X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=f53567e9a6c92521a46914d32b2fd65f2bc16e63;hb=47cadb06ccb7dbe87fdda243544a1ecb06fb104f;hp=59aabdfcdbce0f24f803a23044b7fb4892641add;hpb=8e680dc37b34060770f3573bebbd4e34de1c230e;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index 59aabdf..f53567e 100644 --- a/regexec.c +++ b/regexec.c @@ -37,6 +37,7 @@ # define Perl_regprop my_regprop /* *These* symbols are masked to allow static link. */ # define Perl_pregexec my_pregexec +# define Perl_reginitcolors my_reginitcolors #endif /*SUPPRESS 112*/ @@ -73,6 +74,7 @@ */ #include "EXTERN.h" #include "perl.h" + #include "regcomp.h" #define RF_tainted 1 /* tainted information used? */ @@ -105,6 +107,9 @@ static bool reginclass _((char *p, I32 c)); static bool reginclassutf8 _((regnode *f, U8* p)); static CHECKPOINT regcppush _((I32 parenfloor)); static char * regcppop _((void)); +static char * regcp_set_to _((I32 ss)); +static void cache_re _((regexp *prog)); +static void restore_pos _((void *arg)); #endif #define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c)) @@ -113,10 +118,16 @@ static char * regcppop _((void)); #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) -static char * reghop _((unsigned char *pos, I32 off)); -static char * reghopmaybe _((unsigned char *pos, I32 off)); -#define HOP(pos,off) (UTF ? reghop(pos, off) : (pos + off)) -#define HOPMAYBE(pos,off) (UTF ? reghopmaybe(pos, off) : (pos + off)) +#ifndef PERL_OBJECT +static U8 * reghop _((U8 *pos, I32 off)); +static U8 * reghopmaybe _((U8 *pos, I32 off)); +#endif +#define reghop_c(pos,off) ((char*)reghop((U8*)pos, off)) +#define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off)) +#define HOP(pos,off) (UTF ? reghop((U8*)pos, off) : (U8*)(pos + off)) +#define HOPMAYBE(pos,off) (UTF ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off)) +#define HOPc(pos,off) ((char*)HOP(pos,off)) +#define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off)) STATIC CHECKPOINT regcppush(I32 parenfloor) @@ -195,6 +206,27 @@ regcppop(void) return input; } +STATIC char * +regcp_set_to(I32 ss) +{ + dTHR; + I32 tmp = PL_savestack_ix; + + PL_savestack_ix = ss; + regcppop(); + PL_savestack_ix = tmp; + return Nullch; +} + +typedef struct re_cc_state +{ + I32 ss; + regnode *node; + struct re_cc_state *prev; + CURCUR *cc; + regexp *re; +} re_cc_state; + #define regcpblow(cp) LEAVE_SCOPE(cp) /* @@ -216,13 +248,38 @@ pregexec(register regexp *prog, char *stringarg, register char *strend, regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, nosave ? 0 : REXEC_COPY_STR); } - + +STATIC void +cache_re(regexp *prog) +{ + dTHR; + PL_regprecomp = prog->precomp; /* Needed for FAIL. */ +#ifdef DEBUGGING + PL_regprogram = prog->program; +#endif + PL_regnpar = prog->nparens; + PL_regdata = prog->data; + PL_reg_re = prog; +} + +STATIC void +restore_pos(void *arg) +{ + dTHR; + if (PL_reg_eval_set) { + PL_reg_magic->mg_len = PL_reg_oldpos; + PL_reg_eval_set = 0; + PL_curpm = PL_reg_oldcurpm; + } +} + + /* - regexec_flags - match a regexp against a string */ I32 regexec_flags(register regexp *prog, char *stringarg, register char *strend, - char *strbeg, I32 minend, SV *screamer, void *data, U32 flags) + char *strbeg, I32 minend, SV *sv, void *data, U32 flags) /* strend: pointer to null at end of string */ /* strbeg: real beginning of string */ /* minend: end of match must be >=minend after stringarg. */ @@ -248,10 +305,9 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cc.oldcc = 0; PL_regcc = &cc; - PL_regprecomp = prog->precomp; /* Needed for error messages. */ + cache_re(prog); #ifdef DEBUGGING PL_regnarrate = PL_debug & 512; - PL_regprogram = prog->program; #endif /* Be paranoid... */ @@ -272,11 +328,10 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, } /* Check validity of program. */ - if (UCHARAT(prog->program) != MAGIC) { - FAIL("corrupted regexp program"); + if (UCHARAT(prog->program) != REG_MAGIC) { + croak("corrupted regexp program"); } - PL_regnpar = prog->nparens; PL_reg_flags = 0; PL_reg_eval_set = 0; @@ -286,6 +341,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, /* Mark beginning of line for ^ and lookbehind. */ PL_regbol = startpos; PL_bostr = strbeg; + PL_reg_sv = sv; /* Mark end of line for $ (and such) */ PL_regeol = strend; @@ -293,6 +349,9 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, /* see how far we have to get to not match where we matched before */ PL_regtill = startpos+minend; + /* We start without call_cc context. */ + PL_reg_call_cc = 0; + /* If there is a "must appear" string, look for it. */ s = startpos; if (!