X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=48797065b3cc95f0b1eebfcebad51b42eabf3c05;hb=f63ceb1c48810cc43268c83de36dc71834702c3b;hp=f431cccefc9e6a8c857f981cadcc49d43c3bbce5;hpb=79072805bf63abe5b5978b5928ab00d360ea3e7f;p=p5sagit%2Fp5-mst-13.2.git diff --git a/regexec.c b/regexec.c index f431ccc..4879706 100644 --- a/regexec.c +++ b/regexec.c @@ -1,3 +1,10 @@ +/* regexec.c + */ + +/* + * "One Ring to rule them all, One Ring to find them..." + */ + /* NOTE: this is derived from Henry Spencer's regexp code, and should not * confused with the original package (see point 3 below). Thanks, Henry! */ @@ -7,34 +14,35 @@ * blame Henry for some of the lack of readability. */ -/* $RCSfile: regexec.c,v $$Revision: 4.1 $$Date: 92/08/07 18:26:32 $ - * - * $Log: regexec.c,v $ - * Revision 4.1 92/08/07 18:26:32 lwall - * - * Revision 4.0.1.4 92/06/08 15:25:50 lwall - * patch20: pattern modifiers i and g didn't interact right - * patch20: in some cases $` and $' didn't get set by match - * patch20: /x{0}/ was wrongly interpreted as /x{0,}/ - * - * Revision 4.0.1.3 91/11/05 18:23:55 lwall - * patch11: prepared for ctype implementations that don't define isascii() - * patch11: initial .* in pattern had dependency on value of $* - * - * Revision 4.0.1.2 91/06/07 11:50:33 lwall - * patch4: new copyright notice - * patch4: // wouldn't use previous pattern if it started with a null character - * - * Revision 4.0.1.1 91/04/12 09:07:39 lwall - * patch1: regexec only allocated space for 9 subexpresssions - * - * Revision 4.0 91/03/20 01:39:16 lwall - * 4.0 baseline. - * - */ +/* The names of the functions have been changed from regcomp and + * regexec to pregcomp and pregexec in order to avoid conflicts + * 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 */ +# ifndef 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 +/* *These* symbols are masked to allow static link. */ +# define Perl_pregexec my_pregexec +# define Perl_reginitcolors my_reginitcolors +#endif + /*SUPPRESS 112*/ /* - * regcomp and regexec -- regsub and regerror are not used in perl + * pregcomp and pregexec -- regsub and regerror are not used in perl * * Copyright (c) 1986 by University of Toronto. * Written by Henry Spencer. Not derived from licensed software. @@ -55,7 +63,7 @@ * **** Alterations to Henry's code are... **** - **** Copyright (c) 1991, Larry Wall + **** Copyright (c) 1991-1998, 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. @@ -66,848 +74,2725 @@ */ #include "EXTERN.h" #include "perl.h" + #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 */ + #ifndef STATIC #define STATIC static #endif -#ifdef DEBUGGING -I32 regnarrate = 0; -#endif - -/* - * regexec and friends - */ +#ifndef PERL_OBJECT +typedef I32 CHECKPOINT; /* * Forwards. */ -STATIC I32 regtry(); -STATIC I32 regmatch(); -STATIC I32 regrepeat(); -/* - - regexec - match a regexp against a string - */ -I32 -regexec(prog, stringarg, strend, strbeg, minend, screamer, safebase) -register regexp *prog; -char *stringarg; -register char *strend; /* pointer to null at end of string */ -char *strbeg; /* real beginning of string */ -I32 minend; /* end of match must be at least minend after stringarg */ -SV *screamer; -I32 safebase; /* no need to remember string in subbase */ -{ - register char *s; - register I32 i; - register char *c; - register char *string = stringarg; - register I32 tmp; - I32 minlen = 0; /* must match at least this many chars */ - I32 dontbother = 0; /* how many characters not to try at end */ - - /* Be paranoid... */ - if (prog == NULL || string == NULL) { - fatal("NULL regexp parameter"); - return(0); - } - - if (string == strbeg) /* is ^ valid at stringarg? */ - regprev = '\n'; - else { - regprev = stringarg[-1]; - if (!multiline && regprev == '\n') - regprev = '\0'; /* force ^ to NOT match */ - } - regprecomp = prog->precomp; - /* Check validity of program. */ - if (UCHARAT(prog->program) != MAGIC) { - FAIL("corrupted regexp program"); - } - - if (prog->do_folding) { - i = strend - string; - New(1101,c,i+1,char); - Copy(string, c, i+1, char); - string = c; - strend = string + i; - for (s = string; s < strend; s++) - if (isUPPER(*s)) - *s = tolower(*s); - } - - /* If there is a "must appear" string, look for it. */ - s = string; - if (prog->regmust != Nullsv && - (!(prog->reganch & ROPT_ANCH) - || (multiline && prog->regback >= 0)) ) { - if (stringarg == strbeg && screamer) { - if (screamfirst[BmRARE(prog->regmust)] >= 0) - s = screaminstr(screamer,prog->regmust); - else - s = Nullch; - } -#ifndef lint - else - s = fbm_instr((unsigned char*)s, (unsigned char*)strend, - prog->regmust); +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)); + +static bool reginclass _((char *p, I32 c)); +static bool reginclassutf8 _((regnode *f, U8* p)); +static CHECKPOINT regcppush _((I32 parenfloor)); +static char * regcppop _((void)); +static char * regcp_set_to _((I32 ss)); +static void cache_re _((regexp *prog)); +static void restore_pos _((void *arg)); #endif - if (!s) { - ++BmUSEFUL(prog->regmust); /* hooray */ - goto phooey; /* not present */ - } - else if (prog->regback >= 0) { - s -= prog->regback; - if (s < string) - s = string; - minlen = prog->regback + SvCUR(prog->regmust); - } - else if (--BmUSEFUL(prog->regmust) < 0) { /* boo */ - sv_free(prog->regmust); - prog->regmust = Nullsv; /* disable regmust */ - s = string; - } - else { - s = string; - minlen = SvCUR(prog->regmust); - } - } - /* Mark beginning of line for ^ . */ - regbol = string; +#define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c)) +#define REGINCLASSUTF8(f,p) (ARG1(f) ? reginclassutf8(f,p) : swash_fetch((SV*)PL_regdata->data[ARG2(f)],p)) - /* Mark end of line for $ (and such) */ - regeol = strend; +#define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) +#define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) - /* see how far we have to get to not match where we matched before */ - regtill = string+minend; +#ifndef PERL_OBJECT +static U8 * reghop _((U8 *pos, I32 off)); +static U8 * reghopmaybe _((U8 *pos, I32 off)); +#endif +#define reghop_c(pos,off) ((char*)reghop((U8*)pos, off)) +#define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off)) +#define HOP(pos,off) (UTF ? reghop((U8*)pos, off) : (U8*)(pos + off)) +#define HOPMAYBE(pos,off) (UTF ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off)) +#define