186,7 +2263,105 @@ typedef union re_unwind_t { #define sayNO_SILENT goto do_no #define saySAME(x) if (x) goto yes; else goto no -#define REPORT_CODE_OFF 24 +#define POSCACHE_SUCCESS 0 /* caching success rather than failure */ +#define POSCACHE_SEEN 1 /* we know what we're caching */ +#define POSCACHE_START 2 /* the real cache: this bit maps to pos 0 */ +#define CACHEsayYES STMT_START { \ + if (cache_offset | cache_bit) { \ + if (!(PL_reg_poscache[0] & (1<states[ state ].wordnum ) { \ + if ( !accepted ) { \ + ENTER; \ + SAVETMPS; \ + bufflen = TRIE_INITAL_ACCEPT_BUFFLEN ; \ + sv_accept_buff=NEWSV( 1234, \ + bufflen * sizeof(reg_trie_accepted) - 1 ); \ + SvCUR_set( sv_accept_buff, sizeof(reg_trie_accepted) ); \ + SvPOK_on( sv_accept_buff ); \ + sv_2mortal( sv_accept_buff ); \ + accept_buff = (reg_trie_accepted*)SvPV_nolen( sv_accept_buff );\ + } else { \ + if ( accepted >= bufflen ) { \ + bufflen *= 2; \ + accept_buff =(reg_trie_accepted*)SvGROW( sv_accept_buff, \ + bufflen * sizeof(reg_trie_accepted) ); \ + } \ + SvCUR_set( sv_accept_buff,SvCUR( sv_accept_buff ) \ + + sizeof( reg_trie_accepted ) ); \ + } \ + accept_buff[ accepted ].wordnum = trie->states[ state ].wordnum; \ + accept_buff[ accepted ].endpos = uc; \ + ++accepted; \ + } } STMT_END + +#define TRIE_HANDLE_CHAR STMT_START { \ + if ( uvc < 256 ) { \ + charid = trie->charmap[ uvc ]; \ + } else { \ + charid = 0; \ + if( trie->widecharmap ) { \ + SV** svpp = (SV**)NULL; \ + svpp = hv_fetch( trie->widecharmap, (char*)&uvc, \ + sizeof( UV ), 0 ); \ + if ( svpp ) { \ + charid = (U16)SvIV( *svpp ); \ + } \ + } \ + } \ + if ( charid && \ + ( base + charid > trie->uniquecharcount ) && \ + ( base + charid - 1 - trie->uniquecharcount < trie->lasttrans) && \ + trie->trans[ base + charid - 1 - trie->uniquecharcount ].check == state ) \ + { \ + state = trie->trans[ base + charid - 1 - trie->uniquecharcount ].next; \ + } else { \ + state = 0; \ + } \ + uc += len; \ + } STMT_END /* - regmatch - main matching routine @@ -2217,6 +2392,13 @@ S_regmatch(pTHX_ regnode *prog) register I32 c1 = 0, c2 = 0, paren; /* case fold search, parenth */ int minmod = 0, sw = 0, logical = 0; I32 unwind = 0; + + /* used by the trie code */ + SV *sv_accept_buff = 0; /* accepting states we have traversed */ + reg_trie_accepted *accept_buff = 0; /* "" */ + reg_trie_data *trie; /* what trie are we using right now */ + U32 accepted = 0; /* how many accepting states we have seen*/ + #if 0 I32 firstcp = PL_savestack_ix; #endif @@ -2225,18 +2407,23 @@ S_regmatch(pTHX_ regnode *prog) SV *dsv0 = PERL_DEBUG_PAD_ZERO(0); SV *dsv1 = PERL_DEBUG_PAD_ZERO(1); SV *dsv2 = PERL_DEBUG_PAD_ZERO(2); + + SV *re_debug_flags = NULL; #endif + GET_RE_DEBUG_FLAGS; + #ifdef DEBUGGING PL_regindent++; #endif + /* Note that nextchr is a byte even in UTF */ nextchr = UCHARAT(locinput); scan = prog; while (scan != NULL) { - DEBUG_r( { + DEBUG_EXECUTE_r( { SV *prop = sv_newmortal(); int docolor = *PL_colors[0]; int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ @@ -2266,17 +2453,17 @@ S_regmatch(pTHX_ regnode *prog) regprop(prop, scan); { char *s0 = - do_utf8 ? + do_utf8 && OP(scan) != CANY ? pv_uni_display(dsv0, (U8*)(locinput - pref_len), pref0_len, 60, UNI_DISPLAY_REGEX) : locinput - pref_len; int len0 = do_utf8 ? strlen(s0) : pref0_len; - char *s1 = do_utf8 ? + char *s1 = do_utf8 && OP(scan) != CANY ? pv_uni_display(dsv1, (U8*)(locinput - pref_len + pref0_len), 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 ? + char *s2 = do_utf8 && OP(scan) != CANY ? pv_uni_display(dsv2, (U8*)locinput, PL_regeol - locinput, 60, UNI_DISPLAY_REGEX) : locinput; @@ -2307,8 +2494,7 @@ S_regmatch(pTHX_ regnode *prog) switch (OP(scan)) { case BOL: - if (locinput == PL_bostr || (PL_multiline && - (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) + if (locinput == PL_bostr) { /* regtill = regbol; */ break; @@ -2330,12 +2516,8 @@ S_regmatch(pTHX_ regnode *prog) break; sayNO; case EOL: - if (PL_multiline) - goto meol; - else goto seol; case MEOL: - meol: if ((nextchr || locinput < PL_regeol) && nextchr != '\n') sayNO; break; @@ -2379,10 +2561,235 @@ S_regmatch(pTHX_ regnode *prog) else nextchr = UCHARAT(++locinput); break; + + + + /* + traverse the TRIE keeping track of all accepting states + we transition through until we get to a failing node. + + we use two slightly different pieces of code to handle + the traversal depending on whether its case sensitive or + not. we reuse the accept code however. (this should probably + be turned into a macro.) + + */ + case TRIEF: + case TRIEFL: + { + + U32 uniflags = ckWARN( WARN_UTF8 ) ? 0 : UTF8_ALLOW_ANY; + U8 *uc = ( U8* )locinput; + U32 state = 1; + U16 charid = 0; + U32 base = 0; + UV uvc = 0; + STRLEN len = 0; + STRLEN foldlen = 0; + U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; + U8 *uscan = (U8*)NULL; + STRLEN bufflen=0; + accepted = 0; + + trie = (reg_trie_data*)PL_regdata->data[ ARG( scan ) ]; + + while ( state && uc <= (U8*)PL_regeol ) { + + TRIE_CHECK_STATE_IS_ACCEPTING; + + base = trie->states[ state ].trans.base; + + DEBUG_TRIE_EXECUTE_r( + PerlIO_printf( Perl_debug_log, + "%*s %sState: %4"UVxf", Base: %4"UVxf", Accepted: %4"UVxf" ", + REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], + (UV)state, (UV)base, (UV)accepted ); + ); + + if ( base ) { + + if ( do_utf8 || UTF ) { + if ( foldlen>0 ) { + uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); + foldlen -= len; + uscan += len; + len=0; + } else { + uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); + uvc = to_uni_fold( uvc, foldbuf, &foldlen ); + foldlen -= UNISKIP( uvc ); + uscan = foldbuf + UNISKIP( uvc ); + } + } else { + uvc = (UV)*uc; + len = 1; + } + + TRIE_HANDLE_CHAR; + + } else { + state = 0; + } + DEBUG_TRIE_EXECUTE_r( + PerlIO_printf( Perl_debug_log, + "Charid:%3x CV:%4"UVxf" After State: %4"UVxf"%s\n", + charid, uvc, (UV)state, PL_colors[5] ); + ); + } + if ( !accepted ) { + sayNO; + } else { + goto TrieAccept; + } + } + /* unreached codepoint: we jump into the middle of the next case + from previous if blocks */ + case TRIE: + { + U32 uniflags = ckWARN( WARN_UTF8 ) ? 0 : UTF8_ALLOW_ANY; + U8 *uc = (U8*)locinput; + U32 state = 1; + U16 charid = 0; + U32 base = 0; + UV uvc = 0; + STRLEN len = 0; + STRLEN bufflen = 0; + accepted = 0; + + trie = (reg_trie_data*)PL_regdata->data[ ARG( scan ) ]; + + while ( state && uc <= (U8*)PL_regeol ) { + + TRIE_CHECK_STATE_IS_ACCEPTING; + + base = trie->states[ state ].trans.base; + + DEBUG_TRIE_EXECUTE_r( + PerlIO_printf( Perl_debug_log, + "%*s %sState: %4"UVxf", Base: %4"UVxf", Accepted: %4"UVxf" ", + REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], + (UV)state, (UV)base, (UV)accepted ); + ); + + if ( base ) { + + if ( do_utf8 || UTF ) { + uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); + } else { + uvc = (U32)*uc; + len = 1; + } + + TRIE_HANDLE_CHAR; + + } else { + state = 0; + } + DEBUG_TRIE_EXECUTE_r( + PerlIO_printf( Perl_debug_log, + "Charid:%3x CV:%4"UVxf" After State: %4"UVxf"%s\n", + charid, uvc, (UV)state, PL_colors[5] ); + ); + } + if ( !accepted ) { + sayNO; + } + } + + + /* + There was at least one accepting state that we + transitioned through. Presumably the number of accepting + states is going to be low, typically one or two. So we + simply scan through to find the one with lowest wordnum. + Once we find it, we swap the last state into its place + and decrement the size. We then try to match the rest of + the pattern at the point where the word ends, if we + succeed then we end the loop, otherwise the loop + eventually terminates once all of the accepting states + have been tried. + */ + TrieAccept: + { + int gotit = 0; + + if ( accepted == 1 ) { + DEBUG_EXECUTE_r({ + SV **tmp = av_fetch( trie->words, accept_buff[ 0 ].wordnum-1, 0 ); + PerlIO_printf( Perl_debug_log, + "%*s %sonly one match : #%d <%s>%s\n", + REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], + accept_buff[ 0 ].wordnum, + tmp ? SvPV_nolen( *tmp ) : "not compiled under -Dr", + PL_colors[5] ); + }); + PL_reginput = (char *)accept_buff[ 0 ].endpos; + /* in this case we free tmps/leave before we call regmatch + as we wont be using accept_buff again. */ + FREETMPS; + LEAVE; + gotit = regmatch( scan + NEXT_OFF( scan ) ); + } else { + DEBUG_EXECUTE_r( + PerlIO_printf( Perl_debug_log,"%*s %sgot %"IVdf" possible matches%s\n", + REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], (IV)accepted, + PL_colors[5] ); + ); + while ( !gotit && accepted-- ) { + U32 best = 0; + U32 cur; + for( cur = 1 ; cur <= accepted ; cur++ ) { + DEBUG_TRIE_EXECUTE_r( + PerlIO_printf( Perl_debug_log, + "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n", + REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], + (IV)best, accept_buff[ best ].wordnum, (IV)cur, + accept_buff[ cur ].wordnum, PL_colors[5] ); + ); + + if ( accept_buff[ cur ].wordnum < accept_buff[ best ].wordnum ) + best = cur; + } + DEBUG_EXECUTE_r({ + SV **tmp = av_fetch( trie->words, accept_buff[ best ].wordnum - 1, 0 ); + PerlIO_printf( Perl_debug_log, "%*s %strying alternation #%d <%s> at 0x%p%s\n", + REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], + accept_buff[best].wordnum, + tmp ? SvPV_nolen( *tmp ) : "not compiled under -Dr",scan, + PL_colors[5] ); + }); + if ( best= PL_regeol) sayNO; if (NATIVE_TO_UNI(*(U8*)s) != - utf8_to_uvuni((U8*)l, &ulen)) + utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY)) sayNO; l += ulen; s ++; @@ -2406,7 +2815,9 @@ S_regmatch(pTHX_ regnode *prog) if (l >= PL_regeol) sayNO; if (NATIVE_TO_UNI(*((U8*)l)) != - utf8_to_uvuni((U8*)s, &ulen)) + utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY)) sayNO; s += ulen; l ++; @@ -2439,7 +2850,7 @@ S_regmatch(pTHX_ regnode *prog) char *l = locinput; char *e = PL_regeol; - if (ibcmp_utf8(s, 0, ln, UTF, + if (ibcmp_utf8(s, 0, ln, (bool)UTF, l, &e, 0, do_utf8)) { /* One more case for the sharp s: * pack("U0U*", 0xDF) =~ /ss/i, @@ -2565,11 +2976,9 @@ S_regmatch(pTHX_ regnode *prog) if (locinput == PL_bostr) ln = '\n'; else { - U8 *r = reghop3((U8*)locinput, -1, (U8*)PL_reg_starttry); + U8 *r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); - ln = locinput > (char*)r ? - utf8n_to_uvchr(r, locinput - (char*)r, 0, 0) : - utf8n_to_uvchr(locinput, UTF8SKIP(locinput), 0, 0); + ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0); } if (OP(scan) == BOUND || OP(scan) == NBOUND) { ln = isALNUM_uni(ln); @@ -2723,7 +3132,7 @@ S_regmatch(pTHX_ regnode *prog) n = ARG(scan); /* which paren pair */ ln = PL_regstartp[n]; PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ - if (*PL_reglastparen < n || ln == -1) + if ((I32)*PL_reglastparen < n || ln == -1) sayNO; /* Do not match unless seen CLOSEn. */ if (ln == PL_regendp[n]) break; @@ -2739,8 +3148,8 @@ S_regmatch(pTHX_ regnode *prog) */ if (OP(scan) == REFF) { STRLEN ulen1, ulen2; - U8 tmpbuf1[UTF8_MAXLEN_UCLC+1]; - U8 tmpbuf2[UTF8_MAXLEN_UCLC+1]; + U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; + U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; while (s < e) { if (l >= PL_regeol) sayNO; @@ -2786,13 +3195,14 @@ S_regmatch(pTHX_ regnode *prog) dSP; OP_4tree *oop = PL_op; COP *ocurcop = PL_curcop; - SV **ocurpad = PL_curpad; + PAD *old_comppad; SV *ret; + struct regexp *oreg = PL_reg_re; 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)) ); - PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 2]); + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); + PAD_SAVE_LOCAL(old_comppad, (PAD*)PL_regdata->data[n + 2]); PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; { @@ -2800,7 +3210,7 @@ S_regmatch(pTHX_ regnode *prog) CALLRUNOPS(aTHX); /* Scalar context. */ SPAGAIN; if (SP == before) - ret = Nullsv; /* protect against empty (?