5 * One Ring to rule them all, One Ring to find them
7 * [p.v of _The Lord of the Rings_, opening poem]
8 * [p.50 of _The Lord of the Rings_, I/iii: "The Shadow of the Past"]
9 * [p.254 of _The Lord of the Rings_, II/ii: "The Council of Elrond"]
12 /* This file contains functions for executing a regular expression. See
13 * also regcomp.c which funnily enough, contains functions for compiling
14 * a regular expression.
16 * This file is also copied at build time to ext/re/re_exec.c, where
17 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
18 * This causes the main functions to be compiled under new names and with
19 * debugging support added, which makes "use re 'debug'" work.
22 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
23 * confused with the original package (see point 3 below). Thanks, Henry!
26 /* Additional note: this code is very heavily munged from Henry's version
27 * in places. In some spots I've traded clarity for efficiency, so don't
28 * blame Henry for some of the lack of readability.
31 /* The names of the functions have been changed from regcomp and
32 * regexec to pregcomp and pregexec in order to avoid conflicts
33 * with the POSIX routines of the same names.
36 #ifdef PERL_EXT_RE_BUILD
41 * pregcomp and pregexec -- regsub and regerror are not used in perl
43 * Copyright (c) 1986 by University of Toronto.
44 * Written by Henry Spencer. Not derived from licensed software.
46 * Permission is granted to anyone to use this software for any
47 * purpose on any computer system, and to redistribute it freely,
48 * subject to the following restrictions:
50 * 1. The author is not responsible for the consequences of use of
51 * this software, no matter how awful, even if they arise
54 * 2. The origin of this software must not be misrepresented, either
55 * by explicit claim or by omission.
57 * 3. Altered versions must be plainly marked as such, and must not
58 * be misrepresented as being the original software.
60 **** Alterations to Henry's code are...
62 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
63 **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
64 **** by Larry Wall and others
66 **** You may distribute under the terms of either the GNU General Public
67 **** License or the Artistic License, as specified in the README file.
69 * Beware that some of this code is subtly aware of the way operator
70 * precedence is structured in regular expressions. Serious changes in
71 * regular-expression syntax might require a total rethink.
74 #define PERL_IN_REGEXEC_C
78 #ifdef PERL_IN_XSUB_RE
84 #define RF_tainted 1 /* tainted information used? */
85 #define RF_warned 2 /* warned about big count? */
87 #define RF_utf8 8 /* Pattern contains multibyte chars? */
89 #define UTF ((PL_reg_flags & RF_utf8) != 0)
91 #define RS_init 1 /* eval environment created */
92 #define RS_set 2 /* replsv value is set */
98 #define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c)))
104 #define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv))
105 #define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b)
107 #define HOPc(pos,off) \
108 (char *)(PL_reg_match_utf8 \
109 ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \
111 #define HOPBACKc(pos, off) \
112 (char*)(PL_reg_match_utf8\
113 ? reghopmaybe3((U8*)pos, -off, (U8*)PL_bostr) \
114 : (pos - off >= PL_bostr) \
118 #define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off))
119 #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
121 /* these are unrolled below in the CCC_TRY_XXX defined */
122 #define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \
123 if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END
125 /* Doesn't do an assert to verify that is correct */
126 #define LOAD_UTF8_CHARCLASS_NO_CHECK(class) STMT_START { \
127 if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)" "); LEAVE; } } STMT_END
129 #define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a")
130 #define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0")
131 #define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ")
133 #define LOAD_UTF8_CHARCLASS_GCB() /* Grapheme cluster boundaries */ \
134 LOAD_UTF8_CHARCLASS(X_begin, " "); \
135 LOAD_UTF8_CHARCLASS(X_non_hangul, "A"); \
136 /* These are utf8 constants, and not utf-ebcdic constants, so the \
137 * assert should likely and hopefully fail on an EBCDIC machine */ \
138 LOAD_UTF8_CHARCLASS(X_extend, "\xcc\x80"); /* U+0300 */ \
140 /* No asserts are done for these, in case called on an early \
141 * Unicode version in which they map to nothing */ \
142 LOAD_UTF8_CHARCLASS_NO_CHECK(X_prepend);/* U+0E40 "\xe0\xb9\x80" */ \
143 LOAD_UTF8_CHARCLASS_NO_CHECK(X_L); /* U+1100 "\xe1\x84\x80" */ \
144 LOAD_UTF8_CHARCLASS_NO_CHECK(X_LV); /* U+AC00 "\xea\xb0\x80" */ \
145 LOAD_UTF8_CHARCLASS_NO_CHECK(X_LVT); /* U+AC01 "\xea\xb0\x81" */ \
146 LOAD_UTF8_CHARCLASS_NO_CHECK(X_LV_LVT_V);/* U+AC01 "\xea\xb0\x81" */\
147 LOAD_UTF8_CHARCLASS_NO_CHECK(X_T); /* U+11A8 "\xe1\x86\xa8" */ \
148 LOAD_UTF8_CHARCLASS_NO_CHECK(X_V) /* U+1160 "\xe1\x85\xa0" */
151 We dont use PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS as the direct test
152 so that it is possible to override the option here without having to
153 rebuild the entire core. as we are required to do if we change regcomp.h
154 which is where PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS is defined.
156 #if PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS
157 #define BROKEN_UNICODE_CHARCLASS_MAPPINGS
160 #ifdef BROKEN_UNICODE_CHARCLASS_MAPPINGS
161 #define LOAD_UTF8_CHARCLASS_PERL_WORD() LOAD_UTF8_CHARCLASS_ALNUM()
162 #define LOAD_UTF8_CHARCLASS_PERL_SPACE() LOAD_UTF8_CHARCLASS_SPACE()
163 #define LOAD_UTF8_CHARCLASS_POSIX_DIGIT() LOAD_UTF8_CHARCLASS_DIGIT()
164 #define RE_utf8_perl_word PL_utf8_alnum
165 #define RE_utf8_perl_space PL_utf8_space
166 #define RE_utf8_posix_digit PL_utf8_digit
167 #define perl_word alnum
168 #define perl_space space
169 #define posix_digit digit
171 #define LOAD_UTF8_CHARCLASS_PERL_WORD() LOAD_UTF8_CHARCLASS(perl_word,"a")
172 #define LOAD_UTF8_CHARCLASS_PERL_SPACE() LOAD_UTF8_CHARCLASS(perl_space," ")
173 #define LOAD_UTF8_CHARCLASS_POSIX_DIGIT() LOAD_UTF8_CHARCLASS(posix_digit,"0")
174 #define RE_utf8_perl_word PL_utf8_perl_word
175 #define RE_utf8_perl_space PL_utf8_perl_space
176 #define RE_utf8_posix_digit PL_utf8_posix_digit
180 #define CCC_TRY_AFF(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC,LCFUNC) \
182 PL_reg_flags |= RF_tainted; \
187 if (do_utf8 && UTF8_IS_CONTINUED(nextchr)) { \
188 if (!CAT2(PL_utf8_,CLASS)) { \
192 ok=CAT2(is_utf8_,CLASS)((const U8*)STR); \
196 if (!(OP(scan) == NAME \
197 ? (bool)swash_fetch(CAT2(PL_utf8_,CLASS), (U8*)locinput, do_utf8) \
198 : LCFUNC_utf8((U8*)locinput))) \
202 locinput += PL_utf8skip[nextchr]; \
203 nextchr = UCHARAT(locinput); \
206 if (!(OP(scan) == NAME ? FUNC(nextchr) : LCFUNC(nextchr))) \
208 nextchr = UCHARAT(++locinput); \
211 #define CCC_TRY_NEG(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC,LCFUNC) \
213 PL_reg_flags |= RF_tainted; \
216 if (!nextchr && locinput >= PL_regeol) \
218 if (do_utf8 && UTF8_IS_CONTINUED(nextchr)) { \
219 if (!CAT2(PL_utf8_,CLASS)) { \
223 ok=CAT2(is_utf8_,CLASS)((const U8*)STR); \
227 if ((OP(scan) == NAME \
228 ? (bool)swash_fetch(CAT2(PL_utf8_,CLASS), (U8*)locinput, do_utf8) \
229 : LCFUNC_utf8((U8*)locinput))) \
233 locinput += PL_utf8skip[nextchr]; \
234 nextchr = UCHARAT(locinput); \
237 if ((OP(scan) == NAME ? FUNC(nextchr) : LCFUNC(nextchr))) \
239 nextchr = UCHARAT(++locinput); \
246 /* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */
248 /* for use after a quantifier and before an EXACT-like node -- japhy */
249 /* it would be nice to rework regcomp.sym to generate this stuff. sigh */
250 #define JUMPABLE(rn) ( \
252 (OP(rn) == CLOSE && (!cur_eval || cur_eval->u.eval.close_paren != ARG(rn))) || \
254 OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
255 OP(rn) == PLUS || OP(rn) == MINMOD || \
256 OP(rn) == KEEPS || (PL_regkind[OP(rn)] == VERB) || \
257 (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \
259 #define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT)
261 #define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF )
264 /* Currently these are only used when PL_regkind[OP(rn)] == EXACT so
265 we don't need this definition. */
266 #define IS_TEXT(rn) ( OP(rn)==EXACT || OP(rn)==REF || OP(rn)==NREF )
267 #define IS_TEXTF(rn) ( OP(rn)==EXACTF || OP(rn)==REFF || OP(rn)==NREFF )
268 #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL )
271 /* ... so we use this as its faster. */
272 #define IS_TEXT(rn) ( OP(rn)==EXACT )
273 #define IS_TEXTF(rn) ( OP(rn)==EXACTF )
274 #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL )
279 Search for mandatory following text node; for lookahead, the text must
280 follow but for lookbehind (rn->flags != 0) we skip to the next step.
282 #define FIND_NEXT_IMPT(rn) STMT_START { \
283 while (JUMPABLE(rn)) { \
284 const OPCODE type = OP(rn); \
285 if (type == SUSPEND || PL_regkind[type] == CURLY) \
286 rn = NEXTOPER(NEXTOPER(rn)); \
287 else if (type == PLUS) \
289 else if (type == IFMATCH) \
290 rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
291 else rn += NEXT_OFF(rn); \
296 static void restore_pos(pTHX_ void *arg);
299 S_regcppush(pTHX_ I32 parenfloor)
302 const int retval = PL_savestack_ix;
303 #define REGCP_PAREN_ELEMS 4
304 const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
306 GET_RE_DEBUG_FLAGS_DECL;
308 if (paren_elems_to_push < 0)
309 Perl_croak(aTHX_ "panic: paren_elems_to_push < 0");
311 #define REGCP_OTHER_ELEMS 7
312 SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS);
314 for (p = PL_regsize; p > parenfloor; p--) {
315 /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
316 SSPUSHINT(PL_regoffs[p].end);
317 SSPUSHINT(PL_regoffs[p].start);
318 SSPUSHPTR(PL_reg_start_tmp[p]);
320 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
321 " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n",
322 (UV)p, (IV)PL_regoffs[p].start,
323 (IV)(PL_reg_start_tmp[p] - PL_bostr),
324 (IV)PL_regoffs[p].end
327 /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
328 SSPUSHPTR(PL_regoffs);
329 SSPUSHINT(PL_regsize);
330 SSPUSHINT(*PL_reglastparen);
331 SSPUSHINT(*PL_reglastcloseparen);
332 SSPUSHPTR(PL_reginput);
333 #define REGCP_FRAME_ELEMS 2
334 /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and
335 * are needed for the regexp context stack bookkeeping. */
336 SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
337 SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */
342 /* These are needed since we do not localize EVAL nodes: */
343 #define REGCP_SET(cp) \
345 PerlIO_printf(Perl_debug_log, \
346 " Setting an EVAL scope, savestack=%"IVdf"\n", \
347 (IV)PL_savestack_ix)); \
350 #define REGCP_UNWIND(cp) \
352 if (cp != PL_savestack_ix) \
353 PerlIO_printf(Perl_debug_log, \
354 " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
355 (IV)(cp), (IV)PL_savestack_ix)); \
359 S_regcppop(pTHX_ const regexp *rex)
364 GET_RE_DEBUG_FLAGS_DECL;
366 PERL_ARGS_ASSERT_REGCPPOP;
368 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
370 assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
371 i = SSPOPINT; /* Parentheses elements to pop. */
372 input = (char *) SSPOPPTR;
373 *PL_reglastcloseparen = SSPOPINT;
374 *PL_reglastparen = SSPOPINT;
375 PL_regsize = SSPOPINT;
376 PL_regoffs=(regexp_paren_pair *) SSPOPPTR;
379 /* Now restore the parentheses context. */
380 for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
381 i > 0; i -= REGCP_PAREN_ELEMS) {
383 U32 paren = (U32)SSPOPINT;
384 PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
385 PL_regoffs[paren].start = SSPOPINT;
387 if (paren <= *PL_reglastparen)
388 PL_regoffs[paren].end = tmps;
390 PerlIO_printf(Perl_debug_log,
391 " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n",
392 (UV)paren, (IV)PL_regoffs[paren].start,
393 (IV)(PL_reg_start_tmp[paren] - PL_bostr),
394 (IV)PL_regoffs[paren].end,
395 (paren > *PL_reglastparen ? "(no)" : ""));
399 if (*PL_reglastparen + 1 <= rex->nparens) {
400 PerlIO_printf(Perl_debug_log,
401 " restoring \\%"IVdf"..\\%"IVdf" to undef\n",
402 (IV)(*PL_reglastparen + 1), (IV)rex->nparens);
406 /* It would seem that the similar code in regtry()
407 * already takes care of this, and in fact it is in
408 * a better location to since this code can #if 0-ed out
409 * but the code in regtry() is needed or otherwise tests
410 * requiring null fields (pat.t#187 and split.t#{13,14}
411 * (as of patchlevel 7877) will fail. Then again,
412 * this code seems to be necessary or otherwise
413 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
414 * --jhi updated by dapm */
415 for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) {
417 PL_regoffs[i].start = -1;
418 PL_regoffs[i].end = -1;
424 #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
427 * pregexec and friends
430 #ifndef PERL_IN_XSUB_RE
432 - pregexec - match a regexp against a string
435 Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, register char *strend,
436 char *strbeg, I32 minend, SV *screamer, U32 nosave)
437 /* strend: pointer to null at end of string */
438 /* strbeg: real beginning of string */
439 /* minend: end of match must be >=minend after stringarg. */
440 /* nosave: For optimizations. */
442 PERL_ARGS_ASSERT_PREGEXEC;
445 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
446 nosave ? 0 : REXEC_COPY_STR);
451 * Need to implement the following flags for reg_anch:
453 * USE_INTUIT_NOML - Useful to call re_intuit_start() first
455 * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer
456 * INTUIT_AUTORITATIVE_ML
457 * INTUIT_ONCE_NOML - Intuit can match in one location only.
460 * Another flag for this function: SECOND_TIME (so that float substrs
461 * with giant delta may be not rechecked).
464 /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
466 /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend.
467 Otherwise, only SvCUR(sv) is used to get strbeg. */
469 /* XXXX We assume that strpos is strbeg unless sv. */
471 /* XXXX Some places assume that there is a fixed substring.
472 An update may be needed if optimizer marks as "INTUITable"
473 RExen without fixed substrings. Similarly, it is assumed that
474 lengths of all the strings are no more than minlen, thus they
475 cannot come from lookahead.
476 (Or minlen should take into account lookahead.)
477 NOTE: Some of this comment is not correct. minlen does now take account
478 of lookahead/behind. Further research is required. -- demerphq
482 /* A failure to find a constant substring means that there is no need to make
483 an expensive call to REx engine, thus we celebrate a failure. Similarly,
484 finding a substring too deep into the string means that less calls to
485 regtry() should be needed.
487 REx compiler's optimizer found 4 possible hints:
488 a) Anchored substring;
490 c) Whether we are anchored (beginning-of-line or \G);
491 d) First node (of those at offset 0) which may distingush positions;
492 We use a)b)d) and multiline-part of c), and try to find a position in the
493 string which does not contradict any of them.
496 /* Most of decisions we do here should have been done at compile time.
497 The nodes of the REx which we used for the search should have been
498 deleted from the finite automaton. */
501 Perl_re_intuit_start(pTHX_ REGEXP * const rx, SV *sv, char *strpos,
502 char *strend, const U32 flags, re_scream_pos_data *data)
505 struct regexp *const prog = (struct regexp *)SvANY(rx);
506 register I32 start_shift = 0;
507 /* Should be nonnegative! */
508 register I32 end_shift = 0;
513 const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */
515 register char *other_last = NULL; /* other substr checked before this */
516 char *check_at = NULL; /* check substr found at this pos */
517 const I32 multiline = prog->extflags & RXf_PMf_MULTILINE;
518 RXi_GET_DECL(prog,progi);
520 const char * const i_strpos = strpos;
522 GET_RE_DEBUG_FLAGS_DECL;
524 PERL_ARGS_ASSERT_RE_INTUIT_START;
526 RX_MATCH_UTF8_set(rx,do_utf8);
529 PL_reg_flags |= RF_utf8;
532 debug_start_match(rx, do_utf8, strpos, strend,
533 sv ? "Guessing start of match in sv for"
534 : "Guessing start of match in string for");
537 /* CHR_DIST() would be more correct here but it makes things slow. */
538 if (prog->minlen > strend - strpos) {
539 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
540 "String too short... [re_intuit_start]\n"));
544 strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
547 if (!prog->check_utf8 && prog->check_substr)
548 to_utf8_substr(prog);
549 check = prog->check_utf8;
551 if (!prog->check_substr && prog->check_utf8)
552 to_byte_substr(prog);
553 check = prog->check_substr;
555 if (check == &PL_sv_undef) {
556 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
557 "Non-utf8 string cannot match utf8 check string\n"));
560 if (prog->extflags & RXf_ANCH) { /* Match at beg-of-str or after \n */
561 ml_anch = !( (prog->extflags & RXf_ANCH_SINGLE)
562 || ( (prog->extflags & RXf_ANCH_BOL)
563 && !multiline ) ); /* Check after \n? */
566 if ( !(prog->extflags & RXf_ANCH_GPOS) /* Checked by the caller */
567 && !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */
568 /* SvCUR is not set on references: SvRV and SvPVX_const overlap */
570 && (strpos != strbeg)) {
571 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
574 if (prog->check_offset_min == prog->check_offset_max &&
575 !(prog->extflags & RXf_CANY_SEEN)) {
576 /* Substring at constant offset from beg-of-str... */
579 s = HOP3c(strpos, prog->check_offset_min, strend);
582 slen = SvCUR(check); /* >= 1 */
584 if ( strend - s > slen || strend - s < slen - 1
585 || (strend - s == slen && strend[-1] != '\n')) {
586 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
589 /* Now should match s[0..slen-2] */
591 if (slen && (*SvPVX_const(check) != *s
593 && memNE(SvPVX_const(check), s, slen)))) {
595 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
599 else if (*SvPVX_const(check) != *s
600 || ((slen = SvCUR(check)) > 1
601 && memNE(SvPVX_const(check), s, slen)))
604 goto success_at_start;
607 /* Match is anchored, but substr is not anchored wrt beg-of-str. */
609 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
610 end_shift = prog->check_end_shift;
613 const I32 end = prog->check_offset_max + CHR_SVLEN(check)
614 - (SvTAIL(check) != 0);
615 const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
617 if (end_shift < eshift)
621 else { /* Can match at random position */
624 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
625 end_shift = prog->check_end_shift;
627 /* end shift should be non negative here */
630 #ifdef QDEBUGGING /* 7/99: reports of failure (with the older version) */
632 Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ",
633 (IV)end_shift, RX_PRECOMP(prog));
637 /* Find a possible match in the region s..strend by looking for
638 the "check" substring in the region corrected by start/end_shift. */
641 I32 srch_start_shift = start_shift;
642 I32 srch_end_shift = end_shift;
643 if (srch_start_shift < 0 && strbeg - s > srch_start_shift) {
644 srch_end_shift -= ((strbeg - s) - srch_start_shift);
645 srch_start_shift = strbeg - s;
647 DEBUG_OPTIMISE_MORE_r({
648 PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n",
649 (IV)prog->check_offset_min,
650 (IV)srch_start_shift,
652 (IV)prog->check_end_shift);
655 if (flags & REXEC_SCREAM) {
656 I32 p = -1; /* Internal iterator of scream. */
657 I32 * const pp = data ? data->scream_pos : &p;
659 if (PL_screamfirst[BmRARE(check)] >= 0
660 || ( BmRARE(check) == '\n'
661 && (BmPREVIOUS(check) == SvCUR(check) - 1)
663 s = screaminstr(sv, check,
664 srch_start_shift + (s - strbeg), srch_end_shift, pp, 0);
667 /* we may be pointing at the wrong string */
668 if (s && RXp_MATCH_COPIED(prog))
669 s = strbeg + (s - SvPVX_const(sv));
671 *data->scream_olds = s;
676 if (prog->extflags & RXf_CANY_SEEN) {
677 start_point= (U8*)(s + srch_start_shift);
678 end_point= (U8*)(strend - srch_end_shift);
680 start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend);
681 end_point= HOP3(strend, -srch_end_shift, strbeg);
683 DEBUG_OPTIMISE_MORE_r({
684 PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n",
685 (int)(end_point - start_point),
686 (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point),
690 s = fbm_instr( start_point, end_point,
691 check, multiline ? FBMrf_MULTILINE : 0);
694 /* Update the count-of-usability, remove useless subpatterns,
698 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
699 SvPVX_const(check), RE_SV_DUMPLEN(check), 30);
700 PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s",
701 (s ? "Found" : "Did not find"),
702 (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)
703 ? "anchored" : "floating"),
706 (s ? " at offset " : "...\n") );
711 /* Finish the diagnostic message */
712 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
714 /* XXX dmq: first branch is for positive lookbehind...
