X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=f69c36002c7aa89c698eae32a102ddf41d6d32bf;hb=7e107e90b7bd52c7fb110ac98da6bb7ab38e8959;hp=4380fd8916d14bd7335eb3ee00b5e55a39239665;hpb=6795a346dd967331b0b6120505d98d9c47937054;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index 4380fd8..f69c360 100644 --- a/regexec.c +++ b/regexec.c @@ -87,7 +87,7 @@ #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 */ @@ -239,7 +239,7 @@ S_regcppop(pTHX) ); } DEBUG_r( - if (*PL_reglastparen + 1 <= PL_regnpar) { + 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 +256,8 @@ 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; } @@ -431,7 +431,8 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, ); }); - if (prog->minlen > CHR_DIST((U8*)strend, (U8*)strpos)) { + /* CHR_DIST() would be more correct here but it makes things slow. */ + if (prog->minlen > strend - strpos) { DEBUG_r(PerlIO_printf(Perl_debug_log, "String too short... [re_intuit_start]\n")); goto fail; @@ -594,7 +595,7 @@ 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 */; @@ -714,7 +715,7 @@ 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 @@ -1045,7 +1046,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta 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 { @@ -1057,7 +1058,7 @@ 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; } @@ -1084,7 +1085,7 @@ 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 { @@ -1096,7 +1097,7 @@ 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; } @@ -2134,7 +2135,7 @@ 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; } @@ -2384,7 +2385,7 @@ S_regmatch(pTHX_ regnode *prog) case EXACT: s = STRING(scan); ln = STR_LEN(scan); - if (do_utf8 != (UTF!=0)) { + if (do_utf8 != UTF) { /* The target and the pattern have differing utf8ness. */ char *l = locinput; char *e = s + ln; @@ -2445,7 +2446,7 @@ 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, @@ -2727,7 +2728,7 @@ 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; @@ -2790,13 +2791,13 @@ S_regmatch(pTHX_ regnode *prog) dSP; OP_4tree *oop = PL_op; COP *ocurcop = PL_curcop; - SV **ocurpad = PL_curpad; + PAD *old_comppad; 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)) ); - PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 2]); + PAD_SAVE_LOCAL(old_comppad, (PAD*)PL_regdata->data[n + 2]); PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; { @@ -2804,7 +2805,7 @@ 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; @@ -2812,7 +2813,7 @@ 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. */ @@ -2820,6 +2821,7 @@ S_regmatch(pTHX_ regnode *prog) MAGIC *mg = Null(MAGIC*); re_cc_state state; CHECKPOINT cp, lastcp; + int toggleutf; if(SvROK(ret) || SvRMAGICAL(ret)) { SV *sv = SvROK(ret) ? SvRV(ret) : ret; @@ -2840,6 +2842,7 @@ 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))) @@ -2872,6 +2875,9 @@ 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; @@ -2886,6 +2892,7 @@ 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; @@ -2902,6 +2909,7 @@ 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; @@ -2926,13 +2934,13 @@ 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 */ @@ -3034,7 +3042,7 @@ 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; @@ -3070,10 +3078,10 @@ S_regmatch(pTHX_ regnode *prog) DEBUG_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. */ @@ -3118,7 +3126,7 @@ S_regmatch(pTHX_ regnode *prog) if (PL_reg_leftiter-- == 0) { I32 size = (PL_reg_maxiter + 7)/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; } @@ -3303,7 +3311,7 @@ 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; @@ -3337,7 +3345,7 @@ S_regmatch(pTHX_ regnode *prog) ln = PL_regstartp[n]; /* assume yes if we haven't seen CLOSEn */ if ( - *PL_reglastparen < n || + (I32)*PL_reglastparen < n || ln == -1 || ln == PL_regendp[n] ) { @@ -3419,7 +3427,7 @@ S_regmatch(pTHX_ regnode *prog) ln = PL_regstartp[n]; /* assume yes if we haven't seen CLOSEn */ if ( - *PL_reglastparen < n || + (I32)*PL_reglastparen < n || ln == -1 || ln == PL_regendp[n] ) { @@ -3479,7 +3487,7 @@ 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 */ @@ -3528,7 +3536,7 @@ S_regmatch(pTHX_ regnode *prog) ln = PL_regstartp[n]; /* assume yes if we haven't seen CLOSEn */ if ( - *PL_reglastparen < n || + (I32)*PL_reglastparen < n || ln == -1 || ln == PL_regendp[n] ) { @@ -3627,7 +3635,7 @@ S_regmatch(pTHX_ regnode *prog) utf8n_to_uvchr((U8*)locinput, UTF8_MAXLEN, &len, ckWARN(WARN_UTF8) ? - 0 : UTF8_ALLOW_ANY) != c1) { + 0 : UTF8_ALLOW_ANY) != (UV)c1) { locinput += len; count++; } @@ -3639,7 +3647,7 @@ S_regmatch(pTHX_ regnode *prog) UTF8_MAXLEN, &len, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); - if (c == c1 || c == c2) + if (c == (UV)c1 || c == (UV)c2) break; locinput += len; count++; @@ -3679,16 +3687,16 @@ S_regmatch(pTHX_ regnode *prog) 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. */ @@ -3730,7 +3738,7 @@ S_regmatch(pTHX_ regnode *prog) 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); @@ -3753,7 +3761,7 @@ S_regmatch(pTHX_ regnode *prog) 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); @@ -3989,7 +3997,9 @@ 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: @@ -4283,15 +4293,16 @@ 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) { @@ -4326,12 +4337,13 @@ 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 ? utf8n_to_uvchr(p, UTF8_MAXLEN, &len, - ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY) : *p; + if (do_utf8 && !UTF8_IS_INVARIANT(c)) + c = utf8n_to_uvchr(p, UTF8_MAXLEN, &len, + ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c)); if (do_utf8 || (flags & ANYOF_UNICODE)) { @@ -4384,7 +4396,7 @@ 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; @@ -4526,7 +4538,7 @@ 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); + SvSetSV(sv, prog->float_substr); sv_utf8_upgrade(sv); if (SvTAIL(prog->float_substr)) SvTAIL_on(sv); @@ -4535,7 +4547,7 @@ S_to_utf8_substr(pTHX_ register regexp *prog) } if (prog->anchored_substr && !prog->anchored_utf8) { prog->anchored_utf8 = sv = NEWSV(118, 0); - SvSetMagicSV(sv, prog->anchored_substr); + SvSetSV(sv, prog->anchored_substr); sv_utf8_upgrade(sv); if (SvTAIL(prog->anchored_substr)) SvTAIL_on(sv); @@ -4550,7 +4562,7 @@ 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); + SvSetSV(sv, prog->float_utf8); if (sv_utf8_downgrade(sv, TRUE)) { if (SvTAIL(prog->float_utf8)) SvTAIL_on(sv); @@ -4563,7 +4575,7 @@ S_to_byte_substr(pTHX_ register regexp *prog) } if (prog->anchored_utf8 && !prog->anchored_substr) { prog->anchored_substr = sv = NEWSV(118, 0); - SvSetMagicSV(sv, prog->anchored_utf8); + SvSetSV(sv, prog->anchored_utf8); if (sv_utf8_downgrade(sv, TRUE)) { if (SvTAIL(prog->anchored_utf8)) SvTAIL_on(sv);