{}) blocks. */ + ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */ else { ret = POPs; PUTBACK; @@ -2808,7 +3218,7 @@ S_regmatch(pTHX_ regnode *prog) } PL_op = oop; - PL_curpad = ocurpad; + PAD_RESTORE_LOCAL(old_comppad); PL_curcop = ocurcop; if (logical) { if (logical == 2) { /* Postponed subexpression. */ @@ -2816,13 +3226,18 @@ S_regmatch(pTHX_ regnode *prog) MAGIC *mg = Null(MAGIC*); re_cc_state state; CHECKPOINT cp, lastcp; - - if(SvROK(ret) || SvRMAGICAL(ret)) { - SV *sv = SvROK(ret) ? SvRV(ret) : ret; - - if(SvMAGICAL(sv)) - mg = mg_find(sv, PERL_MAGIC_qr); + int toggleutf; + register SV *sv; + + if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret))) + mg = mg_find(sv, PERL_MAGIC_qr); + else if (SvSMAGICAL(ret)) { + if (SvGMAGICAL(ret)) + sv_unmagic(ret, PERL_MAGIC_qr); + else + mg = mg_find(ret, PERL_MAGIC_qr); } + if (mg) { re = (regexp *)mg->mg_obj; (void)ReREFCNT_inc(re); @@ -2836,16 +3251,18 @@ S_regmatch(pTHX_ regnode *prog) I32 onpar = PL_regnpar; Zero(&pm, 1, PMOP); + if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8; re = CALLREGCOMP(aTHX_ t, t + len, &pm); if (!(SvFLAGS(ret) - & (SVs_TEMP | SVs_PADTMP | SVf_READONLY))) + & (SVs_TEMP | SVs_PADTMP | SVf_READONLY + | SVs_GMG))) sv_magic(ret,(SV*)ReREFCNT_inc(re), PERL_MAGIC_qr,0,0); PL_regprecomp = oprecomp; PL_regsize = osize; PL_regnpar = onpar; } - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "Entering embedded `%s%.60s%s%s'\n", PL_colors[0], @@ -2868,6 +3285,9 @@ S_regmatch(pTHX_ regnode *prog) *PL_reglastcloseparen = 0; PL_reg_call_cc = &state; PL_reginput = locinput; + toggleutf = ((PL_reg_flags & RF_utf8) != 0) ^ + ((re->reganch & ROPT_UTF8) != 0); + if (toggleutf) PL_reg_flags ^= RF_utf8; /* XXXX This is too dramatic a measure... */ PL_reg_maxiter = 0; @@ -2882,6 +3302,7 @@ S_regmatch(pTHX_ regnode *prog) PL_regcc = state.cc; PL_reg_re = state.re; cache_re(PL_reg_re); + if (toggleutf) PL_reg_flags ^= RF_utf8; /* XXXX This is too dramatic a measure... */ PL_reg_maxiter = 0; @@ -2898,6 +3319,7 @@ S_regmatch(pTHX_ regnode *prog) PL_regcc = state.cc; PL_reg_re = state.re; cache_re(PL_reg_re); + if (toggleutf) PL_reg_flags ^= RF_utf8; /* XXXX This is too dramatic a measure... */ PL_reg_maxiter = 0; @@ -2908,8 +3330,10 @@ S_regmatch(pTHX_ regnode *prog) sw = SvTRUE(ret); logical = 0; } - else + else { sv_setsv(save_scalar(PL_replgv), ret); + cache_re(oreg); + } break; } case OPEN: @@ -2922,13 +3346,13 @@ S_regmatch(pTHX_ regnode *prog) n = ARG(scan); /* which paren pair */ PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr; PL_regendp[n] = locinput - PL_bostr; - if (n > *PL_reglastparen) + if (n > (I32)*PL_reglastparen) *PL_reglastparen = n; *PL_reglastcloseparen = n; break; case GROUPP: n = ARG(scan); /* which paren pair */ - sw = (*PL_reglastparen >= n && PL_regendp[n] != -1); + sw = ((I32)*PL_reglastparen >= n && PL_regendp[n] != -1); break; case IFTHEN: PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ @@ -3030,7 +3454,7 @@ S_regmatch(pTHX_ regnode *prog) PL_regcc = &cc; /* XXXX Probably it is better to teach regpush to support parenfloor > PL_regsize... */ - if (parenfloor > *PL_reglastparen) + if (parenfloor > (I32)*PL_reglastparen) parenfloor = *PL_reglastparen; /* Pessimization... */ cc.parenfloor = parenfloor; cc.cur = -1; @@ -3060,16 +3484,17 @@ S_regmatch(pTHX_ regnode *prog) CHECKPOINT cp, lastcp; CURCUR* cc = PL_regcc; char *lastloc = cc->lastloc; /* Detection of 0-len. */ + I32 cache_offset = 0, cache_bit = 0; n = cc->cur + 1; /* how many we know we matched */ PL_reginput = locinput; - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, - "%*s %ld out of %ld..