5 * "One Ring to rule them all, One Ring to find them..."
8 /* This file contains functions for executing a regular expression. See
9 * also regcomp.c which funnily enough, contains functions for compiling
10 * a regular expression.
12 * This file is also copied at build time to ext/re/re_exec.c, where
13 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
14 * This causes the main functions to be compiled under new names and with
15 * debugging support added, which makes "use re 'debug'" work.
18 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
19 * confused with the original package (see point 3 below). Thanks, Henry!
22 /* Additional note: this code is very heavily munged from Henry's version
23 * in places. In some spots I've traded clarity for efficiency, so don't
24 * blame Henry for some of the lack of readability.
27 /* The names of the functions have been changed from regcomp and
28 * regexec to pregcomp and pregexec in order to avoid conflicts
29 * with the POSIX routines of the same names.
32 #ifdef PERL_EXT_RE_BUILD
37 * pregcomp and pregexec -- regsub and regerror are not used in perl
39 * Copyright (c) 1986 by University of Toronto.
40 * Written by Henry Spencer. Not derived from licensed software.
42 * Permission is granted to anyone to use this software for any
43 * purpose on any computer system, and to redistribute it freely,
44 * subject to the following restrictions:
46 * 1. The author is not responsible for the consequences of use of
47 * this software, no matter how awful, even if they arise
50 * 2. The origin of this software must not be misrepresented, either
51 * by explicit claim or by omission.
53 * 3. Altered versions must be plainly marked as such, and must not
54 * be misrepresented as being the original software.
56 **** Alterations to Henry's code are...
58 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
59 **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007 by Larry Wall and others
61 **** You may distribute under the terms of either the GNU General Public
62 **** License or the Artistic License, as specified in the README file.
64 * Beware that some of this code is subtly aware of the way operator
65 * precedence is structured in regular expressions. Serious changes in
66 * regular-expression syntax might require a total rethink.
69 #define PERL_IN_REGEXEC_C
72 #ifdef PERL_IN_XSUB_RE
78 #define RF_tainted 1 /* tainted information used? */
79 #define RF_warned 2 /* warned about big count? */
81 #define RF_utf8 8 /* Pattern contains multibyte chars? */
83 #define UTF ((PL_reg_flags & RF_utf8) != 0)
85 #define RS_init 1 /* eval environment created */
86 #define RS_set 2 /* replsv value is set */
92 #define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c)))
98 #define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv))
99 #define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b)
101 #define HOPc(pos,off) \
102 (char *)(PL_reg_match_utf8 \
103 ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \
105 #define HOPBACKc(pos, off) \
106 (char*)(PL_reg_match_utf8\
107 ? reghopmaybe3((U8*)pos, -off, (U8*)PL_bostr) \
108 : (pos - off >= PL_bostr) \
112 #define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off))
113 #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
115 #define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \
116 if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END
117 #define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a")
118 #define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0")
119 #define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ")
120 #define LOAD_UTF8_CHARCLASS_MARK() LOAD_UTF8_CHARCLASS(mark, "\xcd\x86")
122 /* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */
124 /* for use after a quantifier and before an EXACT-like node -- japhy */
125 #define JUMPABLE(rn) ( \
126 OP(rn) == OPEN || OP(rn) == CLOSE || OP(rn) == EVAL || \
127 OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
128 OP(rn) == PLUS || OP(rn) == MINMOD || \
129 (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \
132 #define HAS_TEXT(rn) ( \
133 PL_regkind[OP(rn)] == EXACT || PL_regkind[OP(rn)] == REF \
137 Search for mandatory following text node; for lookahead, the text must
138 follow but for lookbehind (rn->flags != 0) we skip to the next step.
140 #define FIND_NEXT_IMPT(rn) STMT_START { \
141 while (JUMPABLE(rn)) { \
142 const OPCODE type = OP(rn); \
143 if (type == SUSPEND || PL_regkind[type] == CURLY) \
144 rn = NEXTOPER(NEXTOPER(rn)); \
145 else if (type == PLUS) \
147 else if (type == IFMATCH) \
148 rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
149 else rn += NEXT_OFF(rn); \
154 static void restore_pos(pTHX_ void *arg);
157 S_regcppush(pTHX_ I32 parenfloor)
160 const int retval = PL_savestack_ix;
161 #define REGCP_PAREN_ELEMS 4
162 const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
164 GET_RE_DEBUG_FLAGS_DECL;
166 if (paren_elems_to_push < 0)
167 Perl_croak(aTHX_ "panic: paren_elems_to_push < 0");
169 #define REGCP_OTHER_ELEMS 8
170 SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS);
172 for (p = PL_regsize; p > parenfloor; p--) {
173 /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
174 SSPUSHINT(PL_regendp[p]);
175 SSPUSHINT(PL_regstartp[p]);
176 SSPUSHPTR(PL_reg_start_tmp[p]);
178 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
179 " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n",
180 (UV)p, (IV)PL_regstartp[p],
181 (IV)(PL_reg_start_tmp[p] - PL_bostr),
185 /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
186 SSPUSHPTR(PL_regstartp);
187 SSPUSHPTR(PL_regendp);
188 SSPUSHINT(PL_regsize);
189 SSPUSHINT(*PL_reglastparen);
190 SSPUSHINT(*PL_reglastcloseparen);
191 SSPUSHPTR(PL_reginput);
192 #define REGCP_FRAME_ELEMS 2
193 /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and
194 * are needed for the regexp context stack bookkeeping. */
195 SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
196 SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */
201 /* These are needed since we do not localize EVAL nodes: */
202 #define REGCP_SET(cp) \
204 PerlIO_printf(Perl_debug_log, \
205 " Setting an EVAL scope, savestack=%"IVdf"\n", \
206 (IV)PL_savestack_ix)); \
209 #define REGCP_UNWIND(cp) \
211 if (cp != PL_savestack_ix) \
212 PerlIO_printf(Perl_debug_log, \
213 " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
214 (IV)(cp), (IV)PL_savestack_ix)); \
218 S_regcppop(pTHX_ const regexp *rex)
224 GET_RE_DEBUG_FLAGS_DECL;
226 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
228 assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
229 i = SSPOPINT; /* Parentheses elements to pop. */
230 input = (char *) SSPOPPTR;
231 *PL_reglastcloseparen = SSPOPINT;
232 *PL_reglastparen = SSPOPINT;
233 PL_regsize = SSPOPINT;
234 PL_regendp=(I32 *) SSPOPPTR;
235 PL_regstartp=(I32 *) SSPOPPTR;
238 /* Now restore the parentheses context. */
239 for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
240 i > 0; i -= REGCP_PAREN_ELEMS) {
242 U32 paren = (U32)SSPOPINT;
243 PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
244 PL_regstartp[paren] = SSPOPINT;
246 if (paren <= *PL_reglastparen)
247 PL_regendp[paren] = tmps;
249 PerlIO_printf(Perl_debug_log,
250 " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n",
251 (UV)paren, (IV)PL_regstartp[paren],
252 (IV)(PL_reg_start_tmp[paren] - PL_bostr),
253 (IV)PL_regendp[paren],
254 (paren > *PL_reglastparen ? "(no)" : ""));
258 if (*PL_reglastparen + 1 <= rex->nparens) {
259 PerlIO_printf(Perl_debug_log,
260 " restoring \\%"IVdf"..\\%"IVdf" to undef\n",
261 (IV)(*PL_reglastparen + 1), (IV)rex->nparens);
265 /* It would seem that the similar code in regtry()
266 * already takes care of this, and in fact it is in
267 * a better location to since this code can #if 0-ed out
268 * but the code in regtry() is needed or otherwise tests
269 * requiring null fields (pat.t#187 and split.t#{13,14}
270 * (as of patchlevel 7877) will fail. Then again,
271 * this code seems to be necessary or otherwise
272 * building DynaLoader will fail:
273 * "Error: '*' not in typemap in DynaLoader.xs, line 164"
275 for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) {
277 PL_regstartp[i] = -1;
284 #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
287 * pregexec and friends
290 #ifndef PERL_IN_XSUB_RE
292 - pregexec - match a regexp against a string
295 Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend,
296 char *strbeg, I32 minend, SV *screamer, U32 nosave)
297 /* strend: pointer to null at end of string */
298 /* strbeg: real beginning of string */
299 /* minend: end of match must be >=minend after stringarg. */
300 /* nosave: For optimizations. */
303 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
304 nosave ? 0 : REXEC_COPY_STR);
309 * Need to implement the following flags for reg_anch:
311 * USE_INTUIT_NOML - Useful to call re_intuit_start() first
313 * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer
314 * INTUIT_AUTORITATIVE_ML
315 * INTUIT_ONCE_NOML - Intuit can match in one location only.
318 * Another flag for this function: SECOND_TIME (so that float substrs
319 * with giant delta may be not rechecked).
322 /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
324 /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend.
325 Otherwise, only SvCUR(sv) is used to get strbeg. */
327 /* XXXX We assume that strpos is strbeg unless sv. */
329 /* XXXX Some places assume that there is a fixed substring.
330 An update may be needed if optimizer marks as "INTUITable"
331 RExen without fixed substrings. Similarly, it is assumed that
332 lengths of all the strings are no more than minlen, thus they
333 cannot come from lookahead.
334 (Or minlen should take into account lookahead.)
335 NOTE: Some of this comment is not correct. minlen does now take account
336 of lookahead/behind. Further research is required. -- demerphq
340 /* A failure to find a constant substring means that there is no need to make
341 an expensive call to REx engine, thus we celebrate a failure. Similarly,
342 finding a substring too deep into the string means that less calls to
343 regtry() should be needed.
345 REx compiler's optimizer found 4 possible hints:
346 a) Anchored substring;
348 c) Whether we are anchored (beginning-of-line or \G);
349 d) First node (of those at offset 0) which may distingush positions;
350 We use a)b)d) and multiline-part of c), and try to find a position in the
351 string which does not contradict any of them.
354 /* Most of decisions we do here should have been done at compile time.
355 The nodes of the REx which we used for the search should have been
356 deleted from the finite automaton. */
359 Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
360 char *strend, U32 flags, re_scream_pos_data *data)
363 register I32 start_shift = 0;
364 /* Should be nonnegative! */
365 register I32 end_shift = 0;
370 const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */
372 register char *other_last = NULL; /* other substr checked before this */
373 char *check_at = NULL; /* check substr found at this pos */
374 const I32 multiline = prog->extflags & RXf_PMf_MULTILINE;
375 RXi_GET_DECL(prog,progi);
377 const char * const i_strpos = strpos;
380 GET_RE_DEBUG_FLAGS_DECL;
382 RX_MATCH_UTF8_set(prog,do_utf8);
384 if (prog->extflags & RXf_UTF8) {
385 PL_reg_flags |= RF_utf8;
388 debug_start_match(prog, do_utf8, strpos, strend,
389 sv ? "Guessing start of match in sv for"
390 : "Guessing start of match in string for");
393 /* CHR_DIST() would be more correct here but it makes things slow. */
394 if (prog->minlen > strend - strpos) {
395 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
396 "String too short... [re_intuit_start]\n"));
400 strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
403 if (!prog->check_utf8 && prog->check_substr)
404 to_utf8_substr(prog);
405 check = prog->check_utf8;
407 if (!prog->check_substr && prog->check_utf8)
408 to_byte_substr(prog);
409 check = prog->check_substr;
411 if (check == &PL_sv_undef) {
412 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
413 "Non-utf8 string cannot match utf8 check string\n"));
416 if (prog->extflags & RXf_ANCH) { /* Match at beg-of-str or after \n */
417 ml_anch = !( (prog->extflags & RXf_ANCH_SINGLE)
418 || ( (prog->extflags & RXf_ANCH_BOL)
419 && !multiline ) ); /* Check after \n? */
422 if ( !(prog->extflags & RXf_ANCH_GPOS) /* Checked by the caller */
423 && !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */
424 /* SvCUR is not set on references: SvRV and SvPVX_const overlap */
426 && (strpos != strbeg)) {
427 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
430 if (prog->check_offset_min == prog->check_offset_max &&
431 !(prog->extflags & RXf_CANY_SEEN)) {
432 /* Substring at constant offset from beg-of-str... */
435 s = HOP3c(strpos, prog->check_offset_min, strend);
438 slen = SvCUR(check); /* >= 1 */
440 if ( strend - s > slen || strend - s < slen - 1
441 || (strend - s == slen && strend[-1] != '\n')) {
442 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
445 /* Now should match s[0..slen-2] */
447 if (slen && (*SvPVX_const(check) != *s
449 && memNE(SvPVX_const(check), s, slen)))) {
451 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
455 else if (*SvPVX_const(check) != *s
456 || ((slen = SvCUR(check)) > 1
457 && memNE(SvPVX_const(check), s, slen)))
460 goto success_at_start;
463 /* Match is anchored, but substr is not anchored wrt beg-of-str. */
465 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
466 end_shift = prog->check_end_shift;
469 const I32 end = prog->check_offset_max + CHR_SVLEN(check)
470 - (SvTAIL(check) != 0);
471 const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
473 if (end_shift < eshift)
477 else { /* Can match at random position */
480 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
481 end_shift = prog->check_end_shift;
483 /* end shift should be non negative here */
486 #ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */
488 Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ",
489 (IV)end_shift, prog->precomp);
493 /* Find a possible match in the region s..strend by looking for
494 the "check" substring in the region corrected by start/end_shift. */
497 I32 srch_start_shift = start_shift;
498 I32 srch_end_shift = end_shift;
499 if (srch_start_shift < 0 && strbeg - s > srch_start_shift) {
500 srch_end_shift -= ((strbeg - s) - srch_start_shift);
501 srch_start_shift = strbeg - s;
503 DEBUG_OPTIMISE_MORE_r({
504 PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n",
505 (IV)prog->check_offset_min,
506 (IV)srch_start_shift,
508 (IV)prog->check_end_shift);
511 if (flags & REXEC_SCREAM) {
512 I32 p = -1; /* Internal iterator of scream. */
513 I32 * const pp = data ? data->scream_pos : &p;
515 if (PL_screamfirst[BmRARE(check)] >= 0
516 || ( BmRARE(check) == '\n'
517 && (BmPREVIOUS(check) == SvCUR(check) - 1)
519 s = screaminstr(sv, check,
520 srch_start_shift + (s - strbeg), srch_end_shift, pp, 0);
523 /* we may be pointing at the wrong string */
524 if (s && RX_MATCH_COPIED(prog))
525 s = strbeg + (s - SvPVX_const(sv));
527 *data->scream_olds = s;
532 if (prog->extflags & RXf_CANY_SEEN) {
533 start_point= (U8*)(s + srch_start_shift);
534 end_point= (U8*)(strend - srch_end_shift);
536 start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend);
537 end_point= HOP3(strend, -srch_end_shift, strbeg);
539 DEBUG_OPTIMISE_MORE_r({
540 PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n",
541 (int)(end_point - start_point),
542 (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point),
546 s = fbm_instr( start_point, end_point,
547 check, multiline ? FBMrf_MULTILINE : 0);
550 /* Update the count-of-usability, remove useless subpatterns,
554 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
555 SvPVX_const(check), RE_SV_DUMPLEN(check), 30);
556 PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s",
557 (s ? "Found" : "Did not find"),
558 (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)
559 ? "anchored" : "floating"),
562 (s ? " at offset " : "...\n") );
567 /* Finish the diagnostic message */
568 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
570 /* XXX dmq: first branch is for positive lookbehind...
571 Our check string is offset from the beginning of the pattern.
572 So we need to do any stclass tests offset forward from that
581 /* Got a candidate. Check MBOL anchoring, and the *other* substr.
582 Start with the other substr.
583 XXXX no SCREAM optimization yet - and a very coarse implementation
584 XXXX /ttx+/ results in anchored="ttx", floating="x". floating will
585 *always* match. Probably should be marked during compile...
586 Probably it is right to do no SCREAM here...
589 if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8)
590 : (prog->float_substr && prog->anchored_substr))
592 /* Take into account the "other" substring. */
593 /* XXXX May be hopelessly wrong for UTF... */
596 if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) {
599 char * const last = HOP3c(s, -start_shift, strbeg);
601 char * const saved_s = s;
604 t = s - prog->check_offset_max;
605 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
607 || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos))
612 t = HOP3c(t, prog->anchored_offset, strend);
613 if (t < other_last) /* These positions already checked */
615 last2 = last1 = HOP3c(strend, -prog->minlen, strbeg);
618 /* XXXX It is not documented what units *_offsets are in.
