X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/00db4c452819e776e72467584ab4e9617d012f7b..517db07721f121446b16672c63a0ca8d753c8eea:/regexec.c diff --git a/regexec.c b/regexec.c index b6d2ca4..4df60ac 100644 --- a/regexec.c +++ b/regexec.c @@ -19,6 +19,30 @@ * with the POSIX routines of the same names. */ +#ifdef PERL_EXT_RE_BUILD +/* need to replace pregcomp et al, so enable that */ +# ifndef PERL_IN_XSUB_RE +# define PERL_IN_XSUB_RE +# endif +/* need access to debugger hooks */ +# if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING) +# define DEBUGGING +# endif +#endif + +#ifdef PERL_IN_XSUB_RE +/* We *really* need to overwrite these symbols: */ +# define Perl_regexec_flags my_regexec +# define Perl_regdump my_regdump +# define Perl_regprop my_regprop +# define Perl_re_intuit_start my_re_intuit_start +/* *These* symbols are masked to allow static link. */ +# define Perl_pregexec my_pregexec +# define Perl_reginitcolors my_reginitcolors + +# define PERL_NO_GET_CONTEXT +#endif + /*SUPPRESS 112*/ /* * pregcomp and pregexec -- regsub and regerror are not used in perl @@ -42,7 +66,7 @@ * **** Alterations to Henry's code are... **** - **** Copyright (c) 1991-1997, Larry Wall + **** Copyright (c) 1991-1999, Larry Wall **** **** You may distribute under the terms of either the GNU General Public **** License or the Artistic License, as specified in the README file. @@ -52,12 +76,23 @@ * regular-expression syntax might require a total rethink. */ #include "EXTERN.h" +#define PERL_IN_REGEXEC_C #include "perl.h" + +#ifdef PERL_IN_XSUB_RE +# if defined(PERL_CAPI) || defined(PERL_OBJECT) +# include "XSUB.h" +# endif +#endif + #include "regcomp.h" #define RF_tainted 1 /* tainted information used? */ #define RF_warned 2 /* warned about big count? */ #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 RS_init 1 /* eval environment created */ #define RS_set 2 /* replsv value is set */ @@ -66,93 +101,124 @@ #define STATIC static #endif -#ifndef PERL_OBJECT -typedef I32 CHECKPOINT; - /* * Forwards. */ -static I32 regmatch _((regnode *prog)); -static I32 regrepeat _((regnode *p, I32 max)); -static I32 regrepeat_hard _((regnode *p, I32 max, I32 *lp)); -static I32 regtry _((regexp *prog, char *startpos)); +#define REGINCLASS(p,c) (ANYOF_FLAGS(p) ? reginclass(p,c) : ANYOF_BITMAP_TEST(p,c)) +#define REGINCLASSUTF8(f,p) (ARG1(f) ? reginclassutf8(f,p) : swash_fetch((SV*)PL_regdata->data[ARG2(f)],p)) + +#define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) +#define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) + +#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 void restore_pos(pTHXo_ void *arg); -static bool reginclass _((char *p, I32 c)); -static CHECKPOINT regcppush _((I32 parenfloor)); -static char * regcppop _((void)); -#endif -#define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c)) STATIC CHECKPOINT -regcppush(I32 parenfloor) +S_regcppush(pTHX_ I32 parenfloor) { dTHR; - int retval = savestack_ix; - int i = (regsize - parenfloor) * 4; + int retval = PL_savestack_ix; + int i = (PL_regsize - parenfloor) * 4; int p; SSCHECK(i + 5); - for (p = regsize; p > parenfloor; p--) { - SSPUSHPTR(regendp[p]); - SSPUSHPTR(regstartp[p]); - SSPUSHPTR(reg_start_tmp[p]); + for (p = PL_regsize; p > parenfloor; p--) { + SSPUSHINT(PL_regendp[p]); + SSPUSHINT(PL_regstartp[p]); + SSPUSHPTR(PL_reg_start_tmp[p]); SSPUSHINT(p); } - SSPUSHINT(regsize); - SSPUSHINT(*reglastparen); - SSPUSHPTR(reginput); + SSPUSHINT(PL_regsize); + SSPUSHINT(*PL_reglastparen); + SSPUSHPTR(PL_reginput); SSPUSHINT(i + 3); SSPUSHINT(SAVEt_REGCONTEXT); return retval; } /* These are needed since we do not localize EVAL nodes: */ -# define REGCP_SET DEBUG_r(PerlIO_printf(Perl_debug_log, " Setting an EVAL scope, savestack=%i\n", savestack_ix)); lastcp = savestack_ix -# define REGCP_UNWIND DEBUG_r(lastcp != savestack_ix ? PerlIO_printf(Perl_debug_log," Clearing an EVAL scope, savestack=%i..%i\n", lastcp, savestack_ix) : 0); regcpblow(lastcp) +# define REGCP_SET DEBUG_r(PerlIO_printf(Perl_debug_log, \ + " Setting an EVAL scope, savestack=%i\n", \ + PL_savestack_ix)); lastcp = PL_savestack_ix + +# define REGCP_UNWIND DEBUG_r(lastcp != PL_savestack_ix ? \ + PerlIO_printf(Perl_debug_log, \ + " Clearing an EVAL scope, savestack=%i..%i\n", \ + lastcp, PL_savestack_ix) : 0); regcpblow(lastcp) STATIC char * -regcppop(void) +S_regcppop(pTHX) { dTHR; I32 i = SSPOPINT; U32 paren = 0; char *input; - char *tmps; + I32 tmps; assert(i == SAVEt_REGCONTEXT); i = SSPOPINT; input = (char *) SSPOPPTR; - *reglastparen = SSPOPINT; - regsize = SSPOPINT; + *PL_reglastparen = SSPOPINT; + PL_regsize = SSPOPINT; for (i -= 3; i > 0; i -= 4) { paren = (U32)SSPOPINT; - reg_start_tmp[paren] = (char *) SSPOPPTR; - regstartp[paren] = (char *) SSPOPPTR; - tmps = (char*)SSPOPPTR; - if (paren <= *reglastparen) - regendp[paren] = tmps; + PL_reg_start_tmp[paren] = (char *) SSPOPPTR; + PL_regstartp[paren] = SSPOPINT; + tmps = SSPOPINT; + if (paren <= *PL_reglastparen) + PL_regendp[paren] = tmps; DEBUG_r( - PerlIO_printf(Perl_debug_log, " restoring \\%d to %d(%d)..%d%s\n", - paren, regstartp[paren] - regbol, - reg_start_tmp[paren] - regbol, - regendp[paren] - regbol, - (paren > *reglastparen ? "(no)" : "")); + PerlIO_printf(Perl_debug_log, + " restoring \\%d to %d(%d)..%d%s\n", + paren, PL_regstartp[paren], + PL_reg_start_tmp[paren] - PL_bostr, + PL_regendp[paren], + (paren > *PL_reglastparen ? "(no)" : "")); ); } DEBUG_r( - if (*reglastparen + 1 <= regnpar) { - PerlIO_printf(Perl_debug_log, " restoring \\%d..\\%d to undef\n", - *reglastparen + 1, regnpar); + if (*PL_reglastparen + 1 <= PL_regnpar) { + PerlIO_printf(Perl_debug_log, + " restoring \\%d..\\%d to undef\n", + *PL_reglastparen + 1, PL_regnpar); } ); - for (paren = *reglastparen + 1; paren <= regnpar; paren++) { - if (paren > regsize) - regstartp[paren] = Nullch; - regendp[paren] = Nullch; + for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) { + if (paren > PL_regsize) + PL_regstartp[paren] = -1; + PL_regendp[paren] = -1; } return input; } +STATIC char * +S_regcp_set_to(pTHX_ 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) /* @@ -163,7 +229,8 @@ regcppop(void) - pregexec - match a regexp against a string */ I32 -pregexec(register regexp *prog, char *stringarg, register char *strend, char *strbeg, I32 minend, SV *screamer, U32 nosave) +Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, + char *strbeg, I32 minend, SV *screamer, U32 nosave) /* strend: pointer to null at end of string */ /* strbeg: real beginning of string */ /* minend: end of match must be >=minend after stringarg. */ @@ -173,18 +240,220 @@ pregexec(register regexp *prog, char *stringarg, register char *strend, char *st regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, nosave ? 0 : REXEC_COPY_STR); } - + +STATIC void +S_cache_re(pTHX_ 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; +} + +/* + * Need to implement the following flags for reg_anch: + * + * USE_INTUIT_NOML - Useful to call re_intuit_start() first + * USE_INTUIT_ML + * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer + * INTUIT_AUTORITATIVE_ML + * INTUIT_ONCE_NOML - Intuit can match in one location only. + * INTUIT_ONCE_ML + * + * Another flag for this function: SECOND_TIME (so that float substrs + * with giant delta may be not rechecked). + */ + +/* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ + +/* If SCREAM, then sv should be compatible with strpos and strend. + Otherwise, only SvCUR(sv) is used to get strbeg. */ + +/* XXXX We assume that strpos is strbeg unless sv. */ + +char * +Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, + char *strend, U32 flags, re_scream_pos_data *data) +{ + I32 start_shift; + /* Should be nonnegative! */ + I32 end_shift; + char *s; + char *t; + I32 ml_anch; + + DEBUG_r( if (!PL_colorset) reginitcolors() ); + DEBUG_r(PerlIO_printf(Perl_debug_log, + "%sGuessing start of match:%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 - strpos > 60 ? 60 : strend - strpos), + strpos, PL_colors[1], + (strend - strpos > 60 ? "..." : "")) + ); + + if (prog->minlen > strend - strpos) + goto fail; + + /* XXXX Move further down? */ + start_shift = prog->check_offset_min; /* okay to underestimate on CC */ + /* Should be nonnegative! */ + end_shift = prog->minlen - start_shift - + CHR_SVLEN(prog->check_substr) + (SvTAIL(prog->check_substr) != 0); + + if (prog->reganch & ROPT_ANCH) { + ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE) + || ( (prog->reganch & ROPT_ANCH_BOL) + && !PL_multiline ) ); + + if ((prog->check_offset_min == prog->check_offset_max) && !ml_anch) { + /* Anchored... */ + I32 slen; + + if ( !(prog->reganch & ROPT_ANCH_GPOS) /* Checked by the caller */ + && (sv && (strpos + SvCUR(sv) != strend)) ) + goto fail; + + s = (char*)HOP((U8*)strpos, prog->check_offset_min); + if (SvTAIL(prog->check_substr)) { + slen = SvCUR(prog->check_substr); /* >= 1 */ + + if ( strend - s > slen || strend - s < slen - 1 ) { + s = Nullch; + goto finish; + } + if ( strend - s == slen && strend[-1] != '\n') { + s = Nullch; + goto finish; + } + /* Now should match s[0..slen-2] */ + slen--; + if (slen && (*SvPVX(prog->check_substr) != *s + || (slen > 1 + && memNE(SvPVX(prog->check_substr), s, slen)))) + s = Nullch; + } + else if (*SvPVX(prog->check_substr) != *s + || ((slen = SvCUR(prog->check_substr)) > 1 + && memNE(SvPVX(prog->check_substr), s, slen))) + s = Nullch; + else + s = strpos; + goto finish; + } + s = strpos; + if (!ml_anch && (s + prog->check_offset_max < strend - prog->minlen)) + end_shift += strend - s - prog->minlen - prog->check_offset_max; + } + else { + ml_anch = 0; + s = strpos; + } + + restart: + if (flags & REXEC_SCREAM) { + SV *c = prog->check_substr; + char *strbeg = SvPVX(sv); /* XXXX Assume PV_force() on SCREAM! */ + I32 p = -1; /* Internal iterator of scream. */ + I32 *pp = data ? data->scream_pos : &p; + + if (PL_screamfirst[BmRARE(c)] >= 0 + || ( BmRARE(c) == '\n' + && (BmPREVIOUS(c) == SvCUR(c) - 1) + && SvTAIL(c) )) + s = screaminstr(sv, prog->check_substr, + start_shift + (strpos - strbeg), end_shift, pp, 0); + else + s = Nullch; + if (data) + *data->scream_olds = s; + } + else + s = fbm_instr((unsigned char*)s + start_shift, + (unsigned char*)strend - end_shift, + prog->check_substr, PL_multiline ? FBMrf_MULTILINE : 0); + + /* Update the count-of-usability, remove useless subpatterns, + unshift s. */ + finish: + if (!s) { + ++BmUSEFUL(prog->check_substr); /* hooray */ + goto fail; /* not present */ + } + else if (s - strpos > prog->check_offset_max && + ((prog->reganch & ROPT_UTF8) + ? ((t = reghopmaybe_c(s, -(prog->check_offset_max))) + && t >= strpos) + : (t = s - prog->check_offset_max) != 0) ) { + if (ml_anch && t[-1] != '\n') { + find_anchor: + while (t < strend - end_shift - prog->minlen) { + if (*t == '\n') { + if (t < s - prog->check_offset_min) { + s = t + 1; + goto set_useful; + } + s = t + 1; + goto restart; + } + t++; + } + s = Nullch; + goto finish; + } + s = t; + set_useful: + ++BmUSEFUL(prog->check_substr); /* hooray/2 */ + } + else { + if (ml_anch && sv + && (strpos + SvCUR(sv) != strend) && strpos[-1] != '\n') { + t = strpos; + goto find_anchor; + } + if (!