(flags & REXEC_CHECKED) @@ -305,9 +364,9 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, start_shift = prog->check_offset_min; /* okay to underestimate on CC */ /* Should be nonnegative! */ end_shift = minlen - start_shift - CHR_SVLEN(prog->check_substr); - if (screamer) { + if (flags & REXEC_SCREAM) { if (PL_screamfirst[BmRARE(prog->check_substr)] >= 0) - s = screaminstr(screamer, prog->check_substr, + s = screaminstr(sv, prog->check_substr, start_shift + (stringarg - strbeg), end_shift, &scream_pos, 0); else @@ -324,7 +383,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, } else if (s - stringarg > prog->check_offset_max && (UTF - ? ((t = reghopmaybe(s, -(prog->check_offset_max))) && t >= stringarg) + ? ((t = reghopmaybe_c(s, -(prog->check_offset_max))) && t >= stringarg) : (t = s - prog->check_offset_max) != 0 ) ) @@ -344,41 +403,55 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, s = startpos; } - DEBUG_r( - PerlIO_printf(Perl_debug_log, - "Matching `%.60s%s' against `%.*s%s'\n", - prog->precomp, + DEBUG_r(if (!PL_colorset) reginitcolors()); + DEBUG_r(PerlIO_printf(Perl_debug_log, + "%sMatching%s `%s%.60s%s%s' against `%s%.*s%s%s'\n", + PL_colors[4],PL_colors[5],PL_colors[0], + prog->precomp, + PL_colors[1], (strlen(prog->precomp) > 60 ? "..." : ""), + PL_colors[0], (strend - startpos > 60 ? 60 : strend - startpos), - startpos, + startpos, PL_colors[1], (strend - startpos > 60 ? "..." : "")) ); - PL_regdata = prog->data; + if (prog->reganch & ROPT_GPOS_SEEN) { + MAGIC *mg; + + if (!(flags & REXEC_IGNOREPOS) && sv && SvTYPE(sv) >= SVt_PVMG + && SvMAGIC(sv) && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0) + PL_reg_ganch = strbeg + mg->mg_len; + else + PL_reg_ganch = startpos; + } /* Simplest case: anchored match need be tried only once. */ /* [unless only anchor is BOL and multiline is set] */ - if (prog->reganch & ROPT_ANCH) { + if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { if (regtry(prog, startpos)) goto got_it; - else if (!(prog->reganch & ROPT_ANCH_GPOS) && - (PL_multiline || (prog->reganch & ROPT_IMPLICIT) - || (prog->reganch & ROPT_ANCH_MBOL))) + else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT) + || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */ { if (minlen) dontbother = minlen - 1; - strend = HOP(strend, -dontbother); + strend = HOPc(strend, -dontbother); /* for multiline we only have to try after newlines */ if (s > startpos) s--; while (s < strend) { - if (*s++ == '\n') { /* don't need utf8skip here */ + if (*s++ == '\n') { /* don't need PL_utf8skip here */ if (s < strend && regtry(prog, s)) goto got_it; } } } goto phooey; + } else if (prog->reganch & ROPT_ANCH_GPOS) { + if (regtry(prog, PL_reg_ganch)) + goto got_it; + goto phooey; } /* Messy cases: unanchored match. */ @@ -420,11 +493,11 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, I32 back_min = prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset; I32 delta = back_max - back_min; - char *last = HOP(strend, 0-(CHR_SVLEN(must) + back_min)); /* Cannot start after this */ + char *last = HOPc(strend, 0-(CHR_SVLEN(must) + back_min)); /* Cannot start after this */ char *last1; /* Last position checked before */ if (s > PL_bostr) - last1 = HOP(s, -1); + last1 = HOPc(s, -1); else last1 = s - 1; /* bogus */ @@ -432,21 +505,21 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, check_substr==must. */ scream_pos = -1; dontbother = end_shift; - strend = HOP(strend, -dontbother); + strend = HOPc(strend, -dontbother); while ( (s <= last) && - (screamer - ? (s = screaminstr(screamer, must, HOP(s, back_min) - strbeg, + ((flags & REXEC_SCREAM) + ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg, end_shift, &scream_pos, 0)) : (s = fbm_instr((unsigned char*)HOP(s, back_min), (unsigned char*)strend, must, 0))) ) { - if (HOP(s, -back_max) > last1) { - last1 = HOP(s, -back_min); - s = HOP(s, -back_max); + if (HOPc(s, -back_max) > last1) { + last1 = HOPc(s, -back_min); + s = HOPc(s, -back_max); } else { - char *t = (last1 >= PL_bostr) ? HOP(last1, 1) : last1 + 1; + char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; - last1 = HOP(s, -back_min); + last1 = HOPc(s, -back_min); s = t; } if (UTF) { @@ -472,7 +545,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, if (minlen) dontbother = minlen - 1; - strend = HOP(strend, -dontbother); /* don't bother with what can't match */ + strend = HOPc(strend, -dontbother); /* don't bother with what can't match */ tmp = 1; /* We know what class it must start with. */ switch (OP(c)) { @@ -531,12 +604,15 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, case BOUNDUTF8: if (minlen) { dontbother++; - strend = reghop(strend, -1); + strend = reghop_c(strend, -1); } - tmp = (I32)(s != startpos) ? utf8_to_uv(reghop(s, -1), 0) : PL_regprev; + tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev; tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0); while (s < strend) { - if (tmp == !