%ld cc=%lx\n", + "%*s %ld out of %ld..%ld cc=%"UVxf"\n", REPORT_CODE_OFF+PL_regindent*2, "", (long)n, (long)cc->min, - (long)cc->max, (long)cc) + (long)cc->max, PTR2UV(cc)) ); /* If degenerate scan matches "", assume scan done. */ @@ -3078,7 +3503,7 @@ S_regmatch(pTHX_ regnode *prog) PL_regcc = cc->oldcc; if (PL_regcc) ln = PL_regcc->cur; - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s empty match detected, try continuation...\n", REPORT_CODE_OFF+PL_regindent*2, "") @@ -3112,9 +3537,9 @@ S_regmatch(pTHX_ regnode *prog) PL_reg_leftiter = PL_reg_maxiter; } if (PL_reg_leftiter-- == 0) { - I32 size = (PL_reg_maxiter + 7)/8; + I32 size = (PL_reg_maxiter + 7 + POSCACHE_START)/8; if (PL_reg_poscache) { - if (PL_reg_poscache_size < size) { + if ((I32)PL_reg_poscache_size < size) { Renew(PL_reg_poscache, size, char); PL_reg_poscache_size = size; } @@ -3124,27 +3549,33 @@ S_regmatch(pTHX_ regnode *prog) PL_reg_poscache_size = size; Newz(29, PL_reg_poscache, size, char); } - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%sDetected a super-linear match, switching on caching%s...\n", PL_colors[4], PL_colors[5]) ); } if (PL_reg_leftiter < 0) { - I32 o = locinput - PL_bostr, b; - - o = (scan->flags & 0xf) - 1 + o * (scan->flags>>4); - b = o % 8; - o /= 8; - if (PL_reg_poscache[o] & (1<flags & 0xf) - 1 + POSCACHE_START + + cache_offset * (scan->flags>>4); + cache_bit = cache_offset % 8; + cache_offset /= 8; + if (PL_reg_poscache[cache_offset] & (1<next)) { regcpblow(cp); - sayYES; /* All done. */ + CACHEsayYES; /* All done. */ } REGCP_UNWIND(lastcp); regcppop(); @@ -3174,10 +3605,10 @@ S_regmatch(pTHX_ regnode *prog) "Complex regular subexpression recursion", REG_INFTY - 1); } - sayNO; + CACHEsayNO; } - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s trying longer...\n", REPORT_CODE_OFF+PL_regindent*2, "") @@ -3190,13 +3621,13 @@ S_regmatch(pTHX_ regnode *prog) REGCP_SET(lastcp); if (regmatch(cc->scan)) { regcpblow(cp); - sayYES; + CACHEsayYES; } REGCP_UNWIND(lastcp); regcppop(); cc->cur = n - 1; cc->lastloc = lastloc; - sayNO; + CACHEsayNO; } /* Prefer scan over next for maximal matching. */ @@ -3208,12 +3639,12 @@ S_regmatch(pTHX_ regnode *prog) REGCP_SET(lastcp); if (regmatch(cc->scan)) { regcpblow(cp); - sayYES; + CACHEsayYES; } REGCP_UNWIND(lastcp); regcppop(); /* Restore some previous $s? */ PL_reginput = locinput; - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s failed, try continuation...\n", REPORT_CODE_OFF+PL_regindent*2, "") @@ -3232,13 +3663,13 @@ S_regmatch(pTHX_ regnode *prog) if (PL_regcc) ln = PL_regcc->cur; if (regmatch(cc->next)) - sayYES; + CACHEsayYES; if (PL_regcc) PL_regcc->cur = ln; PL_regcc = cc; cc->cur = n - 1; cc->lastloc = lastloc; - sayNO; + CACHEsayNO; } /* NOT REACHED */ case BRANCHJ: @@ -3291,7 +3722,7 @@ S_regmatch(pTHX_ regnode *prog) CHECKPOINT lastcp; /* We suppose that the next guy does not need - backtracking: in particular, it is of constant length, + backtracking: in particular, it is of constant non-zero length, and has no parenths to influence future backrefs. */ ln = ARG1(scan); /* min to match */ n = ARG2(scan); /* max to match */ @@ -3299,7 +3730,7 @@ S_regmatch(pTHX_ regnode *prog) if (paren) { if (paren > PL_regsize) PL_regsize = paren; - if (paren > *PL_reglastparen) + if (paren > (I32)*PL_reglastparen) *PL_reglastparen = paren; } scan = NEXTOPER(scan) + NODE_STEP_REGNODE; @@ -3310,15 +3741,6 @@ S_regmatch(pTHX_ regnode *prog) minmod = 0; if (ln && regrepeat_hard(scan, ln, &l) < ln) sayNO; - /* if we matched something zero-length we don't need to - backtrack - capturing parens are already defined, so - the caveat in the maximal case doesn't