(prog->reganch & ROPT_NAUGHTY) + && --BmUSEFUL(prog->check_substr) < 0 + && prog->check_substr == prog->float_substr) { /* boo */ + /* If flags & SOMETHING - do not do it many times on the same match */ + SvREFCNT_dec(prog->check_substr); + prog->check_substr = Nullsv; /* disable */ + prog->float_substr = Nullsv; /* clear */ + s = strpos; + prog->reganch &= ~RE_USE_INTUIT; + } + else + s = strpos; + } + + DEBUG_r(PerlIO_printf(Perl_debug_log, "%sFound%s at offset %ld\n", + PL_colors[4],PL_colors[5], (long)(s - strpos)) ); + return s; + fail: + DEBUG_r(PerlIO_printf(Perl_debug_log, "%sNot found...%s\n", + PL_colors[4],PL_colors[5])); + return Nullch; +} + /* - 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) +Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, + 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. */ /* data: May be used for some additional optimizations. */ /* nosave: For optimizations. */ { + dTHR; register char *s; register regnode *c; register char *startpos = stringarg; @@ -193,25 +462,24 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha I32 dontbother = 0; /* how many characters not to try at end */ CURCUR cc; I32 start_shift = 0; /* Offset of the start to find - constant substr. */ - I32 end_shift = 0; /* Same for the end. */ + constant substr. */ /* CC */ + I32 end_shift = 0; /* Same for the end. */ /* CC */ I32 scream_pos = -1; /* Internal iterator of scream. */ char *scream_olds; - SV* oreplsv = GvSV(replgv); + SV* oreplsv = GvSV(PL_replgv); cc.cur = 0; cc.oldcc = 0; - regcc = &cc; + PL_regcc = &cc; - regprecomp = prog->precomp; /* Needed for error messages. */ + cache_re(prog); #ifdef DEBUGGING - regnarrate = debug & 512; - regprogram = prog->program; + PL_regnarrate = PL_debug & 512; #endif /* Be paranoid... */ if (prog == NULL || startpos == NULL) { - croak("NULL regexp parameter"); + Perl_croak(aTHX_ "NULL regexp parameter"); return 0; } @@ -219,120 +487,147 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha if (strend - startpos < minlen) goto phooey; if (startpos == strbeg) /* is ^ valid at stringarg? */ - regprev = '\n'; + PL_regprev = '\n'; else { - regprev = stringarg[-1]; - if (!multiline && regprev == '\n') - regprev = '\0'; /* force ^ to NOT match */ + PL_regprev = (U32)stringarg[-1]; + if (!PL_multiline && PL_regprev == '\n') + PL_regprev = '\0'; /* force ^ to NOT match */ } /* Check validity of program. */ - if (UCHARAT(prog->program) != MAGIC) { - FAIL("corrupted regexp program"); + if (UCHARAT(prog->program) != REG_MAGIC) { + Perl_croak(aTHX_ "corrupted regexp program"); } - regnpar = prog->nparens; - reg_flags = 0; - reg_eval_set = 0; + PL_reg_flags = 0; + PL_reg_eval_set = 0; - /* If there is a "must appear" string, look for it. */ - s = startpos; - if (!(flags & REXEC_CHECKED) - && prog->check_substr != Nullsv && - !(prog->reganch & ROPT_ANCH_GPOS) && - (!(prog->reganch & (ROPT_ANCH_BOL | ROPT_ANCH_MBOL)) - || (multiline && prog->check_substr == prog->anchored_substr)) ) - { - start_shift = prog->check_offset_min; - /* Should be nonnegative! */ - end_shift = minlen - start_shift - SvCUR(prog->check_substr); - if (screamer) { - if (screamfirst[BmRARE(prog->check_substr)] >= 0) - s = screaminstr(screamer, prog->check_substr, - start_shift + (stringarg - strbeg), - end_shift, &scream_pos, 0); - else - s = Nullch; - scream_olds = s; - } - else - s = fbm_instr((unsigned char*)s + start_shift, - (unsigned char*)strend - end_shift, - prog->check_substr); - if (!s) { - ++BmUSEFUL(prog->check_substr); /* hooray */ - goto phooey; /* not present */ - } else if ((s - stringarg) > prog->check_offset_max) { - ++BmUSEFUL(prog->check_substr); /* hooray/2 */ - s -= prog->check_offset_max; - } else if (!prog->naughty - && --BmUSEFUL(prog->check_substr) < 0 - && prog->check_substr == prog->float_substr) { /* boo */ - SvREFCNT_dec(prog->check_substr); - prog->check_substr = Nullsv; /* disable */ - prog->float_substr = Nullsv; /* clear */ - s = startpos; - } else s = startpos; - } + if (prog->reganch & ROPT_UTF8) + PL_reg_flags |= RF_utf8; /* Mark beginning of line for ^ and lookbehind. */ - regbol = startpos; - bostr = strbeg; + PL_regbol = startpos; + PL_bostr = strbeg; + PL_reg_sv = sv; /* Mark end of line for $ (and such) */ - regeol = strend; + PL_regeol = strend; /* see how far we have to get to not match where we matched before */ - regtill = startpos+minend; + PL_regtill = startpos+minend; - DEBUG_r( - PerlIO_printf(Perl_debug_log, - "Matching `%.60s%s' against `%.*s%s'\n", - prog->precomp, + /* We start without call_cc context. */ + PL_reg_call_cc = 0; + + /* If there is a "must appear" string, look for it. */ + s = startpos; + + 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; + if (prog->reganch & ROPT_ANCH_GPOS) { + if (s > PL_reg_ganch) + goto phooey; + s = PL_reg_ganch; + } + } + + if (!(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) { + re_scream_pos_data d; + + d.scream_olds = &scream_olds; + d.scream_pos = &scream_pos; + s = re_intuit_start(prog, sv, s, strend, flags, &d); + if (!s) + goto phooey; /* not present */ + } + + 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 ? "..." : "")) ); /* Simplest case: anchored match need be tried only once. */ /* [unless only anchor is BOL and multiline is set] */ - if (prog->reganch & ROPT_ANCH) { - if (regtry(prog, startpos)) + if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { + if (s == startpos && regtry(prog, startpos)) goto got_it; - else if (!(prog->reganch & ROPT_ANCH_GPOS) && - (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? */ { + char *end; + if (minlen) dontbother = minlen - 1; - strend -= dontbother; + end = HOPc(strend, -dontbother) - 1; /* for multiline we only have to try after newlines */ - if (s > startpos) - s--; - while (s < strend) { - if (*s++ == '\n') { - if (s < strend && regtry(prog, s)) + if (prog->check_substr) { + while (1) { + if (regtry(prog, s)) goto got_it; - } + if (s >= end) + goto phooey; + s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL); + if (!s) + goto phooey; + } + } else { + if (s > startpos) + s--; + while (s < end) { + if (*s++ == '\n') { /* don't need PL_utf8skip here */ + if (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. */ if (prog->anchored_substr && prog->reganch & ROPT_SKIP) { /* we have /x+whatever/ */ - /* it must be a one character string */ + /* it must be a one character string (XXXX Except UTF?) */ char ch = SvPVX(prog->anchored_substr)[0]; - while (s < strend) { - if (*s == ch) { - if (regtry(prog, s)) goto got_it; - s++; - while (s < strend && *s == ch) + if (UTF) { + while (s < strend) { + if (*s == ch) { + if (regtry(prog, s)) goto got_it; + s += UTF8SKIP(s); + while (s < strend && *s == ch) + s += UTF8SKIP(s); + } + s += UTF8SKIP(s); + } + } + else { + while (s < strend) { + if (*s == ch) { + if (regtry(prog, s)) goto got_it; s++; + while (s < strend && *s == ch) + s++; + } + s++; } - s++; } } /*SUPPRESS 560*/ @@ -346,50 +641,83 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha I32 back_min = prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset; I32 delta = back_max - back_min; - char *last = strend - SvCUR(must) - back_min; /* Cannot start after this */ - char *last1 = s - 1; /* Last position checked before */ + char *last = HOPc(strend, /* Cannot start after this */ + -(I32)(CHR_SVLEN(must) + - (SvTAIL(must) != 0) + back_min)); + char *last1; /* Last position checked before */ + + if (s > PL_bostr) + last1 = HOPc(s, -1); + else + last1 = s - 1; /* bogus */ /* XXXX check_substr already used to find `s', can optimize if check_substr==must. */ scream_pos = -1; dontbother = end_shift; - strend -= dontbother; + strend = HOPc(strend, -dontbother); while ( (s <= last) && - (screamer - ? (s = screaminstr(screamer, must, 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*)s + back_min, - (unsigned char*)strend, must))) ) { - if (s - back_max > last1) { - last1 = s - back_min; - s = s - back_max; - } else { - char *t = last1 + 1; + : (s = fbm_instr((unsigned char*)HOP(s, back_min), + (unsigned char*)strend, must, + PL_multiline ? FBMrf_MULTILINE : 0))) ) { + if (HOPc(s, -back_max) > last1) { + last1 = HOPc(s, -back_min); + s = HOPc(s, -back_max); + } + else { + char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; - last1 = s - back_min; + last1 = HOPc(s, -back_min); s = t; } - while (s <= last1) { - if (regtry(prog, s)) - goto got_it; - s++; + if (UTF) { + while (s <= last1) { + if (regtry(prog, s)) + goto got_it; + s += UTF8SKIP(s); + } + } + else { + while (s <= last1) { + if (regtry(prog, s)) + goto got_it; + s++; + } } } goto phooey; - } else if (c = prog->regstclass) { + } + else if (c = prog->regstclass) { I32 doevery = (prog->reganch & ROPT_SKIP) == 0; - char *Class; + char *cc; if (minlen) dontbother = minlen - 1; - 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)) { + case ANYOFUTF8: + cc = (char *) OPERAND(c); + while (s < strend) { + if (REGINCLASSUTF8(c, (U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; case ANYOF: - Class = (char *) OPERAND(c); + cc = (char *) OPERAND(c); while (s < strend) { - if (REGINCLASS(Class, *s)) { + if (REGINCLASS(cc, *s)) { if (tmp && regtry(prog, s)) goto got_it; else @@ -401,12 +729,14 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha } break; case BOUNDL: - reg_flags |= RF_tainted; + PL_reg_flags |= RF_tainted; /* FALL THROUGH */ case BOUND: - if (minlen) - dontbother++,strend--; - tmp = (s != startpos) ? UCHARAT(s - 1) : regprev; + if (minlen) { + dontbother++; + strend -= 1; + } + tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev; tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); while (s < strend) { if (tmp == !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { @@ -419,13 +749,39 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha if ((minlen || tmp) && regtry(prog,s)) goto got_it; break; + case BOUNDLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case BOUNDUTF8: + if (minlen) { + dontbother++; + strend = reghop_c(strend, -1); + } + 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, (U8*)s) : + isALNUM_LC_utf8((U8*)s))) + { + tmp = !tmp; + if (regtry(prog, s)) + goto got_it; + } + s += UTF8SKIP(s); + } + if ((minlen || tmp) && regtry(prog,s)) + goto got_it; + break; case NBOUNDL: - reg_flags |= RF_tainted; + PL_reg_flags |= RF_tainted; /* FALL THROUGH */ case NBOUND: - if (minlen) - dontbother++,strend--; - tmp = (s != startpos) ? UCHARAT(s - 1) : regprev; + if (minlen) { + dontbother++; + strend -= 1; + } + tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev; tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); while (s < strend) { if (tmp == !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) @@ -437,6 +793,28 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha if ((minlen || !tmp) && regtry(prog,s)) goto got_it; break; + case NBOUNDLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NBOUNDUTF8: + if (minlen) { + dontbother++; + strend = reghop_c(strend, -1); + } + 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, (U8*)s) : + isALNUM_LC_utf8((U8*)s))) + tmp = !tmp; + else if (regtry(prog, s)) + goto got_it; + s += UTF8SKIP(s); + } + if ((minlen || !tmp) && regtry(prog,s)) + goto got_it; + break; case ALNUM: while (s < strend) { if (isALNUM(*s)) { @@ -450,8 +828,21 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha s++; } break; + case ALNUMUTF8: + while (s < strend) { + if (swash_fetch(PL_utf8_alnum, (U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; case ALNUML: - reg_flags |= RF_tainted; + PL_reg_flags |= RF_tainted; while (s < strend) { if (isALNUM_LC(*s)) { if (tmp && regtry(prog, s)) @@ -464,6 +855,20 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha s++; } break; + case ALNUMLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isALNUM_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; case NALNUM: while (s < strend) { if (!isALNUM(*s)) { @@ -477,8 +882,21 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha s++; } break; + case NALNUMUTF8: + while (s < strend) { + if (!swash_fetch(PL_utf8_alnum, (U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; case NALNUML: - reg_flags |= RF_tainted; + PL_reg_flags |= RF_tainted; while (s < strend) { if (!isALNUM_LC(*s)) { if (tmp && regtry(prog, s)) @@ -491,6 +909,20 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha s++; } break; + case NALNUMLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isALNUM_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; case SPACE: while (s < strend) { if (isSPACE(*s)) { @@ -504,8 +936,21 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha s++; } break; + case SPACEUTF8: + while (s < strend) { + if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; case SPACEL: - reg_flags |= RF_tainted; + PL_reg_flags |= RF_tainted; while (s < strend) { if (isSPACE_LC(*s)) { if (tmp && regtry(prog, s)) @@ -518,6 +963,20 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha s++; } break; + case SPACELUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; case NSPACE: while (s < strend) { if (!isSPACE(*s)) { @@ -531,10 +990,9 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha s++; } break; - case NSPACEL: - reg_flags |= RF_tainted; + case NSPACEUTF8: while (s < strend) { - if (!isSPACE_LC(*s)) { + if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) { if (tmp && regtry(prog, s)) goto got_it; else @@ -542,12 +1000,13 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha } else tmp = 1; - s++; + s += UTF8SKIP(s); } break; - case DIGIT: + case NSPACEL: + PL_reg_flags |= RF_tainted; while (s < strend) { - if (isDIGIT(*s)) { + if (!isSPACE_LC(*s)) { if (tmp && regtry(prog, s)) goto got_it; else @@ -558,9 +1017,10 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha s++; } break; - case NDIGIT: + case NSPACELUTF8: + PL_reg_flags |= RF_tainted; while (s < strend) { - if (!isDIGIT(*s)) { + if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) { if (tmp && regtry(prog, s)) goto got_it; else @@ -568,409 +1028,2184 @@ regexec_flags(register regexp *prog, char *stringarg, register char *strend, cha } else tmp = 1; - s++; + s += UTF8SKIP(s); } break; - } - } - else { - dontbother = 0; - if (prog->float_substr != Nullsv) { /* Trim the end. */ + case DIGIT: + while (s < strend) { + if (isDIGIT(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case DIGITUTF8: + while (s < strend) { + if (swash_fetch(PL_utf8_digit,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case DIGITL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isDIGIT_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case DIGITLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isDIGIT_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NDIGIT: + while (s < strend) { + if (!isDIGIT(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NDIGITUTF8: + while (s < strend) { + if (!swash_fetch(PL_utf8_digit,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NDIGITL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isDIGIT_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NDIGITLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isDIGIT_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case ALNUMC: + while (s < strend) { + if (isALNUMC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case ALNUMCUTF8: + while (s < strend) { + if (swash_fetch(PL_utf8_alnumc, (U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case ALNUMCL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isALNUMC_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case ALNUMCLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isALNUMC_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NALNUMC: + while (s < strend) { + if (!isALNUMC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NALNUMCUTF8: + while (s < strend) { + if (!swash_fetch(PL_utf8_alnumc, (U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NALNUMCL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isALNUMC_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NALNUMCLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isALNUMC_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case ASCII: + while (s < strend) { + if (isASCII(*(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NASCII: + while (s < strend) { + if (!isASCII(*(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case CNTRL: + while (s < strend) { + if (isCNTRL(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case CNTRLUTF8: + while (s < strend) { + if (swash_fetch(PL_utf8_cntrl,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case CNTRLL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isCNTRL_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case CNTRLLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (*s == ' ' || isCNTRL_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NCNTRL: + while (s < strend) { + if (!isCNTRL(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NCNTRLUTF8: + while (s < strend) { + if (!swash_fetch(PL_utf8_cntrl,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NCNTRLL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isCNTRL_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NCNTRLLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isCNTRL_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case GRAPH: + while (s < strend) { + if (isGRAPH(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case GRAPHUTF8: + while (s < strend) { + if (swash_fetch(PL_utf8_graph,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case GRAPHL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isGRAPH_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case GRAPHLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (*s == ' ' || isGRAPH_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NGRAPH: + while (s < strend) { + if (!isGRAPH(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NGRAPHUTF8: + while (s < strend) { + if (!swash_fetch(PL_utf8_graph,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NGRAPHL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isGRAPH_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NGRAPHLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isGRAPH_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case LOWER: + while (s < strend) { + if (isLOWER(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case LOWERUTF8: + while (s < strend) { + if (swash_fetch(PL_utf8_lower,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case LOWERL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isLOWER_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case LOWERLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (*s == ' ' || isLOWER_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NLOWER: + while (s < strend) { + if (!isLOWER(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NLOWERUTF8: + while (s < strend) { + if (!swash_fetch(PL_utf8_lower,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NLOWERL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isLOWER_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NLOWERLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isLOWER_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case PRINT: + while (s < strend) { + if (isPRINT(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case PRINTUTF8: + while (s < strend) { + if (swash_fetch(PL_utf8_print,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case PRINTL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isPRINT_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case PRINTLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (*s == ' ' || isPRINT_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NPRINT: + while (s < strend) { + if (!isPRINT(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NPRINTUTF8: + while (s < strend) { + if (!swash_fetch(PL_utf8_print,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NPRINTL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isPRINT_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NPRINTLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isPRINT_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case PUNCT: + while (s < strend) { + if (isPUNCT(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case PUNCTUTF8: + while (s < strend) { + if (swash_fetch(PL_utf8_punct,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case PUNCTL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isPUNCT_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case PUNCTLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (*s == ' ' || isPUNCT_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NPUNCT: + while (s < strend) { + if (!isPUNCT(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NPUNCTUTF8: + while (s < strend) { + if (!swash_fetch(PL_utf8_punct,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NPUNCTL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isPUNCT_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NPUNCTLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isPUNCT_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case UPPER: + while (s < strend) { + if (isUPPER(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case UPPERUTF8: + while (s < strend) { + if (swash_fetch(PL_utf8_upper,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case UPPERL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isUPPER_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case UPPERLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (*s == ' ' || isUPPER_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NUPPER: + while (s < strend) { + if (!isUPPER(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NUPPERUTF8: + while (s < strend) { + if (!swash_fetch(PL_utf8_upper,(U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case NUPPERL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isUPPER_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NUPPERLUTF8: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isUPPER_LC_utf8((U8*)s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s += UTF8SKIP(s); + } + break; + case XDIGIT: + while (s < strend) { + if (isXDIGIT(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NXDIGIT: + while (s < strend) { + if (!isXDIGIT(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + } + } + else { + dontbother = 0; + if (prog->float_substr != Nullsv) { /* Trim the end. */ char *last; I32 oldpos = scream_pos; - if (screamer) { - last = screaminstr(screamer, prog->float_substr, s - strbeg, - end_shift, &scream_pos, 1); /* last one */ - if (!last) { - last = scream_olds; /* Only one occurence. */ + 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. */ + } + else { + STRLEN len; + char *little = SvPV(prog->float_substr, len); + + if (SvTAIL(prog->float_substr)) { + if (memEQ(strend - len + 1, little, len - 1)) + last = strend - len + 1; + else if (!PL_multiline) + last = memEQ(strend - len, little, len) + ? strend - len : Nullch; + else + goto find_last; + } else { + find_last: + if (len) + last = rninstr(s, strend, little, little + len); + else + last = strend; /* matching `$' */ + } + } + if (last == NULL) goto phooey; /* Should not happen! */ + dontbother = strend - last + prog->float_min_offset; + } + if (minlen && (dontbother < minlen)) + dontbother = minlen - 1; + strend -= dontbother; /* this one's always in bytes! */ + /* We don't know much -- general case. */ + if (UTF) { + for (;;) { + if (regtry(prog, s)) + goto got_it; + if (s >= strend) + break; + s += UTF8SKIP(s); + }; + } + else { + do { + if (regtry(prog, s)) + goto got_it; + } while (s++ < strend); + } + } + + /* Failure. */ + goto phooey; + +got_it: + RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); + + 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(aTHXo_ 0); + } + + /* 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); + } + if (flags & REXEC_COPY_STR) { + I32 i = PL_regeol - startpos + (stringarg - strbeg); + + s = savepvn(strbeg, i); + prog->subbeg = s; + prog->sublen = i; + RX_MATCH_COPIED_on(prog); + } + else { + prog->subbeg = strbeg; + prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ + } + } + + return 1; + +phooey: + if (PL_reg_eval_set) + restore_pos(aTHXo_ 0); + return 0; +} + +/* + - regtry - try match at specific point + */ +STATIC I32 /* 0 failure, 1 success */ +S_regtry(pTHX_ regexp *prog, char *startpos) +{ + dTHR; + register I32 i; + register I32 *sp; + register I32 *ep; + 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", + PL_stack_sp - PL_stack_base); + )); + SAVEINT(cxstack[cxstack_ix].blk_oldsp); + cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; + /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ + SAVETMPS; + /* 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; + if (RX_MATCH_COPIED(prog)) { + /* Here is a serious problem: we cannot rewrite subbeg, + since it may be needed if this match fails. Thus + $` inside (?{}) could fail... */ + PL_reg_oldsaved = prog->subbeg; + PL_reg_oldsavedlen = prog->sublen; + RX_MATCH_COPIED_off(prog); + } + else + PL_reg_oldsaved = Nullch; + prog->subbeg = PL_bostr; + prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ + } + prog->startp[0] = startpos - PL_bostr; + 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) + Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); + else + 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 >= 1; i--) { + *++sp = -1; + *++ep = -1; + } + } + REGCP_SET; + if (regmatch(prog->program + 1)) { + prog->endp[0] = PL_reginput - PL_bostr; + return 1; + } + REGCP_UNWIND; + return 0; +} + +/* + - regmatch - main matching routine + * + * Conceptually the strategy is simple: check to see whether the current + * node matches, call self recursively to see whether the rest matches, + * and then act accordingly. In practice we make some effort to avoid + * recursion, in particular by going through "ordinary" nodes (that don't + * need to know whether the rest of the match failed) by a loop instead of + * by recursion. + */ +/* [lwall] I've hoisted the register declarations to the outer block in order to + * maybe save a little bit of pushing and popping on the stack. It also takes + * advantage of machines that use a register save mask on subroutine entry. + */ +STATIC I32 /* 0 failure, 1 success */ +S_regmatch(pTHX_ regnode *prog) +{ + dTHR; + register regnode *scan; /* Current node. */ + regnode *next; /* Next node. */ + regnode *inner; /* Next node in internal branch. */ + 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 char *locinput = PL_reginput; + register I32 c1, c2, paren; /* case fold search, parenth */ + int minmod = 0, sw = 0, logical = 0; +#ifdef DEBUGGING + PL_regindent++; +#endif + + /* Note that nextchr is a byte even in UTF */ + nextchr = UCHARAT(locinput); + scan = prog; + while (scan != NULL) { +#define sayNO_L (logical ? (logical = 0, sw = 0, goto cont) : sayNO) +#ifdef DEBUGGING +# define sayYES goto yes +# define sayNO goto no +# define saySAME(x) if (x) goto yes; else goto no +# define REPORT_CODE_OFF 24 +#else +# define sayYES return 1 +# define sayNO return 0 +# define saySAME(x) return x +#endif + DEBUG_r( { + SV *prop = sv_newmortal(); + 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%.*s%s>%*s|%3d:%*s%s\n", + locinput - PL_bostr, + 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, + "", + scan - PL_regprogram, PL_regindent*2, "", + SvPVX(prop)); + } ); + + next = scan + NEXT_OFF(scan); + if (next == scan) + next = NULL; + + switch (OP(scan)) { + case BOL: + if (locinput == PL_bostr + ? PL_regprev == '\n' + : (PL_multiline && + (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) + { + /* regtill = regbol; */ + break; + } + sayNO; + case MBOL: + if (locinput == PL_bostr + ? PL_regprev == '\n' + : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) + { + break; + } + sayNO; + case SBOL: + if (locinput == PL_regbol && PL_regprev == '\n') + break; + sayNO; + case GPOS: + if (locinput == PL_reg_ganch) + break; + sayNO; + case EOL: + if (PL_multiline) + goto meol; + else + goto seol; + case MEOL: + meol: + if ((nextchr || locinput < PL_regeol) && nextchr != '\n') + sayNO; + break; + case SEOL: + seol: + if ((nextchr || locinput < PL_regeol) && nextchr != '\n') + sayNO; + if (PL_regeol - locinput > 1) + sayNO; + break; + case EOS: + if (PL_regeol != locinput) + sayNO; + break; + case SANYUTF8: + if (nextchr & 0x80) { + locinput += PL_utf8skip[nextchr]; + if (locinput > PL_regeol) + sayNO; + nextchr = UCHARAT(locinput); + break; + } + if (!nextchr && locinput >= PL_regeol) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case SANY: + if (!nextchr && locinput >= PL_regeol) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case ANYUTF8: + if (nextchr & 0x80) { + locinput += PL_utf8skip[nextchr]; + if (locinput > PL_regeol) + sayNO; + nextchr = UCHARAT(locinput); + break; + } + if (!nextchr && locinput >= PL_regeol || nextchr == '\n') + sayNO; + nextchr = UCHARAT(++locinput); + break; + case REG_ANY: + if (!nextchr && locinput >= PL_regeol || nextchr == '\n') + sayNO; + nextchr = UCHARAT(++locinput); + break; + case EXACT: + s = (char *) OPERAND(scan); + ln = UCHARAT(s++); + /* Inline the first character, for speed. */ + if (UCHARAT(s) != nextchr) + sayNO; + if (PL_regeol - locinput < ln) + sayNO; + if (ln > 1 && memNE(s, locinput, ln)) + sayNO; + locinput += ln; + nextchr = UCHARAT(locinput); + break; + case EXACTFL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case EXACTF: + s = (char *) OPERAND(scan); + ln = UCHARAT(s++); + + if (UTF) { + char *l = locinput; + char *e = s + ln; + c1 = OP(scan) == EXACTF; + while (s < e) { + if (l >= PL_regeol) + sayNO; + if (utf8_to_uv((U8*)s, 0) != (c1 ? + toLOWER_utf8((U8*)l) : + toLOWER_LC_utf8((U8*)l))) + { + sayNO; + } + s += UTF8SKIP(s); + l += UTF8SKIP(l); } - } else { - STRLEN len; - char *little = SvPV(prog->float_substr, len); - last = rninstr(s, strend, little, little + len); + locinput = l; + nextchr = UCHARAT(locinput); + break; } - if (last == NULL) goto phooey; /* Should not happen! */ - dontbother = strend - last - 1; - } - if (minlen && (dontbother < minlen)) - dontbother = minlen - 1; - strend -= dontbother; - /* We don't know much -- general case. */ - do { - if (regtry(prog, s)) - goto got_it; - } while (s++ < strend); - } - - /* Failure. */ - goto phooey; - -got_it: - strend += dontbother; /* uncheat */ - prog->subbeg = strbeg; - prog->subend = strend; - RX_MATCH_TAINTED_SET(prog, reg_flags & RF_tainted); - /* make sure $`, $&, $', and $digit will work later */ - if (strbeg != prog->subbase) { /* second+ //g match. */ - if (!(flags & REXEC_COPY_STR)) { - if (prog->subbase) { - Safefree(prog->subbase); - prog->subbase = Nullch; + /* Inline the first character, for speed. */ + if (UCHARAT(s) != nextchr && + UCHARAT(s) != ((OP(scan) == EXACTF) + ? PL_fold : PL_fold_locale)[nextchr]) + sayNO; + if (PL_regeol - locinput < ln) + sayNO; + if (ln > 1 && (OP(scan) == EXACTF + ? ibcmp(s, locinput, ln) + : ibcmp_locale(s, locinput, ln))) + sayNO; + locinput += ln; + nextchr = UCHARAT(locinput); + break; + case ANYOFUTF8: + s = (char *) OPERAND(scan); + if (!REGINCLASSUTF8(scan, (U8*)locinput)) + sayNO; + if (locinput >= PL_regeol) + sayNO; + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + case ANYOF: + s = (char *) OPERAND(scan); + if (nextchr < 0) + nextchr = UCHARAT(locinput); + if (!REGINCLASS(s, nextchr)) + sayNO; + if (!nextchr && locinput >= PL_regeol) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case ALNUML: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case ALNUM: + if (!nextchr) + sayNO; + if (!(OP(scan) == ALNUM + ? isALNUM(nextchr) : isALNUM_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case ALNUMLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case ALNUMUTF8: + if (!nextchr) + sayNO; + if (nextchr & 0x80) { + if (!(OP(scan) == ALNUMUTF8 + ? swash_fetch(PL_utf8_alnum, (U8*)locinput) + : isALNUM_LC_utf8((U8*)locinput))) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (!(OP(scan) == ALNUMUTF8 + ? isALNUM(nextchr) : isALNUM_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NALNUML: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NALNUM: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (OP(scan) == NALNUM + ? isALNUM(nextchr) : isALNUM_LC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NALNUMLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NALNUMUTF8: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (nextchr & 0x80) { + if (OP(scan) == NALNUMUTF8 + ? swash_fetch(PL_utf8_alnum, (U8*)locinput) + : isALNUM_LC_utf8((U8*)locinput)) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (OP(scan) == NALNUMUTF8 + ? isALNUM(nextchr) : isALNUM_LC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case BOUNDL: + case NBOUNDL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case BOUND: + case NBOUND: + /* was last char in word? */ + ln = (locinput != PL_regbol) ? UCHARAT(locinput - 1) : PL_regprev; + if (OP(scan) == BOUND || OP(scan) == NBOUND) { + ln = isALNUM(ln); + n = isALNUM(nextchr); + } + else { + ln = isALNUM_LC(ln); + n = isALNUM_LC(nextchr); + } + if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL)) + sayNO; + break; + case BOUNDLUTF8: + case NBOUNDLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case BOUNDUTF8: + case NBOUNDUTF8: + /* was last char in word? */ + 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, (U8*)locinput); + } + else { + ln = isALNUM_LC_uni(ln); + n = isALNUM_LC_utf8((U8*)locinput); + } + if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8)) + sayNO; + break; + case SPACEL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case SPACE: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (!(OP(scan) == SPACE + ? isSPACE(nextchr) : isSPACE_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case SPACELUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case SPACEUTF8: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (nextchr & 0x80) { + if (!(OP(scan) == SPACEUTF8 + ? swash_fetch(PL_utf8_space,(U8*)locinput) + : isSPACE_LC_utf8((U8*)locinput))) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (!(OP(scan) == SPACEUTF8 + ? isSPACE(nextchr) : isSPACE_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NSPACEL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NSPACE: + if (!nextchr) + sayNO; + if (OP(scan) == SPACE + ? isSPACE(nextchr) : isSPACE_LC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NSPACELUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NSPACEUTF8: + if (!nextchr) + sayNO; + if (nextchr & 0x80) { + if (OP(scan) == NSPACEUTF8 + ? swash_fetch(PL_utf8_space,(U8*)locinput) + : isSPACE_LC_utf8((U8*)locinput)) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (OP(scan) == NSPACEUTF8 + ? isSPACE(nextchr) : isSPACE_LC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case DIGITL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case DIGIT: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (!(OP(scan) == DIGIT + ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case DIGITLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case DIGITUTF8: + if (!nextchr) + sayNO; + if (nextchr & 0x80) { + if (OP(scan) == NDIGITUTF8 + ? swash_fetch(PL_utf8_digit,(U8*)locinput) + : isDIGIT_LC_utf8((U8*)locinput)) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (!isDIGIT(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NDIGITL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NDIGIT: + if (!nextchr) + sayNO; + if (OP(scan) == DIGIT + ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NDIGITLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NDIGITUTF8: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (nextchr & 0x80) { + if (swash_fetch(PL_utf8_digit,(U8*)locinput)) + sayNO; + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (isDIGIT(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case ALNUMCL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case ALNUMC: + if (!nextchr) + sayNO; + if (!(OP(scan) == ALNUMC + ? isALNUMC(nextchr) : isALNUMC_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case ALNUMCLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case ALNUMCUTF8: + if (!nextchr) + sayNO; + if (nextchr & 0x80) { + if (!(OP(scan) == ALNUMCUTF8 + ? swash_fetch(PL_utf8_alnumc, (U8*)locinput) + : isALNUMC_LC_utf8((U8*)locinput))) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (!(OP(scan) == ALNUMCUTF8 + ? isALNUMC(nextchr) : isALNUMC_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NALNUMCL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NALNUMC: + if (!nextchr) + sayNO; + if (OP(scan) == ALNUMC + ? isALNUMC(nextchr) : isALNUMC_LC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NALNUMCLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NALNUMCUTF8: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (nextchr & 0x80) { + if (swash_fetch(PL_utf8_alnumc,(U8*)locinput)) + sayNO; + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (isALNUMC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case ALPHAL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case ALPHA: + if (!nextchr) + sayNO; + if (!(OP(scan) == ALPHA + ? isALPHA(nextchr) : isALPHA_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case ALPHALUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case ALPHAUTF8: + if (!nextchr) + sayNO; + if (nextchr & 0x80) { + if (!(OP(scan) == ALPHAUTF8 + ? swash_fetch(PL_utf8_alpha, (U8*)locinput) + : isALPHA_LC_utf8((U8*)locinput))) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (!(OP(scan) == ALPHAUTF8 + ? isALPHA(nextchr) : isALPHA_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NALPHAL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NALPHA: + if (!nextchr) + sayNO; + if (OP(scan) == ALPHA + ? isALPHA(nextchr) : isALPHA_LC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NALPHALUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NALPHAUTF8: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (nextchr & 0x80) { + if (swash_fetch(PL_utf8_alpha,(U8*)locinput)) + sayNO; + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; } - } - else { - I32 i = strend - startpos + (stringarg - strbeg); - s = savepvn(strbeg, i); - Safefree(prog->subbase); - prog->subbase = s; - prog->subbeg = prog->subbase; - prog->subend = prog->subbase + i; - s = prog->subbase + (stringarg - strbeg); - for (i = 0; i <= prog->nparens; i++) { - if (prog->endp[i]) { - prog->startp[i] = s + (prog->startp[i] - startpos); - prog->endp[i] = s + (prog->endp[i] - startpos); + if (isALPHA(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case ASCII: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (!isASCII(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NASCII: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (isASCII(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case CNTRLL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case CNTRL: + if (!nextchr) + sayNO; + if (!(OP(scan) == CNTRL + ? isCNTRL(nextchr) : isCNTRL_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case CNTRLLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case CNTRLUTF8: + if (!nextchr) + sayNO; + if (nextchr & 0x80) { + if (!(OP(scan) == CNTRLUTF8 + ? swash_fetch(PL_utf8_cntrl, (U8*)locinput) + : isCNTRL_LC_utf8((U8*)locinput))) + { + sayNO; } - } - } - } - /* Preserve the current value of $^R */ - if (oreplsv != GvSV(replgv)) { - sv_setsv(oreplsv, GvSV(replgv));/* So that when GvSV(replgv) is - restored, the value remains - the same. */ - } - return 1; - -phooey: - return 0; -} - -/* - - regtry - try match at specific point - */ -STATIC I32 /* 0 failure, 1 success */ -regtry(regexp *prog, char *startpos) -{ - dTHR; - register I32 i; - register char **sp; - register char **ep; - CHECKPOINT lastcp; - - if ((prog->reganch & ROPT_EVAL_SEEN) && !reg_eval_set) { - reg_eval_set = RS_init; - DEBUG_r(DEBUG_s( - PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n", stack_sp - stack_base); - )); - SAVEINT(cxstack[cxstack_ix].blk_oldsp); - cxstack[cxstack_ix].blk_oldsp = stack_sp - stack_base; - /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ - SAVETMPS; - /* Apparently this is not needed, judging by wantarray. */ - /* SAVEINT(cxstack[cxstack_ix].blk_gimme); - cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ - } - reginput = startpos; - regstartp = prog->startp; - regendp = prog->endp; - reglastparen = &prog->lastparen; - prog->lastparen = 0; - regsize = 0; - if (reg_start_tmpl <= prog->nparens) { - reg_start_tmpl = prog->nparens*3/2 + 3; - if(reg_start_tmp) - Renew(reg_start_tmp, reg_start_tmpl, char*); - else - New(22,reg_start_tmp, reg_start_tmpl, char*); - } - - sp = prog->startp; - ep = prog->endp; - regdata = prog->data; - if (prog->nparens) { - for (i = prog->nparens; i >= 0; i--) { - *sp++ = NULL; - *ep++ = NULL; - } - } - REGCP_SET; - if (regmatch(prog->program + 1) && reginput >= regtill) { - prog->startp[0] = startpos; - prog->endp[0] = reginput; - return 1; - } - REGCP_UNWIND; - return 0; -} - -/* - - regmatch - main matching routine - * - * Conceptually the strategy is simple: check to see whether the current - * node matches, call self recursively to see whether the rest matches, - * and then act accordingly. In practice we make some effort to avoid - * recursion, in particular by going through "ordinary" nodes (that don't - * need to know whether the rest of the match failed) by a loop instead of - * by recursion. - */ -/* [lwall] I've hoisted the register declarations to the outer block in order to - * maybe save a little bit of pushing and popping on the stack. It also takes - * advantage of machines that use a register save mask on subroutine entry. - */ -STATIC I32 /* 0 failure, 1 success */ -regmatch(regnode *prog) -{ - dTHR; - register regnode *scan; /* Current node. */ - regnode *next; /* Next node. */ - regnode *inner; /* Next node in internal branch. */ - 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 char *locinput = reginput; - register I32 c1, c2, paren; /* case fold search, parenth */ - int minmod = 0, sw = 0, logical = 0; -#ifdef DEBUGGING - regindent++; -#endif - - nextchr = UCHARAT(locinput); - scan = prog; - while (scan != NULL) { -#define sayNO_L (logical ? (logical = 0, sw = 0, goto cont) : sayNO) -#ifdef DEBUGGING -# define sayYES goto yes -# define sayNO goto no -# define saySAME(x) if (x) goto yes; else goto no -# define REPORT_CODE_OFF 24 -#else -# define sayYES return 1 -# define sayNO return 0 -# define saySAME(x) return x -#endif - DEBUG_r( { - SV *prop = sv_newmortal(); - int docolor = *colors[0]; - int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ - int l = (regeol - locinput > taill ? taill : regeol - locinput); - int pref_len = (locinput - bostr > (5 + taill) - l - ? (5 + taill) - l : locinput - bostr); - - if (l + pref_len < (5 + taill) && l < regeol - locinput) - l = ( regeol - locinput > (5 + taill) - pref_len - ? (5 + taill) - pref_len : regeol - locinput); - regprop(prop, scan); - PerlIO_printf(Perl_debug_log, - "%4i <%s%.*s%s%s%s%.*s%s>%*s|%*s%2d%s\n", - locinput - bostr, - colors[2], pref_len, locinput - pref_len, colors[3], - (docolor ? "" : "> <"), - colors[0], l, locinput, colors[1], - 15 - l - pref_len + 1, - "", - regindent*2, "", scan - regprogram, - SvPVX(prop)); - } ); - - next = scan + NEXT_OFF(scan); - if (next == scan) - next = NULL; - - switch (OP(scan)) { - case BOL: - if (locinput == regbol - ? regprev == '\n' - : (multiline && - (nextchr || locinput < regeol) && locinput[-1] == '\n') ) - { - /* regtill = regbol; */ + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); break; } - sayNO; - case MBOL: - if (locinput == regbol - ? regprev == '\n' - : ((nextchr || locinput < regeol) && locinput[-1] == '\n') ) - { + if (!(OP(scan) == CNTRLUTF8 + ? isCNTRL(nextchr) : isCNTRL_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NCNTRLL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NCNTRL: + if (!nextchr) + sayNO; + if (OP(scan) == CNTRL + ? isCNTRL(nextchr) : isCNTRL_LC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NCNTRLLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NCNTRLUTF8: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (nextchr & 0x80) { + if (swash_fetch(PL_utf8_cntrl,(U8*)locinput)) + sayNO; + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); break; } - sayNO; - case SBOL: - if (locinput == regbol && regprev == '\n') + if (isCNTRL(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case GRAPHL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case GRAPH: + if (!nextchr) + sayNO; + if (!(OP(scan) == GRAPH + ? isGRAPH(nextchr) : isGRAPH_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case GRAPHLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case GRAPHUTF8: + if (!nextchr) + sayNO; + if (nextchr & 0x80) { + if (!(OP(scan) == GRAPHUTF8 + ? swash_fetch(PL_utf8_graph, (U8*)locinput) + : isGRAPH_LC_utf8((U8*)locinput))) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); break; - sayNO; - case GPOS: - if (locinput == regbol) + } + if (!(OP(scan) == GRAPHUTF8 + ? isGRAPH(nextchr) : isGRAPH_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NGRAPHL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NGRAPH: + if (!nextchr) + sayNO; + if (OP(scan) == GRAPH + ? isGRAPH(nextchr) : isGRAPH_LC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NGRAPHLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NGRAPHUTF8: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (nextchr & 0x80) { + if (swash_fetch(PL_utf8_graph,(U8*)locinput)) + sayNO; + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); break; - sayNO; - case EOL: - if (multiline) - goto meol; - else - goto seol; - case MEOL: - meol: - if ((nextchr || locinput < regeol) && nextchr != '\n') + } + if (isGRAPH(nextchr)) sayNO; + nextchr = UCHARAT(++locinput); break; - case SEOL: - seol: - if ((nextchr || locinput < regeol) && nextchr != '\n') + case LOWERL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case LOWER: + if (!nextchr) sayNO; - if (regeol - locinput > 1) + if (!(OP(scan) == LOWER + ? isLOWER(nextchr) : isLOWER_LC(nextchr))) sayNO; + nextchr = UCHARAT(++locinput); break; - case SANY: - if (!nextchr && locinput >= regeol) + case LOWERLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case LOWERUTF8: + if (!nextchr) + sayNO; + if (nextchr & 0x80) { + if (!(OP(scan) == LOWERUTF8 + ? swash_fetch(PL_utf8_lower, (U8*)locinput) + : isLOWER_LC_utf8((U8*)locinput))) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (!(OP(scan) == LOWERUTF8 + ? isLOWER(nextchr) : isLOWER_LC(nextchr))) sayNO; nextchr = UCHARAT(++locinput); break; - case ANY: - if (!nextchr && locinput >= regeol || nextchr == '\n') + case NLOWERL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NLOWER: + if (!nextchr) + sayNO; + if (OP(scan) == LOWER + ? isLOWER(nextchr) : isLOWER_LC(nextchr)) sayNO; nextchr = UCHARAT(++locinput); break; - case EXACT: - s = (char *) OPERAND(scan); - ln = UCHARAT(s++); - /* Inline the first character, for speed. */ - if (UCHARAT(s) != nextchr) + case NLOWERLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NLOWERUTF8: + if (!nextchr && locinput >= PL_regeol) sayNO; - if (regeol - locinput < ln) + if (nextchr & 0x80) { + if (swash_fetch(PL_utf8_lower,(U8*)locinput)) + sayNO; + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (isLOWER(nextchr)) sayNO; - if (ln > 1 && memNE(s, locinput, ln)) + nextchr = UCHARAT(++locinput); + break; + case PRINTL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case PRINT: + if (!nextchr) sayNO; - locinput += ln; - nextchr = UCHARAT(locinput); + if (!(OP(scan) == PRINT + ? isPRINT(nextchr) : isPRINT_LC(nextchr))) + sayNO; + nextchr = UCHARAT(++locinput); break; - case EXACTFL: - reg_flags |= RF_tainted; + case PRINTLUTF8: + PL_reg_flags |= RF_tainted; /* FALL THROUGH */ - case EXACTF: - s = (char *) OPERAND(scan); - ln = UCHARAT(s++); - /* Inline the first character, for speed. */ - if (UCHARAT(s) != nextchr && - UCHARAT(s) != ((OP(scan) == EXACTF) - ? fold : fold_locale)[nextchr]) + case PRINTUTF8: + if (!nextchr) sayNO; - if (regeol - locinput < ln) + if (nextchr & 0x80) { + if (!(OP(scan) == PRINTUTF8 + ? swash_fetch(PL_utf8_print, (U8*)locinput) + : isPRINT_LC_utf8((U8*)locinput))) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (!(OP(scan) == PRINTUTF8 + ? isPRINT(nextchr) : isPRINT_LC(nextchr))) sayNO; - if (ln > 1 && (OP(scan) == EXACTF - ? ibcmp(s, locinput, ln) - : ibcmp_locale(s, locinput, ln))) + nextchr = UCHARAT(++locinput); + break; + case NPRINTL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NPRINT: + if (!nextchr) sayNO; - locinput += ln; - nextchr = UCHARAT(locinput); + if (OP(scan) == PRINT + ? isPRINT(nextchr) : isPRINT_LC(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); break; - case ANYOF: - s = (char *) OPERAND(scan); - if (nextchr < 0) + case NPRINTLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NPRINTUTF8: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (nextchr & 0x80) { + if (swash_fetch(PL_utf8_print,(U8*)locinput)) + sayNO; + locinput += PL_utf8skip[nextchr]; nextchr = UCHARAT(locinput); - if (!REGINCLASS(s, nextchr)) + break; + } + if (isPRINT(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case PUNCTL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case PUNCT: + if (!nextchr) sayNO; - if (!nextchr && locinput >= regeol) + if (!(OP(scan) == PUNCT + ? isPUNCT(nextchr) : isPUNCT_LC(nextchr))) sayNO; nextchr = UCHARAT(++locinput); break; - case ALNUML: - reg_flags |= RF_tainted; + case PUNCTLUTF8: + PL_reg_flags |= RF_tainted; /* FALL THROUGH */ - case ALNUM: + case PUNCTUTF8: if (!nextchr) sayNO; - if (!(OP(scan) == ALNUM - ? isALNUM(nextchr) : isALNUM_LC(nextchr))) + if (nextchr & 0x80) { + if (!(OP(scan) == PUNCTUTF8 + ? swash_fetch(PL_utf8_punct, (U8*)locinput) + : isPUNCT_LC_utf8((U8*)locinput))) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (!(OP(scan) == PUNCTUTF8 + ? isPUNCT(nextchr) : isPUNCT_LC(nextchr))) sayNO; nextchr = UCHARAT(++locinput); break; - case NALNUML: - reg_flags |= RF_tainted; + case NPUNCTL: + PL_reg_flags |= RF_tainted; /* FALL THROUGH */ - case NALNUM: - if (!nextchr && locinput >= regeol) + case NPUNCT: + if (!nextchr) sayNO; - if (OP(scan) == NALNUM - ? isALNUM(nextchr) : isALNUM_LC(nextchr)) + if (OP(scan) == PUNCT + ? isPUNCT(nextchr) : isPUNCT_LC(nextchr)) sayNO; nextchr = UCHARAT(++locinput); break; - case BOUNDL: - case NBOUNDL: - reg_flags |= RF_tainted; + case NPUNCTLUTF8: + PL_reg_flags |= RF_tainted; /* FALL THROUGH */ - case BOUND: - case NBOUND: - /* was last char in word? */ - ln = (locinput != regbol) ? UCHARAT(locinput - 1) : regprev; - if (OP(scan) == BOUND || OP(scan) == NBOUND) { - ln = isALNUM(ln); - n = isALNUM(nextchr); - } - else { - ln = isALNUM_LC(ln); - n = isALNUM_LC(nextchr); + case NPUNCTUTF8: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (nextchr & 0x80) { + if (swash_fetch(PL_utf8_punct,(U8*)locinput)) + sayNO; + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; } - if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL)) + if (isPUNCT(nextchr)) sayNO; + nextchr = UCHARAT(++locinput); break; - case SPACEL: - reg_flags |= RF_tainted; + case UPPERL: + PL_reg_flags |= RF_tainted; /* FALL THROUGH */ - case SPACE: - if (!nextchr && locinput >= regeol) + case UPPER: + if (!nextchr) sayNO; - if (!(OP(scan) == SPACE - ? isSPACE(nextchr) : isSPACE_LC(nextchr))) + if (!(OP(scan) == UPPER + ? isUPPER(nextchr) : isUPPER_LC(nextchr))) sayNO; nextchr = UCHARAT(++locinput); break; - case NSPACEL: - reg_flags |= RF_tainted; + case UPPERLUTF8: + PL_reg_flags |= RF_tainted; /* FALL THROUGH */ - case NSPACE: + case UPPERUTF8: if (!nextchr) sayNO; - if (OP(scan) == SPACE - ? isSPACE(nextchr) : isSPACE_LC(nextchr)) + if (nextchr & 0x80) { + if (!(OP(scan) == UPPERUTF8 + ? swash_fetch(PL_utf8_upper, (U8*)locinput) + : isUPPER_LC_utf8((U8*)locinput))) + { + sayNO; + } + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (!(OP(scan) == UPPERUTF8 + ? isUPPER(nextchr) : isUPPER_LC(nextchr))) sayNO; nextchr = UCHARAT(++locinput); break; - case DIGIT: - if (!isDIGIT(nextchr)) + case NUPPERL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NUPPER: + if (!nextchr) + sayNO; + if (OP(scan) == UPPER + ? isUPPER(nextchr) : isUPPER_LC(nextchr)) sayNO; nextchr = UCHARAT(++locinput); break; - case NDIGIT: - if (!nextchr && locinput >= regeol) + case NUPPERLUTF8: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NUPPERUTF8: + if (!nextchr && locinput >= PL_regeol) sayNO; - if (isDIGIT(nextchr)) + if (nextchr & 0x80) { + if (swash_fetch(PL_utf8_upper,(U8*)locinput)) + sayNO; + locinput += PL_utf8skip[nextchr]; + nextchr = UCHARAT(locinput); + break; + } + if (isUPPER(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case XDIGIT: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (!isXDIGIT(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case NXDIGIT: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (isXDIGIT(nextchr)) sayNO; nextchr = UCHARAT(++locinput); break; + case CLUMP: + if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput)) + sayNO; + locinput += PL_utf8skip[nextchr]; + while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput)) + locinput += UTF8SKIP(locinput); + if (locinput > PL_regeol) + sayNO; + nextchr = UCHARAT(locinput); + break; case REFFL: - reg_flags |= RF_tainted; + PL_reg_flags |= RF_tainted; /* FALL THROUGH */ case REF: case REFF: n = ARG(scan); /* which paren pair */ - s = regstartp[n]; - if (*reglastparen < n || !s) + ln = PL_regstartp[n]; + if (*PL_reglastparen < n || ln == -1) sayNO; /* Do not match unless seen CLOSEn. */ - if (s == regendp[n]) + if (ln == PL_regendp[n]) + break; + + s = PL_bostr + ln; + if (UTF && OP(scan) != REF) { /* REF can do byte comparison */ + char *l = locinput; + char *e = PL_bostr + PL_regendp[n]; + /* + * Note that we can't do the "other character" lookup trick as + * in the 8-bit case (no pun intended) because in Unicode we + * have to map both upper and title case to lower case. + */ + if (OP(scan) == REFF) { + while (s < e) { + if (l >= PL_regeol) + sayNO; + if (toLOWER_utf8((U8*)s) != toLOWER_utf8((U8*)l)) + sayNO; + s += UTF8SKIP(s); + l += UTF8SKIP(l); + } + } + else { + while (s < e) { + if (l >= PL_regeol) + sayNO; + if (toLOWER_LC_utf8((U8*)s) != toLOWER_LC_utf8((U8*)l)) + sayNO; + s += UTF8SKIP(s); + l += UTF8SKIP(l); + } + } + locinput = l; + nextchr = UCHARAT(locinput); break; + } + /* Inline the first character, for speed. */ if (UCHARAT(s) != nextchr && (OP(scan) == REF || (UCHARAT(s) != ((OP(scan) == REFF - ? fold : fold_locale)[nextchr])))) + ? PL_fold : PL_fold_locale)[nextchr])))) sayNO; - ln = regendp[n] - s; - if (locinput + ln > regeol) + ln = PL_regendp[n] - ln; + if (locinput + ln > PL_regeol) sayNO; if (ln > 1 && (OP(scan) == REF ? memNE(s, locinput, ln) @@ -990,47 +3225,126 @@ regmatch(regnode *prog) case EVAL: { dSP; - OP_4tree *oop = op; - COP *ocurcop = curcop; - SV **ocurpad = curpad; + OP_4tree *oop = PL_op; + COP *ocurcop = PL_curcop; + SV **ocurpad = PL_curpad; SV *ret; n = ARG(scan); - op = (OP_4tree*)regdata->data[n]; - DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", op) ); - curpad = AvARRAY((AV*)regdata->data[n + 1]); + 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 + 2]); + PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; - CALLRUNOPS(); /* Scalar context. */ + CALLRUNOPS(aTHX); /* 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(aTHX_ 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); - } else - sv_setsv(save_scalar(replgv), ret); - op = oop; - curpad = ocurpad; - curcop = ocurcop; + logical = 0; + } + else + sv_setsv(save_scalar(PL_replgv), ret); break; } case OPEN: n = ARG(scan); /* which paren pair */ - reg_start_tmp[n] = locinput; - if (n > regsize) - regsize = n; + PL_reg_start_tmp[n] = locinput; + if (n > PL_regsize) + PL_regsize = n; break; case CLOSE: n = ARG(scan); /* which paren pair */ - regstartp[n] = reg_start_tmp[n]; - regendp[n] = locinput; - if (n > *reglastparen) - *reglastparen = n; + PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr; + PL_regendp[n] = locinput - PL_bostr; + if (n > *PL_reglastparen) + *PL_reglastparen = n; break; case GROUPP: n = ARG(scan); /* which paren pair */ - sw = (*reglastparen >= n && regendp[n] != NULL); + sw = (*PL_reglastparen >= n && PL_regendp[n] != -1); break; case IFTHEN: if (sw) @@ -1042,17 +3356,17 @@ regmatch(regnode *prog) } break; case LOGICAL: - logical = 1; + logical = scan->flags; break; case CURLYX: { CURCUR cc; - CHECKPOINT cp = savestack_ix; + CHECKPOINT cp = PL_savestack_ix; if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ next += ARG(next); - cc.oldcc = regcc; - regcc = &cc; - cc.parenfloor = *reglastparen; + cc.oldcc = PL_regcc; + PL_regcc = &cc; + cc.parenfloor = *PL_reglastparen; cc.cur = -1; cc.min = ARG1(scan); cc.max = ARG2(scan); @@ -1060,10 +3374,10 @@ regmatch(regnode *prog) cc.next = next; cc.minmod = minmod; cc.lastloc = 0; - reginput = locinput; + PL_reginput = locinput; n = regmatch(PREVOPER(next)); /* start on the WHILEM */ regcpblow(cp); - regcc = cc.oldcc; + PL_regcc = cc.oldcc; saySAME(n); } /* NOT REACHED */ @@ -1078,16 +3392,16 @@ regmatch(regnode *prog) */ CHECKPOINT cp, lastcp; - CURCUR* cc = regcc; + CURCUR* cc = PL_regcc; char *lastloc = cc->lastloc; /* Detection of 0-len. */ n = cc->cur + 1; /* how many we know we matched */ - reginput = locinput; + PL_reginput = locinput; DEBUG_r( PerlIO_printf(Perl_debug_log, "%*s %ld out of %ld..