X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=b5f8f4759ae15b0afcb1e23ce7166acbc6c77b5f;hb=e882dd6753c26e02b7dbf9e7c3e89bae261f600d;hp=d3e347e1df00a040cf10f910aa1def30bf38cd55;hpb=12d33761ed16a63d0b1b4afd6443ea446d40b38e;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index d3e347e..b5f8f47 100644 --- a/regexec.c +++ b/regexec.c @@ -116,6 +116,15 @@ #define HOPc(pos,off) ((char*)HOP(pos,off)) #define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off)) +#define HOPBACK(pos, off) ( \ + (UTF && DO_UTF8(PL_reg_sv)) \ + ? reghopmaybe((U8*)pos, -off) \ + : (pos - off >= PL_bostr) \ + ? (U8*)(pos - off) \ + : (U8*)NULL \ +) +#define HOPBACKc(pos, off) (char*)HOPBACK(pos, off) + #define reghop3_c(pos,off,lim) ((char*)reghop3((U8*)pos, off, (U8*)lim)) #define reghopmaybe3_c(pos,off,lim) ((char*)reghopmaybe3((U8*)pos, off, (U8*)lim)) #define HOP3(pos,off,lim) (DO_UTF8(PL_reg_sv) ? reghop3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off)) @@ -135,7 +144,10 @@ S_regcppush(pTHX_ I32 parenfloor) int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; int p; -#define REGCP_OTHER_ELEMS 5 + if (paren_elems_to_push < 0) + Perl_croak(aTHX_ "panic: paren_elems_to_push < 0"); + +#define REGCP_OTHER_ELEMS 6 SSCHECK(paren_elems_to_push + REGCP_OTHER_ELEMS); for (p = PL_regsize; p > parenfloor; p--) { /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ @@ -147,6 +159,7 @@ S_regcppush(pTHX_ I32 parenfloor) /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ SSPUSHINT(PL_regsize); SSPUSHINT(*PL_reglastparen); + SSPUSHINT(*PL_reglastcloseparen); SSPUSHPTR(PL_reginput); #define REGCP_FRAME_ELEMS 2 /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and @@ -180,6 +193,7 @@ S_regcppop(pTHX) assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ i = SSPOPINT; /* Parentheses elements to pop. */ input = (char *) SSPOPPTR; + *PL_reglastcloseparen = SSPOPINT; *PL_reglastparen = SSPOPINT; PL_regsize = SSPOPINT; @@ -348,17 +362,16 @@ char * Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, char *strend, U32 flags, re_scream_pos_data *data) { - register I32 start_shift; + register I32 start_shift = 0; /* Should be nonnegative! */ - register I32 end_shift; + register I32 end_shift = 0; register char *s; register SV *check; char *strbeg; char *t; I32 ml_anch; - char *tmp; register char *other_last = Nullch; /* other substr checked before this */ - char *check_at; /* check substr found at this pos */ + char *check_at = Nullch; /* check substr found at this pos */ #ifdef DEBUGGING char *i_strpos = strpos; #endif @@ -766,7 +779,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, s = find_byclass(prog, prog->regstclass, s, endpos, startpos, 1); if (!s) { #ifdef DEBUGGING - char *what; + char *what = 0; #endif if (endpos == strend) { DEBUG_r( PerlIO_printf(Perl_debug_log, @@ -827,13 +840,17 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, DEBUG_r( what = "floating" ); goto hop_and_restart; } - DEBUG_r( if (t != s) - PerlIO_printf(Perl_debug_log, + if (t != s) { + DEBUG_r(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, - "Does not contradict STCLASS...