HOPc(pos,off) ((char*)HOP(pos,off)) +#define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off)) - /* Allocate our backreference arrays */ - if ( regmyp_size < prog->nparens + 1 ) { - /* Allocate or enlarge the arrays */ - regmyp_size = prog->nparens + 1; - if ( regmyp_size < 10 ) regmyp_size = 10; /* minimum */ - if ( regmystartp ) { - /* reallocate larger */ - Renew(regmystartp,regmyp_size,char*); - Renew(regmyendp, regmyp_size,char*); - } - else { - /* Initial allocation */ - New(1102,regmystartp,regmyp_size,char*); - New(1102,regmyendp, regmyp_size,char*); - } - - } +STATIC CHECKPOINT +regcppush(I32 parenfloor) +{ + dTHR; + int retval = PL_savestack_ix; + int i = (PL_regsize - parenfloor) * 4; + int p; - /* Simplest case: anchored match need be tried only once. */ - /* [unless multiline is set] */ - if (prog->reganch & ROPT_ANCH) { - if (regtry(prog, string)) - goto got_it; - else if (multiline || (prog->reganch & ROPT_IMPLICIT)) { - if (minlen) - dontbother = minlen - 1; - strend -= dontbother; - /* for multiline we only have to try after newlines */ - if (s > string) - s--; - while (s < strend) { - if (*s++ == '\n') { - if (s < strend && regtry(prog, s)) - goto got_it; - } - } - } - goto phooey; - } - - /* Messy cases: unanchored match. */ - if (prog->regstart) { - if (prog->reganch & ROPT_SKIP) { /* we have /x+whatever/ */ - /* it must be a one character string */ - i = SvPV(prog->regstart)[0]; - while (s < strend) { - if (*s == i) { - if (regtry(prog, s)) - goto got_it; - s++; - while (s < strend && *s == i) - s++; - } - s++; - } - } - else if (SvPOK(prog->regstart) == 3) { - /* We know what string it must start with. */ -#ifndef lint - while ((s = fbm_instr((unsigned char*)s, - (unsigned char*)strend, prog->regstart)) != NULL) -#else - while (s = Nullch) -#endif - { - if (regtry(prog, s)) - goto got_it; - s++; - } - } - else { - c = SvPV(prog->regstart); - while ((s = ninstr(s, strend, - c, c + SvCUR(prog->regstart) )) != NULL) { - if (regtry(prog, s)) - goto got_it; - s++; - } - } - goto phooey; - } - /*SUPPRESS 560*/ - if (c = prog->regstclass) { - I32 doevery = (prog->reganch & ROPT_SKIP) == 0; - - if (minlen) - dontbother = minlen - 1; - strend -= dontbother; /* don't bother with what can't match */ - tmp = 1; - /* We know what class it must start with. */ - switch (OP(c)) { - case ANYOF: - c = OPERAND(c); - while (s < strend) { - i = UCHARAT(s); - if (!(c[i >> 3] & (1 << (i&7)))) { - if (tmp && regtry(prog, s)) - goto got_it; - else - tmp = doevery; - } - else - tmp = 1; - s++; - } - break; - case BOUND: - if (minlen) - dontbother++,strend--; - if (s != string) { - i = s[-1]; - tmp = isALNUM(i); - } - else - tmp = isALNUM(regprev); /* assume not alphanumeric */ - while (s < strend) { - i = *s; - if (tmp != isALNUM(i)) { - tmp = !tmp; - if (regtry(prog, s)) - goto got_it; - } - s++; - } - if ((minlen || tmp) && regtry(prog,s)) - goto got_it; - break; - case NBOUND: - if (minlen) - dontbother++,strend--; - if (s != string) { - i = s[-1]; - tmp = isALNUM(i); - } - else - tmp = isALNUM(regprev); /* assume not alphanumeric */ - while (s < strend) { - i = *s; - if (tmp != isALNUM(i)) - tmp = !tmp; - else if (regtry(prog, s)) - goto got_it; - s++; - } - if ((minlen || !tmp) && regtry(prog,s)) - goto got_it; - break; - case ALNUM: - while (s < strend) { - i = *s; - if (isALNUM(i)) { - if (tmp && regtry(prog, s)) - goto got_it; - else - tmp = doevery; - } - else - tmp = 1; - s++; - } - break; - case NALNUM: - while (s < strend) { - i = *s; - if (!isALNUM(i)) { - if (tmp && regtry(prog, s)) - goto got_it; - else - tmp = doevery; - } - else - tmp = 1; - s++; - } - break; - case SPACE: - while (s < strend) { - if (isSPACE(*s)) { - if (tmp && regtry(prog, s)) - goto got_it; - else - tmp = doevery; - } - else - tmp = 1; - s++; - } - break; - case NSPACE: - while (s < strend) { - if (!isSPACE(*s)) { - if (tmp && regtry(prog, s)) - goto got_it; - else - tmp = doevery; - } - else - tmp = 1; - s++; - } - break; - 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 NDIGIT: - while (s < strend) { - if (!isDIGIT(*s)) { - if (tmp && regtry(prog, s)) - goto got_it; - else - tmp = doevery; - } - else - tmp = 1; - s++; - } - break; - } - } - else { - if (minlen) - dontbother = minlen - 1; - strend -= dontbother; - /* We don't know much -- general case. */ - do { - if (regtry(prog, s)) - goto got_it; - } while (s++ < strend); - } + SSCHECK(i + 5); + for (p = PL_regsize; p > parenfloor; p--) { + SSPUSHPTR(PL_regendp[p]); + SSPUSHPTR(PL_regstartp[p]); + SSPUSHPTR(PL_reg_start_tmp[p]); + SSPUSHINT(p); + } + SSPUSHINT(PL_regsize); + SSPUSHINT(*PL_reglastparen); + SSPUSHPTR(PL_reginput); + SSPUSHINT(i + 3); + SSPUSHINT(SAVEt_REGCONTEXT); + return retval; +} - /* Failure. */ - goto phooey; +/* 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", \ + PL_savestack_ix)); lastcp = PL_savestack_ix - got_it: - prog->subbeg = strbeg; - prog->subend = strend; - if ((!safebase && (prog->nparens || sawampersand)) || prog->do_folding){ - strend += dontbother; /* uncheat */ - if (safebase) /* no need for $digit later */ - s = strbeg; - else if (strbeg != prog->subbase) { - i = strend - string + (stringarg - strbeg); - s = nsavestr(strbeg,i); /* so $digit will work later */ - if (prog->subbase) - Safefree(prog->subbase); - prog->subbeg = prog->subbase = s; - prog->subend = s+i; - } - else { - i = strend - string + (stringarg - strbeg); - prog->subbeg = s = prog->subbase; - prog->subend = s+i; - } - s += (stringarg - strbeg); - for (i = 0; i <= prog->nparens; i++) { - if (prog->endp[i]) { - prog->startp[i] = s + (prog->startp[i] - string); - prog->endp[i] = s + (prog->endp[i] - string); - } - } - if (prog->do_folding) - Safefree(string); - } - return(1); +# 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) - phooey: - if (prog->do_folding) - Safefree(string); - return(0); +STATIC char * +regcppop(void) +{ + dTHR; + I32 i = SSPOPINT; + U32 paren = 0; + char *input; + char *tmps; + assert(i == SAVEt_REGCONTEXT); + i = SSPOPINT; + input = (char *) SSPOPPTR; + *PL_reglastparen = SSPOPINT; + PL_regsize = SSPOPINT; + for (i -= 3; i > 0; i -= 4) { + paren = (U32)SSPOPINT; + PL_reg_start_tmp[paren] = (char *) SSPOPPTR; + PL_regstartp[paren] = (char *) SSPOPPTR; + tmps = (char*)SSPOPPTR; + if (paren <= *PL_reglastparen) + PL_regendp[paren] = tmps; + DEBUG_r( + PerlIO_printf(Perl_debug_log, + " restoring \\%d to %d(%d)..%d%s\n", + paren, PL_regstartp[paren] - PL_regbol, + PL_reg_start_tmp[paren] - PL_regbol, + PL_regendp[paren] - PL_regbol, + (paren > *PL_reglastparen ? "(no)" : "")); + ); + } + DEBUG_r( + if (*PL_reglastparen + 1 <= PL_regnpar) { + PerlIO_printf(Perl_debug_log, + " restoring \\%d..\\%d to undef\n", + *PL_reglastparen + 1, PL_regnpar); + } + ); + for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) { + if (paren > PL_regsize) + PL_regstartp[paren] = Nullch; + PL_regendp[paren] = Nullch; + } + return input; } -/* - - regtry - try match at specific point - */ -static I32 /* 0 failure, 1 success */ -regtry(prog, string) -regexp *prog; -char *string; +STATIC char * +regcp_set_to(I32 ss) { - register I32 i; - register char **sp; - register char **ep; - - reginput = string; - regstartp = prog->startp; - regendp = prog->endp; - reglastparen = &prog->lastparen; - prog->lastparen = 0; - - sp = prog->startp; - ep = prog->endp; - if (prog->nparens) { - for (i = prog->nparens; i >= 0; i--) { - *sp++ = NULL; - *ep++ = NULL; - } - } - if (regmatch(prog->program + 1) && reginput >= regtill) { - prog->startp[0] = string; - prog->endp[0] = reginput; - return(1); - } else - return(0); + 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) + /* - - 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. + * pregexec and friends */ -/* [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. + +/* + - pregexec - match a regexp against a string */ -static I32 /* 0 failure, 1 success */ -regmatch(prog) -char *prog; +I32 +pregexec(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. */ +/* nosave: For optimizations. */ +{ + return + regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, + nosave ? 0 : REXEC_COPY_STR); +} + +STATIC void +cache_re(regexp *prog) { - register char *scan; /* Current node. */ - char *next; /* Next node. */ - register I32 nextchar; - register I32 n; /* no or next */ - register I32 ln; /* len or last */ - register char *s; /* operand or save */ - register char *locinput = reginput; - - nextchar = *locinput; - scan = prog; + dTHR; + PL_regprecomp = prog->precomp; /* Needed for FAIL. */ #ifdef DEBUGGING - if (scan != NULL && regnarrate) - fprintf(stderr, "%s(\n", regprop(scan)); + PL_regprogram = prog->program; #endif - while (scan != NULL) { + PL_regnpar = prog->nparens; + PL_regdata = prog->data; + PL_reg_re = prog; +} + +STATIC void +restore_pos(void *arg) +{ + dTHR; + if (PL_reg_eval_set) { + PL_reg_magic->mg_len = PL_reg_oldpos; + PL_reg_eval_set = 0; + PL_curpm = PL_reg_oldcurpm; + } +} + + +/* + - regexec_flags - match a regexp against a string + */ +I32 +regexec_flags(register regexp *prog, char *stringarg, register char *strend, + char *strbeg, I32 minend, SV *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; + register I32 tmp; + I32 minlen; /* must match at least this many chars */ + 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. */ /* CC */ + I32 end_shift = 0; /* Same for the end. */ /* CC */ + I32 scream_pos = -1; /* Internal iterator of scream. */ + char *scream_olds; + SV* oreplsv = GvSV(PL_replgv); + + cc.cur = 0; + cc.oldcc = 0; + PL_regcc = &cc; + + cache_re(prog); #ifdef DEBUGGING - if (regnarrate) - fprintf(stderr, "%s...\n", regprop(scan)); + PL_regnarrate = PL_debug & 512; #endif -#ifdef REGALIGN - next = scan + NEXT(scan); - if (next == scan) - next = NULL; -#else - next = regnext(scan); -#endif + /* Be paranoid... */ + if (prog == NULL || startpos == NULL) { + croak("NULL regexp parameter"); + return 0; + } - switch (OP(scan)) { - case BOL: - if (locinput == regbol ? regprev == '\n' : - ((nextchar || locinput < regeol) && - locinput[-1] == '\n') ) - { - /* regtill = regbol; */ - break; - } - return(0); - case EOL: - if ((nextchar || locinput < regeol) && nextchar != '\n') - return(0); - if (!multiline && regeol - locinput > 1) - return 0; - /* regtill = regbol; */ - break; - case ANY: - if ((nextchar == '\0' && locinput >= regeol) || - nextchar == '\n') - return(0); - nextchar = *++locinput; - break; - case EXACTLY: - s = OPERAND(scan); - ln = *s++; - /* Inline the first character, for speed. */ - if (*s != nextchar) - return(0); - if (regeol - locinput < ln) - return 0; - if (ln > 1 && bcmp(s, locinput, ln) != 0) - return(0); - locinput += ln; - nextchar = *locinput; - break; - case ANYOF: - s = OPERAND(scan); - if (nextchar < 0) - nextchar = UCHARAT(locinput); - if (s[nextchar >> 3] & (1 << (nextchar&7))) - return(0); - if (!nextchar && locinput >= regeol) - return 0; - nextchar = *++locinput; - break; - case ALNUM: - if (!nextchar) - return(0); - if (!isALNUM(nextchar)) - return(0); - nextchar = *++locinput; - break; - case NALNUM: - if (!nextchar && locinput >= regeol) - return(0); - if (isALNUM(nextchar)) - return(0); - nextchar = *++locinput; - break; - case NBOUND: - case BOUND: - if (locinput == regbol) /* was last char in word? */ - ln = isALNUM(regprev); - else - ln = isALNUM(locinput[-1]); - n = isALNUM(nextchar); /* is next char in word? */ - if ((ln == n) == (OP(scan) == BOUND)) - return(0); - break; - case SPACE: - if (!nextchar && locinput >= regeol) - return(0); - if (!isSPACE(nextchar)) - return(0); - nextchar = *++locinput; - break; - case NSPACE: - if (!nextchar) - return(0); - if (isSPACE(nextchar)) - return(0); - nextchar = *++locinput; - break; - case DIGIT: - if (!isDIGIT(nextchar)) - return(0); - nextchar = *++locinput; - break; - case NDIGIT: - if (!nextchar && locinput >= regeol) - return(0); - if (isDIGIT(nextchar)) - return(0); - nextchar = *++locinput; - break; - case REF: - n = ARG1(scan); /* which paren pair */ - s = regmystartp[n]; - if (!s) - return(0); - if (!regmyendp[n]) - return(0); - if (s == regmyendp[n]) - break; - /* Inline the first character, for speed. */ - if (*s != nextchar) - return(0); - ln = regmyendp[n] - s; - if (locinput + ln > regeol) - return 0; - if (ln > 1 && bcmp(s, locinput, ln) != 0) - return(0); - locinput += ln; - nextchar = *locinput; - break; - - case NOTHING: - break; - case BACK: - break; - case OPEN: - n = ARG1(scan); /* which paren pair */ - reginput = locinput; - - regmystartp[n] = locinput; /* for REF */ - if (regmatch(next)) { - /* - * Don't set startp if some later - * invocation of the same parentheses - * already has. - */ - if (regstartp[n] == NULL) - regstartp[n] = locinput; - return(1); - } else - return(0); - /* NOTREACHED */ - case CLOSE: { - n = ARG1(scan); /* which paren pair */ - reginput = locinput; - - regmyendp[n] = locinput; /* for REF */ - if (regmatch(next)) { - /* - * Don't set endp if some later - * invocation of the same parentheses - * already has. - */ - if (regendp[n] == NULL) { - regendp[n] = locinput; - if (n > *reglastparen) - *reglastparen = n; - } - return(1); - } else - return(0); - } - /*NOTREACHED*/ - case BRANCH: { - if (OP(next) != BRANCH) /* No choice. */ - next = NEXTOPER(scan); /* Avoid recursion. */ - else { - do { - reginput = locinput; - if (regmatch(NEXTOPER(scan))) - return(1); -#ifdef REGALIGN - /*SUPPRESS 560*/ - if (n = NEXT(scan)) - scan += n; - else - scan = NULL; -#else - scan = regnext(scan); -#endif - } while (scan != NULL && OP(scan) == BRANCH); - return(0); - /* NOTREACHED */ - } - } - break; -#ifdef NOTYET - case MINCURLY: - ln = ARG1(scan); /* min to match */ - n = -ARG2(scan); /* max to match */ - scan = NEXTOPER(scan) + 4; - goto repeat; -#endif - case CURLY: - ln = ARG1(scan); /* min to match */ - n = ARG2(scan); /* max to match */ - scan = NEXTOPER(scan) + 4; - goto repeat; - case STAR: - ln = 0; - n = 32767; - scan = NEXTOPER(scan); - goto repeat; - case PLUS: - /* - * Lookahead to avoid useless match attempts - * when we know what character comes next. - */ - ln = 1; - n = 32767; - scan = NEXTOPER(scan); - repeat: - if (OP(next) == EXACTLY) - nextchar = *(OPERAND(next)+1); - else - nextchar = -1000; - reginput = locinput; - if (n < 0) { - n = -n; - while (n >= ln) { - /* If it could work, try it. */ - if (nextchar == -1000 || - *reginput == nextchar) - if (regmatch(next)) - return(1); - /* Couldn't or didn't -- back up. */ - ln++; - reginput = locinput + ln; - } - } - else { - n = regrepeat(scan, n); - if (!multiline && OP(next) == EOL && ln < n) - ln = n; /* why back off? */ - while (n >= ln) { - /* If it could work, try it. */ - if (nextchar == -1000 || - *reginput == nextchar) - if (regmatch(next)) - return(1); - /* Couldn't or didn't -- back up. */ - n--; - reginput = locinput + n; - } - } - return(0); - case END: - reginput = locinput; /* put where regtry can find it */ - return(1); /* Success! */ - default: - printf("%x %d\n",scan,scan[1]); - FAIL("regexp memory corruption"); - } + minlen = prog->minlen; + if (strend - startpos < minlen) goto phooey; - scan = next; - } + if (startpos == strbeg) /* is ^ valid at stringarg? */ + PL_regprev = '\n'; + else { + PL_regprev = (U32)stringarg[-1]; + if (!PL_multiline && PL_regprev == '\n') + PL_regprev = '\0'; /* force ^ to NOT match */ + } - /* - * We get here only if there's trouble -- normally "case END" is - * the terminating point. - */ - FAIL("corrupted regexp pointers"); - /*NOTREACHED*/ -#ifdef lint - return 0; -#endif -} + /* Check validity of program. */ + if (UCHARAT(prog->program) != REG_MAGIC) { + FAIL("corrupted regexp program"); + } -/* - - regrepeat - repeatedly match something simple, report how many - */ -/* - * [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.] - */ -static I32 -regrepeat(p, max) -char *p; -I32 max; -{ - register char *scan; - register char *opnd; - register I32 c; - register char *loceol = regeol; - - scan = reginput; - if (max != 32767 && max < loceol - scan) - loceol = scan + max; - opnd = OPERAND(p); - switch (OP(p)) { - case ANY: - while (scan < loceol && *scan != '\n') - scan++; - break; - case EXACTLY: /* length of string is 1 */ - opnd++; - while (scan < loceol && *opnd == *scan) - scan++; - break; - case ANYOF: - c = UCHARAT(scan); - while (scan < loceol && !(opnd[c >> 3] & (1 << (c & 7)))) { - scan++; - c = UCHARAT(scan); - } - break; - case ALNUM: - while (scan < loceol && isALNUM(*scan)) - scan++; - break; - case NALNUM: - while (scan < loceol && !isALNUM(*scan)) - scan++; - break; - case SPACE: - while (scan < loceol && isSPACE(*scan)) - scan++; - break; - case NSPACE: - while (scan < loceol && !isSPACE(*scan)) - scan++; - break; - case DIGIT: - while (scan < loceol && isDIGIT(*scan)) - scan++; - break; - case NDIGIT: - while (scan < loceol && !isDIGIT(*scan)) - scan++; - break; - default: /* Oh dear. Called inappropriately. */ - FAIL("internal regexp foulup"); - /* NOTREACHED */ + PL_reg_flags = 0; + PL_reg_eval_set = 0; + + if (prog->reganch & ROPT_UTF8) + PL_reg_flags |= RF_utf8; + + /* Mark beginning of line for ^ and lookbehind. */ + PL_regbol = startpos; + PL_bostr = strbeg; + PL_reg_sv = sv; + + /* Mark end of line for $ (and such) */ + PL_regeol = strend; + + /* see how far we have to get to not match where we matched before */ + PL_regtill = startpos+minend; + + /* We start without call_cc context. */ + PL_reg_call_cc = 0; + + /* If there is a "must appear" string, look for it. */ + s = startpos; + if (!(flags & REXEC_CHECKED) + && prog->check_substr != Nullsv && + !(prog->reganch & ROPT_ANCH_GPOS) && + (!(prog->reganch & (ROPT_ANCH_BOL | ROPT_ANCH_MBOL)) + || (PL_multiline && prog->check_substr == prog->anchored_substr)) ) + { + char *t; + start_shift = prog->check_offset_min; /* okay to underestimate on CC */ + /* Should be nonnegative! */ + end_shift = minlen - start_shift - CHR_SVLEN(prog->check_substr); + if (flags & REXEC_SCREAM) { + if (PL_screamfirst[BmRARE(prog->check_substr)] >= 0) + s = screaminstr(sv, 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, 0); + if (!s) { + ++BmUSEFUL(prog->check_substr); /* hooray */ + goto phooey; /* not present */ + } + else if (s - stringarg > prog->check_offset_max && + (UTF + ? ((t = reghopmaybe_c(s, -(prog->check_offset_max))) && t >= stringarg) + : (t = s - prog->check_offset_max) != 0 + ) + ) + { + ++BmUSEFUL(prog->check_substr); /* hooray/2 */ + s = t; } + else if (!(prog->reganch & ROPT_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; + } - c = scan - reginput; - reginput = scan; + 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, PL_colors[1], + (strend - startpos > 60 ? "..." : "")) + ); - return(c); -} + if (prog->reganch & ROPT_GPOS_SEEN) { + MAGIC *mg; + int pos = 0; -/* - - regnext - dig the "next" pointer out of a node - * - * [Note, when REGALIGN is defined there are two places in regmatch() - * that bypass this code for speed.] - */ -char * -regnext(p) -register char *p; -{ - register I32 offset; + if (SvTYPE(sv) >= SVt_PVMG && SvMAGIC(sv) + && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0) + pos = mg->mg_len; + PL_reg_ganch = startpos + pos; + } - if (p == ®dummy) - return(NULL); + /* Simplest case: anchored match need be tried only once. */ + /* [unless only anchor is BOL and multiline is set] */ + if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { + if (regtry(prog, startpos)) + goto got_it; + else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT) + || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */ + { + if (minlen) + dontbother = minlen - 1; + strend = HOPc(strend, -dontbother); + /* for multiline we only have to try after newlines */ + if (s > startpos) + s--; + while (s < strend) { + if (*s++ == '\n') { /* don't need PL_utf8skip here */ + if (s < strend && regtry(prog, s)) + goto got_it; + } + } + } + goto phooey; + } else if (prog->reganch & ROPT_ANCH_GPOS) { + if (regtry(prog, PL_reg_ganch)) + goto got_it; + goto phooey; + } - offset = NEXT(p); - if (offset == 0) - return(NULL); + /* Messy cases: unanchored match. */ + if (prog->anchored_substr && prog->reganch & ROPT_SKIP) { + /* we have /x+whatever/ */ + /* it must be a one character string */ + char ch = SvPVX(prog->anchored_substr)[0]; + 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++; + } + } + } + /*SUPPRESS 560*/ + else if (prog->anchored_substr != Nullsv + || (prog->float_substr != Nullsv + && prog->float_max_offset < strend - s)) { + SV *must = prog->anchored_substr + ? prog->anchored_substr : prog->float_substr; + I32 back_max = + prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset; + I32 back_min = + prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset; + I32 delta = back_max - back_min; + char *last = HOPc(strend, 0-(CHR_SVLEN(must) + back_min)); /* Cannot start after this */ + char *last1; /* Last position checked before */ -#ifdef REGALIGN - return(p+offset); -#else - if (OP(p) == BACK) - return(p-offset); + if (s > PL_bostr) + last1 = HOPc(s, -1); else - return(p+offset); -#endif + 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 = HOPc(strend, -dontbother); + while ( (s <= last) && + ((flags & REXEC_SCREAM) + ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg, + end_shift, &scream_pos, 0)) + : (s = fbm_instr((unsigned char*)HOP(s, back_min), + (unsigned char*)strend, must, 0))) ) { + if (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 = HOPc(s, -back_min); + s = t; + } + 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) { + I32 doevery = (prog->reganch & ROPT_SKIP) == 0; + char *cc; + + if (minlen) + dontbother = minlen - 1; + 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: + cc = (char *) OPERAND(c); + while (s < strend) { + if (REGINCLASS(cc, *s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case BOUNDL: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case BOUND: + 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))) { + tmp = !tmp; + if (regtry(prog, s)) + goto got_it; + } + s++; + } + 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: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case NBOUND: + 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))) + tmp = !tmp; + else if (regtry(prog, s)) + goto got_it; + s++; + } + 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)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + 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: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isALNUM_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + 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)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + 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: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isALNUM_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + 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)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + 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: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (isSPACE_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + 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)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NSPACEUTF8: + 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 NSPACEL: + PL_reg_flags |= RF_tainted; + while (s < strend) { + if (!isSPACE_LC(*s)) { + if (tmp && regtry(prog, s)) + goto got_it; + else + tmp = doevery; + } + else + tmp = 1; + s++; + } + break; + case NSPACELUTF8: + 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 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 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; + } + } + else { + dontbother = 0; + if (prog->float_substr != Nullsv) { /* Trim the end. */ + char *last; + I32 oldpos = scream_pos; + + 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 (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: + prog->subbeg = strbeg; + prog->subend = PL_regeol; /* strend may have been modified */ + RX_MATCH_TAINTED_set(prog, PL_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; + } + } + else { + I32 i = PL_regeol - 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 (PL_reg_eval_set) { + /* Preserve the current value of $^R */ + if (oreplsv != GvSV(PL_replgv)) + sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is + restored, the value remains + the same. */ + restore_pos(0); + } + + return 1; + +phooey: + if (PL_reg_eval_set) + restore_pos(0); + return 0; +} + +/* + - 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) && !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; + prog->subbeg = PL_bostr; + prog->subend = PL_regeol; /* strend may have been modified */ + } + prog->startp[0] = startpos; + PL_reginput = startpos; + PL_regstartp = prog->startp; + PL_regendp = prog->endp; + PL_reglastparen = &prog->lastparen; + prog->lastparen = 0; + PL_regsize = 0; + DEBUG_r(PL_reg_starttry = startpos); + if (PL_reg_start_tmpl <= prog->nparens) { + PL_reg_start_tmpl = prog->nparens*3/2 + 3; + if(PL_reg_start_tmp) + 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 = NULL; + *++ep = NULL; + } + } + REGCP_SET; + if (regmatch(prog->program + 1)) { + prog->endp[0] = PL_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 = 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); + 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; + 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); + } + locinput = l; + nextchr = UCHARAT(locinput); + break; + } + + /* 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 DIGIT: + if (!isDIGIT(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + case DIGITUTF8: + 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 NDIGIT: + if (!nextchr && locinput >= PL_regeol) + sayNO; + if (isDIGIT(nextchr)) + sayNO; + nextchr = UCHARAT(++locinput); + break; + 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 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: + PL_reg_flags |= RF_tainted; + /* FALL THROUGH */ + case REF: + case REFF: + n = ARG(scan); /* which paren pair */ + s = PL_regstartp[n]; + if (*PL_reglastparen < n || !s) + sayNO; /* Do not match unless seen CLOSEn. */ + if (s == PL_regendp[n]) + break; + + if (UTF && OP(scan) != REF) { /* REF can do byte comparison */ + char *l = locinput; + char *e = 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 + ? PL_fold : PL_fold_locale)[nextchr])))) + sayNO; + ln = PL_regendp[n] - s; + if (locinput + ln > PL_regeol) + sayNO; + if (ln > 1 && (OP(scan) == REF + ? memNE(s, locinput, ln) + : (OP(scan) == REFF + ? ibcmp(s, locinput, ln) + : ibcmp_locale(s, locinput, ln)))) + sayNO; + locinput += ln; + nextchr = UCHARAT(locinput); + break; + + case NOTHING: + case TAIL: + break; + case BACK: + break; + case EVAL: + { + dSP; + OP_4tree *oop = PL_op; + COP *ocurcop = PL_curcop; + SV **ocurpad = PL_curpad; + SV *ret; + + n = ARG(scan); + PL_op = (OP_4tree*)PL_regdata->data[n]; + DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", PL_op) ); + PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 1]); + PL_reg_magic->mg_len = locinput - PL_bostr; + PL_regendp[0] = locinput; + + CALLRUNOPS(); /* Scalar context. */ + SPAGAIN; + ret = POPs; + PUTBACK; + + PL_op = oop; + PL_curpad = ocurpad; + PL_curcop = ocurcop; + if (logical) { + if (logical == 2) { /* Postponed subexpression. */ + regexp *re; + MAGIC *mg = Null(MAGIC*); + re_cc_state state; + CURCUR cctmp; + CHECKPOINT cp, lastcp; + + if(SvROK(ret) || SvRMAGICAL(ret)) { + SV *sv = SvROK(ret) ? SvRV(ret) : ret; + + if(SvMAGICAL(sv)) + mg = mg_find(sv, 'r'); + } + if (mg) { + re = (regexp *)mg->mg_obj; + (void)ReREFCNT_inc(re); + } + else { + STRLEN len; + char *t = SvPV(ret, len); + PMOP pm; + char *oprecomp = PL_regprecomp; + I32 osize = PL_regsize; + I32 onpar = PL_regnpar; + + pm.op_pmflags = 0; + re = CALLREGCOMP(t, t + len, &pm); + if (!(SvFLAGS(ret) + & (SVs_TEMP | SVs_PADTMP | SVf_READONLY))) + sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0); + PL_regprecomp = oprecomp; + PL_regsize = osize; + PL_regnpar = onpar; + } + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "Entering embedded `%s%.60s%s%s'\n", + PL_colors[0], + re->precomp, + PL_colors[1], + (strlen(re->precomp) > 60 ? "..." : "")) + ); + state.node = next; + state.prev = PL_reg_call_cc; + state.cc = PL_regcc; + state.re = PL_reg_re; + + cctmp.cur = 0; + cctmp.oldcc = 0; + PL_regcc = &cctmp; + + cp = regcppush(0); /* Save *all* the positions. */ + REGCP_SET; + cache_re(re); + state.ss = PL_savestack_ix; + *PL_reglastparen = 0; + PL_reg_call_cc = &state; + PL_reginput = locinput; + if (regmatch(re->program + 1)) { + ReREFCNT_dec(re); + regcpblow(cp); + sayYES; + } + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") + ); + ReREFCNT_dec(re); + REGCP_UNWIND; + regcppop(); + PL_reg_call_cc = state.prev; + PL_regcc = state.cc; + PL_reg_re = state.re; + cache_re(PL_reg_re); + sayNO; + } + sw = SvTRUE(ret); + logical = 0; + } + else + sv_setsv(save_scalar(PL_replgv), ret); + break; + } + case OPEN: + n = ARG(scan); /* which paren pair */ + PL_reg_start_tmp[n] = locinput; + if (n > PL_regsize) + PL_regsize = n; + break; + case CLOSE: + n = ARG(scan); /* which paren pair */ + PL_regstartp[n] = PL_reg_start_tmp[n]; + PL_regendp[n] = locinput; + if (n > *PL_reglastparen) + *PL_reglastparen = n; + break; + case GROUPP: + n = ARG(scan); /* which paren pair */ + sw = (*PL_reglastparen >= n && PL_regendp[n] != NULL); + break; + case IFTHEN: + if (sw) + next = NEXTOPER(NEXTOPER(scan)); + else { + next = scan + ARG(scan); + if (OP(next) == IFTHEN) /* Fake one. */ + next = NEXTOPER(NEXTOPER(next)); + } + break; + case LOGICAL: + logical = scan->flags; + break; + case CURLYX: { + CURCUR cc; + CHECKPOINT cp = PL_savestack_ix; + + if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ + next += ARG(next); + cc.oldcc = PL_regcc; + PL_regcc = &cc; + cc.parenfloor = *PL_reglastparen; + cc.cur = -1; + cc.min = ARG1(scan); + cc.max = ARG2(scan); + cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; + cc.next = next; + cc.minmod = minmod; + cc.lastloc = 0; + PL_reginput = locinput; + n = regmatch(PREVOPER(next)); /* start on the WHILEM */ + regcpblow(cp); + PL_regcc = cc.oldcc; + saySAME(n); + } + /* NOT REACHED */ + case WHILEM: { + /* + * This is really hard to understand, because after we match + * what we're trying to match, we must make sure the rest of + * the RE is going to match for sure, and to do that we have + * to go back UP the parse tree by recursing ever deeper. And + * if it fails, we have to reset our parent's current state + * that we can try again after backing off. + */ + + CHECKPOINT cp, lastcp; + CURCUR* cc = PL_regcc; + char *lastloc = cc->lastloc; /* Detection of 0-len. */ + + n = cc->cur + 1; /* how many we know we matched */ + PL_reginput = locinput; + + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s %ld out of %ld..%ld cc=%lx\n", + REPORT_CODE_OFF+PL_regindent*2, "", + (long)n, (long)cc->min, + (long)cc->max, (long)cc) + ); + + /* If degenerate scan matches "", assume scan done. */ + + if (locinput == cc->lastloc && n >= cc->min) { + 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+PL_regindent*2, "") + ); + if (regmatch(cc->next)) + sayYES; + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") + ); + PL_regcc->cur = ln; + PL_regcc = cc; + sayNO; + } + + /* First just match a string of min scans. */ + + if (n < cc->min) { + cc->cur = n; + cc->lastloc = locinput; + if (regmatch(cc->scan)) + sayYES; + cc->cur = n - 1; + cc->lastloc = lastloc; + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") + ); + sayNO; + } + + /* Prefer next over scan for minimal matching. */ + + if (cc->minmod) { + PL_regcc = cc->oldcc; + ln = PL_regcc->cur; + cp = regcppush(cc->parenfloor); + REGCP_SET; + if (regmatch(cc->next)) { + regcpblow(cp); + sayYES; /* All done. */ + } + REGCP_UNWIND; + regcppop(); + PL_regcc->cur = ln; + PL_regcc = cc; + + if (n >= cc->max) { /* Maximum greed exceeded? */ + if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY + && !(PL_reg_flags & RF_warned)) { + PL_reg_flags |= RF_warned; + warner(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+PL_regindent*2, "") + ); + /* Try scanning more and see if it helps. */ + PL_reginput = locinput; + cc->cur = n; + cc->lastloc = locinput; + cp = regcppush(cc->parenfloor); + REGCP_SET; + if (regmatch(cc->scan)) { + regcpblow(cp); + sayYES; + } + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") + ); + REGCP_UNWIND; + regcppop(); + cc->cur = n - 1; + cc->lastloc = lastloc; + sayNO; + } + + /* Prefer scan over next for maximal matching. */ + + if (n < cc->max) { /* More greed allowed? */ + cp = regcppush(cc->parenfloor); + cc->cur = n; + cc->lastloc = locinput; + REGCP_SET; + if (regmatch(cc->scan)) { + regcpblow(cp); + sayYES; + } + REGCP_UNWIND; + regcppop(); /* Restore some previous $s? */ + PL_reginput = locinput; + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s failed, try continuation...\n", + REPORT_CODE_OFF+PL_regindent*2, "") + ); + } + if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY + && !(PL_reg_flags & RF_warned)) { + PL_reg_flags |= RF_warned; + warner(WARN_UNSAFE, "%s limit (%d) exceeded", + "Complex regular subexpression recursion", + REG_INFTY - 1); + } + + /* Failed deeper matches of scan, so see if this one works. */ + 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+PL_regindent*2, "") + ); + PL_regcc->cur = ln; + PL_regcc = cc; + cc->cur = n - 1; + cc->lastloc = lastloc; + sayNO; + } + /* NOT REACHED */ + case BRANCHJ: + next = scan + ARG(scan); + if (next == scan) + next = NULL; + inner = NEXTOPER(NEXTOPER(scan)); + goto do_branch; + case BRANCH: + inner = NEXTOPER(scan); + do_branch: + { + CHECKPOINT lastcp; + c1 = OP(scan); + if (OP(next) != c1) /* No choice. */ + next = inner; /* Avoid recursion. */ + else { + int lastparen = *PL_reglastparen; + + REGCP_SET; + do { + PL_reginput = locinput; + if (regmatch(inner)) + sayYES; + REGCP_UNWIND; + for (n = *PL_reglastparen; n > lastparen; n--) + PL_regendp[n] = 0; + *PL_reglastparen = n; + scan = next; + /*SUPPRESS 560*/ + if (n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next))) + next += n; + else + next = NULL; + inner = NEXTOPER(scan); + if (c1 == BRANCHJ) { + inner = NEXTOPER(inner); + } + } while (scan != NULL && OP(scan) == c1); + sayNO; + /* NOTREACHED */ + } + } + break; + case MINMOD: + minmod = 1; + break; + case CURLYM: + { + I32 l = 0; + CHECKPOINT lastcp; + + /* We suppose that the next guy does not need + backtracking: in particular, it is of constant length, + and has no parenths to influence future backrefs. */ + ln = ARG1(scan); /* min to match */ + n = ARG2(scan); /* max to match */ + paren = scan->flags; + if (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. */ + PL_reginput = locinput; + if (minmod) { + minmod = 0; + if (ln && regrepeat_hard(scan, ln, &l) < ln) + sayNO; + if (ln && l == 0 && n >= ln + /* In fact, this is tricky. If paren, then the + fact that we did/didnot match may influence + future execution. */ + && !(paren && ln == 0)) + ln = n; + locinput = PL_reginput; + if (PL_regkind[(U8)OP(next)] == EXACT) { + c1 = UCHARAT(OPERAND(next) + 1); + if (OP(next) == EXACTF) + c2 = PL_fold[c1]; + else if (OP(next) == EXACTFL) + c2 = PL_fold_locale[c1]; + else + c2 = c1; + } + 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(PL_reginput) == c1 || + UCHARAT(PL_reginput) == c2) + { + if (paren) { + if (n) { + PL_regstartp[paren] = HOPc(PL_reginput, -l); + PL_regendp[paren] = PL_reginput; + } + else + PL_regendp[paren] = NULL; + } + if (regmatch(next)) + sayYES; + REGCP_UNWIND; + } + /* Couldn't or didn't -- move forward. */ + PL_reginput = locinput; + if (regrepeat_hard(scan, 1, &l)) { + ln++; + locinput = PL_reginput; + } + else + sayNO; + } + } + else { + n = regrepeat_hard(scan, n, &l); + if (n != 0 && l == 0 + /* In fact, this is tricky. If paren, then the + fact that we did/didnot match may influence + future execution. */ + && !(paren && ln == 0)) + ln = n; + locinput = PL_reginput; + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s matched %ld times, len=%ld...\n", + REPORT_CODE_OFF+PL_regindent*2, "", n, l) + ); + if (n >= ln) { + if (PL_regkind[(U8)OP(next)] == EXACT) { + c1 = UCHARAT(OPERAND(next) + 1); + if (OP(next) == EXACTF) + c2 = PL_fold[c1]; + else if (OP(next) == EXACTFL) + c2 = PL_fold_locale[c1]; + else + c2 = c1; + } + else + c1 = c2 = -1000; + } + REGCP_SET; + while (n >= ln) { + /* If it could work, try it. */ + if (c1 == -1000 || + 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_regendp[paren] = PL_reginput; + } + else + PL_regendp[paren] = NULL; + } + if (regmatch(next)) + sayYES; + REGCP_UNWIND; + } + /* Couldn't or didn't -- back up. */ + n--; + locinput = HOPc(locinput, -l); + PL_reginput = locinput; + } + } + sayNO; + break; + } + case CURLYN: + paren = scan->flags; /* Which paren to set */ + 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); + goto repeat; + case CURLY: + paren = 0; + ln = ARG1(scan); /* min to match */ + n = ARG2(scan); /* max to match */ + scan = NEXTOPER(scan) + NODE_STEP_REGNODE; + goto repeat; + case STAR: + ln = 0; + n = REG_INFTY; + scan = NEXTOPER(scan); + paren = 0; + goto repeat; + case PLUS: + ln = 1; + n = REG_INFTY; + scan = NEXTOPER(scan); + paren = 0; + repeat: + /* + * Lookahead to avoid useless match attempts + * when we know what character comes next. + */ + if (PL_regkind[(U8)OP(next)] == EXACT) { + c1 = UCHARAT(OPERAND(next) + 1); + if (OP(next) == EXACTF) + c2 = PL_fold[c1]; + else if (OP(next) == EXACTFL) + c2 = PL_fold_locale[c1]; + else + c2 = c1; + } + else + c1 = c2 = -1000; + PL_reginput = locinput; + if (minmod) { + CHECKPOINT lastcp; + minmod = 0; + if (ln && regrepeat(scan, ln) < ln) + sayNO; + locinput = PL_reginput; + REGCP_SET; + while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */ + /* If it could work, try it. */ + if (c1 == -1000 || + UCHARAT(PL_reginput) == c1 || + UCHARAT(PL_reginput) == c2) + { + if (paren) { + if (n) { + PL_regstartp[paren] = HOPc(PL_reginput, -1); + PL_regendp[paren] = PL_reginput; + } + else + PL_regendp[paren] = NULL; + } + if (regmatch(next)) + sayYES; + REGCP_UNWIND; + } + /* Couldn't or didn't -- move forward. */ + PL_reginput = locinput; + if (regrepeat(scan, 1)) { + ln++; + locinput = PL_reginput; + } + else + sayNO; + } + } + else { + CHECKPOINT lastcp; + n = regrepeat(scan, n); + 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(PL_reginput) == c1 || + UCHARAT(PL_reginput) == c2) + { + if (paren && n) { + if (n) { + PL_regstartp[paren] = HOPc(PL_reginput, -1); + PL_regendp[paren] = PL_reginput; + } + else + PL_regendp[paren] = NULL; + } + if (regmatch(next)) + sayYES; + REGCP_UNWIND; + } + /* Couldn't or didn't -- back up. */ + n--; + PL_reginput = locinput = HOPc(locinput, -1); + } + } + else { + while (n >= ln) { + /* If it could work, try it. */ + if (c1 == -1000 || + UCHARAT(PL_reginput) == c1 || + UCHARAT(PL_reginput) == c2) + { + if (regmatch(next)) + sayYES; + REGCP_UNWIND; + } + /* Couldn't or didn't -- back up. */ + n--; + PL_reginput = locinput = HOPc(locinput, -1); + } + } + } + sayNO; + break; + case END: + if (PL_reg_call_cc) { + re_cc_state *cur_call_cc = PL_reg_call_cc; + CURCUR *cctmp = PL_regcc; + regexp *re = PL_reg_re; + CHECKPOINT cp, lastcp; + + cp = regcppush(0); /* Save *all* the positions. */ + REGCP_SET; + regcp_set_to(PL_reg_call_cc->ss); /* Restore parens of + the caller. */ + PL_reginput = locinput; /* Make position available to + the callcc. */ + cache_re(PL_reg_call_cc->re); + PL_regcc = PL_reg_call_cc->cc; + PL_reg_call_cc = PL_reg_call_cc->prev; + if (regmatch(cur_call_cc->node)) { + PL_reg_call_cc = cur_call_cc; + regcpblow(cp); + sayYES; + } + REGCP_UNWIND; + regcppop(); + PL_reg_call_cc = cur_call_cc; + PL_regcc = cctmp; + PL_reg_re = re; + cache_re(re); + + DEBUG_r( + PerlIO_printf(Perl_debug_log, + "%*s continuation failed...\n", + REPORT_CODE_OFF+PL_regindent*2, "") + ); + sayNO; + } + if (locinput < PL_regtill) + sayNO; /* Cannot match: too short. */ + /* Fall through */ + 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 (scan->flags) { + s = HOPMAYBEc(locinput, -scan->flags); + if (!s) + goto say_yes; + PL_reginput = s; + } + else + PL_reginput = locinput; + goto do_ifmatch; + case IFMATCH: + n = 1; + if (scan->flags) { + s = HOPMAYBEc(locinput, -scan->flags); + if (!s || s < PL_bostr) + goto say_no; + PL_reginput = s; + } + else + PL_reginput = locinput; + + do_ifmatch: + inner = NEXTOPER(NEXTOPER(scan)); + if (regmatch(inner) != n) { + say_no: + if (logical) { + logical = 0; + sw = 0; + goto do_longjump; + } + else + sayNO; + } + say_yes: + if (logical) { + logical = 0; + sw = 1; + } + if (OP(scan) == SUSPEND) { + locinput = PL_reginput; + nextchr = UCHARAT(locinput); + } + /* FALL THROUGH. */ + case LONGJMP: + do_longjump: + next = scan + ARG(scan); + if (next == scan) + next = NULL; + break; + default: + PerlIO_printf(PerlIO_stderr(), "%lx %d\n", + (unsigned long)scan, OP(scan)); + FAIL("regexp memory corruption"); + } + scan = next; + } + + /* + * We get here only if there's trouble -- normally "case END" is + * the terminating point. + */ + FAIL("corrupted regexp pointers"); + /*NOTREACHED*/ + sayNO; + +yes: +#ifdef DEBUGGING + PL_regindent--; +#endif + return 1; + +no: +#ifdef DEBUGGING + PL_regindent--; +#endif + return 0; +} + +/* + - regrepeat - repeatedly match something simple, report how many + */ +/* + * [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. [Er, except utf8.]] + */ +STATIC I32 +regrepeat(regnode *p, I32 max) +{ + dTHR; + register char *scan; + register char *opnd; + register I32 c; + register char *loceol = PL_regeol; + register I32 hardcount = 0; + + scan = PL_reginput; + if (max != REG_INFTY && max < loceol - scan) + loceol = scan + max; + opnd = (char *) OPERAND(p); + switch (OP(p)) { + 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) + scan++; + break; + case EXACTF: /* length of string is 1 */ + c = UCHARAT(++opnd); + while (scan < loceol && + (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) + scan++; + break; + case EXACTFL: /* length of string is 1 */ + PL_reg_flags |= RF_tainted; + c = UCHARAT(++opnd); + while (scan < loceol && + (UCHARAT(scan) == c || UCHARAT(scan) == 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++; + break; + case ALNUM: + 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: + 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: + 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: + 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: + 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. */ + } + + if (hardcount) + c = hardcount; + else + c = scan - PL_reginput; + PL_reginput = scan; + + DEBUG_r( + { + SV *prop = sv_newmortal(); + + regprop(prop, p); + PerlIO_printf(Perl_debug_log, + "%*s %s can match %ld times out of %ld...\n", + REPORT_CODE_OFF+1, "", SvPVX(prop),c,max); + }); + + return(c); +} + +/* + - regrepeat_hard - repeatedly match something, report total lenth and length + * + * The repeater is supposed to have constant length. + */ + +STATIC I32 +regrepeat_hard(regnode *p, I32 max, I32 *lp) +{ + dTHR; + register char *scan; + register char *start; + 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; + } + } + 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 count; +} + +/* + - regclass - determine if a character falls into a character class + */ + +STATIC bool +reginclass(register char *p, register I32 c) +{ + dTHR; + char flags = *p; + bool match = FALSE; + + c &= 0xFF; + if (ANYOF_TEST(p, c)) + match = TRUE; + else if (flags & ANYOF_FOLD) { + I32 cf; + if (flags & ANYOF_LOCALE) { + PL_reg_flags |= RF_tainted; + cf = PL_fold_locale[c]; + } + else + cf = PL_fold[c]; + if (ANYOF_TEST(p, cf)) + match = TRUE; + } + + if (!match && (flags & ANYOF_ISA)) { + PL_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))) + { + match = TRUE; + } + } + + return (flags & ANYOF_INVERT) ? !match : match; +} + +STATIC bool +reginclassutf8(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; + } + + if (!match && (flags & ANYOF_ISA)) { + PL_reg_flags |= RF_tainted; + + if (((flags & ANYOF_ALNUML) && isALNUM_LC_utf8(p)) || + ((flags & ANYOF_NALNUML) && !isALNUM_LC_utf8(p)) || + ((flags & ANYOF_SPACEL) && isSPACE_LC_utf8(p)) || + ((flags & ANYOF_NSPACEL) && !isSPACE_LC_utf8(p))) + { + match = TRUE; + } + } + + return (flags & ANYOF_INVERT) ? !match : match; +} + +STATIC U8 * +reghop(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 * +reghopmaybe(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; }