X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=a3c7a7d6687f3165d8ddf07a03493024f6325e7e;hb=2a551100939323428e5b071089d183aafb961137;hp=5dbe166af7a0daaa23e95782351c9c0647fa8a89;hpb=d2212429110ea5c7ffe992aa2cb18bde7ce6e83d;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index 5dbe166..a3c7a7d 100644 --- a/regexec.c +++ b/regexec.c @@ -67,7 +67,8 @@ * **** 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, 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 +88,7 @@ #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 +240,7 @@ 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 +257,8 @@ 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; } @@ -400,6 +401,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, char *i_strpos = strpos; SV *dsv = PERL_DEBUG_PAD_ZERO(0); #endif + RX_MATCH_UTF8_set(prog,do_utf8); if (prog->reganch & ROPT_UTF8) { DEBUG_r(PerlIO_printf(Perl_debug_log, @@ -431,7 +433,8 @@ 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; @@ -540,6 +543,9 @@ 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; } @@ -594,7 +600,7 @@ 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 +720,7 @@ 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 @@ -847,10 +853,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, char *startpos = strbeg; t = s; - if (prog->reganch & ROPT_UTF8) { - PL_regdata = prog->data; - PL_bostr = startpos; - } + cache_re(prog); s = find_byclass(prog, prog->regstclass, s, endpos, startpos, 1); if (!s) { #ifdef DEBUGGING @@ -958,25 +961,40 @@ 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 < 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 += UTF8SKIP(s); + } + } + 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: @@ -999,8 +1017,10 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta 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_MAXLEN_UCLC, + 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); + c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN_UCLC, + 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); } else { c1 = *(U8*)m; @@ -1013,7 +1033,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta c1 = *(U8*)m; c2 = PL_fold_locale[c1]; do_exactf: - e = do_utf8 ? s + ln : strend - ln; + e = HOP3c(strend, -(I32)ln, s); if (norun && e < s) e = s; /* Due to minlen logic of intuit() */ @@ -1037,11 +1057,13 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta if (c1 == c2) { while (s <= e) { - c = utf8_to_uvchr((U8*)s, &len); + c = utf8n_to_uvchr((U8*)s, UTF8_MAXLEN, &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 +1075,7 @@ 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 +1084,9 @@ 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_MAXLEN, &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 +1102,7 @@ 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 +1114,7 @@ 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; } @@ -1129,11 +1153,9 @@ 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); @@ -1174,11 +1196,9 @@ 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); @@ -1594,6 +1614,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * SV *dsv0 = PERL_DEBUG_PAD_ZERO(0); SV *dsv1 = PERL_DEBUG_PAD_ZERO(1); #endif + RX_MATCH_UTF8_set(prog,do_utf8); PL_regcc = 0; @@ -1840,6 +1861,9 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char * : (s = fbm_instr((unsigned char*)HOP3(s, back_min, strend), (unsigned char*)strend, must, PL_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_r( did_match = 1 ); if (HOPc(s, -back_max) > last1) { last1 = HOPc(s, -back_min); @@ -1879,9 +1903,12 @@ 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)); + if (PL_regkind[op] != EXACT && op != CANY) + strend = HOPc(strend, -(minlen - 1)); + } DEBUG_r({ SV *prop = sv_newmortal(); char *s0; @@ -1923,6 +1950,9 @@ 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; @@ -1990,17 +2020,28 @@ 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; @@ -2090,6 +2131,9 @@ 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 @@ -2104,6 +2148,7 @@ 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); if (PL_reg_start_tmpl <= prog->nparens) { @@ -2132,7 +2177,7 @@ 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; } @@ -2266,17 +2311,17 @@ 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; @@ -2382,7 +2427,7 @@ 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; @@ -2394,7 +2439,9 @@ S_regmatch(pTHX_ regnode *prog) if (l >= PL_regeol) sayNO; if (NATIVE_TO_UNI(*(U8*)s) != - utf8_to_uvuni((U8*)l, &ulen)) + utf8n_to_uvuni((U8*)l, UTF8_MAXLEN, &ulen, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY)) sayNO; l += ulen; s ++; @@ -2406,7 +2453,9 @@ 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_MAXLEN, &ulen, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY)) sayNO; s += ulen; l ++; @@ -2439,7 +2488,7 @@ 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 +2614,9 @@ 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 +2770,7 @@ 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; @@ -2786,13 +2833,13 @@ 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; { @@ -2800,7 +2847,7 @@ 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 +2855,7 @@ 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 +2863,18 @@ 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); @@ -2836,9 +2888,11 @@ 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; @@ -2868,6 +2922,9 @@ 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 +2939,7 @@ 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 +2956,7 @@ 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; @@ -2922,13 +2981,13 @@ 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 +3089,7 @@ 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; @@ -3066,10 +3125,10 @@ 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. */ @@ -3114,7 +3173,7 @@ 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; } @@ -3299,7 +3358,7 @@ 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; @@ -3333,7 +3392,7 @@ 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] ) { @@ -3415,7 +3474,7 @@ 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] ) { @@ -3475,7 +3534,7 @@ 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 */ @@ -3524,7 +3583,7 @@ 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] ) { @@ -3551,11 +3610,17 @@ S_regmatch(pTHX_ regnode *prog) 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_MAXLEN, 0, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); + c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXLEN, 0, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); } else { - c2 = c1 = utf8_to_uvchr(s, NULL); + c2 = c1 = utf8n_to_uvchr(s, UTF8_MAXLEN, 0, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); } } } @@ -3574,6 +3639,7 @@ 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 +3659,6 @@ 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 +3676,28 @@ 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_MAXLEN, &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_MAXLEN, &len, + ckWARN(WARN_UTF8) ? + 0 : UTF8_ALLOW_ANY); + if (c == (UV)c1 || c == (UV)c2) break; - locinput += len; + locinput += len; + count++; } } } @@ -3644,6 +3719,7 @@ S_regmatch(pTHX_ regnode *prog) locinput += UTF8SKIP(locinput); else locinput++; + count = 1; } } else @@ -3651,20 +3727,23 @@ 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_MAXLEN, 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. */ @@ -3698,12 +3777,15 @@ 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_MAXLEN, 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 +3800,15 @@ 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_MAXLEN, 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); @@ -3959,7 +4044,9 @@ 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: @@ -4253,15 +4340,16 @@ 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 +4384,13 @@ 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_MAXLEN, &len, + ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY); plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c)); if (do_utf8 || (flags & ANYOF_UNICODE)) { @@ -4353,7 +4443,7 @@ 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; @@ -4481,6 +4571,9 @@ restore_pos(pTHX_ void *arg) 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; @@ -4495,7 +4588,7 @@ 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); @@ -4504,7 +4597,7 @@ 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); @@ -4519,7 +4612,7 @@ 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); @@ -4532,7 +4625,7 @@ 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);