715 Our check string is offset from the beginning of the pattern.
716 So we need to do any stclass tests offset forward from that
725 /* Got a candidate. Check MBOL anchoring, and the *other* substr.
726 Start with the other substr.
727 XXXX no SCREAM optimization yet - and a very coarse implementation
728 XXXX /ttx+/ results in anchored="ttx", floating="x". floating will
729 *always* match. Probably should be marked during compile...
730 Probably it is right to do no SCREAM here...
733 if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8)
734 : (prog->float_substr && prog->anchored_substr))
736 /* Take into account the "other" substring. */
737 /* XXXX May be hopelessly wrong for UTF... */
740 if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) {
743 char * const last = HOP3c(s, -start_shift, strbeg);
745 char * const saved_s = s;
748 t = s - prog->check_offset_max;
749 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
751 || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos))
756 t = HOP3c(t, prog->anchored_offset, strend);
757 if (t < other_last) /* These positions already checked */
759 last2 = last1 = HOP3c(strend, -prog->minlen, strbeg);
762 /* XXXX It is not documented what units *_offsets are in.
763 We assume bytes, but this is clearly wrong.
764 Meaning this code needs to be carefully reviewed for errors.
768 /* On end-of-str: see comment below. */
769 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
770 if (must == &PL_sv_undef) {
772 DEBUG_r(must = prog->anchored_utf8); /* for debug */
777 HOP3(HOP3(last1, prog->anchored_offset, strend)
778 + SvCUR(must), -(SvTAIL(must)!=0), strbeg),
780 multiline ? FBMrf_MULTILINE : 0
783 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
784 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
785 PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s",
786 (s ? "Found" : "Contradicts"),
787 quoted, RE_SV_TAIL(must));
792 if (last1 >= last2) {
793 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
794 ", giving up...\n"));
797 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
798 ", trying floating at offset %ld...\n",
799 (long)(HOP3c(saved_s, 1, strend) - i_strpos)));
800 other_last = HOP3c(last1, prog->anchored_offset+1, strend);
801 s = HOP3c(last, 1, strend);
805 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
806 (long)(s - i_strpos)));
807 t = HOP3c(s, -prog->anchored_offset, strbeg);
808 other_last = HOP3c(s, 1, strend);
816 else { /* Take into account the floating substring. */
818 char * const saved_s = s;
821 t = HOP3c(s, -start_shift, strbeg);
823 HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg);
824 if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
825 last = HOP3c(t, prog->float_max_offset, strend);
826 s = HOP3c(t, prog->float_min_offset, strend);
829 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
830 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
831 /* fbm_instr() takes into account exact value of end-of-str
832 if the check is SvTAIL(ed). Since false positives are OK,
833 and end-of-str is not later than strend we are OK. */
834 if (must == &PL_sv_undef) {
836 DEBUG_r(must = prog->float_utf8); /* for debug message */
839 s = fbm_instr((unsigned char*)s,
840 (unsigned char*)last + SvCUR(must)
842 must, multiline ? FBMrf_MULTILINE : 0);
844 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
845 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
846 PerlIO_printf(Perl_debug_log, "%s floating substr %s%s",
847 (s ? "Found" : "Contradicts"),
848 quoted, RE_SV_TAIL(must));
852 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
853 ", giving up...\n"));
856 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
857 ", trying anchored starting at offset %ld...\n",
858 (long)(saved_s + 1 - i_strpos)));
860 s = HOP3c(t, 1, strend);
864 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
865 (long)(s - i_strpos)));
866 other_last = s; /* Fix this later. --Hugo */
876 t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos);
878 DEBUG_OPTIMISE_MORE_r(
879 PerlIO_printf(Perl_debug_log,
880 "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n",
881 (IV)prog->check_offset_min,
882 (IV)prog->check_offset_max,
890 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
892 || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos)))
895 /* Fixed substring is found far enough so that the match
896 cannot start at strpos. */
898 if (ml_anch && t[-1] != '\n') {
899 /* Eventually fbm_*() should handle this, but often
900 anchored_offset is not 0, so this check will not be wasted. */
901 /* XXXX In the code below we prefer to look for "^" even in
902 presence of anchored substrings. And we search even
903 beyond the found float position. These pessimizations
904 are historical artefacts only. */
906 while (t < strend - prog->minlen) {
908 if (t < check_at - prog->check_offset_min) {
909 if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) {
910 /* Since we moved from the found position,
911 we definitely contradict the found anchored
912 substr. Due to the above check we do not
913 contradict "check" substr.
914 Thus we can arrive here only if check substr
915 is float. Redo checking for "other"=="fixed".
918 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
919 PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
920 goto do_other_anchored;
922 /* We don't contradict the found floating substring. */
923 /* XXXX Why not check for STCLASS? */
925 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
926 PL_colors[0], PL_colors[1], (long)(s - i_strpos)));
929 /* Position contradicts check-string */
930 /* XXXX probably better to look for check-string
931 than for "\n", so one should lower the limit for t? */
932 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
933 PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos)));
934 other_last = strpos = s = t + 1;
939 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
940 PL_colors[0], PL_colors[1]));
944 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
945 PL_colors[0], PL_colors[1]));
949 ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */
952 /* The found string does not prohibit matching at strpos,
953 - no optimization of calling REx engine can be performed,
954 unless it was an MBOL and we are not after MBOL,
955 or a future STCLASS check will fail this. */
957 /* Even in this situation we may use MBOL flag if strpos is offset
958 wrt the start of the string. */
959 if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
960 && (strpos != strbeg) && strpos[-1] != '\n'
961 /* May be due to an implicit anchor of m{.*foo} */
962 && !(prog->intflags & PREGf_IMPLICIT))
967 DEBUG_EXECUTE_r( if (ml_anch)
968 PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
969 (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]);
972 if (!(prog->intflags & PREGf_NAUGHTY) /* XXXX If strpos moved? */
974 prog->check_utf8 /* Could be deleted already */
975 && --BmUSEFUL(prog->check_utf8) < 0
976 && (prog->check_utf8 == prog->float_utf8)
978 prog->check_substr /* Could be deleted already */
979 && --BmUSEFUL(prog->check_substr) < 0
980 && (prog->check_substr == prog->float_substr)
983 /* If flags & SOMETHING - do not do it many times on the same match */
984 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
985 /* XXX Does the destruction order has to change with do_utf8? */
986 SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr);
987 SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8);
988 prog->check_substr = prog->check_utf8 = NULL; /* disable */
989 prog->float_substr = prog->float_utf8 = NULL; /* clear */
990 check = NULL; /* abort */
992 /* XXXX This is a remnant of the old implementation. It
993 looks wasteful, since now INTUIT can use many
995 prog->extflags &= ~RXf_USE_INTUIT;
1001 /* Last resort... */
1002 /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
1003 /* trie stclasses are too expensive to use here, we are better off to
1004 leave it to regmatch itself */
1005 if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) {
1006 /* minlen == 0 is possible if regstclass is \b or \B,
1007 and the fixed substr is ''$.
1008 Since minlen is already taken into account, s+1 is before strend;
1009 accidentally, minlen >= 1 guaranties no false positives at s + 1
1010 even for \b or \B. But (minlen? 1 : 0) below assumes that
1011 regstclass does not come from lookahead... */
1012 /* If regstclass takes bytelength more than 1: If charlength==1, OK.
1013 This leaves EXACTF only, which is dealt with in find_byclass(). */
1014 const U8* const str = (U8*)STRING(progi->regstclass);
1015 const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
1016 ? CHR_DIST(str+STR_LEN(progi->regstclass), str)
1019 if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
1020 endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend);
1021 else if (prog->float_substr || prog->float_utf8)
1022 endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend);
1026 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf"\n",
1027 (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg)));
1030 s = find_byclass(prog, progi->regstclass, s, endpos, NULL);
1033 const char *what = NULL;
1035 if (endpos == strend) {
1036 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1037 "Could not match STCLASS...\n") );
1040 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1041 "This position contradicts STCLASS...\n") );
1042 if ((prog->extflags & RXf_ANCH) && !ml_anch)
1044 /* Contradict one of substrings */
1045 if (prog->anchored_substr || prog->anchored_utf8) {
1046 if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) {
1047 DEBUG_EXECUTE_r( what = "anchored" );
1049 s = HOP3c(t, 1, strend);
1050 if (s + start_shift + end_shift > strend) {
1051 /* XXXX Should be taken into account earlier? */
1052 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1053 "Could not match STCLASS...\n") );
1058 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1059 "Looking for %s substr starting at offset %ld...\n",
1060 what, (long)(s + start_shift - i_strpos)) );
1063 /* Have both, check_string is floating */
1064 if (t + start_shift >= check_at) /* Contradicts floating=check */
1065 goto retry_floating_check;
1066 /* Recheck anchored substring, but not floating... */
1070 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1071 "Looking for anchored substr starting at offset %ld...\n",
1072 (long)(other_last - i_strpos)) );
1073 goto do_other_anchored;
1075 /* Another way we could have checked stclass at the
1076 current position only: */
1081 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1082 "Looking for /%s^%s/m starting at offset %ld...\n",
1083 PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
1086 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
1088 /* Check is floating subtring. */
1089 retry_floating_check:
1090 t = check_at - start_shift;
1091 DEBUG_EXECUTE_r( what = "floating" );
1092 goto hop_and_restart;
1095 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1096 "By STCLASS: moving %ld --> %ld\n",
1097 (long)(t - i_strpos), (long)(s - i_strpos))
1101 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1102 "Does not contradict STCLASS...\n");
1107 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
1108 PL_colors[4], (check ? "Guessed" : "Giving up"),
1109 PL_colors[5], (long)(s - i_strpos)) );
1112 fail_finish: /* Substring not found */
1113 if (prog->check_substr || prog->check_utf8) /* could be removed already */
1114 BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
1116 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
1117 PL_colors[4], PL_colors[5]));
1121 #define DECL_TRIE_TYPE(scan) \
1122 const enum { trie_plain, trie_utf8, trie_utf8_fold, trie_latin_utf8_fold } \
1123 trie_type = (scan->flags != EXACT) \
1124 ? (do_utf8 ? trie_utf8_fold : (UTF ? trie_latin_utf8_fold : trie_plain)) \
1125 : (do_utf8 ? trie_utf8 : trie_plain)
1127 #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \
1128 uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \
1129 switch (trie_type) { \
1130 case trie_utf8_fold: \
1131 if ( foldlen>0 ) { \
1132 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
1137 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
1138 uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \
1139 foldlen -= UNISKIP( uvc ); \
1140 uscan = foldbuf + UNISKIP( uvc ); \
1143 case trie_latin_utf8_fold: \
1144 if ( foldlen>0 ) { \
1145 uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
1151 uvc = to_uni_fold( *(U8*)uc, foldbuf, &foldlen ); \
1152 foldlen -= UNISKIP( uvc ); \
1153 uscan = foldbuf + UNISKIP( uvc ); \
1157 uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
1164 charid = trie->charmap[ uvc ]; \
1168 if (widecharmap) { \
1169 SV** const svpp = hv_fetch(widecharmap, \
1170 (char*)&uvc, sizeof(UV), 0); \
1172 charid = (U16)SvIV(*svpp); \
1177 #define REXEC_FBC_EXACTISH_CHECK(CoNd) \
1179 char *my_strend= (char *)strend; \
1182 !ibcmp_utf8(s, &my_strend, 0, do_utf8, \
1183 m, NULL, ln, (bool)UTF)) \
1184 && (!reginfo || regtry(reginfo, &s)) ) \
1187 U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \
1188 uvchr_to_utf8(tmpbuf, c); \
1189 f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \
1191 && (f == c1 || f == c2) \
1193 !ibcmp_utf8(s, &my_strend, 0, do_utf8,\
1194 m, NULL, ln, (bool)UTF)) \
1195 && (!reginfo || regtry(reginfo, &s)) ) \
1201 #define REXEC_FBC_EXACTISH_SCAN(CoNd) \
1205 && (ln == 1 || !(OP(c) == EXACTF \
1207 : ibcmp_locale(s, m, ln))) \
1208 && (!reginfo || regtry(reginfo, &s)) ) \
1214 #define REXEC_FBC_UTF8_SCAN(CoDe) \
1216 while (s + (uskip = UTF8SKIP(s)) <= strend) { \
1222 #define REXEC_FBC_SCAN(CoDe) \
1224 while (s < strend) { \
1230 #define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \
1231 REXEC_FBC_UTF8_SCAN( \
1233 if (tmp && (!reginfo || regtry(reginfo, &s))) \
1242 #define REXEC_FBC_CLASS_SCAN(CoNd) \
1245 if (tmp && (!reginfo || regtry(reginfo, &s))) \
1254 #define REXEC_FBC_TRYIT \
1255 if ((!reginfo || regtry(reginfo, &s))) \
1258 #define REXEC_FBC_CSCAN(CoNdUtF8,CoNd) \
1260 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1263 REXEC_FBC_CLASS_SCAN(CoNd); \
1267 #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \
1270 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1273 REXEC_FBC_CLASS_SCAN(CoNd); \
1277 #define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \
1278 PL_reg_flags |= RF_tainted; \
1280 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1283 REXEC_FBC_CLASS_SCAN(CoNd); \
1287 #define DUMP_EXEC_POS(li,s,doutf8) \
1288 dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8)
1290 /* We know what class REx starts with. Try to find this position... */
1291 /* if reginfo is NULL, its a dryrun */
1292 /* annoyingly all the vars in this routine have different names from their counterparts
1293 in regmatch. /grrr */
1296 S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
1297 const char *strend, regmatch_info *reginfo)
1300 const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
1304 register STRLEN uskip;
1308 register I32 tmp = 1; /* Scratch variable? */
1309 register const bool do_utf8 = PL_reg_match_utf8;
1310 RXi_GET_DECL(prog,progi);
1312 PERL_ARGS_ASSERT_FIND_BYCLASS;
1314 /* We know what class it must start with. */
1318 REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
1319 !UTF8_IS_INVARIANT((U8)s[0]) ?
1320 reginclass(prog, c, (U8*)s, 0, do_utf8) :
1321 REGINCLASS(prog, c, (U8*)s));
1324 while (s < strend) {
1327 if (REGINCLASS(prog, c, (U8*)s) ||
1328 (ANYOF_FOLD_SHARP_S(c, s, strend) &&
1329 /* The assignment of 2 is intentional:
1330 * for the folded sharp s, the skip is 2. */
1331 (skip = SHARP_S_SKIP))) {
1332 if (tmp && (!reginfo || regtry(reginfo, &s)))
1345 if (tmp && (!reginfo || regtry(reginfo, &s)))
1353 ln = STR_LEN(c); /* length to match in octets/bytes */
1354 lnc = (I32) ln; /* length to match in characters */
1356 STRLEN ulen1, ulen2;
1358 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
1359 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
1360 /* used by commented-out code below */
1361 /*const U32 uniflags = UTF8_ALLOW_DEFAULT;*/
1363 /* XXX: Since the node will be case folded at compile
1364 time this logic is a little odd, although im not
1365 sure that its actually wrong. --dmq */
1367 c1 = to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
1368 c2 = to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
1370 /* XXX: This is kinda strange. to_utf8_XYZ returns the
1371 codepoint of the first character in the converted
1372 form, yet originally we did the extra step.
1373 No tests fail by commenting this code out however
1374 so Ive left it out. -- dmq.
1376 c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE,
1378 c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE,
1383 while (sm < ((U8 *) m + ln)) {
1398 c2 = PL_fold_locale[c1];
1400 e = HOP3c(strend, -((I32)lnc), s);
1402 if (!reginfo && e < s)
1403 e = s; /* Due to minlen logic of intuit() */
1405 /* The idea in the EXACTF* cases is to first find the
1406 * first character of the EXACTF* node and then, if
1407 * necessary, case-insensitively compare the full
1408 * text of the node. The c1 and c2 are the first
1409 * characters (though in Unicode it gets a bit
1410 * more complicated because there are more cases
1411 * than just upper and lower: one needs to use
1412 * the so-called folding case for case-insensitive
1413 * matching (called "loose matching" in Unicode).
1414 * ibcmp_utf8() will do just that. */
1416 if (do_utf8 || UTF) {
1418 U8 tmpbuf [UTF8_MAXBYTES+1];
1421 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1423 /* Upper and lower of 1st char are equal -
1424 * probably not a "letter". */
1427 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
1432 REXEC_FBC_EXACTISH_CHECK(c == c1);
1438 c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
1444 /* Handle some of the three Greek sigmas cases.
1445 * Note that not all the possible combinations
1446 * are handled here: some of them are handled
1447 * by the standard folding rules, and some of
1448 * them (the character class or ANYOF cases)
1449 * are handled during compiletime in
1450 * regexec.c:S_regclass(). */
1451 if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA ||
1452 c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA)
1453 c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA;
1455 REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2);
1460 /* Neither pattern nor string are UTF8 */
1462 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
1464 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
1468 PL_reg_flags |= RF_tainted;
1475 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
1476 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
1478 tmp = ((OP(c) == BOUND ?
1479 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1480 LOAD_UTF8_CHARCLASS_ALNUM();
1481 REXEC_FBC_UTF8_SCAN(
1482 if (tmp == !(OP(c) == BOUND ?
1483 (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
1484 isALNUM_LC_utf8((U8*)s)))
1492 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
1493 tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
1496 !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
1502 if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s)))
1506 PL_reg_flags |= RF_tainted;
1513 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
1514 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
1516 tmp = ((OP(c) == NBOUND ?
1517 isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
1518 LOAD_UTF8_CHARCLASS_ALNUM();
1519 REXEC_FBC_UTF8_SCAN(
1520 if (tmp == !(OP(c) == NBOUND ?
1521 (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
1522 isALNUM_LC_utf8((U8*)s)))
1524 else REXEC_FBC_TRYIT;
1528 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
1529 tmp = ((OP(c) == NBOUND ?
1530 isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
1533 !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
1535 else REXEC_FBC_TRYIT;
1538 if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s)))
1542 REXEC_FBC_CSCAN_PRELOAD(
1543 LOAD_UTF8_CHARCLASS_PERL_WORD(),
1544 swash_fetch(RE_utf8_perl_word, (U8*)s, do_utf8),
1548 REXEC_FBC_CSCAN_TAINT(
1549 isALNUM_LC_utf8((U8*)s),
1553 REXEC_FBC_CSCAN_PRELOAD(
1554 LOAD_UTF8_CHARCLASS_PERL_WORD(),
1555 !swash_fetch(RE_utf8_perl_word, (U8*)s, do_utf8),
1559 REXEC_FBC_CSCAN_TAINT(
1560 !isALNUM_LC_utf8((U8*)s),
1564 REXEC_FBC_CSCAN_PRELOAD(
1565 LOAD_UTF8_CHARCLASS_PERL_SPACE(),
1566 *s == ' ' || swash_fetch(RE_utf8_perl_space,(U8*)s, do_utf8),
1570 REXEC_FBC_CSCAN_TAINT(
1571 *s == ' ' || isSPACE_LC_utf8((U8*)s),
1575 REXEC_FBC_CSCAN_PRELOAD(
1576 LOAD_UTF8_CHARCLASS_PERL_SPACE(),
1577 !(*s == ' ' || swash_fetch(RE_utf8_perl_space,(U8*)s, do_utf8)),
1581 REXEC_FBC_CSCAN_TAINT(
1582 !(*s == ' ' || isSPACE_LC_utf8((U8*)s)),
1586 REXEC_FBC_CSCAN_PRELOAD(
1587 LOAD_UTF8_CHARCLASS_POSIX_DIGIT(),
1588 swash_fetch(RE_utf8_posix_digit,(U8*)s, do_utf8),
1592 REXEC_FBC_CSCAN_TAINT(
1593 isDIGIT_LC_utf8((U8*)s),
1597 REXEC_FBC_CSCAN_PRELOAD(
1598 LOAD_UTF8_CHARCLASS_POSIX_DIGIT(),
1599 !swash_fetch(RE_utf8_posix_digit,(U8*)s, do_utf8),
1603 REXEC_FBC_CSCAN_TAINT(
1604 !isDIGIT_LC_utf8((U8*)s),
1610 is_LNBREAK_latin1(s)
1620 !is_VERTWS_latin1(s)
1625 is_HORIZWS_latin1(s)
1629 !is_HORIZWS_utf8(s),
1630 !is_HORIZWS_latin1(s)
1636 /* what trie are we using right now */
1638 = (reg_ac_data*)progi->data->data[ ARG( c ) ];
1640 = (reg_trie_data*)progi->data->data[ aho->trie ];
1641 HV *widecharmap = MUTABLE_HV(progi->data->data[ aho->trie + 1 ]);
1643 const char *last_start = strend - trie->minlen;
1645 const char *real_start = s;
1647 STRLEN maxlen = trie->maxlen;
1649 U8 **points; /* map of where we were in the input string
1650 when reading a given char. For ASCII this
1651 is unnecessary overhead as the relationship
1652 is always 1:1, but for Unicode, especially
1653 case folded Unicode this is not true. */
1654 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1658 GET_RE_DEBUG_FLAGS_DECL;
1660 /* We can't just allocate points here. We need to wrap it in
1661 * an SV so it gets freed properly if there is a croak while
1662 * running the match */
1665 sv_points=newSV(maxlen * sizeof(U8 *));
1666 SvCUR_set(sv_points,
1667 maxlen * sizeof(U8 *));
1668 SvPOK_on(sv_points);
1669 sv_2mortal(sv_points);
1670 points=(U8**)SvPV_nolen(sv_points );
1671 if ( trie_type != trie_utf8_fold
1672 && (trie->bitmap || OP(c)==AHOCORASICKC) )
1675 bitmap=(U8*)trie->bitmap;
1677 bitmap=(U8*)ANYOF_BITMAP(c);
1679 /* this is the Aho-Corasick algorithm modified a touch
1680 to include special handling for long "unknown char"
1681 sequences. The basic idea being that we use AC as long
1682 as we are dealing with a possible matching char, when
1683 we encounter an unknown char (and we have not encountered
1684 an accepting state) we scan forward until we find a legal
1686 AC matching is basically that of trie matching, except
1687 that when we encounter a failing transition, we fall back
1688 to the current states "fail state", and try the current char
1689 again, a process we repeat until we reach the root state,
1690 state 1, or a legal transition. If we fail on the root state
1691 then we can either terminate if we have reached an accepting
1692 state previously, or restart the entire process from the beginning
1696 while (s <= last_start) {
1697 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1705 U8 *uscan = (U8*)NULL;
1706 U8 *leftmost = NULL;
1708 U32 accepted_word= 0;
1712 while ( state && uc <= (U8*)strend ) {
1714 U32 word = aho->states[ state ].wordnum;
1718 DEBUG_TRIE_EXECUTE_r(
1719 if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1720 dump_exec_pos( (char *)uc, c, strend, real_start,
1721 (char *)uc, do_utf8 );
1722 PerlIO_printf( Perl_debug_log,
1723 " Scanning for legal start char...\n");
1726 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1731 if (uc >(U8*)last_start) break;
1735 U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ];
1736 if (!leftmost || lpos < leftmost) {
1737 DEBUG_r(accepted_word=word);
1743 points[pointpos++ % maxlen]= uc;
1744 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
1745 uscan, len, uvc, charid, foldlen,
1747 DEBUG_TRIE_EXECUTE_r({
1748 dump_exec_pos( (char *)uc, c, strend, real_start,
1750 PerlIO_printf(Perl_debug_log,
1751 " Charid:%3u CP:%4"UVxf" ",
1757 word = aho->states[ state ].wordnum;
1759 base = aho->states[ state ].trans.base;
1761 DEBUG_TRIE_EXECUTE_r({
1763 dump_exec_pos( (char *)uc, c, strend, real_start,
1765 PerlIO_printf( Perl_debug_log,
1766 "%sState: %4"UVxf", word=%"UVxf,
1767 failed ? " Fail transition to " : "",
1768 (UV)state, (UV)word);
1773 (base + charid > trie->uniquecharcount )
1774 && (base + charid - 1 - trie->uniquecharcount
1776 && trie->trans[base + charid - 1 -
1777 trie->uniquecharcount].check == state
1778 && (tmp=trie->trans[base + charid - 1 -
1779 trie->uniquecharcount ].next))
1781 DEBUG_TRIE_EXECUTE_r(
1782 PerlIO_printf( Perl_debug_log," - legal\n"));
1787 DEBUG_TRIE_EXECUTE_r(
1788 PerlIO_printf( Perl_debug_log," - fail\n"));
1790 state = aho->fail[state];
1794 /* we must be accepting here */
1795 DEBUG_TRIE_EXECUTE_r(
1796 PerlIO_printf( Perl_debug_log," - accepting\n"));
1805 if (!state) state = 1;
1808 if ( aho->states[ state ].wordnum ) {
1809 U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ];
1810 if (!leftmost || lpos < leftmost) {
1811 DEBUG_r(accepted_word=aho->states[ state ].wordnum);
1816 s = (char*)leftmost;
1817 DEBUG_TRIE_EXECUTE_r({
1819 Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n",
1820 (UV)accepted_word, (IV)(s - real_start)
1823 if (!reginfo || regtry(reginfo, &s)) {
1829 DEBUG_TRIE_EXECUTE_r({
1830 PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n");
1833 DEBUG_TRIE_EXECUTE_r(
1834 PerlIO_printf( Perl_debug_log,"No match.\n"));
1843 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
1853 - regexec_flags - match a regexp against a string
1856 Perl_regexec_flags(pTHX_ REGEXP * const rx, char *stringarg, register char *strend,
1857 char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
1858 /* strend: pointer to null at end of string */
1859 /* strbeg: real beginning of string */
1860 /* minend: end of match must be >=minend after stringarg. */
1861 /* data: May be used for some additional optimizations.
1862 Currently its only used, with a U32 cast, for transmitting
1863 the ganch offset when doing a /g match. This will change */
1864 /* nosave: For optimizations. */
1867 struct regexp *const prog = (struct regexp *)SvANY(rx);
1868 /*register*/ char *s;
1869 register regnode *c;
1870 /*register*/ char *startpos = stringarg;
1871 I32 minlen; /* must match at least this many chars */
1872 I32 dontbother = 0; /* how many characters not to try at end */
1873 I32 end_shift = 0; /* Same for the end. */ /* CC */
1874 I32 scream_pos = -1; /* Internal iterator of scream. */
1875 char *scream_olds = NULL;
1876 const bool do_utf8 = (bool)DO_UTF8(sv);
1878 RXi_GET_DECL(prog,progi);
1879 regmatch_info reginfo; /* create some info to pass to regtry etc */
1880 regexp_paren_pair *swap = NULL;
1881 GET_RE_DEBUG_FLAGS_DECL;
1883 PERL_ARGS_ASSERT_REGEXEC_FLAGS;
1884 PERL_UNUSED_ARG(data);
1886 /* Be paranoid... */
1887 if (prog == NULL || startpos == NULL) {
1888 Perl_croak(aTHX_ "NULL regexp parameter");
1892 multiline = prog->extflags & RXf_PMf_MULTILINE;
1893 reginfo.prog = rx; /* Yes, sorry that this is confusing. */
1895 RX_MATCH_UTF8_set(rx, do_utf8);
1897 debug_start_match(rx, do_utf8, startpos, strend,
1901 minlen = prog->minlen;
1903 if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
1904 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1905 "String too short [regexec_flags]...\n"));
1910 /* Check validity of program. */
1911 if (UCHARAT(progi->program) != REG_MAGIC) {
1912 Perl_croak(aTHX_ "corrupted regexp program");
1916 PL_reg_eval_set = 0;
1920 PL_reg_flags |= RF_utf8;
1922 /* Mark beginning of line for ^ and lookbehind. */
1923 reginfo.bol = startpos; /* XXX not used ??? */
1927 /* Mark end of line for $ (and such) */
1930 /* see how far we have to get to not match where we matched before */
1931 reginfo.till = startpos+minend;
1933 /* If there is a "must appear" string, look for it. */
1936 if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */
1938 if (flags & REXEC_IGNOREPOS){ /* Means: check only at start */
1939 reginfo.ganch = startpos + prog->gofs;
1940 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
1941 "GPOS IGNOREPOS: reginfo.ganch = startpos + %"UVxf"\n",(UV)prog->gofs));
1942 } else if (sv && SvTYPE(sv) >= SVt_PVMG
1944 && (mg = mg_find(sv, PERL_MAGIC_regex_global))
1945 && mg->mg_len >= 0) {
1946 reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */
1947 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
1948 "GPOS MAGIC: reginfo.ganch = strbeg + %"IVdf"\n",(IV)mg->mg_len));
1950 if (prog->extflags & RXf_ANCH_GPOS) {
1951 if (s > reginfo.ganch)
1953 s = reginfo.ganch - prog->gofs;
1954 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
1955 "GPOS ANCH_GPOS: s = ganch - %"UVxf"\n",(UV)prog->gofs));
1961 reginfo.ganch = strbeg + PTR2UV(data);
1962 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
1963 "GPOS DATA: reginfo.ganch= strbeg + %"UVxf"\n",PTR2UV(data)));
1965 } else { /* pos() not defined */
1966 reginfo.ganch = strbeg;
1967 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
1968 "GPOS: reginfo.ganch = strbeg\n"));
1971 if (PL_curpm && (PM_GETRE(PL_curpm) == rx)) {
1972 /* We have to be careful. If the previous successful match
1973 was from this regex we don't want a subsequent partially
1974 successful match to clobber the old results.
1975 So when we detect this possibility we add a swap buffer
1976 to the re, and switch the buffer each match. If we fail
1977 we switch it back, otherwise we leave it swapped.
1980 /* do we need a save destructor here for eval dies? */
1981 Newxz(prog->offs, (prog->nparens + 1), regexp_paren_pair);
1983 if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) {
1984 re_scream_pos_data d;
1986 d.scream_olds = &scream_olds;
1987 d.scream_pos = &scream_pos;
1988 s = re_intuit_start(rx, sv, s, strend, flags, &d);
1990 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
1991 goto phooey; /* not present */
1997 /* Simplest case: anchored match need be tried only once. */
1998 /* [unless only anchor is BOL and multiline is set] */
1999 if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) {
2000 if (s == startpos && regtry(®info, &startpos))
2002 else if (multiline || (prog->intflags & PREGf_IMPLICIT)
2003 || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */
2008 dontbother = minlen - 1;
2009 end = HOP3c(strend, -dontbother, strbeg) - 1;
2010 /* for multiline we only have to try after newlines */
2011 if (prog->check_substr || prog->check_utf8) {
2012 /* because of the goto we can not easily reuse the macros for bifurcating the
2013 unicode/non-unicode match modes here like we do elsewhere - demerphq */
2016 goto after_try_utf8;
2018 if (regtry(®info, &s)) {
2025 if (prog->extflags & RXf_USE_INTUIT) {
2026 s = re_intuit_start(rx, sv, s + UTF8SKIP(s), strend, flags, NULL);
2035 } /* end search for check string in unicode */
2037 if (s == startpos) {
2038 goto after_try_latin;
2041 if (regtry(®info, &s)) {
2048 if (prog->extflags & RXf_USE_INTUIT) {
2049 s = re_intuit_start(rx, sv, s + 1, strend, flags, NULL);
2058 } /* end search for check string in latin*/
2059 } /* end search for check string */
2060 else { /* search for newline */
2062 /*XXX: The s-- is almost definitely wrong here under unicode - demeprhq*/
2065 /* We can use a more efficient search as newlines are the same in unicode as they are in latin */
2067 if (*s++ == '\n') { /* don't need PL_utf8skip here */
2068 if (regtry(®info, &s))
2072 } /* end search for newline */
2073 } /* end anchored/multiline check string search */
2075 } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK))
2077 /* the warning about reginfo.ganch being used without intialization
2078 is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN
2079 and we only enter this block when the same bit is set. */
2080 char *tmp_s = reginfo.ganch - prog->gofs;
2082 if (tmp_s >= strbeg && regtry(®info, &tmp_s))
2087 /* Messy cases: unanchored match. */
2088 if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) {
2089 /* we have /x+whatever/ */
2090 /* it must be a one character string (XXXX Except UTF?) */
2095 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
2096 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
2097 ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0];
2102 DEBUG_EXECUTE_r( did_match = 1 );
2103 if (regtry(®info, &s)) goto got_it;
2105 while (s < strend && *s == ch)
2113 DEBUG_EXECUTE_r( did_match = 1 );
2114 if (regtry(®info, &s)) goto got_it;
2116 while (s < strend && *s == ch)
2121 DEBUG_EXECUTE_r(if (!did_match)
2122 PerlIO_printf(Perl_debug_log,
2123 "Did not find anchored character...\n")
2126 else if (prog->anchored_substr != NULL
2127 || prog->anchored_utf8 != NULL
2128 || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
2129 && prog->float_max_offset < strend - s)) {
2134 char *last1; /* Last position checked before */
2138 if (prog->anchored_substr || prog->anchored_utf8) {
2139 if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
2140 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
2141 must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
2142 back_max = back_min = prog->anchored_offset;
2144 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
2145 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
2146 must = do_utf8 ? prog->float_utf8 : prog->float_substr;
2147 back_max = prog->float_max_offset;
2148 back_min = prog->float_min_offset;
2152 if (must == &PL_sv_undef)
2153 /* could not downgrade utf8 check substring, so must fail */
2159 last = HOP3c(strend, /* Cannot start after this */
2160 -(I32)(CHR_SVLEN(must)
2161 - (SvTAIL(must) != 0) + back_min), strbeg);
2164 last1 = HOPc(s, -1);
2166 last1 = s - 1; /* bogus */
2168 /* XXXX check_substr already used to find "s", can optimize if
2169 check_substr==must. */
2171 dontbother = end_shift;
2172 strend = HOPc(strend, -dontbother);
2173 while ( (s <= last) &&
2174 ((flags & REXEC_SCREAM)
2175 ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg,
2176 end_shift, &scream_pos, 0))
2177 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)),
2178 (unsigned char*)strend, must,
2179 multiline ? FBMrf_MULTILINE : 0))) ) {
2180 /* we may be pointing at the wrong string */
2181 if ((flags & REXEC_SCREAM) && RXp_MATCH_COPIED(prog))
2182 s = strbeg + (s - SvPVX_const(sv));
2183 DEBUG_EXECUTE_r( did_match = 1 );
2184 if (HOPc(s, -back_max) > last1) {
2185 last1 = HOPc(s, -back_min);
2186 s = HOPc(s, -back_max);
2189 char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
2191 last1 = HOPc(s, -back_min);
2195 while (s <= last1) {
2196 if (regtry(®info, &s))
2202 while (s <= last1) {
2203 if (regtry(®info, &s))
2209 DEBUG_EXECUTE_r(if (!did_match) {
2210 RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0),
2211 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
2212 PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n",
2213 ((must == prog->anchored_substr || must == prog->anchored_utf8)
2214 ? "anchored" : "floating"),
2215 quoted, RE_SV_TAIL(must));
2219 else if ( (c = progi->regstclass) ) {
2221 const OPCODE op = OP(progi->regstclass);
2222 /* don't bother with what can't match */
2223 if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE)
2224 strend = HOPc(strend, -(minlen - 1));
2227 SV * const prop = sv_newmortal();
2228 regprop(prog, prop, c);
2230 RE_PV_QUOTED_DECL(quoted,do_utf8,PERL_DEBUG_PAD_ZERO(1),
2232 PerlIO_printf(Perl_debug_log,
2233 "Matching stclass %.*s against %s (%d chars)\n",
2234 (int)SvCUR(prop), SvPVX_const(prop),
2235 quoted, (int)(strend - s));
2238 if (find_byclass(prog, c, s, strend, ®info))
2240 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n"));
2244 if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
2249 if (!(do_utf8 ? prog->float_utf8 : prog->float_substr))
2250 do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
2251 float_real = do_utf8 ? prog->float_utf8 : prog->float_substr;
2253 if (flags & REXEC_SCREAM) {
2254 last = screaminstr(sv, float_real, s - strbeg,
2255 end_shift, &scream_pos, 1); /* last one */
2257 last = scream_olds; /* Only one occurrence. */
2258 /* we may be pointing at the wrong string */
2259 else if (RXp_MATCH_COPIED(prog))
2260 s = strbeg + (s - SvPVX_const(sv));
2264 const char * const little = SvPV_const(float_real, len);
2266 if (SvTAIL(float_real)) {
2267 if (memEQ(strend - len + 1, little, len - 1))
2268 last = strend - len + 1;
2269 else if (!multiline)
2270 last = memEQ(strend - len, little, len)
2271 ? strend - len : NULL;
2277 last = rninstr(s, strend, little, little + len);
2279 last = strend; /* matching "$" */
2284 PerlIO_printf(Perl_debug_log,
2285 "%sCan't trim the tail, match fails (should not happen)%s\n",
2286 PL_colors[4], PL_colors[5]));
2287 goto phooey; /* Should not happen! */
2289 dontbother = strend - last + prog->float_min_offset;
2291 if (minlen && (dontbother < minlen))
2292 dontbother = minlen - 1;
2293 strend -= dontbother; /* this one's always in bytes! */
2294 /* We don't know much -- general case. */
2297 if (regtry(®info, &s))
2306 if (regtry(®info, &s))
2308 } while (s++ < strend);
2317 RX_MATCH_TAINTED_set(rx, PL_reg_flags & RF_tainted);
2319 if (PL_reg_eval_set)
2320 restore_pos(aTHX_ prog);
2321 if (RXp_PAREN_NAMES(prog))
2322 (void)hv_iterinit(RXp_PAREN_NAMES(prog));
2324 /* make sure $`, $&, $', and $digit will work later */
2325 if ( !(flags & REXEC_NOT_FIRST) ) {
2326 RX_MATCH_COPY_FREE(rx);
2327 if (flags & REXEC_COPY_STR) {
2328 const I32 i = PL_regeol - startpos + (stringarg - strbeg);
2329 #ifdef PERL_OLD_COPY_ON_WRITE
2331 || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
2333 PerlIO_printf(Perl_debug_log,
2334 "Copy on write: regexp capture, type %d\n",
2337 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
2338 prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
2339 assert (SvPOKp(prog->saved_copy));
2343 RX_MATCH_COPIED_on(rx);
2344 s = savepvn(strbeg, i);
2350 prog->subbeg = strbeg;
2351 prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
2358 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
2359 PL_colors[4], PL_colors[5]));
2360 if (PL_reg_eval_set)
2361 restore_pos(aTHX_ prog);
2363 /* we failed :-( roll it back */
2364 Safefree(prog->offs);
2373 - regtry - try match at specific point
2375 STATIC I32 /* 0 failure, 1 success */
2376 S_regtry(pTHX_ regmatch_info *reginfo, char **startpos)
2380 REGEXP *const rx = reginfo->prog;
2381 regexp *const prog = (struct regexp *)SvANY(rx);
2382 RXi_GET_DECL(prog,progi);
2383 GET_RE_DEBUG_FLAGS_DECL;
2385 PERL_ARGS_ASSERT_REGTRY;
2387 reginfo->cutpoint=NULL;
2389 if ((prog->extflags & RXf_EVAL_SEEN) && !PL_reg_eval_set) {
2392 PL_reg_eval_set = RS_init;
2393 DEBUG_EXECUTE_r(DEBUG_s(
2394 PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
2395 (IV)(PL_stack_sp - PL_stack_base));
2398 cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
2399 /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
2401 /* Apparently this is not needed, judging by wantarray. */
2402 /* SAVEI8(cxstack[cxstack_ix].blk_gimme);
2403 cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
2406 /* Make $_ available to executed code. */
2407 if (reginfo->sv != DEFSV) {
2409 DEFSV_set(reginfo->sv);
2412 if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv)
2413 && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) {
2414 /* prepare for quick setting of pos */
2415 #ifdef PERL_OLD_COPY_ON_WRITE
2416 if (SvIsCOW(reginfo->sv))
2417 sv_force_normal_flags(reginfo->sv, 0);
2419 mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global,
2420 &PL_vtbl_mglob, NULL, 0);
2424 PL_reg_oldpos = mg->mg_len;
2425 SAVEDESTRUCTOR_X(restore_pos, prog);
2427 if (!PL_reg_curpm) {
2428 Newxz(PL_reg_curpm, 1, PMOP);
2431 SV* const repointer = &PL_sv_undef;
2432 /* this regexp is also owned by the new PL_reg_curpm, which
2433 will try to free it. */
2434 av_push(PL_regex_padav, repointer);
2435 PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav);
2436 PL_regex_pad = AvARRAY(PL_regex_padav);
2441 /* It seems that non-ithreads works both with and without this code.
2442 So for efficiency reasons it seems best not to have the code
2443 compiled when it is not needed. */
2444 /* This is safe against NULLs: */
2445 ReREFCNT_dec(PM_GETRE(PL_reg_curpm));
2446 /* PM_reg_curpm owns a reference to this regexp. */
2449 PM_SETRE(PL_reg_curpm, rx);
2450 PL_reg_oldcurpm = PL_curpm;
2451 PL_curpm = PL_reg_curpm;
2452 if (RXp_MATCH_COPIED(prog)) {
2453 /* Here is a serious problem: we cannot rewrite subbeg,
2454 since it may be needed if this match fails. Thus
2455 $` inside (?{}) could fail... */
2456 PL_reg_oldsaved = prog->subbeg;
2457 PL_reg_oldsavedlen = prog->sublen;
2458 #ifdef PERL_OLD_COPY_ON_WRITE
2459 PL_nrs = prog->saved_copy;
2461 RXp_MATCH_COPIED_off(prog);
2464 PL_reg_oldsaved = NULL;
2465 prog->subbeg = PL_bostr;
2466 prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
2468 DEBUG_EXECUTE_r(PL_reg_starttry = *startpos);
2469 prog->offs[0].start = *startpos - PL_bostr;
2470 PL_reginput = *startpos;
2471 PL_reglastparen = &prog->lastparen;
2472 PL_reglastcloseparen = &prog->lastcloseparen;
2473 prog->lastparen = 0;
2474 prog->lastcloseparen = 0;
2476 PL_regoffs = prog->offs;
2477 if (PL_reg_start_tmpl <= prog->nparens) {
2478 PL_reg_start_tmpl = prog->nparens*3/2 + 3;
2479 if(PL_reg_start_tmp)
2480 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2482 Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
2485 /* XXXX What this code is doing here?!!! There should be no need
2486 to do this again and again, PL_reglastparen should take care of
2489 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
2490 * Actually, the code in regcppop() (which Ilya may be meaning by
2491 * PL_reglastparen), is not needed at all by the test suite
2492 * (op/regexp, op/pat, op/split), but that code is needed otherwise
2493 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
2494 * Meanwhile, this code *is* needed for the
2495 * above-mentioned test suite tests to succeed. The common theme
2496 * on those tests seems to be returning null fields from matches.
2497 * --jhi updated by dapm */
2499 if (prog->nparens) {
2500 regexp_paren_pair *pp = PL_regoffs;
2502 for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) {
2510 if (regmatch(reginfo, progi->program + 1)) {
2511 PL_regoffs[0].end = PL_reginput - PL_bostr;
2514 if (reginfo->cutpoint)
2515 *startpos= reginfo->cutpoint;
2516 REGCP_UNWIND(lastcp);
2521 #define sayYES goto yes
2522 #define sayNO goto no
2523 #define sayNO_SILENT goto no_silent
2525 /* we dont use STMT_START/END here because it leads to
2526 "unreachable code" warnings, which are bogus, but distracting. */
2527 #define CACHEsayNO \
2528 if (ST.cache_mask) \
2529 PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \
2532 /* this is used to determine how far from the left messages like
2533 'failed...' are printed. It should be set such that messages
2534 are inline with the regop output that created them.
2536 #define REPORT_CODE_OFF 32
2539 /* Make sure there is a test for this +1 options in re_tests */
2540 #define TRIE_INITAL_ACCEPT_BUFFLEN 4;
2542 #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
2543 #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */
2545 #define SLAB_FIRST(s) (&(s)->states[0])
2546 #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
2548 /* grab a new slab and return the first slot in it */
2550 STATIC regmatch_state *
2553 #if PERL_VERSION < 9 && !defined(PERL_CORE)
2556 regmatch_slab *s = PL_regmatch_slab->next;
2558 Newx(s, 1, regmatch_slab);
2559 s->prev = PL_regmatch_slab;
2561 PL_regmatch_slab->next = s;
2563 PL_regmatch_slab = s;
2564 return SLAB_FIRST(s);
2568 /* push a new state then goto it */
2570 #define PUSH_STATE_GOTO(state, node) \
2572 st->resume_state = state; \
2575 /* push a new state with success backtracking, then goto it */
2577 #define PUSH_YES_STATE_GOTO(state, node) \
2579 st->resume_state = state; \
2580 goto push_yes_state;
2586 regmatch() - main matching routine
2588 This is basically one big switch statement in a loop. We execute an op,
2589 set 'next' to point the next op, and continue. If we come to a point which
2590 we may need to backtrack to on failure such as (A|B|C), we push a
2591 backtrack state onto the backtrack stack. On failure, we pop the top
2592 state, and re-enter the loop at the state indicated. If there are no more
2593 states to pop, we return failure.
2595 Sometimes we also need to backtrack on success; for example /A+/, where
2596 after successfully matching one A, we need to go back and try to
2597 match another one; similarly for lookahead assertions: if the assertion
2598 completes successfully, we backtrack to the state just before the assertion
2599 and then carry on. In these cases, the pushed state is marked as
2600 'backtrack on success too'. This marking is in fact done by a chain of
2601 pointers, each pointing to the previous 'yes' state. On success, we pop to
2602 the nearest yes state, discarding any intermediate failure-only states.
2603 Sometimes a yes state is pushed just to force some cleanup code to be
2604 called at the end of a successful match or submatch; e.g. (??{$re}) uses
2605 it to free the inner regex.
2607 Note that failure backtracking rewinds the cursor position, while
2608 success backtracking leaves it alone.
2610 A pattern is complete when the END op is executed, while a subpattern
2611 such as (?=foo) is complete when the SUCCESS op is executed. Both of these
2612 ops trigger the "pop to last yes state if any, otherwise return true"
2615 A common convention in this function is to use A and B to refer to the two
2616 subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is
2617 the subpattern to be matched possibly multiple times, while B is the entire
2618 rest of the pattern. Variable and state names reflect this convention.
2620 The states in the main switch are the union of ops and failure/success of
2621 substates associated with with that op. For example, IFMATCH is the op
2622 that does lookahead assertions /(?=A)B/ and so the IFMATCH state means
2623 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just
2624 successfully matched A and IFMATCH_A_fail is a state saying that we have
2625 just failed to match A. Resume states always come in pairs. The backtrack
2626 state we push is marked as 'IFMATCH_A', but when that is popped, we resume
2627 at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking
2628 on success or failure.
2630 The struct that holds a backtracking state is actually a big union, with
2631 one variant for each major type of op. The variable st points to the
2632 top-most backtrack struct. To make the code clearer, within each
2633 block of code we #define ST to alias the relevant union.
2635 Here's a concrete example of a (vastly oversimplified) IFMATCH
2641 #define ST st->u.ifmatch
2643 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2644 ST.foo = ...; // some state we wish to save
2646 // push a yes backtrack state with a resume value of
2647 // IFMATCH_A/IFMATCH_A_fail, then continue execution at the
2649 PUSH_YES_STATE_GOTO(IFMATCH_A, A);
2652 case IFMATCH_A: // we have successfully executed A; now continue with B
2654 bar = ST.foo; // do something with the preserved value
2657 case IFMATCH_A_fail: // A failed, so the assertion failed
2658 ...; // do some housekeeping, then ...
2659 sayNO; // propagate the failure
2666 For any old-timers reading this who are familiar with the old recursive
2667 approach, the code above is equivalent to:
2669 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2678 ...; // do some housekeeping, then ...
2679 sayNO; // propagate the failure
2682 The topmost backtrack state, pointed to by st, is usually free. If you
2683 want to claim it, populate any ST.foo fields in it with values you wish to
2684 save, then do one of
2686 PUSH_STATE_GOTO(resume_state, node);
2687 PUSH_YES_STATE_GOTO(resume_state, node);
2689 which sets that backtrack state's resume value to 'resume_state', pushes a
2690 new free entry to the top of the backtrack stack, then goes to 'node'.
2691 On backtracking, the free slot is popped, and the saved state becomes the
2692 new free state. An ST.foo field in this new top state can be temporarily
2693 accessed to retrieve values, but once the main loop is re-entered, it
2694 becomes available for reuse.
2696 Note that the depth of the backtrack stack constantly increases during the
2697 left-to-right execution of the pattern, rather than going up and down with
2698 the pattern nesting. For example the stack is at its maximum at Z at the
2699 end of the pattern, rather than at X in the following:
2701 /(((X)+)+)+....(Y)+....Z/
2703 The only exceptions to this are lookahead/behind assertions and the cut,
2704 (?>A), which pop all the backtrack states associated with A before
2707 Bascktrack state structs are allocated in slabs of about 4K in size.
2708 PL_regmatch_state and st always point to the currently active state,
2709 and PL_regmatch_slab points to the slab currently containing
2710 PL_regmatch_state. The first time regmatch() is called, the first slab is
2711 allocated, and is never freed until interpreter destruction. When the slab
2712 is full, a new one is allocated and chained to the end. At exit from
2713 regmatch(), slabs allocated since entry are freed.
2718 #define DEBUG_STATE_pp(pp) \
2720 DUMP_EXEC_POS(locinput, scan, do_utf8); \
2721 PerlIO_printf(Perl_debug_log, \
2722 " %*s"pp" %s%s%s%s%s\n", \
2724 PL_reg_name[st->resume_state], \
2725 ((st==yes_state||st==mark_state) ? "[" : ""), \
2726 ((st==yes_state) ? "Y" : ""), \
2727 ((st==mark_state) ? "M" : ""), \
2728 ((st==yes_state||st==mark_state) ? "]" : "") \
2733 #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1)
2738 S_debug_start_match(pTHX_ const REGEXP *prog, const bool do_utf8,
2739 const char *start, const char *end, const char *blurb)
2741 const bool utf8_pat = RX_UTF8(prog) ? 1 : 0;
2743 PERL_ARGS_ASSERT_DEBUG_START_MATCH;
2748 RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0),
2749 RX_PRECOMP_const(prog), RX_PRELEN(prog), 60);
2751 RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1),
2752 start, end - start, 60);
2754 PerlIO_printf(Perl_debug_log,
2755 "%s%s REx%s %s against %s\n",
2756 PL_colors[4], blurb, PL_colors[5], s0, s1);
2758 if (do_utf8||utf8_pat)
2759 PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n",
2760 utf8_pat ? "pattern" : "",
2761 utf8_pat && do_utf8 ? " and " : "",
2762 do_utf8 ? "string" : ""
2768 S_dump_exec_pos(pTHX_ const char *locinput,
2769 const regnode *scan,
2770 const char *loc_regeol,
2771 const char *loc_bostr,
2772 const char *loc_reg_starttry,
2775 const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
2776 const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
2777 int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
2778 /* The part of the string before starttry has one color
2779 (pref0_len chars), between starttry and current
2780 position another one (pref_len - pref0_len chars),
2781 after the current position the third one.
2782 We assume that pref0_len <= pref_len, otherwise we
2783 decrease pref0_len. */
2784 int pref_len = (locinput - loc_bostr) > (5 + taill) - l
2785 ? (5 + taill) - l : locinput - loc_bostr;
2788 PERL_ARGS_ASSERT_DUMP_EXEC_POS;
2790 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
2792 pref0_len = pref_len - (locinput - loc_reg_starttry);
2793 if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
2794 l = ( loc_regeol - locinput > (5 + taill) - pref_len
2795 ? (5 + taill) - pref_len : loc_regeol - locinput);
2796 while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
2800 if (pref0_len > pref_len)
2801 pref0_len = pref_len;
2803 const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0;
2805 RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
2806 (locinput - pref_len),pref0_len, 60, 4, 5);
2808 RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
2809 (locinput - pref_len + pref0_len),
2810 pref_len - pref0_len, 60, 2, 3);
2812 RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
2813 locinput, loc_regeol - locinput, 10, 0, 1);
2815 const STRLEN tlen=len0+len1+len2;
2816 PerlIO_printf(Perl_debug_log,
2817 "%4"IVdf" <%.*s%.*s%s%.*s>%*s|",
2818 (IV)(locinput - loc_bostr),
2821 (docolor ? "" : "> <"),
2823 (int)(tlen > 19 ? 0 : 19 - tlen),
2830 /* reg_check_named_buff_matched()
2831 * Checks to see if a named buffer has matched. The data array of
2832 * buffer numbers corresponding to the buffer is expected to reside
2833 * in the regexp->data->data array in the slot stored in the ARG() of
2834 * node involved. Note that this routine doesn't actually care about the
2835 * name, that information is not preserved from compilation to execution.
2836 * Returns the index of the leftmost defined buffer with the given name
2837 * or 0 if non of the buffers matched.
2840 S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan)
2843 RXi_GET_DECL(rex,rexi);
2844 SV *sv_dat= MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
2845 I32 *nums=(I32*)SvPVX(sv_dat);
2847 PERL_ARGS_ASSERT_REG_CHECK_NAMED_BUFF_MATCHED;
2849 for ( n=0; n<SvIVX(sv_dat); n++ ) {
2850 if ((I32)*PL_reglastparen >= nums[n] &&
2851 PL_regoffs[nums[n]].end != -1)
2860 /* free all slabs above current one - called during LEAVE_SCOPE */
2863 S_clear_backtrack_stack(pTHX_ void *p)
2865 regmatch_slab *s = PL_regmatch_slab->next;
2870 PL_regmatch_slab->next = NULL;
2872 regmatch_slab * const osl = s;
2879 #define SETREX(Re1,Re2) \
2880 if (PL_reg_eval_set) PM_SETRE((PL_reg_curpm), (Re2)); \
2883 STATIC I32 /* 0 failure, 1 success */
2884 S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
2886 #if PERL_VERSION < 9 && !defined(PERL_CORE)
2890 register const bool do_utf8 = PL_reg_match_utf8;
2891 const U32 uniflags = UTF8_ALLOW_DEFAULT;
2892 REGEXP *rex_sv = reginfo->prog;
2893 regexp *rex = (struct regexp *)SvANY(rex_sv);
2894 RXi_GET_DECL(rex,rexi);
2896 /* the current state. This is a cached copy of PL_regmatch_state */
2897 register regmatch_state *st;
2898 /* cache heavy used fields of st in registers */
2899 register regnode *scan;
2900 register regnode *next;
2901 register U32 n = 0; /* general value; init to avoid compiler warning */
2902 register I32 ln = 0; /* len or last; init to avoid compiler warning */
2903 register char *locinput = PL_reginput;
2904 register I32 nextchr; /* is always set to UCHARAT(locinput) */
2906 bool result = 0; /* return value of S_regmatch */
2907 int depth = 0; /* depth of backtrack stack */
2908 U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */
2909 const U32 max_nochange_depth =
2910 (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ?
2911 3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH;
2912 regmatch_state *yes_state = NULL; /* state to pop to on success of
2914 /* mark_state piggy backs on the yes_state logic so that when we unwind
2915 the stack on success we can update the mark_state as we go */
2916 regmatch_state *mark_state = NULL; /* last mark state we have seen */
2917 regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */
2918 struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */
2920 bool no_final = 0; /* prevent failure from backtracking? */
2921 bool do_cutgroup = 0; /* no_final only until next branch/trie entry */
2922 char *startpoint = PL_reginput;
2923 SV *popmark = NULL; /* are we looking for a mark? */
2924 SV *sv_commit = NULL; /* last mark name seen in failure */
2925 SV *sv_yes_mark = NULL; /* last mark name we have seen
2926 during a successfull match */
2927 U32 lastopen = 0; /* last open we saw */
2928 bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0;
2929 SV* const oreplsv = GvSV(PL_replgv);
2930 /* these three flags are set by various ops to signal information to
2931 * the very next op. They have a useful lifetime of exactly one loop
2932 * iteration, and are not preserved or restored by state pushes/pops
2934 bool sw = 0; /* the condition value in (?(cond)a|b) */
2935 bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */
2936 int logical = 0; /* the following EVAL is:
2940 or the following IFMATCH/UNLESSM is:
2941 false: plain (?=foo)
2942 true: used as a condition: (?(?=foo))
2945 GET_RE_DEBUG_FLAGS_DECL;
2948 PERL_ARGS_ASSERT_REGMATCH;
2950 DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({
2951 PerlIO_printf(Perl_debug_log,"regmatch start\n");
2953 /* on first ever call to regmatch, allocate first slab */
2954 if (!PL_regmatch_slab) {
2955 Newx(PL_regmatch_slab, 1, regmatch_slab);
2956 PL_regmatch_slab->prev = NULL;
2957 PL_regmatch_slab->next = NULL;
2958 PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
2961 oldsave = PL_savestack_ix;
2962 SAVEDESTRUCTOR_X(S_clear_backtrack_stack, NULL);
2963 SAVEVPTR(PL_regmatch_slab);
2964 SAVEVPTR(PL_regmatch_state);
2966 /* grab next free state slot */
2967 st = ++PL_regmatch_state;
2968 if (st > SLAB_LAST(PL_regmatch_slab))
2969 st = PL_regmatch_state = S_push_slab(aTHX);
2971 /* Note that nextchr is a byte even in UTF */
2972 nextchr = UCHARAT(locinput);
2974 while (scan != NULL) {
2977 SV * const prop = sv_newmortal();
2978 regnode *rnext=regnext(scan);
2979 DUMP_EXEC_POS( locinput, scan, do_utf8 );
2980 regprop(rex, prop, scan);
2982 PerlIO_printf(Perl_debug_log,
2983 "%3"IVdf":%*s%s(%"IVdf")\n",
2984 (IV)(scan - rexi->program), depth*2, "",
2986 (PL_regkind[OP(scan)] == END || !rnext) ?
2987 0 : (IV)(rnext - rexi->program));
2990 next = scan + NEXT_OFF(scan);
2993 state_num = OP(scan);
2995 REH_CALL_EXEC_NODE_HOOK(rex, scan, reginfo, st);
2998 assert(PL_reglastparen == &rex->lastparen);
2999 assert(PL_reglastcloseparen == &rex->lastcloseparen);
3000 assert(PL_regoffs == rex->offs);
3002 switch (state_num) {
3004 if (locinput == PL_bostr)
3006 /* reginfo->till = reginfo->bol; */
3011 if (locinput == PL_bostr ||
3012 ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n'))
3018 if (locinput == PL_bostr)
3022 if (locinput == reginfo->ganch)
3027 /* update the startpoint */
3028 st->u.keeper.val = PL_regoffs[0].start;
3029 PL_reginput = locinput;
3030 PL_regoffs[0].start = locinput - PL_bostr;
3031 PUSH_STATE_GOTO(KEEPS_next, next);
3033 case KEEPS_next_fail:
3034 /* rollback the start point change */
3035 PL_regoffs[0].start = st->u.keeper.val;
3041 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
3046 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
3048 if (PL_regeol - locinput > 1)
3052 if (PL_regeol != locinput)
3056 if (!nextchr && locinput >= PL_regeol)
3059 locinput += PL_utf8skip[nextchr];
3060 if (locinput > PL_regeol)
3062 nextchr = UCHARAT(locinput);
3065 nextchr = UCHARAT(++locinput);
3068 if (!nextchr && locinput >= PL_regeol)
3070 nextchr = UCHARAT(++locinput);
3073 if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
3076 locinput += PL_utf8skip[nextchr];
3077 if (locinput > PL_regeol)
3079 nextchr = UCHARAT(locinput);
3082 nextchr = UCHARAT(++locinput);
3086 #define ST st->u.trie
3088 /* In this case the charclass data is available inline so
3089 we can fail fast without a lot of extra overhead.
3091 if (scan->flags == EXACT || !do_utf8) {
3092 if(!ANYOF_BITMAP_TEST(scan, *locinput)) {
3094 PerlIO_printf(Perl_debug_log,
3095 "%*s %sfailed to match trie start class...%s\n",
3096 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3105 /* what type of TRIE am I? (utf8 makes this contextual) */
3106 DECL_TRIE_TYPE(scan);
3108 /* what trie are we using right now */
3109 reg_trie_data * const trie
3110 = (reg_trie_data*)rexi->data->data[ ARG( scan ) ];
3111 HV * widecharmap = MUTABLE_HV(rexi->data->data[ ARG( scan ) + 1 ]);
3112 U32 state = trie->startstate;
3114 if (trie->bitmap && trie_type != trie_utf8_fold &&
3115 !TRIE_BITMAP_TEST(trie,*locinput)
3117 if (trie->states[ state ].wordnum) {
3119 PerlIO_printf(Perl_debug_log,
3120 "%*s %smatched empty string...%s\n",
3121 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3126 PerlIO_printf(Perl_debug_log,
3127 "%*s %sfailed to match trie start class...%s\n",
3128 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3135 U8 *uc = ( U8* )locinput;
3139 U8 *uscan = (U8*)NULL;
3141 SV *sv_accept_buff = NULL;
3142 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
3144 ST.accepted = 0; /* how many accepting states we have seen */
3146 ST.jump = trie->jump;
3149 traverse the TRIE keeping track of all accepting states
3150 we transition through until we get to a failing node.
3153 while ( state && uc <= (U8*)PL_regeol ) {
3154 U32 base = trie->states[ state ].trans.base;
3157 /* We use charid to hold the wordnum as we don't use it
3158 for charid until after we have done the wordnum logic.
3159 We define an alias just so that the wordnum logic reads
3162 #define got_wordnum charid
3163 got_wordnum = trie->states[ state ].wordnum;
3165 if ( got_wordnum ) {
3166 if ( ! ST.accepted ) {
3168 SAVETMPS; /* XXX is this necessary? dmq */
3169 bufflen = TRIE_INITAL_ACCEPT_BUFFLEN;
3170 sv_accept_buff=newSV(bufflen *
3171 sizeof(reg_trie_accepted) - 1);
3172 SvCUR_set(sv_accept_buff, 0);
3173 SvPOK_on(sv_accept_buff);
3174 sv_2mortal(sv_accept_buff);
3177 (reg_trie_accepted*)SvPV_nolen(sv_accept_buff );
3180 if (ST.accepted >= bufflen) {
3182 ST.accept_buff =(reg_trie_accepted*)
3183 SvGROW(sv_accept_buff,
3184 bufflen * sizeof(reg_trie_accepted));
3186 SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff)
3187 + sizeof(reg_trie_accepted));
3190 ST.accept_buff[ST.accepted].wordnum = got_wordnum;
3191 ST.accept_buff[ST.accepted].endpos = uc;
3193 } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum]));
3197 DEBUG_TRIE_EXECUTE_r({
3198 DUMP_EXEC_POS( (char *)uc, scan, do_utf8 );
3199 PerlIO_printf( Perl_debug_log,
3200 "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ",
3201 2+depth * 2, "", PL_colors[4],
3202 (UV)state, (UV)ST.accepted );
3206 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
3207 uscan, len, uvc, charid, foldlen,
3211 (base + charid > trie->uniquecharcount )
3212 && (base + charid - 1 - trie->uniquecharcount
3214 && trie->trans[base + charid - 1 -
3215 trie->uniquecharcount].check == state)
3217 state = trie->trans[base + charid - 1 -
3218 trie->uniquecharcount ].next;
3229 DEBUG_TRIE_EXECUTE_r(
3230 PerlIO_printf( Perl_debug_log,
3231 "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n",
3232 charid, uvc, (UV)state, PL_colors[5] );
3239 PerlIO_printf( Perl_debug_log,
3240 "%*s %sgot %"IVdf" possible matches%s\n",
3241 REPORT_CODE_OFF + depth * 2, "",
3242 PL_colors[4], (IV)ST.accepted, PL_colors[5] );
3245 goto trie_first_try; /* jump into the fail handler */
3247 case TRIE_next_fail: /* we failed - try next alterative */
3249 REGCP_UNWIND(ST.cp);
3250 for (n = *PL_reglastparen; n > ST.lastparen; n--)
3251 PL_regoffs[n].end = -1;
3252 *PL_reglastparen = n;
3261 ST.lastparen = *PL_reglastparen;
3264 if ( ST.accepted == 1 ) {
3265 /* only one choice left - just continue */
3267 AV *const trie_words
3268 = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]);
3269 SV ** const tmp = av_fetch( trie_words,
3270 ST.accept_buff[ 0 ].wordnum-1, 0 );
3271 SV *sv= tmp ? sv_newmortal() : NULL;
3273 PerlIO_printf( Perl_debug_log,
3274 "%*s %sonly one match left: #%d <%s>%s\n",
3275 REPORT_CODE_OFF+depth*2, "", PL_colors[4],
3276 ST.accept_buff[ 0 ].wordnum,
3277 tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
3278 PL_colors[0], PL_colors[1],
3279 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)
3281 : "not compiled under -Dr",
3284 PL_reginput = (char *)ST.accept_buff[ 0 ].endpos;
3285 /* in this case we free tmps/leave before we call regmatch
3286 as we wont be using accept_buff again. */
3288 locinput = PL_reginput;
3289 nextchr = UCHARAT(locinput);
3290 if ( !ST.jump || !ST.jump[ST.accept_buff[0].wordnum])
3293 scan = ST.me + ST.jump[ST.accept_buff[0].wordnum];
3294 if (!has_cutgroup) {
3299 PUSH_YES_STATE_GOTO(TRIE_next, scan);
3302 continue; /* execute rest of RE */
3305 if ( !ST.accepted-- ) {
3307 PerlIO_printf( Perl_debug_log,
3308 "%*s %sTRIE failed...%s\n",
3309 REPORT_CODE_OFF+depth*2, "",
3320 There are at least two accepting states left. Presumably
3321 the number of accepting states is going to be low,
3322 typically two. So we simply scan through to find the one
3323 with lowest wordnum. Once we find it, we swap the last
3324 state into its place and decrement the size. We then try to
3325 match the rest of the pattern at the point where the word
3326 ends. If we succeed, control just continues along the
3327 regex; if we fail we return here to try the next accepting
3334 for( cur = 1 ; cur <= ST.accepted ; cur++ ) {
3335 DEBUG_TRIE_EXECUTE_r(
3336 PerlIO_printf( Perl_debug_log,
3337 "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n",
3338 REPORT_CODE_OFF + depth * 2, "", PL_colors[4],
3339 (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur,
3340 ST.accept_buff[ cur ].wordnum, PL_colors[5] );
3343 if (ST.accept_buff[cur].wordnum <
3344 ST.accept_buff[best].wordnum)
3349 AV *const trie_words
3350 = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]);
3351 SV ** const tmp = av_fetch( trie_words,
3352 ST.accept_buff[ best ].wordnum - 1, 0 );
3353 regnode *nextop=(!ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) ?
3355 ST.me + ST.jump[ST.accept_buff[best].wordnum];
3356 SV *sv= tmp ? sv_newmortal() : NULL;
3358 PerlIO_printf( Perl_debug_log,
3359 "%*s %strying alternation #%d <%s> at node #%d %s\n",
3360 REPORT_CODE_OFF+depth*2, "", PL_colors[4],
3361 ST.accept_buff[best].wordnum,
3362 tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
3363 PL_colors[0], PL_colors[1],
3364 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)
3365 ) : "not compiled under -Dr",
3366 REG_NODE_NUM(nextop),
3370 if ( best<ST.accepted ) {
3371 reg_trie_accepted tmp = ST.accept_buff[ best ];
3372 ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ];
3373 ST.accept_buff[ ST.accepted ] = tmp;
3376 PL_reginput = (char *)ST.accept_buff[ best ].endpos;
3377 if ( !ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) {
3380 scan = ST.me + ST.jump[ST.accept_buff[best].wordnum];
3382 PUSH_YES_STATE_GOTO(TRIE_next, scan);
3387 /* we dont want to throw this away, see bug 57042*/
3388 if (oreplsv != GvSV(PL_replgv))
3389 sv_setsv(oreplsv, GvSV(PL_replgv));
3396 char *s = STRING(scan);
3398 if (do_utf8 != UTF) {
3399 /* The target and the pattern have differing utf8ness. */
3401 const char * const e = s + ln;
3404 /* The target is utf8, the pattern is not utf8. */
3409 if (NATIVE_TO_UNI(*(U8*)s) !=
3410 utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
3418 /* The target is not utf8, the pattern is utf8. */
3423 if (NATIVE_TO_UNI(*((U8*)l)) !=
3424 utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
3432 nextchr = UCHARAT(locinput);
3435 /* The target and the pattern have the same utf8ness. */
3436 /* Inline the first character, for speed. */
3437 if (UCHARAT(s) != nextchr)
3439 if (PL_regeol - locinput < ln)
3441 if (ln > 1 && memNE(s, locinput, ln))
3444 nextchr = UCHARAT(locinput);
3448 PL_reg_flags |= RF_tainted;
3451 char * const s = STRING(scan);
3454 if (do_utf8 || UTF) {
3455 /* Either target or the pattern are utf8. */
3456 const char * const l = locinput;
3457 char *e = PL_regeol;
3459 if (ibcmp_utf8(s, 0, ln, (bool)UTF,
3460 l, &e, 0, do_utf8)) {
3461 /* One more case for the sharp s:
3462 * pack("U0U*", 0xDF) =~ /ss/i,
3463 * the 0xC3 0x9F are the UTF-8
3464 * byte sequence for the U+00DF. */
3467 toLOWER(s[0]) == 's' &&
3469 toLOWER(s[1]) == 's' &&
3476 nextchr = UCHARAT(locinput);
3480 /* Neither the target and the pattern are utf8. */
3482 /* Inline the first character, for speed. */
3483 if (UCHARAT(s) != nextchr &&
3484 UCHARAT(s) != ((OP(scan) == EXACTF)
3485 ? PL_fold : PL_fold_locale)[nextchr])
3487 if (PL_regeol - locinput < ln)
3489 if (ln > 1 && (OP(scan) == EXACTF
3490 ? ibcmp(s, locinput, ln)
3491 : ibcmp_locale(s, locinput, ln)))
3494 nextchr = UCHARAT(locinput);
3499 PL_reg_flags |= RF_tainted;
3503 /* was last char in word? */
3505 if (locinput == PL_bostr)
3508 const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);
3510 ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags);
3512 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
3513 ln = isALNUM_uni(ln);
3514 LOAD_UTF8_CHARCLASS_ALNUM();
3515 n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8);
3518 ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln));
3519 n = isALNUM_LC_utf8((U8*)locinput);
3523 ln = (locinput != PL_bostr) ?
3524 UCHARAT(locinput - 1) : '\n';
3525 if (OP(scan) == BOUND || OP(scan) == NBOUND) {
3527 n = isALNUM(nextchr);
3530 ln = isALNUM_LC(ln);
3531 n = isALNUM_LC(nextchr);
3534 if (((!ln) == (!n)) == (OP(scan) == BOUND ||
3535 OP(scan) == BOUNDL))
3540 STRLEN inclasslen = PL_regeol - locinput;
3542 if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8))
3544 if (locinput >= PL_regeol)
3546 locinput += inclasslen ? inclasslen : UTF8SKIP(locinput);
3547 nextchr = UCHARAT(locinput);
3552 nextchr = UCHARAT(locinput);
3553 if (!REGINCLASS(rex, scan, (U8*)locinput))
3555 if (!nextchr && locinput >= PL_regeol)
3557 nextchr = UCHARAT(++locinput);
3561 /* If we might have the case of the German sharp s
3562 * in a casefolding Unicode character class. */
3564 if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) {
3565 locinput += SHARP_S_SKIP;
3566 nextchr = UCHARAT(locinput);
3571 /* Special char classes - The defines start on line 129 or so */
3572 CCC_TRY_AFF( ALNUM, ALNUML, perl_word, "a", isALNUM_LC_utf8, isALNUM, isALNUM_LC);
3573 CCC_TRY_NEG(NALNUM, NALNUML, perl_word, "a", isALNUM_LC_utf8, isALNUM, isALNUM_LC);
3575 CCC_TRY_AFF( SPACE, SPACEL, perl_space, " ", isSPACE_LC_utf8, isSPACE, isSPACE_LC);
3576 CCC_TRY_NEG(NSPACE, NSPACEL, perl_space, " ", isSPACE_LC_utf8, isSPACE, isSPACE_LC);
3578 CCC_TRY_AFF( DIGIT, DIGITL, posix_digit, "0", isDIGIT_LC_utf8, isDIGIT, isDIGIT_LC);
3579 CCC_TRY_NEG(NDIGIT, NDIGITL, posix_digit, "0", isDIGIT_LC_utf8, isDIGIT, isDIGIT_LC);
3581 case CLUMP: /* Match \X: logical Unicode character. This is defined as
3582 a Unicode extended Grapheme Cluster */
3583 /* From http://www.unicode.org/reports/tr29 (5.2 version). An
3584 extended Grapheme Cluster is:
3587 | Prepend* Begin Extend*
3590 Begin is (Hangul-syllable | ! Control)
3591 Extend is (Grapheme_Extend | Spacing_Mark)
3592 Control is [ GCB_Control CR LF ]
3594 The discussion below shows how the code for CLUMP is derived
3595 from this regex. Note that most of these concepts are from
3596 property values of the Grapheme Cluster Boundary (GCB) property.
3597 No code point can have multiple property values for a given
3598 property. Thus a code point in Prepend can't be in Control, but
3599 it must be in !Control. This is why Control above includes
3600 GCB_Control plus CR plus LF. The latter two are used in the GCB
3601 property separately, and so can't be in GCB_Control, even though
3602 they logically are controls. Control is not the same as gc=cc,
3603 but includes format and other characters as well.
3605 The Unicode definition of Hangul-syllable is:
3607 | (L* ( ( V | LV ) V* | LVT ) T*)
3610 Each of these is a value for the GCB property, and hence must be
3611 disjoint, so the order they are tested is immaterial, so the
3612 above can safely be changed to
3615 | (L* ( LVT | ( V | LV ) V*) T*)
3617 The last two terms can be combined like this:
3619 | (( LVT | ( V | LV ) V*) T*))
3621 And refactored into this:
3622 L* (L | LVT T* | V V* T* | LV V* T*)
3624 That means that if we have seen any L's at all we can quit
3625 there, but if the next character is a LVT, a V or and LV we
3628 There is a subtlety with Prepend* which showed up in testing.
3629 Note that the Begin, and only the Begin is required in:
3630 | Prepend* Begin Extend*
3631 Also, Begin contains '! Control'. A Prepend must be a '!
3632 Control', which means it must be a Begin. What it comes down to
3633 is that if we match Prepend* and then find no suitable Begin
3634 afterwards, that if we backtrack the last Prepend, that one will
3635 be a suitable Begin.
3638 if (locinput >= PL_regeol)
3642 /* Match either CR LF or '.', as all the other possibilities
3644 locinput++; /* Match the . or CR */
3646 && locinput < PL_regeol
3647 && UCHARAT(locinput) == '\n') locinput++;
3651 /* Utf8: See if is ( CR LF ); already know that locinput <
3652 * PL_regeol, so locinput+1 is in bounds */
3653 if (nextchr == '\r' && UCHARAT(locinput + 1) == '\n') {
3657 /* In case have to backtrack to beginning, then match '.' */
3658 char *starting = locinput;
3660 /* In case have to backtrack the last prepend */
3661 char *previous_prepend = 0;
3663 LOAD_UTF8_CHARCLASS_GCB();
3665 /* Match (prepend)* */
3666 while (locinput < PL_regeol
3667 && swash_fetch(PL_utf8_X_prepend,
3668 (U8*)locinput, do_utf8))
3670 previous_prepend = locinput;
3671 locinput += UTF8SKIP(locinput);
3674 /* As noted above, if we matched a prepend character, but
3675 * the next thing won't match, back off the last prepend we
3676 * matched, as it is guaranteed to match the begin */
3677 if (previous_prepend
3678 && (locinput >= PL_regeol
3679 || ! swash_fetch(PL_utf8_X_begin,
3680 (U8*)locinput, do_utf8)))
3682 locinput = previous_prepend;
3685 /* Note that here we know PL_regeol > locinput, as we
3686 * tested that upon input to this switch case, and if we
3687 * moved locinput forward, we tested the result just above
3688 * and it either passed, or we backed off so that it will
3690 if (! swash_fetch(PL_utf8_X_begin, (U8*)locinput, do_utf8)) {
3692 /* Here did not match the required 'Begin' in the
3693 * second term. So just match the very first
3694 * character, the '.' of the final term of the regex */
3695 locinput = starting + UTF8SKIP(starting);
3698 /* Here is the beginning of a character that can have
3699 * an extender. It is either a hangul syllable, or a
3701 if (swash_fetch(PL_utf8_X_non_hangul,
3702 (U8*)locinput, do_utf8))
3705 /* Here not a Hangul syllable, must be a
3706 * ('! * Control') */
3707 locinput += UTF8SKIP(locinput);
3710 /* Here is a Hangul syllable. It can be composed
3711 * of several individual characters. One
3712 * possibility is T+ */
3713 if (swash_fetch(PL_utf8_X_T,
3714 (U8*)locinput, do_utf8))
3716 while (locinput < PL_regeol
3717 && swash_fetch(PL_utf8_X_T,
3718 (U8*)locinput, do_utf8))
3720 locinput += UTF8SKIP(locinput);
3724 /* Here, not T+, but is a Hangul. That means
3725 * it is one of the others: L, LV, LVT or V,
3727 * L* (L | LVT T* | V V* T* | LV V* T*) */
3730 while (locinput < PL_regeol
3731 && swash_fetch(PL_utf8_X_L,
3732 (U8*)locinput, do_utf8))
3734 locinput += UTF8SKIP(locinput);
3737 /* Here, have exhausted L*. If the next
3738 * character is not an LV, LVT nor V, it means
3739 * we had to have at least one L, so matches L+
3740 * in the original equation, we have a complete
3741 * hangul syllable. Are done. */
3743 if (locinput < PL_regeol
3744 && swash_fetch(PL_utf8_X_LV_LVT_V,
3745 (U8*)locinput, do_utf8))
3748 /* Otherwise keep going. Must be LV, LVT
3749 * or V. See if LVT */
3750 if (swash_fetch(PL_utf8_X_LVT,
3751 (U8*)locinput, do_utf8))
3753 locinput += UTF8SKIP(locinput);
3756 /* Must be V or LV. Take it, then
3758 locinput += UTF8SKIP(locinput);
3759 while (locinput < PL_regeol
3760 && swash_fetch(PL_utf8_X_V,
3761 (U8*)locinput, do_utf8))
3763 locinput += UTF8SKIP(locinput);
3767 /* And any of LV, LVT, or V can be followed
3769 while (locinput < PL_regeol
3770 && swash_fetch(PL_utf8_X_T,
3774 locinput += UTF8SKIP(locinput);
3780 /* Match any extender */
3781 while (locinput < PL_regeol
3782 && swash_fetch(PL_utf8_X_extend,
3783 (U8*)locinput, do_utf8))
3785 locinput += UTF8SKIP(locinput);
3789 if (locinput > PL_regeol) sayNO;
3791 nextchr = UCHARAT(locinput);
3798 PL_reg_flags |= RF_tainted;
3803 n = reg_check_named_buff_matched(rex,scan);
3806 type = REF + ( type - NREF );
3813 PL_reg_flags |= RF_tainted;
3817 n = ARG(scan); /* which paren pair */
3820 ln = PL_regoffs[n].start;
3821 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
3822 if (*PL_reglastparen < n || ln == -1)
3823 sayNO; /* Do not match unless seen CLOSEn. */
3824 if (ln == PL_regoffs[n].end)
3828 if (do_utf8 && type != REF) { /* REF can do byte comparison */
3830 const char *e = PL_bostr + PL_regoffs[n].end;
3832 * Note that we can't do the "other character" lookup trick as
3833 * in the 8-bit case (no pun intended) because in Unicode we
3834 * have to map both upper and title case to lower case.
3838 STRLEN ulen1, ulen2;
3839 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
3840 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
3844 toLOWER_utf8((U8*)s, tmpbuf1, &ulen1);
3845 toLOWER_utf8((U8*)l, tmpbuf2, &ulen2);
3846 if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
3853 nextchr = UCHARAT(locinput);
3857 /* Inline the first character, for speed. */
3858 if (UCHARAT(s) != nextchr &&
3860 (UCHARAT(s) != (type == REFF
3861 ? PL_fold : PL_fold_locale)[nextchr])))
3863 ln = PL_regoffs[n].end - ln;
3864 if (locinput + ln > PL_regeol)
3866 if (ln > 1 && (type == REF
3867 ? memNE(s, locinput, ln)
3869 ? ibcmp(s, locinput, ln)
3870 : ibcmp_locale(s, locinput, ln))))
3873 nextchr = UCHARAT(locinput);
3883 #define ST st->u.eval
3888 regexp_internal *rei;
3889 regnode *startpoint;
3892 case GOSUB: /* /(...(?1))/ /(...(?&foo))/ */
3893 if (cur_eval && cur_eval->locinput==locinput) {
3894 if (cur_eval->u.eval.close_paren == (U32)ARG(scan))
3895 Perl_croak(aTHX_ "Infinite recursion in regex");
3896 if ( ++nochange_depth > max_nochange_depth )
3898 "Pattern subroutine nesting without pos change"
3899 " exceeded limit in regex");
3906 (void)ReREFCNT_inc(rex_sv);
3907 if (OP(scan)==GOSUB) {
3908 startpoint = scan + ARG2L(scan);
3909 ST.close_paren = ARG(scan);
3911 startpoint = rei->program+1;
3914 goto eval_recurse_doit;
3916 case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */
3917 if (cur_eval && cur_eval->locinput==locinput) {
3918 if ( ++nochange_depth > max_nochange_depth )
3919 Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex");
3924 /* execute the code in the {...} */
3926 SV ** const before = SP;
3927 OP_4tree * const oop = PL_op;
3928 COP * const ocurcop = PL_curcop;
3930 char *saved_regeol = PL_regeol;
3933 PL_op = (OP_4tree*)rexi->data->data[n];
3934 DEBUG_STATE_r( PerlIO_printf(Perl_debug_log,
3935 " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
3936 PAD_SAVE_LOCAL(old_comppad, (PAD*)rexi->data->data[n + 2]);
3937 PL_regoffs[0].end = PL_reg_magic->mg_len = locinput - PL_bostr;
3940 SV *sv_mrk = get_sv("REGMARK", 1);
3941 sv_setsv(sv_mrk, sv_yes_mark);
3944 CALLRUNOPS(aTHX); /* Scalar context. */
3947 ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */
3954 PAD_RESTORE_LOCAL(old_comppad);
3955 PL_curcop = ocurcop;
3956 PL_regeol = saved_regeol;
3959 sv_setsv(save_scalar(PL_replgv), ret);
3963 if (logical == 2) { /* Postponed subexpression: /(??{...})/ */
3966 /* extract RE object from returned value; compiling if
3972 SV *const sv = SvRV(ret);
3974 if (SvTYPE(sv) == SVt_REGEXP) {
3976 } else if (SvSMAGICAL(sv)) {
3977 mg = mg_find(sv, PERL_MAGIC_qr);
3980 } else if (SvTYPE(ret) == SVt_REGEXP) {
3982 } else if (SvSMAGICAL(ret)) {
3983 if (SvGMAGICAL(ret)) {
3984 /* I don't believe that there is ever qr magic
3986 assert(!mg_find(ret, PERL_MAGIC_qr));
3987 sv_unmagic(ret, PERL_MAGIC_qr);
3990 mg = mg_find(ret, PERL_MAGIC_qr);
3991 /* testing suggests mg only ends up non-NULL for
3992 scalars who were upgraded and compiled in the
3993 else block below. In turn, this is only
3994 triggered in the "postponed utf8 string" tests
4000 rx = (REGEXP *) mg->mg_obj; /*XXX:dmq*/
4004 rx = reg_temp_copy(NULL, rx);
4008 const I32 osize = PL_regsize;
4011 assert (SvUTF8(ret));
4012 } else if (SvUTF8(ret)) {
4013 /* Not doing UTF-8, despite what the SV says. Is
4014 this only if we're trapped in use 'bytes'? */
4015 /* Make a copy of the octet sequence, but without
4016 the flag on, as the compiler now honours the
4017 SvUTF8 flag on ret. */
4019 const char *const p = SvPV(ret, len);
4020 ret = newSVpvn_flags(p, len, SVs_TEMP);
4022 rx = CALLREGCOMP(ret, pm_flags);
4024 & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
4026 /* This isn't a first class regexp. Instead, it's
4027 caching a regexp onto an existing, Perl visible
4029 sv_magic(ret, MUTABLE_SV(rx), PERL_MAGIC_qr, 0, 0);
4034 re = (struct regexp *)SvANY(rx);
4036 RXp_MATCH_COPIED_off(re);
4037 re->subbeg = rex->subbeg;
4038 re->sublen = rex->sublen;
4041 debug_start_match(re_sv, do_utf8, locinput, PL_regeol,
4042 "Matching embedded");
4044 startpoint = rei->program + 1;
4045 ST.close_paren = 0; /* only used for GOSUB */
4046 /* borrowed from regtry */
4047 if (PL_reg_start_tmpl <= re->nparens) {
4048 PL_reg_start_tmpl = re->nparens*3/2 + 3;
4049 if(PL_reg_start_tmp)
4050 Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
4052 Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
4055 eval_recurse_doit: /* Share code with GOSUB below this line */
4056 /* run the pattern returned from (??{...}) */
4057 ST.cp = regcppush(0); /* Save *all* the positions. */
4058 REGCP_SET(ST.lastcp);
4060 PL_regoffs = re->offs; /* essentially NOOP on GOSUB */
4062 /* see regtry, specifically PL_reglast(?:close)?paren is a pointer! (i dont know why) :dmq */
4063 PL_reglastparen = &re->lastparen;
4064 PL_reglastcloseparen = &re->lastcloseparen;
4066 re->lastcloseparen = 0;
4068 PL_reginput = locinput;
4071 /* XXXX This is too dramatic a measure... */
4074 ST.toggle_reg_flags = PL_reg_flags;
4076 PL_reg_flags |= RF_utf8;
4078 PL_reg_flags &= ~RF_utf8;
4079 ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */
4081 ST.prev_rex = rex_sv;
4082 ST.prev_curlyx = cur_curlyx;
4083 SETREX(rex_sv,re_sv);
4088 ST.prev_eval = cur_eval;
4090 /* now continue from first node in postoned RE */
4091 PUSH_YES_STATE_GOTO(EVAL_AB, startpoint);
4094 /* logical is 1, /(?(?{...})X|Y)/ */
4095 sw = (bool)SvTRUE(ret);
4100 case EVAL_AB: /* cleanup after a successful (??{A})B */
4101 /* note: this is called twice; first after popping B, then A */
4102 PL_reg_flags ^= ST.toggle_reg_flags;
4103 ReREFCNT_dec(rex_sv);
4104 SETREX(rex_sv,ST.prev_rex);
4105 rex = (struct regexp *)SvANY(rex_sv);
4106 rexi = RXi_GET(rex);
4108 cur_eval = ST.prev_eval;
4109 cur_curlyx = ST.prev_curlyx;
4111 /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */
4112 PL_reglastparen = &rex->lastparen;
4113 PL_reglastcloseparen = &rex->lastcloseparen;
4114 /* also update PL_regoffs */
4115 PL_regoffs = rex->offs;
4117 /* XXXX This is too dramatic a measure... */
4119 if ( nochange_depth )
4124 case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */
4125 /* note: this is called twice; first after popping B, then A */
4126 PL_reg_flags ^= ST.toggle_reg_flags;
4127 ReREFCNT_dec(rex_sv);
4128 SETREX(rex_sv,ST.prev_rex);
4129 rex = (struct regexp *)SvANY(rex_sv);
4130 rexi = RXi_GET(rex);
4131 /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */
4132 PL_reglastparen = &rex->lastparen;
4133 PL_reglastcloseparen = &rex->lastcloseparen;
4135 PL_reginput = locinput;
4136 REGCP_UNWIND(ST.lastcp);
4138 cur_eval = ST.prev_eval;
4139 cur_curlyx = ST.prev_curlyx;
4140 /* XXXX This is too dramatic a measure... */
4142 if ( nochange_depth )
4148 n = ARG(scan); /* which paren pair */
4149 PL_reg_start_tmp[n] = locinput;
4155 n = ARG(scan); /* which paren pair */
4156 PL_regoffs[n].start = PL_reg_start_tmp[n] - PL_bostr;
4157 PL_regoffs[n].end = locinput - PL_bostr;
4158 /*if (n > PL_regsize)
4160 if (n > *PL_reglastparen)
4161 *PL_reglastparen = n;
4162 *PL_reglastcloseparen = n;
4163 if (cur_eval && cur_eval->u.eval.close_paren == n) {
4171 cursor && OP(cursor)!=END;
4172 cursor=regnext(cursor))
4174 if ( OP(cursor)==CLOSE ){
4176 if ( n <= lastopen ) {
4178 = PL_reg_start_tmp[n] - PL_bostr;
4179 PL_regoffs[n].end = locinput - PL_bostr;
4180 /*if (n > PL_regsize)
4182 if (n > *PL_reglastparen)
4183 *PL_reglastparen = n;
4184 *PL_reglastcloseparen = n;
4185 if ( n == ARG(scan) || (cur_eval &&
4186 cur_eval->u.eval.close_paren == n))
4195 n = ARG(scan); /* which paren pair */
4196 sw = (bool)(*PL_reglastparen >= n && PL_regoffs[n].end != -1);
4199 /* reg_check_named_buff_matched returns 0 for no match */
4200 sw = (bool)(0 < reg_check_named_buff_matched(rex,scan));
4204 sw = (cur_eval && (!n || cur_eval->u.eval.close_paren == n));
4210 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
4212 next = NEXTOPER(NEXTOPER(scan));
4214 next = scan + ARG(scan);
4215 if (OP(next) == IFTHEN) /* Fake one. */
4216 next = NEXTOPER(NEXTOPER(next));
4220 logical = scan->flags;
4223 /*******************************************************************
4225 The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/
4226 pattern, where A and B are subpatterns. (For simple A, CURLYM or
4227 STAR/PLUS/CURLY/CURLYN are used instead.)
4229 A*B is compiled as <CURLYX><A><WHILEM><B>
4231 On entry to the subpattern, CURLYX is called. This pushes a CURLYX
4232 state, which contains the current count, initialised to -1. It also sets
4233 cur_curlyx to point to this state, with any previous value saved in the
4236 CURLYX then jumps straight to the WHILEM op, rather than executing A,
4237 since the pattern may possibly match zero times (i.e. it's a while {} loop
4238 rather than a do {} while loop).
4240 Each entry to WHILEM represents a successful match of A. The count in the
4241 CURLYX block is incremented, another WHILEM state is pushed, and execution
4242 passes to A or B depending on greediness and the current count.
4244 For example, if matching against the string a1a2a3b (where the aN are
4245 substrings that match /A/), then the match progresses as follows: (the
4246 pushed states are interspersed with the bits of strings matched so far):
4249 <CURLYX cnt=0><WHILEM>
4250 <CURLYX cnt=1><WHILEM> a1 <WHILEM>
4251 <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM>
4252 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM>
4253 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b
4255 (Contrast this with something like CURLYM, which maintains only a single
4259 a1 <CURLYM cnt=1> a2
4260 a1 a2 <CURLYM cnt=2> a3
4261 a1 a2 a3 <CURLYM cnt=3> b
4264 Each WHILEM state block marks a point to backtrack to upon partial failure
4265 of A or B, and also contains some minor state data related to that
4266 iteration. The CURLYX block, pointed to by cur_curlyx, contains the
4267 overall state, such as the count, and pointers to the A and B ops.
4269 This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx
4270 must always point to the *current* CURLYX block, the rules are:
4272 When executing CURLYX, save the old cur_curlyx in the CURLYX state block,
4273 and set cur_curlyx to point the new block.
4275 When popping the CURLYX block after a successful or unsuccessful match,
4276 restore the previous cur_curlyx.
4278 When WHILEM is about to execute B, save the current cur_curlyx, and set it
4279 to the outer one saved in the CURLYX block.
4281 When popping the WHILEM block after a successful or unsuccessful B match,
4282 restore the previous cur_curlyx.
4284 Here's an example for the pattern (AI* BI)*BO
4285 I and O refer to inner and outer, C and W refer to CURLYX and WHILEM:
4288 curlyx backtrack stack
4289 ------ ---------------
4291 CO <CO prev=NULL> <WO>
4292 CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
4293 CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
4294 NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo
4296 At this point the pattern succeeds, and we work back down the stack to
4297 clean up, restoring as we go:
4299 CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
4300 CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
4301 CO <CO prev=NULL> <WO>
4304 *******************************************************************/
4306 #define ST st->u.curlyx
4308 case CURLYX: /* start of /A*B/ (for complex A) */
4310 /* No need to save/restore up to this paren */
4311 I32 parenfloor = scan->flags;
4313 assert(next); /* keep Coverity happy */
4314 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
4317 /* XXXX Probably it is better to teach regpush to support
4318 parenfloor > PL_regsize... */
4319 if (parenfloor > (I32)*PL_reglastparen)
4320 parenfloor = *PL_reglastparen; /* Pessimization... */
4322 ST.prev_curlyx= cur_curlyx;
4324 ST.cp = PL_savestack_ix;
4326 /* these fields contain the state of the current curly.
4327 * they are accessed by subsequent WHILEMs */
4328 ST.parenfloor = parenfloor;
4329 ST.min = ARG1(scan);
4330 ST.max = ARG2(scan);
4331 ST.A = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
4335 ST.count = -1; /* this will be updated by WHILEM */
4336 ST.lastloc = NULL; /* this will be updated by WHILEM */
4338 PL_reginput = locinput;
4339 PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next));
4343 case CURLYX_end: /* just finished matching all of A*B */
4344 cur_curlyx = ST.prev_curlyx;
4348 case CURLYX_end_fail: /* just failed to match all of A*B */
4350 cur_curlyx = ST.prev_curlyx;
4356 #define ST st->u.whilem
4358 case WHILEM: /* just matched an A in /A*B/ (for complex A) */
4360 /* see the discussion above about CURLYX/WHILEM */
4362 assert(cur_curlyx); /* keep Coverity happy */
4363 n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */
4364 ST.save_lastloc = cur_curlyx->u.curlyx.lastloc;
4365 ST.cache_offset = 0;
4368 PL_reginput = locinput;
4370 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4371 "%*s whilem: matched %ld out of %ld..%ld\n",
4372 REPORT_CODE_OFF+depth*2, "", (long)n,
4373 (long)cur_curlyx->u.curlyx.min,
4374 (long)cur_curlyx->u.curlyx.max)
4377 /* First just match a string of min A's. */
4379 if (n < cur_curlyx->u.curlyx.min) {
4380 cur_curlyx->u.curlyx.lastloc = locinput;
4381 PUSH_STATE_GOTO(WHILEM_A_pre, cur_curlyx->u.curlyx.A);
4385 /* If degenerate A matches "", assume A done. */
4387 if (locinput == cur_curlyx->u.curlyx.lastloc) {
4388 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4389 "%*s whilem: empty match detected, trying continuation...\n",
4390 REPORT_CODE_OFF+depth*2, "")
4392 goto do_whilem_B_max;
4395 /* super-linear cache processing */
4399 if (!PL_reg_maxiter) {
4400 /* start the countdown: Postpone detection until we
4401 * know the match is not *that* much linear. */
4402 PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4);
4403 /* possible overflow for long strings and many CURLYX's */
4404 if (PL_reg_maxiter < 0)
4405 PL_reg_maxiter = I32_MAX;
4406 PL_reg_leftiter = PL_reg_maxiter;
4409 if (PL_reg_leftiter-- == 0) {
4410 /* initialise cache */
4411 const I32 size = (PL_reg_maxiter + 7)/8;
4412 if (PL_reg_poscache) {
4413 if ((I32)PL_reg_poscache_size < size) {
4414 Renew(PL_reg_poscache, size, char);
4415 PL_reg_poscache_size = size;
4417 Zero(PL_reg_poscache, size, char);
4420 PL_reg_poscache_size = size;
4421 Newxz(PL_reg_poscache, size, char);
4423 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4424 "%swhilem: Detected a super-linear match, switching on caching%s...\n",
4425 PL_colors[4], PL_colors[5])
4429 if (PL_reg_leftiter < 0) {
4430 /* have we already failed at this position? */
4432 offset = (scan->flags & 0xf) - 1
4433 + (locinput - PL_bostr) * (scan->flags>>4);
4434 mask = 1 << (offset % 8);
4436 if (PL_reg_poscache[offset] & mask) {
4437 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4438 "%*s whilem: (cache) already tried at this position...\n",
4439 REPORT_CODE_OFF+depth*2, "")
4441 sayNO; /* cache records failure */
4443 ST.cache_offset = offset;
4444 ST.cache_mask = mask;
4448 /* Prefer B over A for minimal matching. */
4450 if (cur_curlyx->u.curlyx.minmod) {
4451 ST.save_curlyx = cur_curlyx;
4452 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
4453 ST.cp = regcppush(ST.save_curlyx->u.curlyx.parenfloor);
4454 REGCP_SET(ST.lastcp);
4455 PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B);
4459 /* Prefer A over B for maximal matching. */
4461 if (n < cur_curlyx->u.curlyx.max) { /* More greed allowed? */
4462 ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor);
4463 cur_curlyx->u.curlyx.lastloc = locinput;
4464 REGCP_SET(ST.lastcp);
4465 PUSH_STATE_GOTO(WHILEM_A_max, cur_curlyx->u.curlyx.A);
4468 goto do_whilem_B_max;
4472 case WHILEM_B_min: /* just matched B in a minimal match */
4473 case WHILEM_B_max: /* just matched B in a maximal match */
4474 cur_curlyx = ST.save_curlyx;
4478 case WHILEM_B_max_fail: /* just failed to match B in a maximal match */
4479 cur_curlyx = ST.save_curlyx;
4480 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
4481 cur_curlyx->u.curlyx.count--;
4485 case WHILEM_A_min_fail: /* just failed to match A in a minimal match */
4486 REGCP_UNWIND(ST.lastcp);
4489 case WHILEM_A_pre_fail: /* just failed to match even minimal A */
4490 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
4491 cur_curlyx->u.curlyx.count--;
4495 case WHILEM_A_max_fail: /* just failed to match A in a maximal match */
4496 REGCP_UNWIND(ST.lastcp);
4497 regcppop(rex); /* Restore some previous $<digit>s? */
4498 PL_reginput = locinput;
4499 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
4500 "%*s whilem: failed, trying continuation...\n",
4501 REPORT_CODE_OFF+depth*2, "")
4504 if (cur_curlyx->u.curlyx.count >= REG_INFTY
4505 && ckWARN(WARN_REGEXP)
4506 && !(PL_reg_flags & RF_warned))
4508 PL_reg_flags |= RF_warned;
4509 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded",
4510 "Complex regular subexpression recursion",
4515 ST.save_curlyx = cur_curlyx;
4516 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
4517 PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B);
4520 case WHILEM_B_min_fail: /* just failed to match B in a minimal match */
4521 cur_curlyx = ST.save_curlyx;
4522 REGCP_UNWIND(ST.lastcp);
4525 if (cur_curlyx->u.curlyx.count >= cur_curlyx->u.curlyx.max) {
4526 /* Maximum greed exceeded */
4527 if (cur_curlyx->u.curlyx.count >= REG_INFTY
4528 && ckWARN(WARN_REGEXP)
4529 && !(PL_reg_flags & RF_warned))
4531 PL_reg_flags |= RF_warned;
4532 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
4533 "%s limit (%d) exceeded",
4534 "Complex regular subexpression recursion",
4537 cur_curlyx->u.curlyx.count--;
4541 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
4542 "%*s trying longer...\n", REPORT_CODE_OFF+depth*2, "")
4544 /* Try grabbing another A and see if it helps. */
4545 PL_reginput = locinput;
4546 cur_curlyx->u.curlyx.lastloc = locinput;
4547 ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor);
4548 REGCP_SET(ST.lastcp);
4549 PUSH_STATE_GOTO(WHILEM_A_min, ST.save_curlyx->u.curlyx.A);
4553 #define ST st->u.branch
4555 case BRANCHJ: /* /(...|A|...)/ with long next pointer */
4556 next = scan + ARG(scan);
4559 scan = NEXTOPER(scan);
4562 case BRANCH: /* /(...|A|...)/ */
4563 scan = NEXTOPER(scan); /* scan now points to inner node */
4564 ST.lastparen = *PL_reglastparen;
4565 ST.next_branch = next;
4567 PL_reginput = locinput;
4569 /* Now go into the branch */
4571 PUSH_YES_STATE_GOTO(BRANCH_next, scan);
4573 PUSH_STATE_GOTO(BRANCH_next, scan);
4577 PL_reginput = locinput;
4578 sv_yes_mark = st->u.mark.mark_name = scan->flags ? NULL :
4579 MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
4580 PUSH_STATE_GOTO(CUTGROUP_next,next);
4582 case CUTGROUP_next_fail:
4585 if (st->u.mark.mark_name)
4586 sv_commit = st->u.mark.mark_name;
4592 case BRANCH_next_fail: /* that branch failed; try the next, if any */
4597 REGCP_UNWIND(ST.cp);
4598 for (n = *PL_reglastparen; n > ST.lastparen; n--)
4599 PL_regoffs[n].end = -1;
4600 *PL_reglastparen = n;
4601 /*dmq: *PL_reglastcloseparen = n; */
4602 scan = ST.next_branch;
4603 /* no more branches? */
4604 if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) {
4606 PerlIO_printf( Perl_debug_log,
4607 "%*s %sBRANCH failed...%s\n",
4608 REPORT_CODE_OFF+depth*2, "",
4614 continue; /* execute next BRANCH[J] op */
4622 #define ST st->u.curlym
4624 case CURLYM: /* /A{m,n}B/ where A is fixed-length */
4626 /* This is an optimisation of CURLYX that enables us to push
4627 * only a single backtracking state, no matter how many matches
4628 * there are in {m,n}. It relies on the pattern being constant
4629 * length, with no parens to influence future backrefs
4633 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
4635 /* if paren positive, emulate an OPEN/CLOSE around A */
4637 U32 paren = ST.me->flags;
4638 if (paren > PL_regsize)
4640 if (paren > *PL_reglastparen)
4641 *PL_reglastparen = paren;
4642 scan += NEXT_OFF(scan); /* Skip former OPEN. */
4650 ST.c1 = CHRTEST_UNINIT;
4653 if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */
4656 curlym_do_A: /* execute the A in /A{m,n}B/ */
4657 PL_reginput = locinput;
4658 PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */
4661 case CURLYM_A: /* we've just matched an A */
4662 locinput = st->locinput;
4663 nextchr = UCHARAT(locinput);
4666 /* after first match, determine A's length: u.curlym.alen */
4667 if (ST.count == 1) {
4668 if (PL_reg_match_utf8) {
4670 while (s < PL_reginput) {
4676 ST.alen = PL_reginput - locinput;
4679 ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me);
4682 PerlIO_printf(Perl_debug_log,
4683 "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n",
4684 (int)(REPORT_CODE_OFF+(depth*2)), "",
4685 (IV) ST.count, (IV)ST.alen)
4688 locinput = PL_reginput;
4690 if (cur_eval && cur_eval->u.eval.close_paren &&
4691 cur_eval->u.eval.close_paren == (U32)ST.me->flags)
4695 I32 max = (ST.minmod ? ARG1(ST.me) : ARG2(ST.me));
4696 if ( max == REG_INFTY || ST.count < max )
4697 goto curlym_do_A; /* try to match another A */
4699 goto curlym_do_B; /* try to match B */
4701 case CURLYM_A_fail: /* just failed to match an A */
4702 REGCP_UNWIND(ST.cp);
4704 if (ST.minmod || ST.count < ARG1(ST.me) /* min*/
4705 || (cur_eval && cur_eval->u.eval.close_paren &&
4706 cur_eval->u.eval.close_paren == (U32)ST.me->flags))
4709 curlym_do_B: /* execute the B in /A{m,n}B/ */
4710 PL_reginput = locinput;
4711 if (ST.c1 == CHRTEST_UNINIT) {
4712 /* calculate c1 and c2 for possible match of 1st char
4713 * following curly */
4714 ST.c1 = ST.c2 = CHRTEST_VOID;
4715 if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) {
4716 regnode *text_node = ST.B;
4717 if (! HAS_TEXT(text_node))
4718 FIND_NEXT_IMPT(text_node);
4721 (HAS_TEXT(text_node) && PL_regkind[OP(text_node)] == EXACT)
4723 But the former is redundant in light of the latter.
4725 if this changes back then the macro for
4726 IS_TEXT and friends need to change.
4728 if (PL_regkind[OP(text_node)] == EXACT)
4731 ST.c1 = (U8)*STRING(text_node);
4733 (IS_TEXTF(text_node))
4735 : (IS_TEXTFL(text_node))
4736 ? PL_fold_locale[ST.c1]
4743 PerlIO_printf(Perl_debug_log,
4744 "%*s CURLYM trying tail with matches=%"IVdf"...\n",
4745 (int)(REPORT_CODE_OFF+(depth*2)),
4748 if (ST.c1 != CHRTEST_VOID
4749 && UCHARAT(PL_reginput) != ST.c1
4750 && UCHARAT(PL_reginput) != ST.c2)
4752 /* simulate B failing */
4754 PerlIO_printf(Perl_debug_log,
4755 "%*s CURLYM Fast bail c1=%"IVdf" c2=%"IVdf"\n",
4756 (int)(REPORT_CODE_OFF+(depth*2)),"",
4759 state_num = CURLYM_B_fail;
4760 goto reenter_switch;
4764 /* mark current A as captured */
4765 I32 paren = ST.me->flags;
4767 PL_regoffs[paren].start
4768 = HOPc(PL_reginput, -ST.alen) - PL_bostr;
4769 PL_regoffs[paren].end = PL_reginput - PL_bostr;
4770 /*dmq: *PL_reglastcloseparen = paren; */
4773 PL_regoffs[paren].end = -1;
4774 if (cur_eval && cur_eval->u.eval.close_paren &&
4775 cur_eval->u.eval.close_paren == (U32)ST.me->flags)
4784 PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */
4787 case CURLYM_B_fail: /* just failed to match a B */
4788 REGCP_UNWIND(ST.cp);
4790 I32 max = ARG2(ST.me);
4791 if (max != REG_INFTY && ST.count == max)
4793 goto curlym_do_A; /* try to match a further A */
4795 /* backtrack one A */
4796 if (ST.count == ARG1(ST.me) /* min */)
4799 locinput = HOPc(locinput, -ST.alen);
4800 goto curlym_do_B; /* try to match B */
4803 #define ST st->u.curly
4805 #define CURLY_SETPAREN(paren, success) \
4808 PL_regoffs[paren].start = HOPc(locinput, -1) - PL_bostr; \
4809 PL_regoffs[paren].end = locinput - PL_bostr; \
4810 *PL_reglastcloseparen = paren; \
4813 PL_regoffs[paren].end = -1; \
4816 case STAR: /* /A*B/ where A is width 1 */
4820 scan = NEXTOPER(scan);
4822 case PLUS: /* /A+B/ where A is width 1 */
4826 scan = NEXTOPER(scan);
4828 case CURLYN: /* /(A){m,n}B/ where A is width 1 */
4829 ST.paren = scan->flags; /* Which paren to set */
4830 if (ST.paren > PL_regsize)
4831 PL_regsize = ST.paren;
4832 if (ST.paren > *PL_reglastparen)
4833 *PL_reglastparen = ST.paren;
4834 ST.min = ARG1(scan); /* min to match */
4835 ST.max = ARG2(scan); /* max to match */
4836 if (cur_eval && cur_eval->u.eval.close_paren &&
4837 cur_eval->u.eval.close_paren == (U32)ST.paren) {
4841 scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE);
4843 case CURLY: /* /A{m,n}B/ where A is width 1 */
4845 ST.min = ARG1(scan); /* min to match */
4846 ST.max = ARG2(scan); /* max to match */
4847 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
4850 * Lookahead to avoid useless match attempts
4851 * when we know what character comes next.
4853 * Used to only do .*x and .*?x, but now it allows
4854 * for )'s, ('s and (?{ ... })'s to be in the way
4855 * of the quantifier and the EXACT-like node. -- japhy
4858 if (ST.min > ST.max) /* XXX make this a compile-time check? */
4860 if (HAS_TEXT(next) || JUMPABLE(next)) {
4862 regnode *text_node = next;
4864 if (! HAS_TEXT(text_node))
4865 FIND_NEXT_IMPT(text_node);
4867 if (! HAS_TEXT(text_node))
4868 ST.c1 = ST.c2 = CHRTEST_VOID;
4870 if ( PL_regkind[OP(text_node)] != EXACT ) {
4871 ST.c1 = ST.c2 = CHRTEST_VOID;
4872 goto assume_ok_easy;
4875 s = (U8*)STRING(text_node);
4877 /* Currently we only get here when
4879 PL_rekind[OP(text_node)] == EXACT
4881 if this changes back then the macro for IS_TEXT and
4882 friends need to change. */
4885 if (IS_TEXTF(text_node))
4886 ST.c2 = PL_fold[ST.c1];
4887 else if (IS_TEXTFL(text_node))
4888 ST.c2 = PL_fold_locale[ST.c1];
4891 if (IS_TEXTF(text_node)) {
4892 STRLEN ulen1, ulen2;
4893 U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
4894 U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
4896 to_utf8_lower((U8*)s, tmpbuf1, &ulen1);
4897 to_utf8_upper((U8*)s, tmpbuf2, &ulen2);
4899 ST.c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN, 0,
4901 0 : UTF8_ALLOW_ANY);
4902 ST.c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN, 0,
4904 0 : UTF8_ALLOW_ANY);
4906 ST.c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0,
4908 ST.c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0,
4913 ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0,
4920 ST.c1 = ST.c2 = CHRTEST_VOID;
4925 PL_reginput = locinput;
4928 if (ST.min && regrepeat(rex, ST.A, ST.min, depth) < ST.min)
4931 locinput = PL_reginput;
4933 if (ST.c1 == CHRTEST_VOID)
4934 goto curly_try_B_min;
4936 ST.oldloc = locinput;
4938 /* set ST.maxpos to the furthest point along the
4939 * string that could possibly match */
4940 if (ST.max == REG_INFTY) {
4941 ST.maxpos = PL_regeol - 1;
4943 while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos))
4947 int m = ST.max - ST.min;
4948 for (ST.maxpos = locinput;
4949 m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--)
4950 ST.maxpos += UTF8SKIP(ST.maxpos);
4953 ST.maxpos = locinput + ST.max - ST.min;
4954 if (ST.maxpos >= PL_regeol)
4955 ST.maxpos = PL_regeol - 1;
4957 goto curly_try_B_min_known;
4961 ST.count = regrepeat(rex, ST.A, ST.max, depth);
4962 locinput = PL_reginput;
4963 if (ST.count < ST.min)
4965 if ((ST.count > ST.min)
4966 && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL))
4968 /* A{m,n} must come at the end of the string, there's
4969 * no point in backing off ... */
4971 /* ...except that $ and \Z can match before *and* after
4972 newline at the end. Consider "\n\n" =~ /\n+\Z\n/.
4973 We may back off by one in this case. */
4974 if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS)
4978 goto curly_try_B_max;
4983 case CURLY_B_min_known_fail:
4984 /* failed to find B in a non-greedy match where c1,c2 valid */
4985 if (ST.paren && ST.count)
4986 PL_regoffs[ST.paren].end = -1;
4988 PL_reginput = locinput; /* Could be reset... */
4989 REGCP_UNWIND(ST.cp);
4990 /* Couldn't or didn't -- move forward. */
4991 ST.oldloc = locinput;
4993 locinput += UTF8SKIP(locinput);
4997 curly_try_B_min_known:
4998 /* find the next place where 'B' could work, then call B */
5002 n = (ST.oldloc == locinput) ? 0 : 1;
5003 if (ST.c1 == ST.c2) {
5005 /* set n to utf8_distance(oldloc, locinput) */
5006 while (locinput <= ST.maxpos &&
5007 utf8n_to_uvchr((U8*)locinput,
5008 UTF8_MAXBYTES, &len,
5009 uniflags) != (UV)ST.c1) {
5015 /* set n to utf8_distance(oldloc, locinput) */
5016 while (locinput <= ST.maxpos) {
5018 const UV c = utf8n_to_uvchr((U8*)locinput,
5019 UTF8_MAXBYTES, &len,
5021 if (c == (UV)ST.c1 || c == (UV)ST.c2)
5029 if (ST.c1 == ST.c2) {
5030 while (locinput <= ST.maxpos &&
5031 UCHARAT(locinput) != ST.c1)
5035 while (locinput <= ST.maxpos
5036 && UCHARAT(locinput) != ST.c1
5037 && UCHARAT(locinput) != ST.c2)
5040 n = locinput - ST.oldloc;
5042 if (locinput > ST.maxpos)
5044 /* PL_reginput == oldloc now */
5047 if (regrepeat(rex, ST.A, n, depth) < n)
5050 PL_reginput = locinput;
5051 CURLY_SETPAREN(ST.paren, ST.count);
5052 if (cur_eval && cur_eval->u.eval.close_paren &&
5053 cur_eval->u.eval.close_paren == (U32)ST.paren) {
5056 PUSH_STATE_GOTO(CURLY_B_min_known, ST.B);
5061 case CURLY_B_min_fail:
5062 /* failed to find B in a non-greedy match where c1,c2 invalid */
5063 if (ST.paren && ST.count)
5064 PL_regoffs[ST.paren].end = -1;
5066 REGCP_UNWIND(ST.cp);
5067 /* failed -- move forward one */
5068 PL_reginput = locinput;
5069 if (regrepeat(rex, ST.A, 1, depth)) {
5071 locinput = PL_reginput;
5072 if (ST.count <= ST.max || (ST.max == REG_INFTY &&
5073 ST.count > 0)) /* count overflow ? */
5076 CURLY_SETPAREN(ST.paren, ST.count);
5077 if (cur_eval && cur_eval->u.eval.close_paren &&
5078 cur_eval->u.eval.close_paren == (U32)ST.paren) {
5081 PUSH_STATE_GOTO(CURLY_B_min, ST.B);
5089 /* a successful greedy match: now try to match B */
5090 if (cur_eval && cur_eval->u.eval.close_paren &&
5091 cur_eval->u.eval.close_paren == (U32)ST.paren) {
5096 if (ST.c1 != CHRTEST_VOID)
5097 c = do_utf8 ? utf8n_to_uvchr((U8*)PL_reginput,
5098 UTF8_MAXBYTES, 0, uniflags)
5099 : (UV) UCHARAT(PL_reginput);
5100 /* If it could work, try it. */
5101 if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) {
5102 CURLY_SETPAREN(ST.paren, ST.count);
5103 PUSH_STATE_GOTO(CURLY_B_max, ST.B);
5108 case CURLY_B_max_fail:
5109 /* failed to find B in a greedy match */
5110 if (ST.paren && ST.count)
5111 PL_regoffs[ST.paren].end = -1;
5113 REGCP_UNWIND(ST.cp);
5115 if (--ST.count < ST.min)
5117 PL_reginput = locinput = HOPc(locinput, -1);
5118 goto curly_try_B_max;
5125 /* we've just finished A in /(??{A})B/; now continue with B */
5127 st->u.eval.toggle_reg_flags
5128 = cur_eval->u.eval.toggle_reg_flags;
5129 PL_reg_flags ^= st->u.eval.toggle_reg_flags;
5131 st->u.eval.prev_rex = rex_sv; /* inner */
5132 SETREX(rex_sv,cur_eval->u.eval.prev_rex);
5133 rex = (struct regexp *)SvANY(rex_sv);
5134 rexi = RXi_GET(rex);
5135 cur_curlyx = cur_eval->u.eval.prev_curlyx;
5136 ReREFCNT_inc(rex_sv);
5137 st->u.eval.cp = regcppush(0); /* Save *all* the positions. */
5139 /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */
5140 PL_reglastparen = &rex->lastparen;
5141 PL_reglastcloseparen = &rex->lastcloseparen;
5143 REGCP_SET(st->u.eval.lastcp);
5144 PL_reginput = locinput;
5146 /* Restore parens of the outer rex without popping the
5148 tmpix = PL_savestack_ix;
5149 PL_savestack_ix = cur_eval->u.eval.lastcp;
5151 PL_savestack_ix = tmpix;
5153 st->u.eval.prev_eval = cur_eval;
5154 cur_eval = cur_eval->u.eval.prev_eval;
5156 PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ... %"UVxf"\n",
5157 REPORT_CODE_OFF+depth*2, "",PTR2UV(cur_eval)););
5158 if ( nochange_depth )
5161 PUSH_YES_STATE_GOTO(EVAL_AB,
5162 st->u.eval.prev_eval->u.eval.B); /* match B */
5165 if (locinput < reginfo->till) {
5166 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
5167 "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n",
5169 (long)(locinput - PL_reg_starttry),
5170 (long)(reginfo->till - PL_reg_starttry),
5173 sayNO_SILENT; /* Cannot match: too short. */
5175 PL_reginput = locinput; /* put where regtry can find it */
5176 sayYES; /* Success! */
5178 case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */
5180 PerlIO_printf(Perl_debug_log,
5181 "%*s %ssubpattern success...%s\n",
5182 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]));
5183 PL_reginput = locinput; /* put where regtry can find it */
5184 sayYES; /* Success! */
5187 #define ST st->u.ifmatch
5189 case SUSPEND: /* (?>A) */
5191 PL_reginput = locinput;
5194 case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */
5196 goto ifmatch_trivial_fail_test;
5198 case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */
5200 ifmatch_trivial_fail_test:
5202 char * const s = HOPBACKc(locinput, scan->flags);
5207 sw = 1 - (bool)ST.wanted;
5211 next = scan + ARG(scan);
5219 PL_reginput = locinput;
5223 ST.logical = logical;
5224 logical = 0; /* XXX: reset state of logical once it has been saved into ST */
5226 /* execute body of (?...A) */
5227 PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan)));
5230 case IFMATCH_A_fail: /* body of (?...A) failed */
5231 ST.wanted = !ST.wanted;
5234 case IFMATCH_A: /* body of (?...A) succeeded */
5236 sw = (bool)ST.wanted;
5238 else if (!ST.wanted)
5241 if (OP(ST.me) == SUSPEND)
5242 locinput = PL_reginput;
5244 locinput = PL_reginput = st->locinput;
5245 nextchr = UCHARAT(locinput);
5247 scan = ST.me + ARG(ST.me);
5250 continue; /* execute B */
5255 next = scan + ARG(scan);
5260 reginfo->cutpoint = PL_regeol;
5263 PL_reginput = locinput;
5265 sv_yes_mark = sv_commit = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
5266 PUSH_STATE_GOTO(COMMIT_next,next);
5268 case COMMIT_next_fail:
5275 #define ST st->u.mark
5277 ST.prev_mark = mark_state;
5278 ST.mark_name = sv_commit = sv_yes_mark
5279 = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
5281 ST.mark_loc = PL_reginput = locinput;
5282 PUSH_YES_STATE_GOTO(MARKPOINT_next,next);
5284 case MARKPOINT_next:
5285 mark_state = ST.prev_mark;
5288 case MARKPOINT_next_fail:
5289 if (popmark && sv_eq(ST.mark_name,popmark))
5291 if (ST.mark_loc > startpoint)
5292 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
5293 popmark = NULL; /* we found our mark */
5294 sv_commit = ST.mark_name;
5297 PerlIO_printf(Perl_debug_log,
5298 "%*s %ssetting cutpoint to mark:%"SVf"...%s\n",
5299 REPORT_CODE_OFF+depth*2, "",
5300 PL_colors[4], SVfARG(sv_commit), PL_colors[5]);
5303 mark_state = ST.prev_mark;
5304 sv_yes_mark = mark_state ?
5305 mark_state->u.mark.mark_name : NULL;
5309 PL_reginput = locinput;
5311 /* (*SKIP) : if we fail we cut here*/
5312 ST.mark_name = NULL;
5313 ST.mark_loc = locinput;
5314 PUSH_STATE_GOTO(SKIP_next,next);
5316 /* (*SKIP:NAME) : if there is a (*MARK:NAME) fail where it was,
5317 otherwise do nothing. Meaning we need to scan
5319 regmatch_state *cur = mark_state;
5320 SV *find = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
5323 if ( sv_eq( cur->u.mark.mark_name,
5326 ST.mark_name = find;
5327 PUSH_STATE_GOTO( SKIP_next, next );
5329 cur = cur->u.mark.prev_mark;
5332 /* Didn't find our (*MARK:NAME) so ignore this (*SKIP:NAME) */
5334 case SKIP_next_fail:
5336 /* (*CUT:NAME) - Set up to search for the name as we
5337 collapse the stack*/
5338 popmark = ST.mark_name;
5340 /* (*CUT) - No name, we cut here.*/
5341 if (ST.mark_loc > startpoint)
5342 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
5343 /* but we set sv_commit to latest mark_name if there
5344 is one so they can test to see how things lead to this
5347 sv_commit=mark_state->u.mark.mark_name;
5355 if ( n == (U32)what_len_TRICKYFOLD(locinput,do_utf8,ln) ) {
5357 } else if ( 0xDF == n && !do_utf8 && !UTF ) {
5360 U8 folded[UTF8_MAXBYTES_CASE+1];
5362 const char * const l = locinput;
5363 char *e = PL_regeol;
5364 to_uni_fold(n, folded, &foldlen);
5366 if (ibcmp_utf8((const char*) folded, 0, foldlen, 1,
5367 l, &e, 0, do_utf8)) {
5372 nextchr = UCHARAT(locinput);
5375 if ((n=is_LNBREAK(locinput,do_utf8))) {
5377 nextchr = UCHARAT(locinput);
5382 #define CASE_CLASS(nAmE) \
5384 if ((n=is_##nAmE(locinput,do_utf8))) { \
5386 nextchr = UCHARAT(locinput); \
5391 if ((n=is_##nAmE(locinput,do_utf8))) { \
5394 locinput += UTF8SKIP(locinput); \
5395 nextchr = UCHARAT(locinput); \
5400 CASE_CLASS(HORIZWS);
5404 PerlIO_printf(Perl_error_log, "%"UVxf" %d\n",
5405 PTR2UV(scan), OP(scan));
5406 Perl_croak(aTHX_ "regexp memory corruption");
5410 /* switch break jumps here */
5411 scan = next; /* prepare to execute the next op and ... */
5412 continue; /* ... jump back to the top, reusing st */
5416 /* push a state that backtracks on success */
5417 st->u.yes.prev_yes_state = yes_state;
5421 /* push a new regex state, then continue at scan */
5423 regmatch_state *newst;
5426 regmatch_state *cur = st;
5427 regmatch_state *curyes = yes_state;
5429 regmatch_slab *slab = PL_regmatch_slab;
5430 for (;curd > -1;cur--,curd--) {
5431 if (cur < SLAB_FIRST(slab)) {
5433 cur = SLAB_LAST(slab);
5435 PerlIO_printf(Perl_error_log, "%*s#%-3d %-10s %s\n",
5436 REPORT_CODE_OFF + 2 + depth * 2,"",
5437 curd, PL_reg_name[cur->resume_state],
5438 (curyes == cur) ? "yes" : ""
5441 curyes = cur->u.yes.prev_yes_state;
5444 DEBUG_STATE_pp("push")
5447 st->locinput = locinput;
5449 if (newst > SLAB_LAST(PL_regmatch_slab))
5450 newst = S_push_slab(aTHX);
5451 PL_regmatch_state = newst;
5453 locinput = PL_reginput;
5454 nextchr = UCHARAT(locinput);
5462 * We get here only if there's trouble -- normally "case END" is
5463 * the terminating point.
5465 Perl_croak(aTHX_ "corrupted regexp pointers");
5471 /* we have successfully completed a subexpression, but we must now
5472 * pop to the state marked by yes_state and continue from there */
5473 assert(st != yes_state);
5475 while (st != yes_state) {
5477 if (st < SLAB_FIRST(PL_regmatch_slab)) {
5478 PL_regmatch_slab = PL_regmatch_slab->prev;
5479 st = SLAB_LAST(PL_regmatch_slab);
5483 DEBUG_STATE_pp("pop (no final)");
5485 DEBUG_STATE_pp("pop (yes)");
5491 while (yes_state < SLAB_FIRST(PL_regmatch_slab)
5492 || yes_state > SLAB_LAST(PL_regmatch_slab))
5494 /* not in this slab, pop slab */
5495 depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1);
5496 PL_regmatch_slab = PL_regmatch_slab->prev;
5497 st = SLAB_LAST(PL_regmatch_slab);
5499 depth -= (st - yes_state);
5502 yes_state = st->u.yes.prev_yes_state;
5503 PL_regmatch_state = st;
5506 locinput= st->locinput;
5507 nextchr = UCHARAT(locinput);
5509 state_num = st->resume_state + no_final;
5510 goto reenter_switch;
5513 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n",
5514 PL_colors[4], PL_colors[5]));
5516 if (PL_reg_eval_set) {
5517 /* each successfully executed (?{...}) block does the equivalent of
5518 * local $^R = do {...}
5519 * When popping the save stack, all these locals would be undone;
5520 * bypass this by setting the outermost saved $^R to the latest
5522 if (oreplsv != GvSV(PL_replgv))
5523 sv_setsv(oreplsv, GvSV(PL_replgv));
5530 PerlIO_printf(Perl_debug_log,
5531 "%*s %sfailed...%s\n",
5532 REPORT_CODE_OFF+depth*2, "",
5533 PL_colors[4], PL_colors[5])
5545 /* there's a previous state to backtrack to */
5547 if (st < SLAB_FIRST(PL_regmatch_slab)) {
5548 PL_regmatch_slab = PL_regmatch_slab->prev;
5549 st = SLAB_LAST(PL_regmatch_slab);
5551 PL_regmatch_state = st;
5552 locinput= st->locinput;
5553 nextchr = UCHARAT(locinput);
5555 DEBUG_STATE_pp("pop");
5557 if (yes_state == st)
5558 yes_state = st->u.yes.prev_yes_state;
5560 state_num = st->resume_state + 1; /* failure = success + 1 */
5561 goto reenter_switch;
5566 if (rex->intflags & PREGf_VERBARG_SEEN) {
5567 SV *sv_err = get_sv("REGERROR", 1);
5568 SV *sv_mrk = get_sv("REGMARK", 1);
5570 sv_commit = &PL_sv_no;
5572 sv_yes_mark = &PL_sv_yes;
5575 sv_commit = &PL_sv_yes;
5576 sv_yes_mark = &PL_sv_no;
5578 sv_setsv(sv_err, sv_commit);
5579 sv_setsv(sv_mrk, sv_yes_mark);
5582 /* clean up; in particular, free all slabs above current one */
5583 LEAVE_SCOPE(oldsave);
5589 - regrepeat - repeatedly match something simple, report how many
5592 * [This routine now assumes that it will only match on things of length 1.
5593 * That was true before, but now we assume scan - reginput is the count,
5594 * rather than incrementing count on every character. [Er, except utf8.]]
5597 S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max, int depth)
5600 register char *scan;
5602 register char *loceol = PL_regeol;
5603 register I32 hardcount = 0;
5604 register bool do_utf8 = PL_reg_match_utf8;
5606 PERL_UNUSED_ARG(depth);
5609 PERL_ARGS_ASSERT_REGREPEAT;
5612 if (max == REG_INFTY)
5614 else if (max < loceol - scan)
5615 loceol = scan + max;
5620 while (scan < loceol && hardcount < max && *scan != '\n') {
5621 scan += UTF8SKIP(scan);
5625 while (scan < loceol && *scan != '\n')
5632 while (scan < loceol && hardcount < max) {
5633 scan += UTF8SKIP(scan);
5643 case EXACT: /* length of string is 1 */
5645 while (scan < loceol && UCHARAT(scan) == c)
5648 case EXACTF: /* length of string is 1 */
5650 while (scan < loceol &&
5651 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c]))
5654 case EXACTFL: /* length of string is 1 */
5655 PL_reg_flags |= RF_tainted;
5657 while (scan < loceol &&
5658 (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c]))
5664 while (hardcount < max && scan < loceol &&
5665 reginclass(prog, p, (U8*)scan, 0, do_utf8)) {
5666 scan += UTF8SKIP(scan);
5670 while (scan < loceol && REGINCLASS(prog, p, (U8*)scan))
5677 LOAD_UTF8_CHARCLASS_ALNUM();
5678 while (hardcount < max && scan < loceol &&
5679 swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) {
5680 scan += UTF8SKIP(scan);
5684 while (scan < loceol && isALNUM(*scan))
5689 PL_reg_flags |= RF_tainted;
5692 while (hardcount < max && scan < loceol &&
5693 isALNUM_LC_utf8((U8*)scan)) {
5694 scan += UTF8SKIP(scan);
5698 while (scan < loceol && isALNUM_LC(*scan))
5705 LOAD_UTF8_CHARCLASS_ALNUM();
5706 while (hardcount < max && scan < loceol &&
5707 !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) {
5708 scan += UTF8SKIP(scan);
5712 while (scan < loceol && !isALNUM(*scan))
5717 PL_reg_flags |= RF_tainted;
5720 while (hardcount < max && scan < loceol &&
5721 !isALNUM_LC_utf8((U8*)scan)) {
5722 scan += UTF8SKIP(scan);
5726 while (scan < loceol && !isALNUM_LC(*scan))
5733 LOAD_UTF8_CHARCLASS_SPACE();
5734 while (hardcount < max && scan < loceol &&
5736 swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) {
5737 scan += UTF8SKIP(scan);
5741 while (scan < loceol && isSPACE(*scan))
5746 PL_reg_flags |= RF_tainted;
5749 while (hardcount < max && scan < loceol &&
5750 (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
5751 scan += UTF8SKIP(scan);
5755 while (scan < loceol && isSPACE_LC(*scan))
5762 LOAD_UTF8_CHARCLASS_SPACE();
5763 while (hardcount < max && scan < loceol &&
5765 swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) {
5766 scan += UTF8SKIP(scan);
5770 while (scan < loceol && !isSPACE(*scan))
5775 PL_reg_flags |= RF_tainted;
5778 while (hardcount < max && scan < loceol &&
5779 !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
5780 scan += UTF8SKIP(scan);
5784 while (scan < loceol && !isSPACE_LC(*scan))
5791 LOAD_UTF8_CHARCLASS_DIGIT();
5792 while (hardcount < max && scan < loceol &&
5793 swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) {
5794 scan += UTF8SKIP(scan);
5798 while (scan < loceol && isDIGIT(*scan))
5805 LOAD_UTF8_CHARCLASS_DIGIT();
5806 while (hardcount < max && scan < loceol &&
5807 !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) {
5808 scan += UTF8SKIP(scan);
5812 while (scan < loceol && !isDIGIT(*scan))
5818 while (hardcount < max && scan < loceol && (c=is_LNBREAK_utf8(scan))) {
5824 LNBREAK can match two latin chars, which is ok,
5825 because we have a null terminated string, but we
5826 have to use hardcount in this situation
5828 while (scan < loceol && (c=is_LNBREAK_latin1(scan))) {
5837 while (hardcount < max && scan < loceol && (c=is_HORIZWS_utf8(scan))) {
5842 while (scan < loceol && is_HORIZWS_latin1(scan))
5849 while (hardcount < max && scan < loceol && !is_HORIZWS_utf8(scan)) {
5850 scan += UTF8SKIP(scan);
5854 while (scan < loceol && !is_HORIZWS_latin1(scan))
5862 while (hardcount < max && scan < loceol && (c=is_VERTWS_utf8(scan))) {
5867 while (scan < loceol && is_VERTWS_latin1(scan))
5875 while (hardcount < max && scan < loceol && !is_VERTWS_utf8(scan)) {
5876 scan += UTF8SKIP(scan);
5880 while (scan < loceol && !is_VERTWS_latin1(scan))
5886 default: /* Called on something of 0 width. */
5887 break; /* So match right here or not at all. */
5893 c = scan - PL_reginput;
5897 GET_RE_DEBUG_FLAGS_DECL;
5899 SV * const prop = sv_newmortal();
5900 regprop(prog, prop, p);
5901 PerlIO_printf(Perl_debug_log,
5902 "%*s %s can match %"IVdf" times out of %"IVdf"...\n",
5903 REPORT_CODE_OFF + depth*2, "", SvPVX_const(prop),(IV)c,(IV)max);
5911 #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION)
5913 - regclass_swash - prepare the utf8 swash
5917 Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
5923 RXi_GET_DECL(prog,progi);
5924 const struct reg_data * const data = prog ? progi->data : NULL;
5926 PERL_ARGS_ASSERT_REGCLASS_SWASH;
5928 if (data && data->count) {
5929 const U32 n = ARG(node);
5931 if (data->what[n] == 's') {
5932 SV * const rv = MUTABLE_SV(data->data[n]);
5933 AV * const av = MUTABLE_AV(SvRV(rv));
5934 SV **const ary = AvARRAY(av);
5937 /* See the end of regcomp.c:S_regclass() for
5938 * documentation of these array elements. */
5941 a = SvROK(ary[1]) ? &ary[1] : NULL;
5942 b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : NULL;
5946 else if (si && doinit) {
5947 sw = swash_init("utf8", "", si, 1, 0);
5948 (void)av_store(av, 1, sw);
5965 - reginclass - determine if a character falls into a character class
5967 The n is the ANYOF regnode, the p is the target string, lenp
5968 is pointer to the maximum length of how far to go in the p
5969 (if the lenp is zero, UTF8SKIP(p) is used),
5970 do_utf8 tells whether the target string is in UTF-8.
5975 S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8)
5978 const char flags = ANYOF_FLAGS(n);
5984 PERL_ARGS_ASSERT_REGINCLASS;
5986 if (do_utf8 && !UTF8_IS_INVARIANT(c)) {
5987 c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len,
5988 (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV)
5989 | UTF8_ALLOW_FFFF | UTF8_CHECK_ONLY);
5990 /* see [perl #37836] for UTF8_ALLOW_ANYUV; [perl #38293] for
5991 * UTF8_ALLOW_FFFF */
5992 if (len == (STRLEN)-1)
5993 Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)");
5996 plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c));
5997 if (do_utf8 || (flags & ANYOF_UNICODE)) {
6000 if (do_utf8 && !ANYOF_RUNTIME(n)) {
6001 if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c))
6004 if (!match && do_utf8 && (flags & ANYOF_UNICODE_ALL) && c >= 256)
6008 SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av);
6016 utf8_p = bytes_to_utf8(p, &len);
6018 if (swash_fetch(sw, utf8_p, 1))
6020 else if (flags & ANYOF_FOLD) {
6021 if (!match && lenp && av) {
6023 for (i = 0; i <= av_len(av); i++) {
6024 SV* const sv = *av_fetch(av, i, FALSE);
6026 const char * const s = SvPV_const(sv, len);
6027 if (len <= plen && memEQ(s, (char*)utf8_p, len)) {
6035 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
6038 to_utf8_fold(utf8_p, tmpbuf, &tmplen);
6039 if (swash_fetch(sw, tmpbuf, 1))
6044 /* If we allocated a string above, free it */
6045 if (! do_utf8) Safefree(utf8_p);
6048 if (match && lenp && *lenp == 0)
6049 *lenp = UNISKIP(NATIVE_TO_UNI(c));
6051 if (!match && c < 256) {
6052 if (ANYOF_BITMAP_TEST(n, c))
6054 else if (flags & ANYOF_FOLD) {
6057 if (flags & ANYOF_LOCALE) {
6058 PL_reg_flags |= RF_tainted;
6059 f = PL_fold_locale[c];
6063 if (f != c && ANYOF_BITMAP_TEST(n, f))
6067 if (!match && (flags & ANYOF_CLASS)) {
6068 PL_reg_flags |= RF_tainted;
6070 (ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) ||
6071 (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) ||
6072 (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) ||
6073 (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) ||
6074 (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) ||
6075 (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) ||
6076 (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) ||
6077 (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) ||
6078 (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) ||
6079 (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) ||
6080 (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) ||
6081 (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) ||
6082 (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) ||
6083 (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) ||
6084 (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) ||
6085 (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) ||
6086 (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) ||
6087 (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) ||
6088 (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) ||
6089 (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) ||
6090 (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) ||
6091 (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) ||
6092 (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) ||
6093 (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) ||
6094 (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) ||
6095 (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) ||
6096 (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) ||
6097 (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) ||
6098 (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) ||
6099 (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c))
6100 ) /* How's that for a conditional? */
6107 return (flags & ANYOF_INVERT) ? !match : match;
6111 S_reghop3(U8 *s, I32 off, const U8* lim)
6115 PERL_ARGS_ASSERT_REGHOP3;
6118 while (off-- && s < lim) {
6119 /* XXX could check well-formedness here */
6124 while (off++ && s > lim) {
6126 if (UTF8_IS_CONTINUED(*s)) {
6127 while (s > lim && UTF8_IS_CONTINUATION(*s))
6130 /* XXX could check well-formedness here */
6137 /* there are a bunch of places where we use two reghop3's that should
6138 be replaced with this routine. but since thats not done yet
6139 we ifdef it out - dmq
6142 S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim)
6146 PERL_ARGS_ASSERT_REGHOP4;
6149 while (off-- && s < rlim) {
6150 /* XXX could check well-formedness here */
6155 while (off++ && s > llim) {
6157 if (UTF8_IS_CONTINUED(*s)) {
6158 while (s > llim && UTF8_IS_CONTINUATION(*s))
6161 /* XXX could check well-formedness here */
6169 S_reghopmaybe3(U8* s, I32 off, const U8* lim)
6173 PERL_ARGS_ASSERT_REGHOPMAYBE3;
6176 while (off-- && s < lim) {
6177 /* XXX could check well-formedness here */
6184 while (off++ && s > lim) {
6186 if (UTF8_IS_CONTINUED(*s)) {
6187 while (s > lim && UTF8_IS_CONTINUATION(*s))
6190 /* XXX could check well-formedness here */
6199 restore_pos(pTHX_ void *arg)
6202 regexp * const rex = (regexp *)arg;
6203 if (PL_reg_eval_set) {
6204 if (PL_reg_oldsaved) {
6205 rex->subbeg = PL_reg_oldsaved;
6206 rex->sublen = PL_reg_oldsavedlen;
6207 #ifdef PERL_OLD_COPY_ON_WRITE
6208 rex->saved_copy = PL_nrs;
6210 RXp_MATCH_COPIED_on(rex);
6212 PL_reg_magic->mg_len = PL_reg_oldpos;
6213 PL_reg_eval_set = 0;
6214 PL_curpm = PL_reg_oldcurpm;
6219 S_to_utf8_substr(pTHX_ register regexp *prog)
6223 PERL_ARGS_ASSERT_TO_UTF8_SUBSTR;
6226 if (prog->substrs->data[i].substr
6227 && !prog->substrs->data[i].utf8_substr) {
6228 SV* const sv = newSVsv(prog->substrs->data[i].substr);
6229 prog->substrs->data[i].utf8_substr = sv;
6230 sv_utf8_upgrade(sv);
6231 if (SvVALID(prog->substrs->data[i].substr)) {
6232 const U8 flags = BmFLAGS(prog->substrs->data[i].substr);
6233 if (flags & FBMcf_TAIL) {
6234 /* Trim the trailing \n that fbm_compile added last
6236 SvCUR_set(sv, SvCUR(sv) - 1);
6237 /* Whilst this makes the SV technically "invalid" (as its
6238 buffer is no longer followed by "\0") when fbm_compile()
6239 adds the "\n" back, a "\0" is restored. */
6241 fbm_compile(sv, flags);
6243 if (prog->substrs->data[i].substr == prog->check_substr)
6244 prog->check_utf8 = sv;
6250 S_to_byte_substr(pTHX_ register regexp *prog)
6255 PERL_ARGS_ASSERT_TO_BYTE_SUBSTR;
6258 if (prog->substrs->data[i].utf8_substr
6259 && !prog->substrs->data[i].substr) {
6260 SV* sv = newSVsv(prog->substrs->data[i].utf8_substr);
6261 if (sv_utf8_downgrade(sv, TRUE)) {
6262 if (SvVALID(prog->substrs->data[i].utf8_substr)) {
6264 = BmFLAGS(prog->substrs->data[i].utf8_substr);
6265 if (flags & FBMcf_TAIL) {
6266 /* Trim the trailing \n that fbm_compile added last
6268 SvCUR_set(sv, SvCUR(sv) - 1);
6270 fbm_compile(sv, flags);
6276 prog->substrs->data[i].substr = sv;
6277 if (prog->substrs->data[i].utf8_substr == prog->check_utf8)
6278 prog->check_substr = sv;
6285 * c-indentation-style: bsd
6287 * indent-tabs-mode: t
6290 * ex: set ts=8 sts=4 sw=4 noet: