X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=93fadab5843a208e017d04be72f54737d5051bee;hb=e061ec81dcb60a85554e31a39475b89528de42f0;hp=363e2055896a3ed2484537845b013a5219e23c7a;hpb=24d786f4d2806834028ce32abc1769da2e945f9b;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index 363e205..93fadab 100644 --- a/regexec.c +++ b/regexec.c @@ -1007,15 +1007,16 @@ Perl_re_intuit_start(pTHX_ REGEXP * const rx, SV *sv, char *strpos, #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \ uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \ + UV uvc_unfolded = 0; \ switch (trie_type) { \ case trie_utf8_fold: \ if ( foldlen>0 ) { \ - uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ + uvc_unfolded = 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_unfolded = uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \ foldlen -= UNISKIP( uvc ); \ uscan = foldbuf + UNISKIP( uvc ); \ @@ -1054,6 +1055,9 @@ uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \ charid = (U16)SvIV(*svpp); \ } \ } \ + if (!charid && trie_type == trie_utf8_fold && !UTF) { \ + charid = trie->charmap[uvc_unfolded]; \ + } \ } STMT_END #define REXEC_FBC_EXACTISH_CHECK(CoNd) \ @@ -2250,7 +2254,7 @@ S_regtry(pTHX_ regmatch_info *reginfo, char **startpos) /* Make $_ available to executed code. */ if (reginfo->sv != DEFSV) { SAVE_DEFSV; - DEFSV = reginfo->sv; + DEFSV_set(reginfo->sv); } if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv) @@ -2590,7 +2594,7 @@ S_debug_start_match(pTHX_ const REGEXP *prog, const bool do_utf8, reginitcolors(); { RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0), - RX_PRECOMP(prog), RX_PRELEN(prog), 60); + RX_PRECOMP_const(prog), RX_PRELEN(prog), 60); RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1), start, end - start, 60); @@ -2837,6 +2841,11 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) state_num = OP(scan); reenter_switch: + + assert(PL_reglastparen == &rex->lastparen); + assert(PL_reglastcloseparen == &rex->lastcloseparen); + assert(PL_regoffs == rex->offs); + switch (state_num) { case BOL: if (locinput == PL_bostr) @@ -3003,7 +3012,7 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) if ( got_wordnum ) { if ( ! ST.accepted ) { ENTER; - /* SAVETMPS; */ /* XXX is this necessary? dmq */ + SAVETMPS; /* XXX is this necessary? dmq */ bufflen = TRIE_INITAL_ACCEPT_BUFFLEN; sv_accept_buff=newSV(bufflen * sizeof(reg_trie_accepted) - 1); @@ -3222,6 +3231,9 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) } /* NOTREACHED */ case TRIE_next: + /* we dont want to throw this away, see bug 57042*/ + if (oreplsv != GvSV(PL_replgv)) + sv_setsv(oreplsv, GvSV(PL_replgv)); FREETMPS; LEAVE; sayYES; @@ -3882,9 +3894,12 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) regcpblow(ST.cp); cur_eval = ST.prev_eval; cur_curlyx = ST.prev_curlyx; - + + /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */ PL_reglastparen = &rex->lastparen; PL_reglastcloseparen = &rex->lastcloseparen; + /* also update PL_regoffs */ + PL_regoffs = rex->offs; /* XXXX This is too dramatic a measure... */ PL_reg_maxiter = 0; @@ -3900,6 +3915,7 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) SETREX(rex_sv,ST.prev_rex); rex = (struct regexp *)SvANY(rex_sv); rexi = RXi_GET(rex); + /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */ PL_reglastparen = &rex->lastparen; PL_reglastcloseparen = &rex->lastcloseparen; @@ -4462,8 +4478,11 @@ NULL cur_eval->u.eval.close_paren == (U32)ST.me->flags) goto fake_end; - if ( ST.count < (ST.minmod ? ARG1(ST.me) : ARG2(ST.me)) ) - goto curlym_do_A; /* try to match another A */ + { + I32 max = (ST.minmod ? ARG1(ST.me) : ARG2(ST.me)); + if ( max == REG_INFTY || ST.count < max ) + goto curlym_do_A; /* try to match another A */ + } goto curlym_do_B; /* try to match B */ case CURLYM_A_fail: /* just failed to match an A */ @@ -4902,6 +4921,11 @@ NULL cur_curlyx = cur_eval->u.eval.prev_curlyx; ReREFCNT_inc(rex_sv); st->u.eval.cp = regcppush(0); /* Save *all* the positions. */ + + /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */ + PL_reglastparen = &rex->lastparen; + PL_reglastcloseparen = &rex->lastcloseparen; + REGCP_SET(st->u.eval.lastcp); PL_reginput = locinput; @@ -5768,7 +5792,14 @@ S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av); if (sw) { - if (swash_fetch(sw, p, do_utf8)) + U8 * utf8_p; + if (do_utf8) { + utf8_p = (U8 *) p; + } else { + STRLEN len = 1; + utf8_p = bytes_to_utf8(p, &len); + } + if (swash_fetch(sw, utf8_p, 1)) match = TRUE; else if (flags & ANYOF_FOLD) { if (!match && lenp && av) { @@ -5777,8 +5808,7 @@ S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const SV* const sv = *av_fetch(av, i, FALSE); STRLEN len; const char * const s = SvPV_const(sv, len); - - if (len <= plen && memEQ(s, (char*)p, len)) { + if (len <= plen && memEQ(s, (char*)utf8_p, len)) { *lenp = len; match = TRUE; break; @@ -5787,13 +5817,16 @@ S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const } if (!match) { U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; - STRLEN tmplen; - to_utf8_fold(p, tmpbuf, &tmplen); - if (swash_fetch(sw, tmpbuf, do_utf8)) + STRLEN tmplen; + to_utf8_fold(utf8_p, tmpbuf, &tmplen); + if (swash_fetch(sw, tmpbuf, 1)) match = TRUE; } } + + /* If we allocated a string above, free it */ + if (! do_utf8) Safefree(utf8_p); } } if (match && lenp && *lenp == 0)