apply - - XXXX if ln == 0, we can redo this check first time - through the following loop - */ - if (ln && l == 0) - n = ln; /* don't backtrack */ locinput = PL_reginput; if (HAS_TEXT(next) || JUMPABLE(next)) { regnode *text_node = next; @@ -3328,19 +3750,8 @@ S_regmatch(pTHX_ regnode *prog) if (! HAS_TEXT(text_node)) c1 = c2 = -1000; else { 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); + c1 = c2 = -1000; + goto assume_ok_MM; } else { c1 = (U8)*STRING(text_node); } if (OP(text_node) == EXACTF || OP(text_node) == REFF) @@ -3355,8 +3766,7 @@ S_regmatch(pTHX_ regnode *prog) 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 ? */ + while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */ /* If it could work, try it. */ if (c1 == -1000 || UCHARAT(PL_reginput) == c1 || @@ -3387,15 +3797,8 @@ S_regmatch(pTHX_ regnode *prog) } else { n = regrepeat_hard(scan, n, &l); - /* if we matched something zero-length we don't need to - backtrack, unless the minimum count is zero and we - are capturing the result - in that case the capture - being defined or not may affect later execution - */ - if (n != 0 && l == 0 && !(paren && ln == 0)) - ln = n; /* don't backtrack */ locinput = PL_reginput; - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s matched %"IVdf" times, len=%"IVdf"...\n", (int)(REPORT_CODE_OFF+PL_regindent*2), "", @@ -3410,19 +3813,8 @@ S_regmatch(pTHX_ regnode *prog) if (! HAS_TEXT(text_node)) c1 = c2 = -1000; else { 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); + c1 = c2 = -1000; + goto assume_ok_REG; } else { c1 = (U8)*STRING(text_node); } @@ -3445,7 +3837,7 @@ S_regmatch(pTHX_ regnode *prog) UCHARAT(PL_reginput) == c1 || UCHARAT(PL_reginput) == c2) { - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s trying tail with n=%"IVdf"...\n", (int)(REPORT_CODE_OFF+PL_regindent*2), "", (IV)n) @@ -3475,7 +3867,7 @@ S_regmatch(pTHX_ regnode *prog) paren = scan->flags; /* Which paren to set */ if (paren > PL_regsize) PL_regsize = paren; - if (paren > *PL_reglastparen) + if (paren > (I32)*PL_reglastparen) *PL_reglastparen = paren; ln = ARG1(scan); /* min to match */ n = ARG2(scan); /* max to match */ @@ -3519,19 +3911,8 @@ S_regmatch(pTHX_ regnode *prog) if (! HAS_TEXT(text_node)) c1 = c2 = -1000; else { 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; + c1 = c2 = -1000; + goto assume_ok_easy; } else { s = (U8*)STRING(text_node); } @@ -3545,17 +3926,23 @@ S_regmatch(pTHX_ regnode *prog) else { /* UTF */ if (OP(text_node) == EXACTF || OP(text_node) == REFF) { STRLEN ulen1, ulen2; - U8 tmpbuf1[UTF8_MAXLEN_UCLC+1]; - U8 tmpbuf2[UTF8_MAXLEN_UCLC+1]; + U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; + U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; to_utf8_lower((U8*)s, tmpbuf1, &ulen1); to_utf8_upper((U8*)s, tmpbuf2, &ulen2); - c1 = utf8_to_uvuni(tmpbuf1, 0); - c2 = utf8_to_uvuni(tmpbuf2, 0); + c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); + c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); } else { - c2 = c1 = utf8_to_uvchr(s, NULL); + c2 = c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); } } } @@ -3574,6 +3961,7 @@ S_regmatch(pTHX_ regnode *prog) if (c1 != -1000) { char *e; /* Should not check after this */ char *old = locinput; + int count = 0; if (n == REG_INFTY) { e = PL_regeol - 1; @@ -3593,7 +3981,6 @@ S_regmatch(pTHX_ regnode *prog) e = PL_regeol - 1; } while (1) { - int count; /* Find place 'next' could work */ if (!do_utf8) { if (c1 == c2) { @@ -3611,18 +3998,28 @@ S_regmatch(pTHX_ regnode *prog) else { STRLEN len; if (c1 == c2) { - for (count = 0; - locinput <= e && - utf8_to_uvchr((U8*)locinput, &len) != c1; - count++) + /* count initialised to + * utf8_distance(old, locinput) */ + while (locinput <= e && + utf8n_to_uvchr((U8*)locinput, + UTF8_MAXBYTES, &len, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY) != (UV)c1) { locinput += len; - + count++; + } } else { - for (count = 0; locinput <= e; count++) { - UV c = utf8_to_uvchr((U8*)locinput, &len); - if (c == c1 || c == c2) + /* count initialised to + * utf8_distance(old, locinput) */ + while (locinput <= e) { + UV c = utf8n_to_uvchr((U8*)locinput, + UTF8_MAXBYTES, &len, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); + if (c == (UV)c1 || c == (UV)c2) break; - locinput += len; + locinput += len; + count++; } } } @@ -3644,6 +4041,7 @@ S_regmatch(pTHX_ regnode *prog) locinput += UTF8SKIP(locinput); else locinput++; + count = 1; } } else @@ -3651,20 +4049,23 @@ S_regmatch(pTHX_ regnode *prog) UV c; if (c1 != -1000) { if (do_utf8) - c = utf8_to_uvchr((U8*)PL_reginput, NULL); + c = utf8n_to_uvchr((U8*)PL_reginput, + UTF8_MAXBYTES, 0, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); else c = UCHARAT(PL_reginput); /* If it could work, try it. */ - if (c == c1 || c == c2) + if (c == (UV)c1 || c == (UV)c2) { - TRYPAREN(paren, n, PL_reginput); + TRYPAREN(paren, ln, PL_reginput); REGCP_UNWIND(lastcp); } } /* If it could work, try it. */ else if (c1 == -1000) { - TRYPAREN(paren, n, PL_reginput); + TRYPAREN(paren, ln, PL_reginput); REGCP_UNWIND(lastcp); } /* Couldn't or didn't -- move forward. */ @@ -3682,7 +4083,7 @@ 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) != MEOL) || + (OP(next) != MEOL || OP(next) == SEOL || OP(next) == EOS)) { ln = n; /* why back off? */ @@ -3698,12 +4099,15 @@ S_regmatch(pTHX_ regnode *prog) while (n >= ln) { if (c1 != -1000) { if (do_utf8) - c = utf8_to_uvchr((U8*)PL_reginput, NULL); + c = utf8n_to_uvchr((U8*)PL_reginput, + UTF8_MAXBYTES, 0, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); else c = UCHARAT(PL_reginput); } /* If it could work, try it. */ - if (c1 == -1000 || c == c1 || c == c2) + if (c1 == -1000 || c == (UV)c1 || c == (UV)c2) { TRYPAREN(paren, n, PL_reginput); REGCP_UNWIND(lastcp); @@ -3718,12 +4122,15 @@ S_regmatch(pTHX_ regnode *prog) while (n >= ln) { if (c1 != -1000) { if (do_utf8) - c = utf8_to_uvchr((U8*)PL_reginput, NULL); + c = utf8n_to_uvchr((U8*)PL_reginput, + UTF8_MAXBYTES, 0, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); else c = UCHARAT(PL_reginput); } /* If it could work, try it. */ - if (c1 == -1000 || c == c1 || c == c2) + if (c1 == -1000 || c == (UV)c1 || c == (UV)c2) { TRYPAREN(paren, n, PL_reginput); REGCP_UNWIND(lastcp); @@ -3764,7 +4171,7 @@ S_regmatch(pTHX_ regnode *prog) PL_reg_re = re; cache_re(re); - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s continuation failed...\n", REPORT_CODE_OFF+PL_regindent*2, "") @@ -3772,7 +4179,7 @@ S_regmatch(pTHX_ regnode *prog) sayNO_SILENT; } if (locinput < PL_regtill) { - DEBUG_r(PerlIO_printf(Perl_debug_log, + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n", PL_colors[4], (long)(locinput - PL_reg_starttry), @@ -3857,15 +4264,15 @@ S_regmatch(pTHX_ regnode *prog) sayNO; yes_loud: - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s %scould match...%s\n", - REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],PL_colors[5]) + REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], PL_colors[5]) ); goto yes; yes_final: - DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n", - PL_colors[4],PL_colors[5])); + DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n", + PL_colors[4], PL_colors[5])); yes: #ifdef DEBUGGING PL_regindent--; @@ -3878,10 +4285,10 @@ yes: return 1; no: - DEBUG_r( + DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, "%*s %sfailed...%s\n", - REPORT_CODE_OFF+PL_regindent*2, "",PL_colors[4],PL_colors[5]) + REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], PL_colors[5]) ); goto do_no; no_final: @@ -3959,7 +4366,9 @@ S_regrepeat(pTHX_ regnode *p, I32 max) register bool do_utf8 = PL_reg_match_utf8; scan = PL_reginput; - if (max != REG_INFTY && max < loceol - scan) + if (max == REG_INFTY) + max = I32_MAX; + else if (max < loceol - scan) loceol = scan + max; switch (OP(p)) { case REG_ANY: @@ -4171,15 +4580,17 @@ S_regrepeat(pTHX_ regnode *p, I32 max) c = scan - PL_reginput; PL_reginput = scan; - DEBUG_r( - { + DEBUG_r({ + SV *re_debug_flags = NULL; SV *prop = sv_newmortal(); - + GET_RE_DEBUG_FLAGS; + DEBUG_EXECUTE_r({ regprop(prop, p); 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); } @@ -4187,7 +4598,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) /* - regrepeat_hard - repeatedly match something, report total lenth and length * - * The repeater is supposed to have constant length. + * The repeater is supposed to have constant non-zero length. */ STATIC I32 @@ -4253,15 +4664,16 @@ Perl_regclass_swash(pTHX_ register regnode* node, bool doinit, SV** listsvp, SV if (PL_regdata->what[n] == 's') { SV *rv = (SV*)PL_regdata->data[n]; AV *av = (AV*)SvRV((SV*)rv); + SV **ary = AvARRAY(av); SV **a, **b; /* 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); - + si = *ary; + a = SvTYPE(ary[1]) == SVt_RV ? &ary[1] : 0; + b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : 0; + if (a) sw = *a; else if (si && doinit) { @@ -4296,11 +4708,13 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, STRLEN* lenp, register b { char flags = ANYOF_FLAGS(n); bool match = FALSE; - UV c; + UV c = *p; STRLEN len = 0; STRLEN plen; - c = do_utf8 ? utf8_to_uvchr(p, &len) : *p; + if (do_utf8 && !UTF8_IS_INVARIANT(c)) + c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len, + ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c)); if (do_utf8 || (flags & ANYOF_UNICODE)) { @@ -4336,7 +4750,7 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, STRLEN* lenp, register b } } if (!match) { - U8 tmpbuf[UTF8_MAXLEN_FOLD+1]; + U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; STRLEN tmplen; to_utf8_fold(p, tmpbuf, &tmplen); @@ -4353,7 +4767,7 @@ S_reginclass(pTHX_ register regnode *n, register U8* p, STRLEN* lenp, register b if (ANYOF_BITMAP_TEST(n, c)) match = TRUE; else if (flags & ANYOF_FOLD) { - I32 f; + U8 f; if (flags & ANYOF_LOCALE) { PL_reg_flags |= RF_tainted; @@ -4477,10 +4891,14 @@ S_reghopmaybe3(pTHX_ U8* s, I32 off, U8* lim) static void restore_pos(pTHX_ void *arg) { + (void)arg; /* unused */ if (PL_reg_eval_set) { if (PL_reg_oldsaved) { PL_reg_re->subbeg = PL_reg_oldsaved; PL_reg_re->sublen = PL_reg_oldsavedlen; +#ifdef PERL_COPY_ON_WRITE + PL_reg_re->saved_copy = PL_nrs; +#endif RX_MATCH_COPIED_on(PL_reg_re); } PL_reg_magic->mg_len = PL_reg_oldpos; @@ -4494,8 +4912,7 @@ 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); + prog->float_utf8 = sv = newSVsv(prog->float_substr); sv_utf8_upgrade(sv); if (SvTAIL(prog->float_substr)) SvTAIL_on(sv); @@ -4503,8 +4920,7 @@ S_to_utf8_substr(pTHX_ register regexp *prog) prog->check_utf8 = sv; } if (prog->anchored_substr && !prog->anchored_utf8) { - prog->anchored_utf8 = sv = NEWSV(118, 0); - SvSetMagicSV(sv, prog->anchored_substr); + prog->anchored_utf8 = sv = newSVsv(prog->anchored_substr); sv_utf8_upgrade(sv); if (SvTAIL(prog->anchored_substr)) SvTAIL_on(sv); @@ -4518,8 +4934,7 @@ 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); + prog->float_substr = sv = newSVsv(prog->float_utf8); if (sv_utf8_downgrade(sv, TRUE)) { if (SvTAIL(prog->float_utf8)) SvTAIL_on(sv); @@ -4531,8 +4946,7 @@ S_to_byte_substr(pTHX_ register regexp *prog) prog->check_substr = sv; } if (prog->anchored_utf8 && !prog->anchored_substr) { - prog->anchored_substr = sv = NEWSV(118, 0); - SvSetMagicSV(sv, prog->anchored_utf8); + prog->anchored_substr = sv = newSVsv(prog->anchored_utf8); if (sv_utf8_downgrade(sv, TRUE)) { if (SvTAIL(prog->anchored_utf8)) SvTAIL_on(sv);