(OP(c) == BOUND ? swash_fetch(PL_utf8_alnum, s) : isALNUM_LC_utf8(s))) { + if (tmp == !(OP(c) == BOUND ? + swash_fetch(PL_utf8_alnum, (U8*)s) : + isALNUM_LC_utf8((U8*)s))) + { tmp = !tmp; if (regtry(prog, s)) goto got_it; @@ -572,12 +648,14 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, case NBOUNDUTF8: if (minlen) { dontbother++; - strend = reghop(strend, -1); + strend = reghop_c(strend, -1); } - tmp = (I32)(s != startpos) ? utf8_to_uv(reghop(s, -1), 0) : PL_regprev; + tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev; tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0); while (s < strend) { - if (tmp == !(OP(c) == NBOUND ? swash_fetch(PL_utf8_alnum, s) : isALNUM_LC_utf8(s))) + if (tmp == !(OP(c) == NBOUND ? + swash_fetch(PL_utf8_alnum, (U8*)s) : + isALNUM_LC_utf8((U8*)s))) tmp = !tmp; else if (regtry(prog, s)) goto got_it; @@ -601,7 +679,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, break; case ALNUMUTF8: while (s < strend) { - if (swash_fetch(PL_utf8_alnum, s)) { + if (swash_fetch(PL_utf8_alnum, (U8*)s)) { if (tmp && regtry(prog, s)) goto got_it; else @@ -629,7 +707,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, case ALNUMLUTF8: PL_reg_flags |= RF_tainted; while (s < strend) { - if (isALNUM_LC_utf8(s)) { + if (isALNUM_LC_utf8((U8*)s)) { if (tmp && regtry(prog, s)) goto got_it; else @@ -655,7 +733,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, break; case NALNUMUTF8: while (s < strend) { - if (!swash_fetch(PL_utf8_alnum, s)) { + if (!swash_fetch(PL_utf8_alnum, (U8*)s)) { if (tmp && regtry(prog, s)) goto got_it; else @@ -683,7 +761,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, case NALNUMLUTF8: PL_reg_flags |= RF_tainted; while (s < strend) { - if (!isALNUM_LC_utf8(s)) { + if (!isALNUM_LC_utf8((U8*)s)) { if (tmp && regtry(prog, s)) goto got_it; else @@ -709,7 +787,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, break; case SPACEUTF8: while (s < strend) { - if (*s == ' ' || swash_fetch(PL_utf8_space,s)) { + if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) { if (tmp && regtry(prog, s)) goto got_it; else @@ -737,7 +815,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, case SPACELUTF8: PL_reg_flags |= RF_tainted; while (s < strend) { - if (*s == ' ' || isSPACE_LC_utf8(s)) { + if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) { if (tmp && regtry(prog, s)) goto got_it; else @@ -763,7 +841,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, break; case NSPACEUTF8: while (s < strend) { - if (!(*s == ' ' || swash_fetch(PL_utf8_space,s))) { + if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) { if (tmp && regtry(prog, s)) goto got_it; else @@ -791,7 +869,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, case NSPACELUTF8: PL_reg_flags |= RF_tainted; while (s < strend) { - if (!(*s == ' ' || isSPACE_LC_utf8(s))) { + if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) { if (tmp && regtry(prog, s)) goto got_it; else @@ -817,7 +895,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, break; case DIGITUTF8: while (s < strend) { - if (swash_fetch(PL_utf8_digit,s)) { + if (swash_fetch(PL_utf8_digit,(U8*)s)) { if (tmp && regtry(prog, s)) goto got_it; else @@ -843,7 +921,7 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, break; case NDIGITUTF8: while (s < strend) { - if (!swash_fetch(PL_utf8_digit,s)) { + if (!swash_fetch(PL_utf8_digit,(U8*)s)) { if (tmp && regtry(prog, s)) goto got_it; else @@ -862,8 +940,8 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, char *last; I32 oldpos = scream_pos; - if (screamer) { - last = screaminstr(screamer, prog->float_substr, s - strbeg, + if (flags & REXEC_SCREAM) { + last = screaminstr(sv, prog->float_substr, s - strbeg, end_shift, &scream_pos, 1); /* last one */ if (!last) { last = scream_olds; /* Only one occurence. */ @@ -933,15 +1011,20 @@ got_it: } } } - /* Preserve the current value of $^R */ - if (oreplsv != GvSV(PL_replgv)) { - sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is - restored, the value remains - the same. */ + if (PL_reg_eval_set) { + /* Preserve the current value of $^R */ + if (oreplsv != GvSV(PL_replgv)) + sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is + restored, the value remains + the same. */ + restore_pos(0); } + return 1; phooey: + if (PL_reg_eval_set) + restore_pos(0); return 0; } @@ -958,6 +1041,8 @@ regtry(regexp *prog, char *startpos) CHECKPOINT lastcp; if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) { + MAGIC *mg; + PL_reg_eval_set = RS_init; DEBUG_r(DEBUG_s( PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n", @@ -970,13 +1055,42 @@ regtry(regexp *prog, char *startpos) /* Apparently this is not needed, judging by wantarray. */ /* SAVEINT(cxstack[cxstack_ix].blk_gimme); cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ + + if (PL_reg_sv) { + /* Make $_ available to executed code. */ + if (PL_reg_sv != DEFSV) { + /* SAVE_DEFSV does *not* suffice here for USE_THREADS */ + SAVESPTR(DEFSV); + DEFSV = PL_reg_sv; + } + + if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv) + && (mg = mg_find(PL_reg_sv, 'g')))) { + /* prepare for quick setting of pos */ + sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0); + mg = mg_find(PL_reg_sv, 'g'); + mg->mg_len = -1; + } + PL_reg_magic = mg; + PL_reg_oldpos = mg->mg_len; + SAVEDESTRUCTOR(restore_pos, 0); + } + if (!PL_reg_curpm) + New(22,PL_reg_curpm, 1, PMOP); + PL_reg_curpm->op_pmregexp = prog; + PL_reg_oldcurpm = PL_curpm; + PL_curpm = PL_reg_curpm; + prog->subbeg = PL_bostr; + prog->subend = PL_regeol; /* strend may have been modified */ } + prog->startp[0] = startpos; PL_reginput = startpos; PL_regstartp = prog->startp; PL_regendp = prog->endp; PL_reglastparen = &prog->lastparen; prog->lastparen = 0; PL_regsize = 0; + DEBUG_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) @@ -985,17 +1099,19 @@ regtry(regexp *prog, char *startpos) New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*); } + /* XXXX What this code is doing here?!!! There should be no need + to do this again and again, PL_reglastparen should take care of + this! */ sp = prog->startp; ep = prog->endp; if (prog->nparens) { - for (i = prog->nparens; i >= 0; i--) { - *sp++ = NULL; - *ep++ = NULL; + for (i = prog->nparens; i >= 1; i--) { + *++sp = NULL; + *++ep = NULL; } } REGCP_SET; if (regmatch(prog->program + 1)) { - prog->startp[0] = startpos; prog->endp[0] = PL_reginput; return 1; } @@ -1056,17 +1172,31 @@ regmatch(regnode *prog) int docolor = *PL_colors[0]; int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ int l = (PL_regeol - locinput > taill ? taill : PL_regeol - locinput); + /* The part of the string before starttry has one color + (pref0_len chars), between starttry and current + position another one (pref_len - pref0_len chars), + after the current position the third one. + We assume that pref0_len <= pref_len, otherwise we + decrease pref0_len. */ int pref_len = (locinput - PL_bostr > (5 + taill) - l ? (5 + taill) - l : locinput - PL_bostr); + int pref0_len = pref_len - (locinput - PL_reg_starttry); if (l + pref_len < (5 + taill) && l < PL_regeol - locinput) l = ( PL_regeol - locinput > (5 + taill) - pref_len ? (5 + taill) - pref_len : PL_regeol - locinput); + if (pref0_len < 0) + pref0_len = 0; + if (pref0_len > pref_len) + pref0_len = pref_len; regprop(prop, scan); PerlIO_printf(Perl_debug_log, - "%4i <%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n", + "%4i <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n", locinput - PL_bostr, - PL_colors[2], pref_len, locinput - pref_len, PL_colors[3], + PL_colors[4], pref0_len, + locinput - pref_len, PL_colors[5], + PL_colors[2], pref_len - pref0_len, + locinput - pref_len + pref0_len, PL_colors[3], (docolor ? "" : "> <"), PL_colors[0], l, locinput, PL_colors[1], 15 - l - pref_len + 1, @@ -1103,7 +1233,7 @@ regmatch(regnode *prog) break; sayNO; case GPOS: - if (locinput == PL_regbol) + if (locinput == PL_reg_ganch) break; sayNO; case EOL: @@ -1129,7 +1259,7 @@ regmatch(regnode *prog) break; case SANYUTF8: if (nextchr & 0x80) { - locinput += utf8skip[nextchr]; + locinput += PL_utf8skip[nextchr]; if (locinput > PL_regeol) sayNO; nextchr = UCHARAT(locinput); @@ -1146,7 +1276,7 @@ regmatch(regnode *prog) break; case ANYUTF8: if (nextchr & 0x80) { - locinput += utf8skip[nextchr]; + locinput += PL_utf8skip[nextchr]; if (locinput > PL_regeol) sayNO; nextchr = UCHARAT(locinput); @@ -1156,7 +1286,7 @@ regmatch(regnode *prog) sayNO; nextchr = UCHARAT(++locinput); break; - case ANY: + case REG_ANY: if (!nextchr && locinput >= PL_regeol || nextchr == '\n') sayNO; nextchr = UCHARAT(++locinput); @@ -1188,8 +1318,12 @@ regmatch(regnode *prog) while (s < e) { if (l >= PL_regeol) sayNO; - if (utf8_to_uv(s, 0) != (c1 ? toLOWER_utf8(l) : toLOWER_LC_utf8(l))) + if (utf8_to_uv((U8*)s, 0) != (c1 ? + toLOWER_utf8((U8*)l) : + toLOWER_LC_utf8((U8*)l))) + { sayNO; + } s += UTF8SKIP(s); l += UTF8SKIP(l); } @@ -1201,7 +1335,7 @@ regmatch(regnode *prog) /* Inline the first character, for speed. */ if (UCHARAT(s) != nextchr && UCHARAT(s) != ((OP(scan) == EXACTF) - ? fold : fold_locale)[nextchr]) + ? PL_fold : PL_fold_locale)[nextchr]) sayNO; if (PL_regeol - locinput < ln) sayNO; @@ -1218,7 +1352,7 @@ regmatch(regnode *prog) sayNO; if (locinput >= PL_regeol) sayNO; - locinput += utf8skip[nextchr]; + locinput += PL_utf8skip[nextchr]; nextchr = UCHARAT(locinput); break; case ANYOF: @@ -1250,9 +1384,12 @@ regmatch(regnode *prog) sayNO; if (nextchr & 0x80) { if (!(OP(scan) == ALNUMUTF8 - ? swash_fetch(PL_utf8_alnum, locinput) : isALNUM_LC_utf8(locinput))) + ? swash_fetch(PL_utf8_alnum, (U8*)locinput) + : isALNUM_LC_utf8((U8*)locinput))) + { sayNO; - locinput += utf8skip[nextchr]; + } + locinput += PL_utf8skip[nextchr]; nextchr = UCHARAT(locinput); break; } @@ -1280,9 +1417,12 @@ regmatch(regnode *prog) sayNO; if (nextchr & 0x80) { if (OP(scan) == NALNUMUTF8 - ? swash_fetch(PL_utf8_alnum, locinput) : isALNUM_LC_utf8(locinput)) + ? swash_fetch(PL_utf8_alnum, (U8*)locinput) + : isALNUM_LC_utf8((U8*)locinput)) + { sayNO; - locinput += utf8skip[nextchr]; + } + locinput += PL_utf8skip[nextchr]; nextchr = UCHARAT(locinput); break; } @@ -1317,14 +1457,15 @@ regmatch(regnode *prog) case BOUNDUTF8: case NBOUNDUTF8: /* was last char in word? */ - ln = (locinput != PL_regbol) ? utf8_to_uv(reghop(locinput, -1), 0) : PL_regprev; + ln = (locinput != PL_regbol) + ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev; if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) { ln = isALNUM_uni(ln); - n = swash_fetch(PL_utf8_alnum, locinput); + n = swash_fetch(PL_utf8_alnum, (U8*)locinput); } else { ln = isALNUM_LC_uni(ln); - n = isALNUM_LC_utf8(locinput); + n = isALNUM_LC_utf8((U8*)locinput); } if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8)) sayNO; @@ -1348,9 +1489,12 @@ regmatch(regnode *prog) sayNO; if (nextchr & 0x80) { if (!(OP(scan) == SPACEUTF8 - ? swash_fetch(PL_utf8_space,locinput) : isSPACE_LC_utf8(locinput))) + ? swash_fetch(PL_utf8_space,(U8*)locinput) + : isSPACE_LC_utf8((U8*)locinput))) + { sayNO; - locinput += utf8skip[nextchr]; + } + locinput += PL_utf8skip[nextchr]; nextchr = UCHARAT(locinput); break; } @@ -1378,9 +1522,12 @@ regmatch(regnode *prog) sayNO; if (nextchr & 0x80) { if (OP(scan) == NSPACEUTF8 - ? swash_fetch(PL_utf8_space,locinput) : isSPACE_LC_utf8(locinput)) + ? swash_fetch(PL_utf8_space,(U8*)locinput) + : isSPACE_LC_utf8((U8*)locinput)) + { sayNO; - locinput += utf8skip[nextchr]; + } + locinput += PL_utf8skip[nextchr]; nextchr = UCHARAT(locinput); break; } @@ -1396,9 +1543,9 @@ regmatch(regnode *prog) break; case DIGITUTF8: if (nextchr & 0x80) { - if (!(swash_fetch(PL_utf8_digit,locinput))) + if (!(swash_fetch(PL_utf8_digit,(U8*)locinput))) sayNO; - locinput += utf8skip[nextchr]; + locinput += PL_utf8skip[nextchr]; nextchr = UCHARAT(locinput); break; } @@ -1417,9 +1564,9 @@ regmatch(regnode *prog) if (!nextchr && locinput >= PL_regeol) sayNO; if (nextchr & 0x80) { - if (swash_fetch(PL_utf8_digit,locinput)) + if (swash_fetch(PL_utf8_digit,(U8*)locinput)) sayNO; - locinput += utf8skip[nextchr]; + locinput += PL_utf8skip[nextchr]; nextchr = UCHARAT(locinput); break; } @@ -1428,10 +1575,10 @@ regmatch(regnode *prog) nextchr = UCHARAT(++locinput); break; case CLUMP: - if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark, locinput)) + if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput)) sayNO; - locinput += utf8skip[nextchr]; - while (locinput < PL_regeol && swash_fetch(PL_utf8_mark, locinput)) + locinput += PL_utf8skip[nextchr]; + while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput)) locinput += UTF8SKIP(locinput); if (locinput > PL_regeol) sayNO; @@ -1461,7 +1608,7 @@ regmatch(regnode *prog) while (s < e) { if (l >= PL_regeol) sayNO; - if (toLOWER_utf8(s) != toLOWER_utf8(l)) + if (toLOWER_utf8((U8*)s) != toLOWER_utf8((U8*)l)) sayNO; s += UTF8SKIP(s); l += UTF8SKIP(l); @@ -1471,7 +1618,7 @@ regmatch(regnode *prog) while (s < e) { if (l >= PL_regeol) sayNO; - if (toLOWER_LC_utf8(s) != toLOWER_LC_utf8(l)) + if (toLOWER_LC_utf8((U8*)s) != toLOWER_LC_utf8((U8*)l)) sayNO; s += UTF8SKIP(s); l += UTF8SKIP(l); @@ -1486,7 +1633,7 @@ regmatch(regnode *prog) if (UCHARAT(s) != nextchr && (OP(scan) == REF || (UCHARAT(s) != ((OP(scan) == REFF - ? fold : fold_locale)[nextchr])))) + ? PL_fold : PL_fold_locale)[nextchr])))) sayNO; ln = PL_regendp[n] - s; if (locinput + ln > PL_regeol) @@ -1518,21 +1665,100 @@ regmatch(regnode *prog) PL_op = (OP_4tree*)PL_regdata->data[n]; DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", PL_op) ); PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 1]); + PL_reg_magic->mg_len = locinput - PL_bostr; + PL_regendp[0] = locinput; CALLRUNOPS(); /* Scalar context. */ SPAGAIN; ret = POPs; PUTBACK; + PL_op = oop; + PL_curpad = ocurpad; + PL_curcop = ocurcop; if (logical) { - logical = 0; + if (logical == 2) { /* Postponed subexpression. */ + regexp *re; + MAGIC *mg = Null(MAGIC*); + re_cc_state state; + CURCUR cctmp; + CHECKPOINT cp, lastcp; + + if(SvROK(ret) || SvRMAGICAL(ret)) { + SV *sv = SvROK(ret) ? SvRV(ret) : ret; + + if(SvMAGICAL(sv)) + mg = mg_find(sv, 'r'); + } + if (mg) { + re = (regexp *)mg->mg_obj; + (void)ReREFCNT_inc(re); + } + else { + STRLEN len; + char *t = SvPV(ret, len); + PMOP pm; + char *oprecomp = PL_regprecomp; + I32 osize = PL_regsize; + I32 onpar = PL_regnpar; + + pm.op_pmflags = 0; + re = CALLREGCOMP(t, t + len, &pm); + if (!