\n") ); + (long)(t - i_strpos), (long)(s - i_strpos)) + ); + } + else { + DEBUG_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", @@ -951,7 +968,8 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta else { U8 *r = reghop3((U8*)s, -1, (U8*)startpos); - tmp = (I32)utf8n_to_uvchr(r, s - (char*)r, 0, 0); + if (s > (char*)r) + tmp = (I32)utf8n_to_uvchr(r, s - (char*)r, 0, 0); } tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); @@ -994,7 +1012,8 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta else { U8 *r = reghop3((U8*)s, -1, (U8*)startpos); - tmp = (I32)utf8n_to_uvchr(r, s - (char*)r, 0, 0); + if (s > (char*)r) + tmp = (I32)utf8n_to_uvchr(r, s - (char*)r, 0, 0); } tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); @@ -1465,7 +1484,8 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * PL_reg_ganch = startpos; else if (sv && SvTYPE(sv) >= SVt_PVMG && SvMAGIC(sv) - && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0) { + && (mg = mg_find(sv, PERL_MAGIC_regex_global)) + && mg->mg_len >= 0) { PL_reg_ganch = strbeg + mg->mg_len; /* Defined pos() */ if (prog->reganch & ROPT_ANCH_GPOS) { if (s > PL_reg_ganch) @@ -1583,9 +1603,10 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * s++; } } - DEBUG_r(did_match || + DEBUG_r(if (!did_match) PerlIO_printf(Perl_debug_log, - "Did not find anchored character...\n")); + "Did not find anchored character...\n") + ); } /*SUPPRESS 560*/ else if (do_utf8 == (UTF!=0) && @@ -1649,14 +1670,16 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * } } } - DEBUG_r(did_match || - PerlIO_printf(Perl_debug_log, "Did not find %s substr `%s%.*s%s'%s...\n", + DEBUG_r(if (!did_match) + PerlIO_printf(Perl_debug_log, + "Did not find %s substr `%s%.*s%s'%s...\n", ((must == prog->anchored_substr) ? "anchored" : "floating"), PL_colors[0], (int)(SvCUR(must) - (SvTAIL(must)!=0)), SvPVX(must), - PL_colors[1], (SvTAIL(must) ? "$" : ""))); + PL_colors[1], (SvTAIL(must) ? "$" : "")) + ); goto phooey; } else if ((c = prog->regstclass)) { @@ -1816,10 +1839,11 @@ S_regtry(pTHX_ regexp *prog, char *startpos) } if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv) - && (mg = mg_find(PL_reg_sv, 'g')))) { + && (mg = mg_find(PL_reg_sv, PERL_MAGIC_regex_global)))) { /* prepare for quick setting of pos */ - sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0); - mg = mg_find(PL_reg_sv, 'g'); + sv_magic(PL_reg_sv, (SV*)0, + PERL_MAGIC_regex_global, Nullch, 0); + mg = mg_find(PL_reg_sv, PERL_MAGIC_regex_global); mg->mg_len = -1; } PL_reg_magic = mg; @@ -1828,7 +1852,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) } if (!PL_reg_curpm) Newz(22,PL_reg_curpm, 1, PMOP); - PL_reg_curpm->op_pmregexp = prog; + PM_SETRE(PL_reg_curpm, prog); PL_reg_oldcurpm = PL_curpm; PL_curpm = PL_reg_curpm; if (RX_MATCH_COPIED(prog)) { @@ -1849,6 +1873,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos) PL_regstartp = prog->startp; PL_regendp = prog->endp; PL_reglastparen = &prog->lastparen; + PL_reglastcloseparen = &prog->lastcloseparen; prog->lastparen = 0; PL_regsize = 0; DEBUG_r(PL_reg_starttry = startpos); @@ -1946,13 +1971,15 @@ S_regmatch(pTHX_ regnode *prog) register I32 nextchr; /* renamed nextchr - nextchar colides with function of same name */ register I32 n; /* no or next */ - register I32 ln; /* len or last */ - register char *s; /* operand or save */ + register I32 ln = 0; /* len or last */ + register char *s = Nullch; /* operand or save */ register char *locinput = PL_reginput; - register I32 c1, c2, paren; /* case fold search, parenth */ + register I32 c1 = 0, c2 = 0, paren; /* case fold search, parenth */ int minmod = 0, sw = 0, logical = 0; I32 unwind = 0; +#if 0 I32 firstcp = PL_savestack_ix; +#endif register bool do_utf8 = DO_UTF8(PL_reg_sv); #ifdef DEBUGGING @@ -2197,6 +2224,7 @@ S_regmatch(pTHX_ regnode *prog) if (!nextchr) sayNO; if (do_utf8) { + LOAD_UTF8_CHARCLASS(alnum,"a"); if (!(OP(scan) == ALNUM ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) : isALNUM_LC_utf8((U8*)locinput))) @@ -2493,7 +2521,7 @@ S_regmatch(pTHX_ regnode *prog) SV *sv = SvROK(ret) ? SvRV(ret) : ret; if(SvMAGICAL(sv)) - mg = mg_find(sv, 'r'); + mg = mg_find(sv, PERL_MAGIC_qr); } if (mg) { re = (regexp *)mg->mg_obj; @@ -2511,7 +2539,8 @@ S_regmatch(pTHX_ regnode *prog) re = CALLREGCOMP(aTHX_ t, t + len, &pm); if (!(SvFLAGS(ret) & (SVs_TEMP | SVs_PADTMP | SVf_READONLY))) - sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0); + sv_magic(ret,(SV*)ReREFCNT_inc(re), + PERL_MAGIC_qr,0,0); PL_regprecomp = oprecomp; PL_regsize = osize; PL_regnpar = onpar; @@ -2536,6 +2565,7 @@ S_regmatch(pTHX_ regnode *prog) cache_re(re); state.ss = PL_savestack_ix; *PL_reglastparen = 0; + *PL_reglastcloseparen = 0; PL_reg_call_cc = &state; PL_reginput = locinput; @@ -2593,6 +2623,7 @@ S_regmatch(pTHX_ regnode *prog) PL_regendp[n] = locinput - PL_bostr; if (n > *PL_reglastparen) *PL_reglastparen = n; + *PL_reglastcloseparen = n; break; case GROUPP: n = ARG(scan); /* which paren pair */ @@ -2919,7 +2950,6 @@ S_regmatch(pTHX_ regnode *prog) inner = NEXTOPER(scan); do_branch: { - CHECKPOINT lastcp; c1 = OP(scan); if (OP(next) != c1) /* No choice. */ next = inner; /* Avoid recursion. */ @@ -3270,7 +3300,7 @@ S_regmatch(pTHX_ regnode *prog) } REGCP_SET(lastcp); if (paren) { - UV c; + UV c = 0; while (n >= ln) { if (c1 != -1000) { if (do_utf8) @@ -3290,7 +3320,7 @@ S_regmatch(pTHX_ regnode *prog) } } else { - UV c; + UV c = 0; while (n >= ln) { if (c1 != -1000) { if (do_utf8) @@ -3368,20 +3398,10 @@ S_regmatch(pTHX_ regnode *prog) case UNLESSM: n = 0; if (scan->flags) { - if (UTF) { /* XXXX This is absolutely - broken, we read before - start of string. */ - s = HOPMAYBEc(locinput, -scan->flags); - if (!s) - goto say_yes; - PL_reginput = s; - } - else { - if (locinput < PL_bostr + scan->flags) - goto say_yes; - PL_reginput = locinput - scan->flags; - goto do_ifmatch; - } + s = HOPBACKc(locinput, scan->flags); + if (!s) + goto say_yes; + PL_reginput = s; } else PL_reginput = locinput; @@ -3389,20 +3409,10 @@ S_regmatch(pTHX_ regnode *prog) case IFMATCH: n = 1; if (scan->flags) { - if (UTF) { /* XXXX This is absolutely - broken, we read before - start of string. */ - s = HOPMAYBEc(locinput, -scan->flags); - if (!s || s < PL_bostr) - goto say_no; - PL_reginput = s; - } - else { - if (locinput < PL_bostr + scan->flags) - goto say_no; - PL_reginput = locinput - scan->flags; - goto do_ifmatch; - } + s = HOPBACKc(locinput, scan->flags); + if (!s) + goto say_no; + PL_reginput = s; } else PL_reginput = locinput; @@ -3778,7 +3788,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max) STATIC I32 S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp) { - register char *scan; + register char *scan = Nullch; register char *start; register char *loceol = PL_regeol; I32 l = 0;