%ld cc=%lx\n", - REPORT_CODE_OFF+regindent*2, "", + REPORT_CODE_OFF+PL_regindent*2, "", (long)n, (long)cc->min, (long)cc->max, (long)cc) ); @@ -1095,18 +3409,22 @@ regmatch(regnode *prog) /* If degenerate scan matches "", assume scan done. */ if (locinput == cc->lastloc && n >= cc->min) { - regcc = cc->oldcc; - ln = regcc->cur; + PL_regcc = cc->oldcc; + ln = PL_regcc->cur; DEBUG_r( - PerlIO_printf(Perl_debug_log, "%*s empty match detected, try continuation...\n", REPORT_CODE_OFF+regindent*2, "") + PerlIO_printf(Perl_debug_log, + "%*s empty match detected, try continuation...\n", + REPORT_CODE_OFF+PL_regindent*2, "") ); if (regmatch(cc->next)) sayYES; DEBUG_r( - PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") + PerlIO_printf(Perl_debug_log, + "%*s failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") ); - regcc->cur = ln; - regcc = cc; + PL_regcc->cur = ln; + PL_regcc = cc; sayNO; } @@ -1120,7 +3438,9 @@ regmatch(regnode *prog) cc->cur = n - 1; cc->lastloc = lastloc; DEBUG_r( - PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") + PerlIO_printf(Perl_debug_log, + "%*s failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") ); sayNO; } @@ -1128,8 +3448,8 @@ regmatch(regnode *prog) /* Prefer next over scan for minimal matching. */ if (cc->minmod) { - regcc = cc->oldcc; - ln = regcc->cur; + PL_regcc = cc->oldcc; + ln = PL_regcc->cur; cp = regcppush(cc->parenfloor); REGCP_SET; if (regmatch(cc->next)) { @@ -1138,23 +3458,27 @@ regmatch(regnode *prog) } REGCP_UNWIND; regcppop(); - regcc->cur = ln; - regcc = cc; + PL_regcc->cur = ln; + PL_regcc = cc; if (n >= cc->max) { /* Maximum greed exceeded? */ - if (dowarn && n >= REG_INFTY - && !(reg_flags & RF_warned)) { - reg_flags |= RF_warned; - warn("count exceeded %d", REG_INFTY - 1); + if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY + && !(PL_reg_flags & RF_warned)) { + PL_reg_flags |= RF_warned; + Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded", + "Complex regular subexpression recursion", + REG_INFTY - 1); } sayNO; } DEBUG_r( - PerlIO_printf(Perl_debug_log, "%*s trying longer...\n", REPORT_CODE_OFF+regindent*2, "") + PerlIO_printf(Perl_debug_log, + "%*s trying longer...\n", + REPORT_CODE_OFF+PL_regindent*2, "") ); /* Try scanning more and see if it helps. */ - reginput = locinput; + PL_reginput = locinput; cc->cur = n; cc->lastloc = locinput; cp = regcppush(cc->parenfloor); @@ -1164,7 +3488,9 @@ regmatch(regnode *prog) sayYES; } DEBUG_r( - PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") + PerlIO_printf(Perl_debug_log, + "%*s failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") ); REGCP_UNWIND; regcppop(); @@ -1186,26 +3512,32 @@ regmatch(regnode *prog) } REGCP_UNWIND; regcppop(); /* Restore some previous $s? */ - reginput = locinput; + PL_reginput = locinput; DEBUG_r( - PerlIO_printf(Perl_debug_log, "%*s failed, try continuation...\n", REPORT_CODE_OFF+regindent*2, "") + PerlIO_printf(Perl_debug_log, + "%*s failed, try continuation...\n", + REPORT_CODE_OFF+PL_regindent*2, "") ); } - if (dowarn && n >= REG_INFTY && !(reg_flags & RF_warned)) { - reg_flags |= RF_warned; - warn("count exceeded %d", REG_INFTY - 1); + if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY + && !(PL_reg_flags & RF_warned)) { + PL_reg_flags |= RF_warned; + Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded", + "Complex regular subexpression recursion", + REG_INFTY - 1); } /* Failed deeper matches of scan, so see if this one works. */ - regcc = cc->oldcc; - ln = regcc->cur; + PL_regcc = cc->oldcc; + ln = PL_regcc->cur; if (regmatch(cc->next)) sayYES; DEBUG_r( - PerlIO_printf(Perl_debug_log, "%*s failed...\n", REPORT_CODE_OFF+regindent*2, "") + PerlIO_printf(Perl_debug_log, "%*s failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") ); - regcc->cur = ln; - regcc = cc; + PL_regcc->cur = ln; + PL_regcc = cc; cc->cur = n - 1; cc->lastloc = lastloc; sayNO; @@ -1226,17 +3558,17 @@ regmatch(regnode *prog) if (OP(next) != c1) /* No choice. */ next = inner; /* Avoid recursion. */ else { - int lastparen = *reglastparen; + int lastparen = *PL_reglastparen; REGCP_SET; do { - reginput = locinput; + PL_reginput = locinput; if (regmatch(inner)) sayYES; REGCP_UNWIND; - for (n = *reglastparen; n > lastparen; n--) - regendp[n] = 0; - *reglastparen = n; + for (n = *PL_reglastparen; n > lastparen; n--) + PL_regendp[n] = -1; + *PL_reglastparen = n; scan = next; /*SUPPRESS 560*/ if (n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next))) @@ -1268,15 +3600,15 @@ regmatch(regnode *prog) n = ARG2(scan); /* max to match */ paren = scan->flags; if (paren) { - if (paren > regsize) - regsize = paren; - if (paren > *reglastparen) - *reglastparen = paren; + if (paren > PL_regsize) + PL_regsize = paren; + if (paren > *PL_reglastparen) + *PL_reglastparen = paren; } scan = NEXTOPER(scan) + NODE_STEP_REGNODE; if (paren) scan += NEXT_OFF(scan); /* Skip former OPEN. */ - reginput = locinput; + PL_reginput = locinput; if (minmod) { minmod = 0; if (ln && regrepeat_hard(scan, ln, &l) < ln) @@ -1287,46 +3619,50 @@ regmatch(regnode *prog) future execution. */ && !(paren && ln == 0)) ln = n; - locinput = reginput; - if (regkind[(U8)OP(next)] == EXACT) { + locinput = PL_reginput; + 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; - } else + } + else c1 = c2 = -1000; REGCP_SET; /* This may be improved if l == 0. */ while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */ /* If it could work, try it. */ if (c1 == -1000 || - UCHARAT(reginput) == c1 || - UCHARAT(reginput) == c2) + UCHARAT(PL_reginput) == c1 || + UCHARAT(PL_reginput) == c2) { if (paren) { if (n) { - regstartp[paren] = reginput - l; - regendp[paren] = reginput; - } else - regendp[paren] = NULL; + PL_regstartp[paren] = + HOPc(PL_reginput, -l) - PL_bostr; + PL_regendp[paren] = PL_reginput - PL_bostr; + } + else + PL_regendp[paren] = -1; } if (regmatch(next)) sayYES; REGCP_UNWIND; } /* Couldn't or didn't -- move forward. */ - reginput = locinput; + PL_reginput = locinput; if (regrepeat_hard(scan, 1, &l)) { ln++; - locinput = reginput; + locinput = PL_reginput; } else sayNO; } - } else { + } + else { n = regrepeat_hard(scan, n, &l); if (n != 0 && l == 0 /* In fact, this is tricky. If paren, then the @@ -1334,47 +3670,53 @@ regmatch(regnode *prog) future execution. */ && !(paren && ln == 0)) ln = n; - locinput = reginput; + locinput = PL_reginput; DEBUG_r( - PerlIO_printf(Perl_debug_log, "%*s matched %ld times, len=%ld...\n", REPORT_CODE_OFF+regindent*2, "", n, l) + PerlIO_printf(Perl_debug_log, + "%*s matched %ld times, len=%ld...\n", + 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; - } else + } + else c1 = c2 = -1000; } REGCP_SET; while (n >= ln) { /* If it could work, try it. */ if (c1 == -1000 || - UCHARAT(reginput) == c1 || - UCHARAT(reginput) == c2) - { - DEBUG_r( - PerlIO_printf(Perl_debug_log, "%*s trying tail with n=%ld...\n", REPORT_CODE_OFF+regindent*2, "", n) - ); - if (paren) { - if (n) { - regstartp[paren] = reginput - l; - regendp[paren] = reginput; - } else - regendp[paren] = NULL; + UCHARAT(PL_reginput) == c1 || + UCHARAT(PL_reginput) == c2) + { + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s trying tail with n=%ld...\n", + REPORT_CODE_OFF+PL_regindent*2, "", n) + ); + if (paren) { + if (n) { + PL_regstartp[paren] = HOPc(PL_reginput, -l) - PL_bostr; + PL_regendp[paren] = PL_reginput - PL_bostr; } - if (regmatch(next)) - sayYES; - REGCP_UNWIND; + else + PL_regendp[paren] = -1; } + if (regmatch(next)) + sayYES; + REGCP_UNWIND; + } /* Couldn't or didn't -- back up. */ n--; - locinput -= l; - reginput = locinput; + locinput = HOPc(locinput, -l); + PL_reginput = locinput; } } sayNO; @@ -1382,10 +3724,10 @@ regmatch(regnode *prog) } case CURLYN: paren = scan->flags; /* Which paren to set */ - if (paren > regsize) - regsize = paren; - if (paren > *reglastparen) - *reglastparen = paren; + if (paren > PL_regsize) + PL_regsize = paren; + if (paren > *PL_reglastparen) + *PL_reglastparen = paren; ln = ARG1(scan); /* min to match */ n = ARG2(scan); /* max to match */ scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); @@ -1412,70 +3754,119 @@ 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; } else c1 = c2 = -1000; - reginput = locinput; + PL_reginput = locinput; if (minmod) { CHECKPOINT lastcp; minmod = 0; if (ln && regrepeat(scan, ln) < ln) 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_bostr; + PL_regendp[paren] = locinput - PL_bostr; + } + else + PL_regendp[paren] = -1; + } + 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 || - UCHARAT(reginput) == c1 || - UCHARAT(reginput) == c2) + UCHARAT(PL_reginput) == c1 || + UCHARAT(PL_reginput) == c2) { if (paren) { if (n) { - regstartp[paren] = reginput - 1; - regendp[paren] = reginput; - } else - regendp[paren] = NULL; + PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr; + PL_regendp[paren] = PL_reginput - PL_bostr; + } + else + PL_regendp[paren] = -1; } if (regmatch(next)) sayYES; REGCP_UNWIND; } /* Couldn't or didn't -- move forward. */ - reginput = locinput + ln; + PL_reginput = locinput; if (regrepeat(scan, 1)) { ln++; - reginput = locinput + ln; - } else + locinput = PL_reginput; + } + else sayNO; } } else { CHECKPOINT lastcp; n = regrepeat(scan, n); - if (ln < n && regkind[(U8)OP(next)] == EOL && - (!multiline || OP(next) == SEOL)) + locinput = PL_reginput; + if (ln < n && PL_regkind[(U8)OP(next)] == EOL && + (!PL_multiline || OP(next) == SEOL)) ln = n; /* why back off? */ REGCP_SET; if (paren) { while (n >= ln) { /* If it could work, try it. */ if (c1 == -1000 || - UCHARAT(reginput) == c1 || - UCHARAT(reginput) == c2) + UCHARAT(PL_reginput) == c1 || + UCHARAT(PL_reginput) == c2) { if (paren && n) { if (n) { - regstartp[paren] = reginput - 1; - regendp[paren] = reginput; - } else - regendp[paren] = NULL; + PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr; + PL_regendp[paren] = PL_reginput - PL_bostr; + } + else + PL_regendp[paren] = -1; } if (regmatch(next)) sayYES; @@ -1483,14 +3874,15 @@ regmatch(regnode *prog) } /* Couldn't or didn't -- back up. */ n--; - reginput = locinput + n; + PL_reginput = locinput = HOPc(locinput, -1); } - } else { + } + else { while (n >= ln) { /* If it could work, try it. */ if (c1 == -1000 || - UCHARAT(reginput) == c1 || - UCHARAT(reginput) == c2) + UCHARAT(PL_reginput) == c1 || + UCHARAT(PL_reginput) == c2) { if (regmatch(next)) sayYES; @@ -1498,30 +3890,100 @@ regmatch(regnode *prog) } /* Couldn't or didn't -- back up. */ n--; - reginput = locinput + n; + PL_reginput = locinput = HOPc(locinput, -1); } } } sayNO; break; - case SUCCEED: case END: - reginput = locinput; /* put where regtry can find it */ + 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 */ + case SUCCEED: + PL_reginput = locinput; /* put where regtry can find it */ sayYES; /* Success! */ case SUSPEND: n = 1; + PL_reginput = locinput; goto do_ifmatch; case UNLESSM: n = 0; - if (locinput < bostr + scan->flags) - goto say_yes; + 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; + } + } + else + PL_reginput = locinput; goto do_ifmatch; case IFMATCH: n = 1; - if (locinput < bostr + scan->flags) - goto say_no; + 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; + } + } + else + PL_reginput = locinput; + do_ifmatch: - reginput = locinput - scan->flags; inner = NEXTOPER(NEXTOPER(scan)); if (regmatch(inner) != n) { say_no: @@ -1529,7 +3991,8 @@ regmatch(regnode *prog) logical = 0; sw = 0; goto do_longjump; - } else + } + else sayNO; } say_yes: @@ -1538,7 +4001,7 @@ regmatch(regnode *prog) sw = 1; } if (OP(scan) == SUSPEND) { - locinput = reginput; + locinput = PL_reginput; nextchr = UCHARAT(locinput); } /* FALL THROUGH. */ @@ -1551,7 +4014,7 @@ regmatch(regnode *prog) default: PerlIO_printf(PerlIO_stderr(), "%lx %d\n", (unsigned long)scan, OP(scan)); - FAIL("regexp memory corruption"); + Perl_croak(aTHX_ "regexp memory corruption"); } scan = next; } @@ -1560,19 +4023,19 @@ regmatch(regnode *prog) * We get here only if there's trouble -- normally "case END" is * the terminating point. */ - FAIL("corrupted regexp pointers"); + Perl_croak(aTHX_ "corrupted regexp pointers"); /*NOTREACHED*/ sayNO; yes: #ifdef DEBUGGING - regindent--; + PL_regindent--; #endif return 1; no: #ifdef DEBUGGING - regindent--; + PL_regindent--; #endif return 0; } @@ -1583,28 +4046,44 @@ no: /* * [This routine now assumes that it will only match on things of length 1. * That was true before, but now we assume scan - reginput is the count, - * rather than incrementing count on every character.] + * rather than incrementing count on every character. [Er, except utf8.]] */ STATIC I32 -regrepeat(regnode *p, I32 max) +S_regrepeat(pTHX_ regnode *p, I32 max) { + dTHR; register char *scan; register char *opnd; register I32 c; - register char *loceol = regeol; + register char *loceol = PL_regeol; + register I32 hardcount = 0; - scan = reginput; + scan = PL_reginput; if (max != REG_INFTY && max < loceol - scan) loceol = scan + max; opnd = (char *) OPERAND(p); switch (OP(p)) { - case ANY: + case REG_ANY: while (scan < loceol && *scan != '\n') scan++; break; case SANY: scan = loceol; break; + case ANYUTF8: + loceol = PL_regeol; + while (scan < loceol && *scan != '\n') { + scan += UTF8SKIP(scan); + hardcount++; + } + break; + case SANYUTF8: + loceol = PL_regeol; + while (scan < loceol) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; case EXACT: /* length of string is 1 */ c = UCHARAT(++opnd); while (scan < loceol && UCHARAT(scan) == c) @@ -1613,16 +4092,23 @@ 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 */ - reg_flags |= RF_tainted; + 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: + loceol = PL_regeol; + while (scan < loceol && REGINCLASSUTF8(p, (U8*)scan)) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; case ANYOF: while (scan < loceol && REGINCLASS(opnd, *scan)) scan++; @@ -1631,52 +4117,131 @@ regrepeat(regnode *p, I32 max) while (scan < loceol && isALNUM(*scan)) scan++; break; + case ALNUMUTF8: + loceol = PL_regeol; + while (scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan)) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; case ALNUML: - reg_flags |= RF_tainted; + PL_reg_flags |= RF_tainted; while (scan < loceol && isALNUM_LC(*scan)) scan++; break; + case ALNUMLUTF8: + PL_reg_flags |= RF_tainted; + loceol = PL_regeol; + while (scan < loceol && isALNUM_LC_utf8((U8*)scan)) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; + break; case NALNUM: while (scan < loceol && !isALNUM(*scan)) scan++; break; + case NALNUMUTF8: + loceol = PL_regeol; + while (scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan)) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; case NALNUML: - reg_flags |= RF_tainted; + PL_reg_flags |= RF_tainted; while (scan < loceol && !isALNUM_LC(*scan)) scan++; break; + case NALNUMLUTF8: + PL_reg_flags |= RF_tainted; + loceol = PL_regeol; + while (scan < loceol && !isALNUM_LC_utf8((U8*)scan)) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; case SPACE: while (scan < loceol && isSPACE(*scan)) scan++; break; + case SPACEUTF8: + loceol = PL_regeol; + while (scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; case SPACEL: - reg_flags |= RF_tainted; + PL_reg_flags |= RF_tainted; while (scan < loceol && isSPACE_LC(*scan)) scan++; break; + case SPACELUTF8: + PL_reg_flags |= RF_tainted; + loceol = PL_regeol; + while (scan < loceol && (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; case NSPACE: while (scan < loceol && !isSPACE(*scan)) scan++; break; + case NSPACEUTF8: + loceol = PL_regeol; + while (scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; case NSPACEL: - reg_flags |= RF_tainted; + PL_reg_flags |= RF_tainted; while (scan < loceol && !isSPACE_LC(*scan)) scan++; break; + case NSPACELUTF8: + PL_reg_flags |= RF_tainted; + loceol = PL_regeol; + while (scan < loceol && !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; case DIGIT: while (scan < loceol && isDIGIT(*scan)) scan++; break; + case DIGITUTF8: + loceol = PL_regeol; + while (scan < loceol && swash_fetch(PL_utf8_digit,(U8*)scan)) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; + break; case NDIGIT: while (scan < loceol && !isDIGIT(*scan)) scan++; break; + case NDIGITUTF8: + loceol = PL_regeol; + while (scan < loceol && !swash_fetch(PL_utf8_digit,(U8*)scan)) { + scan += UTF8SKIP(scan); + hardcount++; + } + break; default: /* Called on something of 0 width. */ break; /* So match right here or not at all. */ } - c = scan - reginput; - reginput = scan; + if (hardcount) + c = hardcount; + else + c = scan - PL_reginput; + PL_reginput = scan; DEBUG_r( { @@ -1698,64 +4263,108 @@ regrepeat(regnode *p, I32 max) */ STATIC I32 -regrepeat_hard(regnode *p, I32 max, I32 *lp) +S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp) { + dTHR; register char *scan; register char *start; - register char *loceol = regeol; - I32 l = -1; - - start = reginput; - while (reginput < loceol && (scan = reginput, regmatch(p))) { - if (l == -1) { - *lp = l = reginput - start; - if (max != REG_INFTY && l*max < loceol - scan) - loceol = scan + l*max; - if (l == 0) { - return max; + register char *loceol = PL_regeol; + I32 l = 0; + I32 count = 0, res = 1; + + if (!max) + return 0; + + start = PL_reginput; + if (UTF) { + while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) { + if (!count++) { + l = 0; + while (start < PL_reginput) { + l++; + start += UTF8SKIP(start); + } + *lp = l; + if (l == 0) + return max; } + if (count == max) + return count; } } - if (reginput < loceol) - reginput = scan; - else - scan = reginput; + else { + 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) + loceol = scan + l*max; + if (l == 0) + return max; + } + } + } + if (!res) + PL_reginput = scan; - return (scan - start)/l; + return count; } /* - - regclass - determine if a character falls into a character class + - reginclass - determine if a character falls into a character class */ STATIC bool -reginclass(register char *p, register I32 c) +S_reginclass(pTHX_ register char *p, register I32 c) { - char flags = *p; + dTHR; + char flags = ANYOF_FLAGS(p); bool match = FALSE; c &= 0xFF; - if (ANYOF_TEST(p, c)) + if (ANYOF_BITMAP_TEST(p, c)) match = TRUE; else if (flags & ANYOF_FOLD) { I32 cf; if (flags & ANYOF_LOCALE) { - reg_flags |= RF_tainted; - cf = fold_locale[c]; + PL_reg_flags |= RF_tainted; + cf = PL_fold_locale[c]; } else - cf = fold[c]; - if (ANYOF_TEST(p, cf)) + cf = PL_fold[c]; + if (ANYOF_BITMAP_TEST(p, cf)) match = TRUE; } - if (!match && (flags & ANYOF_ISA)) { - reg_flags |= RF_tainted; - - if (((flags & ANYOF_ALNUML) && isALNUM_LC(c)) || - ((flags & ANYOF_NALNUML) && !isALNUM_LC(c)) || - ((flags & ANYOF_SPACEL) && isSPACE_LC(c)) || - ((flags & ANYOF_NSPACEL) && !isSPACE_LC(c))) + if (!match && (flags & ANYOF_CLASS)) { + PL_reg_flags |= RF_tainted; + if ( + (ANYOF_CLASS_TEST(p, ANYOF_ALNUM) && isALNUM_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NALNUM) && !isALNUM_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_SPACE) && isSPACE_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NSPACE) && !isSPACE_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_DIGIT) && isDIGIT_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NDIGIT) && !isDIGIT_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_ALNUMC) && isALNUMC_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NALNUMC) && !isALNUMC_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_ALPHA) && isALPHA_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NALPHA) && !isALPHA_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_ASCII) && isASCII(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NASCII) && !isASCII(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_CNTRL) && isCNTRL_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NCNTRL) && !isCNTRL_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_GRAPH) && isGRAPH_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NGRAPH) && !isGRAPH_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_LOWER) && isLOWER_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NLOWER) && !isLOWER_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_PRINT) && isPRINT_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NPRINT) && !isPRINT_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_PUNCT) && isPUNCT_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NPUNCT) && !isPUNCT_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_UPPER) && isUPPER_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NUPPER) && !isUPPER_LC(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_XDIGIT) && isXDIGIT(c)) || + (ANYOF_CLASS_TEST(p, ANYOF_NXDIGIT) && !isXDIGIT(c)) + ) /* How's that for a conditional? */ { match = TRUE; } @@ -1764,5 +4373,102 @@ reginclass(register char *p, register I32 c) return (flags & ANYOF_INVERT) ? !match : match; } +STATIC bool +S_reginclassutf8(pTHX_ regnode *f, U8 *p) +{ + dTHR; + char flags = ARG1(f); + bool match = FALSE; + SV *sv = (SV*)PL_regdata->data[ARG2(f)]; + + if (swash_fetch(sv, p)) + match = TRUE; + else if (flags & ANYOF_FOLD) { + I32 cf; + U8 tmpbuf[10]; + if (flags & ANYOF_LOCALE) { + PL_reg_flags |= RF_tainted; + uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p)); + } + else + uv_to_utf8(tmpbuf, toLOWER_utf8(p)); + if (swash_fetch(sv, tmpbuf)) + match = TRUE; + } + + /* UTF8 combined with ANYOF_CLASS is ill-defined. */ + + return (flags & ANYOF_INVERT) ? !match : match; +} + +STATIC U8 * +S_reghop(pTHX_ U8 *s, I32 off) +{ + dTHR; + if (off >= 0) { + while (off-- && s < (U8*)PL_regeol) + s += UTF8SKIP(s); + } + else { + while (off++) { + if (s > (U8*)PL_bostr) { + s--; + if (*s & 0x80) { + while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80) + s--; + } /* XXX could check well-formedness here */ + } + } + } + return s; +} + +STATIC U8 * +S_reghopmaybe(pTHX_ U8* s, I32 off) +{ + dTHR; + if (off >= 0) { + while (off-- && s < (U8*)PL_regeol) + s += UTF8SKIP(s); + if (off >= 0) + return 0; + } + else { + while (off++) { + if (s > (U8*)PL_bostr) { + s--; + if (*s & 0x80) { + while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80) + s--; + } /* XXX could check well-formedness here */ + } + else + break; + } + if (off <= 0) + return 0; + } + return s; +} + +#ifdef PERL_OBJECT +#define NO_XSLOCKS +#include "XSUB.h" +#endif +static void +restore_pos(pTHXo_ void *arg) +{ + dTHR; + if (PL_reg_eval_set) { + if (PL_reg_oldsaved) { + PL_reg_re->subbeg = PL_reg_oldsaved; + PL_reg_re->sublen = PL_reg_oldsavedlen; + RX_MATCH_COPIED_on(PL_reg_re); + } + PL_reg_magic->mg_len = PL_reg_oldpos; + PL_reg_eval_set = 0; + PL_curpm = PL_reg_oldcurpm; + } +}