(SvFLAGS(ret) + & (SVs_TEMP | SVs_PADTMP | SVf_READONLY))) + sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0); + PL_regprecomp = oprecomp; + PL_regsize = osize; + PL_regnpar = onpar; + } + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "Entering embedded `%s%.60s%s%s'\n", + PL_colors[0], + re->precomp, + PL_colors[1], + (strlen(re->precomp) > 60 ? "..." : "")) + ); + state.node = next; + state.prev = PL_reg_call_cc; + state.cc = PL_regcc; + state.re = PL_reg_re; + + cctmp.cur = 0; + cctmp.oldcc = 0; + PL_regcc = &cctmp; + + cp = regcppush(0); /* Save *all* the positions. */ + REGCP_SET; + cache_re(re); + state.ss = PL_savestack_ix; + *PL_reglastparen = 0; + PL_reg_call_cc = &state; + PL_reginput = locinput; + if (regmatch(re->program + 1)) { + ReREFCNT_dec(re); + regcpblow(cp); + sayYES; + } + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") + ); + ReREFCNT_dec(re); + REGCP_UNWIND; + regcppop(); + PL_reg_call_cc = state.prev; + PL_regcc = state.cc; + PL_reg_re = state.re; + cache_re(PL_reg_re); + sayNO; + } sw = SvTRUE(ret); + logical = 0; } else sv_setsv(save_scalar(PL_replgv), ret); - PL_op = oop; - PL_curpad = ocurpad; - PL_curcop = ocurcop; break; } case OPEN: @@ -1562,7 +1788,7 @@ regmatch(regnode *prog) } break; case LOGICAL: - logical = 1; + logical = scan->flags; break; case CURLYX: { CURCUR cc; @@ -1668,10 +1894,10 @@ regmatch(regnode *prog) PL_regcc = cc; if (n >= cc->max) { /* Maximum greed exceeded? */ - if (PL_dowarn && n >= REG_INFTY + if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY && !(PL_reg_flags & RF_warned)) { PL_reg_flags |= RF_warned; - warn("%s limit (%d) exceeded", + warner(WARN_UNSAFE, "%s limit (%d) exceeded", "Complex regular subexpression recursion", REG_INFTY - 1); } @@ -1725,9 +1951,10 @@ regmatch(regnode *prog) REPORT_CODE_OFF+PL_regindent*2, "") ); } - if (PL_dowarn && n >= REG_INFTY && !(PL_reg_flags & RF_warned)) { + if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY + && !(PL_reg_flags & RF_warned)) { PL_reg_flags |= RF_warned; - warn("%s limit (%d) exceeded", + warner(WARN_UNSAFE, "%s limit (%d) exceeded", "Complex regular subexpression recursion", REG_INFTY - 1); } @@ -1825,12 +2052,12 @@ regmatch(regnode *prog) && !(paren && ln == 0)) ln = n; locinput = PL_reginput; - if (regkind[(U8)OP(next)] == EXACT) { + if (PL_regkind[(U8)OP(next)] == EXACT) { c1 = UCHARAT(OPERAND(next) + 1); if (OP(next) == EXACTF) - c2 = fold[c1]; + c2 = PL_fold[c1]; else if (OP(next) == EXACTFL) - c2 = fold_locale[c1]; + c2 = PL_fold_locale[c1]; else c2 = c1; } @@ -1846,7 +2073,7 @@ regmatch(regnode *prog) { if (paren) { if (n) { - PL_regstartp[paren] = HOP(PL_reginput, -l); + PL_regstartp[paren] = HOPc(PL_reginput, -l); PL_regendp[paren] = PL_reginput; } else @@ -1881,12 +2108,12 @@ regmatch(regnode *prog) REPORT_CODE_OFF+PL_regindent*2, "", n, l) ); if (n >= ln) { - if (regkind[(U8)OP(next)] == EXACT) { + if (PL_regkind[(U8)OP(next)] == EXACT) { c1 = UCHARAT(OPERAND(next) + 1); if (OP(next) == EXACTF) - c2 = fold[c1]; + c2 = PL_fold[c1]; else if (OP(next) == EXACTFL) - c2 = fold_locale[c1]; + c2 = PL_fold_locale[c1]; else c2 = c1; } @@ -1907,7 +2134,7 @@ regmatch(regnode *prog) ); if (paren) { if (n) { - PL_regstartp[paren] = HOP(PL_reginput, -l); + PL_regstartp[paren] = HOPc(PL_reginput, -l); PL_regendp[paren] = PL_reginput; } else @@ -1919,7 +2146,7 @@ regmatch(regnode *prog) } /* Couldn't or didn't -- back up. */ n--; - locinput = HOP(locinput, -l); + locinput = HOPc(locinput, -l); PL_reginput = locinput; } } @@ -1958,12 +2185,12 @@ regmatch(regnode *prog) * Lookahead to avoid useless match attempts * when we know what character comes next. */ - if (regkind[(U8)OP(next)] == EXACT) { + if (PL_regkind[(U8)OP(next)] == EXACT) { c1 = UCHARAT(OPERAND(next) + 1); if (OP(next) == EXACTF) - c2 = fold[c1]; + c2 = PL_fold[c1]; else if (OP(next) == EXACTFL) - c2 = fold_locale[c1]; + c2 = PL_fold_locale[c1]; else c2 = c1; } @@ -1977,6 +2204,50 @@ regmatch(regnode *prog) sayNO; locinput = PL_reginput; REGCP_SET; + if (c1 != -1000) { + char *e = locinput + n - ln; /* Should not check after this */ + char *old = locinput; + + if (e >= PL_regeol || (n == REG_INFTY)) + e = PL_regeol - 1; + while (1) { + /* Find place 'next' could work */ + if (c1 == c2) { + while (locinput <= e && *locinput != c1) + locinput++; + } else { + while (locinput <= e + && *locinput != c1 + && *locinput != c2) + locinput++; + } + if (locinput > e) + sayNO; + /* PL_reginput == old now */ + if (locinput != old) { + ln = 1; /* Did some */ + if (regrepeat(scan, locinput - old) < + locinput - old) + sayNO; + } + /* PL_reginput == locinput now */ + if (paren) { + if (ln) { + PL_regstartp[paren] = HOPc(locinput, -1); + PL_regendp[paren] = locinput; + } + else + PL_regendp[paren] = NULL; + } + if (regmatch(next)) + sayYES; + PL_reginput = locinput; /* Could be reset... */ + REGCP_UNWIND; + /* Couldn't or didn't -- move forward. */ + old = locinput++; + } + } + else while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */ /* If it could work, try it. */ if (c1 == -1000 || @@ -1985,7 +2256,7 @@ regmatch(regnode *prog) { if (paren) { if (n) { - PL_regstartp[paren] = HOP(PL_reginput, -1); + PL_regstartp[paren] = HOPc(PL_reginput, -1); PL_regendp[paren] = PL_reginput; } else @@ -2009,7 +2280,7 @@ regmatch(regnode *prog) CHECKPOINT lastcp; n = regrepeat(scan, n); locinput = PL_reginput; - if (ln < n && regkind[(U8)OP(next)] == EOL && + if (ln < n && PL_regkind[(U8)OP(next)] == EOL && (!PL_multiline || OP(next) == SEOL)) ln = n; /* why back off? */ REGCP_SET; @@ -2022,7 +2293,7 @@ regmatch(regnode *prog) { if (paren && n) { if (n) { - PL_regstartp[paren] = HOP(PL_reginput, -1); + PL_regstartp[paren] = HOPc(PL_reginput, -1); PL_regendp[paren] = PL_reginput; } else @@ -2034,7 +2305,7 @@ regmatch(regnode *prog) } /* Couldn't or didn't -- back up. */ n--; - PL_reginput = locinput = HOP(locinput, -1); + PL_reginput = locinput = HOPc(locinput, -1); } } else { @@ -2050,13 +2321,47 @@ regmatch(regnode *prog) } /* Couldn't or didn't -- back up. */ n--; - PL_reginput = locinput = HOP(locinput, -1); + PL_reginput = locinput = HOPc(locinput, -1); } } } sayNO; break; case END: + if (PL_reg_call_cc) { + re_cc_state *cur_call_cc = PL_reg_call_cc; + CURCUR *cctmp = PL_regcc; + regexp *re = PL_reg_re; + CHECKPOINT cp, lastcp; + + cp = regcppush(0); /* Save *all* the positions. */ + REGCP_SET; + regcp_set_to(PL_reg_call_cc->ss); /* Restore parens of + the caller. */ + PL_reginput = locinput; /* Make position available to + the callcc. */ + cache_re(PL_reg_call_cc->re); + PL_regcc = PL_reg_call_cc->cc; + PL_reg_call_cc = PL_reg_call_cc->prev; + if (regmatch(cur_call_cc->node)) { + PL_reg_call_cc = cur_call_cc; + regcpblow(cp); + sayYES; + } + REGCP_UNWIND; + regcppop(); + PL_reg_call_cc = cur_call_cc; + PL_regcc = cctmp; + PL_reg_re = re; + cache_re(re); + + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s continuation failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") + ); + sayNO; + } if (locinput < PL_regtill) sayNO; /* Cannot match: too short. */ /* Fall through */ @@ -2065,14 +2370,25 @@ regmatch(regnode *prog) sayYES; /* Success! */ case SUSPEND: n = 1; + PL_reginput = locinput; goto do_ifmatch; case UNLESSM: n = 0; if (scan->flags) { - s = HOPMAYBE(locinput, -scan->flags); - if (!s) - goto say_yes; - PL_reginput = s; + 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; + } } else PL_reginput = locinput; @@ -2080,10 +2396,20 @@ regmatch(regnode *prog) case IFMATCH: n = 1; if (scan->flags) { - s = HOPMAYBE(locinput, -scan->flags); - if (!s) - goto say_no; - PL_reginput = s; + 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; + } } else PL_reginput = locinput; @@ -2119,7 +2445,7 @@ regmatch(regnode *prog) default: PerlIO_printf(PerlIO_stderr(), "%lx %d\n", (unsigned long)scan, OP(scan)); - FAIL("regexp memory corruption"); + croak("regexp memory corruption"); } scan = next; } @@ -2128,7 +2454,7 @@ regmatch(regnode *prog) * We get here only if there's trouble -- normally "case END" is * the terminating point. */ - FAIL("corrupted regexp pointers"); + croak("corrupted regexp pointers"); /*NOTREACHED*/ sayNO; @@ -2168,7 +2494,7 @@ regrepeat(regnode *p, I32 max) loceol = scan + max; opnd = (char *) OPERAND(p); switch (OP(p)) { - case ANY: + case REG_ANY: while (scan < loceol && *scan != '\n') scan++; break; @@ -2197,14 +2523,14 @@ regrepeat(regnode *p, I32 max) case EXACTF: /* length of string is 1 */ c = UCHARAT(++opnd); while (scan < loceol && - (UCHARAT(scan) == c || UCHARAT(scan) == fold[c])) + (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) scan++; break; case EXACTFL: /* length of string is 1 */ PL_reg_flags |= RF_tainted; c = UCHARAT(++opnd); while (scan < loceol && - (UCHARAT(scan) == c || UCHARAT(scan) == fold_locale[c])) + (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c])) scan++; break; case ANYOFUTF8: @@ -2224,7 +2550,7 @@ regrepeat(regnode *p, I32 max) break; case ALNUMUTF8: loceol = PL_regeol; - while (scan < loceol && swash_fetch(PL_utf8_alnum, scan)) { + while (scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -2237,7 +2563,7 @@ regrepeat(regnode *p, I32 max) case ALNUMLUTF8: PL_reg_flags |= RF_tainted; loceol = PL_regeol; - while (scan < loceol && isALNUM_LC_utf8(scan)) { + while (scan < loceol && isALNUM_LC_utf8((U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -2249,7 +2575,7 @@ regrepeat(regnode *p, I32 max) break; case NALNUMUTF8: loceol = PL_regeol; - while (scan < loceol && !swash_fetch(PL_utf8_alnum, scan)) { + while (scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -2262,7 +2588,7 @@ regrepeat(regnode *p, I32 max) case NALNUMLUTF8: PL_reg_flags |= RF_tainted; loceol = PL_regeol; - while (scan < loceol && !isALNUM_LC_utf8(scan)) { + while (scan < loceol && !isALNUM_LC_utf8((U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -2273,7 +2599,7 @@ regrepeat(regnode *p, I32 max) break; case SPACEUTF8: loceol = PL_regeol; - while (scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,scan))) { + while (scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { scan += UTF8SKIP(scan); hardcount++; } @@ -2286,7 +2612,7 @@ regrepeat(regnode *p, I32 max) case SPACELUTF8: PL_reg_flags |= RF_tainted; loceol = PL_regeol; - while (scan < loceol && (*scan == ' ' || isSPACE_LC_utf8(scan))) { + while (scan < loceol && (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { scan += UTF8SKIP(scan); hardcount++; } @@ -2297,7 +2623,7 @@ regrepeat(regnode *p, I32 max) break; case NSPACEUTF8: loceol = PL_regeol; - while (scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,scan))) { + while (scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { scan += UTF8SKIP(scan); hardcount++; } @@ -2310,7 +2636,7 @@ regrepeat(regnode *p, I32 max) case NSPACELUTF8: PL_reg_flags |= RF_tainted; loceol = PL_regeol; - while (scan < loceol && !(*scan == ' ' || isSPACE_LC_utf8(scan))) { + while (scan < loceol && !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { scan += UTF8SKIP(scan); hardcount++; } @@ -2321,7 +2647,7 @@ regrepeat(regnode *p, I32 max) break; case DIGITUTF8: loceol = PL_regeol; - while (scan < loceol && swash_fetch(PL_utf8_digit,scan)) { + while (scan < loceol && swash_fetch(PL_utf8_digit,(U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -2333,7 +2659,7 @@ regrepeat(regnode *p, I32 max) break; case NDIGITUTF8: loceol = PL_regeol; - while (scan < loceol && !swash_fetch(PL_utf8_digit,scan)) { + while (scan < loceol && !swash_fetch(PL_utf8_digit,(U8*)scan)) { scan += UTF8SKIP(scan); hardcount++; } @@ -2375,14 +2701,14 @@ regrepeat_hard(regnode *p, I32 max, I32 *lp) register char *start; register char *loceol = PL_regeol; I32 l = 0; - I32 count = 0; + I32 count = 0, res = 1; if (!max) return 0; start = PL_reginput; if (UTF) { - while (PL_reginput < loceol && (scan = PL_reginput, regmatch(p))) { + while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) { if (!count++) { l = 0; while (start < PL_reginput) { @@ -2398,7 +2724,7 @@ regrepeat_hard(regnode *p, I32 max, I32 *lp) } } else { - while (PL_reginput < loceol && (scan = PL_reginput, regmatch(p))) { + while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) { if (!count++) { *lp = l = PL_reginput - start; if (max != REG_INFTY && l*max < loceol - scan) @@ -2408,14 +2734,14 @@ regrepeat_hard(regnode *p, I32 max, I32 *lp) } } } - if (PL_reginput < loceol) + if (!res) PL_reginput = scan; return count; } /* - - regclass - determine if a character falls into a character class + - reginclass - determine if a character falls into a character class */ STATIC bool @@ -2432,10 +2758,10 @@ reginclass(register char *p, register I32 c) I32 cf; if (flags & ANYOF_LOCALE) { PL_reg_flags |= RF_tainted; - cf = fold_locale[c]; + cf = PL_fold_locale[c]; } else - cf = fold[c]; + cf = PL_fold[c]; if (ANYOF_TEST(p, cf)) match = TRUE; } @@ -2467,7 +2793,7 @@ reginclassutf8(regnode *f, U8 *p) match = TRUE; else if (flags & ANYOF_FOLD) { I32 cf; - char tmpbuf[10]; + U8 tmpbuf[10]; if (flags & ANYOF_LOCALE) { PL_reg_flags |= RF_tainted; uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p)); @@ -2493,8 +2819,8 @@ reginclassutf8(regnode *f, U8 *p) return (flags & ANYOF_INVERT) ? !match : match; } -STATIC char * -reghop(unsigned char *s, I32 off) +STATIC U8 * +reghop(U8 *s, I32 off) { dTHR; if (off >= 0) { @@ -2515,8 +2841,8 @@ reghop(unsigned char *s, I32 off) return s; } -STATIC char * -reghopmaybe(unsigned char *s, I32 off) +STATIC U8 * +reghopmaybe(U8* s, I32 off) { dTHR; if (off >= 0) {