619 We assume bytes, but this is clearly wrong.
620 Meaning this code needs to be carefully reviewed for errors.
624 /* On end-of-str: see comment below. */
625 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
626 if (must == &PL_sv_undef) {
628 DEBUG_r(must = prog->anchored_utf8); /* for debug */
633 HOP3(HOP3(last1, prog->anchored_offset, strend)
634 + SvCUR(must), -(SvTAIL(must)!=0), strbeg),
636 multiline ? FBMrf_MULTILINE : 0
639 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
640 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
641 PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s",
642 (s ? "Found" : "Contradicts"),
643 quoted, RE_SV_TAIL(must));
648 if (last1 >= last2) {
649 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
650 ", giving up...\n"));
653 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
654 ", trying floating at offset %ld...\n",
655 (long)(HOP3c(saved_s, 1, strend) - i_strpos)));
656 other_last = HOP3c(last1, prog->anchored_offset+1, strend);
657 s = HOP3c(last, 1, strend);
661 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
662 (long)(s - i_strpos)));
663 t = HOP3c(s, -prog->anchored_offset, strbeg);
664 other_last = HOP3c(s, 1, strend);
672 else { /* Take into account the floating substring. */
674 char * const saved_s = s;
677 t = HOP3c(s, -start_shift, strbeg);
679 HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg);
680 if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
681 last = HOP3c(t, prog->float_max_offset, strend);
682 s = HOP3c(t, prog->float_min_offset, strend);
685 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
686 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
687 /* fbm_instr() takes into account exact value of end-of-str
688 if the check is SvTAIL(ed). Since false positives are OK,
689 and end-of-str is not later than strend we are OK. */
690 if (must == &PL_sv_undef) {
692 DEBUG_r(must = prog->float_utf8); /* for debug message */
695 s = fbm_instr((unsigned char*)s,
696 (unsigned char*)last + SvCUR(must)
698 must, multiline ? FBMrf_MULTILINE : 0);
700 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
701 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
702 PerlIO_printf(Perl_debug_log, "%s floating substr %s%s",
703 (s ? "Found" : "Contradicts"),
704 quoted, RE_SV_TAIL(must));
708 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
709 ", giving up...\n"));
712 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
713 ", trying anchored starting at offset %ld...\n",
714 (long)(saved_s + 1 - i_strpos)));
716 s = HOP3c(t, 1, strend);
720 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
721 (long)(s - i_strpos)));
722 other_last = s; /* Fix this later. --Hugo */
732 t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos);
734 DEBUG_OPTIMISE_MORE_r(
735 PerlIO_printf(Perl_debug_log,
736 "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n",
737 (IV)prog->check_offset_min,
738 (IV)prog->check_offset_max,
746 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
748 || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos)))
751 /* Fixed substring is found far enough so that the match
752 cannot start at strpos. */
754 if (ml_anch && t[-1] != '\n') {
755 /* Eventually fbm_*() should handle this, but often
756 anchored_offset is not 0, so this check will not be wasted. */
757 /* XXXX In the code below we prefer to look for "^" even in
758 presence of anchored substrings. And we search even
759 beyond the found float position. These pessimizations
760 are historical artefacts only. */
762 while (t < strend - prog->minlen) {
764 if (t < check_at - prog->check_offset_min) {
765 if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) {
766 /* Since we moved from the found position,
767 we definitely contradict the found anchored
768 substr. Due to the above check we do not
769 contradict "check" substr.
770 Thus we can arrive here only if check substr
771 is float. Redo checking for "other"=="fixed".
774 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
775 PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
776 goto do_other_anchored;
778 /* We don't contradict the found floating substring. */
779 /* XXXX Why not check for STCLASS? */
781 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
782 PL_colors[0], PL_colors[1], (long)(s - i_strpos)));
785 /* Position contradicts check-string */
786 /* XXXX probably better to look for check-string
787 than for "\n", so one should lower the limit for t? */
788 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
789 PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos)));
790 other_last = strpos = s = t + 1;
795 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
796 PL_colors[0], PL_colors[1]));
800 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
801 PL_colors[0], PL_colors[1]));
805 ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */
808 /* The found string does not prohibit matching at strpos,
809 - no optimization of calling REx engine can be performed,
810 unless it was an MBOL and we are not after MBOL,
811 or a future STCLASS check will fail this. */
813 /* Even in this situation we may use MBOL flag if strpos is offset
814 wrt the start of the string. */
815 if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
816 && (strpos != strbeg) && strpos[-1] != '\n'
817 /* May be due to an implicit anchor of m{.*foo} */
818 && !(prog->intflags & PREGf_IMPLICIT))
823 DEBUG_EXECUTE_r( if (ml_anch)
824 PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
825 (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]);
828 if (!(prog->intflags & PREGf_NAUGHTY) /* XXXX If strpos moved? */
830 prog->check_utf8 /* Could be deleted already */
831 && --BmUSEFUL(prog->check_utf8) < 0
832 && (prog->check_utf8 == prog->float_utf8)
834 prog->check_substr /* Could be deleted already */
835 && --BmUSEFUL(prog->check_substr) < 0
836 && (prog->check_substr == prog->float_substr)
839 /* If flags & SOMETHING - do not do it many times on the same match */
840 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
841 SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr);
842 if (do_utf8 ? prog->check_substr : prog->check_utf8)
843 SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8);
844 prog->check_substr = prog->check_utf8 = NULL; /* disable */
845 prog->float_substr = prog->float_utf8 = NULL; /* clear */
846 check = NULL; /* abort */
848 /* XXXX This is a remnant of the old implementation. It
849 looks wasteful, since now INTUIT can use many
851 prog->extflags &= ~RXf_USE_INTUIT;
858 /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
859 /* trie stclasses are too expensive to use here, we are better off to
860 leave it to regmatch itself */
861 if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) {
862 /* minlen == 0 is possible if regstclass is \b or \B,
863 and the fixed substr is ''$.
864 Since minlen is already taken into account, s+1 is before strend;
865 accidentally, minlen >= 1 guaranties no false positives at s + 1
866 even for \b or \B. But (minlen? 1 : 0) below assumes that
867 regstclass does not come from lookahead... */
868 /* If regstclass takes bytelength more than 1: If charlength==1, OK.
869 This leaves EXACTF only, which is dealt with in find_byclass(). */
870 const U8* const str = (U8*)STRING(progi->regstclass);
871 const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
872 ? CHR_DIST(str+STR_LEN(progi->regstclass), str)
875 if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
876 endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend);
877 else if (prog->float_substr || prog->float_utf8)
878 endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend);
882 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf"\n",
883 (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg)));
886 s = find_byclass(prog, progi->regstclass, s, endpos, NULL);
889 const char *what = NULL;
891 if (endpos == strend) {
892 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
893 "Could not match STCLASS...\n") );
896 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
897 "This position contradicts STCLASS...\n") );
898 if ((prog->extflags & RXf_ANCH) && !ml_anch)
900 /* Contradict one of substrings */
901 if (prog->anchored_substr || prog->anchored_utf8) {
902 if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) {
903 DEBUG_EXECUTE_r( what = "anchored" );
905 s = HOP3c(t, 1, strend);
906 if (s + start_shift + end_shift > strend) {
907 /* XXXX Should be taken into account earlier? */
908 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
909 "Could not match STCLASS...\n") );
914 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
915 "Looking for %s substr starting at offset %ld...\n",
916 what, (long)(s + start_shift - i_strpos)) );
919 /* Have both, check_string is floating */
920 if (t + start_shift >= check_at) /* Contradicts floating=check */
921 goto retry_floating_check;
922 /* Recheck anchored substring, but not floating... */
926 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
927 "Looking for anchored substr starting at offset %ld...\n",
928 (long)(other_last - i_strpos)) );
929 goto do_other_anchored;
931 /* Another way we could have checked stclass at the
932 current position only: */
937 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
938 "Looking for /%s^%s/m starting at offset %ld...\n",
939 PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
942 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
944 /* Check is floating subtring. */
945 retry_floating_check:
946 t = check_at - start_shift;
947 DEBUG_EXECUTE_r( what = "floating" );
948 goto hop_and_restart;
951 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
952 "By STCLASS: moving %ld --> %ld\n",
953 (long)(t - i_strpos), (long)(s - i_strpos))
957 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
958 "Does not contradict STCLASS...\n");
963 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
964 PL_colors[4], (check ? "Guessed" : "Giving up"),
965 PL_colors[5], (long)(s - i_strpos)) );
968 fail_finish: /* Substring not found */
969 if (prog->check_substr || prog->check_utf8) /* could be removed already */
970 BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
972 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
973 PL_colors[4], PL_colors[5]));
979 #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \
980 uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \
981 switch (trie_type) { \
982 case trie_utf8_fold: \
984 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
989 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
990 uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \
991 foldlen -= UNISKIP( uvc ); \
992 uscan = foldbuf + UNISKIP( uvc ); \
996 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
1004 charid = trie->charmap[ uvc ]; \
1008 if (widecharmap) { \
1009 SV** const svpp = hv_fetch(widecharmap, \
1010 (char*)&uvc, sizeof(UV), 0); \
1012 charid = (U16)SvIV(*svpp); \
1017 #define REXEC_FBC_EXACTISH_CHECK(CoNd) \
1020 ibcmp_utf8(s, NULL, 0, do_utf8, \
1021 m, NULL, ln, (bool)UTF)) \
1022 && (!reginfo || regtry(reginfo, &s)) ) \
1025 U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \
1026 uvchr_to_utf8(tmpbuf, c); \
1027 f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \
1029 && (f == c1 || f == c2) \
1030 && (ln == foldlen || \
1031 !ibcmp_utf8((char *) foldbuf, \
1032 NULL, foldlen, do_utf8, \
1034 NULL, ln, (bool)UTF)) \
1035 && (!reginfo || regtry(reginfo, &s)) ) \
1040 #define REXEC_FBC_EXACTISH_SCAN(CoNd) \
1044 && (ln == 1 || !(OP(c) == EXACTF \
1046 : ibcmp_locale(s, m, ln))) \
1047 && (!reginfo || regtry(reginfo, &s)) ) \
1053 #define REXEC_FBC_UTF8_SCAN(CoDe) \
1055 while (s + (uskip = UTF8SKIP(s)) <= strend) { \
1061 #define REXEC_FBC_SCAN(CoDe) \
1063 while (s < strend) { \
1069 #define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \
1070 REXEC_FBC_UTF8_SCAN( \
1072 if (tmp && (!reginfo || regtry(reginfo, &s))) \
1081 #define REXEC_FBC_CLASS_SCAN(CoNd) \
1084 if (tmp && (!reginfo || regtry(reginfo, &s))) \
1093 #define REXEC_FBC_TRYIT \
1094 if ((!reginfo || regtry(reginfo, &s))) \
1097 #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \
1100 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1103 REXEC_FBC_CLASS_SCAN(CoNd); \
1107 #define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \
1108 PL_reg_flags |= RF_tainted; \
1110 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1113 REXEC_FBC_CLASS_SCAN(CoNd); \
1117 #define DUMP_EXEC_POS(li,s,doutf8) \
1118 dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8)
1120 /* We know what class REx starts with. Try to find this position... */
1121 /* if reginfo is NULL, its a dryrun */
1122 /* annoyingly all the vars in this routine have different names from their counterparts
1123 in regmatch. /grrr */
1126 S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
1127 const char *strend, regmatch_info *reginfo)
1130 const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
1134 register STRLEN uskip;
1138 register I32 tmp = 1; /* Scratch variable? */
1139 register const bool do_utf8 = PL_reg_match_utf8;
1140 RXi_GET_DECL(prog,progi);
1142 /* We know what class it must start with. */
1146 REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
1147 !UTF8_IS_INVARIANT((U8)s[0]) ?
1148 reginclass(prog, c, (U8*)s, 0, do_utf8) :
1149 REGINCLASS(prog, c, (U8*)s));
1152 while (s < strend) {
1155 if (REGINCLASS(prog, c, (U8*)s) ||
1156 (ANYOF_FOLD_SHARP_S(c, s, strend) &&
1157 /* The assignment of 2 is intentional:
1158 * for the folded sharp s, the skip is 2. */
1159 (skip = SHARP_S_SKIP))) {
1160 if (tmp && (!reginfo || regtry(reginfo, &s)))
1173 if (tmp && (!reginfo || regtry(reginfo, &s)))
1181 ln = STR_LEN(c); /* length to match in octets/bytes */
1182 lnc = (I32) ln; /* length to match in characters */
1184 STRLEN ulen1, ulen2;
1186 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
1187 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
1188 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1190 to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
1191 to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
1193 c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE,
1195 c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE,
1198 while (sm < ((U8 *) m + ln)) {
1213 c2 = PL_fold_locale[c1];
1215 e = HOP3c(strend, -((I32)lnc), s);
1217 if (!reginfo && e < s)
1218 e = s; /* Due to minlen logic of intuit() */
1220 /* The idea in the EXACTF* cases is to first find the
1221 * first character of the EXACTF* node and then, if
1222 * necessary, case-insensitively compare the full
1223 * text of the node. The c1 and c2 are the first
1224 * characters (though in Unicode it gets a bit
1225 * more complicated because there are more cases
1226 * than just upper and lower: one needs to use
1227 * the so-called folding case for case-insensitive
1228 * matching (called "loose matching" in Unicode).
1229 * ibcmp_utf8() will do just that. */
1233 U8 tmpbuf [UTF8_MAXBYTES+1];
1234 STRLEN len, foldlen;
1235 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1237 /* Upper and lower of 1st char are equal -
1238 * probably not a "letter". */
1240 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
1242 REXEC_FBC_EXACTISH_CHECK(c == c1);
1247 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
1250 /* Handle some of the three Greek sigmas cases.
1251 * Note that not all the possible combinations
1252 * are handled here: some of them are handled
1253 * by the standard folding rules, and some of
1254 * them (the character class or ANYOF cases)
1255 * are handled during compiletime in
1256 * regexec.c:S_regclass(). */
1257 if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA ||
1258 c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA)
1259 c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA;
1261 REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2);
1267 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
1269 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
1273 PL_reg_flags |= RF_tainted;
1280 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
1281 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
1283 tmp = ((OP(c) == BOUND ?
1284 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1285 LOAD_UTF8_CHARCLASS_ALNUM();
1286 REXEC_FBC_UTF8_SCAN(
1287 if (tmp == !(OP(c) == BOUND ?
1288 (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
1289 isALNUM_LC_utf8((U8*)s)))
1297 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
1298 tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
1301 !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
1307 if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s)))
1311 PL_reg_flags |= RF_tainted;
1318 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
1319 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
1321 tmp = ((OP(c) == NBOUND ?
1322 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1323 LOAD_UTF8_CHARCLASS_ALNUM();
1324 REXEC_FBC_UTF8_SCAN(
1325 if (tmp == !(OP(c) == NBOUND ?
1326 (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
1327 isALNUM_LC_utf8((U8*)s)))
1329 else REXEC_FBC_TRYIT;
1333 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
1334 tmp = ((OP(c) == NBOUND ?
1335 isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
1338 !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
1340 else REXEC_FBC_TRYIT;
1343 if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s)))
1347 REXEC_FBC_CSCAN_PRELOAD(
1348 LOAD_UTF8_CHARCLASS_ALNUM(),
1349 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8),
1353 REXEC_FBC_CSCAN_TAINT(
1354 isALNUM_LC_utf8((U8*)s),
1358 REXEC_FBC_CSCAN_PRELOAD(
1359 LOAD_UTF8_CHARCLASS_ALNUM(),
1360 !swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8),
1364 REXEC_FBC_CSCAN_TAINT(
1365 !isALNUM_LC_utf8((U8*)s),
1369 REXEC_FBC_CSCAN_PRELOAD(
1370 LOAD_UTF8_CHARCLASS_SPACE(),
1371 *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8),
1375 REXEC_FBC_CSCAN_TAINT(
1376 *s == ' ' || isSPACE_LC_utf8((U8*)s),
1380 REXEC_FBC_CSCAN_PRELOAD(
1381 LOAD_UTF8_CHARCLASS_SPACE(),
1382 !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)),
1386 REXEC_FBC_CSCAN_TAINT(
1387 !(*s == ' ' || isSPACE_LC_utf8((U8*)s)),
1391 REXEC_FBC_CSCAN_PRELOAD(
1392 LOAD_UTF8_CHARCLASS_DIGIT(),
1393 swash_fetch(PL_utf8_digit,(U8*)s, do_utf8),
1397 REXEC_FBC_CSCAN_TAINT(
1398 isDIGIT_LC_utf8((U8*)s),
1402 REXEC_FBC_CSCAN_PRELOAD(
1403 LOAD_UTF8_CHARCLASS_DIGIT(),
1404 !swash_fetch(PL_utf8_digit,(U8*)s, do_utf8),
1408 REXEC_FBC_CSCAN_TAINT(
1409 !isDIGIT_LC_utf8((U8*)s),
1415 const enum { trie_plain, trie_utf8, trie_utf8_fold }
1416 trie_type = do_utf8 ?
1417 (c->flags == EXACT ? trie_utf8 : trie_utf8_fold)
1419 /* what trie are we using right now */
1421 = (reg_ac_data*)progi->data->data[ ARG( c ) ];
1423 = (reg_trie_data*)progi->data->data[ aho->trie ];
1424 HV *widecharmap = (HV*) progi->data->data[ aho->trie + 1 ];
1426 const char *last_start = strend - trie->minlen;
1428 const char *real_start = s;
1430 STRLEN maxlen = trie->maxlen;
1432 U8 **points; /* map of where we were in the input string
1433 when reading a given char. For ASCII this
1434 is unnecessary overhead as the relationship
1435 is always 1:1, but for unicode, especially
1436 case folded unicode this is not true. */
1437 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1441 GET_RE_DEBUG_FLAGS_DECL;
1443 /* We can't just allocate points here. We need to wrap it in
1444 * an SV so it gets freed properly if there is a croak while
1445 * running the match */
1448 sv_points=newSV(maxlen * sizeof(U8 *));
1449 SvCUR_set(sv_points,
1450 maxlen * sizeof(U8 *));
1451 SvPOK_on(sv_points);
1452 sv_2mortal(sv_points);
1453 points=(U8**)SvPV_nolen(sv_points );
1454 if ( trie_type != trie_utf8_fold
1455 && (trie->bitmap || OP(c)==AHOCORASICKC) )
1458 bitmap=(U8*)trie->bitmap;
1460 bitmap=(U8*)ANYOF_BITMAP(c);
1462 /* this is the Aho-Corasick algorithm modified a touch
1463 to include special handling for long "unknown char"
1464 sequences. The basic idea being that we use AC as long
1465 as we are dealing with a possible matching char, when
1466 we encounter an unknown char (and we have not encountered
1467 an accepting state) we scan forward until we find a legal
1469 AC matching is basically that of trie matching, except
1470 that when we encounter a failing transition, we fall back
1471 to the current states "fail state", and try the current char
1472 again, a process we repeat until we reach the root state,
1473 state 1, or a legal transition. If we fail on the root state
1474 then we can either terminate if we have reached an accepting
1475 state previously, or restart the entire process from the beginning
1479 while (s <= last_start) {
1480 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1488 U8 *uscan = (U8*)NULL;
1489 U8 *leftmost = NULL;
1491 U32 accepted_word= 0;
1495 while ( state && uc <= (U8*)strend ) {
1497 U32 word = aho->states[ state ].wordnum;
1501 DEBUG_TRIE_EXECUTE_r(
1502 if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1503 dump_exec_pos( (char *)uc, c, strend, real_start,
1504 (char *)uc, do_utf8 );
1505 PerlIO_printf( Perl_debug_log,
1506 " Scanning for legal start char...\n");
1509 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1514 if (uc >(U8*)last_start) break;
1518 U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ];
1519 if (!leftmost || lpos < leftmost) {
1520 DEBUG_r(accepted_word=word);
1526 points[pointpos++ % maxlen]= uc;
1527 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
1528 uscan, len, uvc, charid, foldlen,
1530 DEBUG_TRIE_EXECUTE_r({
1531 dump_exec_pos( (char *)uc, c, strend, real_start,
1533 PerlIO_printf(Perl_debug_log,
1534 " Charid:%3u CP:%4"UVxf" ",
1540 word = aho->states[ state ].wordnum;
1542 base = aho->states[ state ].trans.base;
1544 DEBUG_TRIE_EXECUTE_r({
1546 dump_exec_pos( (char *)uc, c, strend, real_start,
1548 PerlIO_printf( Perl_debug_log,
1549 "%sState: %4"UVxf", word=%"UVxf,
1550 failed ? " Fail transition to " : "",
1551 (UV)state, (UV)word);
1556 (base + charid > trie->uniquecharcount )
1557 && (base + charid - 1 - trie->uniquecharcount
1559 && trie->trans[base + charid - 1 -
1560 trie->uniquecharcount].check == state
1561 && (tmp=trie->trans[base + charid - 1 -
1562 trie->uniquecharcount ].next))
1564 DEBUG_TRIE_EXECUTE_r(
1565 PerlIO_printf( Perl_debug_log," - legal\n"));
1570 DEBUG_TRIE_EXECUTE_r(
1571 PerlIO_printf( Perl_debug_log," - fail\n"));
1573 state = aho->fail[state];
1577 /* we must be accepting here */
1578 DEBUG_TRIE_EXECUTE_r(
1579 PerlIO_printf( Perl_debug_log," - accepting\n"));
1588 if (!state) state = 1;
1591 if ( aho->states[ state ].wordnum ) {
1592 U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ];
1593 if (!leftmost || lpos < leftmost) {
1594 DEBUG_r(accepted_word=aho->states[ state ].wordnum);
1599 s = (char*)leftmost;
1600 DEBUG_TRIE_EXECUTE_r({
1602 Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n",
1603 (UV)accepted_word, (IV)(s - real_start)
1606 if (!reginfo || regtry(reginfo, &s)) {
1612 DEBUG_TRIE_EXECUTE_r({
1613 PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n");
1616 DEBUG_TRIE_EXECUTE_r(
1617 PerlIO_printf( Perl_debug_log,"No match.\n"));
1626 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
1635 - regexec_flags - match a regexp against a string
1638 Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend,
1639 char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
1640 /* strend: pointer to null at end of string */
1641 /* strbeg: real beginning of string */
1642 /* minend: end of match must be >=minend after stringarg. */
1643 /* data: May be used for some additional optimizations.
1644 Currently its only used, with a U32 cast, for transmitting
1645 the ganch offset when doing a /g match. This will change */
1646 /* nosave: For optimizations. */
1649 /*register*/ char *s;
1650 register regnode *c;
1651 /*register*/ char *startpos = stringarg;
1652 I32 minlen; /* must match at least this many chars */
1653 I32 dontbother = 0; /* how many characters not to try at end */
1654 I32 end_shift = 0; /* Same for the end. */ /* CC */
1655 I32 scream_pos = -1; /* Internal iterator of scream. */
1656 char *scream_olds = NULL;
1657 SV* const oreplsv = GvSV(PL_replgv);
1658 const bool do_utf8 = (bool)DO_UTF8(sv);
1660 RXi_GET_DECL(prog,progi);
1661 regmatch_info reginfo; /* create some info to pass to regtry etc */
1663 GET_RE_DEBUG_FLAGS_DECL;
1665 PERL_UNUSED_ARG(data);
1667 /* Be paranoid... */
1668 if (prog == NULL || startpos == NULL) {
1669 Perl_croak(aTHX_ "NULL regexp parameter");
1673 multiline = prog->extflags & RXf_PMf_MULTILINE;
1674 reginfo.prog = prog;
1676 RX_MATCH_UTF8_set(prog, do_utf8);
1678 debug_start_match(prog, do_utf8, startpos, strend,
1682 minlen = prog->minlen;
1684 if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
1685 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1686 "String too short [regexec_flags]...\n"));
1691 /* Check validity of program. */
1692 if (UCHARAT(progi->program) != REG_MAGIC) {
1693 Perl_croak(aTHX_ "corrupted regexp program");
1697 PL_reg_eval_set = 0;
1700 if (prog->extflags & RXf_UTF8)
1701 PL_reg_flags |= RF_utf8;
1703 /* Mark beginning of line for ^ and lookbehind. */
1704 reginfo.bol = startpos; /* XXX not used ??? */
1708 /* Mark end of line for $ (and such) */
1711 /* see how far we have to get to not match where we matched before */
1712 reginfo.till = startpos+minend;
1714 /* If there is a "must appear" string, look for it. */
1717 if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */
1720 if (flags & REXEC_IGNOREPOS) /* Means: check only at start */
1721 reginfo.ganch = startpos + prog->gofs;
1722 else if (sv && SvTYPE(sv) >= SVt_PVMG
1724 && (mg = mg_find(sv, PERL_MAGIC_regex_global))
1725 && mg->mg_len >= 0) {
1726 reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */
1727 if (prog->extflags & RXf_ANCH_GPOS) {
1728 if (s > reginfo.ganch)
1730 s = reginfo.ganch - prog->gofs;
1734 reginfo.ganch = strbeg + PTR2UV(data);
1735 } else /* pos() not defined */
1736 reginfo.ganch = strbeg;
1738 if (PL_curpm && (PM_GETRE(PL_curpm) == prog)) {
1741 /* We have to be careful. If the previous successful match
1742 was from this regex we don't want a subsequent paritally
1743 successful match to clobber the old results.
1744 So when we detect this possibility we add a swap buffer
1745 to the re, and switch the buffer each match. If we fail
1746 we switch it back, otherwise we leave it swapped.
1748 Newxz(progi->swap, 1, regexp_paren_ofs);
1749 /* no need to copy these */
1750 Newxz(progi->swap->startp, prog->nparens + 1, I32);
1751 Newxz(progi->swap->endp, prog->nparens + 1, I32);
1753 t = progi->swap->startp;
1754 progi->swap->startp = prog->startp;
1756 t = progi->swap->endp;
1757 progi->swap->endp = prog->endp;
1760 if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) {
1761 re_scream_pos_data d;
1763 d.scream_olds = &scream_olds;
1764 d.scream_pos = &scream_pos;
1765 s = re_intuit_start(prog, sv, s, strend, flags, &d);
1767 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
1768 goto phooey; /* not present */
1774 /* Simplest case: anchored match need be tried only once. */
1775 /* [unless only anchor is BOL and multiline is set] */
1776 if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) {
1777 if (s == startpos && regtry(®info, &startpos))
1779 else if (multiline || (prog->intflags & PREGf_IMPLICIT)
1780 || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */
1785 dontbother = minlen - 1;
1786 end = HOP3c(strend, -dontbother, strbeg) - 1;
1787 /* for multiline we only have to try after newlines */
1788 if (prog->check_substr || prog->check_utf8) {
1792 if (regtry(®info, &s))
1797 if (prog->extflags & RXf_USE_INTUIT) {
1798 s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL);
1809 if (*s++ == '\n') { /* don't need PL_utf8skip here */
1810 if (regtry(®info, &s))
1817 } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK))
1819 /* the warning about reginfo.ganch being used without intialization
1820 is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN
1821 and we only enter this block when the same bit is set. */
1822 char *tmp_s = reginfo.ganch - prog->gofs;
1823 if (regtry(®info, &tmp_s))
1828 /* Messy cases: unanchored match. */
1829 if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) {
1830 /* we have /x+whatever/ */
1831 /* it must be a one character string (XXXX Except UTF?) */
1836 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
1837 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1838 ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0];
1843 DEBUG_EXECUTE_r( did_match = 1 );
1844 if (regtry(®info, &s)) goto got_it;
1846 while (s < strend && *s == ch)
1854 DEBUG_EXECUTE_r( did_match = 1 );
1855 if (regtry(®info, &s)) goto got_it;
1857 while (s < strend && *s == ch)
1862 DEBUG_EXECUTE_r(if (!did_match)
1863 PerlIO_printf(Perl_debug_log,
1864 "Did not find anchored character...\n")
1867 else if (prog->anchored_substr != NULL
1868 || prog->anchored_utf8 != NULL
1869 || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
1870 && prog->float_max_offset < strend - s)) {
1875 char *last1; /* Last position checked before */
1879 if (prog->anchored_substr || prog->anchored_utf8) {
1880 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
1881 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1882 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
1883 back_max = back_min = prog->anchored_offset;
1885 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
1886 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1887 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
1888 back_max = prog->float_max_offset;
1889 back_min = prog->float_min_offset;
1893 if (must == &PL_sv_undef)
1894 /* could not downgrade utf8 check substring, so must fail */
1900 last = HOP3c(strend, /* Cannot start after this */
1901 -(I32)(CHR_SVLEN(must)
1902 - (SvTAIL(must) != 0) + back_min), strbeg);
1905 last1 = HOPc(s, -1);
1907 last1 = s - 1; /* bogus */
1909 /* XXXX check_substr already used to find "s", can optimize if
1910 check_substr==must. */
1912 dontbother = end_shift;
1913 strend = HOPc(strend, -dontbother);
1914 while ( (s <= last) &&
1915 ((flags & REXEC_SCREAM)
1916 ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg,
1917 end_shift, &scream_pos, 0))
1918 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)),
1919 (unsigned char*)strend, must,
1920 multiline ? FBMrf_MULTILINE : 0))) ) {
1921 /* we may be pointing at the wrong string */
1922 if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog))
1923 s = strbeg + (s - SvPVX_const(sv));
1924 DEBUG_EXECUTE_r( did_match = 1 );
1925 if (HOPc(s, -back_max) > last1) {
1926 last1 = HOPc(s, -back_min);
1927 s = HOPc(s, -back_max);
1930 char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
1932 last1 = HOPc(s, -back_min);
1936 while (s <= last1) {
1937 if (regtry(®info, &s))
1943 while (s <= last1) {
1944 if (regtry(®info, &s))
1950 DEBUG_EXECUTE_r(if (!did_match) {
1951 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
1952 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
1953 PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n",
1954 ((must == prog->anchored_substr || must == prog->anchored_utf8)
1955 ? "anchored" : "floating"),
1956 quoted, RE_SV_TAIL(must));
1960 else if ( (c = progi->regstclass) ) {
1962 const OPCODE op = OP(progi->regstclass);
1963 /* don't bother with what can't match */
1964 if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE)
1965 strend = HOPc(strend, -(minlen - 1));
1968 SV * const prop = sv_newmortal();
1969 regprop(prog, prop, c);
1971 RE_PV_QUOTED_DECL(quoted,UTF,PERL_DEBUG_PAD_ZERO(1),
1973 PerlIO_printf(Perl_debug_log,
1974 "Matching stclass %.*s against %s (%d chars)\n",
1975 (int)SvCUR(prop), SvPVX_const(prop),
1976 quoted, (int)(strend - s));
1979 if (find_byclass(prog, c, s, strend, ®info))
1981 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n"));
1985 if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
1990 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
1991 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
1992 float_real = do_utf8 ? prog->float_utf8 : prog->float_substr;
1994 if (flags & REXEC_SCREAM) {
1995 last = screaminstr(sv, float_real, s - strbeg,
1996 end_shift, &scream_pos, 1); /* last one */
1998 last = scream_olds; /* Only one occurrence. */
1999 /* we may be pointing at the wrong string */
2000 else if (RX_MATCH_COPIED(prog))
2001 s = strbeg + (s - SvPVX_const(sv));
2005 const char * const little = SvPV_const(float_real, len);
2007 if (SvTAIL(float_real)) {
2008 if (memEQ(strend - len + 1, little, len - 1))
2009 last = strend - len + 1;
2010 else if (!multiline)
2011 last = memEQ(strend - len, little, len)
2012 ? strend - len : NULL;
2018 last = rninstr(s, strend, little, little + len);
2020 last = strend; /* matching "$" */
2025 PerlIO_printf(Perl_debug_log,
2026 "%sCan't trim the tail, match fails (should not happen)%s\n",
2027 PL_colors[4], PL_colors[5]));
2028 goto phooey; /* Should not happen! */
2030 dontbother = strend - last + prog->float_min_offset;
2032 if (minlen && (dontbother < minlen))
2033 dontbother = minlen - 1;
2034 strend -= dontbother; /* this one's always in bytes! */
2035 /* We don't know much -- general case. */
2038 if (regtry(®info, &s))
2047 if (regtry(®info, &s))
2049 } while (s++ < strend);
2057 RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted);
2059 if (PL_reg_eval_set) {
2060 /* Preserve the current value of $^R */
2061 if (oreplsv != GvSV(PL_replgv))
2062 sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
2063 restored, the value remains
2065 restore_pos(aTHX_ prog);
2067 if (prog->paren_names)
2068 (void)hv_iterinit(prog->paren_names);
2070 /* make sure $`, $&, $', and $digit will work later */
2071 if ( !(flags & REXEC_NOT_FIRST) ) {
2072 RX_MATCH_COPY_FREE(prog);
2073 if (flags & REXEC_COPY_STR) {
2074 const I32 i = PL_regeol - startpos + (stringarg - strbeg);
2075 #ifdef PERL_OLD_COPY_ON_WRITE
2077 || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
2079 PerlIO_printf(Perl_debug_log,
2080 "Copy on write: regexp capture, type %d\n",
2083 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
2084 prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
2085 assert (SvPOKp(prog->saved_copy));
2089 RX_MATCH_COPIED_on(prog);
2090 s = savepvn(strbeg, i);
2096 prog->subbeg = strbeg;
2097 prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
2104 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
2105 PL_colors[4], PL_colors[5]));
2106 if (PL_reg_eval_set)
2107 restore_pos(aTHX_ prog);
2109 /* we failed :-( roll it back */
2111 t = progi->swap->startp;
2112 progi->swap->startp = prog->startp;
2114 t = progi->swap->endp;
2115 progi->swap->endp = prog->endp;
2123 - regtry - try match at specific point
2125 STATIC I32 /* 0 failure, 1 success */
2126 S_regtry(pTHX_ regmatch_info *reginfo, char **startpos)
2132 regexp *prog = reginfo->prog;
2133 RXi_GET_DECL(prog,progi);
2134 GET_RE_DEBUG_FLAGS_DECL;
2135 reginfo->cutpoint=NULL;
2137 if ((prog->extflags & RXf_EVAL_SEEN) && !PL_reg_eval_set) {
2140 PL_reg_eval_set = RS_init;
2141 DEBUG_EXECUTE_r(DEBUG_s(
2142 PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
2143 (IV)(PL_stack_sp - PL_stack_base));
2146 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
2147 /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
2149 /* Apparently this is not needed, judging by wantarray. */
2150 /* SAVEI8(cxstack[cxstack_ix].blk_gimme);
2151 cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
2154 /* Make $_ available to executed code. */
2155 if (reginfo->sv != DEFSV) {
2157 DEFSV = reginfo->sv;
2160 if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv)
2161 && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) {
2162 /* prepare for quick setting of pos */
2163 #ifdef PERL_OLD_COPY_ON_WRITE
2164 if (SvIsCOW(reginfo->sv))
2165 sv_force_normal_flags(reginfo->sv, 0);
2167 mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global,
2168 &PL_vtbl_mglob, NULL, 0);
2172 PL_reg_oldpos = mg->mg_len;
2173 SAVEDESTRUCTOR_X(restore_pos, prog);
2175 if (!PL_reg_curpm) {
2176 Newxz(PL_reg_curpm, 1, PMOP);
2179 SV* const repointer = newSViv(0);
2180 /* so we know which PL_regex_padav element is PL_reg_curpm */
2181 SvFLAGS(repointer) |= SVf_BREAK;
2182 av_push(PL_regex_padav,repointer);
2183 PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav);
2184 PL_regex_pad = AvARRAY(PL_regex_padav);
2188 PM_SETRE(PL_reg_curpm, prog);
2189 PL_reg_oldcurpm = PL_curpm;
2190 PL_curpm = PL_reg_curpm;
2191 if (RX_MATCH_COPIED(prog)) {
2192 /* Here is a serious problem: we cannot rewrite subbeg,
2193 since it may be needed if this match fails. Thus
2194 $` inside (?{}) could fail... */
2195 PL_reg_oldsaved = prog->subbeg;
2196 PL_reg_oldsavedlen = prog->sublen;
2197 #ifdef PERL_OLD_COPY_ON_WRITE
2198 PL_nrs = prog->saved_copy;
2200 RX_MATCH_COPIED_off(prog);
2203 PL_reg_oldsaved = NULL;
2204 prog->subbeg = PL_bostr;
2205 prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
2207 DEBUG_EXECUTE_r(PL_reg_starttry = *startpos);
2208 prog->startp[0] = *startpos - PL_bostr;
2209 PL_reginput = *startpos;
2210 PL_reglastparen = &prog->lastparen;
2211 PL_reglastcloseparen = &prog->lastcloseparen;
2212 prog->lastparen = 0;
2213 prog->lastcloseparen = 0;
2215 PL_regstartp = prog->startp;
2216 PL_regendp = prog->endp;
2217 if (PL_reg_start_tmpl <= prog->nparens) {
2218 PL_reg_start_tmpl = prog->nparens*3/2 + 3;
2219 if(PL_reg_start_tmp)
2220 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2222 Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2225 /* XXXX What this code is doing here?!!! There should be no need
2226 to do this again and again, PL_reglastparen should take care of
2229 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
2230 * Actually, the code in regcppop() (which Ilya may be meaning by
2231 * PL_reglastparen), is not needed at all by the test suite
2232 * (op/regexp, op/pat, op/split), but that code is needed, oddly
2233 * enough, for building DynaLoader, or otherwise this
2234 * "Error: '*' not in typemap in DynaLoader.xs, line 164"
2235 * will happen. Meanwhile, this code *is* needed for the
2236 * above-mentioned test suite tests to succeed. The common theme
2237 * on those tests seems to be returning null fields from matches.
2242 if (prog->nparens) {
2244 for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) {
2251 if (regmatch(reginfo, progi->program + 1)) {
2252 PL_regendp[0] = PL_reginput - PL_bostr;
2255 if (reginfo->cutpoint)
2256 *startpos= reginfo->cutpoint;
2257 REGCP_UNWIND(lastcp);
2262 #define sayYES goto yes
2263 #define sayNO goto no
2264 #define sayNO_SILENT goto no_silent
2266 /* we dont use STMT_START/END here because it leads to
2267 "unreachable code" warnings, which are bogus, but distracting. */
2268 #define CACHEsayNO \
2269 if (ST.cache_mask) \
2270 PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \
2273 /* this is used to determine how far from the left messages like
2274 'failed...' are printed. It should be set such that messages
2275 are inline with the regop output that created them.
2277 #define REPORT_CODE_OFF 32
2280 /* Make sure there is a test for this +1 options in re_tests */
2281 #define TRIE_INITAL_ACCEPT_BUFFLEN 4;
2283 #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
2284 #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */
2286 #define SLAB_FIRST(s) (&(s)->states[0])
2287 #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
2289 /* grab a new slab and return the first slot in it */
2291 STATIC regmatch_state *
2294 #if PERL_VERSION < 9
2297 regmatch_slab *s = PL_regmatch_slab->next;
2299 Newx(s, 1, regmatch_slab);
2300 s->prev = PL_regmatch_slab;
2302 PL_regmatch_slab->next = s;
2304 PL_regmatch_slab = s;
2305 return SLAB_FIRST(s);
2309 /* push a new state then goto it */
2311 #define PUSH_STATE_GOTO(state, node) \
2313 st->resume_state = state; \
2316 /* push a new state with success backtracking, then goto it */
2318 #define PUSH_YES_STATE_GOTO(state, node) \
2320 st->resume_state = state; \
2321 goto push_yes_state;
2327 regmatch() - main matching routine
2329 This is basically one big switch statement in a loop. We execute an op,
2330 set 'next' to point the next op, and continue. If we come to a point which
2331 we may need to backtrack to on failure such as (A|B|C), we push a
2332 backtrack state onto the backtrack stack. On failure, we pop the top
2333 state, and re-enter the loop at the state indicated. If there are no more
2334 states to pop, we return failure.
2336 Sometimes we also need to backtrack on success; for example /A+/, where
2337 after successfully matching one A, we need to go back and try to
2338 match another one; similarly for lookahead assertions: if the assertion
2339 completes successfully, we backtrack to the state just before the assertion
2340 and then carry on. In these cases, the pushed state is marked as
2341 'backtrack on success too'. This marking is in fact done by a chain of
2342 pointers, each pointing to the previous 'yes' state. On success, we pop to
2343 the nearest yes state, discarding any intermediate failure-only states.
2344 Sometimes a yes state is pushed just to force some cleanup code to be
2345 called at the end of a successful match or submatch; e.g. (??{$re}) uses
2346 it to free the inner regex.
2348 Note that failure backtracking rewinds the cursor position, while
2349 success backtracking leaves it alone.
2351 A pattern is complete when the END op is executed, while a subpattern
2352 such as (?=foo) is complete when the SUCCESS op is executed. Both of these
2353 ops trigger the "pop to last yes state if any, otherwise return true"
2356 A common convention in this function is to use A and B to refer to the two
2357 subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is
2358 the subpattern to be matched possibly multiple times, while B is the entire
2359 rest of the pattern. Variable and state names reflect this convention.
2361 The states in the main switch are the union of ops and failure/success of
2362 substates associated with with that op. For example, IFMATCH is the op
2363 that does lookahead assertions /(?=A)B/ and so the IFMATCH state means
2364 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just
2365 successfully matched A and IFMATCH_A_fail is a state saying that we have
2366 just failed to match A. Resume states always come in pairs. The backtrack
2367 state we push is marked as 'IFMATCH_A', but when that is popped, we resume
2368 at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking
2369 on success or failure.
2371 The struct that holds a backtracking state is actually a big union, with
2372 one variant for each major type of op. The variable st points to the
2373 top-most backtrack struct. To make the code clearer, within each
2374 block of code we #define ST to alias the relevant union.
2376 Here's a concrete example of a (vastly oversimplified) IFMATCH
2382 #define ST st->u.ifmatch
2384 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2385 ST.foo = ...; // some state we wish to save
2387 // push a yes backtrack state with a resume value of
2388 // IFMATCH_A/IFMATCH_A_fail, then continue execution at the
2390 PUSH_YES_STATE_GOTO(IFMATCH_A, A);
2393 case IFMATCH_A: // we have successfully executed A; now continue with B
2395 bar = ST.foo; // do something with the preserved value
2398 case IFMATCH_A_fail: // A failed, so the assertion failed
2399 ...; // do some housekeeping, then ...
2400 sayNO; // propagate the failure
2407 For any old-timers reading this who are familiar with the old recursive
2408 approach, the code above is equivalent to:
2410 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2419 ...; // do some housekeeping, then ...
2420 sayNO; // propagate the failure
2423 The topmost backtrack state, pointed to by st, is usually free. If you
2424 want to claim it, populate any ST.foo fields in it with values you wish to
2425 save, then do one of
2427 PUSH_STATE_GOTO(resume_state, node);
2428 PUSH_YES_STATE_GOTO(resume_state, node);
2430 which sets that backtrack state's resume value to 'resume_state', pushes a
2431 new free entry to the top of the backtrack stack, then goes to 'node'.
2432 On backtracking, the free slot is popped, and the saved state becomes the
2433 new free state. An ST.foo field in this new top state can be temporarily
2434 accessed to retrieve values, but once the main loop is re-entered, it
2435 becomes available for reuse.
2437 Note that the depth of the backtrack stack constantly increases during the
2438 left-to-right execution of the pattern, rather than going up and down with
2439 the pattern nesting. For example the stack is at its maximum at Z at the
2440 end of the pattern, rather than at X in the following:
2442 /(((X)+)+)+....(Y)+....Z/
2444 The only exceptions to this are lookahead/behind assertions and the cut,
2445 (?>A), which pop all the backtrack states associated with A before
2448 Bascktrack state structs are allocated in slabs of about 4K in size.
2449 PL_regmatch_state and st always point to the currently active state,
2450 and PL_regmatch_slab points to the slab currently containing
2451 PL_regmatch_state. The first time regmatch() is called, the first slab is
2452 allocated, and is never freed until interpreter destruction. When the slab
2453 is full, a new one is allocated and chained to the end. At exit from
2454 regmatch(), slabs allocated since entry are freed.
2459 #define DEBUG_STATE_pp(pp) \
2461 DUMP_EXEC_POS(locinput, scan, do_utf8); \
2462 PerlIO_printf(Perl_debug_log, \
2463 " %*s"pp" %s%s%s%s%s\n", \
2465 reg_name[st->resume_state], \
2466 ((st==yes_state||st==mark_state) ? "[" : ""), \
2467 ((st==yes_state) ? "Y" : ""), \
2468 ((st==mark_state) ? "M" : ""), \
2469 ((st==yes_state||st==mark_state) ? "]" : "") \
2474 #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1)
2479 S_debug_start_match(pTHX_ const regexp *prog, const bool do_utf8,
2480 const char *start, const char *end, const char *blurb)
2482 const bool utf8_pat= prog->extflags & RXf_UTF8 ? 1 : 0;
2486 RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0),
2487 prog->precomp, prog->prelen, 60);
2489 RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1),
2490 start, end - start, 60);
2492 PerlIO_printf(Perl_debug_log,
2493 "%s%s REx%s %s against %s\n",
2494 PL_colors[4], blurb, PL_colors[5], s0, s1);
2496 if (do_utf8||utf8_pat)
2497 PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n",
2498 utf8_pat ? "pattern" : "",
2499 utf8_pat && do_utf8 ? " and " : "",
2500 do_utf8 ? "string" : ""
2506 S_dump_exec_pos(pTHX_ const char *locinput,
2507 const regnode *scan,
2508 const char *loc_regeol,
2509 const char *loc_bostr,
2510 const char *loc_reg_starttry,
2513 const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
2514 const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
2515 int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
2516 /* The part of the string before starttry has one color
2517 (pref0_len chars), between starttry and current
2518 position another one (pref_len - pref0_len chars),
2519 after the current position the third one.
2520 We assume that pref0_len <= pref_len, otherwise we
2521 decrease pref0_len. */
2522 int pref_len = (locinput - loc_bostr) > (5 + taill) - l
2523 ? (5 + taill) - l : locinput - loc_bostr;
2526 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
2528 pref0_len = pref_len - (locinput - loc_reg_starttry);
2529 if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
2530 l = ( loc_regeol - locinput > (5 + taill) - pref_len
2531 ? (5 + taill) - pref_len : loc_regeol - locinput);
2532 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
2536 if (pref0_len > pref_len)
2537 pref0_len = pref_len;
2539 const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0;
2541 RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
2542 (locinput - pref_len),pref0_len, 60, 4, 5);
2544 RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
2545 (locinput - pref_len + pref0_len),
2546 pref_len - pref0_len, 60, 2, 3);
2548 RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
2549 locinput, loc_regeol - locinput, 10, 0, 1);
2551 const STRLEN tlen=len0+len1+len2;
2552 PerlIO_printf(Perl_debug_log,
2553 "%4"IVdf" <%.*s%.*s%s%.*s>%*s|",
2554 (IV)(locinput - loc_bostr),
2557 (docolor ? "" : "> <"),
2559 (int)(tlen > 19 ? 0 : 19 - tlen),
2566 /* reg_check_named_buff_matched()
2567 * Checks to see if a named buffer has matched. The data array of
2568 * buffer numbers corresponding to the buffer is expected to reside
2569 * in the regexp->data->data array in the slot stored in the ARG() of
2570 * node involved. Note that this routine doesn't actually care about the
2571 * name, that information is not preserved from compilation to execution.
2572 * Returns the index of the leftmost defined buffer with the given name
2573 * or 0 if non of the buffers matched.
2576 S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan) {
2578 RXi_GET_DECL(rex,rexi);
2579 SV *sv_dat=(SV*)rexi->data->data[ ARG( scan ) ];
2580 I32 *nums=(I32*)SvPVX(sv_dat);
2581 for ( n=0; n<SvIVX(sv_dat); n++ ) {
2582 if ((I32)*PL_reglastparen >= nums[n] &&
2583 PL_regendp[nums[n]] != -1)
2591 STATIC I32 /* 0 failure, 1 success */
2592 S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
2594 #if PERL_VERSION < 9
2598 register const bool do_utf8 = PL_reg_match_utf8;
2599 const U32 uniflags = UTF8_ALLOW_DEFAULT;
2601 regexp *rex = reginfo->prog;
2602 RXi_GET_DECL(rex,rexi);
2604 regmatch_slab *orig_slab;
2605 regmatch_state *orig_state;
2607 /* the current state. This is a cached copy of PL_regmatch_state */
2608 register regmatch_state *st;
2610 /* cache heavy used fields of st in registers */
2611 register regnode *scan;
2612 register regnode *next;
2613 register U32 n = 0; /* general value; init to avoid compiler warning */
2614 register I32 ln = 0; /* len or last; init to avoid compiler warning */
2615 register char *locinput = PL_reginput;
2616 register I32 nextchr; /* is always set to UCHARAT(locinput) */
2618 bool result = 0; /* return value of S_regmatch */
2619 int depth = 0; /* depth of backtrack stack */
2620 int nochange_depth = 0; /* depth of GOSUB recursion with nochange*/
2621 regmatch_state *yes_state = NULL; /* state to pop to on success of
2623 /* mark_state piggy backs on the yes_state logic so that when we unwind
2624 the stack on success we can update the mark_state as we go */
2625 regmatch_state *mark_state = NULL; /* last mark state we have seen */
2627 regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */
2628 struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */
2630 bool no_final = 0; /* prevent failure from backtracking? */
2631 bool do_cutgroup = 0; /* no_final only until next branch/trie entry */
2632 char *startpoint = PL_reginput;
2633 SV *popmark = NULL; /* are we looking for a mark? */
2634 SV *sv_commit = NULL; /* last mark name seen in failure */
2635 SV *sv_yes_mark = NULL; /* last mark name we have seen
2636 during a successfull match */
2637 U32 lastopen = 0; /* last open we saw */
2638 bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0;
2641 /* these three flags are set by various ops to signal information to
2642 * the very next op. They have a useful lifetime of exactly one loop
2643 * iteration, and are not preserved or restored by state pushes/pops
2645 bool sw = 0; /* the condition value in (?(cond)a|b) */
2646 bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */
2647 int logical = 0; /* the following EVAL is:
2651 or the following IFMATCH/UNLESSM is:
2652 false: plain (?=foo)
2653 true: used as a condition: (?(?=foo))
2657 GET_RE_DEBUG_FLAGS_DECL;
2661 PerlIO_printf(Perl_debug_log,"regmatch start\n");
2663 /* on first ever call to regmatch, allocate first slab */
2664 if (!PL_regmatch_slab) {
2665 Newx(PL_regmatch_slab, 1, regmatch_slab);
2666 PL_regmatch_slab->prev = NULL;
2667 PL_regmatch_slab->next = NULL;
2668 PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
2671 /* remember current high-water mark for exit */
2672 /* XXX this should be done with SAVE* instead */
2673 orig_slab = PL_regmatch_slab;
2674 orig_state = PL_regmatch_state;
2676 /* grab next free state slot */
2677 st = ++PL_regmatch_state;
2678 if (st > SLAB_LAST(PL_regmatch_slab))
2679 st = PL_regmatch_state = S_push_slab(aTHX);
2681 /* Note that nextchr is a byte even in UTF */
2682 nextchr = UCHARAT(locinput);
2684 while (scan != NULL) {
2687 SV * const prop = sv_newmortal();
2688 regnode *rnext=regnext(scan);
2689 DUMP_EXEC_POS( locinput, scan, do_utf8 );
2690 regprop(rex, prop, scan);
2692 PerlIO_printf(Perl_debug_log,
2693 "%3"IVdf":%*s%s(%"IVdf")\n",
2694 (IV)(scan - rexi->program), depth*2, "",
2696 (PL_regkind[OP(scan)] == END || !rnext) ?
2697 0 : (IV)(rnext - rexi->program));
2700 next = scan + NEXT_OFF(scan);
2703 state_num = OP(scan);
2706 switch (state_num) {
2708 if (locinput == PL_bostr)
2710 /* reginfo->till = reginfo->bol; */
2715 if (locinput == PL_bostr ||
2716 ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n'))
2722 if (locinput == PL_bostr)
2726 if (locinput == reginfo->ganch)
2732 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
2737 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
2739 if (PL_regeol - locinput > 1)
2743 if (PL_regeol != locinput)
2747 if (!nextchr && locinput >= PL_regeol)
2750 locinput += PL_utf8skip[nextchr];
2751 if (locinput > PL_regeol)
2753 nextchr = UCHARAT(locinput);
2756 nextchr = UCHARAT(++locinput);
2759 if (!nextchr && locinput >= PL_regeol)
2761 nextchr = UCHARAT(++locinput);
2764 if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
2767 locinput += PL_utf8skip[nextchr];
2768 if (locinput > PL_regeol)
2770 nextchr = UCHARAT(locinput);
2773 nextchr = UCHARAT(++locinput);
2777 #define ST st->u.trie
2779 /* In this case the charclass data is available inline so
2780 we can fail fast without a lot of extra overhead.
2782 if (scan->flags == EXACT || !do_utf8) {
2783 if(!ANYOF_BITMAP_TEST(scan, *locinput)) {
2785 PerlIO_printf(Perl_debug_log,
2786 "%*s %sfailed to match trie start class...%s\n",
2787 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
2796 /* what type of TRIE am I? (utf8 makes this contextual) */
2797 const enum { trie_plain, trie_utf8, trie_utf8_fold }
2798 trie_type = do_utf8 ?
2799 (scan->flags == EXACT ? trie_utf8 : trie_utf8_fold)
2802 /* what trie are we using right now */
2803 reg_trie_data * const trie
2804 = (reg_trie_data*)rexi->data->data[ ARG( scan ) ];
2805 HV * widecharmap = (HV *)rexi->data->data[ ARG( scan ) + 1 ];
2806 U32 state = trie->startstate;
2808 if (trie->bitmap && trie_type != trie_utf8_fold &&
2809 !TRIE_BITMAP_TEST(trie,*locinput)
2811 if (trie->states[ state ].wordnum) {
2813 PerlIO_printf(Perl_debug_log,
2814 "%*s %smatched empty string...%s\n",
2815 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
2820 PerlIO_printf(Perl_debug_log,
2821 "%*s %sfailed to match trie start class...%s\n",
2822 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
2829 U8 *uc = ( U8* )locinput;
2833 U8 *uscan = (U8*)NULL;
2835 SV *sv_accept_buff = NULL;
2836 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
2838 ST.accepted = 0; /* how many accepting states we have seen */
2840 ST.jump = trie->jump;
2844 traverse the TRIE keeping track of all accepting states
2845 we transition through until we get to a failing node.
2848 while ( state && uc <= (U8*)PL_regeol ) {
2849 U32 base = trie->states[ state ].trans.base;
2852 /* We use charid to hold the wordnum as we don't use it
2853 for charid until after we have done the wordnum logic.
2854 We define an alias just so that the wordnum logic reads
2857 #define got_wordnum charid
2858 got_wordnum = trie->states[ state ].wordnum;
2860 if ( got_wordnum ) {
2861 if ( ! ST.accepted ) {
2864 bufflen = TRIE_INITAL_ACCEPT_BUFFLEN;
2865 sv_accept_buff=newSV(bufflen *
2866 sizeof(reg_trie_accepted) - 1);
2867 SvCUR_set(sv_accept_buff, 0);
2868 SvPOK_on(sv_accept_buff);
2869 sv_2mortal(sv_accept_buff);
2872 (reg_trie_accepted*)SvPV_nolen(sv_accept_buff );
2875 if (ST.accepted >= bufflen) {
2877 ST.accept_buff =(reg_trie_accepted*)
2878 SvGROW(sv_accept_buff,
2879 bufflen * sizeof(reg_trie_accepted));
2881 SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff)
2882 + sizeof(reg_trie_accepted));
2885 ST.accept_buff[ST.accepted].wordnum = got_wordnum;
2886 ST.accept_buff[ST.accepted].endpos = uc;
2888 } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum]));
2892 DEBUG_TRIE_EXECUTE_r({
2893 DUMP_EXEC_POS( (char *)uc, scan, do_utf8 );
2894 PerlIO_printf( Perl_debug_log,
2895 "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ",
2896 2+depth * 2, "", PL_colors[4],
2897 (UV)state, (UV)ST.accepted );
2901 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
2902 uscan, len, uvc, charid, foldlen,
2906 (base + charid > trie->uniquecharcount )
2907 && (base + charid - 1 - trie->uniquecharcount
2909 && trie->trans[base + charid - 1 -
2910 trie->uniquecharcount].check == state)
2912 state = trie->trans[base + charid - 1 -
2913 trie->uniquecharcount ].next;
2924 DEBUG_TRIE_EXECUTE_r(
2925 PerlIO_printf( Perl_debug_log,
2926 "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n",
2927 charid, uvc, (UV)state, PL_colors[5] );
2934 PerlIO_printf( Perl_debug_log,
2935 "%*s %sgot %"IVdf" possible matches%s\n",
2936 REPORT_CODE_OFF + depth * 2, "",
2937 PL_colors[4], (IV)ST.accepted, PL_colors[5] );
2942 case TRIE_next_fail: /* we failed - try next alterative */
2947 if ( ST.accepted == 1 ) {
2948 /* only one choice left - just continue */
2950 AV *const trie_words
2951 = (AV *) rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET];
2952 SV ** const tmp = av_fetch( trie_words,
2953 ST.accept_buff[ 0 ].wordnum-1, 0 );
2954 SV *sv= tmp ? sv_newmortal() : NULL;
2956 PerlIO_printf( Perl_debug_log,
2957 "%*s %sonly one match left: #%d <%s>%s\n",
2958 REPORT_CODE_OFF+depth*2, "", PL_colors[4],
2959 ST.accept_buff[ 0 ].wordnum,
2960 tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
2961 PL_colors[0], PL_colors[1],
2962 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)
2964 : "not compiled under -Dr",
2967 PL_reginput = (char *)ST.accept_buff[ 0 ].endpos;
2968 /* in this case we free tmps/leave before we call regmatch
2969 as we wont be using accept_buff again. */
2971 locinput = PL_reginput;
2972 nextchr = UCHARAT(locinput);
2973 if ( !ST.jump || !ST.jump[ST.accept_buff[0].wordnum])
2976 scan = ST.me + ST.jump[ST.accept_buff[0].wordnum];
2977 if (!has_cutgroup) {
2982 PUSH_YES_STATE_GOTO(TRIE_next, scan);
2985 continue; /* execute rest of RE */
2988 if (!ST.accepted-- ) {
2990 PerlIO_printf( Perl_debug_log,
2991 "%*s %sTRIE failed...%s\n",
2992 REPORT_CODE_OFF+depth*2, "",
3002 There are at least two accepting states left. Presumably
3003 the number of accepting states is going to be low,
3004 typically two. So we simply scan through to find the one
3005 with lowest wordnum. Once we find it, we swap the last
3006 state into its place and decrement the size. We then try to
3007 match the rest of the pattern at the point where the word
3008 ends. If we succeed, control just continues along the
3009 regex; if we fail we return here to try the next accepting
3016 for( cur = 1 ; cur <= ST.accepted ; cur++ ) {
3017 DEBUG_TRIE_EXECUTE_r(
3018 PerlIO_printf( Perl_debug_log,
3019 "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n",
3020 REPORT_CODE_OFF + depth * 2, "", PL_colors[4],
3021 (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur,
3022 ST.accept_buff[ cur ].wordnum, PL_colors[5] );
3025 if (ST.accept_buff[cur].wordnum <
3026 ST.accept_buff[best].wordnum)
3031 AV *const trie_words
3032 = (AV *) rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET];
3033 SV ** const tmp = av_fetch( trie_words,
3034 ST.accept_buff[ best ].wordnum - 1, 0 );
3035 regnode *nextop=(!ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) ?
3037 ST.me + ST.jump[ST.accept_buff[best].wordnum];
3038 SV *sv= tmp ? sv_newmortal() : NULL;
3040 PerlIO_printf( Perl_debug_log,
3041 "%*s %strying alternation #%d <%s> at node #%d %s\n",
3042 REPORT_CODE_OFF+depth*2, "", PL_colors[4],
3043 ST.accept_buff[best].wordnum,
3044 tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
3045 PL_colors[0], PL_colors[1],
3046 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)
3047 ) : "not compiled under -Dr",
3048 REG_NODE_NUM(nextop),
3052 if ( best<ST.accepted ) {
3053 reg_trie_accepted tmp = ST.accept_buff[ best ];
3054 ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ];
3055 ST.accept_buff[ ST.accepted ] = tmp;
3058 PL_reginput = (char *)ST.accept_buff[ best ].endpos;
3059 if ( !ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) {
3063 scan = ST.me + ST.jump[ST.accept_buff[best].wordnum];
3067 PUSH_YES_STATE_GOTO(TRIE_next, scan);
3070 PUSH_STATE_GOTO(TRIE_next, scan);
3083 char *s = STRING(scan);
3085 if (do_utf8 != UTF) {
3086 /* The target and the pattern have differing utf8ness. */
3088 const char * const e = s + ln;
3091 /* The target is utf8, the pattern is not utf8. */
3096 if (NATIVE_TO_UNI(*(U8*)s) !=
3097 utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
3105 /* The target is not utf8, the pattern is utf8. */
3110 if (NATIVE_TO_UNI(*((U8*)l)) !=
3111 utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
3119 nextchr = UCHARAT(locinput);
3122 /* The target and the pattern have the same utf8ness. */
3123 /* Inline the first character, for speed. */
3124 if (UCHARAT(s) != nextchr)
3126 if (PL_regeol - locinput < ln)
3128 if (ln > 1 && memNE(s, locinput, ln))
3131 nextchr = UCHARAT(locinput);
3135 PL_reg_flags |= RF_tainted;
3138 char * const s = STRING(scan);
3141 if (do_utf8 || UTF) {
3142 /* Either target or the pattern are utf8. */
3143 const char * const l = locinput;
3144 char *e = PL_regeol;
3146 if (ibcmp_utf8(s, 0, ln, (bool)UTF,
3147 l, &e, 0, do_utf8)) {
3148 /* One more case for the sharp s:
3149 * pack("U0U*", 0xDF) =~ /ss/i,
3150 * the 0xC3 0x9F are the UTF-8
3151 * byte sequence for the U+00DF. */
3153 toLOWER(s[0]) == 's' &&
3155 toLOWER(s[1]) == 's' &&
3162 nextchr = UCHARAT(locinput);
3166 /* Neither the target and the pattern are utf8. */
3168 /* Inline the first character, for speed. */
3169 if (UCHARAT(s) != nextchr &&
3170 UCHARAT(s) != ((OP(scan) == EXACTF)
3171 ? PL_fold : PL_fold_locale)[nextchr])
3173 if (PL_regeol - locinput < ln)
3175 if (ln > 1 && (OP(scan) == EXACTF
3176 ? ibcmp(s, locinput, ln)
3177 : ibcmp_locale(s, locinput, ln)))
3180 nextchr = UCHARAT(locinput);
3185 STRLEN inclasslen = PL_regeol - locinput;
3187 if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8))
3189 if (locinput >= PL_regeol)
3191 locinput += inclasslen ? inclasslen : UTF8SKIP(locinput);
3192 nextchr = UCHARAT(locinput);
3197 nextchr = UCHARAT(locinput);
3198 if (!REGINCLASS(rex, scan, (U8*)locinput))
3200 if (!nextchr && locinput >= PL_regeol)
3202 nextchr = UCHARAT(++locinput);
3206 /* If we might have the case of the German sharp s
3207 * in a casefolding Unicode character class. */
3209 if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) {
3210 locinput += SHARP_S_SKIP;
3211 nextchr = UCHARAT(locinput);
3217 PL_reg_flags |= RF_tainted;
3223 LOAD_UTF8_CHARCLASS_ALNUM();
3224 if (!(OP(scan) == ALNUM
3225 ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
3226 : isALNUM_LC_utf8((U8*)locinput)))
3230 locinput += PL_utf8skip[nextchr];
3231 nextchr = UCHARAT(locinput);
3234 if (!(OP(scan) == ALNUM
3235 ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
3237 nextchr = UCHARAT(++locinput);
3240 PL_reg_flags |= RF_tainted;
3243 if (!nextchr && locinput >= PL_regeol)
3246 LOAD_UTF8_CHARCLASS_ALNUM();
3247 if (OP(scan) == NALNUM
3248 ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
3249 : isALNUM_LC_utf8((U8*)locinput))
3253 locinput += PL_utf8skip[nextchr];
3254 nextchr = UCHARAT(locinput);
3257 if (OP(scan) == NALNUM
3258 ? isALNUM(nextchr) : isALNUM_LC(nextchr))
3260 nextchr = UCHARAT(++locinput);
3264 PL_reg_flags |= RF_tainted;
3268 /* was last char in word? */
3270 if (locinput == PL_bostr)
3273 const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);
3275 ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags);
3277 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
3278 ln = isALNUM_uni(ln);
3279 LOAD_UTF8_CHARCLASS_ALNUM();
3280 n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8);
3283 ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln));
3284 n = isALNUM_LC_utf8((U8*)locinput);
3288 ln = (locinput != PL_bostr) ?
3289 UCHARAT(locinput - 1) : '\n';
3290 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
3292 n = isALNUM(nextchr);
3295 ln = isALNUM_LC(ln);
3296 n = isALNUM_LC(nextchr);
3299 if (((!ln) == (!n)) == (OP(scan) == BOUND ||
3300 OP(scan) == BOUNDL))
3304 PL_reg_flags |= RF_tainted;
3310 if (UTF8_IS_CONTINUED(nextchr)) {
3311 LOAD_UTF8_CHARCLASS_SPACE();
3312 if (!(OP(scan) == SPACE
3313 ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
3314 : isSPACE_LC_utf8((U8*)locinput)))
3318 locinput += PL_utf8skip[nextchr];
3319 nextchr = UCHARAT(locinput);
3322 if (!(OP(scan) == SPACE
3323 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
3325 nextchr = UCHARAT(++locinput);
3328 if (!(OP(scan) == SPACE
3329 ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
3331 nextchr = UCHARAT(++locinput);
3335 PL_reg_flags |= RF_tainted;
3338 if (!nextchr && locinput >= PL_regeol)
3341 LOAD_UTF8_CHARCLASS_SPACE();
3342 if (OP(scan) == NSPACE
3343 ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
3344 : isSPACE_LC_utf8((U8*)locinput))
3348 locinput += PL_utf8skip[nextchr];
3349 nextchr = UCHARAT(locinput);
3352 if (OP(scan) == NSPACE
3353 ? isSPACE(nextchr) : isSPACE_LC(nextchr))
3355 nextchr = UCHARAT(++locinput);
3358 PL_reg_flags |= RF_tainted;
3364 LOAD_UTF8_CHARCLASS_DIGIT();
3365 if (!(OP(scan) == DIGIT
3366 ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
3367 : isDIGIT_LC_utf8((U8*)locinput)))
3371 locinput += PL_utf8skip[nextchr];
3372 nextchr = UCHARAT(locinput);
3375 if (!(OP(scan) == DIGIT
3376 ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)))
3378 nextchr = UCHARAT(++locinput);
3381 PL_reg_flags |= RF_tainted;
3384 if (!nextchr && locinput >= PL_regeol)
3387 LOAD_UTF8_CHARCLASS_DIGIT();
3388 if (OP(scan) == NDIGIT
3389 ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
3390 : isDIGIT_LC_utf8((U8*)locinput))
3394 locinput += PL_utf8skip[nextchr];
3395 nextchr = UCHARAT(locinput);
3398 if (OP(scan) == NDIGIT
3399 ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))
3401 nextchr = UCHARAT(++locinput);
3404 if (locinput >= PL_regeol)
3407 LOAD_UTF8_CHARCLASS_MARK();
3408 if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
3410 locinput += PL_utf8skip[nextchr];
3411 while (locinput < PL_regeol &&
3412 swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
3413 locinput += UTF8SKIP(locinput);
3414 if (locinput > PL_regeol)
3419 nextchr = UCHARAT(locinput);
3426 PL_reg_flags |= RF_tainted;
3431 n = reg_check_named_buff_matched(rex,scan);
3434 type = REF + ( type - NREF );
3441 PL_reg_flags |= RF_tainted;
3445 n = ARG(scan); /* which paren pair */
3448 ln = PL_regstartp[n];
3449 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
3450 if (*PL_reglastparen < n || ln == -1)
3451 sayNO; /* Do not match unless seen CLOSEn. */
3452 if (ln == PL_regendp[n])
3456 if (do_utf8 && type != REF) { /* REF can do byte comparison */
3458 const char *e = PL_bostr + PL_regendp[n];
3460 * Note that we can't do the "other character" lookup trick as
3461 * in the 8-bit case (no pun intended) because in Unicode we
3462 * have to map both upper and title case to lower case.
3466 STRLEN ulen1, ulen2;
3467 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
3468 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
3472 toLOWER_utf8((U8*)s, tmpbuf1, &ulen1);
3473 toLOWER_utf8((U8*)l, tmpbuf2, &ulen2);
3474 if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
3481 nextchr = UCHARAT(locinput);
3485 /* Inline the first character, for speed. */
3486 if (UCHARAT(s) != nextchr &&
3488 (UCHARAT(s) != (type == REFF
3489 ? PL_fold : PL_fold_locale)[nextchr])))
3491 ln = PL_regendp[n] - ln;
3492 if (locinput + ln > PL_regeol)
3494 if (ln > 1 && (type == REF
3495 ? memNE(s, locinput, ln)
3497 ? ibcmp(s, locinput, ln)
3498 : ibcmp_locale(s, locinput, ln))))
3501 nextchr = UCHARAT(locinput);
3511 #define ST st->u.eval
3515 regexp_internal *rei;
3516 regnode *startpoint;
3519 case GOSUB: /* /(...(?1))/ */
3520 if (cur_eval && cur_eval->locinput==locinput) {
3521 if (cur_eval->u.eval.close_paren == (U32)ARG(scan))
3522 Perl_croak(aTHX_ "Infinite recursion in regex");
3523 if ( ++nochange_depth > MAX_RECURSE_EVAL_NOCHANGE_DEPTH )
3525 "Pattern subroutine nesting without pos change"
3526 " exceeded limit in regex");
3532 (void)ReREFCNT_inc(rex);
3533 if (OP(scan)==GOSUB) {
3534 startpoint = scan + ARG2L(scan);
3535 ST.close_paren = ARG(scan);
3537 startpoint = rei->program+1;
3540 goto eval_recurse_doit;
3542 case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */
3543 if (cur_eval && cur_eval->locinput==locinput) {
3544 if ( ++nochange_depth > MAX_RECURSE_EVAL_NOCHANGE_DEPTH )
3545 Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex");
3550 /* execute the code in the {...} */
3552 SV ** const before = SP;
3553 OP_4tree * const oop = PL_op;
3554 COP * const ocurcop = PL_curcop;
3558 PL_op = (OP_4tree*)rexi->data->data[n];
3559 DEBUG_STATE_r( PerlIO_printf(Perl_debug_log,
3560 " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
3561 PAD_SAVE_LOCAL(old_comppad, (PAD*)rexi->data->data[n + 2]);
3562 PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr;
3565 SV *sv_mrk = get_sv("REGMARK", 1);
3566 sv_setsv(sv_mrk, sv_yes_mark);
3569 CALLRUNOPS(aTHX); /* Scalar context. */
3572 ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */
3579 PAD_RESTORE_LOCAL(old_comppad);
3580 PL_curcop = ocurcop;
3583 sv_setsv(save_scalar(PL_replgv), ret);
3587 if (logical == 2) { /* Postponed subexpression: /(??{...})/ */
3590 /* extract RE object from returned value; compiling if
3595 if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret)))
3596 mg = mg_find(sv, PERL_MAGIC_qr);
3597 else if (SvSMAGICAL(ret)) {
3598 if (SvGMAGICAL(ret))
3599 sv_unmagic(ret, PERL_MAGIC_qr);
3601 mg = mg_find(ret, PERL_MAGIC_qr);
3605 re = (regexp *)mg->mg_obj;
3606 (void)ReREFCNT_inc(re);
3610 const char * const t = SvPV_const(ret, len);
3612 const I32 osize = PL_regsize;
3615 if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8;
3616 re = CALLREGCOMP((char*)t, (char*)t + len, &pm);
3618 & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
3620 sv_magic(ret,(SV*)ReREFCNT_inc(re),
3627 debug_start_match(re, do_utf8, locinput, PL_regeol,
3628 "Matching embedded");
3630 startpoint = rei->program + 1;
3631 ST.close_paren = 0; /* only used for GOSUB */
3632 /* borrowed from regtry */
3633 if (PL_reg_start_tmpl <= re->nparens) {
3634 PL_reg_start_tmpl = re->nparens*3/2 + 3;
3635 if(PL_reg_start_tmp)
3636 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
3638 Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
3641 eval_recurse_doit: /* Share code with GOSUB below this line */
3642 /* run the pattern returned from (??{...}) */
3643 ST.cp = regcppush(0); /* Save *all* the positions. */
3644 REGCP_SET(ST.lastcp);
3646 PL_regstartp = re->startp; /* essentially NOOP on GOSUB */
3647 PL_regendp = re->endp; /* essentially NOOP on GOSUB */
3649 *PL_reglastparen = 0;
3650 *PL_reglastcloseparen = 0;
3651 PL_reginput = locinput;
3654 /* XXXX This is too dramatic a measure... */
3657 ST.toggle_reg_flags = PL_reg_flags;
3658 if (re->extflags & RXf_UTF8)
3659 PL_reg_flags |= RF_utf8;
3661 PL_reg_flags &= ~RF_utf8;
3662 ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */
3665 ST.prev_curlyx = cur_curlyx;
3670 ST.prev_eval = cur_eval;
3672 /* now continue from first node in postoned RE */
3673 PUSH_YES_STATE_GOTO(EVAL_AB, startpoint);
3676 /* logical is 1, /(?(?{...})X|Y)/ */
3677 sw = (bool)SvTRUE(ret);
3682 case EVAL_AB: /* cleanup after a successful (??{A})B */
3683 /* note: this is called twice; first after popping B, then A */
3684 PL_reg_flags ^= ST.toggle_reg_flags;
3687 rexi = RXi_GET(rex);
3689 cur_eval = ST.prev_eval;
3690 cur_curlyx = ST.prev_curlyx;
3691 /* XXXX This is too dramatic a measure... */
3696 case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */
3697 /* note: this is called twice; first after popping B, then A */
3698 PL_reg_flags ^= ST.toggle_reg_flags;
3701 rexi = RXi_GET(rex);
3702 PL_reginput = locinput;
3703 REGCP_UNWIND(ST.lastcp);
3705 cur_eval = ST.prev_eval;
3706 cur_curlyx = ST.prev_curlyx;
3707 /* XXXX This is too dramatic a measure... */
3713 n = ARG(scan); /* which paren pair */
3714 PL_reg_start_tmp[n] = locinput;
3720 n = ARG(scan); /* which paren pair */
3721 PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr;
3722 PL_regendp[n] = locinput - PL_bostr;
3723 /*if (n > PL_regsize)
3725 if (n > *PL_reglastparen)
3726 *PL_reglastparen = n;
3727 *PL_reglastcloseparen = n;
3728 if (cur_eval && cur_eval->u.eval.close_paren == n) {
3736 cursor && OP(cursor)!=END;
3737 cursor=regnext(cursor))
3739 if ( OP(cursor)==CLOSE ){
3741 if ( n <= lastopen ) {
3742 PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr;
3743 PL_regendp[n] = locinput - PL_bostr;
3744 /*if (n > PL_regsize)
3746 if (n > *PL_reglastparen)
3747 *PL_reglastparen = n;
3748 *PL_reglastcloseparen = n;
3749 if ( n == ARG(scan) || (cur_eval &&
3750 cur_eval->u.eval.close_paren == n))
3759 n = ARG(scan); /* which paren pair */
3760 sw = (bool)(*PL_reglastparen >= n && PL_regendp[n] != -1);
3763 /* reg_check_named_buff_matched returns 0 for no match */
3764 sw = (bool)(0 < reg_check_named_buff_matched(rex,scan));
3768 sw = (cur_eval && (!n || cur_eval->u.eval.close_paren == n));
3774 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
3776 next = NEXTOPER(NEXTOPER(scan));
3778 next = scan + ARG(scan);
3779 if (OP(next) == IFTHEN) /* Fake one. */
3780 next = NEXTOPER(NEXTOPER(next));
3784 logical = scan->flags;
3787 /*******************************************************************
3789 The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/
3790 pattern, where A and B are subpatterns. (For simple A, CURLYM or
3791 STAR/PLUS/CURLY/CURLYN are used instead.)
3793 A*B is compiled as <CURLYX><A><WHILEM><B>
3795 On entry to the subpattern, CURLYX is called. This pushes a CURLYX
3796 state, which contains the current count, initialised to -1. It also sets
3797 cur_curlyx to point to this state, with any previous value saved in the
3800 CURLYX then jumps straight to the WHILEM op, rather than executing A,
3801 since the pattern may possibly match zero times (i.e. it's a while {} loop
3802 rather than a do {} while loop).
3804 Each entry to WHILEM represents a successful match of A. The count in the
3805 CURLYX block is incremented, another WHILEM state is pushed, and execution
3806 passes to A or B depending on greediness and the current count.
3808 For example, if matching against the string a1a2a3b (where the aN are
3809 substrings that match /A/), then the match progresses as follows: (the
3810 pushed states are interspersed with the bits of strings matched so far):
3813 <CURLYX cnt=0><WHILEM>
3814 <CURLYX cnt=1><WHILEM> a1 <WHILEM>
3815 <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM>
3816 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM>
3817 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b
3819 (Contrast this with something like CURLYM, which maintains only a single
3823 a1 <CURLYM cnt=1> a2
3824 a1 a2 <CURLYM cnt=2> a3
3825 a1 a2 a3 <CURLYM cnt=3> b
3828 Each WHILEM state block marks a point to backtrack to upon partial failure
3829 of A or B, and also contains some minor state data related to that
3830 iteration. The CURLYX block, pointed to by cur_curlyx, contains the
3831 overall state, such as the count, and pointers to the A and B ops.
3833 This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx
3834 must always point to the *current* CURLYX block, the rules are:
3836 When executing CURLYX, save the old cur_curlyx in the CURLYX state block,
3837 and set cur_curlyx to point the new block.
3839 When popping the CURLYX block after a successful or unsuccessful match,
3840 restore the previous cur_curlyx.
3842 When WHILEM is about to execute B, save the current cur_curlyx, and set it
3843 to the outer one saved in the CURLYX block.
3845 When popping the WHILEM block after a successful or unsuccessful B match,
3846 restore the previous cur_curlyx.
3848 Here's an example for the pattern (AI* BI)*BO
3849 I and O refer to inner and outer, C and W refer to CURLYX and WHILEM:
3852 curlyx backtrack stack
3853 ------ ---------------
3855 CO <CO prev=NULL> <WO>
3856 CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
3857 CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
3858 NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo
3860 At this point the pattern succeeds, and we work back down the stack to
3861 clean up, restoring as we go:
3863 CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
3864 CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
3865 CO <CO prev=NULL> <WO>
3868 *******************************************************************/
3870 #define ST st->u.curlyx
3872 case CURLYX: /* start of /A*B/ (for complex A) */
3874 /* No need to save/restore up to this paren */
3875 I32 parenfloor = scan->flags;
3877 assert(next); /* keep Coverity happy */
3878 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
3881 /* XXXX Probably it is better to teach regpush to support
3882 parenfloor > PL_regsize... */
3883 if (parenfloor > (I32)*PL_reglastparen)
3884 parenfloor = *PL_reglastparen; /* Pessimization... */
3886 ST.prev_curlyx= cur_curlyx;
3888 ST.cp = PL_savestack_ix;
3890 /* these fields contain the state of the current curly.
3891 * they are accessed by subsequent WHILEMs */
3892 ST.parenfloor = parenfloor;
3893 ST.min = ARG1(scan);
3894 ST.max = ARG2(scan);
3895 ST.A = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
3899 ST.count = -1; /* this will be updated by WHILEM */
3900 ST.lastloc = NULL; /* this will be updated by WHILEM */
3902 PL_reginput = locinput;
3903 PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next));
3907 case CURLYX_end: /* just finished matching all of A*B */
3908 if (PL_reg_eval_set){
3909 SV *pres= GvSV(PL_replgv);
3912 sv_setsv(GvSV(PL_replgv), pres);
3917 cur_curlyx = ST.prev_curlyx;
3921 case CURLYX_end_fail: /* just failed to match all of A*B */
3923 cur_curlyx = ST.prev_curlyx;
3929 #define ST st->u.whilem
3931 case WHILEM: /* just matched an A in /A*B/ (for complex A) */
3933 /* see the discussion above about CURLYX/WHILEM */
3935 assert(cur_curlyx); /* keep Coverity happy */
3936 n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */
3937 ST.save_lastloc = cur_curlyx->u.curlyx.lastloc;
3938 ST.cache_offset = 0;
3941 PL_reginput = locinput;
3943 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
3944 "%*s whilem: matched %ld out of %ld..%ld\n",
3945 REPORT_CODE_OFF+depth*2, "", (long)n,
3946 (long)cur_curlyx->u.curlyx.min,
3947 (long)cur_curlyx->u.curlyx.max)
3950 /* First just match a string of min A's. */
3952 if (n < cur_curlyx->u.curlyx.min) {
3953 cur_curlyx->u.curlyx.lastloc = locinput;
3954 PUSH_STATE_GOTO(WHILEM_A_pre, cur_curlyx->u.curlyx.A);
3958 /* If degenerate A matches "", assume A done. */
3960 if (locinput == cur_curlyx->u.curlyx.lastloc) {
3961 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
3962 "%*s whilem: empty match detected, trying continuation...\n",
3963 REPORT_CODE_OFF+depth*2, "")
3965 goto do_whilem_B_max;
3968 /* super-linear cache processing */
3972 if (!PL_reg_maxiter) {
3973 /* start the countdown: Postpone detection until we
3974 * know the match is not *that* much linear. */
3975 PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4);
3976 /* possible overflow for long strings and many CURLYX's */
3977 if (PL_reg_maxiter < 0)
3978 PL_reg_maxiter = I32_MAX;
3979 PL_reg_leftiter = PL_reg_maxiter;
3982 if (PL_reg_leftiter-- == 0) {
3983 /* initialise cache */
3984 const I32 size = (PL_reg_maxiter + 7)/8;
3985 if (PL_reg_poscache) {
3986 if ((I32)PL_reg_poscache_size < size) {
3987 Renew(PL_reg_poscache, size, char);
3988 PL_reg_poscache_size = size;
3990 Zero(PL_reg_poscache, size, char);
3993 PL_reg_poscache_size = size;
3994 Newxz(PL_reg_poscache, size, char);
3996 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
3997 "%swhilem: Detected a super-linear match, switching on caching%s...\n",
3998 PL_colors[4], PL_colors[5])
4002 if (PL_reg_leftiter < 0) {
4003 /* have we already failed at this position? */
4005 offset = (scan->flags & 0xf) - 1
4006 + (locinput - PL_bostr) * (scan->flags>>4);
4007 mask = 1 << (offset % 8);
4009 if (PL_reg_poscache[offset] & mask) {
4010 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4011 "%*s whilem: (cache) already tried at this position...\n",
4012 REPORT_CODE_OFF+depth*2, "")
4014 sayNO; /* cache records failure */
4016 ST.cache_offset = offset;
4017 ST.cache_mask = mask;
4021 /* Prefer B over A for minimal matching. */
4023 if (cur_curlyx->u.curlyx.minmod) {
4024 ST.save_curlyx = cur_curlyx;
4025 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
4026 ST.cp = regcppush(ST.save_curlyx->u.curlyx.parenfloor);
4027 REGCP_SET(ST.lastcp);
4028 PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B);
4032 /* Prefer A over B for maximal matching. */
4034 if (n < cur_curlyx->u.curlyx.max) { /* More greed allowed? */
4035 ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor);
4036 cur_curlyx->u.curlyx.lastloc = locinput;
4037 REGCP_SET(ST.lastcp);
4038 PUSH_STATE_GOTO(WHILEM_A_max, cur_curlyx->u.curlyx.A);
4041 goto do_whilem_B_max;
4045 case WHILEM_B_min: /* just matched B in a minimal match */
4046 case WHILEM_B_max: /* just matched B in a maximal match */
4047 cur_curlyx = ST.save_curlyx;
4051 case WHILEM_B_max_fail: /* just failed to match B in a maximal match */
4052 cur_curlyx = ST.save_curlyx;
4053 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
4054 cur_curlyx->u.curlyx.count--;
4058 case WHILEM_A_min_fail: /* just failed to match A in a minimal match */
4059 REGCP_UNWIND(ST.lastcp);
4062 case WHILEM_A_pre_fail: /* just failed to match even minimal A */
4063 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
4064 cur_curlyx->u.curlyx.count--;
4068 case WHILEM_A_max_fail: /* just failed to match A in a maximal match */
4069 REGCP_UNWIND(ST.lastcp);
4070 regcppop(rex); /* Restore some previous $<digit>s? */
4071 PL_reginput = locinput;
4072 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
4073 "%*s whilem: failed, trying continuation...\n",
4074 REPORT_CODE_OFF+depth*2, "")
4077 if (cur_curlyx->u.curlyx.count >= REG_INFTY
4078 && ckWARN(WARN_REGEXP)
4079 && !(PL_reg_flags & RF_warned))
4081 PL_reg_flags |= RF_warned;
4082 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded",
4083 "Complex regular subexpression recursion",
4088 ST.save_curlyx = cur_curlyx;
4089 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
4090 PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B);
4093 case WHILEM_B_min_fail: /* just failed to match B in a minimal match */
4094 cur_curlyx = ST.save_curlyx;
4095 REGCP_UNWIND(ST.lastcp);
4098 if (cur_curlyx->u.curlyx.count >= cur_curlyx->u.curlyx.max) {
4099 /* Maximum greed exceeded */
4100 if (cur_curlyx->u.curlyx.count >= REG_INFTY
4101 && ckWARN(WARN_REGEXP)
4102 && !(PL_reg_flags & RF_warned))
4104 PL_reg_flags |= RF_warned;
4105 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
4106 "%s limit (%d) exceeded",
4107 "Complex regular subexpression recursion",
4110 cur_curlyx->u.curlyx.count--;
4114 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
4115 "%*s trying longer...\n", REPORT_CODE_OFF+depth*2, "")
4117 /* Try grabbing another A and see if it helps. */
4118 PL_reginput = locinput;
4119 cur_curlyx->u.curlyx.lastloc = locinput;
4120 ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor);
4121 REGCP_SET(ST.lastcp);
4122 PUSH_STATE_GOTO(WHILEM_A_min, ST.save_curlyx->u.curlyx.A);
4126 #define ST st->u.branch
4128 case BRANCHJ: /* /(...|A|...)/ with long next pointer */
4129 next = scan + ARG(scan);
4132 scan = NEXTOPER(scan);
4135 case BRANCH: /* /(...|A|...)/ */
4136 scan = NEXTOPER(scan); /* scan now points to inner node */
4137 if ((!next || (OP(next) != BRANCH && OP(next) != BRANCHJ))
4140 /* last branch; skip state push and jump direct to node */
4143 ST.lastparen = *PL_reglastparen;
4144 ST.next_branch = next;
4146 PL_reginput = locinput;
4148 /* Now go into the branch */
4150 PUSH_YES_STATE_GOTO(BRANCH_next, scan);
4152 PUSH_STATE_GOTO(BRANCH_next, scan);
4156 PL_reginput = locinput;
4157 sv_yes_mark = st->u.mark.mark_name = scan->flags ? NULL :
4158 (SV*)rexi->data->data[ ARG( scan ) ];
4159 PUSH_STATE_GOTO(CUTGROUP_next,next);
4161 case CUTGROUP_next_fail:
4164 if (st->u.mark.mark_name)
4165 sv_commit = st->u.mark.mark_name;
4171 case BRANCH_next_fail: /* that branch failed; try the next, if any */
4176 REGCP_UNWIND(ST.cp);
4177 for (n = *PL_reglastparen; n > ST.lastparen; n--)
4179 *PL_reglastparen = n;
4180 /*dmq: *PL_reglastcloseparen = n; */
4181 scan = ST.next_branch;
4182 /* no more branches? */
4183 if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) {
4185 PerlIO_printf( Perl_debug_log,
4186 "%*s %sBRANCH failed...%s\n",
4187 REPORT_CODE_OFF+depth*2, "",
4193 continue; /* execute next BRANCH[J] op */
4201 #define ST st->u.curlym
4203 case CURLYM: /* /A{m,n}B/ where A is fixed-length */
4205 /* This is an optimisation of CURLYX that enables us to push
4206 * only a single backtracking state, no matter now many matches
4207 * there are in {m,n}. It relies on the pattern being constant
4208 * length, with no parens to influence future backrefs
4212 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
4214 /* if paren positive, emulate an OPEN/CLOSE around A */
4216 U32 paren = ST.me->flags;
4217 if (paren > PL_regsize)
4219 if (paren > *PL_reglastparen)
4220 *PL_reglastparen = paren;
4221 scan += NEXT_OFF(scan); /* Skip former OPEN. */
4229 ST.c1 = CHRTEST_UNINIT;
4232 if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */
4235 curlym_do_A: /* execute the A in /A{m,n}B/ */
4236 PL_reginput = locinput;
4237 PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */
4240 case CURLYM_A: /* we've just matched an A */
4241 locinput = st->locinput;
4242 nextchr = UCHARAT(locinput);
4245 /* after first match, determine A's length: u.curlym.alen */
4246 if (ST.count == 1) {
4247 if (PL_reg_match_utf8) {
4249 while (s < PL_reginput) {
4255 ST.alen = PL_reginput - locinput;
4258 ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me);
4261 PerlIO_printf(Perl_debug_log,
4262 "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n",
4263 (int)(REPORT_CODE_OFF+(depth*2)), "",
4264 (IV) ST.count, (IV)ST.alen)
4267 locinput = PL_reginput;
4269 if (cur_eval && cur_eval->u.eval.close_paren &&
4270 cur_eval->u.eval.close_paren == (U32)ST.me->flags)
4273 if ( ST.count < (ST.minmod ? ARG1(ST.me) : ARG2(ST.me)) )
4274 goto curlym_do_A; /* try to match another A */
4275 goto curlym_do_B; /* try to match B */
4277 case CURLYM_A_fail: /* just failed to match an A */
4278 REGCP_UNWIND(ST.cp);
4280 if (ST.minmod || ST.count < ARG1(ST.me) /* min*/
4281 || (cur_eval && cur_eval->u.eval.close_paren &&
4282 cur_eval->u.eval.close_paren == (U32)ST.me->flags))
4285 curlym_do_B: /* execute the B in /A{m,n}B/ */
4286 PL_reginput = locinput;
4287 if (ST.c1 == CHRTEST_UNINIT) {
4288 /* calculate c1 and c2 for possible match of 1st char
4289 * following curly */
4290 ST.c1 = ST.c2 = CHRTEST_VOID;
4291 if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) {
4292 regnode *text_node = ST.B;
4293 if (! HAS_TEXT(text_node))
4294 FIND_NEXT_IMPT(text_node);
4295 if (HAS_TEXT(text_node)
4296 && PL_regkind[OP(text_node)] != REF)
4298 ST.c1 = (U8)*STRING(text_node);
4300 (OP(text_node) == EXACTF || OP(text_node) == REFF)
4302 : (OP(text_node) == EXACTFL || OP(text_node) == REFFL)
4303 ? PL_fold_locale[ST.c1]
4310 PerlIO_printf(Perl_debug_log,
4311 "%*s CURLYM trying tail with matches=%"IVdf"...\n",
4312 (int)(REPORT_CODE_OFF+(depth*2)),
4315 if (ST.c1 != CHRTEST_VOID
4316 && UCHARAT(PL_reginput) != ST.c1
4317 && UCHARAT(PL_reginput) != ST.c2)
4319 /* simulate B failing */
4320 state_num = CURLYM_B_fail;
4321 goto reenter_switch;
4325 /* mark current A as captured */
4326 I32 paren = ST.me->flags;
4329 = HOPc(PL_reginput, -ST.alen) - PL_bostr;
4330 PL_regendp[paren] = PL_reginput - PL_bostr;
4331 /*dmq: *PL_reglastcloseparen = paren; */
4334 PL_regendp[paren] = -1;
4335 if (cur_eval && cur_eval->u.eval.close_paren &&
4336 cur_eval->u.eval.close_paren == (U32)ST.me->flags)
4345 PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */
4348 case CURLYM_B_fail: /* just failed to match a B */
4349 REGCP_UNWIND(ST.cp);
4351 if (ST.count == ARG2(ST.me) /* max */)
4353 goto curlym_do_A; /* try to match a further A */
4355 /* backtrack one A */
4356 if (ST.count == ARG1(ST.me) /* min */)
4359 locinput = HOPc(locinput, -ST.alen);
4360 goto curlym_do_B; /* try to match B */
4363 #define ST st->u.curly
4365 #define CURLY_SETPAREN(paren, success) \
4368 PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr; \
4369 PL_regendp[paren] = locinput - PL_bostr; \
4370 *PL_reglastcloseparen = paren; \
4373 PL_regendp[paren] = -1; \
4376 case STAR: /* /A*B/ where A is width 1 */
4380 scan = NEXTOPER(scan);
4382 case PLUS: /* /A+B/ where A is width 1 */
4386 scan = NEXTOPER(scan);
4388 case CURLYN: /* /(A){m,n}B/ where A is width 1 */
4389 ST.paren = scan->flags; /* Which paren to set */
4390 if (ST.paren > PL_regsize)
4391 PL_regsize = ST.paren;
4392 if (ST.paren > *PL_reglastparen)
4393 *PL_reglastparen = ST.paren;
4394 ST.min = ARG1(scan); /* min to match */
4395 ST.max = ARG2(scan); /* max to match */
4396 if (cur_eval && cur_eval->u.eval.close_paren &&
4397 cur_eval->u.eval.close_paren == (U32)ST.paren) {
4401 scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE);
4403 case CURLY: /* /A{m,n}B/ where A is width 1 */
4405 ST.min = ARG1(scan); /* min to match */
4406 ST.max = ARG2(scan); /* max to match */
4407 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
4410 * Lookahead to avoid useless match attempts
4411 * when we know what character comes next.
4413 * Used to only do .*x and .*?x, but now it allows
4414 * for )'s, ('s and (?{ ... })'s to be in the way
4415 * of the quantifier and the EXACT-like node. -- japhy
4418 if (ST.min > ST.max) /* XXX make this a compile-time check? */
4420 if (HAS_TEXT(next) || JUMPABLE(next)) {
4422 regnode *text_node = next;
4424 if (! HAS_TEXT(text_node))
4425 FIND_NEXT_IMPT(text_node);
4427 if (! HAS_TEXT(text_node))
4428 ST.c1 = ST.c2 = CHRTEST_VOID;
4430 if (PL_regkind[OP(text_node)] == REF) {
4431 ST.c1 = ST.c2 = CHRTEST_VOID;
4432 goto assume_ok_easy;
4435 s = (U8*)STRING(text_node);
4439 if (OP(text_node) == EXACTF || OP(text_node) == REFF)
4440 ST.c2 = PL_fold[ST.c1];
4441 else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL)
4442 ST.c2 = PL_fold_locale[ST.c1];
4445 if (OP(text_node) == EXACTF || OP(text_node) == REFF) {
4446 STRLEN ulen1, ulen2;
4447 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
4448 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
4450 to_utf8_lower((U8*)s, tmpbuf1, &ulen1);
4451 to_utf8_upper((U8*)s, tmpbuf2, &ulen2);
4453 ST.c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN, 0,
4455 0 : UTF8_ALLOW_ANY);
4456 ST.c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN, 0,
4458 0 : UTF8_ALLOW_ANY);
4460 ST.c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0,
4462 ST.c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0,
4467 ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0,
4474 ST.c1 = ST.c2 = CHRTEST_VOID;
4479 PL_reginput = locinput;
4482 if (ST.min && regrepeat(rex, ST.A, ST.min, depth) < ST.min)
4485 locinput = PL_reginput;
4487 if (ST.c1 == CHRTEST_VOID)
4488 goto curly_try_B_min;
4490 ST.oldloc = locinput;
4492 /* set ST.maxpos to the furthest point along the
4493 * string that could possibly match */
4494 if (ST.max == REG_INFTY) {
4495 ST.maxpos = PL_regeol - 1;
4497 while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos))
4501 int m = ST.max - ST.min;
4502 for (ST.maxpos = locinput;
4503 m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--)
4504 ST.maxpos += UTF8SKIP(ST.maxpos);
4507 ST.maxpos = locinput + ST.max - ST.min;
4508 if (ST.maxpos >= PL_regeol)
4509 ST.maxpos = PL_regeol - 1;
4511 goto curly_try_B_min_known;
4515 ST.count = regrepeat(rex, ST.A, ST.max, depth);
4516 locinput = PL_reginput;
4517 if (ST.count < ST.min)
4519 if ((ST.count > ST.min)
4520 && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL))
4522 /* A{m,n} must come at the end of the string, there's
4523 * no point in backing off ... */
4525 /* ...except that $ and \Z can match before *and* after
4526 newline at the end. Consider "\n\n" =~ /\n+\Z\n/.
4527 We may back off by one in this case. */
4528 if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS)
4532 goto curly_try_B_max;
4537 case CURLY_B_min_known_fail:
4538 /* failed to find B in a non-greedy match where c1,c2 valid */
4539 if (ST.paren && ST.count)
4540 PL_regendp[ST.paren] = -1;
4542 PL_reginput = locinput; /* Could be reset... */
4543 REGCP_UNWIND(ST.cp);
4544 /* Couldn't or didn't -- move forward. */
4545 ST.oldloc = locinput;
4547 locinput += UTF8SKIP(locinput);
4551 curly_try_B_min_known:
4552 /* find the next place where 'B' could work, then call B */
4556 n = (ST.oldloc == locinput) ? 0 : 1;
4557 if (ST.c1 == ST.c2) {
4559 /* set n to utf8_distance(oldloc, locinput) */
4560 while (locinput <= ST.maxpos &&
4561 utf8n_to_uvchr((U8*)locinput,
4562 UTF8_MAXBYTES, &len,
4563 uniflags) != (UV)ST.c1) {
4569 /* set n to utf8_distance(oldloc, locinput) */
4570 while (locinput <= ST.maxpos) {
4572 const UV c = utf8n_to_uvchr((U8*)locinput,
4573 UTF8_MAXBYTES, &len,
4575 if (c == (UV)ST.c1 || c == (UV)ST.c2)
4583 if (ST.c1 == ST.c2) {
4584 while (locinput <= ST.maxpos &&
4585 UCHARAT(locinput) != ST.c1)
4589 while (locinput <= ST.maxpos
4590 && UCHARAT(locinput) != ST.c1
4591 && UCHARAT(locinput) != ST.c2)
4594 n = locinput - ST.oldloc;
4596 if (locinput > ST.maxpos)
4598 /* PL_reginput == oldloc now */
4601 if (regrepeat(rex, ST.A, n, depth) < n)
4604 PL_reginput = locinput;
4605 CURLY_SETPAREN(ST.paren, ST.count);
4606 if (cur_eval && cur_eval->u.eval.close_paren &&
4607 cur_eval->u.eval.close_paren == (U32)ST.paren) {
4610 PUSH_STATE_GOTO(CURLY_B_min_known, ST.B);
4615 case CURLY_B_min_fail:
4616 /* failed to find B in a non-greedy match where c1,c2 invalid */
4617 if (ST.paren && ST.count)
4618 PL_regendp[ST.paren] = -1;
4620 REGCP_UNWIND(ST.cp);
4621 /* failed -- move forward one */
4622 PL_reginput = locinput;
4623 if (regrepeat(rex, ST.A, 1, depth)) {
4625 locinput = PL_reginput;
4626 if (ST.count <= ST.max || (ST.max == REG_INFTY &&
4627 ST.count > 0)) /* count overflow ? */
4630 CURLY_SETPAREN(ST.paren, ST.count);
4631 if (cur_eval && cur_eval->u.eval.close_paren &&
4632 cur_eval->u.eval.close_paren == (U32)ST.paren) {
4635 PUSH_STATE_GOTO(CURLY_B_min, ST.B);
4643 /* a successful greedy match: now try to match B */
4644 if (cur_eval && cur_eval->u.eval.close_paren &&
4645 cur_eval->u.eval.close_paren == (U32)ST.paren) {
4650 if (ST.c1 != CHRTEST_VOID)
4651 c = do_utf8 ? utf8n_to_uvchr((U8*)PL_reginput,
4652 UTF8_MAXBYTES, 0, uniflags)
4653 : (UV) UCHARAT(PL_reginput);
4654 /* If it could work, try it. */
4655 if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) {
4656 CURLY_SETPAREN(ST.paren, ST.count);
4657 PUSH_STATE_GOTO(CURLY_B_max, ST.B);
4662 case CURLY_B_max_fail:
4663 /* failed to find B in a greedy match */
4664 if (ST.paren && ST.count)
4665 PL_regendp[ST.paren] = -1;
4667 REGCP_UNWIND(ST.cp);
4669 if (--ST.count < ST.min)
4671 PL_reginput = locinput = HOPc(locinput, -1);
4672 goto curly_try_B_max;
4679 /* we've just finished A in /(??{A})B/; now continue with B */
4683 st->u.eval.toggle_reg_flags
4684 = cur_eval->u.eval.toggle_reg_flags;
4685 PL_reg_flags ^= st->u.eval.toggle_reg_flags;
4687 st->u.eval.prev_rex = rex; /* inner */
4688 rex = cur_eval->u.eval.prev_rex; /* outer */
4689 rexi = RXi_GET(rex);
4690 cur_curlyx = cur_eval->u.eval.prev_curlyx;
4692 st->u.eval.cp = regcppush(0); /* Save *all* the positions. */
4693 REGCP_SET(st->u.eval.lastcp);
4694 PL_reginput = locinput;
4696 /* Restore parens of the outer rex without popping the
4698 tmpix = PL_savestack_ix;
4699 PL_savestack_ix = cur_eval->u.eval.lastcp;
4701 PL_savestack_ix = tmpix;
4703 st->u.eval.prev_eval = cur_eval;
4704 cur_eval = cur_eval->u.eval.prev_eval;
4706 PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ... %"UVxf"\n",
4707 REPORT_CODE_OFF+depth*2, "",PTR2UV(cur_eval)););
4708 PUSH_YES_STATE_GOTO(EVAL_AB,
4709 st->u.eval.prev_eval->u.eval.B); /* match B */
4712 if (locinput < reginfo->till) {
4713 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
4714 "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n",
4716 (long)(locinput - PL_reg_starttry),
4717 (long)(reginfo->till - PL_reg_starttry),
4720 sayNO_SILENT; /* Cannot match: too short. */
4722 PL_reginput = locinput; /* put where regtry can find it */
4723 sayYES; /* Success! */
4725 case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */
4727 PerlIO_printf(Perl_debug_log,
4728 "%*s %ssubpattern success...%s\n",
4729 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]));
4730 PL_reginput = locinput; /* put where regtry can find it */
4731 sayYES; /* Success! */
4734 #define ST st->u.ifmatch
4736 case SUSPEND: /* (?>A) */
4738 PL_reginput = locinput;
4741 case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */
4743 goto ifmatch_trivial_fail_test;
4745 case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */
4747 ifmatch_trivial_fail_test:
4749 char * const s = HOPBACKc(locinput, scan->flags);
4754 sw = 1 - (bool)ST.wanted;
4758 next = scan + ARG(scan);
4766 PL_reginput = locinput;
4770 ST.logical = logical;
4771 /* execute body of (?...A) */
4772 PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan)));
4775 case IFMATCH_A_fail: /* body of (?...A) failed */
4776 ST.wanted = !ST.wanted;
4779 case IFMATCH_A: /* body of (?...A) succeeded */
4781 sw = (bool)ST.wanted;
4783 else if (!ST.wanted)
4786 if (OP(ST.me) == SUSPEND)
4787 locinput = PL_reginput;
4789 locinput = PL_reginput = st->locinput;
4790 nextchr = UCHARAT(locinput);
4792 scan = ST.me + ARG(ST.me);
4795 continue; /* execute B */
4800 next = scan + ARG(scan);
4805 reginfo->cutpoint = PL_regeol;
4808 PL_reginput = locinput;
4810 sv_yes_mark = sv_commit = (SV*)rexi->data->data[ ARG( scan ) ];
4811 PUSH_STATE_GOTO(COMMIT_next,next);
4813 case COMMIT_next_fail:
4820 #define ST st->u.mark
4822 ST.prev_mark = mark_state;
4823 ST.mark_name = sv_commit = sv_yes_mark
4824 = (SV*)rexi->data->data[ ARG( scan ) ];
4826 ST.mark_loc = PL_reginput = locinput;
4827 PUSH_YES_STATE_GOTO(MARKPOINT_next,next);
4829 case MARKPOINT_next:
4830 mark_state = ST.prev_mark;
4833 case MARKPOINT_next_fail:
4834 if (popmark && sv_eq(ST.mark_name,popmark))
4836 if (ST.mark_loc > startpoint)
4837 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
4838 popmark = NULL; /* we found our mark */
4839 sv_commit = ST.mark_name;
4842 PerlIO_printf(Perl_debug_log,
4843 "%*s %ssetting cutpoint to mark:%"SVf"...%s\n",
4844 REPORT_CODE_OFF+depth*2, "",
4845 PL_colors[4], SVfARG(sv_commit), PL_colors[5]);
4848 mark_state = ST.prev_mark;
4849 sv_yes_mark = mark_state ?
4850 mark_state->u.mark.mark_name : NULL;
4854 PL_reginput = locinput;
4856 /* (*SKIP) : if we fail we cut here*/
4857 ST.mark_name = NULL;
4858 ST.mark_loc = locinput;
4859 PUSH_STATE_GOTO(SKIP_next,next);
4861 /* (*SKIP:NAME) : if there is a (*MARK:NAME) fail where it was,
4862 otherwise do nothing. Meaning we need to scan
4864 regmatch_state *cur = mark_state;
4865 SV *find = (SV*)rexi->data->data[ ARG( scan ) ];
4868 if ( sv_eq( cur->u.mark.mark_name,
4871 ST.mark_name = find;
4872 PUSH_STATE_GOTO( SKIP_next, next );
4874 cur = cur->u.mark.prev_mark;
4877 /* Didn't find our (*MARK:NAME) so ignore this (*SKIP:NAME) */
4879 case SKIP_next_fail:
4881 /* (*CUT:NAME) - Set up to search for the name as we
4882 collapse the stack*/
4883 popmark = ST.mark_name;
4885 /* (*CUT) - No name, we cut here.*/
4886 if (ST.mark_loc > startpoint)
4887 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
4888 /* but we set sv_commit to latest mark_name if there
4889 is one so they can test to see how things lead to this
4892 sv_commit=mark_state->u.mark.mark_name;
4900 PerlIO_printf(Perl_error_log, "%"UVxf" %d\n",
4901 PTR2UV(scan), OP(scan));
4902 Perl_croak(aTHX_ "regexp memory corruption");
4906 /* switch break jumps here */
4907 scan = next; /* prepare to execute the next op and ... */
4908 continue; /* ... jump back to the top, reusing st */
4912 /* push a state that backtracks on success */
4913 st->u.yes.prev_yes_state = yes_state;
4917 /* push a new regex state, then continue at scan */
4919 regmatch_state *newst;
4922 regmatch_state *cur = st;
4923 regmatch_state *curyes = yes_state;
4925 regmatch_slab *slab = PL_regmatch_slab;
4926 for (;curd > -1;cur--,curd--) {
4927 if (cur < SLAB_FIRST(slab)) {
4929 cur = SLAB_LAST(slab);
4931 PerlIO_printf(Perl_error_log, "%*s#%-3d %-10s %s\n",
4932 REPORT_CODE_OFF + 2 + depth * 2,"",
4933 curd, reg_name[cur->resume_state],
4934 (curyes == cur) ? "yes" : ""
4937 curyes = cur->u.yes.prev_yes_state;
4940 DEBUG_STATE_pp("push")
4943 st->locinput = locinput;
4945 if (newst > SLAB_LAST(PL_regmatch_slab))
4946 newst = S_push_slab(aTHX);
4947 PL_regmatch_state = newst;
4949 locinput = PL_reginput;
4950 nextchr = UCHARAT(locinput);
4958 * We get here only if there's trouble -- normally "case END" is
4959 * the terminating point.
4961 Perl_croak(aTHX_ "corrupted regexp pointers");
4967 /* we have successfully completed a subexpression, but we must now
4968 * pop to the state marked by yes_state and continue from there */
4969 assert(st != yes_state);
4971 while (st != yes_state) {
4973 if (st < SLAB_FIRST(PL_regmatch_slab)) {
4974 PL_regmatch_slab = PL_regmatch_slab->prev;
4975 st = SLAB_LAST(PL_regmatch_slab);
4979 DEBUG_STATE_pp("pop (no final)");
4981 DEBUG_STATE_pp("pop (yes)");
4987 while (yes_state < SLAB_FIRST(PL_regmatch_slab)
4988 || yes_state > SLAB_LAST(PL_regmatch_slab))
4990 /* not in this slab, pop slab */
4991 depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1);
4992 PL_regmatch_slab = PL_regmatch_slab->prev;
4993 st = SLAB_LAST(PL_regmatch_slab);
4995 depth -= (st - yes_state);
4998 yes_state = st->u.yes.prev_yes_state;
4999 PL_regmatch_state = st;
5002 locinput= st->locinput;
5003 nextchr = UCHARAT(locinput);
5005 state_num = st->resume_state + no_final;
5006 goto reenter_switch;
5009 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n",
5010 PL_colors[4], PL_colors[5]));
5017 PerlIO_printf(Perl_debug_log,
5018 "%*s %sfailed...%s\n",
5019 REPORT_CODE_OFF+depth*2, "",
5020 PL_colors[4], PL_colors[5])
5032 /* there's a previous state to backtrack to */
5034 if (st < SLAB_FIRST(PL_regmatch_slab)) {
5035 PL_regmatch_slab = PL_regmatch_slab->prev;
5036 st = SLAB_LAST(PL_regmatch_slab);
5038 PL_regmatch_state = st;
5039 locinput= st->locinput;
5040 nextchr = UCHARAT(locinput);
5042 DEBUG_STATE_pp("pop");
5044 if (yes_state == st)
5045 yes_state = st->u.yes.prev_yes_state;
5047 state_num = st->resume_state + 1; /* failure = success + 1 */
5048 goto reenter_switch;
5053 if (rex->intflags & PREGf_VERBARG_SEEN) {
5054 SV *sv_err = get_sv("REGERROR", 1);
5055 SV *sv_mrk = get_sv("REGMARK", 1);
5057 sv_commit = &PL_sv_no;
5059 sv_yes_mark = &PL_sv_yes;
5062 sv_commit = &PL_sv_yes;
5063 sv_yes_mark = &PL_sv_no;
5065 sv_setsv(sv_err, sv_commit);
5066 sv_setsv(sv_mrk, sv_yes_mark);
5068 /* restore original high-water mark */
5069 PL_regmatch_slab = orig_slab;
5070 PL_regmatch_state = orig_state;
5072 /* free all slabs above current one */
5073 if (orig_slab->next) {
5074 regmatch_slab *sl = orig_slab->next;
5075 orig_slab->next = NULL;
5077 regmatch_slab * const osl = sl;
5087 - regrepeat - repeatedly match something simple, report how many
5090 * [This routine now assumes that it will only match on things of length 1.
5091 * That was true before, but now we assume scan - reginput is the count,
5092 * rather than incrementing count on every character. [Er, except utf8.]]
5095 S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max, int depth)
5098 register char *scan;
5100 register char *loceol = PL_regeol;
5101 register I32 hardcount = 0;
5102 register bool do_utf8 = PL_reg_match_utf8;
5105 if (max == REG_INFTY)
5107 else if (max < loceol - scan)
5108 loceol = scan + max;
5113 while (scan < loceol && hardcount < max && *scan != '\n') {
5114 scan += UTF8SKIP(scan);
5118 while (scan < loceol && *scan != '\n')
5125 while (scan < loceol && hardcount < max) {
5126 scan += UTF8SKIP(scan);
5136 case EXACT: /* length of string is 1 */
5138 while (scan < loceol && UCHARAT(scan) == c)
5141 case EXACTF: /* length of string is 1 */
5143 while (scan < loceol &&
5144 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c]))
5147 case EXACTFL: /* length of string is 1 */
5148 PL_reg_flags |= RF_tainted;
5150 while (scan < loceol &&
5151 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c]))
5157 while (hardcount < max && scan < loceol &&
5158 reginclass(prog, p, (U8*)scan, 0, do_utf8)) {
5159 scan += UTF8SKIP(scan);
5163 while (scan < loceol && REGINCLASS(prog, p, (U8*)scan))
5170 LOAD_UTF8_CHARCLASS_ALNUM();
5171 while (hardcount < max && scan < loceol &&
5172 swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) {
5173 scan += UTF8SKIP(scan);
5177 while (scan < loceol && isALNUM(*scan))
5182 PL_reg_flags |= RF_tainted;
5185 while (hardcount < max && scan < loceol &&
5186 isALNUM_LC_utf8((U8*)scan)) {
5187 scan += UTF8SKIP(scan);
5191 while (scan < loceol && isALNUM_LC(*scan))
5198 LOAD_UTF8_CHARCLASS_ALNUM();
5199 while (hardcount < max && scan < loceol &&
5200 !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) {
5201 scan += UTF8SKIP(scan);
5205 while (scan < loceol && !isALNUM(*scan))
5210 PL_reg_flags |= RF_tainted;
5213 while (hardcount < max && scan < loceol &&
5214 !isALNUM_LC_utf8((U8*)scan)) {
5215 scan += UTF8SKIP(scan);
5219 while (scan < loceol && !isALNUM_LC(*scan))
5226 LOAD_UTF8_CHARCLASS_SPACE();
5227 while (hardcount < max && scan < loceol &&
5229 swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) {
5230 scan += UTF8SKIP(scan);
5234 while (scan < loceol && isSPACE(*scan))
5239 PL_reg_flags |= RF_tainted;
5242 while (hardcount < max && scan < loceol &&
5243 (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
5244 scan += UTF8SKIP(scan);
5248 while (scan < loceol && isSPACE_LC(*scan))
5255 LOAD_UTF8_CHARCLASS_SPACE();
5256 while (hardcount < max && scan < loceol &&
5258 swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) {
5259 scan += UTF8SKIP(scan);
5263 while (scan < loceol && !isSPACE(*scan))
5268 PL_reg_flags |= RF_tainted;
5271 while (hardcount < max && scan < loceol &&
5272 !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
5273 scan += UTF8SKIP(scan);
5277 while (scan < loceol && !isSPACE_LC(*scan))
5284 LOAD_UTF8_CHARCLASS_DIGIT();
5285 while (hardcount < max && scan < loceol &&
5286 swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) {
5287 scan += UTF8SKIP(scan);
5291 while (scan < loceol && isDIGIT(*scan))
5298 LOAD_UTF8_CHARCLASS_DIGIT();
5299 while (hardcount < max && scan < loceol &&
5300 !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) {
5301 scan += UTF8SKIP(scan);
5305 while (scan < loceol && !isDIGIT(*scan))
5309 default: /* Called on something of 0 width. */
5310 break; /* So match right here or not at all. */
5316 c = scan - PL_reginput;
5320 GET_RE_DEBUG_FLAGS_DECL;
5322 SV * const prop = sv_newmortal();
5323 regprop(prog, prop, p);
5324 PerlIO_printf(Perl_debug_log,
5325 "%*s %s can match %"IVdf" times out of %"IVdf"...\n",
5326 REPORT_CODE_OFF + depth*2, "", SvPVX_const(prop),(IV)c,(IV)max);
5334 #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION)
5336 - regclass_swash - prepare the utf8 swash
5340 Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
5346 RXi_GET_DECL(prog,progi);
5347 const struct reg_data * const data = prog ? progi->data : NULL;
5349 if (data && data->count) {
5350 const U32 n = ARG(node);
5352 if (data->what[n] == 's') {
5353 SV * const rv = (SV*)data->data[n];
5354 AV * const av = (AV*)SvRV((SV*)rv);
5355 SV **const ary = AvARRAY(av);
5358 /* See the end of regcomp.c:S_regclass() for
5359 * documentation of these array elements. */
5362 a = SvROK(ary[1]) ? &ary[1] : 0;
5363 b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : 0;
5367 else if (si && doinit) {
5368 sw = swash_init("utf8", "", si, 1, 0);
5369 (void)av_store(av, 1, sw);
5386 - reginclass - determine if a character falls into a character class
5388 The n is the ANYOF regnode, the p is the target string, lenp
5389 is pointer to the maximum length of how far to go in the p
5390 (if the lenp is zero, UTF8SKIP(p) is used),
5391 do_utf8 tells whether the target string is in UTF-8.
5396 S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8)
5399 const char flags = ANYOF_FLAGS(n);
5405 if (do_utf8 && !UTF8_IS_INVARIANT(c)) {
5406 c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len,
5407 (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV) | UTF8_CHECK_ONLY);
5408 /* see [perl #37836] for UTF8_ALLOW_ANYUV */
5409 if (len == (STRLEN)-1)
5410 Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)");
5413 plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c));
5414 if (do_utf8 || (flags & ANYOF_UNICODE)) {
5417 if (do_utf8 && !ANYOF_RUNTIME(n)) {
5418 if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c))
5421 if (!match && do_utf8 && (flags & ANYOF_UNICODE_ALL) && c >= 256)
5425 SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av);
5428 if (swash_fetch(sw, p, do_utf8))
5430 else if (flags & ANYOF_FOLD) {
5431 if (!match && lenp && av) {
5433 for (i = 0; i <= av_len(av); i++) {
5434 SV* const sv = *av_fetch(av, i, FALSE);
5436 const char * const s = SvPV_const(sv, len);
5438 if (len <= plen && memEQ(s, (char*)p, len)) {
5446 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
5449 to_utf8_fold(p, tmpbuf, &tmplen);
5450 if (swash_fetch(sw, tmpbuf, do_utf8))
5456 if (match && lenp && *lenp == 0)
5457 *lenp = UNISKIP(NATIVE_TO_UNI(c));
5459 if (!match && c < 256) {
5460 if (ANYOF_BITMAP_TEST(n, c))
5462 else if (flags & ANYOF_FOLD) {
5465 if (flags & ANYOF_LOCALE) {
5466 PL_reg_flags |= RF_tainted;
5467 f = PL_fold_locale[c];
5471 if (f != c && ANYOF_BITMAP_TEST(n, f))
5475 if (!match && (flags & ANYOF_CLASS)) {
5476 PL_reg_flags |= RF_tainted;
5478 (ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) ||
5479 (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) ||
5480 (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) ||
5481 (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) ||
5482 (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) ||
5483 (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) ||
5484 (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) ||
5485 (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) ||
5486 (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) ||
5487 (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) ||
5488 (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) ||
5489 (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) ||
5490 (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) ||
5491 (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) ||
5492 (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) ||
5493 (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) ||
5494 (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) ||
5495 (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) ||
5496 (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) ||
5497 (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) ||
5498 (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) ||
5499 (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) ||
5500 (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) ||
5501 (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) ||
5502 (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) ||
5503 (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) ||
5504 (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) ||
5505 (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) ||
5506 (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) ||
5507 (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c))
5508 ) /* How's that for a conditional? */
5515 return (flags & ANYOF_INVERT) ? !match : match;
5519 S_reghop3(U8 *s, I32 off, const U8* lim)
5523 while (off-- && s < lim) {
5524 /* XXX could check well-formedness here */
5529 while (off++ && s > lim) {
5531 if (UTF8_IS_CONTINUED(*s)) {
5532 while (s > lim && UTF8_IS_CONTINUATION(*s))
5535 /* XXX could check well-formedness here */
5542 /* there are a bunch of places where we use two reghop3's that should
5543 be replaced with this routine. but since thats not done yet
5544 we ifdef it out - dmq
5547 S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim)
5551 while (off-- && s < rlim) {
5552 /* XXX could check well-formedness here */
5557 while (off++ && s > llim) {
5559 if (UTF8_IS_CONTINUED(*s)) {
5560 while (s > llim && UTF8_IS_CONTINUATION(*s))
5563 /* XXX could check well-formedness here */
5571 S_reghopmaybe3(U8* s, I32 off, const U8* lim)
5575 while (off-- && s < lim) {
5576 /* XXX could check well-formedness here */
5583 while (off++ && s > lim) {
5585 if (UTF8_IS_CONTINUED(*s)) {
5586 while (s > lim && UTF8_IS_CONTINUATION(*s))
5589 /* XXX could check well-formedness here */
5598 restore_pos(pTHX_ void *arg)
5601 regexp * const rex = (regexp *)arg;
5602 if (PL_reg_eval_set) {
5603 if (PL_reg_oldsaved) {
5604 rex->subbeg = PL_reg_oldsaved;
5605 rex->sublen = PL_reg_oldsavedlen;
5606 #ifdef PERL_OLD_COPY_ON_WRITE
5607 rex->saved_copy = PL_nrs;
5609 RX_MATCH_COPIED_on(rex);
5611 PL_reg_magic->mg_len = PL_reg_oldpos;
5612 PL_reg_eval_set = 0;
5613 PL_curpm = PL_reg_oldcurpm;
5618 S_to_utf8_substr(pTHX_ register regexp *prog)
5622 if (prog->substrs->data[i].substr
5623 && !prog->substrs->data[i].utf8_substr) {
5624 SV* const sv = newSVsv(prog->substrs->data[i].substr);
5625 prog->substrs->data[i].utf8_substr = sv;
5626 sv_utf8_upgrade(sv);
5627 if (SvVALID(prog->substrs->data[i].substr))
5629 if (SvTAIL(prog->substrs->data[i].substr))
5631 if (prog->substrs->data[i].substr == prog->check_substr)
5632 prog->check_utf8 = sv;
5638 S_to_byte_substr(pTHX_ register regexp *prog)
5643 if (prog->substrs->data[i].utf8_substr
5644 && !prog->substrs->data[i].substr) {
5645 SV* sv = newSVsv(prog->substrs->data[i].utf8_substr);
5646 if (sv_utf8_downgrade(sv, TRUE)) {
5647 if (SvVALID(prog->substrs->data[i].utf8_substr))
5649 if (SvTAIL(prog->substrs->data[i].utf8_substr))
5655 prog->substrs->data[i].substr = sv;
5656 if (prog->substrs->data[i].utf8_substr == prog->check_utf8)
5657 prog->check_substr = sv;
5664 * c-indentation-style: bsd
5666 * indent-tabs-mode: t
5669 * ex: set ts=8 sts=4 sw=4 noet: