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
77 #ifdef PERL_IN_XSUB_RE
83 #define RF_tainted 1 /* tainted information used? e.g. locale */
84 #define RF_warned 2 /* warned about big count? */
86 #define RF_utf8 8 /* Pattern contains multibyte chars? */
88 #define UTF_PATTERN ((PL_reg_flags & RF_utf8) != 0)
94 /* Valid for non-utf8 strings, non-ANYOFV nodes only: avoids the reginclass
95 * call if there are no complications: i.e., if everything matchable is
96 * straight forward in the bitmap */
97 #define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) \
98 : ANYOF_BITMAP_TEST(p,*(c)))
104 #define CHR_SVLEN(sv) (utf8_target ? 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 */
123 /* Often 'str' is a hard-coded utf8 string instead of utfebcdic. so just
124 * skip the check on EBCDIC platforms */
125 # define LOAD_UTF8_CHARCLASS(class,str) LOAD_UTF8_CHARCLASS_NO_CHECK(class)
127 # define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \
128 if (!CAT2(PL_utf8_,class)) { \
130 ENTER; save_re_context(); \
131 ok=CAT2(is_utf8_,class)((const U8*)str); \
132 assert(ok); assert(CAT2(PL_utf8_,class)); LEAVE; } } STMT_END
135 /* Doesn't do an assert to verify that is correct */
136 #define LOAD_UTF8_CHARCLASS_NO_CHECK(class) STMT_START { \
137 if (!CAT2(PL_utf8_,class)) { \
138 bool throw_away PERL_UNUSED_DECL; \
139 ENTER; save_re_context(); \
140 throw_away = CAT2(is_utf8_,class)((const U8*)" "); \
143 #define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a")
144 #define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0")
145 #define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ")
147 #define LOAD_UTF8_CHARCLASS_GCB() /* Grapheme cluster boundaries */ \
148 LOAD_UTF8_CHARCLASS(X_begin, " "); \
149 LOAD_UTF8_CHARCLASS(X_non_hangul, "A"); \
150 /* These are utf8 constants, and not utf-ebcdic constants, so the \
151 * assert should likely and hopefully fail on an EBCDIC machine */ \
152 LOAD_UTF8_CHARCLASS(X_extend, "\xcc\x80"); /* U+0300 */ \
154 /* No asserts are done for these, in case called on an early \
155 * Unicode version in which they map to nothing */ \
156 LOAD_UTF8_CHARCLASS_NO_CHECK(X_prepend);/* U+0E40 "\xe0\xb9\x80" */ \
157 LOAD_UTF8_CHARCLASS_NO_CHECK(X_L); /* U+1100 "\xe1\x84\x80" */ \
158 LOAD_UTF8_CHARCLASS_NO_CHECK(X_LV); /* U+AC00 "\xea\xb0\x80" */ \
159 LOAD_UTF8_CHARCLASS_NO_CHECK(X_LVT); /* U+AC01 "\xea\xb0\x81" */ \
160 LOAD_UTF8_CHARCLASS_NO_CHECK(X_LV_LVT_V);/* U+AC01 "\xea\xb0\x81" */\
161 LOAD_UTF8_CHARCLASS_NO_CHECK(X_T); /* U+11A8 "\xe1\x86\xa8" */ \
162 LOAD_UTF8_CHARCLASS_NO_CHECK(X_V) /* U+1160 "\xe1\x85\xa0" */
164 #define PLACEHOLDER /* Something for the preprocessor to grab onto */
166 /* The actual code for CCC_TRY, which uses several variables from the routine
167 * it's callable from. It is designed to be the bulk of a case statement.
168 * FUNC is the macro or function to call on non-utf8 targets that indicate if
169 * nextchr matches the class.
170 * UTF8_TEST is the whole test string to use for utf8 targets
171 * LOAD is what to use to test, and if not present to load in the swash for the
173 * POS_OR_NEG is either empty or ! to complement the results of FUNC or
175 * The logic is: Fail if we're at the end-of-string; otherwise if the target is
176 * utf8 and a variant, load the swash if necessary and test using the utf8
177 * test. Advance to the next character if test is ok, otherwise fail; If not
178 * utf8 or an invariant under utf8, use the non-utf8 test, and fail if it
179 * fails, or advance to the next character */
181 #define _CCC_TRY_CODE(POS_OR_NEG, FUNC, UTF8_TEST, CLASS, STR) \
182 if (locinput >= PL_regeol) { \
185 if (utf8_target && UTF8_IS_CONTINUED(nextchr)) { \
186 LOAD_UTF8_CHARCLASS(CLASS, STR); \
187 if (POS_OR_NEG (UTF8_TEST)) { \
190 locinput += PL_utf8skip[nextchr]; \
191 nextchr = UCHARAT(locinput); \
194 if (POS_OR_NEG (FUNC(nextchr))) { \
197 nextchr = UCHARAT(++locinput); \
200 /* Handle the non-locale cases for a character class and its complement. It
201 * calls _CCC_TRY_CODE with a ! to complement the test for the character class.
202 * This is because that code fails when the test succeeds, so we want to have
203 * the test fail so that the code succeeds. The swash is stored in a
204 * predictable PL_ place */
205 #define _CCC_TRY_NONLOCALE(NAME, NNAME, FUNC, \
208 _CCC_TRY_CODE( !, FUNC, \
209 cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS), \
210 (U8*)locinput, TRUE)), \
213 _CCC_TRY_CODE( PLACEHOLDER , FUNC, \
214 cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS), \
215 (U8*)locinput, TRUE)), \
218 /* Generate the case statements for both locale and non-locale character
219 * classes in regmatch for classes that don't have special unicode semantics.
220 * Locales don't use an immediate swash, but an intermediary special locale
221 * function that is called on the pointer to the current place in the input
222 * string. That function will resolve to needing the same swash. One might
223 * think that because we don't know what the locale will match, we shouldn't
224 * check with the swash loading function that it loaded properly; ie, that we
225 * should use LOAD_UTF8_CHARCLASS_NO_CHECK for those, but what is passed to the
226 * regular LOAD_UTF8_CHARCLASS is in non-locale terms, and so locale is
228 #define CCC_TRY(NAME, NNAME, FUNC, \
229 NAMEL, NNAMEL, LCFUNC, LCFUNC_utf8, \
230 NAMEA, NNAMEA, FUNCA, \
233 PL_reg_flags |= RF_tainted; \
234 _CCC_TRY_CODE( !, LCFUNC, LCFUNC_utf8((U8*)locinput), CLASS, STR) \
236 PL_reg_flags |= RF_tainted; \
237 _CCC_TRY_CODE( PLACEHOLDER, LCFUNC, LCFUNC_utf8((U8*)locinput), \
240 if (locinput >= PL_regeol || ! FUNCA(nextchr)) { \
243 /* Matched a utf8-invariant, so don't have to worry about utf8 */ \
244 nextchr = UCHARAT(++locinput); \
247 if (locinput >= PL_regeol || FUNCA(nextchr)) { \
251 locinput += PL_utf8skip[nextchr]; \
252 nextchr = UCHARAT(locinput); \
255 nextchr = UCHARAT(++locinput); \
258 /* Generate the non-locale cases */ \
259 _CCC_TRY_NONLOCALE(NAME, NNAME, FUNC, CLASS, STR)
261 /* This is like CCC_TRY, but has an extra set of parameters for generating case
262 * statements to handle separate Unicode semantics nodes */
263 #define CCC_TRY_U(NAME, NNAME, FUNC, \
264 NAMEL, NNAMEL, LCFUNC, LCFUNC_utf8, \
265 NAMEU, NNAMEU, FUNCU, \
266 NAMEA, NNAMEA, FUNCA, \
268 CCC_TRY(NAME, NNAME, FUNC, \
269 NAMEL, NNAMEL, LCFUNC, LCFUNC_utf8, \
270 NAMEA, NNAMEA, FUNCA, \
272 _CCC_TRY_NONLOCALE(NAMEU, NNAMEU, FUNCU, CLASS, STR)
274 /* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */
276 /* for use after a quantifier and before an EXACT-like node -- japhy */
277 /* it would be nice to rework regcomp.sym to generate this stuff. sigh
279 * NOTE that *nothing* that affects backtracking should be in here, specifically
280 * VERBS must NOT be included. JUMPABLE is used to determine if we can ignore a
281 * node that is in between two EXACT like nodes when ascertaining what the required
282 * "follow" character is. This should probably be moved to regex compile time
283 * although it may be done at run time beause of the REF possibility - more
284 * investigation required. -- demerphq
286 #define JUMPABLE(rn) ( \
288 (OP(rn) == CLOSE && (!cur_eval || cur_eval->u.eval.close_paren != ARG(rn))) || \
290 OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
291 OP(rn) == PLUS || OP(rn) == MINMOD || \
293 (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \
295 #define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT)
297 #define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF )
300 /* Currently these are only used when PL_regkind[OP(rn)] == EXACT so
301 we don't need this definition. */
302 #define IS_TEXT(rn) ( OP(rn)==EXACT || OP(rn)==REF || OP(rn)==NREF )
303 #define IS_TEXTF(rn) ( OP(rn)==EXACTFU || OP(rn)==EXACTFU_SS || OP(rn)==EXACTFU_TRICKYFOLD || OP(rn)==EXACTFA || OP(rn)==EXACTF || OP(rn)==REFF || OP(rn)==NREFF )
304 #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL )
307 /* ... so we use this as its faster. */
308 #define IS_TEXT(rn) ( OP(rn)==EXACT )
309 #define IS_TEXTFU(rn) ( OP(rn)==EXACTFU || OP(rn)==EXACTFU_SS || OP(rn)==EXACTFU_TRICKYFOLD || OP(rn) == EXACTFA)
310 #define IS_TEXTF(rn) ( OP(rn)==EXACTF )
311 #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL )
316 Search for mandatory following text node; for lookahead, the text must
317 follow but for lookbehind (rn->flags != 0) we skip to the next step.
319 #define FIND_NEXT_IMPT(rn) STMT_START { \
320 while (JUMPABLE(rn)) { \
321 const OPCODE type = OP(rn); \
322 if (type == SUSPEND || PL_regkind[type] == CURLY) \
323 rn = NEXTOPER(NEXTOPER(rn)); \
324 else if (type == PLUS) \
326 else if (type == IFMATCH) \
327 rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
328 else rn += NEXT_OFF(rn); \
333 static void restore_pos(pTHX_ void *arg);
335 #define REGCP_PAREN_ELEMS 3
336 #define REGCP_OTHER_ELEMS 3
337 #define REGCP_FRAME_ELEMS 1
338 /* REGCP_FRAME_ELEMS are not part of the REGCP_OTHER_ELEMS and
339 * are needed for the regexp context stack bookkeeping. */
342 S_regcppush(pTHX_ const regexp *rex, I32 parenfloor)
345 const int retval = PL_savestack_ix;
346 const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
347 const UV total_elems = paren_elems_to_push + REGCP_OTHER_ELEMS;
348 const UV elems_shifted = total_elems << SAVE_TIGHT_SHIFT;
350 GET_RE_DEBUG_FLAGS_DECL;
352 PERL_ARGS_ASSERT_REGCPPUSH;
354 if (paren_elems_to_push < 0)
355 Perl_croak(aTHX_ "panic: paren_elems_to_push, %i < 0",
356 paren_elems_to_push);
358 if ((elems_shifted >> SAVE_TIGHT_SHIFT) != total_elems)
359 Perl_croak(aTHX_ "panic: paren_elems_to_push offset %"UVuf
360 " out of range (%lu-%ld)",
361 total_elems, (unsigned long)PL_regsize, (long)parenfloor);
363 SSGROW(total_elems + REGCP_FRAME_ELEMS);
366 if ((int)PL_regsize > (int)parenfloor)
367 PerlIO_printf(Perl_debug_log,
368 "rex=0x%"UVxf" offs=0x%"UVxf": saving capture indices:\n",
373 for (p = parenfloor+1; p <= (I32)PL_regsize; p++) {
374 /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
375 SSPUSHINT(rex->offs[p].end);
376 SSPUSHINT(rex->offs[p].start);
377 SSPUSHINT(rex->offs[p].start_tmp);
378 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
379 " \\%"UVuf": %"IVdf"(%"IVdf")..%"IVdf"\n",
381 (IV)rex->offs[p].start,
382 (IV)rex->offs[p].start_tmp,
386 /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
387 SSPUSHINT(PL_regsize);
388 SSPUSHINT(rex->lastparen);
389 SSPUSHINT(rex->lastcloseparen);
390 SSPUSHUV(SAVEt_REGCONTEXT | elems_shifted); /* Magic cookie. */
395 /* These are needed since we do not localize EVAL nodes: */
396 #define REGCP_SET(cp) \
398 PerlIO_printf(Perl_debug_log, \
399 " Setting an EVAL scope, savestack=%"IVdf"\n", \
400 (IV)PL_savestack_ix)); \
403 #define REGCP_UNWIND(cp) \
405 if (cp != PL_savestack_ix) \
406 PerlIO_printf(Perl_debug_log, \
407 " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
408 (IV)(cp), (IV)PL_savestack_ix)); \
411 #define UNWIND_PAREN(lp, lcp) \
412 for (n = rex->lastparen; n > lp; n--) \
413 rex->offs[n].end = -1; \
414 rex->lastparen = n; \
415 rex->lastcloseparen = lcp;
419 S_regcppop(pTHX_ regexp *rex)
424 GET_RE_DEBUG_FLAGS_DECL;
426 PERL_ARGS_ASSERT_REGCPPOP;
428 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
430 assert((i & SAVE_MASK) == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
431 i >>= SAVE_TIGHT_SHIFT; /* Parentheses elements to pop. */
432 rex->lastcloseparen = SSPOPINT;
433 rex->lastparen = SSPOPINT;
434 PL_regsize = SSPOPINT;
436 i -= REGCP_OTHER_ELEMS;
437 /* Now restore the parentheses context. */
439 if (i || rex->lastparen + 1 <= rex->nparens)
440 PerlIO_printf(Perl_debug_log,
441 "rex=0x%"UVxf" offs=0x%"UVxf": restoring capture indices to:\n",
447 for ( ; i > 0; i -= REGCP_PAREN_ELEMS) {
449 rex->offs[paren].start_tmp = SSPOPINT;
450 rex->offs[paren].start = SSPOPINT;
452 if (paren <= rex->lastparen)
453 rex->offs[paren].end = tmps;
454 DEBUG_BUFFERS_r( PerlIO_printf(Perl_debug_log,
455 " \\%"UVuf": %"IVdf"(%"IVdf")..%"IVdf"%s\n",
457 (IV)rex->offs[paren].start,
458 (IV)rex->offs[paren].start_tmp,
459 (IV)rex->offs[paren].end,
460 (paren > rex->lastparen ? "(skipped)" : ""));
465 /* It would seem that the similar code in regtry()
466 * already takes care of this, and in fact it is in
467 * a better location to since this code can #if 0-ed out
468 * but the code in regtry() is needed or otherwise tests
469 * requiring null fields (pat.t#187 and split.t#{13,14}
470 * (as of patchlevel 7877) will fail. Then again,
471 * this code seems to be necessary or otherwise
472 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
473 * --jhi updated by dapm */
474 for (i = rex->lastparen + 1; i <= rex->nparens; i++) {
476 rex->offs[i].start = -1;
477 rex->offs[i].end = -1;
478 DEBUG_BUFFERS_r( PerlIO_printf(Perl_debug_log,
479 " \\%"UVuf": %s ..-1 undeffing\n",
481 (i > PL_regsize) ? "-1" : " "
487 /* restore the parens and associated vars at savestack position ix,
488 * but without popping the stack */
491 S_regcp_restore(pTHX_ regexp *rex, I32 ix)
493 I32 tmpix = PL_savestack_ix;
494 PL_savestack_ix = ix;
496 PL_savestack_ix = tmpix;
499 #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
502 * pregexec and friends
505 #ifndef PERL_IN_XSUB_RE
507 - pregexec - match a regexp against a string
510 Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, register char *strend,
511 char *strbeg, I32 minend, SV *screamer, U32 nosave)
512 /* strend: pointer to null at end of string */
513 /* strbeg: real beginning of string */
514 /* minend: end of match must be >=minend after stringarg. */
515 /* nosave: For optimizations. */
517 PERL_ARGS_ASSERT_PREGEXEC;
520 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
521 nosave ? 0 : REXEC_COPY_STR);
526 * Need to implement the following flags for reg_anch:
528 * USE_INTUIT_NOML - Useful to call re_intuit_start() first
530 * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer
531 * INTUIT_AUTORITATIVE_ML
532 * INTUIT_ONCE_NOML - Intuit can match in one location only.
535 * Another flag for this function: SECOND_TIME (so that float substrs
536 * with giant delta may be not rechecked).
539 /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
541 /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend.
542 Otherwise, only SvCUR(sv) is used to get strbeg. */
544 /* XXXX We assume that strpos is strbeg unless sv. */
546 /* XXXX Some places assume that there is a fixed substring.
547 An update may be needed if optimizer marks as "INTUITable"
548 RExen without fixed substrings. Similarly, it is assumed that
549 lengths of all the strings are no more than minlen, thus they
550 cannot come from lookahead.
551 (Or minlen should take into account lookahead.)
552 NOTE: Some of this comment is not correct. minlen does now take account
553 of lookahead/behind. Further research is required. -- demerphq
557 /* A failure to find a constant substring means that there is no need to make
558 an expensive call to REx engine, thus we celebrate a failure. Similarly,
559 finding a substring too deep into the string means that less calls to
560 regtry() should be needed.
562 REx compiler's optimizer found 4 possible hints:
563 a) Anchored substring;
565 c) Whether we are anchored (beginning-of-line or \G);
566 d) First node (of those at offset 0) which may distinguish positions;
567 We use a)b)d) and multiline-part of c), and try to find a position in the
568 string which does not contradict any of them.
571 /* Most of decisions we do here should have been done at compile time.
572 The nodes of the REx which we used for the search should have been
573 deleted from the finite automaton. */
576 Perl_re_intuit_start(pTHX_ REGEXP * const rx, SV *sv, char *strpos,
577 char *strend, const U32 flags, re_scream_pos_data *data)
580 struct regexp *const prog = (struct regexp *)SvANY(rx);
581 register I32 start_shift = 0;
582 /* Should be nonnegative! */
583 register I32 end_shift = 0;
588 const bool utf8_target = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */
590 register char *other_last = NULL; /* other substr checked before this */
591 char *check_at = NULL; /* check substr found at this pos */
592 char *checked_upto = NULL; /* how far into the string we have already checked using find_byclass*/
593 const I32 multiline = prog->extflags & RXf_PMf_MULTILINE;
594 RXi_GET_DECL(prog,progi);
596 const char * const i_strpos = strpos;
598 GET_RE_DEBUG_FLAGS_DECL;
600 PERL_ARGS_ASSERT_RE_INTUIT_START;
601 PERL_UNUSED_ARG(flags);
602 PERL_UNUSED_ARG(data);
604 RX_MATCH_UTF8_set(rx,utf8_target);
607 PL_reg_flags |= RF_utf8;
610 debug_start_match(rx, utf8_target, strpos, strend,
611 sv ? "Guessing start of match in sv for"
612 : "Guessing start of match in string for");
615 /* CHR_DIST() would be more correct here but it makes things slow. */
616 if (prog->minlen > strend - strpos) {
617 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
618 "String too short... [re_intuit_start]\n"));
622 strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
625 if (!prog->check_utf8 && prog->check_substr)
626 to_utf8_substr(prog);
627 check = prog->check_utf8;
629 if (!prog->check_substr && prog->check_utf8)
630 to_byte_substr(prog);
631 check = prog->check_substr;
633 if (check == &PL_sv_undef) {
634 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
635 "Non-utf8 string cannot match utf8 check string\n"));
638 if (prog->extflags & RXf_ANCH) { /* Match at beg-of-str or after \n */
639 ml_anch = !( (prog->extflags & RXf_ANCH_SINGLE)
640 || ( (prog->extflags & RXf_ANCH_BOL)
641 && !multiline ) ); /* Check after \n? */
644 if ( !(prog->extflags & RXf_ANCH_GPOS) /* Checked by the caller */
645 && !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */
646 /* SvCUR is not set on references: SvRV and SvPVX_const overlap */
648 && (strpos != strbeg)) {
649 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
652 if (prog->check_offset_min == prog->check_offset_max &&
653 !(prog->extflags & RXf_CANY_SEEN)) {
654 /* Substring at constant offset from beg-of-str... */
657 s = HOP3c(strpos, prog->check_offset_min, strend);
660 slen = SvCUR(check); /* >= 1 */
662 if ( strend - s > slen || strend - s < slen - 1
663 || (strend - s == slen && strend[-1] != '\n')) {
664 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
667 /* Now should match s[0..slen-2] */
669 if (slen && (*SvPVX_const(check) != *s
671 && memNE(SvPVX_const(check), s, slen)))) {
673 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
677 else if (*SvPVX_const(check) != *s
678 || ((slen = SvCUR(check)) > 1
679 && memNE(SvPVX_const(check), s, slen)))
682 goto success_at_start;
685 /* Match is anchored, but substr is not anchored wrt beg-of-str. */
687 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
688 end_shift = prog->check_end_shift;
691 const I32 end = prog->check_offset_max + CHR_SVLEN(check)
692 - (SvTAIL(check) != 0);
693 const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
695 if (end_shift < eshift)
699 else { /* Can match at random position */
702 start_shift = prog->check_offset_min; /* okay to underestimate on CC */
703 end_shift = prog->check_end_shift;
705 /* end shift should be non negative here */
708 #ifdef QDEBUGGING /* 7/99: reports of failure (with the older version) */
710 Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ",
711 (IV)end_shift, RX_PRECOMP(prog));
715 /* Find a possible match in the region s..strend by looking for
716 the "check" substring in the region corrected by start/end_shift. */
719 I32 srch_start_shift = start_shift;
720 I32 srch_end_shift = end_shift;
723 if (srch_start_shift < 0 && strbeg - s > srch_start_shift) {
724 srch_end_shift -= ((strbeg - s) - srch_start_shift);
725 srch_start_shift = strbeg - s;
727 DEBUG_OPTIMISE_MORE_r({
728 PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n",
729 (IV)prog->check_offset_min,
730 (IV)srch_start_shift,
732 (IV)prog->check_end_shift);
735 if (prog->extflags & RXf_CANY_SEEN) {
736 start_point= (U8*)(s + srch_start_shift);
737 end_point= (U8*)(strend - srch_end_shift);
739 start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend);
740 end_point= HOP3(strend, -srch_end_shift, strbeg);
742 DEBUG_OPTIMISE_MORE_r({
743 PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n",
744 (int)(end_point - start_point),
745 (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point),
749 s = fbm_instr( start_point, end_point,
750 check, multiline ? FBMrf_MULTILINE : 0);
752 /* Update the count-of-usability, remove useless subpatterns,
756 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
757 SvPVX_const(check), RE_SV_DUMPLEN(check), 30);
758 PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s",
759 (s ? "Found" : "Did not find"),
760 (check == (utf8_target ? prog->anchored_utf8 : prog->anchored_substr)
761 ? "anchored" : "floating"),
764 (s ? " at offset " : "...\n") );
769 /* Finish the diagnostic message */
770 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
772 /* XXX dmq: first branch is for positive lookbehind...
773 Our check string is offset from the beginning of the pattern.
774 So we need to do any stclass tests offset forward from that
783 /* Got a candidate. Check MBOL anchoring, and the *other* substr.
784 Start with the other substr.
785 XXXX no SCREAM optimization yet - and a very coarse implementation
786 XXXX /ttx+/ results in anchored="ttx", floating="x". floating will
787 *always* match. Probably should be marked during compile...
788 Probably it is right to do no SCREAM here...
791 if (utf8_target ? (prog->float_utf8 && prog->anchored_utf8)
792 : (prog->float_substr && prog->anchored_substr))
794 /* Take into account the "other" substring. */
795 /* XXXX May be hopelessly wrong for UTF... */
798 if (check == (utf8_target ? prog->float_utf8 : prog->float_substr)) {
801 char * const last = HOP3c(s, -start_shift, strbeg);
803 char * const saved_s = s;
806 t = s - prog->check_offset_max;
807 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
809 || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos))
814 t = HOP3c(t, prog->anchored_offset, strend);
815 if (t < other_last) /* These positions already checked */
817 last2 = last1 = HOP3c(strend, -prog->minlen, strbeg);
820 /* XXXX It is not documented what units *_offsets are in.
821 We assume bytes, but this is clearly wrong.
822 Meaning this code needs to be carefully reviewed for errors.
826 /* On end-of-str: see comment below. */
827 must = utf8_target ? prog->anchored_utf8 : prog->anchored_substr;
828 if (must == &PL_sv_undef) {
830 DEBUG_r(must = prog->anchored_utf8); /* for debug */
835 HOP3(HOP3(last1, prog->anchored_offset, strend)
836 + SvCUR(must), -(SvTAIL(must)!=0), strbeg),
838 multiline ? FBMrf_MULTILINE : 0
841 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
842 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
843 PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s",
844 (s ? "Found" : "Contradicts"),
845 quoted, RE_SV_TAIL(must));
850 if (last1 >= last2) {
851 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
852 ", giving up...\n"));
855 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
856 ", trying floating at offset %ld...\n",
857 (long)(HOP3c(saved_s, 1, strend) - i_strpos)));
858 other_last = HOP3c(last1, prog->anchored_offset+1, strend);
859 s = HOP3c(last, 1, strend);
863 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
864 (long)(s - i_strpos)));
865 t = HOP3c(s, -prog->anchored_offset, strbeg);
866 other_last = HOP3c(s, 1, strend);
874 else { /* Take into account the floating substring. */
876 char * const saved_s = s;
879 t = HOP3c(s, -start_shift, strbeg);
881 HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg);
882 if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
883 last = HOP3c(t, prog->float_max_offset, strend);
884 s = HOP3c(t, prog->float_min_offset, strend);
887 /* XXXX It is not documented what units *_offsets are in. Assume bytes. */
888 must = utf8_target ? prog->float_utf8 : prog->float_substr;
889 /* fbm_instr() takes into account exact value of end-of-str
890 if the check is SvTAIL(ed). Since false positives are OK,
891 and end-of-str is not later than strend we are OK. */
892 if (must == &PL_sv_undef) {
894 DEBUG_r(must = prog->float_utf8); /* for debug message */
897 s = fbm_instr((unsigned char*)s,
898 (unsigned char*)last + SvCUR(must)
900 must, multiline ? FBMrf_MULTILINE : 0);
902 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
903 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
904 PerlIO_printf(Perl_debug_log, "%s floating substr %s%s",
905 (s ? "Found" : "Contradicts"),
906 quoted, RE_SV_TAIL(must));
910 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
911 ", giving up...\n"));
914 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
915 ", trying anchored starting at offset %ld...\n",
916 (long)(saved_s + 1 - i_strpos)));
918 s = HOP3c(t, 1, strend);
922 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
923 (long)(s - i_strpos)));
924 other_last = s; /* Fix this later. --Hugo */
934 t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos);
936 DEBUG_OPTIMISE_MORE_r(
937 PerlIO_printf(Perl_debug_log,
938 "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n",
939 (IV)prog->check_offset_min,
940 (IV)prog->check_offset_max,
948 if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
950 || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos)))
953 /* Fixed substring is found far enough so that the match
954 cannot start at strpos. */
956 if (ml_anch && t[-1] != '\n') {
957 /* Eventually fbm_*() should handle this, but often
958 anchored_offset is not 0, so this check will not be wasted. */
959 /* XXXX In the code below we prefer to look for "^" even in
960 presence of anchored substrings. And we search even
961 beyond the found float position. These pessimizations
962 are historical artefacts only. */
964 while (t < strend - prog->minlen) {
966 if (t < check_at - prog->check_offset_min) {
967 if (utf8_target ? prog->anchored_utf8 : prog->anchored_substr) {
968 /* Since we moved from the found position,
969 we definitely contradict the found anchored
970 substr. Due to the above check we do not
971 contradict "check" substr.
972 Thus we can arrive here only if check substr
973 is float. Redo checking for "other"=="fixed".
976 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
977 PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
978 goto do_other_anchored;
980 /* We don't contradict the found floating substring. */
981 /* XXXX Why not check for STCLASS? */
983 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
984 PL_colors[0], PL_colors[1], (long)(s - i_strpos)));
987 /* Position contradicts check-string */
988 /* XXXX probably better to look for check-string
989 than for "\n", so one should lower the limit for t? */
990 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
991 PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos)));
992 other_last = strpos = s = t + 1;
997 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
998 PL_colors[0], PL_colors[1]));
1002 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
1003 PL_colors[0], PL_colors[1]));
1007 ++BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr); /* hooray/5 */
1010 /* The found string does not prohibit matching at strpos,
1011 - no optimization of calling REx engine can be performed,
1012 unless it was an MBOL and we are not after MBOL,
1013 or a future STCLASS check will fail this. */
1015 /* Even in this situation we may use MBOL flag if strpos is offset
1016 wrt the start of the string. */
1017 if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
1018 && (strpos != strbeg) && strpos[-1] != '\n'
1019 /* May be due to an implicit anchor of m{.*foo} */
1020 && !(prog->intflags & PREGf_IMPLICIT))
1025 DEBUG_EXECUTE_r( if (ml_anch)
1026 PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
1027 (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]);
1030 if (!(prog->intflags & PREGf_NAUGHTY) /* XXXX If strpos moved? */
1032 prog->check_utf8 /* Could be deleted already */
1033 && --BmUSEFUL(prog->check_utf8) < 0
1034 && (prog->check_utf8 == prog->float_utf8)
1036 prog->check_substr /* Could be deleted already */
1037 && --BmUSEFUL(prog->check_substr) < 0
1038 && (prog->check_substr == prog->float_substr)
1041 /* If flags & SOMETHING - do not do it many times on the same match */
1042 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
1043 /* XXX Does the destruction order has to change with utf8_target? */
1044 SvREFCNT_dec(utf8_target ? prog->check_utf8 : prog->check_substr);
1045 SvREFCNT_dec(utf8_target ? prog->check_substr : prog->check_utf8);
1046 prog->check_substr = prog->check_utf8 = NULL; /* disable */
1047 prog->float_substr = prog->float_utf8 = NULL; /* clear */
1048 check = NULL; /* abort */
1050 /* XXXX If the check string was an implicit check MBOL, then we need to unset the relevant flag
1051 see http://bugs.activestate.com/show_bug.cgi?id=87173 */
1052 if (prog->intflags & PREGf_IMPLICIT)
1053 prog->extflags &= ~RXf_ANCH_MBOL;
1054 /* XXXX This is a remnant of the old implementation. It
1055 looks wasteful, since now INTUIT can use many
1056 other heuristics. */
1057 prog->extflags &= ~RXf_USE_INTUIT;
1058 /* XXXX What other flags might need to be cleared in this branch? */
1064 /* Last resort... */
1065 /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
1066 /* trie stclasses are too expensive to use here, we are better off to
1067 leave it to regmatch itself */
1068 if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) {
1069 /* minlen == 0 is possible if regstclass is \b or \B,
1070 and the fixed substr is ''$.
1071 Since minlen is already taken into account, s+1 is before strend;
1072 accidentally, minlen >= 1 guaranties no false positives at s + 1
1073 even for \b or \B. But (minlen? 1 : 0) below assumes that
1074 regstclass does not come from lookahead... */
1075 /* If regstclass takes bytelength more than 1: If charlength==1, OK.
1076 This leaves EXACTF-ish only, which are dealt with in find_byclass(). */
1077 const U8* const str = (U8*)STRING(progi->regstclass);
1078 const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
1079 ? CHR_DIST(str+STR_LEN(progi->regstclass), str)
1082 if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
1083 endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend);
1084 else if (prog->float_substr || prog->float_utf8)
1085 endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend);
1089 if (checked_upto < s)
1091 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf" checked_upto: %"IVdf"\n",
1092 (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg), (IV)(checked_upto- strbeg)));
1095 s = find_byclass(prog, progi->regstclass, checked_upto, endpos, NULL);
1100 const char *what = NULL;
1102 if (endpos == strend) {
1103 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1104 "Could not match STCLASS...\n") );
1107 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1108 "This position contradicts STCLASS...\n") );
1109 if ((prog->extflags & RXf_ANCH) && !ml_anch)
1111 checked_upto = HOPBACKc(endpos, start_shift);
1112 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" endpos: %"IVdf" checked_upto: %"IVdf"\n",
1113 (IV)start_shift, (IV)(check_at - strbeg), (IV)(endpos - strbeg), (IV)(checked_upto- strbeg)));
1114 /* Contradict one of substrings */
1115 if (prog->anchored_substr || prog->anchored_utf8) {
1116 if ((utf8_target ? prog->anchored_utf8 : prog->anchored_substr) == check) {
1117 DEBUG_EXECUTE_r( what = "anchored" );
1119 s = HOP3c(t, 1, strend);
1120 if (s + start_shift + end_shift > strend) {
1121 /* XXXX Should be taken into account earlier? */
1122 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1123 "Could not match STCLASS...\n") );
1128 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1129 "Looking for %s substr starting at offset %ld...\n",
1130 what, (long)(s + start_shift - i_strpos)) );
1133 /* Have both, check_string is floating */
1134 if (t + start_shift >= check_at) /* Contradicts floating=check */
1135 goto retry_floating_check;
1136 /* Recheck anchored substring, but not floating... */
1140 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1141 "Looking for anchored substr starting at offset %ld...\n",
1142 (long)(other_last - i_strpos)) );
1143 goto do_other_anchored;
1145 /* Another way we could have checked stclass at the
1146 current position only: */
1151 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1152 "Looking for /%s^%s/m starting at offset %ld...\n",
1153 PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
1156 if (!(utf8_target ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
1158 /* Check is floating substring. */
1159 retry_floating_check:
1160 t = check_at - start_shift;
1161 DEBUG_EXECUTE_r( what = "floating" );
1162 goto hop_and_restart;
1165 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1166 "By STCLASS: moving %ld --> %ld\n",
1167 (long)(t - i_strpos), (long)(s - i_strpos))
1171 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1172 "Does not contradict STCLASS...\n");
1177 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
1178 PL_colors[4], (check ? "Guessed" : "Giving up"),
1179 PL_colors[5], (long)(s - i_strpos)) );
1182 fail_finish: /* Substring not found */
1183 if (prog->check_substr || prog->check_utf8) /* could be removed already */
1184 BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
1186 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
1187 PL_colors[4], PL_colors[5]));
1191 #define DECL_TRIE_TYPE(scan) \
1192 const enum { trie_plain, trie_utf8, trie_utf8_fold, trie_latin_utf8_fold } \
1193 trie_type = ((scan->flags == EXACT) \
1194 ? (utf8_target ? trie_utf8 : trie_plain) \
1195 : (utf8_target ? trie_utf8_fold : trie_latin_utf8_fold))
1197 #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \
1198 uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \
1200 switch (trie_type) { \
1201 case trie_utf8_fold: \
1202 if ( foldlen>0 ) { \
1203 uvc = utf8n_to_uvuni( (const U8*) uscan, UTF8_MAXLEN, &len, uniflags ); \
1208 uvc = to_utf8_fold( (const U8*) uc, foldbuf, &foldlen ); \
1209 len = UTF8SKIP(uc); \
1210 skiplen = UNISKIP( uvc ); \
1211 foldlen -= skiplen; \
1212 uscan = foldbuf + skiplen; \
1215 case trie_latin_utf8_fold: \
1216 if ( foldlen>0 ) { \
1217 uvc = utf8n_to_uvuni( (const U8*) uscan, UTF8_MAXLEN, &len, uniflags ); \
1223 uvc = _to_fold_latin1( (U8) *uc, foldbuf, &foldlen, 1); \
1224 skiplen = UNISKIP( uvc ); \
1225 foldlen -= skiplen; \
1226 uscan = foldbuf + skiplen; \
1230 uvc = utf8n_to_uvuni( (const U8*) uc, UTF8_MAXLEN, &len, uniflags ); \
1237 charid = trie->charmap[ uvc ]; \
1241 if (widecharmap) { \
1242 SV** const svpp = hv_fetch(widecharmap, \
1243 (char*)&uvc, sizeof(UV), 0); \
1245 charid = (U16)SvIV(*svpp); \
1250 #define REXEC_FBC_EXACTISH_SCAN(CoNd) \
1254 && (ln == 1 || folder(s, pat_string, ln)) \
1255 && (!reginfo || regtry(reginfo, &s)) ) \
1261 #define REXEC_FBC_UTF8_SCAN(CoDe) \
1263 while (s + (uskip = UTF8SKIP(s)) <= strend) { \
1269 #define REXEC_FBC_SCAN(CoDe) \
1271 while (s < strend) { \
1277 #define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \
1278 REXEC_FBC_UTF8_SCAN( \
1280 if (tmp && (!reginfo || regtry(reginfo, &s))) \
1289 #define REXEC_FBC_CLASS_SCAN(CoNd) \
1292 if (tmp && (!reginfo || regtry(reginfo, &s))) \
1301 #define REXEC_FBC_TRYIT \
1302 if ((!reginfo || regtry(reginfo, &s))) \
1305 #define REXEC_FBC_CSCAN(CoNdUtF8,CoNd) \
1306 if (utf8_target) { \
1307 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1310 REXEC_FBC_CLASS_SCAN(CoNd); \
1313 #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \
1314 if (utf8_target) { \
1316 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1319 REXEC_FBC_CLASS_SCAN(CoNd); \
1322 #define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \
1323 PL_reg_flags |= RF_tainted; \
1324 if (utf8_target) { \
1325 REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
1328 REXEC_FBC_CLASS_SCAN(CoNd); \
1331 #define DUMP_EXEC_POS(li,s,doutf8) \
1332 dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8)
1335 #define UTF8_NOLOAD(TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
1336 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; \
1337 tmp = TEST_NON_UTF8(tmp); \
1338 REXEC_FBC_UTF8_SCAN( \
1339 if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
1348 #define UTF8_LOAD(TeSt1_UtF8, TeSt2_UtF8, IF_SUCCESS, IF_FAIL) \
1349 if (s == PL_bostr) { \
1353 U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); \
1354 tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); \
1357 LOAD_UTF8_CHARCLASS_ALNUM(); \
1358 REXEC_FBC_UTF8_SCAN( \
1359 if (tmp == ! (TeSt2_UtF8)) { \
1368 /* The only difference between the BOUND and NBOUND cases is that
1369 * REXEC_FBC_TRYIT is called when matched in BOUND, and when non-matched in
1370 * NBOUND. This is accomplished by passing it in either the if or else clause,
1371 * with the other one being empty */
1372 #define FBC_BOUND(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
1373 FBC_BOUND_COMMON(UTF8_LOAD(TEST1_UTF8, TEST2_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)
1375 #define FBC_BOUND_NOLOAD(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
1376 FBC_BOUND_COMMON(UTF8_NOLOAD(TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)
1378 #define FBC_NBOUND(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
1379 FBC_BOUND_COMMON(UTF8_LOAD(TEST1_UTF8, TEST2_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)
1381 #define FBC_NBOUND_NOLOAD(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
1382 FBC_BOUND_COMMON(UTF8_NOLOAD(TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)
1385 /* Common to the BOUND and NBOUND cases. Unfortunately the UTF8 tests need to
1386 * be passed in completely with the variable name being tested, which isn't
1387 * such a clean interface, but this is easier to read than it was before. We
1388 * are looking for the boundary (or non-boundary between a word and non-word
1389 * character. The utf8 and non-utf8 cases have the same logic, but the details
1390 * must be different. Find the "wordness" of the character just prior to this
1391 * one, and compare it with the wordness of this one. If they differ, we have
1392 * a boundary. At the beginning of the string, pretend that the previous
1393 * character was a new-line */
1394 #define FBC_BOUND_COMMON(UTF8_CODE, TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
1395 if (utf8_target) { \
1398 else { /* Not utf8 */ \
1399 tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; \
1400 tmp = TEST_NON_UTF8(tmp); \
1402 if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
1411 if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s))) \
1414 /* We know what class REx starts with. Try to find this position... */
1415 /* if reginfo is NULL, its a dryrun */
1416 /* annoyingly all the vars in this routine have different names from their counterparts
1417 in regmatch. /grrr */
1420 S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
1421 const char *strend, regmatch_info *reginfo)
1424 const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
1425 char *pat_string; /* The pattern's exactish string */
1426 char *pat_end; /* ptr to end char of pat_string */
1427 re_fold_t folder; /* Function for computing non-utf8 folds */
1428 const U8 *fold_array; /* array for folding ords < 256 */
1431 register STRLEN uskip;
1435 register I32 tmp = 1; /* Scratch variable? */
1436 register const bool utf8_target = PL_reg_match_utf8;
1437 UV utf8_fold_flags = 0;
1438 RXi_GET_DECL(prog,progi);
1440 PERL_ARGS_ASSERT_FIND_BYCLASS;
1442 /* We know what class it must start with. */
1446 if (utf8_target || OP(c) == ANYOFV) {
1447 STRLEN inclasslen = strend - s;
1448 REXEC_FBC_UTF8_CLASS_SCAN(
1449 reginclass(prog, c, (U8*)s, &inclasslen, utf8_target));
1452 REXEC_FBC_CLASS_SCAN(REGINCLASS(prog, c, (U8*)s));
1457 if (tmp && (!reginfo || regtry(reginfo, &s)))
1465 if (UTF_PATTERN || utf8_target) {
1466 utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
1467 goto do_exactf_utf8;
1469 fold_array = PL_fold_latin1; /* Latin1 folds are not affected by */
1470 folder = foldEQ_latin1; /* /a, except the sharp s one which */
1471 goto do_exactf_non_utf8; /* isn't dealt with by these */
1476 /* regcomp.c already folded this if pattern is in UTF-8 */
1477 utf8_fold_flags = 0;
1478 goto do_exactf_utf8;
1480 fold_array = PL_fold;
1482 goto do_exactf_non_utf8;
1485 if (UTF_PATTERN || utf8_target) {
1486 utf8_fold_flags = FOLDEQ_UTF8_LOCALE;
1487 goto do_exactf_utf8;
1489 fold_array = PL_fold_locale;
1490 folder = foldEQ_locale;
1491 goto do_exactf_non_utf8;
1495 utf8_fold_flags = FOLDEQ_S2_ALREADY_FOLDED;
1497 goto do_exactf_utf8;
1499 case EXACTFU_TRICKYFOLD:
1501 if (UTF_PATTERN || utf8_target) {
1502 utf8_fold_flags = (UTF_PATTERN) ? FOLDEQ_S2_ALREADY_FOLDED : 0;
1503 goto do_exactf_utf8;
1506 /* Any 'ss' in the pattern should have been replaced by regcomp,
1507 * so we don't have to worry here about this single special case
1508 * in the Latin1 range */
1509 fold_array = PL_fold_latin1;
1510 folder = foldEQ_latin1;
1514 do_exactf_non_utf8: /* Neither pattern nor string are UTF8, and there
1515 are no glitches with fold-length differences
1516 between the target string and pattern */
1518 /* The idea in the non-utf8 EXACTF* cases is to first find the
1519 * first character of the EXACTF* node and then, if necessary,
1520 * case-insensitively compare the full text of the node. c1 is the
1521 * first character. c2 is its fold. This logic will not work for
1522 * Unicode semantics and the german sharp ss, which hence should
1523 * not be compiled into a node that gets here. */
1524 pat_string = STRING(c);
1525 ln = STR_LEN(c); /* length to match in octets/bytes */
1527 /* We know that we have to match at least 'ln' bytes (which is the
1528 * same as characters, since not utf8). If we have to match 3
1529 * characters, and there are only 2 availabe, we know without
1530 * trying that it will fail; so don't start a match past the
1531 * required minimum number from the far end */
1532 e = HOP3c(strend, -((I32)ln), s);
1534 if (!reginfo && e < s) {
1535 e = s; /* Due to minlen logic of intuit() */
1539 c2 = fold_array[c1];
1540 if (c1 == c2) { /* If char and fold are the same */
1541 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
1544 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
1553 /* If one of the operands is in utf8, we can't use the simpler
1554 * folding above, due to the fact that many different characters
1555 * can have the same fold, or portion of a fold, or different-
1557 pat_string = STRING(c);
1558 ln = STR_LEN(c); /* length to match in octets/bytes */
1559 pat_end = pat_string + ln;
1560 lnc = (UTF_PATTERN) /* length to match in characters */
1561 ? utf8_length((U8 *) pat_string, (U8 *) pat_end)
1564 /* We have 'lnc' characters to match in the pattern, but because of
1565 * multi-character folding, each character in the target can match
1566 * up to 3 characters (Unicode guarantees it will never exceed
1567 * this) if it is utf8-encoded; and up to 2 if not (based on the
1568 * fact that the Latin 1 folds are already determined, and the
1569 * only multi-char fold in that range is the sharp-s folding to
1570 * 'ss'. Thus, a pattern character can match as little as 1/3 of a
1571 * string character. Adjust lnc accordingly, rounding up, so that
1572 * if we need to match at least 4+1/3 chars, that really is 5. */
1573 expansion = (utf8_target) ? UTF8_MAX_FOLD_CHAR_EXPAND : 2;
1574 lnc = (lnc + expansion - 1) / expansion;
1576 /* As in the non-UTF8 case, if we have to match 3 characters, and
1577 * only 2 are left, it's guaranteed to fail, so don't start a
1578 * match that would require us to go beyond the end of the string
1580 e = HOP3c(strend, -((I32)lnc), s);
1582 if (!reginfo && e < s) {
1583 e = s; /* Due to minlen logic of intuit() */
1586 /* XXX Note that we could recalculate e to stop the loop earlier,
1587 * as the worst case expansion above will rarely be met, and as we
1588 * go along we would usually find that e moves further to the left.
1589 * This would happen only after we reached the point in the loop
1590 * where if there were no expansion we should fail. Unclear if
1591 * worth the expense */
1594 char *my_strend= (char *)strend;
1595 if (foldEQ_utf8_flags(s, &my_strend, 0, utf8_target,
1596 pat_string, NULL, ln, cBOOL(UTF_PATTERN), utf8_fold_flags)
1597 && (!reginfo || regtry(reginfo, &s)) )
1601 s += (utf8_target) ? UTF8SKIP(s) : 1;
1606 PL_reg_flags |= RF_tainted;
1607 FBC_BOUND(isALNUM_LC,
1608 isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp)),
1609 isALNUM_LC_utf8((U8*)s));
1612 PL_reg_flags |= RF_tainted;
1613 FBC_NBOUND(isALNUM_LC,
1614 isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp)),
1615 isALNUM_LC_utf8((U8*)s));
1618 FBC_BOUND(isWORDCHAR,
1620 cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
1623 FBC_BOUND_NOLOAD(isWORDCHAR_A,
1625 isWORDCHAR_A((U8*)s));
1628 FBC_NBOUND(isWORDCHAR,
1630 cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
1633 FBC_NBOUND_NOLOAD(isWORDCHAR_A,
1635 isWORDCHAR_A((U8*)s));
1638 FBC_BOUND(isWORDCHAR_L1,
1640 cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
1643 FBC_NBOUND(isWORDCHAR_L1,
1645 cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
1648 REXEC_FBC_CSCAN_TAINT(
1649 isALNUM_LC_utf8((U8*)s),
1654 REXEC_FBC_CSCAN_PRELOAD(
1655 LOAD_UTF8_CHARCLASS_ALNUM(),
1656 swash_fetch(PL_utf8_alnum,(U8*)s, utf8_target),
1657 isWORDCHAR_L1((U8) *s)
1661 REXEC_FBC_CSCAN_PRELOAD(
1662 LOAD_UTF8_CHARCLASS_ALNUM(),
1663 swash_fetch(PL_utf8_alnum,(U8*)s, utf8_target),
1668 /* Don't need to worry about utf8, as it can match only a single
1669 * byte invariant character */
1670 REXEC_FBC_CLASS_SCAN( isWORDCHAR_A(*s));
1673 REXEC_FBC_CSCAN_PRELOAD(
1674 LOAD_UTF8_CHARCLASS_ALNUM(),
1675 !swash_fetch(PL_utf8_alnum,(U8*)s, utf8_target),
1676 ! isWORDCHAR_L1((U8) *s)
1680 REXEC_FBC_CSCAN_PRELOAD(
1681 LOAD_UTF8_CHARCLASS_ALNUM(),
1682 !swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target),
1693 REXEC_FBC_CSCAN_TAINT(
1694 !isALNUM_LC_utf8((U8*)s),
1699 REXEC_FBC_CSCAN_PRELOAD(
1700 LOAD_UTF8_CHARCLASS_SPACE(),
1701 *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target),
1706 REXEC_FBC_CSCAN_PRELOAD(
1707 LOAD_UTF8_CHARCLASS_SPACE(),
1708 *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target),
1713 /* Don't need to worry about utf8, as it can match only a single
1714 * byte invariant character */
1715 REXEC_FBC_CLASS_SCAN( isSPACE_A(*s));
1718 REXEC_FBC_CSCAN_TAINT(
1719 isSPACE_LC_utf8((U8*)s),
1724 REXEC_FBC_CSCAN_PRELOAD(
1725 LOAD_UTF8_CHARCLASS_SPACE(),
1726 !( *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target)),
1727 ! isSPACE_L1((U8) *s)
1731 REXEC_FBC_CSCAN_PRELOAD(
1732 LOAD_UTF8_CHARCLASS_SPACE(),
1733 !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target)),
1744 REXEC_FBC_CSCAN_TAINT(
1745 !isSPACE_LC_utf8((U8*)s),
1750 REXEC_FBC_CSCAN_PRELOAD(
1751 LOAD_UTF8_CHARCLASS_DIGIT(),
1752 swash_fetch(PL_utf8_digit,(U8*)s, utf8_target),
1757 /* Don't need to worry about utf8, as it can match only a single
1758 * byte invariant character */
1759 REXEC_FBC_CLASS_SCAN( isDIGIT_A(*s));
1762 REXEC_FBC_CSCAN_TAINT(
1763 isDIGIT_LC_utf8((U8*)s),
1768 REXEC_FBC_CSCAN_PRELOAD(
1769 LOAD_UTF8_CHARCLASS_DIGIT(),
1770 !swash_fetch(PL_utf8_digit,(U8*)s, utf8_target),
1781 REXEC_FBC_CSCAN_TAINT(
1782 !isDIGIT_LC_utf8((U8*)s),
1789 is_LNBREAK_latin1(s)
1801 !is_VERTWS_latin1(s)
1807 is_HORIZWS_latin1(s)
1812 !is_HORIZWS_utf8(s),
1813 !is_HORIZWS_latin1(s)
1820 /* what trie are we using right now */
1822 = (reg_ac_data*)progi->data->data[ ARG( c ) ];
1824 = (reg_trie_data*)progi->data->data[ aho->trie ];
1825 HV *widecharmap = MUTABLE_HV(progi->data->data[ aho->trie + 1 ]);
1827 const char *last_start = strend - trie->minlen;
1829 const char *real_start = s;
1831 STRLEN maxlen = trie->maxlen;
1833 U8 **points; /* map of where we were in the input string
1834 when reading a given char. For ASCII this
1835 is unnecessary overhead as the relationship
1836 is always 1:1, but for Unicode, especially
1837 case folded Unicode this is not true. */
1838 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1842 GET_RE_DEBUG_FLAGS_DECL;
1844 /* We can't just allocate points here. We need to wrap it in
1845 * an SV so it gets freed properly if there is a croak while
1846 * running the match */
1849 sv_points=newSV(maxlen * sizeof(U8 *));
1850 SvCUR_set(sv_points,
1851 maxlen * sizeof(U8 *));
1852 SvPOK_on(sv_points);
1853 sv_2mortal(sv_points);
1854 points=(U8**)SvPV_nolen(sv_points );
1855 if ( trie_type != trie_utf8_fold
1856 && (trie->bitmap || OP(c)==AHOCORASICKC) )
1859 bitmap=(U8*)trie->bitmap;
1861 bitmap=(U8*)ANYOF_BITMAP(c);
1863 /* this is the Aho-Corasick algorithm modified a touch
1864 to include special handling for long "unknown char"
1865 sequences. The basic idea being that we use AC as long
1866 as we are dealing with a possible matching char, when
1867 we encounter an unknown char (and we have not encountered
1868 an accepting state) we scan forward until we find a legal
1870 AC matching is basically that of trie matching, except
1871 that when we encounter a failing transition, we fall back
1872 to the current states "fail state", and try the current char
1873 again, a process we repeat until we reach the root state,
1874 state 1, or a legal transition. If we fail on the root state
1875 then we can either terminate if we have reached an accepting
1876 state previously, or restart the entire process from the beginning
1880 while (s <= last_start) {
1881 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1889 U8 *uscan = (U8*)NULL;
1890 U8 *leftmost = NULL;
1892 U32 accepted_word= 0;
1896 while ( state && uc <= (U8*)strend ) {
1898 U32 word = aho->states[ state ].wordnum;
1902 DEBUG_TRIE_EXECUTE_r(
1903 if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1904 dump_exec_pos( (char *)uc, c, strend, real_start,
1905 (char *)uc, utf8_target );
1906 PerlIO_printf( Perl_debug_log,
1907 " Scanning for legal start char...\n");
1911 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1915 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
1921 if (uc >(U8*)last_start) break;
1925 U8 *lpos= points[ (pointpos - trie->wordinfo[word].len) % maxlen ];
1926 if (!leftmost || lpos < leftmost) {
1927 DEBUG_r(accepted_word=word);
1933 points[pointpos++ % maxlen]= uc;
1934 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
1935 uscan, len, uvc, charid, foldlen,
1937 DEBUG_TRIE_EXECUTE_r({
1938 dump_exec_pos( (char *)uc, c, strend, real_start,
1940 PerlIO_printf(Perl_debug_log,
1941 " Charid:%3u CP:%4"UVxf" ",
1947 word = aho->states[ state ].wordnum;
1949 base = aho->states[ state ].trans.base;
1951 DEBUG_TRIE_EXECUTE_r({
1953 dump_exec_pos( (char *)uc, c, strend, real_start,
1955 PerlIO_printf( Perl_debug_log,
1956 "%sState: %4"UVxf", word=%"UVxf,
1957 failed ? " Fail transition to " : "",
1958 (UV)state, (UV)word);
1964 ( ((offset = base + charid
1965 - 1 - trie->uniquecharcount)) >= 0)
1966 && ((U32)offset < trie->lasttrans)
1967 && trie->trans[offset].check == state
1968 && (tmp=trie->trans[offset].next))
1970 DEBUG_TRIE_EXECUTE_r(
1971 PerlIO_printf( Perl_debug_log," - legal\n"));
1976 DEBUG_TRIE_EXECUTE_r(
1977 PerlIO_printf( Perl_debug_log," - fail\n"));
1979 state = aho->fail[state];
1983 /* we must be accepting here */
1984 DEBUG_TRIE_EXECUTE_r(
1985 PerlIO_printf( Perl_debug_log," - accepting\n"));
1994 if (!state) state = 1;
1997 if ( aho->states[ state ].wordnum ) {
1998 U8 *lpos = points[ (pointpos - trie->wordinfo[aho->states[ state ].wordnum].len) % maxlen ];
1999 if (!leftmost || lpos < leftmost) {
2000 DEBUG_r(accepted_word=aho->states[ state ].wordnum);
2005 s = (char*)leftmost;
2006 DEBUG_TRIE_EXECUTE_r({
2008 Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n",
2009 (UV)accepted_word, (IV)(s - real_start)
2012 if (!reginfo || regtry(reginfo, &s)) {
2018 DEBUG_TRIE_EXECUTE_r({
2019 PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n");
2022 DEBUG_TRIE_EXECUTE_r(
2023 PerlIO_printf( Perl_debug_log,"No match.\n"));
2032 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
2042 - regexec_flags - match a regexp against a string
2045 Perl_regexec_flags(pTHX_ REGEXP * const rx, char *stringarg, register char *strend,
2046 char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
2047 /* strend: pointer to null at end of string */
2048 /* strbeg: real beginning of string */
2049 /* minend: end of match must be >=minend after stringarg. */
2050 /* data: May be used for some additional optimizations.
2051 Currently its only used, with a U32 cast, for transmitting
2052 the ganch offset when doing a /g match. This will change */
2053 /* nosave: For optimizations. */
2056 struct regexp *const prog = (struct regexp *)SvANY(rx);
2057 /*register*/ char *s;
2058 register regnode *c;
2059 /*register*/ char *startpos = stringarg;
2060 I32 minlen; /* must match at least this many chars */
2061 I32 dontbother = 0; /* how many characters not to try at end */
2062 I32 end_shift = 0; /* Same for the end. */ /* CC */
2063 I32 scream_pos = -1; /* Internal iterator of scream. */
2064 char *scream_olds = NULL;
2065 const bool utf8_target = cBOOL(DO_UTF8(sv));
2067 RXi_GET_DECL(prog,progi);
2068 regmatch_info reginfo; /* create some info to pass to regtry etc */
2069 regexp_paren_pair *swap = NULL;
2070 GET_RE_DEBUG_FLAGS_DECL;
2072 PERL_ARGS_ASSERT_REGEXEC_FLAGS;
2073 PERL_UNUSED_ARG(data);
2075 /* Be paranoid... */
2076 if (prog == NULL || startpos == NULL) {
2077 Perl_croak(aTHX_ "NULL regexp parameter");
2081 multiline = prog->extflags & RXf_PMf_MULTILINE;
2082 reginfo.prog = rx; /* Yes, sorry that this is confusing. */
2084 RX_MATCH_UTF8_set(rx, utf8_target);
2086 debug_start_match(rx, utf8_target, startpos, strend,
2090 minlen = prog->minlen;
2092 if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
2093 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2094 "String too short [regexec_flags]...\n"));
2099 /* Check validity of program. */
2100 if (UCHARAT(progi->program) != REG_MAGIC) {
2101 Perl_croak(aTHX_ "corrupted regexp program");
2105 PL_reg_state.re_state_eval_setup_done = FALSE;
2109 PL_reg_flags |= RF_utf8;
2111 /* Mark beginning of line for ^ and lookbehind. */
2112 reginfo.bol = startpos; /* XXX not used ??? */
2116 /* Mark end of line for $ (and such) */
2119 /* see how far we have to get to not match where we matched before */
2120 reginfo.till = startpos+minend;
2122 /* If there is a "must appear" string, look for it. */
2125 if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */
2127 if (flags & REXEC_IGNOREPOS){ /* Means: check only at start */
2128 reginfo.ganch = startpos + prog->gofs;
2129 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
2130 "GPOS IGNOREPOS: reginfo.ganch = startpos + %"UVxf"\n",(UV)prog->gofs));
2131 } else if (sv && SvTYPE(sv) >= SVt_PVMG
2133 && (mg = mg_find(sv, PERL_MAGIC_regex_global))
2134 && mg->mg_len >= 0) {
2135 reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */
2136 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
2137 "GPOS MAGIC: reginfo.ganch = strbeg + %"IVdf"\n",(IV)mg->mg_len));
2139 if (prog->extflags & RXf_ANCH_GPOS) {
2140 if (s > reginfo.ganch)
2142 s = reginfo.ganch - prog->gofs;
2143 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
2144 "GPOS ANCH_GPOS: s = ganch - %"UVxf"\n",(UV)prog->gofs));
2150 reginfo.ganch = strbeg + PTR2UV(data);
2151 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
2152 "GPOS DATA: reginfo.ganch= strbeg + %"UVxf"\n",PTR2UV(data)));
2154 } else { /* pos() not defined */
2155 reginfo.ganch = strbeg;
2156 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
2157 "GPOS: reginfo.ganch = strbeg\n"));
2160 if (PL_curpm && (PM_GETRE(PL_curpm) == rx)) {
2161 /* We have to be careful. If the previous successful match
2162 was from this regex we don't want a subsequent partially
2163 successful match to clobber the old results.
2164 So when we detect this possibility we add a swap buffer
2165 to the re, and switch the buffer each match. If we fail
2166 we switch it back, otherwise we leave it swapped.
2169 /* do we need a save destructor here for eval dies? */
2170 Newxz(prog->offs, (prog->nparens + 1), regexp_paren_pair);
2171 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
2172 "rex=0x%"UVxf" saving offs: orig=0x%"UVxf" new=0x%"UVxf"\n",
2178 if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) {
2179 re_scream_pos_data d;
2181 d.scream_olds = &scream_olds;
2182 d.scream_pos = &scream_pos;
2183 s = re_intuit_start(rx, sv, s, strend, flags, &d);
2185 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
2186 goto phooey; /* not present */
2192 /* Simplest case: anchored match need be tried only once. */
2193 /* [unless only anchor is BOL and multiline is set] */
2194 if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) {
2195 if (s == startpos && regtry(®info, &startpos))
2197 else if (multiline || (prog->intflags & PREGf_IMPLICIT)
2198 || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */
2203 dontbother = minlen - 1;
2204 end = HOP3c(strend, -dontbother, strbeg) - 1;
2205 /* for multiline we only have to try after newlines */
2206 if (prog->check_substr || prog->check_utf8) {
2207 /* because of the goto we can not easily reuse the macros for bifurcating the
2208 unicode/non-unicode match modes here like we do elsewhere - demerphq */
2211 goto after_try_utf8;
2213 if (regtry(®info, &s)) {
2220 if (prog->extflags & RXf_USE_INTUIT) {
2221 s = re_intuit_start(rx, sv, s + UTF8SKIP(s), strend, flags, NULL);
2230 } /* end search for check string in unicode */
2232 if (s == startpos) {
2233 goto after_try_latin;
2236 if (regtry(®info, &s)) {
2243 if (prog->extflags & RXf_USE_INTUIT) {
2244 s = re_intuit_start(rx, sv, s + 1, strend, flags, NULL);
2253 } /* end search for check string in latin*/
2254 } /* end search for check string */
2255 else { /* search for newline */
2257 /*XXX: The s-- is almost definitely wrong here under unicode - demeprhq*/
2260 /* We can use a more efficient search as newlines are the same in unicode as they are in latin */
2261 while (s <= end) { /* note it could be possible to match at the end of the string */
2262 if (*s++ == '\n') { /* don't need PL_utf8skip here */
2263 if (regtry(®info, &s))
2267 } /* end search for newline */
2268 } /* end anchored/multiline check string search */
2270 } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK))
2272 /* the warning about reginfo.ganch being used without initialization
2273 is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN
2274 and we only enter this block when the same bit is set. */
2275 char *tmp_s = reginfo.ganch - prog->gofs;
2277 if (tmp_s >= strbeg && regtry(®info, &tmp_s))
2282 /* Messy cases: unanchored match. */
2283 if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) {
2284 /* we have /x+whatever/ */
2285 /* it must be a one character string (XXXX Except UTF_PATTERN?) */
2290 if (!(utf8_target ? prog->anchored_utf8 : prog->anchored_substr))
2291 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2292 ch = SvPVX_const(utf8_target ? prog->anchored_utf8 : prog->anchored_substr)[0];
2297 DEBUG_EXECUTE_r( did_match = 1 );
2298 if (regtry(®info, &s)) goto got_it;
2300 while (s < strend && *s == ch)
2308 DEBUG_EXECUTE_r( did_match = 1 );
2309 if (regtry(®info, &s)) goto got_it;
2311 while (s < strend && *s == ch)
2316 DEBUG_EXECUTE_r(if (!did_match)
2317 PerlIO_printf(Perl_debug_log,
2318 "Did not find anchored character...\n")
2321 else if (prog->anchored_substr != NULL
2322 || prog->anchored_utf8 != NULL
2323 || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
2324 && prog->float_max_offset < strend - s)) {
2329 char *last1; /* Last position checked before */
2333 if (prog->anchored_substr || prog->anchored_utf8) {
2334 if (!(utf8_target ? prog->anchored_utf8 : prog->anchored_substr))
2335 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2336 must = utf8_target ? prog->anchored_utf8 : prog->anchored_substr;
2337 back_max = back_min = prog->anchored_offset;
2339 if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
2340 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2341 must = utf8_target ? prog->float_utf8 : prog->float_substr;
2342 back_max = prog->float_max_offset;
2343 back_min = prog->float_min_offset;
2347 if (must == &PL_sv_undef)
2348 /* could not downgrade utf8 check substring, so must fail */
2354 last = HOP3c(strend, /* Cannot start after this */
2355 -(I32)(CHR_SVLEN(must)
2356 - (SvTAIL(must) != 0) + back_min), strbeg);
2359 last1 = HOPc(s, -1);
2361 last1 = s - 1; /* bogus */
2363 /* XXXX check_substr already used to find "s", can optimize if
2364 check_substr==must. */
2366 dontbother = end_shift;
2367 strend = HOPc(strend, -dontbother);
2368 while ( (s <= last) &&
2369 (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)),
2370 (unsigned char*)strend, must,
2371 multiline ? FBMrf_MULTILINE : 0)) ) {
2372 DEBUG_EXECUTE_r( did_match = 1 );
2373 if (HOPc(s, -back_max) > last1) {
2374 last1 = HOPc(s, -back_min);
2375 s = HOPc(s, -back_max);
2378 char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
2380 last1 = HOPc(s, -back_min);
2384 while (s <= last1) {
2385 if (regtry(®info, &s))
2391 while (s <= last1) {
2392 if (regtry(®info, &s))
2398 DEBUG_EXECUTE_r(if (!did_match) {
2399 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
2400 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
2401 PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n",
2402 ((must == prog->anchored_substr || must == prog->anchored_utf8)
2403 ? "anchored" : "floating"),
2404 quoted, RE_SV_TAIL(must));
2408 else if ( (c = progi->regstclass) ) {
2410 const OPCODE op = OP(progi->regstclass);
2411 /* don't bother with what can't match */
2412 if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE)
2413 strend = HOPc(strend, -(minlen - 1));
2416 SV * const prop = sv_newmortal();
2417 regprop(prog, prop, c);
2419 RE_PV_QUOTED_DECL(quoted,utf8_target,PERL_DEBUG_PAD_ZERO(1),
2421 PerlIO_printf(Perl_debug_log,
2422 "Matching stclass %.*s against %s (%d bytes)\n",
2423 (int)SvCUR(prop), SvPVX_const(prop),
2424 quoted, (int)(strend - s));
2427 if (find_byclass(prog, c, s, strend, ®info))
2429 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n"));
2433 if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
2440 if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
2441 utf8_target ? to_utf8_substr(prog) : to_byte_substr(prog);
2442 float_real = utf8_target ? prog->float_utf8 : prog->float_substr;
2444 little = SvPV_const(float_real, len);
2445 if (SvTAIL(float_real)) {
2446 /* This means that float_real contains an artificial \n on the end
2447 * due to the presence of something like this: /foo$/
2448 * where we can match both "foo" and "foo\n" at the end of the string.
2449 * So we have to compare the end of the string first against the float_real
2450 * without the \n and then against the full float_real with the string.
2451 * We have to watch out for cases where the string might be smaller
2452 * than the float_real or the float_real without the \n.
2454 char *checkpos= strend - len;
2456 PerlIO_printf(Perl_debug_log,
2457 "%sChecking for float_real.%s\n",
2458 PL_colors[4], PL_colors[5]));
2459 if (checkpos + 1 < strbeg) {
2460 /* can't match, even if we remove the trailing \n string is too short to match */
2462 PerlIO_printf(Perl_debug_log,
2463 "%sString shorter than required trailing substring, cannot match.%s\n",
2464 PL_colors[4], PL_colors[5]));
2466 } else if (memEQ(checkpos + 1, little, len - 1)) {
2467 /* can match, the end of the string matches without the "\n" */
2468 last = checkpos + 1;
2469 } else if (checkpos < strbeg) {
2470 /* cant match, string is too short when the "\n" is included */
2472 PerlIO_printf(Perl_debug_log,
2473 "%sString does not contain required trailing substring, cannot match.%s\n",
2474 PL_colors[4], PL_colors[5]));
2476 } else if (!multiline) {
2477 /* non multiline match, so compare with the "\n" at the end of the string */
2478 if (memEQ(checkpos, little, len)) {
2482 PerlIO_printf(Perl_debug_log,
2483 "%sString does not contain required trailing substring, cannot match.%s\n",
2484 PL_colors[4], PL_colors[5]));
2488 /* multiline match, so we have to search for a place where the full string is located */
2494 last = rninstr(s, strend, little, little + len);
2496 last = strend; /* matching "$" */
2499 /* at one point this block contained a comment which was probably
2500 * incorrect, which said that this was a "should not happen" case.
2501 * Even if it was true when it was written I am pretty sure it is
2502 * not anymore, so I have removed the comment and replaced it with
2505 PerlIO_printf(Perl_debug_log,
2506 "String does not contain required substring, cannot match.\n"
2510 dontbother = strend - last + prog->float_min_offset;
2512 if (minlen && (dontbother < minlen))
2513 dontbother = minlen - 1;
2514 strend -= dontbother; /* this one's always in bytes! */
2515 /* We don't know much -- general case. */
2518 if (regtry(®info, &s))
2527 if (regtry(®info, &s))
2529 } while (s++ < strend);
2539 PerlIO_printf(Perl_debug_log,
2540 "rex=0x%"UVxf" freeing offs: 0x%"UVxf"\n",
2546 RX_MATCH_TAINTED_set(rx, PL_reg_flags & RF_tainted);
2548 if (PL_reg_state.re_state_eval_setup_done)
2549 restore_pos(aTHX_ prog);
2550 if (RXp_PAREN_NAMES(prog))
2551 (void)hv_iterinit(RXp_PAREN_NAMES(prog));
2553 /* make sure $`, $&, $', and $digit will work later */
2554 if ( !(flags & REXEC_NOT_FIRST) ) {
2555 RX_MATCH_COPY_FREE(rx);
2556 if (flags & REXEC_COPY_STR) {
2557 const I32 i = PL_regeol - startpos + (stringarg - strbeg);
2558 #ifdef PERL_OLD_COPY_ON_WRITE
2560 || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
2562 PerlIO_printf(Perl_debug_log,
2563 "Copy on write: regexp capture, type %d\n",
2566 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
2567 prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
2568 assert (SvPOKp(prog->saved_copy));
2572 RX_MATCH_COPIED_on(rx);
2573 s = savepvn(strbeg, i);
2579 prog->subbeg = strbeg;
2580 prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
2587 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
2588 PL_colors[4], PL_colors[5]));
2589 if (PL_reg_state.re_state_eval_setup_done)
2590 restore_pos(aTHX_ prog);
2592 /* we failed :-( roll it back */
2593 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
2594 "rex=0x%"UVxf" rolling back offs: freeing=0x%"UVxf" restoring=0x%"UVxf"\n",
2599 Safefree(prog->offs);
2607 /* Set which rex is pointed to by PL_reg_state, handling ref counting.
2608 * Do inc before dec, in case old and new rex are the same */
2609 #define SET_reg_curpm(Re2) \
2610 if (PL_reg_state.re_state_eval_setup_done) { \
2611 (void)ReREFCNT_inc(Re2); \
2612 ReREFCNT_dec(PM_GETRE(PL_reg_curpm)); \
2613 PM_SETRE((PL_reg_curpm), (Re2)); \
2618 - regtry - try match at specific point
2620 STATIC I32 /* 0 failure, 1 success */
2621 S_regtry(pTHX_ regmatch_info *reginfo, char **startpos)
2625 REGEXP *const rx = reginfo->prog;
2626 regexp *const prog = (struct regexp *)SvANY(rx);
2627 RXi_GET_DECL(prog,progi);
2628 GET_RE_DEBUG_FLAGS_DECL;
2630 PERL_ARGS_ASSERT_REGTRY;
2632 reginfo->cutpoint=NULL;
2634 if ((prog->extflags & RXf_EVAL_SEEN)
2635 && !PL_reg_state.re_state_eval_setup_done)
2639 PL_reg_state.re_state_eval_setup_done = TRUE;
2641 /* Make $_ available to executed code. */
2642 if (reginfo->sv != DEFSV) {
2644 DEFSV_set(reginfo->sv);
2647 if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv)
2648 && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) {
2649 /* prepare for quick setting of pos */
2650 #ifdef PERL_OLD_COPY_ON_WRITE
2651 if (SvIsCOW(reginfo->sv))
2652 sv_force_normal_flags(reginfo->sv, 0);
2654 mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global,
2655 &PL_vtbl_mglob, NULL, 0);
2659 PL_reg_oldpos = mg->mg_len;
2660 SAVEDESTRUCTOR_X(restore_pos, prog);
2662 if (!PL_reg_curpm) {
2663 Newxz(PL_reg_curpm, 1, PMOP);
2666 SV* const repointer = &PL_sv_undef;
2667 /* this regexp is also owned by the new PL_reg_curpm, which
2668 will try to free it. */
2669 av_push(PL_regex_padav, repointer);
2670 PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav);
2671 PL_regex_pad = AvARRAY(PL_regex_padav);
2676 PL_reg_oldcurpm = PL_curpm;
2677 PL_curpm = PL_reg_curpm;
2678 if (RXp_MATCH_COPIED(prog)) {
2679 /* Here is a serious problem: we cannot rewrite subbeg,
2680 since it may be needed if this match fails. Thus
2681 $` inside (?{}) could fail... */
2682 PL_reg_oldsaved = prog->subbeg;
2683 PL_reg_oldsavedlen = prog->sublen;
2684 #ifdef PERL_OLD_COPY_ON_WRITE
2685 PL_nrs = prog->saved_copy;
2687 RXp_MATCH_COPIED_off(prog);
2690 PL_reg_oldsaved = NULL;
2691 prog->subbeg = PL_bostr;
2692 prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
2695 PL_reg_starttry = *startpos;
2697 prog->offs[0].start = *startpos - PL_bostr;
2698 PL_reginput = *startpos;
2699 prog->lastparen = 0;
2700 prog->lastcloseparen = 0;
2703 /* XXXX What this code is doing here?!!! There should be no need
2704 to do this again and again, prog->lastparen should take care of
2707 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
2708 * Actually, the code in regcppop() (which Ilya may be meaning by
2709 * prog->lastparen), is not needed at all by the test suite
2710 * (op/regexp, op/pat, op/split), but that code is needed otherwise
2711 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
2712 * Meanwhile, this code *is* needed for the
2713 * above-mentioned test suite tests to succeed. The common theme
2714 * on those tests seems to be returning null fields from matches.
2715 * --jhi updated by dapm */
2717 if (prog->nparens) {
2718 regexp_paren_pair *pp = prog->offs;
2720 for (i = prog->nparens; i > (I32)prog->lastparen; i--) {
2728 if (regmatch(reginfo, progi->program + 1)) {
2729 prog->offs[0].end = PL_reginput - PL_bostr;
2732 if (reginfo->cutpoint)
2733 *startpos= reginfo->cutpoint;
2734 REGCP_UNWIND(lastcp);
2739 #define sayYES goto yes
2740 #define sayNO goto no
2741 #define sayNO_SILENT goto no_silent
2743 /* we dont use STMT_START/END here because it leads to
2744 "unreachable code" warnings, which are bogus, but distracting. */
2745 #define CACHEsayNO \
2746 if (ST.cache_mask) \
2747 PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \
2750 /* this is used to determine how far from the left messages like
2751 'failed...' are printed. It should be set such that messages
2752 are inline with the regop output that created them.
2754 #define REPORT_CODE_OFF 32
2757 #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
2758 #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */
2760 #define SLAB_FIRST(s) (&(s)->states[0])
2761 #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
2763 /* grab a new slab and return the first slot in it */
2765 STATIC regmatch_state *
2768 #if PERL_VERSION < 9 && !defined(PERL_CORE)
2771 regmatch_slab *s = PL_regmatch_slab->next;
2773 Newx(s, 1, regmatch_slab);
2774 s->prev = PL_regmatch_slab;
2776 PL_regmatch_slab->next = s;
2778 PL_regmatch_slab = s;
2779 return SLAB_FIRST(s);
2783 /* push a new state then goto it */
2785 #define PUSH_STATE_GOTO(state, node) \
2787 st->resume_state = state; \
2790 /* push a new state with success backtracking, then goto it */
2792 #define PUSH_YES_STATE_GOTO(state, node) \
2794 st->resume_state = state; \
2795 goto push_yes_state;
2801 regmatch() - main matching routine
2803 This is basically one big switch statement in a loop. We execute an op,
2804 set 'next' to point the next op, and continue. If we come to a point which
2805 we may need to backtrack to on failure such as (A|B|C), we push a
2806 backtrack state onto the backtrack stack. On failure, we pop the top
2807 state, and re-enter the loop at the state indicated. If there are no more
2808 states to pop, we return failure.
2810 Sometimes we also need to backtrack on success; for example /A+/, where
2811 after successfully matching one A, we need to go back and try to
2812 match another one; similarly for lookahead assertions: if the assertion
2813 completes successfully, we backtrack to the state just before the assertion
2814 and then carry on. In these cases, the pushed state is marked as
2815 'backtrack on success too'. This marking is in fact done by a chain of
2816 pointers, each pointing to the previous 'yes' state. On success, we pop to
2817 the nearest yes state, discarding any intermediate failure-only states.
2818 Sometimes a yes state is pushed just to force some cleanup code to be
2819 called at the end of a successful match or submatch; e.g. (??{$re}) uses
2820 it to free the inner regex.
2822 Note that failure backtracking rewinds the cursor position, while
2823 success backtracking leaves it alone.
2825 A pattern is complete when the END op is executed, while a subpattern
2826 such as (?=foo) is complete when the SUCCESS op is executed. Both of these
2827 ops trigger the "pop to last yes state if any, otherwise return true"
2830 A common convention in this function is to use A and B to refer to the two
2831 subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is
2832 the subpattern to be matched possibly multiple times, while B is the entire
2833 rest of the pattern. Variable and state names reflect this convention.
2835 The states in the main switch are the union of ops and failure/success of
2836 substates associated with with that op. For example, IFMATCH is the op
2837 that does lookahead assertions /(?=A)B/ and so the IFMATCH state means
2838 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just
2839 successfully matched A and IFMATCH_A_fail is a state saying that we have
2840 just failed to match A. Resume states always come in pairs. The backtrack
2841 state we push is marked as 'IFMATCH_A', but when that is popped, we resume
2842 at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking
2843 on success or failure.
2845 The struct that holds a backtracking state is actually a big union, with
2846 one variant for each major type of op. The variable st points to the
2847 top-most backtrack struct. To make the code clearer, within each
2848 block of code we #define ST to alias the relevant union.
2850 Here's a concrete example of a (vastly oversimplified) IFMATCH
2856 #define ST st->u.ifmatch
2858 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2859 ST.foo = ...; // some state we wish to save
2861 // push a yes backtrack state with a resume value of
2862 // IFMATCH_A/IFMATCH_A_fail, then continue execution at the
2864 PUSH_YES_STATE_GOTO(IFMATCH_A, A);
2867 case IFMATCH_A: // we have successfully executed A; now continue with B
2869 bar = ST.foo; // do something with the preserved value
2872 case IFMATCH_A_fail: // A failed, so the assertion failed
2873 ...; // do some housekeeping, then ...
2874 sayNO; // propagate the failure
2881 For any old-timers reading this who are familiar with the old recursive
2882 approach, the code above is equivalent to:
2884 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
2893 ...; // do some housekeeping, then ...
2894 sayNO; // propagate the failure
2897 The topmost backtrack state, pointed to by st, is usually free. If you
2898 want to claim it, populate any ST.foo fields in it with values you wish to
2899 save, then do one of
2901 PUSH_STATE_GOTO(resume_state, node);
2902 PUSH_YES_STATE_GOTO(resume_state, node);
2904 which sets that backtrack state's resume value to 'resume_state', pushes a
2905 new free entry to the top of the backtrack stack, then goes to 'node'.
2906 On backtracking, the free slot is popped, and the saved state becomes the
2907 new free state. An ST.foo field in this new top state can be temporarily
2908 accessed to retrieve values, but once the main loop is re-entered, it
2909 becomes available for reuse.
2911 Note that the depth of the backtrack stack constantly increases during the
2912 left-to-right execution of the pattern, rather than going up and down with
2913 the pattern nesting. For example the stack is at its maximum at Z at the
2914 end of the pattern, rather than at X in the following:
2916 /(((X)+)+)+....(Y)+....Z/
2918 The only exceptions to this are lookahead/behind assertions and the cut,
2919 (?>A), which pop all the backtrack states associated with A before
2922 Backtrack state structs are allocated in slabs of about 4K in size.
2923 PL_regmatch_state and st always point to the currently active state,
2924 and PL_regmatch_slab points to the slab currently containing
2925 PL_regmatch_state. The first time regmatch() is called, the first slab is
2926 allocated, and is never freed until interpreter destruction. When the slab
2927 is full, a new one is allocated and chained to the end. At exit from
2928 regmatch(), slabs allocated since entry are freed.
2933 #define DEBUG_STATE_pp(pp) \
2935 DUMP_EXEC_POS(locinput, scan, utf8_target); \
2936 PerlIO_printf(Perl_debug_log, \
2937 " %*s"pp" %s%s%s%s%s\n", \
2939 PL_reg_name[st->resume_state], \
2940 ((st==yes_state||st==mark_state) ? "[" : ""), \
2941 ((st==yes_state) ? "Y" : ""), \
2942 ((st==mark_state) ? "M" : ""), \
2943 ((st==yes_state||st==mark_state) ? "]" : "") \
2948 #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1)
2953 S_debug_start_match(pTHX_ const REGEXP *prog, const bool utf8_target,
2954 const char *start, const char *end, const char *blurb)
2956 const bool utf8_pat = RX_UTF8(prog) ? 1 : 0;
2958 PERL_ARGS_ASSERT_DEBUG_START_MATCH;
2963 RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0),
2964 RX_PRECOMP_const(prog), RX_PRELEN(prog), 60);
2966 RE_PV_QUOTED_DECL(s1, utf8_target, PERL_DEBUG_PAD_ZERO(1),
2967 start, end - start, 60);
2969 PerlIO_printf(Perl_debug_log,
2970 "%s%s REx%s %s against %s\n",
2971 PL_colors[4], blurb, PL_colors[5], s0, s1);
2973 if (utf8_target||utf8_pat)
2974 PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n",
2975 utf8_pat ? "pattern" : "",
2976 utf8_pat && utf8_target ? " and " : "",
2977 utf8_target ? "string" : ""
2983 S_dump_exec_pos(pTHX_ const char *locinput,
2984 const regnode *scan,
2985 const char *loc_regeol,
2986 const char *loc_bostr,
2987 const char *loc_reg_starttry,
2988 const bool utf8_target)
2990 const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
2991 const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
2992 int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
2993 /* The part of the string before starttry has one color
2994 (pref0_len chars), between starttry and current
2995 position another one (pref_len - pref0_len chars),
2996 after the current position the third one.
2997 We assume that pref0_len <= pref_len, otherwise we
2998 decrease pref0_len. */
2999 int pref_len = (locinput - loc_bostr) > (5 + taill) - l
3000 ? (5 + taill) - l : locinput - loc_bostr;
3003 PERL_ARGS_ASSERT_DUMP_EXEC_POS;
3005 while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
3007 pref0_len = pref_len - (locinput - loc_reg_starttry);
3008 if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
3009 l = ( loc_regeol - locinput > (5 + taill) - pref_len
3010 ? (5 + taill) - pref_len : loc_regeol - locinput);
3011 while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
3015 if (pref0_len > pref_len)
3016 pref0_len = pref_len;
3018 const int is_uni = (utf8_target && OP(scan) != CANY) ? 1 : 0;
3020 RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
3021 (locinput - pref_len),pref0_len, 60, 4, 5);
3023 RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
3024 (locinput - pref_len + pref0_len),
3025 pref_len - pref0_len, 60, 2, 3);
3027 RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
3028 locinput, loc_regeol - locinput, 10, 0, 1);
3030 const STRLEN tlen=len0+len1+len2;
3031 PerlIO_printf(Perl_debug_log,
3032 "%4"IVdf" <%.*s%.*s%s%.*s>%*s|",
3033 (IV)(locinput - loc_bostr),
3036 (docolor ? "" : "> <"),
3038 (int)(tlen > 19 ? 0 : 19 - tlen),
3045 /* reg_check_named_buff_matched()
3046 * Checks to see if a named buffer has matched. The data array of
3047 * buffer numbers corresponding to the buffer is expected to reside
3048 * in the regexp->data->data array in the slot stored in the ARG() of
3049 * node involved. Note that this routine doesn't actually care about the
3050 * name, that information is not preserved from compilation to execution.
3051 * Returns the index of the leftmost defined buffer with the given name
3052 * or 0 if non of the buffers matched.
3055 S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan)
3058 RXi_GET_DECL(rex,rexi);
3059 SV *sv_dat= MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
3060 I32 *nums=(I32*)SvPVX(sv_dat);
3062 PERL_ARGS_ASSERT_REG_CHECK_NAMED_BUFF_MATCHED;
3064 for ( n=0; n<SvIVX(sv_dat); n++ ) {
3065 if ((I32)rex->lastparen >= nums[n] &&
3066 rex->offs[nums[n]].end != -1)
3075 /* free all slabs above current one - called during LEAVE_SCOPE */
3078 S_clear_backtrack_stack(pTHX_ void *p)
3080 regmatch_slab *s = PL_regmatch_slab->next;
3085 PL_regmatch_slab->next = NULL;
3087 regmatch_slab * const osl = s;
3094 STATIC I32 /* 0 failure, 1 success */
3095 S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
3097 #if PERL_VERSION < 9 && !defined(PERL_CORE)
3101 register const bool utf8_target = PL_reg_match_utf8;
3102 const U32 uniflags = UTF8_ALLOW_DEFAULT;
3103 REGEXP *rex_sv = reginfo->prog;
3104 regexp *rex = (struct regexp *)SvANY(rex_sv);
3105 RXi_GET_DECL(rex,rexi);
3107 /* the current state. This is a cached copy of PL_regmatch_state */
3108 register regmatch_state *st;
3109 /* cache heavy used fields of st in registers */
3110 register regnode *scan;
3111 register regnode *next;
3112 register U32 n = 0; /* general value; init to avoid compiler warning */
3113 register I32 ln = 0; /* len or last; init to avoid compiler warning */
3114 register char *locinput = PL_reginput;
3115 register I32 nextchr; /* is always set to UCHARAT(locinput) */
3117 bool result = 0; /* return value of S_regmatch */
3118 int depth = 0; /* depth of backtrack stack */
3119 U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */
3120 const U32 max_nochange_depth =
3121 (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ?
3122 3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH;
3123 regmatch_state *yes_state = NULL; /* state to pop to on success of
3125 /* mark_state piggy backs on the yes_state logic so that when we unwind
3126 the stack on success we can update the mark_state as we go */
3127 regmatch_state *mark_state = NULL; /* last mark state we have seen */
3128 regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */
3129 struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */
3131 bool no_final = 0; /* prevent failure from backtracking? */
3132 bool do_cutgroup = 0; /* no_final only until next branch/trie entry */
3133 char *startpoint = PL_reginput;
3134 SV *popmark = NULL; /* are we looking for a mark? */
3135 SV *sv_commit = NULL; /* last mark name seen in failure */
3136 SV *sv_yes_mark = NULL; /* last mark name we have seen
3137 during a successful match */
3138 U32 lastopen = 0; /* last open we saw */
3139 bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0;
3140 SV* const oreplsv = GvSV(PL_replgv);
3141 /* these three flags are set by various ops to signal information to
3142 * the very next op. They have a useful lifetime of exactly one loop
3143 * iteration, and are not preserved or restored by state pushes/pops
3145 bool sw = 0; /* the condition value in (?(cond)a|b) */
3146 bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */
3147 int logical = 0; /* the following EVAL is:
3151 or the following IFMATCH/UNLESSM is:
3152 false: plain (?=foo)
3153 true: used as a condition: (?(?=foo))
3155 PAD* last_pad = NULL;
3157 I32 gimme = G_SCALAR;
3158 CV *caller_cv = NULL; /* who called us */
3159 CV *last_pushed_cv = NULL; /* most recently called (?{}) CV */
3160 CHECKPOINT runops_cp; /* savestack position before executing EVAL */
3163 GET_RE_DEBUG_FLAGS_DECL;
3166 /* shut up 'may be used uninitialized' compiler warnings for dMULTICALL */
3167 multicall_oldcatch = 0;
3168 multicall_cv = NULL;
3172 PERL_ARGS_ASSERT_REGMATCH;
3174 DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({
3175 PerlIO_printf(Perl_debug_log,"regmatch start\n");
3177 /* on first ever call to regmatch, allocate first slab */
3178 if (!PL_regmatch_slab) {
3179 Newx(PL_regmatch_slab, 1, regmatch_slab);
3180 PL_regmatch_slab->prev = NULL;
3181 PL_regmatch_slab->next = NULL;
3182 PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
3185 oldsave = PL_savestack_ix;
3186 SAVEDESTRUCTOR_X(S_clear_backtrack_stack, NULL);
3187 SAVEVPTR(PL_regmatch_slab);
3188 SAVEVPTR(PL_regmatch_state);
3190 /* grab next free state slot */
3191 st = ++PL_regmatch_state;
3192 if (st > SLAB_LAST(PL_regmatch_slab))
3193 st = PL_regmatch_state = S_push_slab(aTHX);
3195 /* Note that nextchr is a byte even in UTF */
3196 nextchr = UCHARAT(locinput);
3198 while (scan != NULL) {
3201 SV * const prop = sv_newmortal();
3202 regnode *rnext=regnext(scan);
3203 DUMP_EXEC_POS( locinput, scan, utf8_target );
3204 regprop(rex, prop, scan);
3206 PerlIO_printf(Perl_debug_log,
3207 "%3"IVdf":%*s%s(%"IVdf")\n",
3208 (IV)(scan - rexi->program), depth*2, "",
3210 (PL_regkind[OP(scan)] == END || !rnext) ?
3211 0 : (IV)(rnext - rexi->program));
3214 next = scan + NEXT_OFF(scan);
3217 state_num = OP(scan);
3221 switch (state_num) {
3223 if (locinput == PL_bostr)
3225 /* reginfo->till = reginfo->bol; */
3230 if (locinput == PL_bostr ||
3231 ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n'))
3237 if (locinput == PL_bostr)
3241 if (locinput == reginfo->ganch)
3246 /* update the startpoint */
3247 st->u.keeper.val = rex->offs[0].start;
3248 PL_reginput = locinput;
3249 rex->offs[0].start = locinput - PL_bostr;
3250 PUSH_STATE_GOTO(KEEPS_next, next);
3252 case KEEPS_next_fail:
3253 /* rollback the start point change */
3254 rex->offs[0].start = st->u.keeper.val;
3260 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
3265 if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
3267 if (PL_regeol - locinput > 1)
3271 if (PL_regeol != locinput)
3275 if (!nextchr && locinput >= PL_regeol)
3278 locinput += PL_utf8skip[nextchr];
3279 if (locinput > PL_regeol)
3281 nextchr = UCHARAT(locinput);
3284 nextchr = UCHARAT(++locinput);
3287 if (!nextchr && locinput >= PL_regeol)
3289 nextchr = UCHARAT(++locinput);
3292 if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
3295 locinput += PL_utf8skip[nextchr];
3296 if (locinput > PL_regeol)
3298 nextchr = UCHARAT(locinput);
3301 nextchr = UCHARAT(++locinput);
3305 #define ST st->u.trie
3307 /* In this case the charclass data is available inline so
3308 we can fail fast without a lot of extra overhead.
3310 if(!ANYOF_BITMAP_TEST(scan, *locinput)) {
3312 PerlIO_printf(Perl_debug_log,
3313 "%*s %sfailed to match trie start class...%s\n",
3314 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3317 assert(0); /* NOTREACHED */
3321 /* the basic plan of execution of the trie is:
3322 * At the beginning, run though all the states, and
3323 * find the longest-matching word. Also remember the position
3324 * of the shortest matching word. For example, this pattern:
3327 * when matched against the string "abcde", will generate
3328 * accept states for all words except 3, with the longest
3329 * matching word being 4, and the shortest being 1 (with
3330 * the position being after char 1 of the string).
3332 * Then for each matching word, in word order (i.e. 1,2,4,5),
3333 * we run the remainder of the pattern; on each try setting
3334 * the current position to the character following the word,
3335 * returning to try the next word on failure.
3337 * We avoid having to build a list of words at runtime by
3338 * using a compile-time structure, wordinfo[].prev, which
3339 * gives, for each word, the previous accepting word (if any).
3340 * In the case above it would contain the mappings 1->2, 2->0,
3341 * 3->0, 4->5, 5->1. We can use this table to generate, from
3342 * the longest word (4 above), a list of all words, by
3343 * following the list of prev pointers; this gives us the
3344 * unordered list 4,5,1,2. Then given the current word we have
3345 * just tried, we can go through the list and find the
3346 * next-biggest word to try (so if we just failed on word 2,
3347 * the next in the list is 4).
3349 * Since at runtime we don't record the matching position in
3350 * the string for each word, we have to work that out for
3351 * each word we're about to process. The wordinfo table holds
3352 * the character length of each word; given that we recorded
3353 * at the start: the position of the shortest word and its
3354 * length in chars, we just need to move the pointer the
3355 * difference between the two char lengths. Depending on
3356 * Unicode status and folding, that's cheap or expensive.
3358 * This algorithm is optimised for the case where are only a
3359 * small number of accept states, i.e. 0,1, or maybe 2.
3360 * With lots of accepts states, and having to try all of them,
3361 * it becomes quadratic on number of accept states to find all
3366 /* what type of TRIE am I? (utf8 makes this contextual) */
3367 DECL_TRIE_TYPE(scan);
3369 /* what trie are we using right now */
3370 reg_trie_data * const trie
3371 = (reg_trie_data*)rexi->data->data[ ARG( scan ) ];
3372 HV * widecharmap = MUTABLE_HV(rexi->data->data[ ARG( scan ) + 1 ]);
3373 U32 state = trie->startstate;
3375 if (trie->bitmap && !TRIE_BITMAP_TEST(trie,*locinput) ) {
3376 if (trie->states[ state ].wordnum) {
3378 PerlIO_printf(Perl_debug_log,
3379 "%*s %smatched empty string...%s\n",
3380 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3386 PerlIO_printf(Perl_debug_log,
3387 "%*s %sfailed to match trie start class...%s\n",
3388 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
3395 U8 *uc = ( U8* )locinput;
3399 U8 *uscan = (U8*)NULL;
3400 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
3401 U32 charcount = 0; /* how many input chars we have matched */
3402 U32 accepted = 0; /* have we seen any accepting states? */
3404 ST.jump = trie->jump;
3407 ST.longfold = FALSE; /* char longer if folded => it's harder */
3410 /* fully traverse the TRIE; note the position of the
3411 shortest accept state and the wordnum of the longest
3414 while ( state && uc <= (U8*)PL_regeol ) {
3415 U32 base = trie->states[ state ].trans.base;
3419 wordnum = trie->states[ state ].wordnum;
3421 if (wordnum) { /* it's an accept state */
3424 /* record first match position */
3426 ST.firstpos = (U8*)locinput;
3431 ST.firstchars = charcount;
3434 if (!ST.nextword || wordnum < ST.nextword)
3435 ST.nextword = wordnum;
3436 ST.topword = wordnum;
3439 DEBUG_TRIE_EXECUTE_r({
3440 DUMP_EXEC_POS( (char *)uc, scan, utf8_target );
3441 PerlIO_printf( Perl_debug_log,
3442 "%*s %sState: %4"UVxf" Accepted: %c ",
3443 2+depth * 2, "", PL_colors[4],
3444 (UV)state, (accepted ? 'Y' : 'N'));
3447 /* read a char and goto next state */
3450 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
3451 uscan, len, uvc, charid, foldlen,
3458 base + charid - 1 - trie->uniquecharcount)) >= 0)
3460 && ((U32)offset < trie->lasttrans)
3461 && trie->trans[offset].check == state)
3463 state = trie->trans[offset].next;
3474 DEBUG_TRIE_EXECUTE_r(
3475 PerlIO_printf( Perl_debug_log,
3476 "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n",
3477 charid, uvc, (UV)state, PL_colors[5] );
3483 /* calculate total number of accept states */
3488 w = trie->wordinfo[w].prev;
3491 ST.accepted = accepted;
3495 PerlIO_printf( Perl_debug_log,
3496 "%*s %sgot %"IVdf" possible matches%s\n",
3497 REPORT_CODE_OFF + depth * 2, "",
3498 PL_colors[4], (IV)ST.accepted, PL_colors[5] );
3500 goto trie_first_try; /* jump into the fail handler */
3502 assert(0); /* NOTREACHED */
3504 case TRIE_next_fail: /* we failed - try next alternative */
3506 REGCP_UNWIND(ST.cp);
3507 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
3509 if (!--ST.accepted) {
3511 PerlIO_printf( Perl_debug_log,
3512 "%*s %sTRIE failed...%s\n",
3513 REPORT_CODE_OFF+depth*2, "",
3520 /* Find next-highest word to process. Note that this code
3521 * is O(N^2) per trie run (O(N) per branch), so keep tight */
3522 register U16 min = 0;
3524 register U16 const nextword = ST.nextword;
3525 register reg_trie_wordinfo * const wordinfo
3526 = ((reg_trie_data*)rexi->data->data[ARG(ST.me)])->wordinfo;
3527 for (word=ST.topword; word; word=wordinfo[word].prev) {
3528 if (word > nextword && (!min || word < min))
3541 ST.lastparen = rex->lastparen;
3542 ST.lastcloseparen = rex->lastcloseparen;
3546 /* find start char of end of current word */
3548 U32 chars; /* how many chars to skip */
3549 U8 *uc = ST.firstpos;
3550 reg_trie_data * const trie
3551 = (reg_trie_data*)rexi->data->data[ARG(ST.me)];
3553 assert((trie->wordinfo[ST.nextword].len - trie->prefixlen)
3555 chars = (trie->wordinfo[ST.nextword].len - trie->prefixlen)
3559 /* the hard option - fold each char in turn and find
3560 * its folded length (which may be different */
3561 U8 foldbuf[UTF8_MAXBYTES_CASE + 1];
3569 uvc = utf8n_to_uvuni((U8*)uc, UTF8_MAXLEN, &len,
3577 uvc = to_uni_fold(uvc, foldbuf, &foldlen);
3582 uvc = utf8n_to_uvuni(uscan, UTF8_MAXLEN, &len,
3596 PL_reginput = (char *)uc;
3599 scan = ST.me + ((ST.jump && ST.jump[ST.nextword])
3600 ? ST.jump[ST.nextword]
3604 PerlIO_printf( Perl_debug_log,
3605 "%*s %sTRIE matched word #%d, continuing%s\n",
3606 REPORT_CODE_OFF+depth*2, "",
3613 if (ST.accepted > 1 || has_cutgroup) {
3614 PUSH_STATE_GOTO(TRIE_next, scan);
3615 assert(0); /* NOTREACHED */
3617 /* only one choice left - just continue */
3619 AV *const trie_words
3620 = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]);
3621 SV ** const tmp = av_fetch( trie_words,
3623 SV *sv= tmp ? sv_newmortal() : NULL;
3625 PerlIO_printf( Perl_debug_log,
3626 "%*s %sonly one match left, short-circuiting: #%d <%s>%s\n",
3627 REPORT_CODE_OFF+depth*2, "", PL_colors[4],
3629 tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
3630 PL_colors[0], PL_colors[1],
3631 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)|PERL_PV_ESCAPE_NONASCII
3633 : "not compiled under -Dr",
3637 locinput = PL_reginput;
3638 nextchr = UCHARAT(locinput);
3639 continue; /* execute rest of RE */
3640 assert(0); /* NOTREACHED */
3644 char *s = STRING(scan);
3646 if (utf8_target != UTF_PATTERN) {
3647 /* The target and the pattern have differing utf8ness. */
3649 const char * const e = s + ln;
3652 /* The target is utf8, the pattern is not utf8. */
3657 if (NATIVE_TO_UNI(*(U8*)s) !=
3658 utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
3666 /* The target is not utf8, the pattern is utf8. */
3671 if (NATIVE_TO_UNI(*((U8*)l)) !=
3672 utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
3680 nextchr = UCHARAT(locinput);
3683 /* The target and the pattern have the same utf8ness. */
3684 /* Inline the first character, for speed. */
3685 if (UCHARAT(s) != nextchr)
3687 if (PL_regeol - locinput < ln)
3689 if (ln > 1 && memNE(s, locinput, ln))
3692 nextchr = UCHARAT(locinput);
3697 const U8 * fold_array;
3699 U32 fold_utf8_flags;
3701 PL_reg_flags |= RF_tainted;
3702 folder = foldEQ_locale;
3703 fold_array = PL_fold_locale;
3704 fold_utf8_flags = FOLDEQ_UTF8_LOCALE;
3708 case EXACTFU_TRICKYFOLD:
3710 folder = foldEQ_latin1;
3711 fold_array = PL_fold_latin1;
3712 fold_utf8_flags = (UTF_PATTERN) ? FOLDEQ_S1_ALREADY_FOLDED : 0;
3716 folder = foldEQ_latin1;
3717 fold_array = PL_fold_latin1;
3718 fold_utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
3723 fold_array = PL_fold;
3724 fold_utf8_flags = 0;
3730 if (utf8_target || UTF_PATTERN || state_num == EXACTFU_SS) {
3731 /* Either target or the pattern are utf8, or has the issue where
3732 * the fold lengths may differ. */
3733 const char * const l = locinput;
3734 char *e = PL_regeol;
3736 if (! foldEQ_utf8_flags(s, 0, ln, cBOOL(UTF_PATTERN),
3737 l, &e, 0, utf8_target, fold_utf8_flags))
3742 nextchr = UCHARAT(locinput);
3746 /* Neither the target nor the pattern are utf8 */
3747 if (UCHARAT(s) != nextchr &&
3748 UCHARAT(s) != fold_array[nextchr])
3752 if (PL_regeol - locinput < ln)
3754 if (ln > 1 && ! folder(s, locinput, ln))
3757 nextchr = UCHARAT(locinput);
3761 /* XXX Could improve efficiency by separating these all out using a
3762 * macro or in-line function. At that point regcomp.c would no longer
3763 * have to set the FLAGS fields of these */
3766 PL_reg_flags |= RF_tainted;
3774 /* was last char in word? */
3776 && FLAGS(scan) != REGEX_ASCII_RESTRICTED_CHARSET
3777 && FLAGS(scan) != REGEX_ASCII_MORE_RESTRICTED_CHARSET)
3779 if (locinput == PL_bostr)
3782 const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);
3784 ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags);
3786 if (FLAGS(scan) != REGEX_LOCALE_CHARSET) {
3787 ln = isALNUM_uni(ln);
3788 LOAD_UTF8_CHARCLASS_ALNUM();
3789 n = swash_fetch(PL_utf8_alnum, (U8*)locinput, utf8_target);
3792 ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln));
3793 n = isALNUM_LC_utf8((U8*)locinput);
3798 /* Here the string isn't utf8, or is utf8 and only ascii
3799 * characters are to match \w. In the latter case looking at
3800 * the byte just prior to the current one may be just the final
3801 * byte of a multi-byte character. This is ok. There are two
3803 * 1) it is a single byte character, and then the test is doing
3804 * just what it's supposed to.
3805 * 2) it is a multi-byte character, in which case the final
3806 * byte is never mistakable for ASCII, and so the test
3807 * will say it is not a word character, which is the
3808 * correct answer. */
3809 ln = (locinput != PL_bostr) ?
3810 UCHARAT(locinput - 1) : '\n';
3811 switch (FLAGS(scan)) {
3812 case REGEX_UNICODE_CHARSET:
3813 ln = isWORDCHAR_L1(ln);
3814 n = isWORDCHAR_L1(nextchr);
3816 case REGEX_LOCALE_CHARSET:
3817 ln = isALNUM_LC(ln);
3818 n = isALNUM_LC(nextchr);
3820 case REGEX_DEPENDS_CHARSET:
3822 n = isALNUM(nextchr);
3824 case REGEX_ASCII_RESTRICTED_CHARSET:
3825 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
3826 ln = isWORDCHAR_A(ln);
3827 n = isWORDCHAR_A(nextchr);
3830 Perl_croak(aTHX_ "panic: Unexpected FLAGS %u in op %u", FLAGS(scan), OP(scan));
3834 /* Note requires that all BOUNDs be lower than all NBOUNDs in
3836 if (((!ln) == (!n)) == (OP(scan) < NBOUND))
3841 if (utf8_target || state_num == ANYOFV) {
3842 STRLEN inclasslen = PL_regeol - locinput;
3843 if (locinput >= PL_regeol)
3846 if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, utf8_target))
3848 locinput += inclasslen;
3849 nextchr = UCHARAT(locinput);
3854 nextchr = UCHARAT(locinput);
3855 if (!nextchr && locinput >= PL_regeol)
3857 if (!REGINCLASS(rex, scan, (U8*)locinput))
3859 nextchr = UCHARAT(++locinput);
3863 /* Special char classes - The defines start on line 129 or so */
3864 CCC_TRY_U(ALNUM, NALNUM, isWORDCHAR,
3865 ALNUML, NALNUML, isALNUM_LC, isALNUM_LC_utf8,
3866 ALNUMU, NALNUMU, isWORDCHAR_L1,
3867 ALNUMA, NALNUMA, isWORDCHAR_A,
3870 CCC_TRY_U(SPACE, NSPACE, isSPACE,
3871 SPACEL, NSPACEL, isSPACE_LC, isSPACE_LC_utf8,
3872 SPACEU, NSPACEU, isSPACE_L1,
3873 SPACEA, NSPACEA, isSPACE_A,
3876 CCC_TRY(DIGIT, NDIGIT, isDIGIT,
3877 DIGITL, NDIGITL, isDIGIT_LC, isDIGIT_LC_utf8,
3878 DIGITA, NDIGITA, isDIGIT_A,
3881 case CLUMP: /* Match \X: logical Unicode character. This is defined as
3882 a Unicode extended Grapheme Cluster */
3883 /* From http://www.unicode.org/reports/tr29 (5.2 version). An
3884 extended Grapheme Cluster is:
3887 | Prepend* Begin Extend*
3890 Begin is (Hangul-syllable | ! Control)
3891 Extend is (Grapheme_Extend | Spacing_Mark)
3892 Control is [ GCB_Control CR LF ]
3894 The discussion below shows how the code for CLUMP is derived
3895 from this regex. Note that most of these concepts are from
3896 property values of the Grapheme Cluster Boundary (GCB) property.
3897 No code point can have multiple property values for a given
3898 property. Thus a code point in Prepend can't be in Control, but
3899 it must be in !Control. This is why Control above includes
3900 GCB_Control plus CR plus LF. The latter two are used in the GCB
3901 property separately, and so can't be in GCB_Control, even though
3902 they logically are controls. Control is not the same as gc=cc,
3903 but includes format and other characters as well.
3905 The Unicode definition of Hangul-syllable is:
3907 | (L* ( ( V | LV ) V* | LVT ) T*)
3910 Each of these is a value for the GCB property, and hence must be
3911 disjoint, so the order they are tested is immaterial, so the
3912 above can safely be changed to
3915 | (L* ( LVT | ( V | LV ) V*) T*)
3917 The last two terms can be combined like this:
3919 | (( LVT | ( V | LV ) V*) T*))
3921 And refactored into this:
3922 L* (L | LVT T* | V V* T* | LV V* T*)
3924 That means that if we have seen any L's at all we can quit
3925 there, but if the next character is an LVT, a V, or an LV we
3928 There is a subtlety with Prepend* which showed up in testing.
3929 Note that the Begin, and only the Begin is required in:
3930 | Prepend* Begin Extend*
3931 Also, Begin contains '! Control'. A Prepend must be a
3932 '! Control', which means it must also be a Begin. What it
3933 comes down to is that if we match Prepend* and then find no
3934 suitable Begin afterwards, that if we backtrack the last
3935 Prepend, that one will be a suitable Begin.
3938 if (locinput >= PL_regeol)
3940 if (! utf8_target) {
3942 /* Match either CR LF or '.', as all the other possibilities
3944 locinput++; /* Match the . or CR */
3945 if (nextchr == '\r' /* And if it was CR, and the next is LF,
3947 && locinput < PL_regeol
3948 && UCHARAT(locinput) == '\n') locinput++;
3952 /* Utf8: See if is ( CR LF ); already know that locinput <
3953 * PL_regeol, so locinput+1 is in bounds */
3954 if (nextchr == '\r' && UCHARAT(locinput + 1) == '\n') {
3958 /* In case have to backtrack to beginning, then match '.' */
3959 char *starting = locinput;
3961 /* In case have to backtrack the last prepend */
3962 char *previous_prepend = 0;
3964 LOAD_UTF8_CHARCLASS_GCB();
3966 /* Match (prepend)* */
3967 while (locinput < PL_regeol
3968 && swash_fetch(PL_utf8_X_prepend,
3969 (U8*)locinput, utf8_target))
3971 previous_prepend = locinput;
3972 locinput += UTF8SKIP(locinput);
3975 /* As noted above, if we matched a prepend character, but
3976 * the next thing won't match, back off the last prepend we
3977 * matched, as it is guaranteed to match the begin */
3978 if (previous_prepend
3979 && (locinput >= PL_regeol
3980 || ! swash_fetch(PL_utf8_X_begin,
3981 (U8*)locinput, utf8_target)))
3983 locinput = previous_prepend;
3986 /* Note that here we know PL_regeol > locinput, as we
3987 * tested that upon input to this switch case, and if we
3988 * moved locinput forward, we tested the result just above
3989 * and it either passed, or we backed off so that it will
3991 if (! swash_fetch(PL_utf8_X_begin, (U8*)locinput, utf8_target)) {
3993 /* Here did not match the required 'Begin' in the
3994 * second term. So just match the very first
3995 * character, the '.' of the final term of the regex */
3996 locinput = starting + UTF8SKIP(starting);
3999 /* Here is the beginning of a character that can have
4000 * an extender. It is either a hangul syllable, or a
4002 if (swash_fetch(PL_utf8_X_non_hangul,
4003 (U8*)locinput, utf8_target))
4006 /* Here not a Hangul syllable, must be a
4007 * ('! * Control') */
4008 locinput += UTF8SKIP(locinput);
4011 /* Here is a Hangul syllable. It can be composed
4012 * of several individual characters. One
4013 * possibility is T+ */
4014 if (swash_fetch(PL_utf8_X_T,
4015 (U8*)locinput, utf8_target))
4017 while (locinput < PL_regeol
4018 && swash_fetch(PL_utf8_X_T,
4019 (U8*)locinput, utf8_target))
4021 locinput += UTF8SKIP(locinput);
4025 /* Here, not T+, but is a Hangul. That means
4026 * it is one of the others: L, LV, LVT or V,
4028 * L* (L | LVT T* | V V* T* | LV V* T*) */
4031 while (locinput < PL_regeol
4032 && swash_fetch(PL_utf8_X_L,
4033 (U8*)locinput, utf8_target))
4035 locinput += UTF8SKIP(locinput);
4038 /* Here, have exhausted L*. If the next
4039 * character is not an LV, LVT nor V, it means
4040 * we had to have at least one L, so matches L+
4041 * in the original equation, we have a complete
4042 * hangul syllable. Are done. */
4044 if (locinput < PL_regeol
4045 && swash_fetch(PL_utf8_X_LV_LVT_V,
4046 (U8*)locinput, utf8_target))
4049 /* Otherwise keep going. Must be LV, LVT
4050 * or V. See if LVT */
4051 if (swash_fetch(PL_utf8_X_LVT,
4052 (U8*)locinput, utf8_target))
4054 locinput += UTF8SKIP(locinput);
4057 /* Must be V or LV. Take it, then
4059 locinput += UTF8SKIP(locinput);
4060 while (locinput < PL_regeol
4061 && swash_fetch(PL_utf8_X_V,
4062 (U8*)locinput, utf8_target))
4064 locinput += UTF8SKIP(locinput);
4068 /* And any of LV, LVT, or V can be followed
4070 while (locinput < PL_regeol
4071 && swash_fetch(PL_utf8_X_T,
4075 locinput += UTF8SKIP(locinput);
4081 /* Match any extender */
4082 while (locinput < PL_regeol
4083 && swash_fetch(PL_utf8_X_extend,
4084 (U8*)locinput, utf8_target))
4086 locinput += UTF8SKIP(locinput);
4090 if (locinput > PL_regeol) sayNO;
4092 nextchr = UCHARAT(locinput);
4096 { /* The capture buffer cases. The ones beginning with N for the
4097 named buffers just convert to the equivalent numbered and
4098 pretend they were called as the corresponding numbered buffer
4100 /* don't initialize these in the declaration, it makes C++
4105 const U8 *fold_array;
4108 PL_reg_flags |= RF_tainted;
4109 folder = foldEQ_locale;
4110 fold_array = PL_fold_locale;
4112 utf8_fold_flags = FOLDEQ_UTF8_LOCALE;
4116 folder = foldEQ_latin1;
4117 fold_array = PL_fold_latin1;
4119 utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
4123 folder = foldEQ_latin1;
4124 fold_array = PL_fold_latin1;
4126 utf8_fold_flags = 0;
4131 fold_array = PL_fold;
4133 utf8_fold_flags = 0;
4140 utf8_fold_flags = 0;
4143 /* For the named back references, find the corresponding buffer
4145 n = reg_check_named_buff_matched(rex,scan);
4150 goto do_nref_ref_common;
4153 PL_reg_flags |= RF_tainted;
4154 folder = foldEQ_locale;
4155 fold_array = PL_fold_locale;
4156 utf8_fold_flags = FOLDEQ_UTF8_LOCALE;
4160 folder = foldEQ_latin1;
4161 fold_array = PL_fold_latin1;
4162 utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
4166 folder = foldEQ_latin1;
4167 fold_array = PL_fold_latin1;
4168 utf8_fold_flags = 0;
4173 fold_array = PL_fold;
4174 utf8_fold_flags = 0;
4180 utf8_fold_flags = 0;
4184 n = ARG(scan); /* which paren pair */
4187 ln = rex->offs[n].start;
4188 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
4189 if (rex->lastparen < n || ln == -1)
4190 sayNO; /* Do not match unless seen CLOSEn. */
4191 if (ln == rex->offs[n].end)
4195 if (type != REF /* REF can do byte comparison */
4196 && (utf8_target || type == REFFU))
4197 { /* XXX handle REFFL better */
4198 char * limit = PL_regeol;
4200 /* This call case insensitively compares the entire buffer
4201 * at s, with the current input starting at locinput, but
4202 * not going off the end given by PL_regeol, and returns in
4203 * limit upon success, how much of the current input was
4205 if (! foldEQ_utf8_flags(s, NULL, rex->offs[n].end - ln, utf8_target,
4206 locinput, &limit, 0, utf8_target, utf8_fold_flags))
4211 nextchr = UCHARAT(locinput);
4215 /* Not utf8: Inline the first character, for speed. */
4216 if (UCHARAT(s) != nextchr &&
4218 UCHARAT(s) != fold_array[nextchr]))
4220 ln = rex->offs[n].end - ln;
4221 if (locinput + ln > PL_regeol)
4223 if (ln > 1 && (type == REF
4224 ? memNE(s, locinput, ln)
4225 : ! folder(s, locinput, ln)))
4228 nextchr = UCHARAT(locinput);
4238 #define ST st->u.eval
4243 regexp_internal *rei;
4244 regnode *startpoint;
4247 case GOSUB: /* /(...(?1))/ /(...(?&foo))/ */
4248 if (cur_eval && cur_eval->locinput==locinput) {
4249 if (cur_eval->u.eval.close_paren == (U32)ARG(scan))
4250 Perl_croak(aTHX_ "Infinite recursion in regex");
4251 if ( ++nochange_depth > max_nochange_depth )
4253 "Pattern subroutine nesting without pos change"
4254 " exceeded limit in regex");
4261 if (OP(scan)==GOSUB) {
4262 startpoint = scan + ARG2L(scan);
4263 ST.close_paren = ARG(scan);
4265 startpoint = rei->program+1;
4268 goto eval_recurse_doit;
4269 assert(0); /* NOTREACHED */
4270 case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */
4271 if (cur_eval && cur_eval->locinput==locinput) {
4272 if ( ++nochange_depth > max_nochange_depth )
4273 Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex");
4278 /* execute the code in the {...} */
4282 OP * const oop = PL_op;
4283 COP * const ocurcop = PL_curcop;
4285 char *saved_regeol = PL_regeol;
4286 struct re_save_state saved_state;
4289 /* save *all* paren positions */
4291 REGCP_SET(runops_cp);
4293 /* To not corrupt the existing regex state while executing the
4294 * eval we would normally put it on the save stack, like with
4295 * save_re_context. However, re-evals have a weird scoping so we
4296 * can't just add ENTER/LEAVE here. With that, things like
4298 * (?{$a=2})(a(?{local$a=$a+1}))*aak*c(?{$b=$a})
4300 * would break, as they expect the localisation to be unwound
4301 * only when the re-engine backtracks through the bit that
4304 * What we do instead is just saving the state in a local c
4307 Copy(&PL_reg_state, &saved_state, 1, struct re_save_state);
4309 PL_reg_state.re_reparsing = FALSE;
4312 caller_cv = find_runcv(NULL);
4316 if (rexi->data->what[n] == 'r') { /* code from an external qr */
4317 newcv = ((struct regexp *)SvANY(
4318 (REGEXP*)(rexi->data->data[n])
4321 nop = (OP*)rexi->data->data[n+1];
4323 else if (rexi->data->what[n] == 'l') { /* literal code */
4325 nop = (OP*)rexi->data->data[n];
4326 assert(CvDEPTH(newcv));
4329 /* literal with own CV */
4330 assert(rexi->data->what[n] == 'L');
4331 newcv = rex->qr_anoncv;
4332 nop = (OP*)rexi->data->data[n];
4335 /* normally if we're about to execute code from the same
4336 * CV that we used previously, we just use the existing
4337 * CX stack entry. However, its possible that in the
4338 * meantime we may have backtracked, popped from the save
4339 * stack, and undone the SAVECOMPPAD(s) associated with
4340 * PUSH_MULTICALL; in which case PL_comppad no longer
4341 * points to newcv's pad. */
4342 if (newcv != last_pushed_cv || PL_comppad != last_pad)
4344 I32 depth = (newcv == caller_cv) ? 0 : 1;
4345 if (last_pushed_cv) {
4346 CHANGE_MULTICALL_WITHDEPTH(newcv, depth);
4349 PUSH_MULTICALL_WITHDEPTH(newcv, depth);
4351 last_pushed_cv = newcv;
4353 last_pad = PL_comppad;
4355 /* the initial nextstate you would normally execute
4356 * at the start of an eval (which would cause error
4357 * messages to come from the eval), may be optimised
4358 * away from the execution path in the regex code blocks;
4359 * so manually set PL_curcop to it initially */
4361 OP *o = cUNOPx(nop)->op_first;
4362 assert(o->op_type == OP_NULL);
4363 if (o->op_targ == OP_SCOPE) {
4364 o = cUNOPo->op_first;
4367 assert(o->op_targ == OP_LEAVE);
4368 o = cUNOPo->op_first;
4369 assert(o->op_type == OP_ENTER);
4373 if (o->op_type != OP_STUB) {
4374 assert( o->op_type == OP_NEXTSTATE
4375 || o->op_type == OP_DBSTATE
4376 || (o->op_type == OP_NULL
4377 && ( o->op_targ == OP_NEXTSTATE
4378 || o->op_targ == OP_DBSTATE
4382 PL_curcop = (COP*)o;
4387 DEBUG_STATE_r( PerlIO_printf(Perl_debug_log,
4388 " re EVAL PL_op=0x%"UVxf"\n", PTR2UV(nop)) );
4390 rex->offs[0].end = PL_reg_magic->mg_len = locinput - PL_bostr;
4393 SV *sv_mrk = get_sv("REGMARK", 1);
4394 sv_setsv(sv_mrk, sv_yes_mark);
4397 /* we don't use MULTICALL here as we want to call the
4398 * first op of the block of interest, rather than the
4399 * first op of the sub */
4402 CALLRUNOPS(aTHX); /* Scalar context. */
4405 ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */
4411 /* before restoring everything, evaluate the returned
4412 * value, so that 'uninit' warnings don't use the wrong
4413 * PL_op or pad. Also need to process any magic vars
4414 * (e.g. $1) *before* parentheses are restored */
4419 if (logical == 0) /* (?{})/ */
4420 sv_setsv(save_scalar(PL_replgv), ret); /* $^R */
4421 else if (logical == 1) { /* /(?(?{...})X|Y)/ */
4422 sw = cBOOL(SvTRUE(ret));
4425 else { /* /(??{}) */
4426 /* if its overloaded, let the regex compiler handle
4427 * it; otherwise extract regex, or stringify */
4428 if (!SvAMAGIC(ret)) {
4432 if (SvTYPE(sv) == SVt_REGEXP)
4433 re_sv = (REGEXP*) sv;
4434 else if (SvSMAGICAL(sv)) {
4435 MAGIC *mg = mg_find(sv, PERL_MAGIC_qr);
4437 re_sv = (REGEXP *) mg->mg_obj;
4440 /* force any magic, undef warnings here */
4442 ret = sv_mortalcopy(ret);
4443 (void) SvPV_force_nolen(ret);
4449 Copy(&saved_state, &PL_reg_state, 1, struct re_save_state);
4451 /* *** Note that at this point we don't restore
4452 * PL_comppad, (or pop the CxSUB) on the assumption it may
4453 * be used again soon. This is safe as long as nothing
4454 * in the regexp code uses the pad ! */
4456 PL_curcop = ocurcop;
4457 PL_regeol = saved_regeol;
4458 S_regcp_restore(aTHX_ rex, runops_cp);
4464 /* only /(??{})/ from now on */
4467 /* extract RE object from returned value; compiling if
4471 re_sv = reg_temp_copy(NULL, re_sv);
4475 const I32 osize = PL_regsize;
4477 if (SvUTF8(ret) && IN_BYTES) {
4478 /* In use 'bytes': make a copy of the octet
4479 * sequence, but without the flag on */
4481 const char *const p = SvPV(ret, len);
4482 ret = newSVpvn_flags(p, len, SVs_TEMP);
4484 if (rex->intflags & PREGf_USE_RE_EVAL)
4485 pm_flags |= PMf_USE_RE_EVAL;
4487 /* if we got here, it should be an engine which
4488 * supports compiling code blocks and stuff */
4489 assert(rex->engine && rex->engine->op_comp);
4490 assert(!(scan->flags & ~RXf_PMf_COMPILETIME));
4491 re_sv = rex->engine->op_comp(aTHX_ &ret, 1, NULL,
4492 rex->engine, NULL, NULL,
4493 /* copy /msix etc to inner pattern */
4498 & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
4500 /* This isn't a first class regexp. Instead, it's
4501 caching a regexp onto an existing, Perl visible
4503 sv_magic(ret, MUTABLE_SV(re_sv), PERL_MAGIC_qr, 0, 0);
4506 /* safe to do now that any $1 etc has been
4507 * interpolated into the new pattern string and
4509 S_regcp_restore(aTHX_ rex, runops_cp);
4511 re = (struct regexp *)SvANY(re_sv);
4513 RXp_MATCH_COPIED_off(re);
4514 re->subbeg = rex->subbeg;
4515 re->sublen = rex->sublen;
4518 debug_start_match(re_sv, utf8_target, locinput, PL_regeol,
4519 "Matching embedded");
4521 startpoint = rei->program + 1;
4522 ST.close_paren = 0; /* only used for GOSUB */
4524 eval_recurse_doit: /* Share code with GOSUB below this line */
4525 /* run the pattern returned from (??{...}) */
4526 ST.cp = regcppush(rex, 0); /* Save *all* the positions. */
4527 REGCP_SET(ST.lastcp);
4530 re->lastcloseparen = 0;
4532 PL_reginput = locinput;
4535 /* XXXX This is too dramatic a measure... */
4538 ST.toggle_reg_flags = PL_reg_flags;
4540 PL_reg_flags |= RF_utf8;
4542 PL_reg_flags &= ~RF_utf8;
4543 ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */
4545 ST.prev_rex = rex_sv;
4546 ST.prev_curlyx = cur_curlyx;
4548 SET_reg_curpm(rex_sv);
4553 ST.prev_eval = cur_eval;
4555 /* now continue from first node in postoned RE */
4556 PUSH_YES_STATE_GOTO(EVAL_AB, startpoint);
4557 assert(0); /* NOTREACHED */
4560 case EVAL_AB: /* cleanup after a successful (??{A})B */
4561 /* note: this is called twice; first after popping B, then A */
4562 PL_reg_flags ^= ST.toggle_reg_flags;
4563 rex_sv = ST.prev_rex;
4564 SET_reg_curpm(rex_sv);
4565 rex = (struct regexp *)SvANY(rex_sv);
4566 rexi = RXi_GET(rex);
4568 cur_eval = ST.prev_eval;
4569 cur_curlyx = ST.prev_curlyx;
4571 /* XXXX This is too dramatic a measure... */
4573 if ( nochange_depth )
4578 case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */
4579 /* note: this is called twice; first after popping B, then A */
4580 PL_reg_flags ^= ST.toggle_reg_flags;
4581 rex_sv = ST.prev_rex;
4582 SET_reg_curpm(rex_sv);
4583 rex = (struct regexp *)SvANY(rex_sv);
4584 rexi = RXi_GET(rex);
4586 PL_reginput = locinput;
4587 REGCP_UNWIND(ST.lastcp);
4589 cur_eval = ST.prev_eval;
4590 cur_curlyx = ST.prev_curlyx;
4591 /* XXXX This is too dramatic a measure... */
4593 if ( nochange_depth )
4599 n = ARG(scan); /* which paren pair */
4600 rex->offs[n].start_tmp = locinput - PL_bostr;
4603 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
4604 "rex=0x%"UVxf" offs=0x%"UVxf": \\%"UVuf": set %"IVdf" tmp; regsize=%"UVuf"\n",
4608 (IV)rex->offs[n].start_tmp,
4614 /* XXX really need to log other places start/end are set too */
4615 #define CLOSE_CAPTURE \
4616 rex->offs[n].start = rex->offs[n].start_tmp; \
4617 rex->offs[n].end = locinput - PL_bostr; \
4618 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log, \
4619 "rex=0x%"UVxf" offs=0x%"UVxf": \\%"UVuf": set %"IVdf"..%"IVdf"\n", \
4621 PTR2UV(rex->offs), \
4623 (IV)rex->offs[n].start, \
4624 (IV)rex->offs[n].end \
4628 n = ARG(scan); /* which paren pair */
4630 /*if (n > PL_regsize)
4632 if (n > rex->lastparen)
4634 rex->lastcloseparen = n;
4635 if (cur_eval && cur_eval->u.eval.close_paren == n) {
4643 cursor && OP(cursor)!=END;
4644 cursor=regnext(cursor))
4646 if ( OP(cursor)==CLOSE ){
4648 if ( n <= lastopen ) {
4650 /*if (n > PL_regsize)
4652 if (n > rex->lastparen)
4654 rex->lastcloseparen = n;
4655 if ( n == ARG(scan) || (cur_eval &&
4656 cur_eval->u.eval.close_paren == n))
4665 n = ARG(scan); /* which paren pair */
4666 sw = cBOOL(rex->lastparen >= n && rex->offs[n].end != -1);
4669 /* reg_check_named_buff_matched returns 0 for no match */
4670 sw = cBOOL(0 < reg_check_named_buff_matched(rex,scan));
4674 sw = (cur_eval && (!n || cur_eval->u.eval.close_paren == n));
4680 PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
4682 next = NEXTOPER(NEXTOPER(scan));
4684 next = scan + ARG(scan);
4685 if (OP(next) == IFTHEN) /* Fake one. */
4686 next = NEXTOPER(NEXTOPER(next));
4690 logical = scan->flags;
4693 /*******************************************************************
4695 The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/
4696 pattern, where A and B are subpatterns. (For simple A, CURLYM or
4697 STAR/PLUS/CURLY/CURLYN are used instead.)
4699 A*B is compiled as <CURLYX><A><WHILEM><B>
4701 On entry to the subpattern, CURLYX is called. This pushes a CURLYX
4702 state, which contains the current count, initialised to -1. It also sets
4703 cur_curlyx to point to this state, with any previous value saved in the
4706 CURLYX then jumps straight to the WHILEM op, rather than executing A,
4707 since the pattern may possibly match zero times (i.e. it's a while {} loop
4708 rather than a do {} while loop).
4710 Each entry to WHILEM represents a successful match of A. The count in the
4711 CURLYX block is incremented, another WHILEM state is pushed, and execution
4712 passes to A or B depending on greediness and the current count.
4714 For example, if matching against the string a1a2a3b (where the aN are
4715 substrings that match /A/), then the match progresses as follows: (the
4716 pushed states are interspersed with the bits of strings matched so far):
4719 <CURLYX cnt=0><WHILEM>
4720 <CURLYX cnt=1><WHILEM> a1 <WHILEM>
4721 <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM>
4722 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM>
4723 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b
4725 (Contrast this with something like CURLYM, which maintains only a single
4729 a1 <CURLYM cnt=1> a2
4730 a1 a2 <CURLYM cnt=2> a3
4731 a1 a2 a3 <CURLYM cnt=3> b
4734 Each WHILEM state block marks a point to backtrack to upon partial failure
4735 of A or B, and also contains some minor state data related to that
4736 iteration. The CURLYX block, pointed to by cur_curlyx, contains the
4737 overall state, such as the count, and pointers to the A and B ops.
4739 This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx
4740 must always point to the *current* CURLYX block, the rules are:
4742 When executing CURLYX, save the old cur_curlyx in the CURLYX state block,
4743 and set cur_curlyx to point the new block.
4745 When popping the CURLYX block after a successful or unsuccessful match,
4746 restore the previous cur_curlyx.
4748 When WHILEM is about to execute B, save the current cur_curlyx, and set it
4749 to the outer one saved in the CURLYX block.
4751 When popping the WHILEM block after a successful or unsuccessful B match,
4752 restore the previous cur_curlyx.
4754 Here's an example for the pattern (AI* BI)*BO
4755 I and O refer to inner and outer, C and W refer to CURLYX and WHILEM:
4758 curlyx backtrack stack
4759 ------ ---------------
4761 CO <CO prev=NULL> <WO>
4762 CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
4763 CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
4764 NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo
4766 At this point the pattern succeeds, and we work back down the stack to
4767 clean up, restoring as we go:
4769 CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
4770 CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
4771 CO <CO prev=NULL> <WO>
4774 *******************************************************************/
4776 #define ST st->u.curlyx
4778 case CURLYX: /* start of /A*B/ (for complex A) */
4780 /* No need to save/restore up to this paren */
4781 I32 parenfloor = scan->flags;
4783 assert(next); /* keep Coverity happy */
4784 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
4787 /* XXXX Probably it is better to teach regpush to support
4788 parenfloor > PL_regsize... */
4789 if (parenfloor > (I32)rex->lastparen)
4790 parenfloor = rex->lastparen; /* Pessimization... */
4792 ST.prev_curlyx= cur_curlyx;
4794 ST.cp = PL_savestack_ix;
4796 /* these fields contain the state of the current curly.
4797 * they are accessed by subsequent WHILEMs */
4798 ST.parenfloor = parenfloor;
4803 ST.count = -1; /* this will be updated by WHILEM */
4804 ST.lastloc = NULL; /* this will be updated by WHILEM */
4806 PL_reginput = locinput;
4807 PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next));
4808 assert(0); /* NOTREACHED */
4811 case CURLYX_end: /* just finished matching all of A*B */
4812 cur_curlyx = ST.prev_curlyx;
4814 assert(0); /* NOTREACHED */
4816 case CURLYX_end_fail: /* just failed to match all of A*B */
4818 cur_curlyx = ST.prev_curlyx;
4820 assert(0); /* NOTREACHED */
4824 #define ST st->u.whilem
4826 case WHILEM: /* just matched an A in /A*B/ (for complex A) */
4828 /* see the discussion above about CURLYX/WHILEM */
4830 int min = ARG1(cur_curlyx->u.curlyx.me);
4831 int max = ARG2(cur_curlyx->u.curlyx.me);
4832 regnode *A = NEXTOPER(cur_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS;
4834 assert(cur_curlyx); /* keep Coverity happy */
4835 n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */
4836 ST.save_lastloc = cur_curlyx->u.curlyx.lastloc;
4837 ST.cache_offset = 0;
4840 PL_reginput = locinput;
4842 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4843 "%*s whilem: matched %ld out of %d..%d\n",
4844 REPORT_CODE_OFF+depth*2, "", (long)n, min, max)
4847 /* First just match a string of min A's. */
4850 ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor);
4851 cur_curlyx->u.curlyx.lastloc = locinput;
4852 REGCP_SET(ST.lastcp);
4854 PUSH_STATE_GOTO(WHILEM_A_pre, A);
4855 assert(0); /* NOTREACHED */
4858 /* If degenerate A matches "", assume A done. */
4860 if (locinput == cur_curlyx->u.curlyx.lastloc) {
4861 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4862 "%*s whilem: empty match detected, trying continuation...\n",
4863 REPORT_CODE_OFF+depth*2, "")
4865 goto do_whilem_B_max;
4868 /* super-linear cache processing */
4872 if (!PL_reg_maxiter) {
4873 /* start the countdown: Postpone detection until we
4874 * know the match is not *that* much linear. */
4875 PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4);
4876 /* possible overflow for long strings and many CURLYX's */
4877 if (PL_reg_maxiter < 0)
4878 PL_reg_maxiter = I32_MAX;
4879 PL_reg_leftiter = PL_reg_maxiter;
4882 if (PL_reg_leftiter-- == 0) {
4883 /* initialise cache */
4884 const I32 size = (PL_reg_maxiter + 7)/8;
4885 if (PL_reg_poscache) {
4886 if ((I32)PL_reg_poscache_size < size) {
4887 Renew(PL_reg_poscache, size, char);
4888 PL_reg_poscache_size = size;
4890 Zero(PL_reg_poscache, size, char);
4893 PL_reg_poscache_size = size;
4894 Newxz(PL_reg_poscache, size, char);
4896 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4897 "%swhilem: Detected a super-linear match, switching on caching%s...\n",
4898 PL_colors[4], PL_colors[5])
4902 if (PL_reg_leftiter < 0) {
4903 /* have we already failed at this position? */
4905 offset = (scan->flags & 0xf) - 1
4906 + (locinput - PL_bostr) * (scan->flags>>4);
4907 mask = 1 << (offset % 8);
4909 if (PL_reg_poscache[offset] & mask) {
4910 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
4911 "%*s whilem: (cache) already tried at this position...\n",
4912 REPORT_CODE_OFF+depth*2, "")
4914 sayNO; /* cache records failure */
4916 ST.cache_offset = offset;
4917 ST.cache_mask = mask;
4921 /* Prefer B over A for minimal matching. */
4923 if (cur_curlyx->u.curlyx.minmod) {
4924 ST.save_curlyx = cur_curlyx;
4925 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
4926 ST.cp = regcppush(rex, ST.save_curlyx->u.curlyx.parenfloor);
4927 REGCP_SET(ST.lastcp);
4928 PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B);
4929 assert(0); /* NOTREACHED */
4932 /* Prefer A over B for maximal matching. */
4934 if (n < max) { /* More greed allowed? */
4935 ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor);
4936 cur_curlyx->u.curlyx.lastloc = locinput;
4937 REGCP_SET(ST.lastcp);
4938 PUSH_STATE_GOTO(WHILEM_A_max, A);
4939 assert(0); /* NOTREACHED */
4941 goto do_whilem_B_max;
4943 assert(0); /* NOTREACHED */
4945 case WHILEM_B_min: /* just matched B in a minimal match */
4946 case WHILEM_B_max: /* just matched B in a maximal match */
4947 cur_curlyx = ST.save_curlyx;
4949 assert(0); /* NOTREACHED */
4951 case WHILEM_B_max_fail: /* just failed to match B in a maximal match */
4952 cur_curlyx = ST.save_curlyx;
4953 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
4954 cur_curlyx->u.curlyx.count--;
4956 assert(0); /* NOTREACHED */
4958 case WHILEM_A_min_fail: /* just failed to match A in a minimal match */
4960 case WHILEM_A_pre_fail: /* just failed to match even minimal A */
4961 REGCP_UNWIND(ST.lastcp);
4963 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
4964 cur_curlyx->u.curlyx.count--;
4966 assert(0); /* NOTREACHED */
4968 case WHILEM_A_max_fail: /* just failed to match A in a maximal match */
4969 REGCP_UNWIND(ST.lastcp);
4970 regcppop(rex); /* Restore some previous $<digit>s? */
4971 PL_reginput = locinput;
4972 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
4973 "%*s whilem: failed, trying continuation...\n",
4974 REPORT_CODE_OFF+depth*2, "")
4977 if (cur_curlyx->u.curlyx.count >= REG_INFTY
4978 && ckWARN(WARN_REGEXP)
4979 && !(PL_reg_flags & RF_warned))
4981 PL_reg_flags |= RF_warned;
4982 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
4983 "Complex regular subexpression recursion limit (%d) "
4989 ST.save_curlyx = cur_curlyx;
4990 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
4991 PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B);
4992 assert(0); /* NOTREACHED */
4994 case WHILEM_B_min_fail: /* just failed to match B in a minimal match */
4995 cur_curlyx = ST.save_curlyx;
4996 REGCP_UNWIND(ST.lastcp);
4999 if (cur_curlyx->u.curlyx.count >= /*max*/ARG2(cur_curlyx->u.curlyx.me)) {
5000 /* Maximum greed exceeded */
5001 if (cur_curlyx->u.curlyx.count >= REG_INFTY
5002 && ckWARN(WARN_REGEXP)
5003 && !(PL_reg_flags & RF_warned))
5005 PL_reg_flags |= RF_warned;
5006 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
5007 "Complex regular subexpression recursion "
5008 "limit (%d) exceeded",
5011 cur_curlyx->u.curlyx.count--;
5015 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
5016 "%*s trying longer...\n", REPORT_CODE_OFF+depth*2, "")
5018 /* Try grabbing another A and see if it helps. */
5019 PL_reginput = locinput;
5020 cur_curlyx->u.curlyx.lastloc = locinput;
5021 ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor);
5022 REGCP_SET(ST.lastcp);
5023 PUSH_STATE_GOTO(WHILEM_A_min,
5024 /*A*/ NEXTOPER(ST.save_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS);
5025 assert(0); /* NOTREACHED */
5028 #define ST st->u.branch
5030 case BRANCHJ: /* /(...|A|...)/ with long next pointer */
5031 next = scan + ARG(scan);
5034 scan = NEXTOPER(scan);
5037 case BRANCH: /* /(...|A|...)/ */
5038 scan = NEXTOPER(scan); /* scan now points to inner node */
5039 ST.lastparen = rex->lastparen;
5040 ST.lastcloseparen = rex->lastcloseparen;
5041 ST.next_branch = next;
5043 PL_reginput = locinput;
5045 /* Now go into the branch */
5047 PUSH_YES_STATE_GOTO(BRANCH_next, scan);
5049 PUSH_STATE_GOTO(BRANCH_next, scan);
5051 assert(0); /* NOTREACHED */
5053 PL_reginput = locinput;
5054 sv_yes_mark = st->u.mark.mark_name = scan->flags ? NULL :
5055 MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
5056 PUSH_STATE_GOTO(CUTGROUP_next,next);
5057 assert(0); /* NOTREACHED */
5058 case CUTGROUP_next_fail:
5061 if (st->u.mark.mark_name)
5062 sv_commit = st->u.mark.mark_name;
5064 assert(0); /* NOTREACHED */
5067 assert(0); /* NOTREACHED */
5068 case BRANCH_next_fail: /* that branch failed; try the next, if any */
5073 REGCP_UNWIND(ST.cp);
5074 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
5075 scan = ST.next_branch;
5076 /* no more branches? */
5077 if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) {
5079 PerlIO_printf( Perl_debug_log,
5080 "%*s %sBRANCH failed...%s\n",
5081 REPORT_CODE_OFF+depth*2, "",
5087 continue; /* execute next BRANCH[J] op */
5088 assert(0); /* NOTREACHED */
5095 #define ST st->u.curlym
5097 case CURLYM: /* /A{m,n}B/ where A is fixed-length */
5099 /* This is an optimisation of CURLYX that enables us to push
5100 * only a single backtracking state, no matter how many matches
5101 * there are in {m,n}. It relies on the pattern being constant
5102 * length, with no parens to influence future backrefs
5106 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
5108 ST.lastparen = rex->lastparen;
5109 ST.lastcloseparen = rex->lastcloseparen;
5111 /* if paren positive, emulate an OPEN/CLOSE around A */
5113 U32 paren = ST.me->flags;
5114 if (paren > PL_regsize)
5116 scan += NEXT_OFF(scan); /* Skip former OPEN. */
5124 ST.c1 = CHRTEST_UNINIT;
5127 if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */
5130 curlym_do_A: /* execute the A in /A{m,n}B/ */
5131 PL_reginput = locinput;
5132 PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */
5133 assert(0); /* NOTREACHED */
5135 case CURLYM_A: /* we've just matched an A */
5136 locinput = st->locinput;
5137 nextchr = UCHARAT(locinput);
5140 /* after first match, determine A's length: u.curlym.alen */
5141 if (ST.count == 1) {
5142 if (PL_reg_match_utf8) {
5144 while (s < PL_reginput) {
5150 ST.alen = PL_reginput - locinput;
5153 ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me);
5156 PerlIO_printf(Perl_debug_log,
5157 "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n",
5158 (int)(REPORT_CODE_OFF+(depth*2)), "",
5159 (IV) ST.count, (IV)ST.alen)
5162 locinput = PL_reginput;
5164 if (cur_eval && cur_eval->u.eval.close_paren &&
5165 cur_eval->u.eval.close_paren == (U32)ST.me->flags)
5169 I32 max = (ST.minmod ? ARG1(ST.me) : ARG2(ST.me));
5170 if ( max == REG_INFTY || ST.count < max )
5171 goto curlym_do_A; /* try to match another A */
5173 goto curlym_do_B; /* try to match B */
5175 case CURLYM_A_fail: /* just failed to match an A */
5176 REGCP_UNWIND(ST.cp);
5178 if (ST.minmod || ST.count < ARG1(ST.me) /* min*/
5179 || (cur_eval && cur_eval->u.eval.close_paren &&
5180 cur_eval->u.eval.close_paren == (U32)ST.me->flags))
5183 curlym_do_B: /* execute the B in /A{m,n}B/ */
5184 PL_reginput = locinput;
5185 if (ST.c1 == CHRTEST_UNINIT) {
5186 /* calculate c1 and c2 for possible match of 1st char
5187 * following curly */
5188 ST.c1 = ST.c2 = CHRTEST_VOID;
5189 if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) {
5190 regnode *text_node = ST.B;
5191 if (! HAS_TEXT(text_node))
5192 FIND_NEXT_IMPT(text_node);
5195 (HAS_TEXT(text_node) && PL_regkind[OP(text_node)] == EXACT)
5197 But the former is redundant in light of the latter.
5199 if this changes back then the macro for
5200 IS_TEXT and friends need to change.
5202 if (PL_regkind[OP(text_node)] == EXACT)
5205 ST.c1 = (U8)*STRING(text_node);
5206 switch (OP(text_node)) {
5207 case EXACTF: ST.c2 = PL_fold[ST.c1]; break;
5210 case EXACTFU_TRICKYFOLD:
5211 case EXACTFU: ST.c2 = PL_fold_latin1[ST.c1]; break;
5212 case EXACTFL: ST.c2 = PL_fold_locale[ST.c1]; break;
5213 default: ST.c2 = ST.c1;
5220 PerlIO_printf(Perl_debug_log,
5221 "%*s CURLYM trying tail with matches=%"IVdf"...\n",
5222 (int)(REPORT_CODE_OFF+(depth*2)),
5225 if (ST.c1 != CHRTEST_VOID
5226 && UCHARAT(PL_reginput) != ST.c1
5227 && UCHARAT(PL_reginput) != ST.c2)
5229 /* simulate B failing */
5231 PerlIO_printf(Perl_debug_log,
5232 "%*s CURLYM Fast bail c1=%"IVdf" c2=%"IVdf"\n",
5233 (int)(REPORT_CODE_OFF+(depth*2)),"",
5236 state_num = CURLYM_B_fail;
5237 goto reenter_switch;
5241 /* emulate CLOSE: mark current A as captured */
5242 I32 paren = ST.me->flags;
5244 rex->offs[paren].start
5245 = HOPc(PL_reginput, -ST.alen) - PL_bostr;
5246 rex->offs[paren].end = PL_reginput - PL_bostr;
5247 if ((U32)paren > rex->lastparen)
5248 rex->lastparen = paren;
5249 rex->lastcloseparen = paren;
5252 rex->offs[paren].end = -1;
5253 if (cur_eval && cur_eval->u.eval.close_paren &&
5254 cur_eval->u.eval.close_paren == (U32)ST.me->flags)
5263 PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */
5264 assert(0); /* NOTREACHED */
5266 case CURLYM_B_fail: /* just failed to match a B */
5267 REGCP_UNWIND(ST.cp);
5268 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
5270 I32 max = ARG2(ST.me);
5271 if (max != REG_INFTY && ST.count == max)
5273 goto curlym_do_A; /* try to match a further A */
5275 /* backtrack one A */
5276 if (ST.count == ARG1(ST.me) /* min */)
5279 locinput = HOPc(locinput, -ST.alen);
5280 goto curlym_do_B; /* try to match B */
5283 #define ST st->u.curly
5285 #define CURLY_SETPAREN(paren, success) \
5288 rex->offs[paren].start = HOPc(locinput, -1) - PL_bostr; \
5289 rex->offs[paren].end = locinput - PL_bostr; \
5290 if (paren > rex->lastparen) \
5291 rex->lastparen = paren; \
5292 rex->lastcloseparen = paren; \
5295 rex->offs[paren].end = -1; \
5296 rex->lastparen = ST.lastparen; \
5297 rex->lastcloseparen = ST.lastcloseparen; \
5301 case STAR: /* /A*B/ where A is width 1 */
5305 scan = NEXTOPER(scan);
5307 case PLUS: /* /A+B/ where A is width 1 */
5311 scan = NEXTOPER(scan);
5313 case CURLYN: /* /(A){m,n}B/ where A is width 1 */
5314 ST.paren = scan->flags; /* Which paren to set */
5315 ST.lastparen = rex->lastparen;
5316 ST.lastcloseparen = rex->lastcloseparen;
5317 if (ST.paren > PL_regsize)
5318 PL_regsize = ST.paren;
5319 ST.min = ARG1(scan); /* min to match */
5320 ST.max = ARG2(scan); /* max to match */
5321 if (cur_eval && cur_eval->u.eval.close_paren &&
5322 cur_eval->u.eval.close_paren == (U32)ST.paren) {
5326 scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE);
5328 case CURLY: /* /A{m,n}B/ where A is width 1 */
5330 ST.min = ARG1(scan); /* min to match */
5331 ST.max = ARG2(scan); /* max to match */
5332 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
5335 * Lookahead to avoid useless match attempts
5336 * when we know what character comes next.
5338 * Used to only do .*x and .*?x, but now it allows
5339 * for )'s, ('s and (?{ ... })'s to be in the way
5340 * of the quantifier and the EXACT-like node. -- japhy
5343 if (ST.min > ST.max) /* XXX make this a compile-time check? */
5345 if (HAS_TEXT(next) || JUMPABLE(next)) {
5347 regnode *text_node = next;
5349 if (! HAS_TEXT(text_node))
5350 FIND_NEXT_IMPT(text_node);
5352 if (! HAS_TEXT(text_node))
5353 ST.c1 = ST.c2 = CHRTEST_VOID;
5355 if ( PL_regkind[OP(text_node)] != EXACT ) {
5356 ST.c1 = ST.c2 = CHRTEST_VOID;
5357 goto assume_ok_easy;
5360 s = (U8*)STRING(text_node);
5362 /* Currently we only get here when
5364 PL_rekind[OP(text_node)] == EXACT
5366 if this changes back then the macro for IS_TEXT and
5367 friends need to change. */
5370 switch (OP(text_node)) {
5371 case EXACTF: ST.c2 = PL_fold[ST.c1]; break;
5374 case EXACTFU_TRICKYFOLD:
5375 case EXACTFU: ST.c2 = PL_fold_latin1[ST.c1]; break;
5376 case EXACTFL: ST.c2 = PL_fold_locale[ST.c1]; break;
5377 default: ST.c2 = ST.c1; break;
5380 else { /* UTF_PATTERN */
5381 if (IS_TEXTFU(text_node) || IS_TEXTF(text_node)) {
5383 U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
5385 to_utf8_fold((U8*)s, tmpbuf, &ulen);
5386 ST.c1 = ST.c2 = utf8n_to_uvchr(tmpbuf, UTF8_MAXLEN, 0,
5390 ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0,
5397 ST.c1 = ST.c2 = CHRTEST_VOID;
5402 PL_reginput = locinput;
5405 if (ST.min && regrepeat(rex, ST.A, ST.min, depth) < ST.min)
5408 locinput = PL_reginput;
5410 if (ST.c1 == CHRTEST_VOID)
5411 goto curly_try_B_min;
5413 ST.oldloc = locinput;
5415 /* set ST.maxpos to the furthest point along the
5416 * string that could possibly match */
5417 if (ST.max == REG_INFTY) {
5418 ST.maxpos = PL_regeol - 1;
5420 while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos))
5423 else if (utf8_target) {
5424 int m = ST.max - ST.min;
5425 for (ST.maxpos = locinput;
5426 m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--)
5427 ST.maxpos += UTF8SKIP(ST.maxpos);
5430 ST.maxpos = locinput + ST.max - ST.min;
5431 if (ST.maxpos >= PL_regeol)
5432 ST.maxpos = PL_regeol - 1;
5434 goto curly_try_B_min_known;
5438 ST.count = regrepeat(rex, ST.A, ST.max, depth);
5439 locinput = PL_reginput;
5440 if (ST.count < ST.min)
5442 if ((ST.count > ST.min)
5443 && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL))
5445 /* A{m,n} must come at the end of the string, there's
5446 * no point in backing off ... */
5448 /* ...except that $ and \Z can match before *and* after
5449 newline at the end. Consider "\n\n" =~ /\n+\Z\n/.
5450 We may back off by one in this case. */
5451 if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS)
5455 goto curly_try_B_max;
5457 assert(0); /* NOTREACHED */
5460 case CURLY_B_min_known_fail:
5461 /* failed to find B in a non-greedy match where c1,c2 valid */
5463 PL_reginput = locinput; /* Could be reset... */
5464 REGCP_UNWIND(ST.cp);
5466 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
5468 /* Couldn't or didn't -- move forward. */
5469 ST.oldloc = locinput;
5471 locinput += UTF8SKIP(locinput);
5475 curly_try_B_min_known:
5476 /* find the next place where 'B' could work, then call B */
5480 n = (ST.oldloc == locinput) ? 0 : 1;
5481 if (ST.c1 == ST.c2) {
5483 /* set n to utf8_distance(oldloc, locinput) */
5484 while (locinput <= ST.maxpos &&
5485 utf8n_to_uvchr((U8*)locinput,
5486 UTF8_MAXBYTES, &len,
5487 uniflags) != (UV)ST.c1) {
5493 /* set n to utf8_distance(oldloc, locinput) */
5494 while (locinput <= ST.maxpos) {
5496 const UV c = utf8n_to_uvchr((U8*)locinput,
5497 UTF8_MAXBYTES, &len,
5499 if (c == (UV)ST.c1 || c == (UV)ST.c2)
5507 if (ST.c1 == ST.c2) {
5508 while (locinput <= ST.maxpos &&
5509 UCHARAT(locinput) != ST.c1)
5513 while (locinput <= ST.maxpos
5514 && UCHARAT(locinput) != ST.c1
5515 && UCHARAT(locinput) != ST.c2)
5518 n = locinput - ST.oldloc;
5520 if (locinput > ST.maxpos)
5522 /* PL_reginput == oldloc now */
5525 if (regrepeat(rex, ST.A, n, depth) < n)
5528 PL_reginput = locinput;
5529 CURLY_SETPAREN(ST.paren, ST.count);
5530 if (cur_eval && cur_eval->u.eval.close_paren &&
5531 cur_eval->u.eval.close_paren == (U32)ST.paren) {
5534 PUSH_STATE_GOTO(CURLY_B_min_known, ST.B);
5536 assert(0); /* NOTREACHED */
5539 case CURLY_B_min_fail:
5540 /* failed to find B in a non-greedy match where c1,c2 invalid */
5542 REGCP_UNWIND(ST.cp);
5544 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
5546 /* failed -- move forward one */
5547 PL_reginput = locinput;
5548 if (regrepeat(rex, ST.A, 1, depth)) {
5550 locinput = PL_reginput;
5551 if (ST.count <= ST.max || (ST.max == REG_INFTY &&
5552 ST.count > 0)) /* count overflow ? */
5555 CURLY_SETPAREN(ST.paren, ST.count);
5556 if (cur_eval && cur_eval->u.eval.close_paren &&
5557 cur_eval->u.eval.close_paren == (U32)ST.paren) {
5560 PUSH_STATE_GOTO(CURLY_B_min, ST.B);
5564 assert(0); /* NOTREACHED */
5568 /* a successful greedy match: now try to match B */
5569 if (cur_eval && cur_eval->u.eval.close_paren &&
5570 cur_eval->u.eval.close_paren == (U32)ST.paren) {
5575 if (ST.c1 != CHRTEST_VOID)
5576 c = utf8_target ? utf8n_to_uvchr((U8*)PL_reginput,
5577 UTF8_MAXBYTES, 0, uniflags)
5578 : (UV) UCHARAT(PL_reginput);
5579 /* If it could work, try it. */
5580 if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) {
5581 CURLY_SETPAREN(ST.paren, ST.count);
5582 PUSH_STATE_GOTO(CURLY_B_max, ST.B);
5583 assert(0); /* NOTREACHED */
5587 case CURLY_B_max_fail:
5588 /* failed to find B in a greedy match */
5590 REGCP_UNWIND(ST.cp);
5592 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
5595 if (--ST.count < ST.min)
5597 PL_reginput = locinput = HOPc(locinput, -1);
5598 goto curly_try_B_max;
5605 /* we've just finished A in /(??{A})B/; now continue with B */
5606 st->u.eval.toggle_reg_flags
5607 = cur_eval->u.eval.toggle_reg_flags;
5608 PL_reg_flags ^= st->u.eval.toggle_reg_flags;
5610 st->u.eval.prev_rex = rex_sv; /* inner */
5611 st->u.eval.cp = regcppush(rex, 0); /* Save *all* the positions. */
5612 rex_sv = cur_eval->u.eval.prev_rex;
5613 SET_reg_curpm(rex_sv);
5614 rex = (struct regexp *)SvANY(rex_sv);
5615 rexi = RXi_GET(rex);
5616 cur_curlyx = cur_eval->u.eval.prev_curlyx;
5618 REGCP_SET(st->u.eval.lastcp);
5619 PL_reginput = locinput;
5621 /* Restore parens of the outer rex without popping the
5623 S_regcp_restore(aTHX_ rex, cur_eval->u.eval.lastcp);
5625 st->u.eval.prev_eval = cur_eval;
5626 cur_eval = cur_eval->u.eval.prev_eval;
5628 PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ... %"UVxf"\n",
5629 REPORT_CODE_OFF+depth*2, "",PTR2UV(cur_eval)););
5630 if ( nochange_depth )
5633 PUSH_YES_STATE_GOTO(EVAL_AB,
5634 st->u.eval.prev_eval->u.eval.B); /* match B */
5637 if (locinput < reginfo->till) {
5638 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
5639 "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n",
5641 (long)(locinput - PL_reg_starttry),
5642 (long)(reginfo->till - PL_reg_starttry),
5645 sayNO_SILENT; /* Cannot match: too short. */
5647 PL_reginput = locinput; /* put where regtry can find it */
5648 sayYES; /* Success! */
5650 case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */
5652 PerlIO_printf(Perl_debug_log,
5653 "%*s %ssubpattern success...%s\n",
5654 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]));
5655 PL_reginput = locinput; /* put where regtry can find it */
5656 sayYES; /* Success! */
5659 #define ST st->u.ifmatch
5661 case SUSPEND: /* (?>A) */
5663 PL_reginput = locinput;
5666 case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */
5668 goto ifmatch_trivial_fail_test;
5670 case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */
5672 ifmatch_trivial_fail_test:
5674 char * const s = HOPBACKc(locinput, scan->flags);
5679 sw = 1 - cBOOL(ST.wanted);
5683 next = scan + ARG(scan);
5691 PL_reginput = locinput;
5695 ST.logical = logical;
5696 logical = 0; /* XXX: reset state of logical once it has been saved into ST */
5698 /* execute body of (?...A) */
5699 PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan)));
5700 assert(0); /* NOTREACHED */
5702 case IFMATCH_A_fail: /* body of (?...A) failed */
5703 ST.wanted = !ST.wanted;
5706 case IFMATCH_A: /* body of (?...A) succeeded */
5708 sw = cBOOL(ST.wanted);
5710 else if (!ST.wanted)
5713 if (OP(ST.me) == SUSPEND)
5714 locinput = PL_reginput;
5716 locinput = PL_reginput = st->locinput;
5717 nextchr = UCHARAT(locinput);
5719 scan = ST.me + ARG(ST.me);
5722 continue; /* execute B */
5727 next = scan + ARG(scan);
5732 reginfo->cutpoint = PL_regeol;
5735 PL_reginput = locinput;
5737 sv_yes_mark = sv_commit = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
5738 PUSH_STATE_GOTO(COMMIT_next,next);
5739 assert(0); /* NOTREACHED */
5740 case COMMIT_next_fail:
5745 assert(0); /* NOTREACHED */
5747 #define ST st->u.mark
5749 ST.prev_mark = mark_state;
5750 ST.mark_name = sv_commit = sv_yes_mark
5751 = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
5753 ST.mark_loc = PL_reginput = locinput;
5754 PUSH_YES_STATE_GOTO(MARKPOINT_next,next);
5755 assert(0); /* NOTREACHED */
5756 case MARKPOINT_next:
5757 mark_state = ST.prev_mark;
5759 assert(0); /* NOTREACHED */
5760 case MARKPOINT_next_fail:
5761 if (popmark && sv_eq(ST.mark_name,popmark))
5763 if (ST.mark_loc > startpoint)
5764 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
5765 popmark = NULL; /* we found our mark */
5766 sv_commit = ST.mark_name;
5769 PerlIO_printf(Perl_debug_log,
5770 "%*s %ssetting cutpoint to mark:%"SVf"...%s\n",
5771 REPORT_CODE_OFF+depth*2, "",
5772 PL_colors[4], SVfARG(sv_commit), PL_colors[5]);
5775 mark_state = ST.prev_mark;
5776 sv_yes_mark = mark_state ?
5777 mark_state->u.mark.mark_name : NULL;
5779 assert(0); /* NOTREACHED */
5781 PL_reginput = locinput;
5783 /* (*SKIP) : if we fail we cut here*/
5784 ST.mark_name = NULL;
5785 ST.mark_loc = locinput;
5786 PUSH_STATE_GOTO(SKIP_next,next);
5788 /* (*SKIP:NAME) : if there is a (*MARK:NAME) fail where it was,
5789 otherwise do nothing. Meaning we need to scan
5791 regmatch_state *cur = mark_state;
5792 SV *find = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
5795 if ( sv_eq( cur->u.mark.mark_name,
5798 ST.mark_name = find;
5799 PUSH_STATE_GOTO( SKIP_next, next );
5801 cur = cur->u.mark.prev_mark;
5804 /* Didn't find our (*MARK:NAME) so ignore this (*SKIP:NAME) */
5806 case SKIP_next_fail:
5808 /* (*CUT:NAME) - Set up to search for the name as we
5809 collapse the stack*/
5810 popmark = ST.mark_name;
5812 /* (*CUT) - No name, we cut here.*/
5813 if (ST.mark_loc > startpoint)
5814 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
5815 /* but we set sv_commit to latest mark_name if there
5816 is one so they can test to see how things lead to this
5819 sv_commit=mark_state->u.mark.mark_name;
5823 assert(0); /* NOTREACHED */
5826 if ((n=is_LNBREAK(locinput,utf8_target))) {
5828 nextchr = UCHARAT(locinput);
5833 #define CASE_CLASS(nAmE) \
5835 if (locinput >= PL_regeol) \
5837 if ((n=is_##nAmE(locinput,utf8_target))) { \
5839 nextchr = UCHARAT(locinput); \
5844 if (locinput >= PL_regeol) \
5846 if ((n=is_##nAmE(locinput,utf8_target))) { \
5849 locinput += UTF8SKIP(locinput); \
5850 nextchr = UCHARAT(locinput); \
5855 CASE_CLASS(HORIZWS);
5859 PerlIO_printf(Perl_error_log, "%"UVxf" %d\n",
5860 PTR2UV(scan), OP(scan));
5861 Perl_croak(aTHX_ "regexp memory corruption");
5865 /* switch break jumps here */
5866 scan = next; /* prepare to execute the next op and ... */
5867 continue; /* ... jump back to the top, reusing st */
5868 assert(0); /* NOTREACHED */
5871 /* push a state that backtracks on success */
5872 st->u.yes.prev_yes_state = yes_state;
5876 /* push a new regex state, then continue at scan */
5878 regmatch_state *newst;
5881 regmatch_state *cur = st;
5882 regmatch_state *curyes = yes_state;
5884 regmatch_slab *slab = PL_regmatch_slab;
5885 for (;curd > -1;cur--,curd--) {
5886 if (cur < SLAB_FIRST(slab)) {
5888 cur = SLAB_LAST(slab);
5890 PerlIO_printf(Perl_error_log, "%*s#%-3d %-10s %s\n",
5891 REPORT_CODE_OFF + 2 + depth * 2,"",
5892 curd, PL_reg_name[cur->resume_state],
5893 (curyes == cur) ? "yes" : ""
5896 curyes = cur->u.yes.prev_yes_state;
5899 DEBUG_STATE_pp("push")
5902 st->locinput = locinput;
5904 if (newst > SLAB_LAST(PL_regmatch_slab))
5905 newst = S_push_slab(aTHX);
5906 PL_regmatch_state = newst;
5908 locinput = PL_reginput;
5909 nextchr = UCHARAT(locinput);
5912 assert(0); /* NOTREACHED */
5917 * We get here only if there's trouble -- normally "case END" is
5918 * the terminating point.
5920 Perl_croak(aTHX_ "corrupted regexp pointers");
5926 /* we have successfully completed a subexpression, but we must now
5927 * pop to the state marked by yes_state and continue from there */
5928 assert(st != yes_state);
5930 while (st != yes_state) {
5932 if (st < SLAB_FIRST(PL_regmatch_slab)) {
5933 PL_regmatch_slab = PL_regmatch_slab->prev;
5934 st = SLAB_LAST(PL_regmatch_slab);
5938 DEBUG_STATE_pp("pop (no final)");
5940 DEBUG_STATE_pp("pop (yes)");
5946 while (yes_state < SLAB_FIRST(PL_regmatch_slab)
5947 || yes_state > SLAB_LAST(PL_regmatch_slab))
5949 /* not in this slab, pop slab */
5950 depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1);
5951 PL_regmatch_slab = PL_regmatch_slab->prev;
5952 st = SLAB_LAST(PL_regmatch_slab);
5954 depth -= (st - yes_state);
5957 yes_state = st->u.yes.prev_yes_state;
5958 PL_regmatch_state = st;
5961 locinput= st->locinput;
5962 nextchr = UCHARAT(locinput);
5964 state_num = st->resume_state + no_final;
5965 goto reenter_switch;
5968 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n",
5969 PL_colors[4], PL_colors[5]));
5971 if (PL_reg_state.re_state_eval_setup_done) {
5972 /* each successfully executed (?{...}) block does the equivalent of
5973 * local $^R = do {...}
5974 * When popping the save stack, all these locals would be undone;
5975 * bypass this by setting the outermost saved $^R to the latest
5977 if (oreplsv != GvSV(PL_replgv))
5978 sv_setsv(oreplsv, GvSV(PL_replgv));
5985 PerlIO_printf(Perl_debug_log,
5986 "%*s %sfailed...%s\n",
5987 REPORT_CODE_OFF+depth*2, "",
5988 PL_colors[4], PL_colors[5])
6000 /* there's a previous state to backtrack to */
6002 if (st < SLAB_FIRST(PL_regmatch_slab)) {
6003 PL_regmatch_slab = PL_regmatch_slab->prev;
6004 st = SLAB_LAST(PL_regmatch_slab);
6006 PL_regmatch_state = st;
6007 locinput= st->locinput;
6008 nextchr = UCHARAT(locinput);
6010 DEBUG_STATE_pp("pop");
6012 if (yes_state == st)
6013 yes_state = st->u.yes.prev_yes_state;
6015 state_num = st->resume_state + 1; /* failure = success + 1 */
6016 goto reenter_switch;
6021 if (rex->intflags & PREGf_VERBARG_SEEN) {
6022 SV *sv_err = get_sv("REGERROR", 1);
6023 SV *sv_mrk = get_sv("REGMARK", 1);
6025 sv_commit = &PL_sv_no;
6027 sv_yes_mark = &PL_sv_yes;
6030 sv_commit = &PL_sv_yes;
6031 sv_yes_mark = &PL_sv_no;
6033 sv_setsv(sv_err, sv_commit);
6034 sv_setsv(sv_mrk, sv_yes_mark);
6038 if (last_pushed_cv) {
6043 /* clean up; in particular, free all slabs above current one */
6044 LEAVE_SCOPE(oldsave);
6050 - regrepeat - repeatedly match something simple, report how many
6053 * [This routine now assumes that it will only match on things of length 1.
6054 * That was true before, but now we assume scan - reginput is the count,
6055 * rather than incrementing count on every character. [Er, except utf8.]]
6058 S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max, int depth)
6061 register char *scan;
6063 register char *loceol = PL_regeol;
6064 register I32 hardcount = 0;
6065 register bool utf8_target = PL_reg_match_utf8;
6068 PERL_UNUSED_ARG(depth);
6071 PERL_ARGS_ASSERT_REGREPEAT;
6074 if (max == REG_INFTY)
6076 else if (max < loceol - scan)
6077 loceol = scan + max;
6082 while (scan < loceol && hardcount < max && *scan != '\n') {
6083 scan += UTF8SKIP(scan);
6087 while (scan < loceol && *scan != '\n')
6094 while (scan < loceol && hardcount < max) {
6095 scan += UTF8SKIP(scan);
6106 /* To get here, EXACTish nodes must have *byte* length == 1. That
6107 * means they match only characters in the string that can be expressed
6108 * as a single byte. For non-utf8 strings, that means a simple match.
6109 * For utf8 strings, the character matched must be an invariant, or
6110 * downgradable to a single byte. The pattern's utf8ness is
6111 * irrelevant, as since it's a single byte, it either isn't utf8, or if
6112 * it is, it's an invariant */
6115 assert(! UTF_PATTERN || UNI_IS_INVARIANT(c));
6117 if (! utf8_target || UNI_IS_INVARIANT(c)) {
6118 while (scan < loceol && UCHARAT(scan) == c) {
6124 /* Here, the string is utf8, and the pattern char is different
6125 * in utf8 than not, so can't compare them directly. Outside the
6126 * loop, find the two utf8 bytes that represent c, and then
6127 * look for those in sequence in the utf8 string */
6128 U8 high = UTF8_TWO_BYTE_HI(c);
6129 U8 low = UTF8_TWO_BYTE_LO(c);
6132 while (hardcount < max
6133 && scan + 1 < loceol
6134 && UCHARAT(scan) == high
6135 && UCHARAT(scan + 1) == low)
6143 utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
6147 PL_reg_flags |= RF_tainted;
6148 utf8_flags = FOLDEQ_UTF8_LOCALE;
6156 case EXACTFU_TRICKYFOLD:
6158 utf8_flags = (UTF_PATTERN) ? FOLDEQ_S2_ALREADY_FOLDED : 0;
6160 /* The comments for the EXACT case above apply as well to these fold
6165 assert(! UTF_PATTERN || UNI_IS_INVARIANT(c));
6167 if (utf8_target || OP(p) == EXACTFU_SS) { /* Use full Unicode fold matching */
6168 char *tmpeol = loceol;
6169 while (hardcount < max
6170 && foldEQ_utf8_flags(scan, &tmpeol, 0, utf8_target,
6171 STRING(p), NULL, 1, cBOOL(UTF_PATTERN), utf8_flags))
6178 /* XXX Note that the above handles properly the German sharp s in
6179 * the pattern matching ss in the string. But it doesn't handle
6180 * properly cases where the string contains say 'LIGATURE ff' and
6181 * the pattern is 'f+'. This would require, say, a new function or
6182 * revised interface to foldEQ_utf8(), in which the maximum number
6183 * of characters to match could be passed and it would return how
6184 * many actually did. This is just one of many cases where
6185 * multi-char folds don't work properly, and so the fix is being
6191 /* Here, the string isn't utf8 and c is a single byte; and either
6192 * the pattern isn't utf8 or c is an invariant, so its utf8ness
6193 * doesn't affect c. Can just do simple comparisons for exact or
6196 case EXACTF: folded = PL_fold[c]; break;
6198 case EXACTFU_TRICKYFOLD:
6199 case EXACTFU: folded = PL_fold_latin1[c]; break;
6200 case EXACTFL: folded = PL_fold_locale[c]; break;
6201 default: Perl_croak(aTHX_ "panic: Unexpected op %u", OP(p));
6203 while (scan < loceol &&
6204 (UCHARAT(scan) == c || UCHARAT(scan) == folded))
6212 if (utf8_target || OP(p) == ANYOFV) {
6215 inclasslen = loceol - scan;
6216 while (hardcount < max
6217 && ((inclasslen = loceol - scan) > 0)
6218 && reginclass(prog, p, (U8*)scan, &inclasslen, utf8_target))
6224 while (scan < loceol && REGINCLASS(prog, p, (U8*)scan))
6232 LOAD_UTF8_CHARCLASS_ALNUM();
6233 while (hardcount < max && scan < loceol &&
6234 swash_fetch(PL_utf8_alnum, (U8*)scan, utf8_target))
6236 scan += UTF8SKIP(scan);
6240 while (scan < loceol && isWORDCHAR_L1((U8) *scan)) {
6248 while (scan < loceol && isALNUM((U8) *scan)) {
6253 while (scan < loceol && isWORDCHAR_A((U8) *scan)) {
6258 PL_reg_flags |= RF_tainted;
6261 while (hardcount < max && scan < loceol &&
6262 isALNUM_LC_utf8((U8*)scan)) {
6263 scan += UTF8SKIP(scan);
6267 while (scan < loceol && isALNUM_LC(*scan))
6277 LOAD_UTF8_CHARCLASS_ALNUM();
6278 while (hardcount < max && scan < loceol &&
6279 ! swash_fetch(PL_utf8_alnum, (U8*)scan, utf8_target))
6281 scan += UTF8SKIP(scan);
6285 while (scan < loceol && ! isWORDCHAR_L1((U8) *scan)) {
6292 goto utf8_Nwordchar;
6293 while (scan < loceol && ! isALNUM((U8) *scan)) {
6299 while (scan < loceol && ! isWORDCHAR_A((U8) *scan)) {
6300 scan += UTF8SKIP(scan);
6304 while (scan < loceol && ! isWORDCHAR_A((U8) *scan)) {
6310 PL_reg_flags |= RF_tainted;
6313 while (hardcount < max && scan < loceol &&
6314 !isALNUM_LC_utf8((U8*)scan)) {
6315 scan += UTF8SKIP(scan);
6319 while (scan < loceol && !isALNUM_LC(*scan))
6329 LOAD_UTF8_CHARCLASS_SPACE();
6330 while (hardcount < max && scan < loceol &&
6332 swash_fetch(PL_utf8_space,(U8*)scan, utf8_target)))
6334 scan += UTF8SKIP(scan);
6340 while (scan < loceol && isSPACE_L1((U8) *scan)) {
6349 while (scan < loceol && isSPACE((U8) *scan)) {
6354 while (scan < loceol && isSPACE_A((U8) *scan)) {
6359 PL_reg_flags |= RF_tainted;
6362 while (hardcount < max && scan < loceol &&
6363 isSPACE_LC_utf8((U8*)scan)) {
6364 scan += UTF8SKIP(scan);
6368 while (scan < loceol && isSPACE_LC(*scan))
6378 LOAD_UTF8_CHARCLASS_SPACE();
6379 while (hardcount < max && scan < loceol &&
6381 swash_fetch(PL_utf8_space,(U8*)scan, utf8_target)))
6383 scan += UTF8SKIP(scan);
6389 while (scan < loceol && ! isSPACE_L1((U8) *scan)) {
6398 while (scan < loceol && ! isSPACE((U8) *scan)) {
6404 while (scan < loceol && ! isSPACE_A((U8) *scan)) {
6405 scan += UTF8SKIP(scan);
6409 while (scan < loceol && ! isSPACE_A((U8) *scan)) {
6415 PL_reg_flags |= RF_tainted;
6418 while (hardcount < max && scan < loceol &&
6419 !isSPACE_LC_utf8((U8*)scan)) {
6420 scan += UTF8SKIP(scan);
6424 while (scan < loceol && !isSPACE_LC(*scan))
6431 LOAD_UTF8_CHARCLASS_DIGIT();
6432 while (hardcount < max && scan < loceol &&
6433 swash_fetch(PL_utf8_digit, (U8*)scan, utf8_target)) {
6434 scan += UTF8SKIP(scan);
6438 while (scan < loceol && isDIGIT(*scan))
6443 while (scan < loceol && isDIGIT_A((U8) *scan)) {
6448 PL_reg_flags |= RF_tainted;
6451 while (hardcount < max && scan < loceol &&
6452 isDIGIT_LC_utf8((U8*)scan)) {
6453 scan += UTF8SKIP(scan);
6457 while (scan < loceol && isDIGIT_LC(*scan))
6464 LOAD_UTF8_CHARCLASS_DIGIT();
6465 while (hardcount < max && scan < loceol &&
6466 !swash_fetch(PL_utf8_digit, (U8*)scan, utf8_target)) {
6467 scan += UTF8SKIP(scan);
6471 while (scan < loceol && !isDIGIT(*scan))
6477 while (scan < loceol && ! isDIGIT_A((U8) *scan)) {
6478 scan += UTF8SKIP(scan);
6482 while (scan < loceol && ! isDIGIT_A((U8) *scan)) {
6488 PL_reg_flags |= RF_tainted;
6491 while (hardcount < max && scan < loceol &&
6492 !isDIGIT_LC_utf8((U8*)scan)) {
6493 scan += UTF8SKIP(scan);
6497 while (scan < loceol && !isDIGIT_LC(*scan))
6504 while (hardcount < max && scan < loceol && (c=is_LNBREAK_utf8(scan))) {
6510 LNBREAK can match two latin chars, which is ok,
6511 because we have a null terminated string, but we
6512 have to use hardcount in this situation
6514 while (scan < loceol && (c=is_LNBREAK_latin1(scan))) {
6523 while (hardcount < max && scan < loceol && (c=is_HORIZWS_utf8(scan))) {
6528 while (scan < loceol && is_HORIZWS_latin1(scan))
6535 while (hardcount < max && scan < loceol && !is_HORIZWS_utf8(scan)) {
6536 scan += UTF8SKIP(scan);
6540 while (scan < loceol && !is_HORIZWS_latin1(scan))
6548 while (hardcount < max && scan < loceol && (c=is_VERTWS_utf8(scan))) {
6553 while (scan < loceol && is_VERTWS_latin1(scan))
6561 while (hardcount < max && scan < loceol && !is_VERTWS_utf8(scan)) {
6562 scan += UTF8SKIP(scan);
6566 while (scan < loceol && !is_VERTWS_latin1(scan))
6572 default: /* Called on something of 0 width. */
6573 break; /* So match right here or not at all. */
6579 c = scan - PL_reginput;
6583 GET_RE_DEBUG_FLAGS_DECL;
6585 SV * const prop = sv_newmortal();
6586 regprop(prog, prop, p);
6587 PerlIO_printf(Perl_debug_log,
6588 "%*s %s can match %"IVdf" times out of %"IVdf"...\n",
6589 REPORT_CODE_OFF + depth*2, "", SvPVX_const(prop),(IV)c,(IV)max);
6597 #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION)
6599 - regclass_swash - prepare the utf8 swash. Wraps the shared core version to
6600 create a copy so that changes the caller makes won't change the shared one
6603 Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
6605 PERL_ARGS_ASSERT_REGCLASS_SWASH;
6606 return newSVsv(core_regclass_swash(prog, node, doinit, listsvp, altsvp));
6611 S_core_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
6613 /* Returns the swash for the input 'node' in the regex 'prog'.
6614 * If <doinit> is true, will attempt to create the swash if not already
6616 * If <listsvp> is non-null, will return the swash initialization string in
6618 * If <altsvp> is non-null, will return the alternates to the regular swash
6620 * Tied intimately to how regcomp.c sets up the data structure */
6628 RXi_GET_DECL(prog,progi);
6629 const struct reg_data * const data = prog ? progi->data : NULL;
6631 PERL_ARGS_ASSERT_CORE_REGCLASS_SWASH;
6633 assert(ANYOF_NONBITMAP(node));
6635 if (data && data->count) {
6636 const U32 n = ARG(node);
6638 if (data->what[n] == 's') {
6639 SV * const rv = MUTABLE_SV(data->data[n]);
6640 AV * const av = MUTABLE_AV(SvRV(rv));
6641 SV **const ary = AvARRAY(av);
6642 bool invlist_has_user_defined_property;
6644 si = *ary; /* ary[0] = the string to initialize the swash with */
6646 /* Elements 3 and 4 are either both present or both absent. [3] is
6647 * any inversion list generated at compile time; [4] indicates if
6648 * that inversion list has any user-defined properties in it. */
6649 if (av_len(av) >= 3) {
6651 invlist_has_user_defined_property = cBOOL(SvUV(ary[4]));
6655 invlist_has_user_defined_property = FALSE;
6658 /* Element [1] is reserved for the set-up swash. If already there,
6659 * return it; if not, create it and store it there */
6660 if (SvROK(ary[1])) {
6663 else if (si && doinit) {
6665 sw = _core_swash_init("utf8", /* the utf8 package */
6669 0, /* not from tr/// */
6670 FALSE, /* is error if can't find
6673 invlist_has_user_defined_property);
6674 (void)av_store(av, 1, sw);
6677 /* Element [2] is for any multi-char folds. Note that is a
6678 * fundamentally flawed design, because can't backtrack and try
6679 * again. See [perl #89774] */
6680 if (SvTYPE(ary[2]) == SVt_PVAV) {
6687 SV* matches_string = newSVpvn("", 0);
6690 /* Use the swash, if any, which has to have incorporated into it all
6694 && SvTYPE(SvRV(sw)) == SVt_PVHV
6695 && (invlistsvp = hv_fetchs(MUTABLE_HV(SvRV(sw)), "INVLIST", FALSE)))
6697 invlist = *invlistsvp;
6699 else if (si && si != &PL_sv_undef) {
6701 /* If no swash, use the input nitialization string, if available */
6702 sv_catsv(matches_string, si);
6705 /* Add the inversion list to whatever we have. This may have come from
6706 * the swash, or from an input parameter */
6708 sv_catsv(matches_string, _invlist_contents(invlist));
6710 *listsvp = matches_string;
6720 - reginclass - determine if a character falls into a character class
6722 n is the ANYOF regnode
6723 p is the target string
6724 lenp is pointer to the maximum number of bytes of how far to go in p
6725 (This is assumed wthout checking to always be at least the current
6727 utf8_target tells whether p is in UTF-8.
6729 Returns true if matched; false otherwise. If lenp is not NULL, on return
6730 from a successful match, the value it points to will be updated to how many
6731 bytes in p were matched. If there was no match, the value is undefined,
6732 possibly changed from the input.
6734 Note that this can be a synthetic start class, a combination of various
6735 nodes, so things you think might be mutually exclusive, such as locale,
6736 aren't. It can match both locale and non-locale
6741 S_reginclass(pTHX_ const regexp * const prog, register const regnode * const n, register const U8* const p, STRLEN* lenp, register const bool utf8_target)
6744 const char flags = ANYOF_FLAGS(n);
6750 PERL_ARGS_ASSERT_REGINCLASS;
6752 /* If c is not already the code point, get it */
6753 if (utf8_target && !UTF8_IS_INVARIANT(c)) {
6754 c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &c_len,
6755 (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV)
6756 | UTF8_ALLOW_FFFF | UTF8_CHECK_ONLY);
6757 /* see [perl #37836] for UTF8_ALLOW_ANYUV; [perl #38293] for
6758 * UTF8_ALLOW_FFFF */
6759 if (c_len == (STRLEN)-1)
6760 Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)");
6766 /* Use passed in max length, or one character if none passed in or less
6767 * than one character. And assume will match just one character. This is
6768 * overwritten later if matched more. */
6770 maxlen = (*lenp > c_len) ? *lenp : c_len;
6778 /* If this character is potentially in the bitmap, check it */
6780 if (ANYOF_BITMAP_TEST(n, c))
6782 else if (flags & ANYOF_NON_UTF8_LATIN1_ALL
6789 else if (flags & ANYOF_LOCALE) {
6790 PL_reg_flags |= RF_tainted;
6792 if ((flags & ANYOF_LOC_NONBITMAP_FOLD)
6793 && ANYOF_BITMAP_TEST(n, PL_fold_locale[c]))
6797 else if (ANYOF_CLASS_TEST_ANY_SET(n) &&
6798 ((ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) ||
6799 (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) ||
6800 (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) ||
6801 (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) ||
6802 (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) ||
6803 (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) ||
6804 (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) ||
6805 (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) ||
6806 (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) ||
6807 (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) ||
6808 (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII_LC(c)) ||
6809 (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII_LC(c)) ||
6810 (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) ||
6811 (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) ||
6812 (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) ||
6813 (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) ||
6814 (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) ||
6815 (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) ||
6816 (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) ||
6817 (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) ||
6818 (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) ||
6819 (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) ||
6820 (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) ||
6821 (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) ||
6822 (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) ||
6823 (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) ||
6824 (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) ||
6825 (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) ||
6826 (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK_LC(c)) ||
6827 (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK_LC(c))
6828 ) /* How's that for a conditional? */
6835 /* If the bitmap didn't (or couldn't) match, and something outside the
6836 * bitmap could match, try that. Locale nodes specifiy completely the
6837 * behavior of code points in the bit map (otherwise, a utf8 target would
6838 * cause them to be treated as Unicode and not locale), except in
6839 * the very unlikely event when this node is a synthetic start class, which
6840 * could be a combination of locale and non-locale nodes. So allow locale
6841 * to match for the synthetic start class, which will give a false
6842 * positive that will be resolved when the match is done again as not part
6843 * of the synthetic start class */
6845 if (utf8_target && (flags & ANYOF_UNICODE_ALL) && c >= 256) {
6846 match = TRUE; /* Everything above 255 matches */
6848 else if (ANYOF_NONBITMAP(n)
6849 && ((flags & ANYOF_NONBITMAP_NON_UTF8)
6852 || (! (flags & ANYOF_LOCALE))
6853 || (flags & ANYOF_IS_SYNTHETIC)))))
6856 SV * const sw = core_regclass_swash(prog, n, TRUE, 0, (SV**)&av);
6864 /* Not utf8. Convert as much of the string as available up
6865 * to the limit of how far the (single) character in the
6866 * pattern can possibly match (no need to go further). If
6867 * the node is a straight ANYOF or not folding, it can't
6868 * match more than one. Otherwise, It can match up to how
6869 * far a single char can fold to. Since not utf8, each
6870 * character is a single byte, so the max it can be in
6871 * bytes is the same as the max it can be in characters */
6872 STRLEN len = (OP(n) == ANYOF
6873 || ! (flags & ANYOF_LOC_NONBITMAP_FOLD))
6875 : (maxlen < UTF8_MAX_FOLD_CHAR_EXPAND)
6877 : UTF8_MAX_FOLD_CHAR_EXPAND;
6878 utf8_p = bytes_to_utf8(p, &len);
6881 if (swash_fetch(sw, utf8_p, TRUE))
6883 else if (flags & ANYOF_LOC_NONBITMAP_FOLD) {
6885 /* Here, we need to test if the fold of the target string
6886 * matches. The non-multi char folds have all been moved to
6887 * the compilation phase, and the multi-char folds have
6888 * been stored by regcomp into 'av'; we linearly check to
6889 * see if any match the target string (folded). We know
6890 * that the originals were each one character, but we don't
6891 * currently know how many characters/bytes each folded to,
6892 * except we do know that there are small limits imposed by
6893 * Unicode. XXX A performance enhancement would be to have
6894 * regcomp.c store the max number of chars/bytes that are
6895 * in an av entry, as, say the 0th element. Even better
6896 * would be to have a hash of the few characters that can
6897 * start a multi-char fold to the max number of chars of
6900 * If there is a match, we will need to advance (if lenp is
6901 * specified) the match pointer in the target string. But
6902 * what we are comparing here isn't that string directly,
6903 * but its fold, whose length may differ from the original.
6904 * As we go along in constructing the fold, therefore, we
6905 * create a map so that we know how many bytes in the
6906 * source to advance given that we have matched a certain
6907 * number of bytes in the fold. This map is stored in
6908 * 'map_fold_len_back'. Let n mean the number of bytes in
6909 * the fold of the first character that we are folding.
6910 * Then map_fold_len_back[n] is set to the number of bytes
6911 * in that first character. Similarly let m be the
6912 * corresponding number for the second character to be
6913 * folded. Then map_fold_len_back[n+m] is set to the
6914 * number of bytes occupied by the first two source
6915 * characters. ... */
6916 U8 map_fold_len_back[UTF8_MAXBYTES_CASE+1] = { 0 };
6917 U8 folded[UTF8_MAXBYTES_CASE+1];
6918 STRLEN foldlen = 0; /* num bytes in fold of 1st char */
6919 STRLEN total_foldlen = 0; /* num bytes in fold of all
6922 if (OP(n) == ANYOF || maxlen == 1 || ! lenp || ! av) {
6924 /* Here, only need to fold the first char of the target
6925 * string. It the source wasn't utf8, is 1 byte long */
6926 to_utf8_fold(utf8_p, folded, &foldlen);
6927 total_foldlen = foldlen;
6928 map_fold_len_back[foldlen] = (utf8_target)
6934 /* Here, need to fold more than the first char. Do so
6935 * up to the limits */
6936 U8* source_ptr = utf8_p; /* The source for the fold
6939 U8* folded_ptr = folded;
6940 U8* e = utf8_p + maxlen; /* Can't go beyond last
6941 available byte in the
6945 i < UTF8_MAX_FOLD_CHAR_EXPAND && source_ptr < e;
6949 /* Fold the next character */
6950 U8 this_char_folded[UTF8_MAXBYTES_CASE+1];
6951 STRLEN this_char_foldlen;
6952 to_utf8_fold(source_ptr,
6954 &this_char_foldlen);
6956 /* Bail if it would exceed the byte limit for
6957 * folding a single char. */
6958 if (this_char_foldlen + folded_ptr - folded >
6964 /* Add the fold of this character */
6965 Copy(this_char_folded,
6969 source_ptr += UTF8SKIP(source_ptr);
6970 folded_ptr += this_char_foldlen;
6971 total_foldlen = folded_ptr - folded;
6973 /* Create map from the number of bytes in the fold
6974 * back to the number of bytes in the source. If
6975 * the source isn't utf8, the byte count is just
6976 * the number of characters so far */
6977 map_fold_len_back[total_foldlen]
6979 ? source_ptr - utf8_p
6986 /* Do the linear search to see if the fold is in the list
6987 * of multi-char folds. */
6990 for (i = 0; i <= av_len(av); i++) {
6991 SV* const sv = *av_fetch(av, i, FALSE);
6993 const char * const s = SvPV_const(sv, len);
6995 if (len <= total_foldlen
6996 && memEQ(s, (char*)folded, len)
6998 /* If 0, means matched a partial char. See
7000 && map_fold_len_back[len])
7003 /* Advance the target string ptr to account for
7004 * this fold, but have to translate from the
7005 * folded length to the corresponding source
7008 *lenp = map_fold_len_back[len];
7017 /* If we allocated a string above, free it */
7018 if (! utf8_target) Safefree(utf8_p);
7023 return (flags & ANYOF_INVERT) ? !match : match;
7027 S_reghop3(U8 *s, I32 off, const U8* lim)
7029 /* return the position 'off' UTF-8 characters away from 's', forward if
7030 * 'off' >= 0, backwards if negative. But don't go outside of position
7031 * 'lim', which better be < s if off < 0 */
7035 PERL_ARGS_ASSERT_REGHOP3;
7038 while (off-- && s < lim) {
7039 /* XXX could check well-formedness here */
7044 while (off++ && s > lim) {
7046 if (UTF8_IS_CONTINUED(*s)) {
7047 while (s > lim && UTF8_IS_CONTINUATION(*s))
7050 /* XXX could check well-formedness here */
7057 /* there are a bunch of places where we use two reghop3's that should
7058 be replaced with this routine. but since thats not done yet
7059 we ifdef it out - dmq
7062 S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim)
7066 PERL_ARGS_ASSERT_REGHOP4;
7069 while (off-- && s < rlim) {
7070 /* XXX could check well-formedness here */
7075 while (off++ && s > llim) {
7077 if (UTF8_IS_CONTINUED(*s)) {
7078 while (s > llim && UTF8_IS_CONTINUATION(*s))
7081 /* XXX could check well-formedness here */
7089 S_reghopmaybe3(U8* s, I32 off, const U8* lim)
7093 PERL_ARGS_ASSERT_REGHOPMAYBE3;
7096 while (off-- && s < lim) {
7097 /* XXX could check well-formedness here */
7104 while (off++ && s > lim) {
7106 if (UTF8_IS_CONTINUED(*s)) {
7107 while (s > lim && UTF8_IS_CONTINUATION(*s))
7110 /* XXX could check well-formedness here */
7119 restore_pos(pTHX_ void *arg)
7122 regexp * const rex = (regexp *)arg;
7123 if (PL_reg_state.re_state_eval_setup_done) {
7124 if (PL_reg_oldsaved) {
7125 rex->subbeg = PL_reg_oldsaved;
7126 rex->sublen = PL_reg_oldsavedlen;
7127 #ifdef PERL_OLD_COPY_ON_WRITE
7128 rex->saved_copy = PL_nrs;
7130 RXp_MATCH_COPIED_on(rex);
7132 PL_reg_magic->mg_len = PL_reg_oldpos;
7133 PL_reg_state.re_state_eval_setup_done = FALSE;
7134 PL_curpm = PL_reg_oldcurpm;
7139 S_to_utf8_substr(pTHX_ register regexp *prog)
7143 PERL_ARGS_ASSERT_TO_UTF8_SUBSTR;
7146 if (prog->substrs->data[i].substr
7147 && !prog->substrs->data[i].utf8_substr) {
7148 SV* const sv = newSVsv(prog->substrs->data[i].substr);
7149 prog->substrs->data[i].utf8_substr = sv;
7150 sv_utf8_upgrade(sv);
7151 if (SvVALID(prog->substrs->data[i].substr)) {
7152 if (SvTAIL(prog->substrs->data[i].substr)) {
7153 /* Trim the trailing \n that fbm_compile added last
7155 SvCUR_set(sv, SvCUR(sv) - 1);
7156 /* Whilst this makes the SV technically "invalid" (as its
7157 buffer is no longer followed by "\0") when fbm_compile()
7158 adds the "\n" back, a "\0" is restored. */
7159 fbm_compile(sv, FBMcf_TAIL);
7163 if (prog->substrs->data[i].substr == prog->check_substr)
7164 prog->check_utf8 = sv;
7170 S_to_byte_substr(pTHX_ register regexp *prog)
7175 PERL_ARGS_ASSERT_TO_BYTE_SUBSTR;
7178 if (prog->substrs->data[i].utf8_substr
7179 && !prog->substrs->data[i].substr) {
7180 SV* sv = newSVsv(prog->substrs->data[i].utf8_substr);
7181 if (sv_utf8_downgrade(sv, TRUE)) {
7182 if (SvVALID(prog->substrs->data[i].utf8_substr)) {
7183 if (SvTAIL(prog->substrs->data[i].utf8_substr)) {
7184 /* Trim the trailing \n that fbm_compile added last
7186 SvCUR_set(sv, SvCUR(sv) - 1);
7187 fbm_compile(sv, FBMcf_TAIL);
7195 prog->substrs->data[i].substr = sv;
7196 if (prog->substrs->data[i].utf8_substr == prog->check_utf8)
7197 prog->check_substr = sv;
7204 * c-indentation-style: bsd
7206 * indent-tabs-mode: nil
7209 * ex: set ts=8 sts=4 sw=4 et: