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
40 #define B_ON_NON_UTF8_LOCALE_IS_WRONG \
41 "Use of \\b{} for non-UTF-8 locale is wrong. Assuming a UTF-8 locale"
44 * pregcomp and pregexec -- regsub and regerror are not used in perl
46 * Copyright (c) 1986 by University of Toronto.
47 * Written by Henry Spencer. Not derived from licensed software.
49 * Permission is granted to anyone to use this software for any
50 * purpose on any computer system, and to redistribute it freely,
51 * subject to the following restrictions:
53 * 1. The author is not responsible for the consequences of use of
54 * this software, no matter how awful, even if they arise
57 * 2. The origin of this software must not be misrepresented, either
58 * by explicit claim or by omission.
60 * 3. Altered versions must be plainly marked as such, and must not
61 * be misrepresented as being the original software.
63 **** Alterations to Henry's code are...
65 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
66 **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
67 **** by Larry Wall and others
69 **** You may distribute under the terms of either the GNU General Public
70 **** License or the Artistic License, as specified in the README file.
72 * Beware that some of this code is subtly aware of the way operator
73 * precedence is structured in regular expressions. Serious changes in
74 * regular-expression syntax might require a total rethink.
77 #define PERL_IN_REGEXEC_C
80 #ifdef PERL_IN_XSUB_RE
86 #include "inline_invlist.c"
87 #include "unicode_constants.h"
90 /* At least one required character in the target string is expressible only in
92 static const char* const non_utf8_target_but_utf8_required
93 = "Can't match, because target string needs to be in UTF-8\n";
96 #define NON_UTF8_TARGET_BUT_UTF8_REQUIRED(target) STMT_START { \
97 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s", non_utf8_target_but_utf8_required));\
101 #define HAS_NONLATIN1_FOLD_CLOSURE(i) _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(i)
104 #define STATIC static
107 /* Valid only for non-utf8 strings: avoids the reginclass
108 * call if there are no complications: i.e., if everything matchable is
109 * straight forward in the bitmap */
110 #define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,c+1,0) \
111 : ANYOF_BITMAP_TEST(p,*(c)))
117 #define CHR_SVLEN(sv) (utf8_target ? sv_len_utf8(sv) : SvCUR(sv))
118 #define CHR_DIST(a,b) (reginfo->is_utf8_target ? utf8_distance(a,b) : a - b)
120 #define HOPc(pos,off) \
121 (char *)(reginfo->is_utf8_target \
122 ? reghop3((U8*)pos, off, \
123 (U8*)(off >= 0 ? reginfo->strend : reginfo->strbeg)) \
126 #define HOPBACKc(pos, off) \
127 (char*)(reginfo->is_utf8_target \
128 ? reghopmaybe3((U8*)pos, -off, (U8*)(reginfo->strbeg)) \
129 : (pos - off >= reginfo->strbeg) \
133 #define HOP3(pos,off,lim) (reginfo->is_utf8_target ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off))
134 #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
136 /* lim must be +ve. Returns NULL on overshoot */
137 #define HOPMAYBE3(pos,off,lim) \
138 (reginfo->is_utf8_target \
139 ? reghopmaybe3((U8*)pos, off, (U8*)(lim)) \
140 : ((U8*)pos + off <= lim) \
144 /* like HOP3, but limits the result to <= lim even for the non-utf8 case.
145 * off must be >=0; args should be vars rather than expressions */
146 #define HOP3lim(pos,off,lim) (reginfo->is_utf8_target \
147 ? reghop3((U8*)(pos), off, (U8*)(lim)) \
148 : (U8*)((pos + off) > lim ? lim : (pos + off)))
150 #define HOP4(pos,off,llim, rlim) (reginfo->is_utf8_target \
151 ? reghop4((U8*)(pos), off, (U8*)(llim), (U8*)(rlim)) \
153 #define HOP4c(pos,off,llim, rlim) ((char*)HOP4(pos,off,llim, rlim))
155 #define NEXTCHR_EOS -10 /* nextchr has fallen off the end */
156 #define NEXTCHR_IS_EOS (nextchr < 0)
158 #define SET_nextchr \
159 nextchr = ((locinput < reginfo->strend) ? UCHARAT(locinput) : NEXTCHR_EOS)
161 #define SET_locinput(p) \
166 #define LOAD_UTF8_CHARCLASS(swash_ptr, property_name, invlist) STMT_START { \
168 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST; \
169 swash_ptr = _core_swash_init("utf8", property_name, &PL_sv_undef, \
170 1, 0, invlist, &flags); \
175 /* If in debug mode, we test that a known character properly matches */
177 # define LOAD_UTF8_CHARCLASS_DEBUG_TEST(swash_ptr, \
180 utf8_char_in_property) \
181 LOAD_UTF8_CHARCLASS(swash_ptr, property_name, invlist); \
182 assert(swash_fetch(swash_ptr, (U8 *) utf8_char_in_property, TRUE));
184 # define LOAD_UTF8_CHARCLASS_DEBUG_TEST(swash_ptr, \
187 utf8_char_in_property) \
188 LOAD_UTF8_CHARCLASS(swash_ptr, property_name, invlist)
191 #define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS_DEBUG_TEST( \
192 PL_utf8_swash_ptrs[_CC_WORDCHAR], \
194 PL_XPosix_ptrs[_CC_WORDCHAR], \
195 LATIN_CAPITAL_LETTER_SHARP_S_UTF8);
197 #define PLACEHOLDER /* Something for the preprocessor to grab onto */
198 /* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */
200 /* for use after a quantifier and before an EXACT-like node -- japhy */
201 /* it would be nice to rework regcomp.sym to generate this stuff. sigh
203 * NOTE that *nothing* that affects backtracking should be in here, specifically
204 * VERBS must NOT be included. JUMPABLE is used to determine if we can ignore a
205 * node that is in between two EXACT like nodes when ascertaining what the required
206 * "follow" character is. This should probably be moved to regex compile time
207 * although it may be done at run time beause of the REF possibility - more
208 * investigation required. -- demerphq
210 #define JUMPABLE(rn) ( \
212 (OP(rn) == CLOSE && (!cur_eval || cur_eval->u.eval.close_paren != ARG(rn))) || \
214 OP(rn) == SUSPEND || OP(rn) == IFMATCH || \
215 OP(rn) == PLUS || OP(rn) == MINMOD || \
217 (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \
219 #define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT)
221 #define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF )
224 /* Currently these are only used when PL_regkind[OP(rn)] == EXACT so
225 we don't need this definition. XXX These are now out-of-sync*/
226 #define IS_TEXT(rn) ( OP(rn)==EXACT || OP(rn)==REF || OP(rn)==NREF )
227 #define IS_TEXTF(rn) ( OP(rn)==EXACTFU || OP(rn)==EXACTFU_SS || OP(rn)==EXACTFA || OP(rn)==EXACTFA_NO_TRIE || OP(rn)==EXACTF || OP(rn)==REFF || OP(rn)==NREFF )
228 #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL )
231 /* ... so we use this as its faster. */
232 #define IS_TEXT(rn) ( OP(rn)==EXACT || OP(rn)==EXACTL )
233 #define IS_TEXTFU(rn) ( OP(rn)==EXACTFU || OP(rn)==EXACTFLU8 || OP(rn)==EXACTFU_SS || OP(rn) == EXACTFA || OP(rn) == EXACTFA_NO_TRIE)
234 #define IS_TEXTF(rn) ( OP(rn)==EXACTF )
235 #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL )
240 Search for mandatory following text node; for lookahead, the text must
241 follow but for lookbehind (rn->flags != 0) we skip to the next step.
243 #define FIND_NEXT_IMPT(rn) STMT_START { \
244 while (JUMPABLE(rn)) { \
245 const OPCODE type = OP(rn); \
246 if (type == SUSPEND || PL_regkind[type] == CURLY) \
247 rn = NEXTOPER(NEXTOPER(rn)); \
248 else if (type == PLUS) \
250 else if (type == IFMATCH) \
251 rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \
252 else rn += NEXT_OFF(rn); \
256 #define SLAB_FIRST(s) (&(s)->states[0])
257 #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
259 static void S_setup_eval_state(pTHX_ regmatch_info *const reginfo);
260 static void S_cleanup_regmatch_info_aux(pTHX_ void *arg);
261 static regmatch_state * S_push_slab(pTHX);
263 #define REGCP_PAREN_ELEMS 3
264 #define REGCP_OTHER_ELEMS 3
265 #define REGCP_FRAME_ELEMS 1
266 /* REGCP_FRAME_ELEMS are not part of the REGCP_OTHER_ELEMS and
267 * are needed for the regexp context stack bookkeeping. */
270 S_regcppush(pTHX_ const regexp *rex, I32 parenfloor, U32 maxopenparen)
272 const int retval = PL_savestack_ix;
273 const int paren_elems_to_push =
274 (maxopenparen - parenfloor) * REGCP_PAREN_ELEMS;
275 const UV total_elems = paren_elems_to_push + REGCP_OTHER_ELEMS;
276 const UV elems_shifted = total_elems << SAVE_TIGHT_SHIFT;
278 GET_RE_DEBUG_FLAGS_DECL;
280 PERL_ARGS_ASSERT_REGCPPUSH;
282 if (paren_elems_to_push < 0)
283 Perl_croak(aTHX_ "panic: paren_elems_to_push, %i < 0, maxopenparen: %i parenfloor: %i REGCP_PAREN_ELEMS: %u",
284 (int)paren_elems_to_push, (int)maxopenparen,
285 (int)parenfloor, (unsigned)REGCP_PAREN_ELEMS);
287 if ((elems_shifted >> SAVE_TIGHT_SHIFT) != total_elems)
288 Perl_croak(aTHX_ "panic: paren_elems_to_push offset %"UVuf
289 " out of range (%lu-%ld)",
291 (unsigned long)maxopenparen,
294 SSGROW(total_elems + REGCP_FRAME_ELEMS);
297 if ((int)maxopenparen > (int)parenfloor)
298 PerlIO_printf(Perl_debug_log,
299 "rex=0x%"UVxf" offs=0x%"UVxf": saving capture indices:\n",
304 for (p = parenfloor+1; p <= (I32)maxopenparen; p++) {
305 /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
306 SSPUSHIV(rex->offs[p].end);
307 SSPUSHIV(rex->offs[p].start);
308 SSPUSHINT(rex->offs[p].start_tmp);
309 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
310 " \\%"UVuf": %"IVdf"(%"IVdf")..%"IVdf"\n",
312 (IV)rex->offs[p].start,
313 (IV)rex->offs[p].start_tmp,
317 /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */
318 SSPUSHINT(maxopenparen);
319 SSPUSHINT(rex->lastparen);
320 SSPUSHINT(rex->lastcloseparen);
321 SSPUSHUV(SAVEt_REGCONTEXT | elems_shifted); /* Magic cookie. */
326 /* These are needed since we do not localize EVAL nodes: */
327 #define REGCP_SET(cp) \
329 PerlIO_printf(Perl_debug_log, \
330 " Setting an EVAL scope, savestack=%"IVdf"\n", \
331 (IV)PL_savestack_ix)); \
334 #define REGCP_UNWIND(cp) \
336 if (cp != PL_savestack_ix) \
337 PerlIO_printf(Perl_debug_log, \
338 " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
339 (IV)(cp), (IV)PL_savestack_ix)); \
342 #define UNWIND_PAREN(lp, lcp) \
343 for (n = rex->lastparen; n > lp; n--) \
344 rex->offs[n].end = -1; \
345 rex->lastparen = n; \
346 rex->lastcloseparen = lcp;
350 S_regcppop(pTHX_ regexp *rex, U32 *maxopenparen_p)
354 GET_RE_DEBUG_FLAGS_DECL;
356 PERL_ARGS_ASSERT_REGCPPOP;
358 /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
360 assert((i & SAVE_MASK) == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */
361 i >>= SAVE_TIGHT_SHIFT; /* Parentheses elements to pop. */
362 rex->lastcloseparen = SSPOPINT;
363 rex->lastparen = SSPOPINT;
364 *maxopenparen_p = SSPOPINT;
366 i -= REGCP_OTHER_ELEMS;
367 /* Now restore the parentheses context. */
369 if (i || rex->lastparen + 1 <= rex->nparens)
370 PerlIO_printf(Perl_debug_log,
371 "rex=0x%"UVxf" offs=0x%"UVxf": restoring capture indices to:\n",
376 paren = *maxopenparen_p;
377 for ( ; i > 0; i -= REGCP_PAREN_ELEMS) {
379 rex->offs[paren].start_tmp = SSPOPINT;
380 rex->offs[paren].start = SSPOPIV;
382 if (paren <= rex->lastparen)
383 rex->offs[paren].end = tmps;
384 DEBUG_BUFFERS_r( PerlIO_printf(Perl_debug_log,
385 " \\%"UVuf": %"IVdf"(%"IVdf")..%"IVdf"%s\n",
387 (IV)rex->offs[paren].start,
388 (IV)rex->offs[paren].start_tmp,
389 (IV)rex->offs[paren].end,
390 (paren > rex->lastparen ? "(skipped)" : ""));
395 /* It would seem that the similar code in regtry()
396 * already takes care of this, and in fact it is in
397 * a better location to since this code can #if 0-ed out
398 * but the code in regtry() is needed or otherwise tests
399 * requiring null fields (pat.t#187 and split.t#{13,14}
400 * (as of patchlevel 7877) will fail. Then again,
401 * this code seems to be necessary or otherwise
402 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
403 * --jhi updated by dapm */
404 for (i = rex->lastparen + 1; i <= rex->nparens; i++) {
405 if (i > *maxopenparen_p)
406 rex->offs[i].start = -1;
407 rex->offs[i].end = -1;
408 DEBUG_BUFFERS_r( PerlIO_printf(Perl_debug_log,
409 " \\%"UVuf": %s ..-1 undeffing\n",
411 (i > *maxopenparen_p) ? "-1" : " "
417 /* restore the parens and associated vars at savestack position ix,
418 * but without popping the stack */
421 S_regcp_restore(pTHX_ regexp *rex, I32 ix, U32 *maxopenparen_p)
423 I32 tmpix = PL_savestack_ix;
424 PL_savestack_ix = ix;
425 regcppop(rex, maxopenparen_p);
426 PL_savestack_ix = tmpix;
429 #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
432 S_isFOO_lc(pTHX_ const U8 classnum, const U8 character)
434 /* Returns a boolean as to whether or not 'character' is a member of the
435 * Posix character class given by 'classnum' that should be equivalent to a
436 * value in the typedef '_char_class_number'.
438 * Ideally this could be replaced by a just an array of function pointers
439 * to the C library functions that implement the macros this calls.
440 * However, to compile, the precise function signatures are required, and
441 * these may vary from platform to to platform. To avoid having to figure
442 * out what those all are on each platform, I (khw) am using this method,
443 * which adds an extra layer of function call overhead (unless the C
444 * optimizer strips it away). But we don't particularly care about
445 * performance with locales anyway. */
447 switch ((_char_class_number) classnum) {
448 case _CC_ENUM_ALPHANUMERIC: return isALPHANUMERIC_LC(character);
449 case _CC_ENUM_ALPHA: return isALPHA_LC(character);
450 case _CC_ENUM_ASCII: return isASCII_LC(character);
451 case _CC_ENUM_BLANK: return isBLANK_LC(character);
452 case _CC_ENUM_CASED: return isLOWER_LC(character)
453 || isUPPER_LC(character);
454 case _CC_ENUM_CNTRL: return isCNTRL_LC(character);
455 case _CC_ENUM_DIGIT: return isDIGIT_LC(character);
456 case _CC_ENUM_GRAPH: return isGRAPH_LC(character);
457 case _CC_ENUM_LOWER: return isLOWER_LC(character);
458 case _CC_ENUM_PRINT: return isPRINT_LC(character);
459 case _CC_ENUM_PUNCT: return isPUNCT_LC(character);
460 case _CC_ENUM_SPACE: return isSPACE_LC(character);
461 case _CC_ENUM_UPPER: return isUPPER_LC(character);
462 case _CC_ENUM_WORDCHAR: return isWORDCHAR_LC(character);
463 case _CC_ENUM_XDIGIT: return isXDIGIT_LC(character);
464 default: /* VERTSPACE should never occur in locales */
465 Perl_croak(aTHX_ "panic: isFOO_lc() has an unexpected character class '%d'", classnum);
468 NOT_REACHED; /* NOTREACHED */
473 S_isFOO_utf8_lc(pTHX_ const U8 classnum, const U8* character)
475 /* Returns a boolean as to whether or not the (well-formed) UTF-8-encoded
476 * 'character' is a member of the Posix character class given by 'classnum'
477 * that should be equivalent to a value in the typedef
478 * '_char_class_number'.
480 * This just calls isFOO_lc on the code point for the character if it is in
481 * the range 0-255. Outside that range, all characters use Unicode
482 * rules, ignoring any locale. So use the Unicode function if this class
483 * requires a swash, and use the Unicode macro otherwise. */
485 PERL_ARGS_ASSERT_ISFOO_UTF8_LC;
487 if (UTF8_IS_INVARIANT(*character)) {
488 return isFOO_lc(classnum, *character);
490 else if (UTF8_IS_DOWNGRADEABLE_START(*character)) {
491 return isFOO_lc(classnum,
492 TWO_BYTE_UTF8_TO_NATIVE(*character, *(character + 1)));
495 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(character, character + UTF8SKIP(character));
497 if (classnum < _FIRST_NON_SWASH_CC) {
499 /* Initialize the swash unless done already */
500 if (! PL_utf8_swash_ptrs[classnum]) {
501 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
502 PL_utf8_swash_ptrs[classnum] =
503 _core_swash_init("utf8",
506 PL_XPosix_ptrs[classnum], &flags);
509 return cBOOL(swash_fetch(PL_utf8_swash_ptrs[classnum], (U8 *)
511 TRUE /* is UTF */ ));
514 switch ((_char_class_number) classnum) {
515 case _CC_ENUM_SPACE: return is_XPERLSPACE_high(character);
516 case _CC_ENUM_BLANK: return is_HORIZWS_high(character);
517 case _CC_ENUM_XDIGIT: return is_XDIGIT_high(character);
518 case _CC_ENUM_VERTSPACE: return is_VERTWS_high(character);
522 return FALSE; /* Things like CNTRL are always below 256 */
526 * pregexec and friends
529 #ifndef PERL_IN_XSUB_RE
531 - pregexec - match a regexp against a string
534 Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, char *strend,
535 char *strbeg, SSize_t minend, SV *screamer, U32 nosave)
536 /* stringarg: the point in the string at which to begin matching */
537 /* strend: pointer to null at end of string */
538 /* strbeg: real beginning of string */
539 /* minend: end of match must be >= minend bytes after stringarg. */
540 /* screamer: SV being matched: only used for utf8 flag, pos() etc; string
541 * itself is accessed via the pointers above */
542 /* nosave: For optimizations. */
544 PERL_ARGS_ASSERT_PREGEXEC;
547 regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
548 nosave ? 0 : REXEC_COPY_STR);
554 /* re_intuit_start():
556 * Based on some optimiser hints, try to find the earliest position in the
557 * string where the regex could match.
559 * rx: the regex to match against
560 * sv: the SV being matched: only used for utf8 flag; the string
561 * itself is accessed via the pointers below. Note that on
562 * something like an overloaded SV, SvPOK(sv) may be false
563 * and the string pointers may point to something unrelated to
565 * strbeg: real beginning of string
566 * strpos: the point in the string at which to begin matching
567 * strend: pointer to the byte following the last char of the string
568 * flags currently unused; set to 0
569 * data: currently unused; set to NULL
571 * The basic idea of re_intuit_start() is to use some known information
572 * about the pattern, namely:
574 * a) the longest known anchored substring (i.e. one that's at a
575 * constant offset from the beginning of the pattern; but not
576 * necessarily at a fixed offset from the beginning of the
578 * b) the longest floating substring (i.e. one that's not at a constant
579 * offset from the beginning of the pattern);
580 * c) Whether the pattern is anchored to the string; either
581 * an absolute anchor: /^../, or anchored to \n: /^.../m,
582 * or anchored to pos(): /\G/;
583 * d) A start class: a real or synthetic character class which
584 * represents which characters are legal at the start of the pattern;
586 * to either quickly reject the match, or to find the earliest position
587 * within the string at which the pattern might match, thus avoiding
588 * running the full NFA engine at those earlier locations, only to
589 * eventually fail and retry further along.
591 * Returns NULL if the pattern can't match, or returns the address within
592 * the string which is the earliest place the match could occur.
594 * The longest of the anchored and floating substrings is called 'check'
595 * and is checked first. The other is called 'other' and is checked
596 * second. The 'other' substring may not be present. For example,
598 * /(abc|xyz)ABC\d{0,3}DEFG/
602 * check substr (float) = "DEFG", offset 6..9 chars
603 * other substr (anchored) = "ABC", offset 3..3 chars
606 * Be aware that during the course of this function, sometimes 'anchored'
607 * refers to a substring being anchored relative to the start of the
608 * pattern, and sometimes to the pattern itself being anchored relative to
609 * the string. For example:
611 * /\dabc/: "abc" is anchored to the pattern;
612 * /^\dabc/: "abc" is anchored to the pattern and the string;
613 * /\d+abc/: "abc" is anchored to neither the pattern nor the string;
614 * /^\d+abc/: "abc" is anchored to neither the pattern nor the string,
615 * but the pattern is anchored to the string.
619 Perl_re_intuit_start(pTHX_
622 const char * const strbeg,
626 re_scream_pos_data *data)
628 struct regexp *const prog = ReANY(rx);
629 SSize_t start_shift = prog->check_offset_min;
630 /* Should be nonnegative! */
631 SSize_t end_shift = 0;
632 /* current lowest pos in string where the regex can start matching */
633 char *rx_origin = strpos;
635 const bool utf8_target = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */
636 U8 other_ix = 1 - prog->substrs->check_ix;
638 char *other_last = strpos;/* latest pos 'other' substr already checked to */
639 char *check_at = NULL; /* check substr found at this pos */
640 const I32 multiline = prog->extflags & RXf_PMf_MULTILINE;
641 RXi_GET_DECL(prog,progi);
642 regmatch_info reginfo_buf; /* create some info to pass to find_byclass */
643 regmatch_info *const reginfo = ®info_buf;
644 GET_RE_DEBUG_FLAGS_DECL;
646 PERL_ARGS_ASSERT_RE_INTUIT_START;
647 PERL_UNUSED_ARG(flags);
648 PERL_UNUSED_ARG(data);
650 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
651 "Intuit: trying to determine minimum start position...\n"));
653 /* for now, assume that all substr offsets are positive. If at some point
654 * in the future someone wants to do clever things with look-behind and
655 * -ve offsets, they'll need to fix up any code in this function
656 * which uses these offsets. See the thread beginning
657 * <20140113145929.GF27210@iabyn.com>
659 assert(prog->substrs->data[0].min_offset >= 0);
660 assert(prog->substrs->data[0].max_offset >= 0);
661 assert(prog->substrs->data[1].min_offset >= 0);
662 assert(prog->substrs->data[1].max_offset >= 0);
663 assert(prog->substrs->data[2].min_offset >= 0);
664 assert(prog->substrs->data[2].max_offset >= 0);
666 /* for now, assume that if both present, that the floating substring
667 * doesn't start before the anchored substring.
668 * If you break this assumption (e.g. doing better optimisations
669 * with lookahead/behind), then you'll need to audit the code in this
670 * function carefully first
673 ! ( (prog->anchored_utf8 || prog->anchored_substr)
674 && (prog->float_utf8 || prog->float_substr))
675 || (prog->float_min_offset >= prog->anchored_offset));
677 /* byte rather than char calculation for efficiency. It fails
678 * to quickly reject some cases that can't match, but will reject
679 * them later after doing full char arithmetic */
680 if (prog->minlen > strend - strpos) {
681 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
682 " String too short...\n"));
686 RX_MATCH_UTF8_set(rx,utf8_target);
687 reginfo->is_utf8_target = cBOOL(utf8_target);
688 reginfo->info_aux = NULL;
689 reginfo->strbeg = strbeg;
690 reginfo->strend = strend;
691 reginfo->is_utf8_pat = cBOOL(RX_UTF8(rx));
693 /* not actually used within intuit, but zero for safety anyway */
694 reginfo->poscache_maxiter = 0;
697 if (!prog->check_utf8 && prog->check_substr)
698 to_utf8_substr(prog);
699 check = prog->check_utf8;
701 if (!prog->check_substr && prog->check_utf8) {
702 if (! to_byte_substr(prog)) {
703 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(fail);
706 check = prog->check_substr;
709 /* dump the various substring data */
710 DEBUG_OPTIMISE_MORE_r({
712 for (i=0; i<=2; i++) {
713 SV *sv = (utf8_target ? prog->substrs->data[i].utf8_substr
714 : prog->substrs->data[i].substr);
718 PerlIO_printf(Perl_debug_log,
719 " substrs[%d]: min=%"IVdf" max=%"IVdf" end shift=%"IVdf
720 " useful=%"IVdf" utf8=%d [%s]\n",
722 (IV)prog->substrs->data[i].min_offset,
723 (IV)prog->substrs->data[i].max_offset,
724 (IV)prog->substrs->data[i].end_shift,
731 if (prog->intflags & PREGf_ANCH) { /* Match at \G, beg-of-str or after \n */
733 /* ml_anch: check after \n?
735 * A note about PREGf_IMPLICIT: on an un-anchored pattern beginning
736 * with /.*.../, these flags will have been added by the
738 * /.*abc/, /.*abc/m: PREGf_IMPLICIT | PREGf_ANCH_MBOL
739 * /.*abc/s: PREGf_IMPLICIT | PREGf_ANCH_SBOL
741 ml_anch = (prog->intflags & PREGf_ANCH_MBOL)
742 && !(prog->intflags & PREGf_IMPLICIT);
744 if (!ml_anch && !(prog->intflags & PREGf_IMPLICIT)) {
745 /* we are only allowed to match at BOS or \G */
747 /* trivially reject if there's a BOS anchor and we're not at BOS.
749 * Note that we don't try to do a similar quick reject for
750 * \G, since generally the caller will have calculated strpos
751 * based on pos() and gofs, so the string is already correctly
752 * anchored by definition; and handling the exceptions would
753 * be too fiddly (e.g. REXEC_IGNOREPOS).
755 if ( strpos != strbeg
756 && (prog->intflags & PREGf_ANCH_SBOL))
758 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
759 " Not at start...\n"));
763 /* in the presence of an anchor, the anchored (relative to the
764 * start of the regex) substr must also be anchored relative
765 * to strpos. So quickly reject if substr isn't found there.
766 * This works for \G too, because the caller will already have
767 * subtracted gofs from pos, and gofs is the offset from the
768 * \G to the start of the regex. For example, in /.abc\Gdef/,
769 * where substr="abcdef", pos()=3, gofs=4, offset_min=1:
770 * caller will have set strpos=pos()-4; we look for the substr
771 * at position pos()-4+1, which lines up with the "a" */
773 if (prog->check_offset_min == prog->check_offset_max
774 && !(prog->intflags & PREGf_CANY_SEEN))
776 /* Substring at constant offset from beg-of-str... */
777 SSize_t slen = SvCUR(check);
778 char *s = HOP3c(strpos, prog->check_offset_min, strend);
780 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
781 " Looking for check substr at fixed offset %"IVdf"...\n",
782 (IV)prog->check_offset_min));
785 /* In this case, the regex is anchored at the end too.
786 * Unless it's a multiline match, the lengths must match
787 * exactly, give or take a \n. NB: slen >= 1 since
788 * the last char of check is \n */
790 && ( strend - s > slen
791 || strend - s < slen - 1
792 || (strend - s == slen && strend[-1] != '\n')))
794 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
795 " String too long...\n"));
798 /* Now should match s[0..slen-2] */
801 if (slen && (*SvPVX_const(check) != *s
802 || (slen > 1 && memNE(SvPVX_const(check), s, slen))))
804 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
805 " String not equal...\n"));
810 goto success_at_start;
815 end_shift = prog->check_end_shift;
817 #ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */
819 Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ",
820 (IV)end_shift, RX_PRECOMP(prog));
825 /* This is the (re)entry point of the main loop in this function.
826 * The goal of this loop is to:
827 * 1) find the "check" substring in the region rx_origin..strend
828 * (adjusted by start_shift / end_shift). If not found, reject
830 * 2) If it exists, look for the "other" substr too if defined; for
831 * example, if the check substr maps to the anchored substr, then
832 * check the floating substr, and vice-versa. If not found, go
833 * back to (1) with rx_origin suitably incremented.
834 * 3) If we find an rx_origin position that doesn't contradict
835 * either of the substrings, then check the possible additional
836 * constraints on rx_origin of /^.../m or a known start class.
837 * If these fail, then depending on which constraints fail, jump
838 * back to here, or to various other re-entry points further along
839 * that skip some of the first steps.
840 * 4) If we pass all those tests, update the BmUSEFUL() count on the
841 * substring. If the start position was determined to be at the
842 * beginning of the string - so, not rejected, but not optimised,
843 * since we have to run regmatch from position 0 - decrement the
844 * BmUSEFUL() count. Otherwise increment it.
848 /* first, look for the 'check' substring */
854 DEBUG_OPTIMISE_MORE_r({
855 PerlIO_printf(Perl_debug_log,
856 " At restart: rx_origin=%"IVdf" Check offset min: %"IVdf
857 " Start shift: %"IVdf" End shift %"IVdf
858 " Real end Shift: %"IVdf"\n",
859 (IV)(rx_origin - strpos),
860 (IV)prog->check_offset_min,
863 (IV)prog->check_end_shift);
866 if (prog->intflags & PREGf_CANY_SEEN) {
867 start_point= (U8*)(rx_origin + start_shift);
868 end_point= (U8*)(strend - end_shift);
869 if (start_point > end_point)
872 end_point = HOP3(strend, -end_shift, strbeg);
873 start_point = HOPMAYBE3(rx_origin, start_shift, end_point);
879 /* If the regex is absolutely anchored to either the start of the
880 * string (SBOL) or to pos() (ANCH_GPOS), then
881 * check_offset_max represents an upper bound on the string where
882 * the substr could start. For the ANCH_GPOS case, we assume that
883 * the caller of intuit will have already set strpos to
884 * pos()-gofs, so in this case strpos + offset_max will still be
885 * an upper bound on the substr.
888 && prog->intflags & PREGf_ANCH
889 && prog->check_offset_max != SSize_t_MAX)
891 SSize_t len = SvCUR(check) - !!SvTAIL(check);
892 const char * const anchor =
893 (prog->intflags & PREGf_ANCH_GPOS ? strpos : strbeg);
895 /* do a bytes rather than chars comparison. It's conservative;
896 * so it skips doing the HOP if the result can't possibly end
897 * up earlier than the old value of end_point.
899 if ((char*)end_point - anchor > prog->check_offset_max) {
900 end_point = HOP3lim((U8*)anchor,
901 prog->check_offset_max,
907 DEBUG_OPTIMISE_MORE_r({
908 PerlIO_printf(Perl_debug_log, " fbm_instr len=%d str=<%.*s>\n",
909 (int)(end_point - start_point),
910 (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point),
914 check_at = fbm_instr( start_point, end_point,
915 check, multiline ? FBMrf_MULTILINE : 0);
917 /* Update the count-of-usability, remove useless subpatterns,
921 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
922 SvPVX_const(check), RE_SV_DUMPLEN(check), 30);
923 PerlIO_printf(Perl_debug_log, " %s %s substr %s%s%s",
924 (check_at ? "Found" : "Did not find"),
925 (check == (utf8_target ? prog->anchored_utf8 : prog->anchored_substr)
926 ? "anchored" : "floating"),
929 (check_at ? " at offset " : "...\n") );
934 /* Finish the diagnostic message */
935 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(check_at - strpos)) );
937 /* set rx_origin to the minimum position where the regex could start
938 * matching, given the constraint of the just-matched check substring.
939 * But don't set it lower than previously.
942 if (check_at - rx_origin > prog->check_offset_max)
943 rx_origin = HOP3c(check_at, -prog->check_offset_max, rx_origin);
947 /* now look for the 'other' substring if defined */
949 if (utf8_target ? prog->substrs->data[other_ix].utf8_substr
950 : prog->substrs->data[other_ix].substr)
952 /* Take into account the "other" substring. */
956 struct reg_substr_datum *other;
959 other = &prog->substrs->data[other_ix];
961 /* if "other" is anchored:
962 * we've previously found a floating substr starting at check_at.
963 * This means that the regex origin must lie somewhere
964 * between min (rx_origin): HOP3(check_at, -check_offset_max)
965 * and max: HOP3(check_at, -check_offset_min)
966 * (except that min will be >= strpos)
967 * So the fixed substr must lie somewhere between
968 * HOP3(min, anchored_offset)
969 * HOP3(max, anchored_offset) + SvCUR(substr)
972 /* if "other" is floating
973 * Calculate last1, the absolute latest point where the
974 * floating substr could start in the string, ignoring any
975 * constraints from the earlier fixed match. It is calculated
978 * strend - prog->minlen (in chars) is the absolute latest
979 * position within the string where the origin of the regex
980 * could appear. The latest start point for the floating
981 * substr is float_min_offset(*) on from the start of the
982 * regex. last1 simply combines thee two offsets.
984 * (*) You might think the latest start point should be
985 * float_max_offset from the regex origin, and technically
986 * you'd be correct. However, consider
988 * Here, float min, max are 3,5 and minlen is 7.
989 * This can match either
993 * In the first case, the regex matches minlen chars; in the
994 * second, minlen+1, in the third, minlen+2.
995 * In the first case, the floating offset is 3 (which equals
996 * float_min), in the second, 4, and in the third, 5 (which
997 * equals float_max). In all cases, the floating string bcd
998 * can never start more than 4 chars from the end of the
999 * string, which equals minlen - float_min. As the substring
1000 * starts to match more than float_min from the start of the
1001 * regex, it makes the regex match more than minlen chars,
1002 * and the two cancel each other out. So we can always use
1003 * float_min - minlen, rather than float_max - minlen for the
1004 * latest position in the string.
1006 * Note that -minlen + float_min_offset is equivalent (AFAIKT)
1007 * to CHR_SVLEN(must) - !!SvTAIL(must) + prog->float_end_shift
1010 assert(prog->minlen >= other->min_offset);
1011 last1 = HOP3c(strend,
1012 other->min_offset - prog->minlen, strbeg);
1014 if (other_ix) {/* i.e. if (other-is-float) */
1015 /* last is the latest point where the floating substr could
1016 * start, *given* any constraints from the earlier fixed
1017 * match. This constraint is that the floating string starts
1018 * <= float_max_offset chars from the regex origin (rx_origin).
1019 * If this value is less than last1, use it instead.
1021 assert(rx_origin <= last1);
1023 /* this condition handles the offset==infinity case, and
1024 * is a short-cut otherwise. Although it's comparing a
1025 * byte offset to a char length, it does so in a safe way,
1026 * since 1 char always occupies 1 or more bytes,
1027 * so if a string range is (last1 - rx_origin) bytes,
1028 * it will be less than or equal to (last1 - rx_origin)
1029 * chars; meaning it errs towards doing the accurate HOP3
1030 * rather than just using last1 as a short-cut */
1031 (last1 - rx_origin) < other->max_offset
1033 : (char*)HOP3lim(rx_origin, other->max_offset, last1);
1036 assert(strpos + start_shift <= check_at);
1037 last = HOP4c(check_at, other->min_offset - start_shift,
1041 s = HOP3c(rx_origin, other->min_offset, strend);
1042 if (s < other_last) /* These positions already checked */
1045 must = utf8_target ? other->utf8_substr : other->substr;
1046 assert(SvPOK(must));
1049 (unsigned char*)last + SvCUR(must) - (SvTAIL(must)!=0),
1051 multiline ? FBMrf_MULTILINE : 0
1054 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
1055 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
1056 PerlIO_printf(Perl_debug_log, " %s %s substr %s%s",
1057 s ? "Found" : "Contradicts",
1058 other_ix ? "floating" : "anchored",
1059 quoted, RE_SV_TAIL(must));
1064 /* last1 is latest possible substr location. If we didn't
1065 * find it before there, we never will */
1066 if (last >= last1) {
1067 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1068 ", giving up...\n"));
1072 /* try to find the check substr again at a later
1073 * position. Maybe next time we'll find the "other" substr
1075 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1076 ", trying %s at offset %ld...\n",
1077 (other_ix ? "floating" : "anchored"),
1078 (long)(HOP3c(check_at, 1, strend) - strpos)));
1080 other_last = HOP3c(last, 1, strend) /* highest failure */;
1082 other_ix /* i.e. if other-is-float */
1083 ? HOP3c(rx_origin, 1, strend)
1084 : HOP4c(last, 1 - other->min_offset, strbeg, strend);
1088 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
1089 (long)(s - strpos)));
1091 if (other_ix) { /* if (other-is-float) */
1092 /* other_last is set to s, not s+1, since its possible for
1093 * a floating substr to fail first time, then succeed
1094 * second time at the same floating position; e.g.:
1095 * "-AB--AABZ" =~ /\wAB\d*Z/
1096 * The first time round, anchored and float match at
1097 * "-(AB)--AAB(Z)" then fail on the initial \w character
1098 * class. Second time round, they match at "-AB--A(AB)(Z)".
1103 rx_origin = HOP3c(s, -other->min_offset, strbeg);
1104 other_last = HOP3c(s, 1, strend);
1109 DEBUG_OPTIMISE_MORE_r(
1110 PerlIO_printf(Perl_debug_log,
1111 " Check-only match: offset min:%"IVdf" max:%"IVdf
1112 " check_at:%"IVdf" rx_origin:%"IVdf" rx_origin-check_at:%"IVdf
1113 " strend-strpos:%"IVdf"\n",
1114 (IV)prog->check_offset_min,
1115 (IV)prog->check_offset_max,
1116 (IV)(check_at-strpos),
1117 (IV)(rx_origin-strpos),
1118 (IV)(rx_origin-check_at),
1124 postprocess_substr_matches:
1126 /* handle the extra constraint of /^.../m if present */
1128 if (ml_anch && rx_origin != strbeg && rx_origin[-1] != '\n') {
1131 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1132 " looking for /^/m anchor"));
1134 /* we have failed the constraint of a \n before rx_origin.
1135 * Find the next \n, if any, even if it's beyond the current
1136 * anchored and/or floating substrings. Whether we should be
1137 * scanning ahead for the next \n or the next substr is debatable.
1138 * On the one hand you'd expect rare substrings to appear less
1139 * often than \n's. On the other hand, searching for \n means
1140 * we're effectively flipping been check_substr and "\n" on each
1141 * iteration as the current "rarest" string candidate, which
1142 * means for example that we'll quickly reject the whole string if
1143 * hasn't got a \n, rather than trying every substr position
1147 s = HOP3c(strend, - prog->minlen, strpos);
1148 if (s <= rx_origin ||
1149 ! ( rx_origin = (char *)memchr(rx_origin, '\n', s - rx_origin)))
1151 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1152 " Did not find /%s^%s/m...\n",
1153 PL_colors[0], PL_colors[1]));
1157 /* earliest possible origin is 1 char after the \n.
1158 * (since *rx_origin == '\n', it's safe to ++ here rather than
1159 * HOP(rx_origin, 1)) */
1162 if (prog->substrs->check_ix == 0 /* check is anchored */
1163 || rx_origin >= HOP3c(check_at, - prog->check_offset_min, strpos))
1165 /* Position contradicts check-string; either because
1166 * check was anchored (and thus has no wiggle room),
1167 * or check was float and rx_origin is above the float range */
1168 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1169 " Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
1170 PL_colors[0], PL_colors[1], (long)(rx_origin - strpos)));
1174 /* if we get here, the check substr must have been float,
1175 * is in range, and we may or may not have had an anchored
1176 * "other" substr which still contradicts */
1177 assert(prog->substrs->check_ix); /* check is float */
1179 if (utf8_target ? prog->anchored_utf8 : prog->anchored_substr) {
1180 /* whoops, the anchored "other" substr exists, so we still
1181 * contradict. On the other hand, the float "check" substr
1182 * didn't contradict, so just retry the anchored "other"
1184 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1185 " Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
1186 PL_colors[0], PL_colors[1],
1187 (long)(rx_origin - strpos),
1188 (long)(rx_origin - strpos + prog->anchored_offset)));
1189 goto do_other_substr;
1192 /* success: we don't contradict the found floating substring
1193 * (and there's no anchored substr). */
1194 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1195 " Found /%s^%s/m at offset %ld...\n",
1196 PL_colors[0], PL_colors[1], (long)(rx_origin - strpos)));
1199 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1200 " (multiline anchor test skipped)\n"));
1206 /* if we have a starting character class, then test that extra constraint.
1207 * (trie stclasses are too expensive to use here, we are better off to
1208 * leave it to regmatch itself) */
1210 if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) {
1211 const U8* const str = (U8*)STRING(progi->regstclass);
1213 /* XXX this value could be pre-computed */
1214 const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
1215 ? (reginfo->is_utf8_pat
1216 ? utf8_distance(str + STR_LEN(progi->regstclass), str)
1217 : STR_LEN(progi->regstclass))
1221 /* latest pos that a matching float substr constrains rx start to */
1222 char *rx_max_float = NULL;
1224 /* if the current rx_origin is anchored, either by satisfying an
1225 * anchored substring constraint, or a /^.../m constraint, then we
1226 * can reject the current origin if the start class isn't found
1227 * at the current position. If we have a float-only match, then
1228 * rx_origin is constrained to a range; so look for the start class
1229 * in that range. if neither, then look for the start class in the
1230 * whole rest of the string */
1232 /* XXX DAPM it's not clear what the minlen test is for, and why
1233 * it's not used in the floating case. Nothing in the test suite
1234 * causes minlen == 0 here. See <20140313134639.GS12844@iabyn.com>.
1235 * Here are some old comments, which may or may not be correct:
1237 * minlen == 0 is possible if regstclass is \b or \B,
1238 * and the fixed substr is ''$.
1239 * Since minlen is already taken into account, rx_origin+1 is
1240 * before strend; accidentally, minlen >= 1 guaranties no false
1241 * positives at rx_origin + 1 even for \b or \B. But (minlen? 1 :
1242 * 0) below assumes that regstclass does not come from lookahead...
1243 * If regstclass takes bytelength more than 1: If charlength==1, OK.
1244 * This leaves EXACTF-ish only, which are dealt with in
1248 if (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
1249 endpos= HOP3c(rx_origin, (prog->minlen ? cl_l : 0), strend);
1250 else if (prog->float_substr || prog->float_utf8) {
1251 rx_max_float = HOP3c(check_at, -start_shift, strbeg);
1252 endpos= HOP3c(rx_max_float, cl_l, strend);
1257 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1258 " looking for class: start_shift: %"IVdf" check_at: %"IVdf
1259 " rx_origin: %"IVdf" endpos: %"IVdf"\n",
1260 (IV)start_shift, (IV)(check_at - strbeg),
1261 (IV)(rx_origin - strbeg), (IV)(endpos - strbeg)));
1263 s = find_byclass(prog, progi->regstclass, rx_origin, endpos,
1266 if (endpos == strend) {
1267 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1268 " Could not match STCLASS...\n") );
1271 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1272 " This position contradicts STCLASS...\n") );
1273 if ((prog->intflags & PREGf_ANCH) && !ml_anch
1274 && !(prog->intflags & PREGf_IMPLICIT))
1277 /* Contradict one of substrings */
1278 if (prog->anchored_substr || prog->anchored_utf8) {
1279 if (prog->substrs->check_ix == 1) { /* check is float */
1280 /* Have both, check_string is floating */
1281 assert(rx_origin + start_shift <= check_at);
1282 if (rx_origin + start_shift != check_at) {
1283 /* not at latest position float substr could match:
1284 * Recheck anchored substring, but not floating.
1285 * The condition above is in bytes rather than
1286 * chars for efficiency. It's conservative, in
1287 * that it errs on the side of doing 'goto
1288 * do_other_substr', where a more accurate
1289 * char-based calculation will be done */
1290 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1291 " Looking for anchored substr starting at offset %ld...\n",
1292 (long)(other_last - strpos)) );
1293 goto do_other_substr;
1301 /* In the presence of ml_anch, we might be able to
1302 * find another \n without breaking the current float
1305 /* strictly speaking this should be HOP3c(..., 1, ...),
1306 * but since we goto a block of code that's going to
1307 * search for the next \n if any, its safe here */
1309 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1310 " Looking for /%s^%s/m starting at offset %ld...\n",
1311 PL_colors[0], PL_colors[1],
1312 (long)(rx_origin - strpos)) );
1313 goto postprocess_substr_matches;
1316 /* strictly speaking this can never be true; but might
1317 * be if we ever allow intuit without substrings */
1318 if (!(utf8_target ? prog->float_utf8 : prog->float_substr))
1321 rx_origin = rx_max_float;
1324 /* at this point, any matching substrings have been
1325 * contradicted. Start again... */
1327 rx_origin = HOP3c(rx_origin, 1, strend);
1329 /* uses bytes rather than char calculations for efficiency.
1330 * It's conservative: it errs on the side of doing 'goto restart',
1331 * where there is code that does a proper char-based test */
1332 if (rx_origin + start_shift + end_shift > strend) {
1333 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1334 " Could not match STCLASS...\n") );
1337 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
1338 " Looking for %s substr starting at offset %ld...\n",
1339 (prog->substrs->check_ix ? "floating" : "anchored"),
1340 (long)(rx_origin + start_shift - strpos)) );
1346 if (rx_origin != s) {
1347 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1348 " By STCLASS: moving %ld --> %ld\n",
1349 (long)(rx_origin - strpos), (long)(s - strpos))
1353 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1354 " Does not contradict STCLASS...\n");
1359 /* Decide whether using the substrings helped */
1361 if (rx_origin != strpos) {
1362 /* Fixed substring is found far enough so that the match
1363 cannot start at strpos. */
1365 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " try at offset...\n"));
1366 ++BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr); /* hooray/5 */
1369 /* The found rx_origin position does not prohibit matching at
1370 * strpos, so calling intuit didn't gain us anything. Decrement
1371 * the BmUSEFUL() count on the check substring, and if we reach
1373 if (!(prog->intflags & PREGf_NAUGHTY)
1375 prog->check_utf8 /* Could be deleted already */
1376 && --BmUSEFUL(prog->check_utf8) < 0
1377 && (prog->check_utf8 == prog->float_utf8)
1379 prog->check_substr /* Could be deleted already */
1380 && --BmUSEFUL(prog->check_substr) < 0
1381 && (prog->check_substr == prog->float_substr)
1384 /* If flags & SOMETHING - do not do it many times on the same match */
1385 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " ... Disabling check substring...\n"));
1386 /* XXX Does the destruction order has to change with utf8_target? */
1387 SvREFCNT_dec(utf8_target ? prog->check_utf8 : prog->check_substr);
1388 SvREFCNT_dec(utf8_target ? prog->check_substr : prog->check_utf8);
1389 prog->check_substr = prog->check_utf8 = NULL; /* disable */
1390 prog->float_substr = prog->float_utf8 = NULL; /* clear */
1391 check = NULL; /* abort */
1392 /* XXXX This is a remnant of the old implementation. It
1393 looks wasteful, since now INTUIT can use many
1394 other heuristics. */
1395 prog->extflags &= ~RXf_USE_INTUIT;
1399 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
1400 "Intuit: %sSuccessfully guessed:%s match at offset %ld\n",
1401 PL_colors[4], PL_colors[5], (long)(rx_origin - strpos)) );
1405 fail_finish: /* Substring not found */
1406 if (prog->check_substr || prog->check_utf8) /* could be removed already */
1407 BmUSEFUL(utf8_target ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
1409 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
1410 PL_colors[4], PL_colors[5]));
1415 #define DECL_TRIE_TYPE(scan) \
1416 const enum { trie_plain, trie_utf8, trie_utf8_fold, trie_latin_utf8_fold, \
1417 trie_utf8_exactfa_fold, trie_latin_utf8_exactfa_fold, \
1418 trie_utf8l, trie_flu8 } \
1419 trie_type = ((scan->flags == EXACT) \
1420 ? (utf8_target ? trie_utf8 : trie_plain) \
1421 : (scan->flags == EXACTL) \
1422 ? (utf8_target ? trie_utf8l : trie_plain) \
1423 : (scan->flags == EXACTFA) \
1425 ? trie_utf8_exactfa_fold \
1426 : trie_latin_utf8_exactfa_fold) \
1427 : (scan->flags == EXACTFLU8 \
1431 : trie_latin_utf8_fold)))
1433 #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, uvc, charid, foldlen, foldbuf, uniflags) \
1436 U8 flags = FOLD_FLAGS_FULL; \
1437 switch (trie_type) { \
1439 _CHECK_AND_WARN_PROBLEMATIC_LOCALE; \
1440 if (utf8_target && UTF8_IS_ABOVE_LATIN1(*uc)) { \
1441 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(uc, uc + UTF8SKIP(uc)); \
1443 goto do_trie_utf8_fold; \
1444 case trie_utf8_exactfa_fold: \
1445 flags |= FOLD_FLAGS_NOMIX_ASCII; \
1447 case trie_utf8_fold: \
1448 do_trie_utf8_fold: \
1449 if ( foldlen>0 ) { \
1450 uvc = utf8n_to_uvchr( (const U8*) uscan, UTF8_MAXLEN, &len, uniflags ); \
1455 uvc = _to_utf8_fold_flags( (const U8*) uc, foldbuf, &foldlen, flags); \
1456 len = UTF8SKIP(uc); \
1457 skiplen = UNISKIP( uvc ); \
1458 foldlen -= skiplen; \
1459 uscan = foldbuf + skiplen; \
1462 case trie_latin_utf8_exactfa_fold: \
1463 flags |= FOLD_FLAGS_NOMIX_ASCII; \
1465 case trie_latin_utf8_fold: \
1466 if ( foldlen>0 ) { \
1467 uvc = utf8n_to_uvchr( (const U8*) uscan, UTF8_MAXLEN, &len, uniflags ); \
1473 uvc = _to_fold_latin1( (U8) *uc, foldbuf, &foldlen, flags); \
1474 skiplen = UNISKIP( uvc ); \
1475 foldlen -= skiplen; \
1476 uscan = foldbuf + skiplen; \
1480 _CHECK_AND_WARN_PROBLEMATIC_LOCALE; \
1481 if (utf8_target && UTF8_IS_ABOVE_LATIN1(*uc)) { \
1482 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(uc, uc + UTF8SKIP(uc)); \
1486 uvc = utf8n_to_uvchr( (const U8*) uc, UTF8_MAXLEN, &len, uniflags ); \
1493 charid = trie->charmap[ uvc ]; \
1497 if (widecharmap) { \
1498 SV** const svpp = hv_fetch(widecharmap, \
1499 (char*)&uvc, sizeof(UV), 0); \
1501 charid = (U16)SvIV(*svpp); \
1506 #define DUMP_EXEC_POS(li,s,doutf8) \
1507 dump_exec_pos(li,s,(reginfo->strend),(reginfo->strbeg), \
1510 #define REXEC_FBC_EXACTISH_SCAN(COND) \
1514 && (ln == 1 || folder(s, pat_string, ln)) \
1515 && (reginfo->intuit || regtry(reginfo, &s)) )\
1521 #define REXEC_FBC_UTF8_SCAN(CODE) \
1523 while (s < strend) { \
1529 #define REXEC_FBC_SCAN(CODE) \
1531 while (s < strend) { \
1537 #define REXEC_FBC_UTF8_CLASS_SCAN(COND) \
1538 REXEC_FBC_UTF8_SCAN( /* Loops while (s < strend) */ \
1540 if (tmp && (reginfo->intuit || regtry(reginfo, &s))) \
1549 #define REXEC_FBC_CLASS_SCAN(COND) \
1550 REXEC_FBC_SCAN( /* Loops while (s < strend) */ \
1552 if (tmp && (reginfo->intuit || regtry(reginfo, &s))) \
1561 #define REXEC_FBC_CSCAN(CONDUTF8,COND) \
1562 if (utf8_target) { \
1563 REXEC_FBC_UTF8_CLASS_SCAN(CONDUTF8); \
1566 REXEC_FBC_CLASS_SCAN(COND); \
1569 /* The three macros below are slightly different versions of the same logic.
1571 * The first is for /a and /aa when the target string is UTF-8. This can only
1572 * match ascii, but it must advance based on UTF-8. The other two handle the
1573 * non-UTF-8 and the more generic UTF-8 cases. In all three, we are looking
1574 * for the boundary (or non-boundary) between a word and non-word character.
1575 * The utf8 and non-utf8 cases have the same logic, but the details must be
1576 * different. Find the "wordness" of the character just prior to this one, and
1577 * compare it with the wordness of this one. If they differ, we have a
1578 * boundary. At the beginning of the string, pretend that the previous
1579 * character was a new-line.
1581 * All these macros uncleanly have side-effects with each other and outside
1582 * variables. So far it's been too much trouble to clean-up
1584 * TEST_NON_UTF8 is the macro or function to call to test if its byte input is
1585 * a word character or not.
1586 * IF_SUCCESS is code to do if it finds that we are at a boundary between
1588 * IF_FAIL is code to do if we aren't at a boundary between word/non-word
1590 * Exactly one of the two IF_FOO parameters is a no-op, depending on whether we
1591 * are looking for a boundary or for a non-boundary. If we are looking for a
1592 * boundary, we want IF_FAIL to be the no-op, and for IF_SUCCESS to go out and
1593 * see if this tentative match actually works, and if so, to quit the loop
1594 * here. And vice-versa if we are looking for a non-boundary.
1596 * 'tmp' below in the next three macros in the REXEC_FBC_SCAN and
1597 * REXEC_FBC_UTF8_SCAN loops is a loop invariant, a bool giving the return of
1598 * TEST_NON_UTF8(s-1). To see this, note that that's what it is defined to be
1599 * at entry to the loop, and to get to the IF_FAIL branch, tmp must equal
1600 * TEST_NON_UTF8(s), and in the opposite branch, IF_SUCCESS, tmp is that
1601 * complement. But in that branch we complement tmp, meaning that at the
1602 * bottom of the loop tmp is always going to be equal to TEST_NON_UTF8(s),
1603 * which means at the top of the loop in the next iteration, it is
1604 * TEST_NON_UTF8(s-1) */
1605 #define FBC_UTF8_A(TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
1606 tmp = (s != reginfo->strbeg) ? UCHARAT(s - 1) : '\n'; \
1607 tmp = TEST_NON_UTF8(tmp); \
1608 REXEC_FBC_UTF8_SCAN( /* advances s while s < strend */ \
1609 if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
1611 IF_SUCCESS; /* Is a boundary if values for s-1 and s differ */ \
1618 /* Like FBC_UTF8_A, but TEST_UV is a macro which takes a UV as its input, and
1619 * TEST_UTF8 is a macro that for the same input code points returns identically
1620 * to TEST_UV, but takes a pointer to a UTF-8 encoded string instead */
1621 #define FBC_UTF8(TEST_UV, TEST_UTF8, IF_SUCCESS, IF_FAIL) \
1622 if (s == reginfo->strbeg) { \
1625 else { /* Back-up to the start of the previous character */ \
1626 U8 * const r = reghop3((U8*)s, -1, (U8*)reginfo->strbeg); \
1627 tmp = utf8n_to_uvchr(r, (U8*) reginfo->strend - r, \
1628 0, UTF8_ALLOW_DEFAULT); \
1630 tmp = TEST_UV(tmp); \
1631 LOAD_UTF8_CHARCLASS_ALNUM(); \
1632 REXEC_FBC_UTF8_SCAN( /* advances s while s < strend */ \
1633 if (tmp == ! (TEST_UTF8((U8 *) s))) { \
1642 /* Like the above two macros. UTF8_CODE is the complete code for handling
1643 * UTF-8. Common to the BOUND and NBOUND cases, set-up by the FBC_BOUND, etc
1645 #define FBC_BOUND_COMMON(UTF8_CODE, TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
1646 if (utf8_target) { \
1649 else { /* Not utf8 */ \
1650 tmp = (s != reginfo->strbeg) ? UCHARAT(s - 1) : '\n'; \
1651 tmp = TEST_NON_UTF8(tmp); \
1652 REXEC_FBC_SCAN( /* advances s while s < strend */ \
1653 if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
1662 /* Here, things have been set up by the previous code so that tmp is the \
1663 * return of TEST_NON_UTF(s-1) or TEST_UTF8(s-1) (depending on the \
1664 * utf8ness of the target). We also have to check if this matches against \
1665 * the EOS, which we treat as a \n (which is the same value in both UTF-8 \
1666 * or non-UTF8, so can use the non-utf8 test condition even for a UTF-8 \
1668 if (tmp == ! TEST_NON_UTF8('\n')) { \
1675 /* This is the macro to use when we want to see if something that looks like it
1676 * could match, actually does, and if so exits the loop */
1677 #define REXEC_FBC_TRYIT \
1678 if ((reginfo->intuit || regtry(reginfo, &s))) \
1681 /* The only difference between the BOUND and NBOUND cases is that
1682 * REXEC_FBC_TRYIT is called when matched in BOUND, and when non-matched in
1683 * NBOUND. This is accomplished by passing it as either the if or else clause,
1684 * with the other one being empty (PLACEHOLDER is defined as empty).
1686 * The TEST_FOO parameters are for operating on different forms of input, but
1687 * all should be ones that return identically for the same underlying code
1689 #define FBC_BOUND(TEST_NON_UTF8, TEST_UV, TEST_UTF8) \
1691 FBC_UTF8(TEST_UV, TEST_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), \
1692 TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)
1694 #define FBC_BOUND_A(TEST_NON_UTF8) \
1696 FBC_UTF8_A(TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), \
1697 TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)
1699 #define FBC_NBOUND(TEST_NON_UTF8, TEST_UV, TEST_UTF8) \
1701 FBC_UTF8(TEST_UV, TEST_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), \
1702 TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)
1704 #define FBC_NBOUND_A(TEST_NON_UTF8) \
1706 FBC_UTF8_A(TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), \
1707 TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)
1709 /* Takes a pointer to an inversion list, a pointer to its corresponding
1710 * inversion map, and a code point, and returns the code point's value
1711 * according to the two arrays. It assumes that all code points have a value.
1712 * This is used as the base macro for macros for particular properties */
1713 #define _generic_GET_BREAK_VAL_CP(invlist, invmap, cp) \
1714 invmap[_invlist_search(invlist, cp)]
1716 /* Same as above, but takes begin, end ptrs to a UTF-8 encoded string instead
1717 * of a code point, returning the value for the first code point in the string.
1718 * And it takes the particular macro name that finds the desired value given a
1719 * code point. Merely convert the UTF-8 to code point and call the cp macro */
1720 #define _generic_GET_BREAK_VAL_UTF8(cp_macro, pos, strend) \
1721 (__ASSERT_(pos < strend) \
1722 /* Note assumes is valid UTF-8 */ \
1723 (cp_macro(utf8_to_uvchr_buf((pos), (strend), NULL))))
1725 /* Returns the GCB value for the input code point */
1726 #define getGCB_VAL_CP(cp) \
1727 _generic_GET_BREAK_VAL_CP( \
1729 Grapheme_Cluster_Break_invmap, \
1732 /* Returns the GCB value for the first code point in the UTF-8 encoded string
1733 * bounded by pos and strend */
1734 #define getGCB_VAL_UTF8(pos, strend) \
1735 _generic_GET_BREAK_VAL_UTF8(getGCB_VAL_CP, pos, strend)
1738 /* Returns the SB value for the input code point */
1739 #define getSB_VAL_CP(cp) \
1740 _generic_GET_BREAK_VAL_CP( \
1742 Sentence_Break_invmap, \
1745 /* Returns the SB value for the first code point in the UTF-8 encoded string
1746 * bounded by pos and strend */
1747 #define getSB_VAL_UTF8(pos, strend) \
1748 _generic_GET_BREAK_VAL_UTF8(getSB_VAL_CP, pos, strend)
1750 /* Returns the WB value for the input code point */
1751 #define getWB_VAL_CP(cp) \
1752 _generic_GET_BREAK_VAL_CP( \
1754 Word_Break_invmap, \
1757 /* Returns the WB value for the first code point in the UTF-8 encoded string
1758 * bounded by pos and strend */
1759 #define getWB_VAL_UTF8(pos, strend) \
1760 _generic_GET_BREAK_VAL_UTF8(getWB_VAL_CP, pos, strend)
1762 /* We know what class REx starts with. Try to find this position... */
1763 /* if reginfo->intuit, its a dryrun */
1764 /* annoyingly all the vars in this routine have different names from their counterparts
1765 in regmatch. /grrr */
1767 S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
1768 const char *strend, regmatch_info *reginfo)
1771 const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
1772 char *pat_string; /* The pattern's exactish string */
1773 char *pat_end; /* ptr to end char of pat_string */
1774 re_fold_t folder; /* Function for computing non-utf8 folds */
1775 const U8 *fold_array; /* array for folding ords < 256 */
1781 I32 tmp = 1; /* Scratch variable? */
1782 const bool utf8_target = reginfo->is_utf8_target;
1783 UV utf8_fold_flags = 0;
1784 const bool is_utf8_pat = reginfo->is_utf8_pat;
1785 bool to_complement = FALSE; /* Invert the result? Taking the xor of this
1786 with a result inverts that result, as 0^1 =
1788 _char_class_number classnum;
1790 RXi_GET_DECL(prog,progi);
1792 PERL_ARGS_ASSERT_FIND_BYCLASS;
1794 /* We know what class it must start with. */
1797 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
1801 REXEC_FBC_UTF8_CLASS_SCAN(
1802 reginclass(prog, c, (U8*)s, (U8*) strend, utf8_target));
1805 REXEC_FBC_CLASS_SCAN(REGINCLASS(prog, c, (U8*)s));
1810 if (tmp && (reginfo->intuit || regtry(reginfo, &s)))
1817 case EXACTFA_NO_TRIE: /* This node only generated for non-utf8 patterns */
1818 assert(! is_utf8_pat);
1821 if (is_utf8_pat || utf8_target) {
1822 utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
1823 goto do_exactf_utf8;
1825 fold_array = PL_fold_latin1; /* Latin1 folds are not affected by */
1826 folder = foldEQ_latin1; /* /a, except the sharp s one which */
1827 goto do_exactf_non_utf8; /* isn't dealt with by these */
1829 case EXACTF: /* This node only generated for non-utf8 patterns */
1830 assert(! is_utf8_pat);
1832 utf8_fold_flags = 0;
1833 goto do_exactf_utf8;
1835 fold_array = PL_fold;
1837 goto do_exactf_non_utf8;
1840 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
1841 if (is_utf8_pat || utf8_target || IN_UTF8_CTYPE_LOCALE) {
1842 utf8_fold_flags = FOLDEQ_LOCALE;
1843 goto do_exactf_utf8;
1845 fold_array = PL_fold_locale;
1846 folder = foldEQ_locale;
1847 goto do_exactf_non_utf8;
1851 utf8_fold_flags = FOLDEQ_S2_ALREADY_FOLDED;
1853 goto do_exactf_utf8;
1856 if (! utf8_target) { /* All code points in this node require
1857 UTF-8 to express. */
1860 utf8_fold_flags = FOLDEQ_LOCALE | FOLDEQ_S2_ALREADY_FOLDED
1861 | FOLDEQ_S2_FOLDS_SANE;
1862 goto do_exactf_utf8;
1865 if (is_utf8_pat || utf8_target) {
1866 utf8_fold_flags = is_utf8_pat ? FOLDEQ_S2_ALREADY_FOLDED : 0;
1867 goto do_exactf_utf8;
1870 /* Any 'ss' in the pattern should have been replaced by regcomp,
1871 * so we don't have to worry here about this single special case
1872 * in the Latin1 range */
1873 fold_array = PL_fold_latin1;
1874 folder = foldEQ_latin1;
1878 do_exactf_non_utf8: /* Neither pattern nor string are UTF8, and there
1879 are no glitches with fold-length differences
1880 between the target string and pattern */
1882 /* The idea in the non-utf8 EXACTF* cases is to first find the
1883 * first character of the EXACTF* node and then, if necessary,
1884 * case-insensitively compare the full text of the node. c1 is the
1885 * first character. c2 is its fold. This logic will not work for
1886 * Unicode semantics and the german sharp ss, which hence should
1887 * not be compiled into a node that gets here. */
1888 pat_string = STRING(c);
1889 ln = STR_LEN(c); /* length to match in octets/bytes */
1891 /* We know that we have to match at least 'ln' bytes (which is the
1892 * same as characters, since not utf8). If we have to match 3
1893 * characters, and there are only 2 availabe, we know without
1894 * trying that it will fail; so don't start a match past the
1895 * required minimum number from the far end */
1896 e = HOP3c(strend, -((SSize_t)ln), s);
1898 if (reginfo->intuit && e < s) {
1899 e = s; /* Due to minlen logic of intuit() */
1903 c2 = fold_array[c1];
1904 if (c1 == c2) { /* If char and fold are the same */
1905 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
1908 REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
1916 /* If one of the operands is in utf8, we can't use the simpler folding
1917 * above, due to the fact that many different characters can have the
1918 * same fold, or portion of a fold, or different- length fold */
1919 pat_string = STRING(c);
1920 ln = STR_LEN(c); /* length to match in octets/bytes */
1921 pat_end = pat_string + ln;
1922 lnc = is_utf8_pat /* length to match in characters */
1923 ? utf8_length((U8 *) pat_string, (U8 *) pat_end)
1926 /* We have 'lnc' characters to match in the pattern, but because of
1927 * multi-character folding, each character in the target can match
1928 * up to 3 characters (Unicode guarantees it will never exceed
1929 * this) if it is utf8-encoded; and up to 2 if not (based on the
1930 * fact that the Latin 1 folds are already determined, and the
1931 * only multi-char fold in that range is the sharp-s folding to
1932 * 'ss'. Thus, a pattern character can match as little as 1/3 of a
1933 * string character. Adjust lnc accordingly, rounding up, so that
1934 * if we need to match at least 4+1/3 chars, that really is 5. */
1935 expansion = (utf8_target) ? UTF8_MAX_FOLD_CHAR_EXPAND : 2;
1936 lnc = (lnc + expansion - 1) / expansion;
1938 /* As in the non-UTF8 case, if we have to match 3 characters, and
1939 * only 2 are left, it's guaranteed to fail, so don't start a
1940 * match that would require us to go beyond the end of the string
1942 e = HOP3c(strend, -((SSize_t)lnc), s);
1944 if (reginfo->intuit && e < s) {
1945 e = s; /* Due to minlen logic of intuit() */
1948 /* XXX Note that we could recalculate e to stop the loop earlier,
1949 * as the worst case expansion above will rarely be met, and as we
1950 * go along we would usually find that e moves further to the left.
1951 * This would happen only after we reached the point in the loop
1952 * where if there were no expansion we should fail. Unclear if
1953 * worth the expense */
1956 char *my_strend= (char *)strend;
1957 if (foldEQ_utf8_flags(s, &my_strend, 0, utf8_target,
1958 pat_string, NULL, ln, is_utf8_pat, utf8_fold_flags)
1959 && (reginfo->intuit || regtry(reginfo, &s)) )
1963 s += (utf8_target) ? UTF8SKIP(s) : 1;
1969 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
1970 if (FLAGS(c) != TRADITIONAL_BOUND) {
1971 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
1972 B_ON_NON_UTF8_LOCALE_IS_WRONG);
1976 FBC_BOUND(isWORDCHAR_LC, isWORDCHAR_LC_uvchr, isWORDCHAR_LC_utf8);
1980 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
1981 if (FLAGS(c) != TRADITIONAL_BOUND) {
1982 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
1983 B_ON_NON_UTF8_LOCALE_IS_WRONG);
1987 FBC_NBOUND(isWORDCHAR_LC, isWORDCHAR_LC_uvchr, isWORDCHAR_LC_utf8);
1990 case BOUND: /* regcomp.c makes sure that this only has the traditional \b
1992 assert(FLAGS(c) == TRADITIONAL_BOUND);
1994 FBC_BOUND(isWORDCHAR, isWORDCHAR_uni, isWORDCHAR_utf8);
1997 case BOUNDA: /* regcomp.c makes sure that this only has the traditional \b
1999 assert(FLAGS(c) == TRADITIONAL_BOUND);
2001 FBC_BOUND_A(isWORDCHAR_A);
2004 case NBOUND: /* regcomp.c makes sure that this only has the traditional \b
2006 assert(FLAGS(c) == TRADITIONAL_BOUND);
2008 FBC_NBOUND(isWORDCHAR, isWORDCHAR_uni, isWORDCHAR_utf8);
2011 case NBOUNDA: /* regcomp.c makes sure that this only has the traditional \b
2013 assert(FLAGS(c) == TRADITIONAL_BOUND);
2015 FBC_NBOUND_A(isWORDCHAR_A);
2019 if ((bound_type) FLAGS(c) == TRADITIONAL_BOUND) {
2020 FBC_NBOUND(isWORDCHAR_L1, isWORDCHAR_uni, isWORDCHAR_utf8);
2031 switch((bound_type) FLAGS(c)) {
2032 case TRADITIONAL_BOUND:
2033 FBC_BOUND(isWORDCHAR_L1, isWORDCHAR_uni, isWORDCHAR_utf8);
2036 if (s == reginfo->strbeg) { /* GCB always matches at begin and
2038 if (to_complement ^ cBOOL(reginfo->intuit
2039 || regtry(reginfo, &s)))
2043 s += (utf8_target) ? UTF8SKIP(s) : 1;
2047 PL_GCB_enum before = getGCB_VAL_UTF8(
2049 (U8*)(reginfo->strbeg)),
2050 (U8*) reginfo->strend);
2051 while (s < strend) {
2052 PL_GCB_enum after = getGCB_VAL_UTF8((U8*) s,
2053 (U8*) reginfo->strend);
2054 if (to_complement ^ isGCB(before, after)) {
2055 if (reginfo->intuit || regtry(reginfo, &s)) {
2063 else { /* Not utf8. Everything is a GCB except between CR and
2065 while (s < strend) {
2066 if (to_complement ^ (UCHARAT(s - 1) != '\r'
2067 || UCHARAT(s) != '\n'))
2069 if (reginfo->intuit || regtry(reginfo, &s)) {
2077 if (to_complement ^ cBOOL(reginfo->intuit || regtry(reginfo, &s))) {
2083 if (s == reginfo->strbeg) { /* SB always matches at beginning */
2085 ^ cBOOL(reginfo->intuit || regtry(reginfo, &s)))
2090 /* Didn't match. Go try at the next position */
2091 s += (utf8_target) ? UTF8SKIP(s) : 1;
2095 PL_SB_enum before = getSB_VAL_UTF8(reghop3((U8*)s,
2097 (U8*)(reginfo->strbeg)),
2098 (U8*) reginfo->strend);
2099 while (s < strend) {
2100 PL_SB_enum after = getSB_VAL_UTF8((U8*) s,
2101 (U8*) reginfo->strend);
2102 if (to_complement ^ isSB(before,
2104 (U8*) reginfo->strbeg,
2106 (U8*) reginfo->strend,
2109 if (reginfo->intuit || regtry(reginfo, &s)) {
2117 else { /* Not utf8. */
2118 PL_SB_enum before = getSB_VAL_CP((U8) *(s -1));
2119 while (s < strend) {
2120 PL_SB_enum after = getSB_VAL_CP((U8) *s);
2121 if (to_complement ^ isSB(before,
2123 (U8*) reginfo->strbeg,
2125 (U8*) reginfo->strend,
2128 if (reginfo->intuit || regtry(reginfo, &s)) {
2137 /* Here are at the final position in the target string. The SB
2138 * value is always true here, so matches, depending on other
2140 if (to_complement ^ cBOOL(reginfo->intuit
2141 || regtry(reginfo, &s)))
2149 if (s == reginfo->strbeg) {
2150 if (to_complement ^ cBOOL(reginfo->intuit
2151 || regtry(reginfo, &s)))
2155 s += (utf8_target) ? UTF8SKIP(s) : 1;
2159 /* We are at a boundary between char_sub_0 and char_sub_1.
2160 * We also keep track of the value for char_sub_-1 as we
2161 * loop through the line. Context may be needed to make a
2162 * determination, and if so, this can save having to
2164 PL_WB_enum previous = PL_WB_UNKNOWN;
2165 PL_WB_enum before = getWB_VAL_UTF8(
2168 (U8*)(reginfo->strbeg)),
2169 (U8*) reginfo->strend);
2170 while (s < strend) {
2171 PL_WB_enum after = getWB_VAL_UTF8((U8*) s,
2172 (U8*) reginfo->strend);
2173 if (to_complement ^ isWB(previous,
2176 (U8*) reginfo->strbeg,
2178 (U8*) reginfo->strend,
2181 if (reginfo->intuit || regtry(reginfo, &s)) {
2190 else { /* Not utf8. */
2191 PL_WB_enum previous = PL_WB_UNKNOWN;
2192 PL_WB_enum before = getWB_VAL_CP((U8) *(s -1));
2193 while (s < strend) {
2194 PL_WB_enum after = getWB_VAL_CP((U8) *s);
2195 if (to_complement ^ isWB(previous,
2198 (U8*) reginfo->strbeg,
2200 (U8*) reginfo->strend,
2203 if (reginfo->intuit || regtry(reginfo, &s)) {
2213 if (to_complement ^ cBOOL(reginfo->intuit
2214 || regtry(reginfo, &s)))
2224 REXEC_FBC_CSCAN(is_LNBREAK_utf8_safe(s, strend),
2225 is_LNBREAK_latin1_safe(s, strend)
2229 /* The argument to all the POSIX node types is the class number to pass to
2230 * _generic_isCC() to build a mask for searching in PL_charclass[] */
2237 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
2238 REXEC_FBC_CSCAN(to_complement ^ cBOOL(isFOO_utf8_lc(FLAGS(c), (U8 *) s)),
2239 to_complement ^ cBOOL(isFOO_lc(FLAGS(c), *s)));
2254 /* The complement of something that matches only ASCII matches all
2255 * non-ASCII, plus everything in ASCII that isn't in the class. */
2256 REXEC_FBC_UTF8_CLASS_SCAN(! isASCII_utf8(s)
2257 || ! _generic_isCC_A(*s, FLAGS(c)));
2266 /* Don't need to worry about utf8, as it can match only a single
2267 * byte invariant character. */
2268 REXEC_FBC_CLASS_SCAN(
2269 to_complement ^ cBOOL(_generic_isCC_A(*s, FLAGS(c))));
2277 if (! utf8_target) {
2278 REXEC_FBC_CLASS_SCAN(to_complement ^ cBOOL(_generic_isCC(*s,
2284 classnum = (_char_class_number) FLAGS(c);
2285 if (classnum < _FIRST_NON_SWASH_CC) {
2286 while (s < strend) {
2288 /* We avoid loading in the swash as long as possible, but
2289 * should we have to, we jump to a separate loop. This
2290 * extra 'if' statement is what keeps this code from being
2291 * just a call to REXEC_FBC_UTF8_CLASS_SCAN() */
2292 if (UTF8_IS_ABOVE_LATIN1(*s)) {
2293 goto found_above_latin1;
2295 if ((UTF8_IS_INVARIANT(*s)
2296 && to_complement ^ cBOOL(_generic_isCC((U8) *s,
2298 || (UTF8_IS_DOWNGRADEABLE_START(*s)
2299 && to_complement ^ cBOOL(
2300 _generic_isCC(TWO_BYTE_UTF8_TO_NATIVE(*s,
2304 if (tmp && (reginfo->intuit || regtry(reginfo, &s)))
2316 else switch (classnum) { /* These classes are implemented as
2318 case _CC_ENUM_SPACE:
2319 REXEC_FBC_UTF8_CLASS_SCAN(
2320 to_complement ^ cBOOL(isSPACE_utf8(s)));
2323 case _CC_ENUM_BLANK:
2324 REXEC_FBC_UTF8_CLASS_SCAN(
2325 to_complement ^ cBOOL(isBLANK_utf8(s)));
2328 case _CC_ENUM_XDIGIT:
2329 REXEC_FBC_UTF8_CLASS_SCAN(
2330 to_complement ^ cBOOL(isXDIGIT_utf8(s)));
2333 case _CC_ENUM_VERTSPACE:
2334 REXEC_FBC_UTF8_CLASS_SCAN(
2335 to_complement ^ cBOOL(isVERTWS_utf8(s)));
2338 case _CC_ENUM_CNTRL:
2339 REXEC_FBC_UTF8_CLASS_SCAN(
2340 to_complement ^ cBOOL(isCNTRL_utf8(s)));
2344 Perl_croak(aTHX_ "panic: find_byclass() node %d='%s' has an unexpected character class '%d'", OP(c), PL_reg_name[OP(c)], classnum);
2345 NOT_REACHED; /* NOTREACHED */
2350 found_above_latin1: /* Here we have to load a swash to get the result
2351 for the current code point */
2352 if (! PL_utf8_swash_ptrs[classnum]) {
2353 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2354 PL_utf8_swash_ptrs[classnum] =
2355 _core_swash_init("utf8",
2358 PL_XPosix_ptrs[classnum], &flags);
2361 /* This is a copy of the loop above for swash classes, though using the
2362 * FBC macro instead of being expanded out. Since we've loaded the
2363 * swash, we don't have to check for that each time through the loop */
2364 REXEC_FBC_UTF8_CLASS_SCAN(
2365 to_complement ^ cBOOL(_generic_utf8(
2368 swash_fetch(PL_utf8_swash_ptrs[classnum],
2376 /* what trie are we using right now */
2377 reg_ac_data *aho = (reg_ac_data*)progi->data->data[ ARG( c ) ];
2378 reg_trie_data *trie = (reg_trie_data*)progi->data->data[ aho->trie ];
2379 HV *widecharmap = MUTABLE_HV(progi->data->data[ aho->trie + 1 ]);
2381 const char *last_start = strend - trie->minlen;
2383 const char *real_start = s;
2385 STRLEN maxlen = trie->maxlen;
2387 U8 **points; /* map of where we were in the input string
2388 when reading a given char. For ASCII this
2389 is unnecessary overhead as the relationship
2390 is always 1:1, but for Unicode, especially
2391 case folded Unicode this is not true. */
2392 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
2396 GET_RE_DEBUG_FLAGS_DECL;
2398 /* We can't just allocate points here. We need to wrap it in
2399 * an SV so it gets freed properly if there is a croak while
2400 * running the match */
2403 sv_points=newSV(maxlen * sizeof(U8 *));
2404 SvCUR_set(sv_points,
2405 maxlen * sizeof(U8 *));
2406 SvPOK_on(sv_points);
2407 sv_2mortal(sv_points);
2408 points=(U8**)SvPV_nolen(sv_points );
2409 if ( trie_type != trie_utf8_fold
2410 && (trie->bitmap || OP(c)==AHOCORASICKC) )
2413 bitmap=(U8*)trie->bitmap;
2415 bitmap=(U8*)ANYOF_BITMAP(c);
2417 /* this is the Aho-Corasick algorithm modified a touch
2418 to include special handling for long "unknown char" sequences.
2419 The basic idea being that we use AC as long as we are dealing
2420 with a possible matching char, when we encounter an unknown char
2421 (and we have not encountered an accepting state) we scan forward
2422 until we find a legal starting char.
2423 AC matching is basically that of trie matching, except that when
2424 we encounter a failing transition, we fall back to the current
2425 states "fail state", and try the current char again, a process
2426 we repeat until we reach the root state, state 1, or a legal
2427 transition. If we fail on the root state then we can either
2428 terminate if we have reached an accepting state previously, or
2429 restart the entire process from the beginning if we have not.
2432 while (s <= last_start) {
2433 const U32 uniflags = UTF8_ALLOW_DEFAULT;
2441 U8 *uscan = (U8*)NULL;
2442 U8 *leftmost = NULL;
2444 U32 accepted_word= 0;
2448 while ( state && uc <= (U8*)strend ) {
2450 U32 word = aho->states[ state ].wordnum;
2454 DEBUG_TRIE_EXECUTE_r(
2455 if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
2456 dump_exec_pos( (char *)uc, c, strend, real_start,
2457 (char *)uc, utf8_target );
2458 PerlIO_printf( Perl_debug_log,
2459 " Scanning for legal start char...\n");
2463 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
2467 while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
2473 if (uc >(U8*)last_start) break;
2477 U8 *lpos= points[ (pointpos - trie->wordinfo[word].len) % maxlen ];
2478 if (!leftmost || lpos < leftmost) {
2479 DEBUG_r(accepted_word=word);
2485 points[pointpos++ % maxlen]= uc;
2486 if (foldlen || uc < (U8*)strend) {
2487 REXEC_TRIE_READ_CHAR(trie_type, trie,
2489 uscan, len, uvc, charid, foldlen,
2491 DEBUG_TRIE_EXECUTE_r({
2492 dump_exec_pos( (char *)uc, c, strend,
2493 real_start, s, utf8_target);
2494 PerlIO_printf(Perl_debug_log,
2495 " Charid:%3u CP:%4"UVxf" ",
2507 word = aho->states[ state ].wordnum;
2509 base = aho->states[ state ].trans.base;
2511 DEBUG_TRIE_EXECUTE_r({
2513 dump_exec_pos( (char *)uc, c, strend, real_start,
2515 PerlIO_printf( Perl_debug_log,
2516 "%sState: %4"UVxf", word=%"UVxf,
2517 failed ? " Fail transition to " : "",
2518 (UV)state, (UV)word);
2524 ( ((offset = base + charid
2525 - 1 - trie->uniquecharcount)) >= 0)
2526 && ((U32)offset < trie->lasttrans)
2527 && trie->trans[offset].check == state
2528 && (tmp=trie->trans[offset].next))
2530 DEBUG_TRIE_EXECUTE_r(
2531 PerlIO_printf( Perl_debug_log," - legal\n"));
2536 DEBUG_TRIE_EXECUTE_r(
2537 PerlIO_printf( Perl_debug_log," - fail\n"));
2539 state = aho->fail[state];
2543 /* we must be accepting here */
2544 DEBUG_TRIE_EXECUTE_r(
2545 PerlIO_printf( Perl_debug_log," - accepting\n"));
2554 if (!state) state = 1;
2557 if ( aho->states[ state ].wordnum ) {
2558 U8 *lpos = points[ (pointpos - trie->wordinfo[aho->states[ state ].wordnum].len) % maxlen ];
2559 if (!leftmost || lpos < leftmost) {
2560 DEBUG_r(accepted_word=aho->states[ state ].wordnum);
2565 s = (char*)leftmost;
2566 DEBUG_TRIE_EXECUTE_r({
2568 Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n",
2569 (UV)accepted_word, (IV)(s - real_start)
2572 if (reginfo->intuit || regtry(reginfo, &s)) {
2578 DEBUG_TRIE_EXECUTE_r({
2579 PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n");
2582 DEBUG_TRIE_EXECUTE_r(
2583 PerlIO_printf( Perl_debug_log,"No match.\n"));
2592 Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
2599 /* set RX_SAVED_COPY, RX_SUBBEG etc.
2600 * flags have same meanings as with regexec_flags() */
2603 S_reg_set_capture_string(pTHX_ REGEXP * const rx,
2610 struct regexp *const prog = ReANY(rx);
2612 if (flags & REXEC_COPY_STR) {
2616 PerlIO_printf(Perl_debug_log,
2617 "Copy on write: regexp capture, type %d\n",
2620 /* Create a new COW SV to share the match string and store
2621 * in saved_copy, unless the current COW SV in saved_copy
2622 * is valid and suitable for our purpose */
2623 if (( prog->saved_copy
2624 && SvIsCOW(prog->saved_copy)
2625 && SvPOKp(prog->saved_copy)
2628 && SvPVX(sv) == SvPVX(prog->saved_copy)))
2630 /* just reuse saved_copy SV */
2631 if (RXp_MATCH_COPIED(prog)) {
2632 Safefree(prog->subbeg);
2633 RXp_MATCH_COPIED_off(prog);
2637 /* create new COW SV to share string */
2638 RX_MATCH_COPY_FREE(rx);
2639 prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
2641 prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
2642 assert (SvPOKp(prog->saved_copy));
2643 prog->sublen = strend - strbeg;
2644 prog->suboffset = 0;
2645 prog->subcoffset = 0;
2650 SSize_t max = strend - strbeg;
2653 if ( (flags & REXEC_COPY_SKIP_POST)
2654 && !(prog->extflags & RXf_PMf_KEEPCOPY) /* //p */
2655 && !(PL_sawampersand & SAWAMPERSAND_RIGHT)
2656 ) { /* don't copy $' part of string */
2659 /* calculate the right-most part of the string covered
2660 * by a capture. Due to look-ahead, this may be to
2661 * the right of $&, so we have to scan all captures */
2662 while (n <= prog->lastparen) {
2663 if (prog->offs[n].end > max)
2664 max = prog->offs[n].end;
2668 max = (PL_sawampersand & SAWAMPERSAND_LEFT)
2669 ? prog->offs[0].start
2671 assert(max >= 0 && max <= strend - strbeg);
2674 if ( (flags & REXEC_COPY_SKIP_PRE)
2675 && !(prog->extflags & RXf_PMf_KEEPCOPY) /* //p */
2676 && !(PL_sawampersand & SAWAMPERSAND_LEFT)
2677 ) { /* don't copy $` part of string */
2680 /* calculate the left-most part of the string covered
2681 * by a capture. Due to look-behind, this may be to
2682 * the left of $&, so we have to scan all captures */
2683 while (min && n <= prog->lastparen) {
2684 if ( prog->offs[n].start != -1
2685 && prog->offs[n].start < min)
2687 min = prog->offs[n].start;
2691 if ((PL_sawampersand & SAWAMPERSAND_RIGHT)
2692 && min > prog->offs[0].end
2694 min = prog->offs[0].end;
2698 assert(min >= 0 && min <= max && min <= strend - strbeg);
2701 if (RX_MATCH_COPIED(rx)) {
2702 if (sublen > prog->sublen)
2704 (char*)saferealloc(prog->subbeg, sublen+1);
2707 prog->subbeg = (char*)safemalloc(sublen+1);
2708 Copy(strbeg + min, prog->subbeg, sublen, char);
2709 prog->subbeg[sublen] = '\0';
2710 prog->suboffset = min;
2711 prog->sublen = sublen;
2712 RX_MATCH_COPIED_on(rx);
2714 prog->subcoffset = prog->suboffset;
2715 if (prog->suboffset && utf8_target) {
2716 /* Convert byte offset to chars.
2717 * XXX ideally should only compute this if @-/@+
2718 * has been seen, a la PL_sawampersand ??? */
2720 /* If there's a direct correspondence between the
2721 * string which we're matching and the original SV,
2722 * then we can use the utf8 len cache associated with
2723 * the SV. In particular, it means that under //g,
2724 * sv_pos_b2u() will use the previously cached
2725 * position to speed up working out the new length of
2726 * subcoffset, rather than counting from the start of
2727 * the string each time. This stops
2728 * $x = "\x{100}" x 1E6; 1 while $x =~ /(.)/g;
2729 * from going quadratic */
2730 if (SvPOKp(sv) && SvPVX(sv) == strbeg)
2731 prog->subcoffset = sv_pos_b2u_flags(sv, prog->subcoffset,
2732 SV_GMAGIC|SV_CONST_RETURN);
2734 prog->subcoffset = utf8_length((U8*)strbeg,
2735 (U8*)(strbeg+prog->suboffset));
2739 RX_MATCH_COPY_FREE(rx);
2740 prog->subbeg = strbeg;
2741 prog->suboffset = 0;
2742 prog->subcoffset = 0;
2743 prog->sublen = strend - strbeg;
2751 - regexec_flags - match a regexp against a string
2754 Perl_regexec_flags(pTHX_ REGEXP * const rx, char *stringarg, char *strend,
2755 char *strbeg, SSize_t minend, SV *sv, void *data, U32 flags)
2756 /* stringarg: the point in the string at which to begin matching */
2757 /* strend: pointer to null at end of string */
2758 /* strbeg: real beginning of string */
2759 /* minend: end of match must be >= minend bytes after stringarg. */
2760 /* sv: SV being matched: only used for utf8 flag, pos() etc; string
2761 * itself is accessed via the pointers above */
2762 /* data: May be used for some additional optimizations.
2763 Currently unused. */
2764 /* flags: For optimizations. See REXEC_* in regexp.h */
2767 struct regexp *const prog = ReANY(rx);
2771 SSize_t minlen; /* must match at least this many chars */
2772 SSize_t dontbother = 0; /* how many characters not to try at end */
2773 const bool utf8_target = cBOOL(DO_UTF8(sv));
2775 RXi_GET_DECL(prog,progi);
2776 regmatch_info reginfo_buf; /* create some info to pass to regtry etc */
2777 regmatch_info *const reginfo = ®info_buf;
2778 regexp_paren_pair *swap = NULL;
2780 GET_RE_DEBUG_FLAGS_DECL;
2782 PERL_ARGS_ASSERT_REGEXEC_FLAGS;
2783 PERL_UNUSED_ARG(data);
2785 /* Be paranoid... */
2786 if (prog == NULL || stringarg == NULL) {
2787 Perl_croak(aTHX_ "NULL regexp parameter");
2791 debug_start_match(rx, utf8_target, stringarg, strend,
2795 startpos = stringarg;
2797 if (prog->intflags & PREGf_GPOS_SEEN) {
2800 /* set reginfo->ganch, the position where \G can match */
2803 (flags & REXEC_IGNOREPOS)
2804 ? stringarg /* use start pos rather than pos() */
2805 : (sv && (mg = mg_find_mglob(sv)) && mg->mg_len >= 0)
2806 /* Defined pos(): */
2807 ? strbeg + MgBYTEPOS(mg, sv, strbeg, strend-strbeg)
2808 : strbeg; /* pos() not defined; use start of string */
2810 DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log,
2811 "GPOS ganch set to strbeg[%"IVdf"]\n", (IV)(reginfo->ganch - strbeg)));
2813 /* in the presence of \G, we may need to start looking earlier in
2814 * the string than the suggested start point of stringarg:
2815 * if prog->gofs is set, then that's a known, fixed minimum
2818 * /ab|c\G/: gofs = 1
2819 * or if the minimum offset isn't known, then we have to go back
2820 * to the start of the string, e.g. /w+\G/
2823 if (prog->intflags & PREGf_ANCH_GPOS) {
2824 startpos = reginfo->ganch - prog->gofs;
2826 ((flags & REXEC_FAIL_ON_UNDERFLOW) ? stringarg : strbeg))
2828 DEBUG_r(PerlIO_printf(Perl_debug_log,
2829 "fail: ganch-gofs before earliest possible start\n"));
2833 else if (prog->gofs) {
2834 if (startpos - prog->gofs < strbeg)
2837 startpos -= prog->gofs;
2839 else if (prog->intflags & PREGf_GPOS_FLOAT)
2843 minlen = prog->minlen;
2844 if ((startpos + minlen) > strend || startpos < strbeg) {
2845 DEBUG_r(PerlIO_printf(Perl_debug_log,
2846 "Regex match can't succeed, so not even tried\n"));
2850 /* at the end of this function, we'll do a LEAVE_SCOPE(oldsave),
2851 * which will call destuctors to reset PL_regmatch_state, free higher
2852 * PL_regmatch_slabs, and clean up regmatch_info_aux and
2853 * regmatch_info_aux_eval */
2855 oldsave = PL_savestack_ix;
2859 if ((prog->extflags & RXf_USE_INTUIT)
2860 && !(flags & REXEC_CHECKED))
2862 s = re_intuit_start(rx, sv, strbeg, startpos, strend,
2867 if (prog->extflags & RXf_CHECK_ALL) {
2868 /* we can match based purely on the result of INTUIT.
2869 * Set up captures etc just for $& and $-[0]
2870 * (an intuit-only match wont have $1,$2,..) */
2871 assert(!prog->nparens);
2873 /* s/// doesn't like it if $& is earlier than where we asked it to
2874 * start searching (which can happen on something like /.\G/) */
2875 if ( (flags & REXEC_FAIL_ON_UNDERFLOW)
2878 /* this should only be possible under \G */
2879 assert(prog->intflags & PREGf_GPOS_SEEN);
2880 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2881 "matched, but failing for REXEC_FAIL_ON_UNDERFLOW\n"));
2885 /* match via INTUIT shouldn't have any captures.
2886 * Let @-, @+, $^N know */
2887 prog->lastparen = prog->lastcloseparen = 0;
2888 RX_MATCH_UTF8_set(rx, utf8_target);
2889 prog->offs[0].start = s - strbeg;
2890 prog->offs[0].end = utf8_target
2891 ? (char*)utf8_hop((U8*)s, prog->minlenret) - strbeg
2892 : s - strbeg + prog->minlenret;
2893 if ( !(flags & REXEC_NOT_FIRST) )
2894 S_reg_set_capture_string(aTHX_ rx,
2896 sv, flags, utf8_target);
2902 multiline = prog->extflags & RXf_PMf_MULTILINE;
2904 if (strend - s < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) {
2905 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
2906 "String too short [regexec_flags]...\n"));
2910 /* Check validity of program. */
2911 if (UCHARAT(progi->program) != REG_MAGIC) {
2912 Perl_croak(aTHX_ "corrupted regexp program");
2915 RX_MATCH_TAINTED_off(rx);
2916 RX_MATCH_UTF8_set(rx, utf8_target);
2918 reginfo->prog = rx; /* Yes, sorry that this is confusing. */
2919 reginfo->intuit = 0;
2920 reginfo->is_utf8_target = cBOOL(utf8_target);
2921 reginfo->is_utf8_pat = cBOOL(RX_UTF8(rx));
2922 reginfo->warned = FALSE;
2923 reginfo->strbeg = strbeg;
2925 reginfo->poscache_maxiter = 0; /* not yet started a countdown */
2926 reginfo->strend = strend;
2927 /* see how far we have to get to not match where we matched before */
2928 reginfo->till = stringarg + minend;
2930 if (prog->extflags & RXf_EVAL_SEEN && SvPADTMP(sv)) {
2931 /* SAVEFREESV, not sv_mortalcopy, as this SV must last until after
2932 S_cleanup_regmatch_info_aux has executed (registered by
2933 SAVEDESTRUCTOR_X below). S_cleanup_regmatch_info_aux modifies
2934 magic belonging to this SV.
2935 Not newSVsv, either, as it does not COW.
2937 reginfo->sv = newSV(0);
2938 SvSetSV_nosteal(reginfo->sv, sv);
2939 SAVEFREESV(reginfo->sv);
2942 /* reserve next 2 or 3 slots in PL_regmatch_state:
2943 * slot N+0: may currently be in use: skip it
2944 * slot N+1: use for regmatch_info_aux struct
2945 * slot N+2: use for regmatch_info_aux_eval struct if we have (?{})'s
2946 * slot N+3: ready for use by regmatch()
2950 regmatch_state *old_regmatch_state;
2951 regmatch_slab *old_regmatch_slab;
2952 int i, max = (prog->extflags & RXf_EVAL_SEEN) ? 2 : 1;
2954 /* on first ever match, allocate first slab */
2955 if (!PL_regmatch_slab) {
2956 Newx(PL_regmatch_slab, 1, regmatch_slab);
2957 PL_regmatch_slab->prev = NULL;
2958 PL_regmatch_slab->next = NULL;
2959 PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab);
2962 old_regmatch_state = PL_regmatch_state;
2963 old_regmatch_slab = PL_regmatch_slab;
2965 for (i=0; i <= max; i++) {
2967 reginfo->info_aux = &(PL_regmatch_state->u.info_aux);
2969 reginfo->info_aux_eval =
2970 reginfo->info_aux->info_aux_eval =
2971 &(PL_regmatch_state->u.info_aux_eval);
2973 if (++PL_regmatch_state > SLAB_LAST(PL_regmatch_slab))
2974 PL_regmatch_state = S_push_slab(aTHX);
2977 /* note initial PL_regmatch_state position; at end of match we'll
2978 * pop back to there and free any higher slabs */
2980 reginfo->info_aux->old_regmatch_state = old_regmatch_state;
2981 reginfo->info_aux->old_regmatch_slab = old_regmatch_slab;
2982 reginfo->info_aux->poscache = NULL;
2984 SAVEDESTRUCTOR_X(S_cleanup_regmatch_info_aux, reginfo->info_aux);
2986 if ((prog->extflags & RXf_EVAL_SEEN))
2987 S_setup_eval_state(aTHX_ reginfo);
2989 reginfo->info_aux_eval = reginfo->info_aux->info_aux_eval = NULL;
2992 /* If there is a "must appear" string, look for it. */
2994 if (PL_curpm && (PM_GETRE(PL_curpm) == rx)) {
2995 /* We have to be careful. If the previous successful match
2996 was from this regex we don't want a subsequent partially
2997 successful match to clobber the old results.
2998 So when we detect this possibility we add a swap buffer
2999 to the re, and switch the buffer each match. If we fail,
3000 we switch it back; otherwise we leave it swapped.
3003 /* do we need a save destructor here for eval dies? */
3004 Newxz(prog->offs, (prog->nparens + 1), regexp_paren_pair);
3005 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
3006 "rex=0x%"UVxf" saving offs: orig=0x%"UVxf" new=0x%"UVxf"\n",
3013 /* Simplest case: anchored match need be tried only once, or with
3014 * MBOL, only at the beginning of each line.
3016 * Note that /.*.../ sets PREGf_IMPLICIT|MBOL, while /.*.../s sets
3017 * PREGf_IMPLICIT|SBOL. The idea is that with /.*.../s, if it doesn't
3018 * match at the start of the string then it won't match anywhere else
3019 * either; while with /.*.../, if it doesn't match at the beginning,
3020 * the earliest it could match is at the start of the next line */
3022 if (prog->intflags & (PREGf_ANCH & ~PREGf_ANCH_GPOS)) {
3025 if (regtry(reginfo, &s))
3028 if (!(prog->intflags & PREGf_ANCH_MBOL))
3031 /* didn't match at start, try at other newline positions */
3034 dontbother = minlen - 1;
3035 end = HOP3c(strend, -dontbother, strbeg) - 1;
3037 /* skip to next newline */
3039 while (s <= end) { /* note it could be possible to match at the end of the string */
3040 /* NB: newlines are the same in unicode as they are in latin */
3043 if (prog->check_substr || prog->check_utf8) {
3044 /* note that with PREGf_IMPLICIT, intuit can only fail
3045 * or return the start position, so it's of limited utility.
3046 * Nevertheless, I made the decision that the potential for
3047 * quick fail was still worth it - DAPM */
3048 s = re_intuit_start(rx, sv, strbeg, s, strend, flags, NULL);
3052 if (regtry(reginfo, &s))
3056 } /* end anchored search */
3058 if (prog->intflags & PREGf_ANCH_GPOS)
3060 /* PREGf_ANCH_GPOS should never be true if PREGf_GPOS_SEEN is not true */
3061 assert(prog->intflags & PREGf_GPOS_SEEN);
3062 /* For anchored \G, the only position it can match from is
3063 * (ganch-gofs); we already set startpos to this above; if intuit
3064 * moved us on from there, we can't possibly succeed */
3065 assert(startpos == reginfo->ganch - prog->gofs);
3066 if (s == startpos && regtry(reginfo, &s))
3071 /* Messy cases: unanchored match. */
3072 if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) {
3073 /* we have /x+whatever/ */
3074 /* it must be a one character string (XXXX Except is_utf8_pat?) */
3080 if (! prog->anchored_utf8) {
3081 to_utf8_substr(prog);
3083 ch = SvPVX_const(prog->anchored_utf8)[0];
3086 DEBUG_EXECUTE_r( did_match = 1 );
3087 if (regtry(reginfo, &s)) goto got_it;
3089 while (s < strend && *s == ch)
3096 if (! prog->anchored_substr) {
3097 if (! to_byte_substr(prog)) {
3098 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
3101 ch = SvPVX_const(prog->anchored_substr)[0];
3104 DEBUG_EXECUTE_r( did_match = 1 );
3105 if (regtry(reginfo, &s)) goto got_it;
3107 while (s < strend && *s == ch)
3112 DEBUG_EXECUTE_r(if (!did_match)
3113 PerlIO_printf(Perl_debug_log,
3114 "Did not find anchored character...\n")
3117 else if (prog->anchored_substr != NULL
3118 || prog->anchored_utf8 != NULL
3119 || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
3120 && prog->float_max_offset < strend - s)) {
3125 char *last1; /* Last position checked before */
3129 if (prog->anchored_substr || prog->anchored_utf8) {
3131 if (! prog->anchored_utf8) {
3132 to_utf8_substr(prog);
3134 must = prog->anchored_utf8;
3137 if (! prog->anchored_substr) {
3138 if (! to_byte_substr(prog)) {
3139 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
3142 must = prog->anchored_substr;
3144 back_max = back_min = prog->anchored_offset;
3147 if (! prog->float_utf8) {
3148 to_utf8_substr(prog);
3150 must = prog->float_utf8;
3153 if (! prog->float_substr) {
3154 if (! to_byte_substr(prog)) {
3155 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
3158 must = prog->float_substr;
3160 back_max = prog->float_max_offset;
3161 back_min = prog->float_min_offset;
3167 last = HOP3c(strend, /* Cannot start after this */
3168 -(SSize_t)(CHR_SVLEN(must)
3169 - (SvTAIL(must) != 0) + back_min), strbeg);
3171 if (s > reginfo->strbeg)
3172 last1 = HOPc(s, -1);
3174 last1 = s - 1; /* bogus */
3176 /* XXXX check_substr already used to find "s", can optimize if
3177 check_substr==must. */
3179 strend = HOPc(strend, -dontbother);
3180 while ( (s <= last) &&
3181 (s = fbm_instr((unsigned char*)HOP4c(s, back_min, strbeg, strend),
3182 (unsigned char*)strend, must,
3183 multiline ? FBMrf_MULTILINE : 0)) ) {
3184 DEBUG_EXECUTE_r( did_match = 1 );
3185 if (HOPc(s, -back_max) > last1) {
3186 last1 = HOPc(s, -back_min);
3187 s = HOPc(s, -back_max);
3190 char * const t = (last1 >= reginfo->strbeg)
3191 ? HOPc(last1, 1) : last1 + 1;
3193 last1 = HOPc(s, -back_min);
3197 while (s <= last1) {
3198 if (regtry(reginfo, &s))
3201 s++; /* to break out of outer loop */
3208 while (s <= last1) {
3209 if (regtry(reginfo, &s))
3215 DEBUG_EXECUTE_r(if (!did_match) {
3216 RE_PV_QUOTED_DECL(quoted, utf8_target, PERL_DEBUG_PAD_ZERO(0),
3217 SvPVX_const(must), RE_SV_DUMPLEN(must), 30);
3218 PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n",
3219 ((must == prog->anchored_substr || must == prog->anchored_utf8)
3220 ? "anchored" : "floating"),
3221 quoted, RE_SV_TAIL(must));
3225 else if ( (c = progi->regstclass) ) {
3227 const OPCODE op = OP(progi->regstclass);
3228 /* don't bother with what can't match */
3229 if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE)
3230 strend = HOPc(strend, -(minlen - 1));
3233 SV * const prop = sv_newmortal();
3234 regprop(prog, prop, c, reginfo, NULL);
3236 RE_PV_QUOTED_DECL(quoted,utf8_target,PERL_DEBUG_PAD_ZERO(1),
3238 PerlIO_printf(Perl_debug_log,
3239 "Matching stclass %.*s against %s (%d bytes)\n",
3240 (int)SvCUR(prop), SvPVX_const(prop),
3241 quoted, (int)(strend - s));
3244 if (find_byclass(prog, c, s, strend, reginfo))
3246 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n"));
3250 if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
3258 if (! prog->float_utf8) {
3259 to_utf8_substr(prog);
3261 float_real = prog->float_utf8;
3264 if (! prog->float_substr) {
3265 if (! to_byte_substr(prog)) {
3266 NON_UTF8_TARGET_BUT_UTF8_REQUIRED(phooey);
3269 float_real = prog->float_substr;
3272 little = SvPV_const(float_real, len);
3273 if (SvTAIL(float_real)) {
3274 /* This means that float_real contains an artificial \n on
3275 * the end due to the presence of something like this:
3276 * /foo$/ where we can match both "foo" and "foo\n" at the
3277 * end of the string. So we have to compare the end of the
3278 * string first against the float_real without the \n and
3279 * then against the full float_real with the string. We
3280 * have to watch out for cases where the string might be
3281 * smaller than the float_real or the float_real without
3283 char *checkpos= strend - len;
3285 PerlIO_printf(Perl_debug_log,
3286 "%sChecking for float_real.%s\n",
3287 PL_colors[4], PL_colors[5]));
3288 if (checkpos + 1 < strbeg) {
3289 /* can't match, even if we remove the trailing \n
3290 * string is too short to match */
3292 PerlIO_printf(Perl_debug_log,
3293 "%sString shorter than required trailing substring, cannot match.%s\n",
3294 PL_colors[4], PL_colors[5]));
3296 } else if (memEQ(checkpos + 1, little, len - 1)) {
3297 /* can match, the end of the string matches without the
3299 last = checkpos + 1;
3300 } else if (checkpos < strbeg) {
3301 /* cant match, string is too short when the "\n" is
3304 PerlIO_printf(Perl_debug_log,
3305 "%sString does not contain required trailing substring, cannot match.%s\n",
3306 PL_colors[4], PL_colors[5]));
3308 } else if (!multiline) {
3309 /* non multiline match, so compare with the "\n" at the
3310 * end of the string */
3311 if (memEQ(checkpos, little, len)) {
3315 PerlIO_printf(Perl_debug_log,
3316 "%sString does not contain required trailing substring, cannot match.%s\n",
3317 PL_colors[4], PL_colors[5]));
3321 /* multiline match, so we have to search for a place
3322 * where the full string is located */
3328 last = rninstr(s, strend, little, little + len);
3330 last = strend; /* matching "$" */
3333 /* at one point this block contained a comment which was
3334 * probably incorrect, which said that this was a "should not
3335 * happen" case. Even if it was true when it was written I am
3336 * pretty sure it is not anymore, so I have removed the comment
3337 * and replaced it with this one. Yves */
3339 PerlIO_printf(Perl_debug_log,
3340 "%sString does not contain required substring, cannot match.%s\n",
3341 PL_colors[4], PL_colors[5]
3345 dontbother = strend - last + prog->float_min_offset;
3347 if (minlen && (dontbother < minlen))
3348 dontbother = minlen - 1;
3349 strend -= dontbother; /* this one's always in bytes! */
3350 /* We don't know much -- general case. */
3353 if (regtry(reginfo, &s))
3362 if (regtry(reginfo, &s))
3364 } while (s++ < strend);
3372 /* s/// doesn't like it if $& is earlier than where we asked it to
3373 * start searching (which can happen on something like /.\G/) */
3374 if ( (flags & REXEC_FAIL_ON_UNDERFLOW)
3375 && (prog->offs[0].start < stringarg - strbeg))
3377 /* this should only be possible under \G */
3378 assert(prog->intflags & PREGf_GPOS_SEEN);
3379 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
3380 "matched, but failing for REXEC_FAIL_ON_UNDERFLOW\n"));
3386 PerlIO_printf(Perl_debug_log,
3387 "rex=0x%"UVxf" freeing offs: 0x%"UVxf"\n",
3394 /* clean up; this will trigger destructors that will free all slabs
3395 * above the current one, and cleanup the regmatch_info_aux
3396 * and regmatch_info_aux_eval sructs */
3398 LEAVE_SCOPE(oldsave);
3400 if (RXp_PAREN_NAMES(prog))
3401 (void)hv_iterinit(RXp_PAREN_NAMES(prog));
3403 /* make sure $`, $&, $', and $digit will work later */
3404 if ( !(flags & REXEC_NOT_FIRST) )
3405 S_reg_set_capture_string(aTHX_ rx,
3406 strbeg, reginfo->strend,
3407 sv, flags, utf8_target);
3412 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
3413 PL_colors[4], PL_colors[5]));
3415 /* clean up; this will trigger destructors that will free all slabs
3416 * above the current one, and cleanup the regmatch_info_aux
3417 * and regmatch_info_aux_eval sructs */
3419 LEAVE_SCOPE(oldsave);
3422 /* we failed :-( roll it back */
3423 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
3424 "rex=0x%"UVxf" rolling back offs: freeing=0x%"UVxf" restoring=0x%"UVxf"\n",
3429 Safefree(prog->offs);
3436 /* Set which rex is pointed to by PL_reg_curpm, handling ref counting.
3437 * Do inc before dec, in case old and new rex are the same */
3438 #define SET_reg_curpm(Re2) \
3439 if (reginfo->info_aux_eval) { \
3440 (void)ReREFCNT_inc(Re2); \
3441 ReREFCNT_dec(PM_GETRE(PL_reg_curpm)); \
3442 PM_SETRE((PL_reg_curpm), (Re2)); \
3447 - regtry - try match at specific point
3449 STATIC I32 /* 0 failure, 1 success */
3450 S_regtry(pTHX_ regmatch_info *reginfo, char **startposp)
3453 REGEXP *const rx = reginfo->prog;
3454 regexp *const prog = ReANY(rx);
3456 RXi_GET_DECL(prog,progi);
3457 GET_RE_DEBUG_FLAGS_DECL;
3459 PERL_ARGS_ASSERT_REGTRY;
3461 reginfo->cutpoint=NULL;
3463 prog->offs[0].start = *startposp - reginfo->strbeg;
3464 prog->lastparen = 0;
3465 prog->lastcloseparen = 0;
3467 /* XXXX What this code is doing here?!!! There should be no need
3468 to do this again and again, prog->lastparen should take care of
3471 /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code.
3472 * Actually, the code in regcppop() (which Ilya may be meaning by
3473 * prog->lastparen), is not needed at all by the test suite
3474 * (op/regexp, op/pat, op/split), but that code is needed otherwise
3475 * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/
3476 * Meanwhile, this code *is* needed for the
3477 * above-mentioned test suite tests to succeed. The common theme
3478 * on those tests seems to be returning null fields from matches.
3479 * --jhi updated by dapm */
3481 if (prog->nparens) {
3482 regexp_paren_pair *pp = prog->offs;
3484 for (i = prog->nparens; i > (I32)prog->lastparen; i--) {
3492 result = regmatch(reginfo, *startposp, progi->program + 1);
3494 prog->offs[0].end = result;
3497 if (reginfo->cutpoint)
3498 *startposp= reginfo->cutpoint;
3499 REGCP_UNWIND(lastcp);
3504 #define sayYES goto yes
3505 #define sayNO goto no
3506 #define sayNO_SILENT goto no_silent
3508 /* we dont use STMT_START/END here because it leads to
3509 "unreachable code" warnings, which are bogus, but distracting. */
3510 #define CACHEsayNO \
3511 if (ST.cache_mask) \
3512 reginfo->info_aux->poscache[ST.cache_offset] |= ST.cache_mask; \
3515 /* this is used to determine how far from the left messages like
3516 'failed...' are printed. It should be set such that messages
3517 are inline with the regop output that created them.
3519 #define REPORT_CODE_OFF 32
3522 #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */
3523 #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */
3524 #define CHRTEST_NOT_A_CP_1 -999
3525 #define CHRTEST_NOT_A_CP_2 -998
3527 /* grab a new slab and return the first slot in it */
3529 STATIC regmatch_state *
3532 #if PERL_VERSION < 9 && !defined(PERL_CORE)
3535 regmatch_slab *s = PL_regmatch_slab->next;
3537 Newx(s, 1, regmatch_slab);
3538 s->prev = PL_regmatch_slab;
3540 PL_regmatch_slab->next = s;
3542 PL_regmatch_slab = s;
3543 return SLAB_FIRST(s);
3547 /* push a new state then goto it */
3549 #define PUSH_STATE_GOTO(state, node, input) \
3550 pushinput = input; \
3552 st->resume_state = state; \
3555 /* push a new state with success backtracking, then goto it */
3557 #define PUSH_YES_STATE_GOTO(state, node, input) \
3558 pushinput = input; \
3560 st->resume_state = state; \
3561 goto push_yes_state;
3568 regmatch() - main matching routine
3570 This is basically one big switch statement in a loop. We execute an op,
3571 set 'next' to point the next op, and continue. If we come to a point which
3572 we may need to backtrack to on failure such as (A|B|C), we push a
3573 backtrack state onto the backtrack stack. On failure, we pop the top
3574 state, and re-enter the loop at the state indicated. If there are no more
3575 states to pop, we return failure.
3577 Sometimes we also need to backtrack on success; for example /A+/, where
3578 after successfully matching one A, we need to go back and try to
3579 match another one; similarly for lookahead assertions: if the assertion
3580 completes successfully, we backtrack to the state just before the assertion
3581 and then carry on. In these cases, the pushed state is marked as
3582 'backtrack on success too'. This marking is in fact done by a chain of
3583 pointers, each pointing to the previous 'yes' state. On success, we pop to
3584 the nearest yes state, discarding any intermediate failure-only states.
3585 Sometimes a yes state is pushed just to force some cleanup code to be
3586 called at the end of a successful match or submatch; e.g. (??{$re}) uses
3587 it to free the inner regex.
3589 Note that failure backtracking rewinds the cursor position, while
3590 success backtracking leaves it alone.
3592 A pattern is complete when the END op is executed, while a subpattern
3593 such as (?=foo) is complete when the SUCCESS op is executed. Both of these
3594 ops trigger the "pop to last yes state if any, otherwise return true"
3597 A common convention in this function is to use A and B to refer to the two
3598 subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is
3599 the subpattern to be matched possibly multiple times, while B is the entire
3600 rest of the pattern. Variable and state names reflect this convention.
3602 The states in the main switch are the union of ops and failure/success of
3603 substates associated with with that op. For example, IFMATCH is the op
3604 that does lookahead assertions /(?=A)B/ and so the IFMATCH state means
3605 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just
3606 successfully matched A and IFMATCH_A_fail is a state saying that we have
3607 just failed to match A. Resume states always come in pairs. The backtrack
3608 state we push is marked as 'IFMATCH_A', but when that is popped, we resume
3609 at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking
3610 on success or failure.
3612 The struct that holds a backtracking state is actually a big union, with
3613 one variant for each major type of op. The variable st points to the
3614 top-most backtrack struct. To make the code clearer, within each
3615 block of code we #define ST to alias the relevant union.
3617 Here's a concrete example of a (vastly oversimplified) IFMATCH
3623 #define ST st->u.ifmatch
3625 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
3626 ST.foo = ...; // some state we wish to save
3628 // push a yes backtrack state with a resume value of
3629 // IFMATCH_A/IFMATCH_A_fail, then continue execution at the
3631 PUSH_YES_STATE_GOTO(IFMATCH_A, A, newinput);
3634 case IFMATCH_A: // we have successfully executed A; now continue with B
3636 bar = ST.foo; // do something with the preserved value
3639 case IFMATCH_A_fail: // A failed, so the assertion failed
3640 ...; // do some housekeeping, then ...
3641 sayNO; // propagate the failure
3648 For any old-timers reading this who are familiar with the old recursive
3649 approach, the code above is equivalent to:
3651 case IFMATCH: // we are executing the IFMATCH op, (?=A)B
3660 ...; // do some housekeeping, then ...
3661 sayNO; // propagate the failure
3664 The topmost backtrack state, pointed to by st, is usually free. If you
3665 want to claim it, populate any ST.foo fields in it with values you wish to
3666 save, then do one of
3668 PUSH_STATE_GOTO(resume_state, node, newinput);
3669 PUSH_YES_STATE_GOTO(resume_state, node, newinput);
3671 which sets that backtrack state's resume value to 'resume_state', pushes a
3672 new free entry to the top of the backtrack stack, then goes to 'node'.
3673 On backtracking, the free slot is popped, and the saved state becomes the
3674 new free state. An ST.foo field in this new top state can be temporarily
3675 accessed to retrieve values, but once the main loop is re-entered, it
3676 becomes available for reuse.
3678 Note that the depth of the backtrack stack constantly increases during the
3679 left-to-right execution of the pattern, rather than going up and down with
3680 the pattern nesting. For example the stack is at its maximum at Z at the
3681 end of the pattern, rather than at X in the following:
3683 /(((X)+)+)+....(Y)+....Z/
3685 The only exceptions to this are lookahead/behind assertions and the cut,
3686 (?>A), which pop all the backtrack states associated with A before
3689 Backtrack state structs are allocated in slabs of about 4K in size.
3690 PL_regmatch_state and st always point to the currently active state,
3691 and PL_regmatch_slab points to the slab currently containing
3692 PL_regmatch_state. The first time regmatch() is called, the first slab is
3693 allocated, and is never freed until interpreter destruction. When the slab
3694 is full, a new one is allocated and chained to the end. At exit from
3695 regmatch(), slabs allocated since entry are freed.
3700 #define DEBUG_STATE_pp(pp) \
3702 DUMP_EXEC_POS(locinput, scan, utf8_target); \
3703 PerlIO_printf(Perl_debug_log, \
3704 " %*s"pp" %s%s%s%s%s\n", \
3706 PL_reg_name[st->resume_state], \
3707 ((st==yes_state||st==mark_state) ? "[" : ""), \
3708 ((st==yes_state) ? "Y" : ""), \
3709 ((st==mark_state) ? "M" : ""), \
3710 ((st==yes_state||st==mark_state) ? "]" : "") \
3715 #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1)
3720 S_debug_start_match(pTHX_ const REGEXP *prog, const bool utf8_target,
3721 const char *start, const char *end, const char *blurb)
3723 const bool utf8_pat = RX_UTF8(prog) ? 1 : 0;
3725 PERL_ARGS_ASSERT_DEBUG_START_MATCH;
3730 RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0),
3731 RX_PRECOMP_const(prog), RX_PRELEN(prog), 60);
3733 RE_PV_QUOTED_DECL(s1, utf8_target, PERL_DEBUG_PAD_ZERO(1),
3734 start, end - start, 60);
3736 PerlIO_printf(Perl_debug_log,
3737 "%s%s REx%s %s against %s\n",
3738 PL_colors[4], blurb, PL_colors[5], s0, s1);
3740 if (utf8_target||utf8_pat)
3741 PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n",
3742 utf8_pat ? "pattern" : "",
3743 utf8_pat && utf8_target ? " and " : "",
3744 utf8_target ? "string" : ""
3750 S_dump_exec_pos(pTHX_ const char *locinput,
3751 const regnode *scan,
3752 const char *loc_regeol,
3753 const char *loc_bostr,
3754 const char *loc_reg_starttry,
3755 const bool utf8_target)
3757 const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4];
3758 const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
3759 int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput);
3760 /* The part of the string before starttry has one color
3761 (pref0_len chars), between starttry and current
3762 position another one (pref_len - pref0_len chars),
3763 after the current position the third one.
3764 We assume that pref0_len <= pref_len, otherwise we
3765 decrease pref0_len. */
3766 int pref_len = (locinput - loc_bostr) > (5 + taill) - l
3767 ? (5 + taill) - l : locinput - loc_bostr;
3770 PERL_ARGS_ASSERT_DUMP_EXEC_POS;
3772 while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
3774 pref0_len = pref_len - (locinput - loc_reg_starttry);
3775 if (l + pref_len < (5 + taill) && l < loc_regeol - locinput)
3776 l = ( loc_regeol - locinput > (5 + taill) - pref_len
3777 ? (5 + taill) - pref_len : loc_regeol - locinput);
3778 while (utf8_target && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
3782 if (pref0_len > pref_len)
3783 pref0_len = pref_len;
3785 const int is_uni = (utf8_target && OP(scan) != CANY) ? 1 : 0;
3787 RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0),
3788 (locinput - pref_len),pref0_len, 60, 4, 5);
3790 RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1),
3791 (locinput - pref_len + pref0_len),
3792 pref_len - pref0_len, 60, 2, 3);
3794 RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2),
3795 locinput, loc_regeol - locinput, 10, 0, 1);
3797 const STRLEN tlen=len0+len1+len2;
3798 PerlIO_printf(Perl_debug_log,
3799 "%4"IVdf" <%.*s%.*s%s%.*s>%*s|",
3800 (IV)(locinput - loc_bostr),
3803 (docolor ? "" : "> <"),
3805 (int)(tlen > 19 ? 0 : 19 - tlen),
3812 /* reg_check_named_buff_matched()
3813 * Checks to see if a named buffer has matched. The data array of
3814 * buffer numbers corresponding to the buffer is expected to reside
3815 * in the regexp->data->data array in the slot stored in the ARG() of
3816 * node involved. Note that this routine doesn't actually care about the
3817 * name, that information is not preserved from compilation to execution.
3818 * Returns the index of the leftmost defined buffer with the given name
3819 * or 0 if non of the buffers matched.
3822 S_reg_check_named_buff_matched(const regexp *rex, const regnode *scan)
3825 RXi_GET_DECL(rex,rexi);
3826 SV *sv_dat= MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
3827 I32 *nums=(I32*)SvPVX(sv_dat);
3829 PERL_ARGS_ASSERT_REG_CHECK_NAMED_BUFF_MATCHED;
3831 for ( n=0; n<SvIVX(sv_dat); n++ ) {
3832 if ((I32)rex->lastparen >= nums[n] &&
3833 rex->offs[nums[n]].end != -1)
3843 S_setup_EXACTISH_ST_c1_c2(pTHX_ const regnode * const text_node, int *c1p,
3844 U8* c1_utf8, int *c2p, U8* c2_utf8, regmatch_info *reginfo)
3846 /* This function determines if there are one or two characters that match
3847 * the first character of the passed-in EXACTish node <text_node>, and if
3848 * so, returns them in the passed-in pointers.
3850 * If it determines that no possible character in the target string can
3851 * match, it returns FALSE; otherwise TRUE. (The FALSE situation occurs if
3852 * the first character in <text_node> requires UTF-8 to represent, and the
3853 * target string isn't in UTF-8.)
3855 * If there are more than two characters that could match the beginning of
3856 * <text_node>, or if more context is required to determine a match or not,
3857 * it sets both *<c1p> and *<c2p> to CHRTEST_VOID.
3859 * The motiviation behind this function is to allow the caller to set up
3860 * tight loops for matching. If <text_node> is of type EXACT, there is
3861 * only one possible character that can match its first character, and so
3862 * the situation is quite simple. But things get much more complicated if
3863 * folding is involved. It may be that the first character of an EXACTFish
3864 * node doesn't participate in any possible fold, e.g., punctuation, so it
3865 * can be matched only by itself. The vast majority of characters that are
3866 * in folds match just two things, their lower and upper-case equivalents.
3867 * But not all are like that; some have multiple possible matches, or match
3868 * sequences of more than one character. This function sorts all that out.
3870 * Consider the patterns A*B or A*?B where A and B are arbitrary. In a
3871 * loop of trying to match A*, we know we can't exit where the thing
3872 * following it isn't a B. And something can't be a B unless it is the
3873 * beginning of B. By putting a quick test for that beginning in a tight
3874 * loop, we can rule out things that can't possibly be B without having to
3875 * break out of the loop, thus avoiding work. Similarly, if A is a single
3876 * character, we can make a tight loop matching A*, using the outputs of
3879 * If the target string to match isn't in UTF-8, and there aren't
3880 * complications which require CHRTEST_VOID, *<c1p> and *<c2p> are set to
3881 * the one or two possible octets (which are characters in this situation)
3882 * that can match. In all cases, if there is only one character that can
3883 * match, *<c1p> and *<c2p> will be identical.
3885 * If the target string is in UTF-8, the buffers pointed to by <c1_utf8>
3886 * and <c2_utf8> will contain the one or two UTF-8 sequences of bytes that
3887 * can match the beginning of <text_node>. They should be declared with at
3888 * least length UTF8_MAXBYTES+1. (If the target string isn't in UTF-8, it is
3889 * undefined what these contain.) If one or both of the buffers are
3890 * invariant under UTF-8, *<c1p>, and *<c2p> will also be set to the
3891 * corresponding invariant. If variant, the corresponding *<c1p> and/or
3892 * *<c2p> will be set to a negative number(s) that shouldn't match any code
3893 * point (unless inappropriately coerced to unsigned). *<c1p> will equal
3894 * *<c2p> if and only if <c1_utf8> and <c2_utf8> are the same. */
3896 const bool utf8_target = reginfo->is_utf8_target;
3898 UV c1 = (UV)CHRTEST_NOT_A_CP_1;
3899 UV c2 = (UV)CHRTEST_NOT_A_CP_2;
3900 bool use_chrtest_void = FALSE;
3901 const bool is_utf8_pat = reginfo->is_utf8_pat;
3903 /* Used when we have both utf8 input and utf8 output, to avoid converting
3904 * to/from code points */
3905 bool utf8_has_been_setup = FALSE;
3909 U8 *pat = (U8*)STRING(text_node);
3910 U8 folded[UTF8_MAX_FOLD_CHAR_EXPAND * UTF8_MAXBYTES_CASE + 1] = { '\0' };
3912 if (OP(text_node) == EXACT || OP(text_node) == EXACTL) {
3914 /* In an exact node, only one thing can be matched, that first
3915 * character. If both the pat and the target are UTF-8, we can just
3916 * copy the input to the output, avoiding finding the code point of
3921 else if (utf8_target) {
3922 Copy(pat, c1_utf8, UTF8SKIP(pat), U8);
3923 Copy(pat, c2_utf8, UTF8SKIP(pat), U8);
3924 utf8_has_been_setup = TRUE;
3927 c2 = c1 = valid_utf8_to_uvchr(pat, NULL);
3930 else { /* an EXACTFish node */
3931 U8 *pat_end = pat + STR_LEN(text_node);
3933 /* An EXACTFL node has at least some characters unfolded, because what
3934 * they match is not known until now. So, now is the time to fold
3935 * the first few of them, as many as are needed to determine 'c1' and
3936 * 'c2' later in the routine. If the pattern isn't UTF-8, we only need
3937 * to fold if in a UTF-8 locale, and then only the Sharp S; everything
3938 * else is 1-1 and isn't assumed to be folded. In a UTF-8 pattern, we
3939 * need to fold as many characters as a single character can fold to,
3940 * so that later we can check if the first ones are such a multi-char
3941 * fold. But, in such a pattern only locale-problematic characters
3942 * aren't folded, so we can skip this completely if the first character
3943 * in the node isn't one of the tricky ones */
3944 if (OP(text_node) == EXACTFL) {
3946 if (! is_utf8_pat) {
3947 if (IN_UTF8_CTYPE_LOCALE && *pat == LATIN_SMALL_LETTER_SHARP_S)
3949 folded[0] = folded[1] = 's';
3951 pat_end = folded + 2;
3954 else if (is_PROBLEMATIC_LOCALE_FOLDEDS_START_utf8(pat)) {
3959 for (i = 0; i < UTF8_MAX_FOLD_CHAR_EXPAND && s < pat_end; i++) {
3961 *(d++) = (U8) toFOLD_LC(*s);
3966 _to_utf8_fold_flags(s,
3969 FOLD_FLAGS_FULL | FOLD_FLAGS_LOCALE);
3980 if ((is_utf8_pat && is_MULTI_CHAR_FOLD_utf8_safe(pat, pat_end))
3981 || (!is_utf8_pat && is_MULTI_CHAR_FOLD_latin1_safe(pat, pat_end)))
3983 /* Multi-character folds require more context to sort out. Also
3984 * PL_utf8_foldclosures used below doesn't handle them, so have to
3985 * be handled outside this routine */
3986 use_chrtest_void = TRUE;
3988 else { /* an EXACTFish node which doesn't begin with a multi-char fold */
3989 c1 = is_utf8_pat ? valid_utf8_to_uvchr(pat, NULL) : *pat;
3991 /* Load the folds hash, if not already done */
3993 if (! PL_utf8_foldclosures) {
3994 _load_PL_utf8_foldclosures();
3997 /* The fold closures data structure is a hash with the keys
3998 * being the UTF-8 of every character that is folded to, like
3999 * 'k', and the values each an array of all code points that
4000 * fold to its key. e.g. [ 'k', 'K', KELVIN_SIGN ].
4001 * Multi-character folds are not included */
4002 if ((! (listp = hv_fetch(PL_utf8_foldclosures,
4007 /* Not found in the hash, therefore there are no folds
4008 * containing it, so there is only a single character that
4012 else { /* Does participate in folds */
4013 AV* list = (AV*) *listp;
4014 if (av_tindex(list) != 1) {
4016 /* If there aren't exactly two folds to this, it is
4017 * outside the scope of this function */
4018 use_chrtest_void = TRUE;
4020 else { /* There are two. Get them */
4021 SV** c_p = av_fetch(list, 0, FALSE);
4023 Perl_croak(aTHX_ "panic: invalid PL_utf8_foldclosures structure");
4027 c_p = av_fetch(list, 1, FALSE);
4029 Perl_croak(aTHX_ "panic: invalid PL_utf8_foldclosures structure");
4033 /* Folds that cross the 255/256 boundary are forbidden
4034 * if EXACTFL (and isnt a UTF8 locale), or EXACTFA and
4035 * one is ASCIII. Since the pattern character is above
4036 * 255, and its only other match is below 256, the only
4037 * legal match will be to itself. We have thrown away
4038 * the original, so have to compute which is the one
4040 if ((c1 < 256) != (c2 < 256)) {
4041 if ((OP(text_node) == EXACTFL
4042 && ! IN_UTF8_CTYPE_LOCALE)
4043 || ((OP(text_node) == EXACTFA
4044 || OP(text_node) == EXACTFA_NO_TRIE)
4045 && (isASCII(c1) || isASCII(c2))))
4058 else /* Here, c1 is <= 255 */
4060 && HAS_NONLATIN1_FOLD_CLOSURE(c1)
4061 && ( ! (OP(text_node) == EXACTFL && ! IN_UTF8_CTYPE_LOCALE))
4062 && ((OP(text_node) != EXACTFA
4063 && OP(text_node) != EXACTFA_NO_TRIE)
4066 /* Here, there could be something above Latin1 in the target
4067 * which folds to this character in the pattern. All such
4068 * cases except LATIN SMALL LETTER Y WITH DIAERESIS have more
4069 * than two characters involved in their folds, so are outside
4070 * the scope of this function */
4071 if (UNLIKELY(c1 == LATIN_SMALL_LETTER_Y_WITH_DIAERESIS)) {
4072 c2 = LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS;
4075 use_chrtest_void = TRUE;
4078 else { /* Here nothing above Latin1 can fold to the pattern
4080 switch (OP(text_node)) {
4082 case EXACTFL: /* /l rules */
4083 c2 = PL_fold_locale[c1];
4086 case EXACTF: /* This node only generated for non-utf8
4088 assert(! is_utf8_pat);
4089 if (! utf8_target) { /* /d rules */
4094 /* /u rules for all these. This happens to work for
4095 * EXACTFA as nothing in Latin1 folds to ASCII */
4096 case EXACTFA_NO_TRIE: /* This node only generated for
4097 non-utf8 patterns */
4098 assert(! is_utf8_pat);
4103 c2 = PL_fold_latin1[c1];
4107 Perl_croak(aTHX_ "panic: Unexpected op %u", OP(text_node));
4108 NOT_REACHED; /* NOTREACHED */
4114 /* Here have figured things out. Set up the returns */
4115 if (use_chrtest_void) {
4116 *c2p = *c1p = CHRTEST_VOID;
4118 else if (utf8_target) {
4119 if (! utf8_has_been_setup) { /* Don't have the utf8; must get it */
4120 uvchr_to_utf8(c1_utf8, c1);
4121 uvchr_to_utf8(c2_utf8, c2);
4124 /* Invariants are stored in both the utf8 and byte outputs; Use
4125 * negative numbers otherwise for the byte ones. Make sure that the
4126 * byte ones are the same iff the utf8 ones are the same */
4127 *c1p = (UTF8_IS_INVARIANT(*c1_utf8)) ? *c1_utf8 : CHRTEST_NOT_A_CP_1;
4128 *c2p = (UTF8_IS_INVARIANT(*c2_utf8))
4131 ? CHRTEST_NOT_A_CP_1
4132 : CHRTEST_NOT_A_CP_2;
4134 else if (c1 > 255) {
4135 if (c2 > 255) { /* both possibilities are above what a non-utf8 string
4140 *c1p = *c2p = c2; /* c2 is the only representable value */
4142 else { /* c1 is representable; see about c2 */
4144 *c2p = (c2 < 256) ? c2 : c1;
4150 /* This creates a single number by combining two, with 'before' being like the
4151 * 10's digit, but this isn't necessarily base 10; it is base however many
4152 * elements of the enum there are */
4153 #define GCBcase(before, after) ((PL_GCB_ENUM_COUNT * before) + after)
4156 S_isGCB(const PL_GCB_enum before, const PL_GCB_enum after)
4158 /* returns a boolean indicating if there is a Grapheme Cluster Boundary
4159 * between the inputs. See http://www.unicode.org/reports/tr29/ */
4161 switch (GCBcase(before, after)) {
4163 /* Break at the start and end of text.
4167 Break before and after controls except between CR and LF
4168 GB4. ( Control | CR | LF ) ÷
4169 GB5. ÷ ( Control | CR | LF )
4171 Otherwise, break everywhere.
4176 /* Do not break between a CR and LF.
4178 case GCBcase(PL_GCB_CR, PL_GCB_LF):
4181 /* Do not break Hangul syllable sequences.
4182 GB6. L × ( L | V | LV | LVT ) */
4183 case GCBcase(PL_GCB_L, PL_GCB_L):
4184 case GCBcase(PL_GCB_L, PL_GCB_V):
4185 case GCBcase(PL_GCB_L, PL_GCB_LV):
4186 case GCBcase(PL_GCB_L, PL_GCB_LVT):
4189 /* GB7. ( LV | V ) × ( V | T ) */
4190 case GCBcase(PL_GCB_LV, PL_GCB_V):
4191 case GCBcase(PL_GCB_LV, PL_GCB_T):
4192 case GCBcase(PL_GCB_V, PL_GCB_V):
4193 case GCBcase(PL_GCB_V, PL_GCB_T):
4196 /* GB8. ( LVT | T) × T */
4197 case GCBcase(PL_GCB_LVT, PL_GCB_T):
4198 case GCBcase(PL_GCB_T, PL_GCB_T):
4201 /* Do not break between regional indicator symbols.
4202 GB8a. Regional_Indicator × Regional_Indicator */
4203 case GCBcase(PL_GCB_Regional_Indicator, PL_GCB_Regional_Indicator):
4206 /* Do not break before extending characters.
4208 case GCBcase(PL_GCB_Other, PL_GCB_Extend):
4209 case GCBcase(PL_GCB_Extend, PL_GCB_Extend):
4210 case GCBcase(PL_GCB_L, PL_GCB_Extend):
4211 case GCBcase(PL_GCB_LV, PL_GCB_Extend):
4212 case GCBcase(PL_GCB_LVT, PL_GCB_Extend):
4213 case GCBcase(PL_GCB_Prepend, PL_GCB_Extend):
4214 case GCBcase(PL_GCB_Regional_Indicator, PL_GCB_Extend):
4215 case GCBcase(PL_GCB_SpacingMark, PL_GCB_Extend):
4216 case GCBcase(PL_GCB_T, PL_GCB_Extend):
4217 case GCBcase(PL_GCB_V, PL_GCB_Extend):
4220 /* Do not break before SpacingMarks, or after Prepend characters.
4221 GB9a. × SpacingMark */
4222 case GCBcase(PL_GCB_Other, PL_GCB_SpacingMark):
4223 case GCBcase(PL_GCB_Extend, PL_GCB_SpacingMark):
4224 case GCBcase(PL_GCB_L, PL_GCB_SpacingMark):
4225 case GCBcase(PL_GCB_LV, PL_GCB_SpacingMark):
4226 case GCBcase(PL_GCB_LVT, PL_GCB_SpacingMark):
4227 case GCBcase(PL_GCB_Prepend, PL_GCB_SpacingMark):
4228 case GCBcase(PL_GCB_Regional_Indicator, PL_GCB_SpacingMark):
4229 case GCBcase(PL_GCB_SpacingMark, PL_GCB_SpacingMark):
4230 case GCBcase(PL_GCB_T, PL_GCB_SpacingMark):
4231 case GCBcase(PL_GCB_V, PL_GCB_SpacingMark):
4234 /* GB9b. Prepend × */
4235 case GCBcase(PL_GCB_Prepend, PL_GCB_Other):
4236 case GCBcase(PL_GCB_Prepend, PL_GCB_L):
4237 case GCBcase(PL_GCB_Prepend, PL_GCB_LV):
4238 case GCBcase(PL_GCB_Prepend, PL_GCB_LVT):
4239 case GCBcase(PL_GCB_Prepend, PL_GCB_Prepend):
4240 case GCBcase(PL_GCB_Prepend, PL_GCB_Regional_Indicator):
4241 case GCBcase(PL_GCB_Prepend, PL_GCB_T):
4242 case GCBcase(PL_GCB_Prepend, PL_GCB_V):
4246 NOT_REACHED; /* NOTREACHED */
4249 #define SBcase(before, after) ((SB_ENUM_COUNT * before) + after)
4252 S_isSB(pTHX_ PL_SB_enum before,
4254 const U8 * const strbeg,
4255 const U8 * const curpos,
4256 const U8 * const strend,
4257 const bool utf8_target)
4259 /* returns a boolean indicating if there is a Sentence Boundary Break
4260 * between the inputs. See http://www.unicode.org/reports/tr29/ */
4262 U8 * lpos = (U8 *) curpos;
4266 PERL_ARGS_ASSERT_ISSB;
4268 /* Break at the start and end of text.
4271 if (before == PL_SB_EDGE || after == PL_SB_EDGE) {
4275 /* SB 3: Do not break within CRLF. */
4276 if (before == PL_SB_CR && after == PL_SB_LF) {
4280 /* Break after paragraph separators. (though why CR and LF are considered
4281 * so is beyond me (khw)
4282 SB4. Sep | CR | LF ÷ */
4283 if (before == PL_SB_Sep || before == PL_SB_CR || before == PL_SB_LF) {
4287 /* Ignore Format and Extend characters, except after sot, Sep, CR, or LF.
4288 * (See Section 6.2, Replacing Ignore Rules.)
4289 SB5. X (Extend | Format)* → X */
4290 if (after == PL_SB_Extend || after == PL_SB_Format) {
4294 if (before == PL_SB_Extend || before == PL_SB_Format) {
4295 before = backup_one_SB(strbeg, &lpos, utf8_target);
4298 /* Do not break after ambiguous terminators like period, if they are
4299 * immediately followed by a number or lowercase letter, if they are
4300 * between uppercase letters, if the first following letter (optionally
4301 * after certain punctuation) is lowercase, or if they are followed by
4302 * "continuation" punctuation such as comma, colon, or semicolon. For
4303 * example, a period may be an abbreviation or numeric period, and thus may
4304 * not mark the end of a sentence.
4306 * SB6. ATerm × Numeric */
4307 if (before == PL_SB_ATerm && after == PL_SB_Numeric) {
4311 /* SB7. Upper ATerm × Upper */
4312 if (before == PL_SB_ATerm && after == PL_SB_Upper) {
4314 if (PL_SB_Upper == backup_one_SB(strbeg, &temp_pos, utf8_target)) {
4319 /* SB8a. (STerm | ATerm) Close* Sp* × (SContinue | STerm | ATerm)
4320 * SB10. (STerm | ATerm) Close* Sp* × ( Sp | Sep | CR | LF ) */
4323 while (backup == PL_SB_Sp) {
4324 backup = backup_one_SB(strbeg, &temp_pos, utf8_target);
4326 while (backup == PL_SB_Close) {
4327 backup = backup_one_SB(strbeg, &temp_pos, utf8_target);
4329 if ((backup == PL_SB_STerm || backup == PL_SB_ATerm)
4330 && ( after == PL_SB_SContinue
4331 || after == PL_SB_STerm
4332 || after == PL_SB_ATerm
4333 || after == PL_SB_Sp
4334 || after == PL_SB_Sep
4335 || after == PL_SB_CR
4336 || after == PL_SB_LF))
4341 /* SB8. ATerm Close* Sp* × ( ¬(OLetter | Upper | Lower | Sep | CR | LF |
4342 * STerm | ATerm) )* Lower */
4343 if (backup == PL_SB_ATerm) {
4344 U8 * rpos = (U8 *) curpos;
4345 PL_SB_enum later = after;
4347 while ( later != PL_SB_OLetter
4348 && later != PL_SB_Upper
4349 && later != PL_SB_Lower
4350 && later != PL_SB_Sep
4351 && later != PL_SB_CR
4352 && later != PL_SB_LF
4353 && later != PL_SB_STerm
4354 && later != PL_SB_ATerm
4355 && later != PL_SB_EDGE)
4357 later = advance_one_SB(&rpos, strend, utf8_target);
4359 if (later == PL_SB_Lower) {
4364 /* Break after sentence terminators, but include closing punctuation,
4365 * trailing spaces, and a paragraph separator (if present). [See note
4367 * SB9. ( STerm | ATerm ) Close* × ( Close | Sp | Sep | CR | LF ) */
4370 while (backup == PL_SB_Close) {
4371 backup = backup_one_SB(strbeg, &temp_pos, utf8_target);
4373 if ((backup == PL_SB_STerm || backup == PL_SB_ATerm)
4374 && ( after == PL_SB_Close
4375 || after == PL_SB_Sp
4376 || after == PL_SB_Sep
4377 || after == PL_SB_CR
4378 || after == PL_SB_LF))
4384 /* SB11. ( STerm | ATerm ) Close* Sp* ( Sep | CR | LF )? ÷ */
4386 backup = backup_one_SB(strbeg, &temp_pos, utf8_target);
4387 if ( backup == PL_SB_Sep
4388 || backup == PL_SB_CR
4389 || backup == PL_SB_LF)
4396 while (backup == PL_SB_Sp) {
4397 backup = backup_one_SB(strbeg, &lpos, utf8_target);
4399 while (backup == PL_SB_Close) {
4400 backup = backup_one_SB(strbeg, &lpos, utf8_target);
4402 if (backup == PL_SB_STerm || backup == PL_SB_ATerm) {
4406 /* Otherwise, do not break.
4413 S_advance_one_SB(pTHX_ U8 ** curpos, const U8 * const strend, const bool utf8_target)
4417 PERL_ARGS_ASSERT_ADVANCE_ONE_SB;
4419 if (*curpos >= strend) {
4425 *curpos += UTF8SKIP(*curpos);
4426 if (*curpos >= strend) {
4429 sb = getSB_VAL_UTF8(*curpos, strend);
4430 } while (sb == PL_SB_Extend || sb == PL_SB_Format);
4435 if (*curpos >= strend) {
4438 sb = getSB_VAL_CP(**curpos);
4439 } while (sb == PL_SB_Extend || sb == PL_SB_Format);
4446 S_backup_one_SB(pTHX_ const U8 * const strbeg, U8 ** curpos, const bool utf8_target)
4450 PERL_ARGS_ASSERT_BACKUP_ONE_SB;
4452 if (*curpos < strbeg) {
4457 U8 * prev_char_pos = reghopmaybe3(*curpos, -1, strbeg);
4458 if (! prev_char_pos) {
4462 /* Back up over Extend and Format. curpos is always just to the right
4463 * of the characater whose value we are getting */
4465 U8 * prev_prev_char_pos;
4466 if ((prev_prev_char_pos = reghopmaybe3((U8 *) prev_char_pos, -1,
4469 sb = getSB_VAL_UTF8(prev_prev_char_pos, prev_char_pos);
4470 *curpos = prev_char_pos;
4471 prev_char_pos = prev_prev_char_pos;
4474 *curpos = (U8 *) strbeg;
4477 } while (sb == PL_SB_Extend || sb == PL_SB_Format);
4481 if (*curpos - 2 < strbeg) {
4482 *curpos = (U8 *) strbeg;
4486 sb = getSB_VAL_CP(*(*curpos - 1));
4487 } while (sb == PL_SB_Extend || sb == PL_SB_Format);
4493 #define WBcase(before, after) ((PL_WB_ENUM_COUNT * before) + after)
4496 S_isWB(pTHX_ PL_WB_enum previous,
4499 const U8 * const strbeg,
4500 const U8 * const curpos,
4501 const U8 * const strend,
4502 const bool utf8_target)
4504 /* Return a boolean as to if the boundary between 'before' and 'after' is
4505 * a Unicode word break, using their published algorithm. Context may be
4506 * needed to make this determination. If the value for the character
4507 * before 'before' is known, it is passed as 'previous'; otherwise that
4508 * should be set to PL_WB_UNKNOWN. The other input parameters give the
4509 * boundaries and current position in the matching of the string. That
4510 * is, 'curpos' marks the position where the character whose wb value is
4511 * 'after' begins. See http://www.unicode.org/reports/tr29/ */
4513 U8 * before_pos = (U8 *) curpos;
4514 U8 * after_pos = (U8 *) curpos;
4516 PERL_ARGS_ASSERT_ISWB;
4518 /* WB1 and WB2: Break at the start and end of text. */
4519 if (before == PL_WB_EDGE || after == PL_WB_EDGE) {
4523 /* WB 3: Do not break within CRLF. */
4524 if (before == PL_WB_CR && after == PL_WB_LF) {
4528 /* WB 3a and WB 3b: Otherwise break before and after Newlines (including CR
4530 if ( before == PL_WB_CR || before == PL_WB_LF || before == PL_WB_Newline
4531 || after == PL_WB_CR || after == PL_WB_LF || after == PL_WB_Newline)
4536 /* Ignore Format and Extend characters, except when they appear at the
4537 * beginning of a region of text.
4538 * WB4. X (Extend | Format)* → X. */
4540 if (after == PL_WB_Extend || after == PL_WB_Format) {
4544 if (before == PL_WB_Extend || before == PL_WB_Format) {
4545 before = backup_one_WB(&previous, strbeg, &before_pos, utf8_target);
4548 switch (WBcase(before, after)) {
4549 /* Otherwise, break everywhere (including around ideographs).
4554 /* Do not break between most letters.
4555 WB5. (ALetter | Hebrew_Letter) × (ALetter | Hebrew_Letter) */
4556 case WBcase(PL_WB_ALetter, PL_WB_ALetter):
4557 case WBcase(PL_WB_ALetter, PL_WB_Hebrew_Letter):
4558 case WBcase(PL_WB_Hebrew_Letter, PL_WB_ALetter):
4559 case WBcase(PL_WB_Hebrew_Letter, PL_WB_Hebrew_Letter):
4562 /* Do not break letters across certain punctuation.
4563 WB6. (ALetter | Hebrew_Letter)
4564 × (MidLetter | MidNumLet | Single_Quote) (ALetter
4566 case WBcase(PL_WB_ALetter, PL_WB_MidLetter):
4567 case WBcase(PL_WB_ALetter, PL_WB_MidNumLet):
4568 case WBcase(PL_WB_ALetter, PL_WB_Single_Quote):
4569 case WBcase(PL_WB_Hebrew_Letter, PL_WB_MidLetter):
4570 case WBcase(PL_WB_Hebrew_Letter, PL_WB_MidNumLet):
4571 /*case WBcase(PL_WB_Hebrew_Letter, PL_WB_Single_Quote):*/
4572 after = advance_one_WB(&after_pos, strend, utf8_target);
4573 return after != PL_WB_ALetter && after != PL_WB_Hebrew_Letter;
4575 /* WB7. (ALetter | Hebrew_Letter) (MidLetter | MidNumLet |
4576 * Single_Quote) × (ALetter | Hebrew_Letter) */
4577 case WBcase(PL_WB_MidLetter, PL_WB_ALetter):
4578 case WBcase(PL_WB_MidLetter, PL_WB_Hebrew_Letter):
4579 case WBcase(PL_WB_MidNumLet, PL_WB_ALetter):
4580 case WBcase(PL_WB_MidNumLet, PL_WB_Hebrew_Letter):
4581 case WBcase(PL_WB_Single_Quote, PL_WB_ALetter):
4582 case WBcase(PL_WB_Single_Quote, PL_WB_Hebrew_Letter):
4584 = backup_one_WB(&previous, strbeg, &before_pos, utf8_target);
4585 return before != PL_WB_ALetter && before != PL_WB_Hebrew_Letter;
4587 /* WB7a. Hebrew_Letter × Single_Quote */
4588 case WBcase(PL_WB_Hebrew_Letter, PL_WB_Single_Quote):
4591 /* WB7b. Hebrew_Letter × Double_Quote Hebrew_Letter */
4592 case WBcase(PL_WB_Hebrew_Letter, PL_WB_Double_Quote):
4593 return advance_one_WB(&after_pos, strend, utf8_target)
4594 != PL_WB_Hebrew_Letter;
4596 /* WB7c. Hebrew_Letter Double_Quote × Hebrew_Letter */
4597 case WBcase(PL_WB_Double_Quote, PL_WB_Hebrew_Letter):
4598 return backup_one_WB(&previous, strbeg, &before_pos, utf8_target)
4599 != PL_WB_Hebrew_Letter;
4601 /* Do not break within sequences of digits, or digits adjacent to
4602 * letters (“3a”, or “A3”).
4603 WB8. Numeric × Numeric */
4604 case WBcase(PL_WB_Numeric, PL_WB_Numeric):
4607 /* WB9. (ALetter | Hebrew_Letter) × Numeric */
4608 case WBcase(PL_WB_ALetter, PL_WB_Numeric):
4609 case WBcase(PL_WB_Hebrew_Letter, PL_WB_Numeric):
4612 /* WB10. Numeric × (ALetter | Hebrew_Letter) */
4613 case WBcase(PL_WB_Numeric, PL_WB_ALetter):
4614 case WBcase(PL_WB_Numeric, PL_WB_Hebrew_Letter):
4617 /* Do not break within sequences, such as “3.2” or “3,456.789”.
4618 WB11. Numeric (MidNum | MidNumLet | Single_Quote) × Numeric
4620 case WBcase(PL_WB_MidNum, PL_WB_Numeric):
4621 case WBcase(PL_WB_MidNumLet, PL_WB_Numeric):
4622 case WBcase(PL_WB_Single_Quote, PL_WB_Numeric):
4623 return backup_one_WB(&previous, strbeg, &before_pos, utf8_target)
4626 /* WB12. Numeric × (MidNum | MidNumLet | Single_Quote) Numeric
4628 case WBcase(PL_WB_Numeric, PL_WB_MidNum):
4629 case WBcase(PL_WB_Numeric, PL_WB_MidNumLet):
4630 case WBcase(PL_WB_Numeric, PL_WB_Single_Quote):
4631 return advance_one_WB(&after_pos, strend, utf8_target)
4634 /* Do not break between Katakana.
4635 WB13. Katakana × Katakana */
4636 case WBcase(PL_WB_Katakana, PL_WB_Katakana):
4639 /* Do not break from extenders.
4640 WB13a. (ALetter | Hebrew_Letter | Numeric | Katakana |
4641 ExtendNumLet) × ExtendNumLet */
4642 case WBcase(PL_WB_ALetter, PL_WB_ExtendNumLet):
4643 case WBcase(PL_WB_Hebrew_Letter, PL_WB_ExtendNumLet):
4644 case WBcase(PL_WB_Numeric, PL_WB_ExtendNumLet):
4645 case WBcase(PL_WB_Katakana, PL_WB_ExtendNumLet):
4646 case WBcase(PL_WB_ExtendNumLet, PL_WB_ExtendNumLet):
4649 /* WB13b. ExtendNumLet × (ALetter | Hebrew_Letter | Numeric
4651 case WBcase(PL_WB_ExtendNumLet, PL_WB_ALetter):
4652 case WBcase(PL_WB_ExtendNumLet, PL_WB_Hebrew_Letter):
4653 case WBcase(PL_WB_ExtendNumLet, PL_WB_Numeric):
4654 case WBcase(PL_WB_ExtendNumLet, PL_WB_Katakana):
4657 /* Do not break between regional indicator symbols.
4658 WB13c. Regional_Indicator × Regional_Indicator */
4659 case WBcase(PL_WB_Regional_Indicator, PL_WB_Regional_Indicator):
4664 NOT_REACHED; /* NOTREACHED */
4668 S_advance_one_WB(pTHX_ U8 ** curpos, const U8 * const strend, const bool utf8_target)
4672 PERL_ARGS_ASSERT_ADVANCE_ONE_WB;
4674 if (*curpos >= strend) {
4680 /* Advance over Extend and Format */
4682 *curpos += UTF8SKIP(*curpos);
4683 if (*curpos >= strend) {
4686 wb = getWB_VAL_UTF8(*curpos, strend);
4687 } while (wb == PL_WB_Extend || wb == PL_WB_Format);
4692 if (*curpos >= strend) {
4695 wb = getWB_VAL_CP(**curpos);
4696 } while (wb == PL_WB_Extend || wb == PL_WB_Format);
4703 S_backup_one_WB(pTHX_ PL_WB_enum * previous, const U8 * const strbeg, U8 ** curpos, const bool utf8_target)
4707 PERL_ARGS_ASSERT_BACKUP_ONE_WB;
4709 /* If we know what the previous character's break value is, don't have
4711 if (*previous != PL_WB_UNKNOWN) {
4713 *previous = PL_WB_UNKNOWN;
4714 /* XXX Note that doesn't change curpos, and maybe should */
4716 /* But we always back up over these two types */
4717 if (wb != PL_WB_Extend && wb != PL_WB_Format) {
4722 if (*curpos < strbeg) {
4727 U8 * prev_char_pos = reghopmaybe3(*curpos, -1, strbeg);
4728 if (! prev_char_pos) {
4732 /* Back up over Extend and Format. curpos is always just to the right
4733 * of the characater whose value we are getting */
4735 U8 * prev_prev_char_pos;
4736 if ((prev_prev_char_pos = reghopmaybe3((U8 *) prev_char_pos,
4740 wb = getWB_VAL_UTF8(prev_prev_char_pos, prev_char_pos);
4741 *curpos = prev_char_pos;
4742 prev_char_pos = prev_prev_char_pos;
4745 *curpos = (U8 *) strbeg;
4748 } while (wb == PL_WB_Extend || wb == PL_WB_Format);
4752 if (*curpos - 2 < strbeg) {
4753 *curpos = (U8 *) strbeg;
4757 wb = getWB_VAL_CP(*(*curpos - 1));
4758 } while (wb == PL_WB_Extend || wb == PL_WB_Format);
4764 /* returns -1 on failure, $+[0] on success */
4766 S_regmatch(pTHX_ regmatch_info *reginfo, char *startpos, regnode *prog)
4768 #if PERL_VERSION < 9 && !defined(PERL_CORE)
4772 const bool utf8_target = reginfo->is_utf8_target;
4773 const U32 uniflags = UTF8_ALLOW_DEFAULT;
4774 REGEXP *rex_sv = reginfo->prog;
4775 regexp *rex = ReANY(rex_sv);
4776 RXi_GET_DECL(rex,rexi);
4777 /* the current state. This is a cached copy of PL_regmatch_state */
4779 /* cache heavy used fields of st in registers */
4782 U32 n = 0; /* general value; init to avoid compiler warning */
4783 SSize_t ln = 0; /* len or last; init to avoid compiler warning */
4784 char *locinput = startpos;
4785 char *pushinput; /* where to continue after a PUSH */
4786 I32 nextchr; /* is always set to UCHARAT(locinput) */
4788 bool result = 0; /* return value of S_regmatch */
4789 int depth = 0; /* depth of backtrack stack */
4790 U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */
4791 const U32 max_nochange_depth =
4792 (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ?
4793 3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH;
4794 regmatch_state *yes_state = NULL; /* state to pop to on success of
4796 /* mark_state piggy backs on the yes_state logic so that when we unwind
4797 the stack on success we can update the mark_state as we go */
4798 regmatch_state *mark_state = NULL; /* last mark state we have seen */
4799 regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */
4800 struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */
4802 bool no_final = 0; /* prevent failure from backtracking? */
4803 bool do_cutgroup = 0; /* no_final only until next branch/trie entry */
4804 char *startpoint = locinput;
4805 SV *popmark = NULL; /* are we looking for a mark? */
4806 SV *sv_commit = NULL; /* last mark name seen in failure */
4807 SV *sv_yes_mark = NULL; /* last mark name we have seen
4808 during a successful match */
4809 U32 lastopen = 0; /* last open we saw */
4810 bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0;
4811 SV* const oreplsv = GvSVn(PL_replgv);
4812 /* these three flags are set by various ops to signal information to
4813 * the very next op. They have a useful lifetime of exactly one loop
4814 * iteration, and are not preserved or restored by state pushes/pops
4816 bool sw = 0; /* the condition value in (?(cond)a|b) */
4817 bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */
4818 int logical = 0; /* the following EVAL is:
4822 or the following IFMATCH/UNLESSM is:
4823 false: plain (?=foo)
4824 true: used as a condition: (?(?=foo))
4826 PAD* last_pad = NULL;
4828 I32 gimme = G_SCALAR;
4829 CV *caller_cv = NULL; /* who called us */
4830 CV *last_pushed_cv = NULL; /* most recently called (?{}) CV */
4831 CHECKPOINT runops_cp; /* savestack position before executing EVAL */
4832 U32 maxopenparen = 0; /* max '(' index seen so far */
4833 int to_complement; /* Invert the result? */
4834 _char_class_number classnum;
4835 bool is_utf8_pat = reginfo->is_utf8_pat;
4840 GET_RE_DEBUG_FLAGS_DECL;
4843 /* protect against undef(*^R) */
4844 SAVEFREESV(SvREFCNT_inc_simple_NN(oreplsv));
4846 /* shut up 'may be used uninitialized' compiler warnings for dMULTICALL */
4847 multicall_oldcatch = 0;
4848 multicall_cv = NULL;
4850 PERL_UNUSED_VAR(multicall_cop);
4851 PERL_UNUSED_VAR(newsp);
4854 PERL_ARGS_ASSERT_REGMATCH;
4856 DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({
4857 PerlIO_printf(Perl_debug_log,"regmatch start\n");
4860 st = PL_regmatch_state;
4862 /* Note that nextchr is a byte even in UTF */
4865 while (scan != NULL) {
4868 SV * const prop = sv_newmortal();
4869 regnode *rnext=regnext(scan);
4870 DUMP_EXEC_POS( locinput, scan, utf8_target );
4871 regprop(rex, prop, scan, reginfo, NULL);
4873 PerlIO_printf(Perl_debug_log,
4874 "%3"IVdf":%*s%s(%"IVdf")\n",
4875 (IV)(scan - rexi->program), depth*2, "",
4877 (PL_regkind[OP(scan)] == END || !rnext) ?
4878 0 : (IV)(rnext - rexi->program));
4881 next = scan + NEXT_OFF(scan);
4884 state_num = OP(scan);
4890 assert(nextchr < 256 && (nextchr >= 0 || nextchr == NEXTCHR_EOS));
4892 switch (state_num) {
4893 case SBOL: /* /^../ and /\A../ */
4894 if (locinput == reginfo->strbeg)
4898 case MBOL: /* /^../m */
4899 if (locinput == reginfo->strbeg ||
4900 (!NEXTCHR_IS_EOS && locinput[-1] == '\n'))
4907 if (locinput == reginfo->ganch)
4911 case KEEPS: /* \K */
4912 /* update the startpoint */
4913 st->u.keeper.val = rex->offs[0].start;
4914 rex->offs[0].start = locinput - reginfo->strbeg;
4915 PUSH_STATE_GOTO(KEEPS_next, next, locinput);
4917 NOT_REACHED; /* NOTREACHED */
4919 case KEEPS_next_fail:
4920 /* rollback the start point change */
4921 rex->offs[0].start = st->u.keeper.val;
4924 NOT_REACHED; /* NOTREACHED */
4926 case MEOL: /* /..$/m */
4927 if (!NEXTCHR_IS_EOS && nextchr != '\n')
4931 case SEOL: /* /..$/ */
4932 if (!NEXTCHR_IS_EOS && nextchr != '\n')
4934 if (reginfo->strend - locinput > 1)
4939 if (!NEXTCHR_IS_EOS)
4943 case SANY: /* /./s */
4946 goto increment_locinput;
4954 case REG_ANY: /* /./ */
4955 if ((NEXTCHR_IS_EOS) || nextchr == '\n')
4957 goto increment_locinput;
4961 #define ST st->u.trie
4962 case TRIEC: /* (ab|cd) with known charclass */
4963 /* In this case the charclass data is available inline so
4964 we can fail fast without a lot of extra overhead.
4966 if(!NEXTCHR_IS_EOS && !ANYOF_BITMAP_TEST(scan, nextchr)) {
4968 PerlIO_printf(Perl_debug_log,
4969 "%*s %sfailed to match trie start class...%s\n",
4970 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
4974 NOT_REACHED; /* NOTREACHED */
4977 case TRIE: /* (ab|cd) */
4978 /* the basic plan of execution of the trie is:
4979 * At the beginning, run though all the states, and
4980 * find the longest-matching word. Also remember the position
4981 * of the shortest matching word. For example, this pattern:
4984 * when matched against the string "abcde", will generate
4985 * accept states for all words except 3, with the longest
4986 * matching word being 4, and the shortest being 2 (with
4987 * the position being after char 1 of the string).
4989 * Then for each matching word, in word order (i.e. 1,2,4,5),
4990 * we run the remainder of the pattern; on each try setting
4991 * the current position to the character following the word,
4992 * returning to try the next word on failure.
4994 * We avoid having to build a list of words at runtime by
4995 * using a compile-time structure, wordinfo[].prev, which
4996 * gives, for each word, the previous accepting word (if any).
4997 * In the case above it would contain the mappings 1->2, 2->0,
4998 * 3->0, 4->5, 5->1. We can use this table to generate, from
4999 * the longest word (4 above), a list of all words, by
5000 * following the list of prev pointers; this gives us the
5001 * unordered list 4,5,1,2. Then given the current word we have
5002 * just tried, we can go through the list and find the
5003 * next-biggest word to try (so if we just failed on word 2,
5004 * the next in the list is 4).
5006 * Since at runtime we don't record the matching position in
5007 * the string for each word, we have to work that out for
5008 * each word we're about to process. The wordinfo table holds
5009 * the character length of each word; given that we recorded
5010 * at the start: the position of the shortest word and its
5011 * length in chars, we just need to move the pointer the
5012 * difference between the two char lengths. Depending on
5013 * Unicode status and folding, that's cheap or expensive.
5015 * This algorithm is optimised for the case where are only a
5016 * small number of accept states, i.e. 0,1, or maybe 2.
5017 * With lots of accepts states, and having to try all of them,
5018 * it becomes quadratic on number of accept states to find all
5023 /* what type of TRIE am I? (utf8 makes this contextual) */
5024 DECL_TRIE_TYPE(scan);
5026 /* what trie are we using right now */
5027 reg_trie_data * const trie
5028 = (reg_trie_data*)rexi->data->data[ ARG( scan ) ];
5029 HV * widecharmap = MUTABLE_HV(rexi->data->data[ ARG( scan ) + 1 ]);
5030 U32 state = trie->startstate;
5032 if (scan->flags == EXACTL || scan->flags == EXACTFLU8) {
5033 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
5035 && UTF8_IS_ABOVE_LATIN1(nextchr)
5036 && scan->flags == EXACTL)
5038 /* We only output for EXACTL, as we let the folder
5039 * output this message for EXACTFLU8 to avoid
5041 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(locinput,
5046 && (NEXTCHR_IS_EOS || !TRIE_BITMAP_TEST(trie, nextchr)))
5048 if (trie->states[ state ].wordnum) {
5050 PerlIO_printf(Perl_debug_log,
5051 "%*s %smatched empty string...%s\n",
5052 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
5058 PerlIO_printf(Perl_debug_log,
5059 "%*s %sfailed to match trie start class...%s\n",
5060 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
5067 U8 *uc = ( U8* )locinput;
5071 U8 *uscan = (U8*)NULL;
5072 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
5073 U32 charcount = 0; /* how many input chars we have matched */
5074 U32 accepted = 0; /* have we seen any accepting states? */
5076 ST.jump = trie->jump;
5079 ST.longfold = FALSE; /* char longer if folded => it's harder */
5082 /* fully traverse the TRIE; note the position of the
5083 shortest accept state and the wordnum of the longest
5086 while ( state && uc <= (U8*)(reginfo->strend) ) {
5087 U32 base = trie->states[ state ].trans.base;
5091 wordnum = trie->states[ state ].wordnum;
5093 if (wordnum) { /* it's an accept state */
5096 /* record first match position */
5098 ST.firstpos = (U8*)locinput;
5103 ST.firstchars = charcount;
5106 if (!ST.nextword || wordnum < ST.nextword)
5107 ST.nextword = wordnum;
5108 ST.topword = wordnum;
5111 DEBUG_TRIE_EXECUTE_r({
5112 DUMP_EXEC_POS( (char *)uc, scan, utf8_target );
5113 PerlIO_printf( Perl_debug_log,
5114 "%*s %sState: %4"UVxf" Accepted: %c ",
5115 2+depth * 2, "", PL_colors[4],
5116 (UV)state, (accepted ? 'Y' : 'N'));
5119 /* read a char and goto next state */
5120 if ( base && (foldlen || uc < (U8*)(reginfo->strend))) {
5122 REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc,
5123 uscan, len, uvc, charid, foldlen,
5130 base + charid - 1 - trie->uniquecharcount)) >= 0)
5132 && ((U32)offset < trie->lasttrans)
5133 && trie->trans[offset].check == state)
5135 state = trie->trans[offset].next;
5146 DEBUG_TRIE_EXECUTE_r(
5147 PerlIO_printf( Perl_debug_log,
5148 "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n",
5149 charid, uvc, (UV)state, PL_colors[5] );
5155 /* calculate total number of accept states */
5160 w = trie->wordinfo[w].prev;
5163 ST.accepted = accepted;
5167 PerlIO_printf( Perl_debug_log,
5168 "%*s %sgot %"IVdf" possible matches%s\n",
5169 REPORT_CODE_OFF + depth * 2, "",
5170 PL_colors[4], (IV)ST.accepted, PL_colors[5] );
5172 goto trie_first_try; /* jump into the fail handler */
5175 NOT_REACHED; /* NOTREACHED */
5177 case TRIE_next_fail: /* we failed - try next alternative */
5181 REGCP_UNWIND(ST.cp);
5182 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
5184 if (!--ST.accepted) {
5186 PerlIO_printf( Perl_debug_log,
5187 "%*s %sTRIE failed...%s\n",
5188 REPORT_CODE_OFF+depth*2, "",
5195 /* Find next-highest word to process. Note that this code
5196 * is O(N^2) per trie run (O(N) per branch), so keep tight */
5199 U16 const nextword = ST.nextword;
5200 reg_trie_wordinfo * const wordinfo
5201 = ((reg_trie_data*)rexi->data->data[ARG(ST.me)])->wordinfo;
5202 for (word=ST.topword; word; word=wordinfo[word].prev) {
5203 if (word > nextword && (!min || word < min))
5216 ST.lastparen = rex->lastparen;
5217 ST.lastcloseparen = rex->lastcloseparen;
5221 /* find start char of end of current word */
5223 U32 chars; /* how many chars to skip */
5224 reg_trie_data * const trie
5225 = (reg_trie_data*)rexi->data->data[ARG(ST.me)];
5227 assert((trie->wordinfo[ST.nextword].len - trie->prefixlen)
5229 chars = (trie->wordinfo[ST.nextword].len - trie->prefixlen)
5234 /* the hard option - fold each char in turn and find
5235 * its folded length (which may be different */
5236 U8 foldbuf[UTF8_MAXBYTES_CASE + 1];
5244 uvc = utf8n_to_uvchr((U8*)uc, UTF8_MAXLEN, &len,
5252 uvc = to_uni_fold(uvc, foldbuf, &foldlen);
5257 uvc = utf8n_to_uvchr(uscan, UTF8_MAXLEN, &len,
5273 scan = ST.me + ((ST.jump && ST.jump[ST.nextword])
5274 ? ST.jump[ST.nextword]
5278 PerlIO_printf( Perl_debug_log,
5279 "%*s %sTRIE matched word #%d, continuing%s\n",
5280 REPORT_CODE_OFF+depth*2, "",
5287 if (ST.accepted > 1 || has_cutgroup) {
5288 PUSH_STATE_GOTO(TRIE_next, scan, (char*)uc);
5290 NOT_REACHED; /* NOTREACHED */
5292 /* only one choice left - just continue */
5294 AV *const trie_words
5295 = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]);
5296 SV ** const tmp = trie_words
5297 ? av_fetch(trie_words, ST.nextword - 1, 0) : NULL;
5298 SV *sv= tmp ? sv_newmortal() : NULL;
5300 PerlIO_printf( Perl_debug_log,
5301 "%*s %sonly one match left, short-circuiting: #%d <%s>%s\n",
5302 REPORT_CODE_OFF+depth*2, "", PL_colors[4],
5304 tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
5305 PL_colors[0], PL_colors[1],
5306 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)|PERL_PV_ESCAPE_NONASCII
5308 : "not compiled under -Dr",
5312 locinput = (char*)uc;
5313 continue; /* execute rest of RE */
5318 case EXACTL: /* /abc/l */
5319 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
5321 /* Complete checking would involve going through every character
5322 * matched by the string to see if any is above latin1. But the
5323 * comparision otherwise might very well be a fast assembly
5324 * language routine, and I (khw) don't think slowing things down
5325 * just to check for this warning is worth it. So this just checks
5326 * the first character */
5327 if (utf8_target && UTF8_IS_ABOVE_LATIN1(*locinput)) {
5328 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(locinput, reginfo->strend);
5331 case EXACT: { /* /abc/ */
5332 char *s = STRING(scan);
5334 if (utf8_target != is_utf8_pat) {
5335 /* The target and the pattern have differing utf8ness. */
5337 const char * const e = s + ln;
5340 /* The target is utf8, the pattern is not utf8.
5341 * Above-Latin1 code points can't match the pattern;
5342 * invariants match exactly, and the other Latin1 ones need
5343 * to be downgraded to a single byte in order to do the
5344 * comparison. (If we could be confident that the target
5345 * is not malformed, this could be refactored to have fewer
5346 * tests by just assuming that if the first bytes match, it
5347 * is an invariant, but there are tests in the test suite
5348 * dealing with (??{...}) which violate this) */
5350 if (l >= reginfo->strend
5351 || UTF8_IS_ABOVE_LATIN1(* (U8*) l))
5355 if (UTF8_IS_INVARIANT(*(U8*)l)) {
5362 if (TWO_BYTE_UTF8_TO_NATIVE(*l, *(l+1)) != * (U8*) s)
5372 /* The target is not utf8, the pattern is utf8. */
5374 if (l >= reginfo->strend
5375 || UTF8_IS_ABOVE_LATIN1(* (U8*) s))
5379 if (UTF8_IS_INVARIANT(*(U8*)s)) {
5386 if (TWO_BYTE_UTF8_TO_NATIVE(*s, *(s+1)) != * (U8*) l)
5398 /* The target and the pattern have the same utf8ness. */
5399 /* Inline the first character, for speed. */
5400 if (reginfo->strend - locinput < ln
5401 || UCHARAT(s) != nextchr
5402 || (ln > 1 && memNE(s, locinput, ln)))
5411 case EXACTFL: { /* /abc/il */
5413 const U8 * fold_array;
5415 U32 fold_utf8_flags;
5417 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
5418 folder = foldEQ_locale;
5419 fold_array = PL_fold_locale;
5420 fold_utf8_flags = FOLDEQ_LOCALE;
5423 case EXACTFLU8: /* /abc/il; but all 'abc' are above 255, so
5424 is effectively /u; hence to match, target
5426 if (! utf8_target) {
5429 fold_utf8_flags = FOLDEQ_LOCALE | FOLDEQ_S1_ALREADY_FOLDED
5430 | FOLDEQ_S1_FOLDS_SANE;
5431 folder = foldEQ_latin1;
5432 fold_array = PL_fold_latin1;
5435 case EXACTFU_SS: /* /\x{df}/iu */
5436 case EXACTFU: /* /abc/iu */
5437 folder = foldEQ_latin1;
5438 fold_array = PL_fold_latin1;
5439 fold_utf8_flags = is_utf8_pat ? FOLDEQ_S1_ALREADY_FOLDED : 0;
5442 case EXACTFA_NO_TRIE: /* This node only generated for non-utf8
5444 assert(! is_utf8_pat);
5446 case EXACTFA: /* /abc/iaa */
5447 folder = foldEQ_latin1;
5448 fold_array = PL_fold_latin1;
5449 fold_utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
5452 case EXACTF: /* /abc/i This node only generated for
5453 non-utf8 patterns */
5454 assert(! is_utf8_pat);
5456 fold_array = PL_fold;
5457 fold_utf8_flags = 0;
5465 || state_num == EXACTFU_SS
5466 || (state_num == EXACTFL && IN_UTF8_CTYPE_LOCALE))
5468 /* Either target or the pattern are utf8, or has the issue where
5469 * the fold lengths may differ. */
5470 const char * const l = locinput;
5471 char *e = reginfo->strend;
5473 if (! foldEQ_utf8_flags(s, 0, ln, is_utf8_pat,
5474 l, &e, 0, utf8_target, fold_utf8_flags))
5482 /* Neither the target nor the pattern are utf8 */
5483 if (UCHARAT(s) != nextchr
5485 && UCHARAT(s) != fold_array[nextchr])
5489 if (reginfo->strend - locinput < ln)
5491 if (ln > 1 && ! folder(s, locinput, ln))
5497 case NBOUNDL: /* /\B/l */
5501 case BOUNDL: /* /\b/l */
5502 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
5504 if (FLAGS(scan) != TRADITIONAL_BOUND) {
5505 if (! IN_UTF8_CTYPE_LOCALE) {
5506 Perl_ck_warner(aTHX_ packWARN(WARN_LOCALE),
5507 B_ON_NON_UTF8_LOCALE_IS_WRONG);
5513 if (locinput == reginfo->strbeg)
5514 ln = isWORDCHAR_LC('\n');
5516 ln = isWORDCHAR_LC_utf8(reghop3((U8*)locinput, -1,
5517 (U8*)(reginfo->strbeg)));
5519 n = (NEXTCHR_IS_EOS)
5520 ? isWORDCHAR_LC('\n')
5521 : isWORDCHAR_LC_utf8((U8*)locinput);
5523 else { /* Here the string isn't utf8 */
5524 ln = (locinput == reginfo->strbeg)
5525 ? isWORDCHAR_LC('\n')
5526 : isWORDCHAR_LC(UCHARAT(locinput - 1));
5527 n = (NEXTCHR_IS_EOS)
5528 ? isWORDCHAR_LC('\n')
5529 : isWORDCHAR_LC(nextchr);
5531 if (to_complement ^ (ln == n)) {
5536 case NBOUND: /* /\B/ */
5540 case BOUND: /* /\b/ */
5544 goto bound_ascii_match_only;
5546 case NBOUNDA: /* /\B/a */
5550 case BOUNDA: /* /\b/a */
5552 bound_ascii_match_only:
5553 /* Here the string isn't utf8, or is utf8 and only ascii characters
5554 * are to match \w. In the latter case looking at the byte just
5555 * prior to the current one may be just the final byte of a
5556 * multi-byte character. This is ok. There are two cases:
5557 * 1) it is a single byte character, and then the test is doing
5558 * just what it's supposed to.
5559 * 2) it is a multi-byte character, in which case the final byte is
5560 * never mistakable for ASCII, and so the test will say it is
5561 * not a word character, which is the correct answer. */
5562 ln = (locinput == reginfo->strbeg)
5563 ? isWORDCHAR_A('\n')
5564 : isWORDCHAR_A(UCHARAT(locinput - 1));
5565 n = (NEXTCHR_IS_EOS)
5566 ? isWORDCHAR_A('\n')
5567 : isWORDCHAR_A(nextchr);
5568 if (to_complement ^ (ln == n)) {
5573 case NBOUNDU: /* /\B/u */
5577 case BOUNDU: /* /\b/u */
5583 switch((bound_type) FLAGS(scan)) {
5584 case TRADITIONAL_BOUND:
5585 ln = (locinput == reginfo->strbeg)
5586 ? isWORDCHAR_L1('\n')
5587 : isWORDCHAR_utf8(reghop3((U8*)locinput, -1,
5588 (U8*)(reginfo->strbeg)));
5589 n = (NEXTCHR_IS_EOS)
5590 ? isWORDCHAR_L1('\n')
5591 : isWORDCHAR_utf8((U8*)locinput);
5592 match = cBOOL(ln != n);
5595 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
5596 match = TRUE; /* GCB always matches at begin and
5600 /* Find the gcb values of previous and current
5601 * chars, then see if is a break point */
5602 match = isGCB(getGCB_VAL_UTF8(
5603 reghop3((U8*)locinput,
5605 (U8*)(reginfo->strbeg)),
5606 (U8*) reginfo->strend),
5607 getGCB_VAL_UTF8((U8*) locinput,
5608 (U8*) reginfo->strend));
5612 case SB_BOUND: /* Always matches at begin and end */
5613 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
5617 match = isSB(getSB_VAL_UTF8(
5618 reghop3((U8*)locinput,
5620 (U8*)(reginfo->strbeg)),
5621 (U8*) reginfo->strend),
5622 getSB_VAL_UTF8((U8*) locinput,
5623 (U8*) reginfo->strend),
5624 (U8*) reginfo->strbeg,
5626 (U8*) reginfo->strend,
5632 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
5636 match = isWB(PL_WB_UNKNOWN,
5638 reghop3((U8*)locinput,
5640 (U8*)(reginfo->strbeg)),
5641 (U8*) reginfo->strend),
5642 getWB_VAL_UTF8((U8*) locinput,
5643 (U8*) reginfo->strend),
5644 (U8*) reginfo->strbeg,
5646 (U8*) reginfo->strend,
5652 else { /* Not utf8 target */
5653 switch((bound_type) FLAGS(scan)) {
5654 case TRADITIONAL_BOUND:
5655 ln = (locinput == reginfo->strbeg)
5656 ? isWORDCHAR_L1('\n')
5657 : isWORDCHAR_L1(UCHARAT(locinput - 1));
5658 n = (NEXTCHR_IS_EOS)
5659 ? isWORDCHAR_L1('\n')
5660 : isWORDCHAR_L1(nextchr);
5661 match = cBOOL(ln != n);
5665 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
5666 match = TRUE; /* GCB always matches at begin and
5669 else { /* Only CR-LF combo isn't a GCB in 0-255
5671 match = UCHARAT(locinput - 1) != '\r'
5672 || UCHARAT(locinput) != '\n';
5676 case SB_BOUND: /* Always matches at begin and end */
5677 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
5681 match = isSB(getSB_VAL_CP(UCHARAT(locinput -1)),
5682 getSB_VAL_CP(UCHARAT(locinput)),
5683 (U8*) reginfo->strbeg,
5685 (U8*) reginfo->strend,
5691 if (locinput == reginfo->strbeg || NEXTCHR_IS_EOS) {
5695 match = isWB(PL_WB_UNKNOWN,
5696 getWB_VAL_CP(UCHARAT(locinput -1)),
5697 getWB_VAL_CP(UCHARAT(locinput)),
5698 (U8*) reginfo->strbeg,
5700 (U8*) reginfo->strend,
5707 if (to_complement ^ ! match) {
5712 case ANYOFL: /* /[abc]/l */
5713 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
5715 case ANYOF: /* /[abc]/ */
5719 if (!reginclass(rex, scan, (U8*)locinput, (U8*)reginfo->strend,
5722 locinput += UTF8SKIP(locinput);
5725 if (!REGINCLASS(rex, scan, (U8*)locinput))
5731 /* The argument (FLAGS) to all the POSIX node types is the class number
5734 case NPOSIXL: /* \W or [:^punct:] etc. under /l */
5738 case POSIXL: /* \w or [:punct:] etc. under /l */
5739 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
5743 /* Use isFOO_lc() for characters within Latin1. (Note that
5744 * UTF8_IS_INVARIANT works even on non-UTF-8 strings, or else
5745 * wouldn't be invariant) */
5746 if (UTF8_IS_INVARIANT(nextchr) || ! utf8_target) {
5747 if (! (to_complement ^ cBOOL(isFOO_lc(FLAGS(scan), (U8) nextchr)))) {
5751 else if (UTF8_IS_DOWNGRADEABLE_START(nextchr)) {
5752 if (! (to_complement ^ cBOOL(isFOO_lc(FLAGS(scan),
5753 (U8) TWO_BYTE_UTF8_TO_NATIVE(nextchr,
5754 *(locinput + 1))))))
5759 else { /* Here, must be an above Latin-1 code point */
5760 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(locinput, reginfo->strend);
5761 goto utf8_posix_above_latin1;
5764 /* Here, must be utf8 */
5765 locinput += UTF8SKIP(locinput);
5768 case NPOSIXD: /* \W or [:^punct:] etc. under /d */
5772 case POSIXD: /* \w or [:punct:] etc. under /d */
5778 case NPOSIXA: /* \W or [:^punct:] etc. under /a */
5780 if (NEXTCHR_IS_EOS) {
5784 /* All UTF-8 variants match */
5785 if (! UTF8_IS_INVARIANT(nextchr)) {
5786 goto increment_locinput;
5792 case POSIXA: /* \w or [:punct:] etc. under /a */
5795 /* We get here through POSIXD, NPOSIXD, and NPOSIXA when not in
5796 * UTF-8, and also from NPOSIXA even in UTF-8 when the current
5797 * character is a single byte */
5800 || ! (to_complement ^ cBOOL(_generic_isCC_A(nextchr,
5806 /* Here we are either not in utf8, or we matched a utf8-invariant,
5807 * so the next char is the next byte */
5811 case NPOSIXU: /* \W or [:^punct:] etc. under /u */
5815 case POSIXU: /* \w or [:punct:] etc. under /u */
5817 if (NEXTCHR_IS_EOS) {
5821 /* Use _generic_isCC() for characters within Latin1. (Note that
5822 * UTF8_IS_INVARIANT works even on non-UTF-8 strings, or else
5823 * wouldn't be invariant) */
5824 if (UTF8_IS_INVARIANT(nextchr) || ! utf8_target) {
5825 if (! (to_complement ^ cBOOL(_generic_isCC(nextchr,
5832 else if (UTF8_IS_DOWNGRADEABLE_START(nextchr)) {
5833 if (! (to_complement
5834 ^ cBOOL(_generic_isCC(TWO_BYTE_UTF8_TO_NATIVE(nextchr,
5842 else { /* Handle above Latin-1 code points */
5843 utf8_posix_above_latin1:
5844 classnum = (_char_class_number) FLAGS(scan);
5845 if (classnum < _FIRST_NON_SWASH_CC) {
5847 /* Here, uses a swash to find such code points. Load if if
5848 * not done already */
5849 if (! PL_utf8_swash_ptrs[classnum]) {
5850 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
5851 PL_utf8_swash_ptrs[classnum]
5852 = _core_swash_init("utf8",
5855 PL_XPosix_ptrs[classnum], &flags);
5857 if (! (to_complement
5858 ^ cBOOL(swash_fetch(PL_utf8_swash_ptrs[classnum],
5859 (U8 *) locinput, TRUE))))
5864 else { /* Here, uses macros to find above Latin-1 code points */
5866 case _CC_ENUM_SPACE:
5867 if (! (to_complement
5868 ^ cBOOL(is_XPERLSPACE_high(locinput))))
5873 case _CC_ENUM_BLANK:
5874 if (! (to_complement
5875 ^ cBOOL(is_HORIZWS_high(locinput))))
5880 case _CC_ENUM_XDIGIT:
5881 if (! (to_complement
5882 ^ cBOOL(is_XDIGIT_high(locinput))))
5887 case _CC_ENUM_VERTSPACE:
5888 if (! (to_complement
5889 ^ cBOOL(is_VERTWS_high(locinput))))
5894 default: /* The rest, e.g. [:cntrl:], can't match
5896 if (! to_complement) {
5902 locinput += UTF8SKIP(locinput);
5906 case CLUMP: /* Match \X: logical Unicode character. This is defined as
5907 a Unicode extended Grapheme Cluster */
5910 if (! utf8_target) {
5912 /* Match either CR LF or '.', as all the other possibilities
5914 locinput++; /* Match the . or CR */
5915 if (nextchr == '\r' /* And if it was CR, and the next is LF,
5917 && locinput < reginfo->strend
5918 && UCHARAT(locinput) == '\n')
5925 /* Get the gcb type for the current character */
5926 PL_GCB_enum prev_gcb = getGCB_VAL_UTF8((U8*) locinput,
5927 (U8*) reginfo->strend);
5929 /* Then scan through the input until we get to the first
5930 * character whose type is supposed to be a gcb with the
5931 * current character. (There is always a break at the
5933 locinput += UTF8SKIP(locinput);
5934 while (locinput < reginfo->strend) {
5935 PL_GCB_enum cur_gcb = getGCB_VAL_UTF8((U8*) locinput,
5936 (U8*) reginfo->strend);
5937 if (isGCB(prev_gcb, cur_gcb)) {
5942 locinput += UTF8SKIP(locinput);
5949 case NREFFL: /* /\g{name}/il */
5950 { /* The capture buffer cases. The ones beginning with N for the
5951 named buffers just convert to the equivalent numbered and
5952 pretend they were called as the corresponding numbered buffer
5954 /* don't initialize these in the declaration, it makes C++
5959 const U8 *fold_array;
5962 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
5963 folder = foldEQ_locale;
5964 fold_array = PL_fold_locale;
5966 utf8_fold_flags = FOLDEQ_LOCALE;
5969 case NREFFA: /* /\g{name}/iaa */
5970 folder = foldEQ_latin1;
5971 fold_array = PL_fold_latin1;
5973 utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
5976 case NREFFU: /* /\g{name}/iu */
5977 folder = foldEQ_latin1;
5978 fold_array = PL_fold_latin1;
5980 utf8_fold_flags = 0;
5983 case NREFF: /* /\g{name}/i */
5985 fold_array = PL_fold;
5987 utf8_fold_flags = 0;
5990 case NREF: /* /\g{name}/ */
5994 utf8_fold_flags = 0;
5997 /* For the named back references, find the corresponding buffer
5999 n = reg_check_named_buff_matched(rex,scan);
6004 goto do_nref_ref_common;
6006 case REFFL: /* /\1/il */
6007 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
6008 folder = foldEQ_locale;
6009 fold_array = PL_fold_locale;
6010 utf8_fold_flags = FOLDEQ_LOCALE;
6013 case REFFA: /* /\1/iaa */
6014 folder = foldEQ_latin1;
6015 fold_array = PL_fold_latin1;
6016 utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
6019 case REFFU: /* /\1/iu */
6020 folder = foldEQ_latin1;
6021 fold_array = PL_fold_latin1;
6022 utf8_fold_flags = 0;
6025 case REFF: /* /\1/i */
6027 fold_array = PL_fold;
6028 utf8_fold_flags = 0;
6031 case REF: /* /\1/ */
6034 utf8_fold_flags = 0;
6038 n = ARG(scan); /* which paren pair */
6041 ln = rex->offs[n].start;
6042 reginfo->poscache_iter = reginfo->poscache_maxiter; /* Void cache */
6043 if (rex->lastparen < n || ln == -1)
6044 sayNO; /* Do not match unless seen CLOSEn. */
6045 if (ln == rex->offs[n].end)
6048 s = reginfo->strbeg + ln;
6049 if (type != REF /* REF can do byte comparison */
6050 && (utf8_target || type == REFFU || type == REFFL))
6052 char * limit = reginfo->strend;
6054 /* This call case insensitively compares the entire buffer
6055 * at s, with the current input starting at locinput, but
6056 * not going off the end given by reginfo->strend, and
6057 * returns in <limit> upon success, how much of the
6058 * current input was matched */
6059 if (! foldEQ_utf8_flags(s, NULL, rex->offs[n].end - ln, utf8_target,
6060 locinput, &limit, 0, utf8_target, utf8_fold_flags))
6068 /* Not utf8: Inline the first character, for speed. */
6069 if (!NEXTCHR_IS_EOS &&
6070 UCHARAT(s) != nextchr &&
6072 UCHARAT(s) != fold_array[nextchr]))
6074 ln = rex->offs[n].end - ln;
6075 if (locinput + ln > reginfo->strend)
6077 if (ln > 1 && (type == REF
6078 ? memNE(s, locinput, ln)
6079 : ! folder(s, locinput, ln)))
6085 case NOTHING: /* null op; e.g. the 'nothing' following
6086 * the '*' in m{(a+|b)*}' */
6088 case TAIL: /* placeholder while compiling (A|B|C) */
6092 #define ST st->u.eval
6097 regexp_internal *rei;
6098 regnode *startpoint;
6100 case GOSTART: /* (?R) */
6101 case GOSUB: /* /(...(?1))/ /(...(?&foo))/ */
6102 if (cur_eval && cur_eval->locinput==locinput) {
6103 if (cur_eval->u.eval.close_paren == (U32)ARG(scan))
6104 Perl_croak(aTHX_ "Infinite recursion in regex");
6105 if ( ++nochange_depth > max_nochange_depth )
6107 "Pattern subroutine nesting without pos change"
6108 " exceeded limit in regex");
6115 if (OP(scan)==GOSUB) {
6116 startpoint = scan + ARG2L(scan);
6117 ST.close_paren = ARG(scan);
6119 startpoint = rei->program+1;
6123 /* Save all the positions seen so far. */
6124 ST.cp = regcppush(rex, 0, maxopenparen);
6125 REGCP_SET(ST.lastcp);
6127 /* and then jump to the code we share with EVAL */
6128 goto eval_recurse_doit;
6131 case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */
6132 if (cur_eval && cur_eval->locinput==locinput) {
6133 if ( ++nochange_depth > max_nochange_depth )
6134 Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex");
6139 /* execute the code in the {...} */
6143 OP * const oop = PL_op;
6144 COP * const ocurcop = PL_curcop;
6148 /* save *all* paren positions */
6149 regcppush(rex, 0, maxopenparen);
6150 REGCP_SET(runops_cp);
6153 caller_cv = find_runcv(NULL);
6157 if (rexi->data->what[n] == 'r') { /* code from an external qr */
6159 (REGEXP*)(rexi->data->data[n])
6162 nop = (OP*)rexi->data->data[n+1];
6164 else if (rexi->data->what[n] == 'l') { /* literal code */
6166 nop = (OP*)rexi->data->data[n];
6167 assert(CvDEPTH(newcv));
6170 /* literal with own CV */
6171 assert(rexi->data->what[n] == 'L');
6172 newcv = rex->qr_anoncv;
6173 nop = (OP*)rexi->data->data[n];
6176 /* normally if we're about to execute code from the same
6177 * CV that we used previously, we just use the existing
6178 * CX stack entry. However, its possible that in the
6179 * meantime we may have backtracked, popped from the save
6180 * stack, and undone the SAVECOMPPAD(s) associated with
6181 * PUSH_MULTICALL; in which case PL_comppad no longer
6182 * points to newcv's pad. */
6183 if (newcv != last_pushed_cv || PL_comppad != last_pad)
6185 U8 flags = (CXp_SUB_RE |
6186 ((newcv == caller_cv) ? CXp_SUB_RE_FAKE : 0));
6187 if (last_pushed_cv) {
6188 CHANGE_MULTICALL_FLAGS(newcv, flags);
6191 PUSH_MULTICALL_FLAGS(newcv, flags);
6193 last_pushed_cv = newcv;
6196 /* these assignments are just to silence compiler
6198 multicall_cop = NULL;
6201 last_pad = PL_comppad;
6203 /* the initial nextstate you would normally execute
6204 * at the start of an eval (which would cause error
6205 * messages to come from the eval), may be optimised
6206 * away from the execution path in the regex code blocks;
6207 * so manually set PL_curcop to it initially */
6209 OP *o = cUNOPx(nop)->op_first;
6210 assert(o->op_type == OP_NULL);
6211 if (o->op_targ == OP_SCOPE) {
6212 o = cUNOPo->op_first;
6215 assert(o->op_targ == OP_LEAVE);
6216 o = cUNOPo->op_first;
6217 assert(o->op_type == OP_ENTER);
6221 if (o->op_type != OP_STUB) {
6222 assert( o->op_type == OP_NEXTSTATE
6223 || o->op_type == OP_DBSTATE
6224 || (o->op_type == OP_NULL
6225 && ( o->op_targ == OP_NEXTSTATE
6226 || o->op_targ == OP_DBSTATE
6230 PL_curcop = (COP*)o;
6235 DEBUG_STATE_r( PerlIO_printf(Perl_debug_log,
6236 " re EVAL PL_op=0x%"UVxf"\n", PTR2UV(nop)) );
6238 rex->offs[0].end = locinput - reginfo->strbeg;
6239 if (reginfo->info_aux_eval->pos_magic)
6240 MgBYTEPOS_set(reginfo->info_aux_eval->pos_magic,
6241 reginfo->sv, reginfo->strbeg,
6242 locinput - reginfo->strbeg);
6245 SV *sv_mrk = get_sv("REGMARK", 1);
6246 sv_setsv(sv_mrk, sv_yes_mark);
6249 /* we don't use MULTICALL here as we want to call the
6250 * first op of the block of interest, rather than the
6251 * first op of the sub */
6252 before = (IV)(SP-PL_stack_base);
6254 CALLRUNOPS(aTHX); /* Scalar context. */
6256 if ((IV)(SP-PL_stack_base) == before)
6257 ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */
6263 /* before restoring everything, evaluate the returned
6264 * value, so that 'uninit' warnings don't use the wrong
6265 * PL_op or pad. Also need to process any magic vars
6266 * (e.g. $1) *before* parentheses are restored */
6271 if (logical == 0) /* (?{})/ */
6272 sv_setsv(save_scalar(PL_replgv), ret); /* $^R */
6273 else if (logical == 1) { /* /(?(?{...})X|Y)/ */
6274 sw = cBOOL(SvTRUE(ret));
6277 else { /* /(??{}) */
6278 /* if its overloaded, let the regex compiler handle
6279 * it; otherwise extract regex, or stringify */
6280 if (SvGMAGICAL(ret))
6281 ret = sv_mortalcopy(ret);
6282 if (!SvAMAGIC(ret)) {
6286 if (SvTYPE(sv) == SVt_REGEXP)
6287 re_sv = (REGEXP*) sv;
6288 else if (SvSMAGICAL(ret)) {
6289 MAGIC *mg = mg_find(ret, PERL_MAGIC_qr);
6291 re_sv = (REGEXP *) mg->mg_obj;
6294 /* force any undef warnings here */
6295 if (!re_sv && !SvPOK(ret) && !SvNIOK(ret)) {
6296 ret = sv_mortalcopy(ret);
6297 (void) SvPV_force_nolen(ret);
6303 /* *** Note that at this point we don't restore
6304 * PL_comppad, (or pop the CxSUB) on the assumption it may
6305 * be used again soon. This is safe as long as nothing
6306 * in the regexp code uses the pad ! */
6308 PL_curcop = ocurcop;
6309 S_regcp_restore(aTHX_ rex, runops_cp, &maxopenparen);
6310 PL_curpm = PL_reg_curpm;
6316 /* only /(??{})/ from now on */
6319 /* extract RE object from returned value; compiling if
6323 re_sv = reg_temp_copy(NULL, re_sv);
6328 if (SvUTF8(ret) && IN_BYTES) {
6329 /* In use 'bytes': make a copy of the octet
6330 * sequence, but without the flag on */
6332 const char *const p = SvPV(ret, len);
6333 ret = newSVpvn_flags(p, len, SVs_TEMP);
6335 if (rex->intflags & PREGf_USE_RE_EVAL)
6336 pm_flags |= PMf_USE_RE_EVAL;
6338 /* if we got here, it should be an engine which
6339 * supports compiling code blocks and stuff */
6340 assert(rex->engine && rex->engine->op_comp);
6341 assert(!(scan->flags & ~RXf_PMf_COMPILETIME));
6342 re_sv = rex->engine->op_comp(aTHX_ &ret, 1, NULL,
6343 rex->engine, NULL, NULL,
6344 /* copy /msixn etc to inner pattern */
6349 & (SVs_TEMP | SVs_GMG | SVf_ROK))
6350 && (!SvPADTMP(ret) || SvREADONLY(ret))) {
6351 /* This isn't a first class regexp. Instead, it's
6352 caching a regexp onto an existing, Perl visible
6354 sv_magic(ret, MUTABLE_SV(re_sv), PERL_MAGIC_qr, 0, 0);
6360 RXp_MATCH_COPIED_off(re);
6361 re->subbeg = rex->subbeg;
6362 re->sublen = rex->sublen;
6363 re->suboffset = rex->suboffset;
6364 re->subcoffset = rex->subcoffset;
6366 re->lastcloseparen = 0;
6369 debug_start_match(re_sv, utf8_target, locinput,
6370 reginfo->strend, "Matching embedded");
6372 startpoint = rei->program + 1;
6373 ST.close_paren = 0; /* only used for GOSUB */
6374 /* Save all the seen positions so far. */
6375 ST.cp = regcppush(rex, 0, maxopenparen);
6376 REGCP_SET(ST.lastcp);
6377 /* and set maxopenparen to 0, since we are starting a "fresh" match */
6379 /* run the pattern returned from (??{...}) */
6381 eval_recurse_doit: /* Share code with GOSUB below this line
6382 * At this point we expect the stack context to be
6383 * set up correctly */
6385 /* invalidate the S-L poscache. We're now executing a
6386 * different set of WHILEM ops (and their associated
6387 * indexes) against the same string, so the bits in the
6388 * cache are meaningless. Setting maxiter to zero forces
6389 * the cache to be invalidated and zeroed before reuse.
6390 * XXX This is too dramatic a measure. Ideally we should
6391 * save the old cache and restore when running the outer
6393 reginfo->poscache_maxiter = 0;
6395 /* the new regexp might have a different is_utf8_pat than we do */
6396 is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(re_sv));
6398 ST.prev_rex = rex_sv;
6399 ST.prev_curlyx = cur_curlyx;
6401 SET_reg_curpm(rex_sv);
6406 ST.prev_eval = cur_eval;
6408 /* now continue from first node in postoned RE */
6409 PUSH_YES_STATE_GOTO(EVAL_AB, startpoint, locinput);
6411 NOT_REACHED; /* NOTREACHED */
6414 case EVAL_AB: /* cleanup after a successful (??{A})B */
6415 /* note: this is called twice; first after popping B, then A */
6416 rex_sv = ST.prev_rex;
6417 is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(rex_sv));
6418 SET_reg_curpm(rex_sv);
6419 rex = ReANY(rex_sv);
6420 rexi = RXi_GET(rex);
6422 /* preserve $^R across LEAVE's. See Bug 121070. */
6423 SV *save_sv= GvSV(PL_replgv);
6424 SvREFCNT_inc(save_sv);
6425 regcpblow(ST.cp); /* LEAVE in disguise */
6426 sv_setsv(GvSV(PL_replgv), save_sv);
6427 SvREFCNT_dec(save_sv);
6429 cur_eval = ST.prev_eval;
6430 cur_curlyx = ST.prev_curlyx;
6432 /* Invalidate cache. See "invalidate" comment above. */
6433 reginfo->poscache_maxiter = 0;
6434 if ( nochange_depth )
6439 case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */
6440 /* note: this is called twice; first after popping B, then A */
6441 rex_sv = ST.prev_rex;
6442 is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(rex_sv));
6443 SET_reg_curpm(rex_sv);
6444 rex = ReANY(rex_sv);
6445 rexi = RXi_GET(rex);
6447 REGCP_UNWIND(ST.lastcp);
6448 regcppop(rex, &maxopenparen);
6449 cur_eval = ST.prev_eval;
6450 cur_curlyx = ST.prev_curlyx;
6451 /* Invalidate cache. See "invalidate" comment above. */
6452 reginfo->poscache_maxiter = 0;
6453 if ( nochange_depth )
6459 n = ARG(scan); /* which paren pair */
6460 rex->offs[n].start_tmp = locinput - reginfo->strbeg;
6461 if (n > maxopenparen)
6463 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log,
6464 "rex=0x%"UVxf" offs=0x%"UVxf": \\%"UVuf": set %"IVdf" tmp; maxopenparen=%"UVuf"\n",
6468 (IV)rex->offs[n].start_tmp,
6474 /* XXX really need to log other places start/end are set too */
6475 #define CLOSE_CAPTURE \
6476 rex->offs[n].start = rex->offs[n].start_tmp; \
6477 rex->offs[n].end = locinput - reginfo->strbeg; \
6478 DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log, \
6479 "rex=0x%"UVxf" offs=0x%"UVxf": \\%"UVuf": set %"IVdf"..%"IVdf"\n", \
6481 PTR2UV(rex->offs), \
6483 (IV)rex->offs[n].start, \
6484 (IV)rex->offs[n].end \
6488 n = ARG(scan); /* which paren pair */
6490 if (n > rex->lastparen)
6492 rex->lastcloseparen = n;
6493 if (cur_eval && cur_eval->u.eval.close_paren == n) {
6498 case ACCEPT: /* (*ACCEPT) */
6502 cursor && OP(cursor)!=END;
6503 cursor=regnext(cursor))
6505 if ( OP(cursor)==CLOSE ){
6507 if ( n <= lastopen ) {
6509 if (n > rex->lastparen)
6511 rex->lastcloseparen = n;
6512 if ( n == ARG(scan) || (cur_eval &&
6513 cur_eval->u.eval.close_paren == n))
6522 case GROUPP: /* (?(1)) */
6523 n = ARG(scan); /* which paren pair */
6524 sw = cBOOL(rex->lastparen >= n && rex->offs[n].end != -1);
6527 case NGROUPP: /* (?(<name>)) */
6528 /* reg_check_named_buff_matched returns 0 for no match */
6529 sw = cBOOL(0 < reg_check_named_buff_matched(rex,scan));
6532 case INSUBP: /* (?(R)) */
6534 sw = (cur_eval && (!n || cur_eval->u.eval.close_paren == n));
6537 case DEFINEP: /* (?(DEFINE)) */
6541 case IFTHEN: /* (?(cond)A|B) */
6542 reginfo->poscache_iter = reginfo->poscache_maxiter; /* Void cache */
6544 next = NEXTOPER(NEXTOPER(scan));
6546 next = scan + ARG(scan);
6547 if (OP(next) == IFTHEN) /* Fake one. */
6548 next = NEXTOPER(NEXTOPER(next));
6552 case LOGICAL: /* modifier for EVAL and IFMATCH */
6553 logical = scan->flags;
6556 /*******************************************************************
6558 The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/
6559 pattern, where A and B are subpatterns. (For simple A, CURLYM or
6560 STAR/PLUS/CURLY/CURLYN are used instead.)
6562 A*B is compiled as <CURLYX><A><WHILEM><B>
6564 On entry to the subpattern, CURLYX is called. This pushes a CURLYX
6565 state, which contains the current count, initialised to -1. It also sets
6566 cur_curlyx to point to this state, with any previous value saved in the
6569 CURLYX then jumps straight to the WHILEM op, rather than executing A,
6570 since the pattern may possibly match zero times (i.e. it's a while {} loop
6571 rather than a do {} while loop).
6573 Each entry to WHILEM represents a successful match of A. The count in the
6574 CURLYX block is incremented, another WHILEM state is pushed, and execution
6575 passes to A or B depending on greediness and the current count.
6577 For example, if matching against the string a1a2a3b (where the aN are
6578 substrings that match /A/), then the match progresses as follows: (the
6579 pushed states are interspersed with the bits of strings matched so far):
6582 <CURLYX cnt=0><WHILEM>
6583 <CURLYX cnt=1><WHILEM> a1 <WHILEM>
6584 <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM>
6585 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM>
6586 <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b
6588 (Contrast this with something like CURLYM, which maintains only a single
6592 a1 <CURLYM cnt=1> a2
6593 a1 a2 <CURLYM cnt=2> a3
6594 a1 a2 a3 <CURLYM cnt=3> b
6597 Each WHILEM state block marks a point to backtrack to upon partial failure
6598 of A or B, and also contains some minor state data related to that
6599 iteration. The CURLYX block, pointed to by cur_curlyx, contains the
6600 overall state, such as the count, and pointers to the A and B ops.
6602 This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx
6603 must always point to the *current* CURLYX block, the rules are:
6605 When executing CURLYX, save the old cur_curlyx in the CURLYX state block,
6606 and set cur_curlyx to point the new block.
6608 When popping the CURLYX block after a successful or unsuccessful match,
6609 restore the previous cur_curlyx.
6611 When WHILEM is about to execute B, save the current cur_curlyx, and set it
6612 to the outer one saved in the CURLYX block.
6614 When popping the WHILEM block after a successful or unsuccessful B match,
6615 restore the previous cur_curlyx.
6617 Here's an example for the pattern (AI* BI)*BO
6618 I and O refer to inner and outer, C and W refer to CURLYX and WHILEM:
6621 curlyx backtrack stack
6622 ------ ---------------
6624 CO <CO prev=NULL> <WO>
6625 CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
6626 CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
6627 NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo
6629 At this point the pattern succeeds, and we work back down the stack to
6630 clean up, restoring as we go:
6632 CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi
6633 CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai
6634 CO <CO prev=NULL> <WO>
6637 *******************************************************************/
6639 #define ST st->u.curlyx
6641 case CURLYX: /* start of /A*B/ (for complex A) */
6643 /* No need to save/restore up to this paren */
6644 I32 parenfloor = scan->flags;
6646 assert(next); /* keep Coverity happy */
6647 if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
6650 /* XXXX Probably it is better to teach regpush to support
6651 parenfloor > maxopenparen ... */
6652 if (parenfloor > (I32)rex->lastparen)
6653 parenfloor = rex->lastparen; /* Pessimization... */
6655 ST.prev_curlyx= cur_curlyx;
6657 ST.cp = PL_savestack_ix;
6659 /* these fields contain the state of the current curly.
6660 * they are accessed by subsequent WHILEMs */
6661 ST.parenfloor = parenfloor;
6666 ST.count = -1; /* this will be updated by WHILEM */
6667 ST.lastloc = NULL; /* this will be updated by WHILEM */
6669 PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next), locinput);
6671 NOT_REACHED; /* NOTREACHED */
6674 case CURLYX_end: /* just finished matching all of A*B */
6675 cur_curlyx = ST.prev_curlyx;
6678 NOT_REACHED; /* NOTREACHED */
6680 case CURLYX_end_fail: /* just failed to match all of A*B */
6682 cur_curlyx = ST.prev_curlyx;
6685 NOT_REACHED; /* NOTREACHED */
6689 #define ST st->u.whilem
6691 case WHILEM: /* just matched an A in /A*B/ (for complex A) */
6693 /* see the discussion above about CURLYX/WHILEM */
6698 assert(cur_curlyx); /* keep Coverity happy */
6700 min = ARG1(cur_curlyx->u.curlyx.me);
6701 max = ARG2(cur_curlyx->u.curlyx.me);
6702 A = NEXTOPER(cur_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS;
6703 n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */
6704 ST.save_lastloc = cur_curlyx->u.curlyx.lastloc;
6705 ST.cache_offset = 0;
6709 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
6710 "%*s whilem: matched %ld out of %d..%d\n",
6711 REPORT_CODE_OFF+depth*2, "", (long)n, min, max)
6714 /* First just match a string of min A's. */
6717 ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor,
6719 cur_curlyx->u.curlyx.lastloc = locinput;
6720 REGCP_SET(ST.lastcp);
6722 PUSH_STATE_GOTO(WHILEM_A_pre, A, locinput);
6724 NOT_REACHED; /* NOTREACHED */
6727 /* If degenerate A matches "", assume A done. */
6729 if (locinput == cur_curlyx->u.curlyx.lastloc) {
6730 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
6731 "%*s whilem: empty match detected, trying continuation...\n",
6732 REPORT_CODE_OFF+depth*2, "")
6734 goto do_whilem_B_max;
6737 /* super-linear cache processing.
6739 * The idea here is that for certain types of CURLYX/WHILEM -
6740 * principally those whose upper bound is infinity (and
6741 * excluding regexes that have things like \1 and other very
6742 * non-regular expresssiony things), then if a pattern like
6743 * /....A*.../ fails and we backtrack to the WHILEM, then we
6744 * make a note that this particular WHILEM op was at string
6745 * position 47 (say) when the rest of pattern failed. Then, if
6746 * we ever find ourselves back at that WHILEM, and at string
6747 * position 47 again, we can just fail immediately rather than
6748 * running the rest of the pattern again.
6750 * This is very handy when patterns start to go
6751 * 'super-linear', like in (a+)*(a+)*(a+)*, where you end up
6752 * with a combinatorial explosion of backtracking.
6754 * The cache is implemented as a bit array, with one bit per
6755 * string byte position per WHILEM op (up to 16) - so its
6756 * between 0.25 and 2x the string size.
6758 * To avoid allocating a poscache buffer every time, we do an
6759 * initially countdown; only after we have executed a WHILEM
6760 * op (string-length x #WHILEMs) times do we allocate the
6763 * The top 4 bits of scan->flags byte say how many different
6764 * relevant CURLLYX/WHILEM op pairs there are, while the
6765 * bottom 4-bits is the identifying index number of this
6771 if (!reginfo->poscache_maxiter) {
6772 /* start the countdown: Postpone detection until we
6773 * know the match is not *that* much linear. */
6774 reginfo->poscache_maxiter
6775 = (reginfo->strend - reginfo->strbeg + 1)
6777 /* possible overflow for long strings and many CURLYX's */
6778 if (reginfo->poscache_maxiter < 0)
6779 reginfo->poscache_maxiter = I32_MAX;
6780 reginfo->poscache_iter = reginfo->poscache_maxiter;
6783 if (reginfo->poscache_iter-- == 0) {
6784 /* initialise cache */
6785 const SSize_t size = (reginfo->poscache_maxiter + 7)/8;
6786 regmatch_info_aux *const aux = reginfo->info_aux;
6787 if (aux->poscache) {
6788 if ((SSize_t)reginfo->poscache_size < size) {
6789 Renew(aux->poscache, size, char);
6790 reginfo->poscache_size = size;
6792 Zero(aux->poscache, size, char);
6795 reginfo->poscache_size = size;
6796 Newxz(aux->poscache, size, char);
6798 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
6799 "%swhilem: Detected a super-linear match, switching on caching%s...\n",
6800 PL_colors[4], PL_colors[5])
6804 if (reginfo->poscache_iter < 0) {
6805 /* have we already failed at this position? */
6806 SSize_t offset, mask;
6808 reginfo->poscache_iter = -1; /* stop eventual underflow */
6809 offset = (scan->flags & 0xf) - 1
6810 + (locinput - reginfo->strbeg)
6812 mask = 1 << (offset % 8);
6814 if (reginfo->info_aux->poscache[offset] & mask) {
6815 DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
6816 "%*s whilem: (cache) already tried at this position...\n",
6817 REPORT_CODE_OFF+depth*2, "")
6819 sayNO; /* cache records failure */
6821 ST.cache_offset = offset;
6822 ST.cache_mask = mask;
6826 /* Prefer B over A for minimal matching. */
6828 if (cur_curlyx->u.curlyx.minmod) {
6829 ST.save_curlyx = cur_curlyx;
6830 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
6831 ST.cp = regcppush(rex, ST.save_curlyx->u.curlyx.parenfloor,
6833 REGCP_SET(ST.lastcp);
6834 PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B,
6837 NOT_REACHED; /* NOTREACHED */
6840 /* Prefer A over B for maximal matching. */
6842 if (n < max) { /* More greed allowed? */
6843 ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor,
6845 cur_curlyx->u.curlyx.lastloc = locinput;
6846 REGCP_SET(ST.lastcp);
6847 PUSH_STATE_GOTO(WHILEM_A_max, A, locinput);
6849 NOT_REACHED; /* NOTREACHED */
6851 goto do_whilem_B_max;
6854 NOT_REACHED; /* NOTREACHED */
6856 case WHILEM_B_min: /* just matched B in a minimal match */
6857 case WHILEM_B_max: /* just matched B in a maximal match */
6858 cur_curlyx = ST.save_curlyx;
6861 NOT_REACHED; /* NOTREACHED */
6863 case WHILEM_B_max_fail: /* just failed to match B in a maximal match */
6864 cur_curlyx = ST.save_curlyx;
6865 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
6866 cur_curlyx->u.curlyx.count--;
6869 NOT_REACHED; /* NOTREACHED */
6871 case WHILEM_A_min_fail: /* just failed to match A in a minimal match */
6873 case WHILEM_A_pre_fail: /* just failed to match even minimal A */
6874 REGCP_UNWIND(ST.lastcp);
6875 regcppop(rex, &maxopenparen);
6876 cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
6877 cur_curlyx->u.curlyx.count--;
6880 NOT_REACHED; /* NOTREACHED */
6882 case WHILEM_A_max_fail: /* just failed to match A in a maximal match */
6883 REGCP_UNWIND(ST.lastcp);
6884 regcppop(rex, &maxopenparen); /* Restore some previous $<digit>s? */
6885 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
6886 "%*s whilem: failed, trying continuation...\n",
6887 REPORT_CODE_OFF+depth*2, "")
6890 if (cur_curlyx->u.curlyx.count >= REG_INFTY
6891 && ckWARN(WARN_REGEXP)
6892 && !reginfo->warned)
6894 reginfo->warned = TRUE;
6895 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
6896 "Complex regular subexpression recursion limit (%d) "
6902 ST.save_curlyx = cur_curlyx;
6903 cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
6904 PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B,
6907 NOT_REACHED; /* NOTREACHED */
6909 case WHILEM_B_min_fail: /* just failed to match B in a minimal match */
6910 cur_curlyx = ST.save_curlyx;
6911 REGCP_UNWIND(ST.lastcp);
6912 regcppop(rex, &maxopenparen);
6914 if (cur_curlyx->u.curlyx.count >= /*max*/ARG2(cur_curlyx->u.curlyx.me)) {
6915 /* Maximum greed exceeded */
6916 if (cur_curlyx->u.curlyx.count >= REG_INFTY
6917 && ckWARN(WARN_REGEXP)
6918 && !reginfo->warned)
6920 reginfo->warned = TRUE;
6921 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
6922 "Complex regular subexpression recursion "
6923 "limit (%d) exceeded",
6926 cur_curlyx->u.curlyx.count--;
6930 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
6931 "%*s trying longer...\n", REPORT_CODE_OFF+depth*2, "")
6933 /* Try grabbing another A and see if it helps. */
6934 cur_curlyx->u.curlyx.lastloc = locinput;
6935 ST.cp = regcppush(rex, cur_curlyx->u.curlyx.parenfloor,
6937 REGCP_SET(ST.lastcp);
6938 PUSH_STATE_GOTO(WHILEM_A_min,
6939 /*A*/ NEXTOPER(ST.save_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS,
6942 NOT_REACHED; /* NOTREACHED */
6945 #define ST st->u.branch
6947 case BRANCHJ: /* /(...|A|...)/ with long next pointer */
6948 next = scan + ARG(scan);
6951 scan = NEXTOPER(scan);
6954 case BRANCH: /* /(...|A|...)/ */
6955 scan = NEXTOPER(scan); /* scan now points to inner node */
6956 ST.lastparen = rex->lastparen;
6957 ST.lastcloseparen = rex->lastcloseparen;
6958 ST.next_branch = next;
6961 /* Now go into the branch */
6963 PUSH_YES_STATE_GOTO(BRANCH_next, scan, locinput);
6965 PUSH_STATE_GOTO(BRANCH_next, scan, locinput);
6968 NOT_REACHED; /* NOTREACHED */
6970 case CUTGROUP: /* /(*THEN)/ */
6971 sv_yes_mark = st->u.mark.mark_name = scan->flags ? NULL :
6972 MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
6973 PUSH_STATE_GOTO(CUTGROUP_next, next, locinput);
6975 NOT_REACHED; /* NOTREACHED */
6977 case CUTGROUP_next_fail:
6980 if (st->u.mark.mark_name)
6981 sv_commit = st->u.mark.mark_name;
6984 NOT_REACHED; /* NOTREACHED */
6989 NOT_REACHED; /* NOTREACHED */
6991 case BRANCH_next_fail: /* that branch failed; try the next, if any */
6996 REGCP_UNWIND(ST.cp);
6997 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
6998 scan = ST.next_branch;
6999 /* no more branches? */
7000 if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) {
7002 PerlIO_printf( Perl_debug_log,
7003 "%*s %sBRANCH failed...%s\n",
7004 REPORT_CODE_OFF+depth*2, "",
7010 continue; /* execute next BRANCH[J] op */
7013 case MINMOD: /* next op will be non-greedy, e.g. A*? */
7018 #define ST st->u.curlym
7020 case CURLYM: /* /A{m,n}B/ where A is fixed-length */
7022 /* This is an optimisation of CURLYX that enables us to push
7023 * only a single backtracking state, no matter how many matches
7024 * there are in {m,n}. It relies on the pattern being constant
7025 * length, with no parens to influence future backrefs
7029 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
7031 ST.lastparen = rex->lastparen;
7032 ST.lastcloseparen = rex->lastcloseparen;
7034 /* if paren positive, emulate an OPEN/CLOSE around A */
7036 U32 paren = ST.me->flags;
7037 if (paren > maxopenparen)
7038 maxopenparen = paren;
7039 scan += NEXT_OFF(scan); /* Skip former OPEN. */
7047 ST.c1 = CHRTEST_UNINIT;
7050 if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */
7053 curlym_do_A: /* execute the A in /A{m,n}B/ */
7054 PUSH_YES_STATE_GOTO(CURLYM_A, ST.A, locinput); /* match A */
7056 NOT_REACHED; /* NOTREACHED */
7058 case CURLYM_A: /* we've just matched an A */
7060 /* after first match, determine A's length: u.curlym.alen */
7061 if (ST.count == 1) {
7062 if (reginfo->is_utf8_target) {
7063 char *s = st->locinput;
7064 while (s < locinput) {
7070 ST.alen = locinput - st->locinput;
7073 ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me);
7076 PerlIO_printf(Perl_debug_log,
7077 "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n",
7078 (int)(REPORT_CODE_OFF+(depth*2)), "",
7079 (IV) ST.count, (IV)ST.alen)
7082 if (cur_eval && cur_eval->u.eval.close_paren &&
7083 cur_eval->u.eval.close_paren == (U32)ST.me->flags)
7087 I32 max = (ST.minmod ? ARG1(ST.me) : ARG2(ST.me));
7088 if ( max == REG_INFTY || ST.count < max )
7089 goto curlym_do_A; /* try to match another A */
7091 goto curlym_do_B; /* try to match B */
7093 case CURLYM_A_fail: /* just failed to match an A */
7094 REGCP_UNWIND(ST.cp);
7096 if (ST.minmod || ST.count < ARG1(ST.me) /* min*/
7097 || (cur_eval && cur_eval->u.eval.close_paren &&
7098 cur_eval->u.eval.close_paren == (U32)ST.me->flags))
7101 curlym_do_B: /* execute the B in /A{m,n}B/ */
7102 if (ST.c1 == CHRTEST_UNINIT) {
7103 /* calculate c1 and c2 for possible match of 1st char
7104 * following curly */
7105 ST.c1 = ST.c2 = CHRTEST_VOID;
7107 if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) {
7108 regnode *text_node = ST.B;
7109 if (! HAS_TEXT(text_node))
7110 FIND_NEXT_IMPT(text_node);
7113 (HAS_TEXT(text_node) && PL_regkind[OP(text_node)] == EXACT)
7115 But the former is redundant in light of the latter.
7117 if this changes back then the macro for
7118 IS_TEXT and friends need to change.
7120 if (PL_regkind[OP(text_node)] == EXACT) {
7121 if (! S_setup_EXACTISH_ST_c1_c2(aTHX_
7122 text_node, &ST.c1, ST.c1_utf8, &ST.c2, ST.c2_utf8,
7132 PerlIO_printf(Perl_debug_log,
7133 "%*s CURLYM trying tail with matches=%"IVdf"...\n",
7134 (int)(REPORT_CODE_OFF+(depth*2)),
7137 if (! NEXTCHR_IS_EOS && ST.c1 != CHRTEST_VOID) {
7138 if (! UTF8_IS_INVARIANT(nextchr) && utf8_target) {
7139 if (memNE(locinput, ST.c1_utf8, UTF8SKIP(locinput))
7140 && memNE(locinput, ST.c2_utf8, UTF8SKIP(locinput)))
7142 /* simulate B failing */
7144 PerlIO_printf(Perl_debug_log,
7145 "%*s CURLYM Fast bail next target=0x%"UVXf" c1=0x%"UVXf" c2=0x%"UVXf"\n",
7146 (int)(REPORT_CODE_OFF+(depth*2)),"",
7147 valid_utf8_to_uvchr((U8 *) locinput, NULL),
7148 valid_utf8_to_uvchr(ST.c1_utf8, NULL),
7149 valid_utf8_to_uvchr(ST.c2_utf8, NULL))
7151 state_num = CURLYM_B_fail;
7152 goto reenter_switch;
7155 else if (nextchr != ST.c1 && nextchr != ST.c2) {
7156 /* simulate B failing */
7158 PerlIO_printf(Perl_debug_log,
7159 "%*s CURLYM Fast bail next target=0x%X c1=0x%X c2=0x%X\n",
7160 (int)(REPORT_CODE_OFF+(depth*2)),"",
7161 (int) nextchr, ST.c1, ST.c2)
7163 state_num = CURLYM_B_fail;
7164 goto reenter_switch;
7169 /* emulate CLOSE: mark current A as captured */
7170 I32 paren = ST.me->flags;
7172 rex->offs[paren].start
7173 = HOPc(locinput, -ST.alen) - reginfo->strbeg;
7174 rex->offs[paren].end = locinput - reginfo->strbeg;
7175 if ((U32)paren > rex->lastparen)
7176 rex->lastparen = paren;
7177 rex->lastcloseparen = paren;
7180 rex->offs[paren].end = -1;
7181 if (cur_eval && cur_eval->u.eval.close_paren &&
7182 cur_eval->u.eval.close_paren == (U32)ST.me->flags)
7191 PUSH_STATE_GOTO(CURLYM_B, ST.B, locinput); /* match B */
7193 NOT_REACHED; /* NOTREACHED */
7195 case CURLYM_B_fail: /* just failed to match a B */
7196 REGCP_UNWIND(ST.cp);
7197 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
7199 I32 max = ARG2(ST.me);
7200 if (max != REG_INFTY && ST.count == max)
7202 goto curlym_do_A; /* try to match a further A */
7204 /* backtrack one A */
7205 if (ST.count == ARG1(ST.me) /* min */)
7208 SET_locinput(HOPc(locinput, -ST.alen));
7209 goto curlym_do_B; /* try to match B */
7212 #define ST st->u.curly
7214 #define CURLY_SETPAREN(paren, success) \
7217 rex->offs[paren].start = HOPc(locinput, -1) - reginfo->strbeg; \
7218 rex->offs[paren].end = locinput - reginfo->strbeg; \
7219 if (paren > rex->lastparen) \
7220 rex->lastparen = paren; \
7221 rex->lastcloseparen = paren; \
7224 rex->offs[paren].end = -1; \
7225 rex->lastparen = ST.lastparen; \
7226 rex->lastcloseparen = ST.lastcloseparen; \
7230 case STAR: /* /A*B/ where A is width 1 char */
7234 scan = NEXTOPER(scan);
7237 case PLUS: /* /A+B/ where A is width 1 char */
7241 scan = NEXTOPER(scan);
7244 case CURLYN: /* /(A){m,n}B/ where A is width 1 char */
7245 ST.paren = scan->flags; /* Which paren to set */
7246 ST.lastparen = rex->lastparen;
7247 ST.lastcloseparen = rex->lastcloseparen;
7248 if (ST.paren > maxopenparen)
7249 maxopenparen = ST.paren;
7250 ST.min = ARG1(scan); /* min to match */
7251 ST.max = ARG2(scan); /* max to match */
7252 if (cur_eval && cur_eval->u.eval.close_paren &&
7253 cur_eval->u.eval.close_paren == (U32)ST.paren) {
7257 scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE);
7260 case CURLY: /* /A{m,n}B/ where A is width 1 char */
7262 ST.min = ARG1(scan); /* min to match */
7263 ST.max = ARG2(scan); /* max to match */
7264 scan = NEXTOPER(scan) + NODE_STEP_REGNODE;
7267 * Lookahead to avoid useless match attempts
7268 * when we know what character comes next.
7270 * Used to only do .*x and .*?x, but now it allows
7271 * for )'s, ('s and (?{ ... })'s to be in the way
7272 * of the quantifier and the EXACT-like node. -- japhy
7275 assert(ST.min <= ST.max);
7276 if (! HAS_TEXT(next) && ! JUMPABLE(next)) {
7277 ST.c1 = ST.c2 = CHRTEST_VOID;
7280 regnode *text_node = next;
7282 if (! HAS_TEXT(text_node))
7283 FIND_NEXT_IMPT(text_node);
7285 if (! HAS_TEXT(text_node))
7286 ST.c1 = ST.c2 = CHRTEST_VOID;
7288 if ( PL_regkind[OP(text_node)] != EXACT ) {
7289 ST.c1 = ST.c2 = CHRTEST_VOID;
7293 /* Currently we only get here when
7295 PL_rekind[OP(text_node)] == EXACT
7297 if this changes back then the macro for IS_TEXT and
7298 friends need to change. */
7299 if (! S_setup_EXACTISH_ST_c1_c2(aTHX_
7300 text_node, &ST.c1, ST.c1_utf8, &ST.c2, ST.c2_utf8,
7312 char *li = locinput;
7315 regrepeat(rex, &li, ST.A, reginfo, ST.min, depth)
7321 if (ST.c1 == CHRTEST_VOID)
7322 goto curly_try_B_min;
7324 ST.oldloc = locinput;
7326 /* set ST.maxpos to the furthest point along the
7327 * string that could possibly match */
7328 if (ST.max == REG_INFTY) {
7329 ST.maxpos = reginfo->strend - 1;
7331 while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos))
7334 else if (utf8_target) {
7335 int m = ST.max - ST.min;
7336 for (ST.maxpos = locinput;
7337 m >0 && ST.maxpos < reginfo->strend; m--)
7338 ST.maxpos += UTF8SKIP(ST.maxpos);
7341 ST.maxpos = locinput + ST.max - ST.min;
7342 if (ST.maxpos >= reginfo->strend)
7343 ST.maxpos = reginfo->strend - 1;
7345 goto curly_try_B_min_known;
7349 /* avoid taking address of locinput, so it can remain
7351 char *li = locinput;
7352 ST.count = regrepeat(rex, &li, ST.A, reginfo, ST.max, depth);
7353 if (ST.count < ST.min)
7356 if ((ST.count > ST.min)
7357 && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL))
7359 /* A{m,n} must come at the end of the string, there's
7360 * no point in backing off ... */
7362 /* ...except that $ and \Z can match before *and* after
7363 newline at the end. Consider "\n\n" =~ /\n+\Z\n/.
7364 We may back off by one in this case. */
7365 if (UCHARAT(locinput - 1) == '\n' && OP(ST.B) != EOS)
7369 goto curly_try_B_max;
7372 NOT_REACHED; /* NOTREACHED */
7374 case CURLY_B_min_known_fail:
7375 /* failed to find B in a non-greedy match where c1,c2 valid */
7377 REGCP_UNWIND(ST.cp);
7379 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
7381 /* Couldn't or didn't -- move forward. */
7382 ST.oldloc = locinput;
7384 locinput += UTF8SKIP(locinput);
7388 curly_try_B_min_known:
7389 /* find the next place where 'B' could work, then call B */
7393 n = (ST.oldloc == locinput) ? 0 : 1;
7394 if (ST.c1 == ST.c2) {
7395 /* set n to utf8_distance(oldloc, locinput) */
7396 while (locinput <= ST.maxpos
7397 && memNE(locinput, ST.c1_utf8, UTF8SKIP(locinput)))
7399 locinput += UTF8SKIP(locinput);
7404 /* set n to utf8_distance(oldloc, locinput) */
7405 while (locinput <= ST.maxpos
7406 && memNE(locinput, ST.c1_utf8, UTF8SKIP(locinput))
7407 && memNE(locinput, ST.c2_utf8, UTF8SKIP(locinput)))
7409 locinput += UTF8SKIP(locinput);
7414 else { /* Not utf8_target */
7415 if (ST.c1 == ST.c2) {
7416 while (locinput <= ST.maxpos &&
7417 UCHARAT(locinput) != ST.c1)
7421 while (locinput <= ST.maxpos
7422 && UCHARAT(locinput) != ST.c1
7423 && UCHARAT(locinput) != ST.c2)
7426 n = locinput - ST.oldloc;
7428 if (locinput > ST.maxpos)
7431 /* In /a{m,n}b/, ST.oldloc is at "a" x m, locinput is
7432 * at b; check that everything between oldloc and
7433 * locinput matches */
7434 char *li = ST.oldloc;
7436 if (regrepeat(rex, &li, ST.A, reginfo, n, depth) < n)
7438 assert(n == REG_INFTY || locinput == li);
7440 CURLY_SETPAREN(ST.paren, ST.count);
7441 if (cur_eval && cur_eval->u.eval.close_paren &&
7442 cur_eval->u.eval.close_paren == (U32)ST.paren) {
7445 PUSH_STATE_GOTO(CURLY_B_min_known, ST.B, locinput);
7448 NOT_REACHED; /* NOTREACHED */
7450 case CURLY_B_min_fail:
7451 /* failed to find B in a non-greedy match where c1,c2 invalid */
7453 REGCP_UNWIND(ST.cp);
7455 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
7457 /* failed -- move forward one */
7459 char *li = locinput;
7460 if (!regrepeat(rex, &li, ST.A, reginfo, 1, depth)) {
7467 if (ST.count <= ST.max || (ST.max == REG_INFTY &&
7468 ST.count > 0)) /* count overflow ? */
7471 CURLY_SETPAREN(ST.paren, ST.count);
7472 if (cur_eval && cur_eval->u.eval.close_paren &&
7473 cur_eval->u.eval.close_paren == (U32)ST.paren) {
7476 PUSH_STATE_GOTO(CURLY_B_min, ST.B, locinput);
7481 NOT_REACHED; /* NOTREACHED */
7484 /* a successful greedy match: now try to match B */
7485 if (cur_eval && cur_eval->u.eval.close_paren &&
7486 cur_eval->u.eval.close_paren == (U32)ST.paren) {
7490 bool could_match = locinput < reginfo->strend;
7492 /* If it could work, try it. */
7493 if (ST.c1 != CHRTEST_VOID && could_match) {
7494 if (! UTF8_IS_INVARIANT(UCHARAT(locinput)) && utf8_target)
7496 could_match = memEQ(locinput,
7501 UTF8SKIP(locinput));
7504 could_match = UCHARAT(locinput) == ST.c1
7505 || UCHARAT(locinput) == ST.c2;
7508 if (ST.c1 == CHRTEST_VOID || could_match) {
7509 CURLY_SETPAREN(ST.paren, ST.count);
7510 PUSH_STATE_GOTO(CURLY_B_max, ST.B, locinput);
7512 NOT_REACHED; /* NOTREACHED */
7517 case CURLY_B_max_fail:
7518 /* failed to find B in a greedy match */
7520 REGCP_UNWIND(ST.cp);
7522 UNWIND_PAREN(ST.lastparen, ST.lastcloseparen);
7525 if (--ST.count < ST.min)
7527 locinput = HOPc(locinput, -1);
7528 goto curly_try_B_max;
7532 case END: /* last op of main pattern */
7535 /* we've just finished A in /(??{A})B/; now continue with B */
7537 st->u.eval.prev_rex = rex_sv; /* inner */
7539 /* Save *all* the positions. */
7540 st->u.eval.cp = regcppush(rex, 0, maxopenparen);
7541 rex_sv = cur_eval->u.eval.prev_rex;
7542 is_utf8_pat = reginfo->is_utf8_pat = cBOOL(RX_UTF8(rex_sv));
7543 SET_reg_curpm(rex_sv);
7544 rex = ReANY(rex_sv);
7545 rexi = RXi_GET(rex);
7546 cur_curlyx = cur_eval->u.eval.prev_curlyx;
7548 REGCP_SET(st->u.eval.lastcp);
7550 /* Restore parens of the outer rex without popping the
7552 S_regcp_restore(aTHX_ rex, cur_eval->u.eval.lastcp,
7555 st->u.eval.prev_eval = cur_eval;
7556 cur_eval = cur_eval->u.eval.prev_eval;
7558 PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ... %"UVxf"\n",
7559 REPORT_CODE_OFF+depth*2, "",PTR2UV(cur_eval)););
7560 if ( nochange_depth )
7563 PUSH_YES_STATE_GOTO(EVAL_AB, st->u.eval.prev_eval->u.eval.B,
7564 locinput); /* match B */
7567 if (locinput < reginfo->till) {
7568 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
7569 "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n",
7571 (long)(locinput - startpos),
7572 (long)(reginfo->till - startpos),
7575 sayNO_SILENT; /* Cannot match: too short. */
7577 sayYES; /* Success! */
7579 case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */
7581 PerlIO_printf(Perl_debug_log,
7582 "%*s %ssubpattern success...%s\n",
7583 REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]));
7584 sayYES; /* Success! */
7587 #define ST st->u.ifmatch
7592 case SUSPEND: /* (?>A) */
7594 newstart = locinput;
7597 case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */
7599 goto ifmatch_trivial_fail_test;
7601 case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */
7603 ifmatch_trivial_fail_test:
7605 char * const s = HOPBACKc(locinput, scan->flags);
7610 sw = 1 - cBOOL(ST.wanted);
7614 next = scan + ARG(scan);
7622 newstart = locinput;
7626 ST.logical = logical;
7627 logical = 0; /* XXX: reset state of logical once it has been saved into ST */
7629 /* execute body of (?...A) */
7630 PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan)), newstart);
7632 NOT_REACHED; /* NOTREACHED */
7635 case IFMATCH_A_fail: /* body of (?...A) failed */
7636 ST.wanted = !ST.wanted;
7639 case IFMATCH_A: /* body of (?...A) succeeded */
7641 sw = cBOOL(ST.wanted);
7643 else if (!ST.wanted)
7646 if (OP(ST.me) != SUSPEND) {
7647 /* restore old position except for (?>...) */
7648 locinput = st->locinput;
7650 scan = ST.me + ARG(ST.me);
7653 continue; /* execute B */
7657 case LONGJMP: /* alternative with many branches compiles to
7658 * (BRANCHJ; EXACT ...; LONGJMP ) x N */
7659 next = scan + ARG(scan);
7664 case COMMIT: /* (*COMMIT) */
7665 reginfo->cutpoint = reginfo->strend;
7668 case PRUNE: /* (*PRUNE) */
7670 sv_yes_mark = sv_commit = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
7671 PUSH_STATE_GOTO(COMMIT_next, next, locinput);
7673 NOT_REACHED; /* NOTREACHED */
7675 case COMMIT_next_fail:
7679 case OPFAIL: /* (*FAIL) */
7682 NOT_REACHED; /* NOTREACHED */
7684 #define ST st->u.mark
7685 case MARKPOINT: /* (*MARK:foo) */
7686 ST.prev_mark = mark_state;
7687 ST.mark_name = sv_commit = sv_yes_mark
7688 = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
7690 ST.mark_loc = locinput;
7691 PUSH_YES_STATE_GOTO(MARKPOINT_next, next, locinput);
7693 NOT_REACHED; /* NOTREACHED */
7695 case MARKPOINT_next:
7696 mark_state = ST.prev_mark;
7699 NOT_REACHED; /* NOTREACHED */
7701 case MARKPOINT_next_fail:
7702 if (popmark && sv_eq(ST.mark_name,popmark))
7704 if (ST.mark_loc > startpoint)
7705 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
7706 popmark = NULL; /* we found our mark */
7707 sv_commit = ST.mark_name;
7710 PerlIO_printf(Perl_debug_log,
7711 "%*s %ssetting cutpoint to mark:%"SVf"...%s\n",
7712 REPORT_CODE_OFF+depth*2, "",
7713 PL_colors[4], SVfARG(sv_commit), PL_colors[5]);
7716 mark_state = ST.prev_mark;
7717 sv_yes_mark = mark_state ?
7718 mark_state->u.mark.mark_name : NULL;
7721 NOT_REACHED; /* NOTREACHED */
7723 case SKIP: /* (*SKIP) */
7725 /* (*SKIP) : if we fail we cut here*/
7726 ST.mark_name = NULL;
7727 ST.mark_loc = locinput;
7728 PUSH_STATE_GOTO(SKIP_next,next, locinput);
7730 /* (*SKIP:NAME) : if there is a (*MARK:NAME) fail where it was,
7731 otherwise do nothing. Meaning we need to scan
7733 regmatch_state *cur = mark_state;
7734 SV *find = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
7737 if ( sv_eq( cur->u.mark.mark_name,
7740 ST.mark_name = find;
7741 PUSH_STATE_GOTO( SKIP_next, next, locinput);
7743 cur = cur->u.mark.prev_mark;
7746 /* Didn't find our (*MARK:NAME) so ignore this (*SKIP:NAME) */
7749 case SKIP_next_fail:
7751 /* (*CUT:NAME) - Set up to search for the name as we
7752 collapse the stack*/
7753 popmark = ST.mark_name;
7755 /* (*CUT) - No name, we cut here.*/
7756 if (ST.mark_loc > startpoint)
7757 reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1);
7758 /* but we set sv_commit to latest mark_name if there
7759 is one so they can test to see how things lead to this
7762 sv_commit=mark_state->u.mark.mark_name;
7767 NOT_REACHED; /* NOTREACHED */
7770 case LNBREAK: /* \R */
7771 if ((n=is_LNBREAK_safe(locinput, reginfo->strend, utf8_target))) {
7778 PerlIO_printf(Perl_error_log, "%"UVxf" %d\n",
7779 PTR2UV(scan), OP(scan));
7780 Perl_croak(aTHX_ "regexp memory corruption");
7782 /* this is a point to jump to in order to increment
7783 * locinput by one character */
7785 assert(!NEXTCHR_IS_EOS);
7787 locinput += PL_utf8skip[nextchr];
7788 /* locinput is allowed to go 1 char off the end, but not 2+ */
7789 if (locinput > reginfo->strend)
7798 /* switch break jumps here */
7799 scan = next; /* prepare to execute the next op and ... */
7800 continue; /* ... jump back to the top, reusing st */
7804 /* push a state that backtracks on success */
7805 st->u.yes.prev_yes_state = yes_state;
7809 /* push a new regex state, then continue at scan */
7811 regmatch_state *newst;
7814 regmatch_state *cur = st;
7815 regmatch_state *curyes = yes_state;
7817 regmatch_slab *slab = PL_regmatch_slab;
7818 for (;curd > -1;cur--,curd--) {
7819 if (cur < SLAB_FIRST(slab)) {
7821 cur = SLAB_LAST(slab);
7823 PerlIO_printf(Perl_error_log, "%*s#%-3d %-10s %s\n",
7824 REPORT_CODE_OFF + 2 + depth * 2,"",
7825 curd, PL_reg_name[cur->resume_state],
7826 (curyes == cur) ? "yes" : ""
7829 curyes = cur->u.yes.prev_yes_state;
7832 DEBUG_STATE_pp("push")
7835 st->locinput = locinput;
7837 if (newst > SLAB_LAST(PL_regmatch_slab))
7838 newst = S_push_slab(aTHX);
7839 PL_regmatch_state = newst;
7841 locinput = pushinput;
7849 * We get here only if there's trouble -- normally "case END" is
7850 * the terminating point.
7852 Perl_croak(aTHX_ "corrupted regexp pointers");
7855 NOT_REACHED; /* NOTREACHED */
7859 /* we have successfully completed a subexpression, but we must now
7860 * pop to the state marked by yes_state and continue from there */
7861 assert(st != yes_state);
7863 while (st != yes_state) {
7865 if (st < SLAB_FIRST(PL_regmatch_slab)) {
7866 PL_regmatch_slab = PL_regmatch_slab->prev;
7867 st = SLAB_LAST(PL_regmatch_slab);
7871 DEBUG_STATE_pp("pop (no final)");
7873 DEBUG_STATE_pp("pop (yes)");
7879 while (yes_state < SLAB_FIRST(PL_regmatch_slab)
7880 || yes_state > SLAB_LAST(PL_regmatch_slab))
7882 /* not in this slab, pop slab */
7883 depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1);
7884 PL_regmatch_slab = PL_regmatch_slab->prev;
7885 st = SLAB_LAST(PL_regmatch_slab);
7887 depth -= (st - yes_state);
7890 yes_state = st->u.yes.prev_yes_state;
7891 PL_regmatch_state = st;
7894 locinput= st->locinput;
7895 state_num = st->resume_state + no_final;
7896 goto reenter_switch;
7899 DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n",
7900 PL_colors[4], PL_colors[5]));
7902 if (reginfo->info_aux_eval) {
7903 /* each successfully executed (?{...}) block does the equivalent of
7904 * local $^R = do {...}
7905 * When popping the save stack, all these locals would be undone;
7906 * bypass this by setting the outermost saved $^R to the latest
7908 /* I dont know if this is needed or works properly now.
7909 * see code related to PL_replgv elsewhere in this file.
7912 if (oreplsv != GvSV(PL_replgv))
7913 sv_setsv(oreplsv, GvSV(PL_replgv));
7920 PerlIO_printf(Perl_debug_log,
7921 "%*s %sfailed...%s\n",
7922 REPORT_CODE_OFF+depth*2, "",
7923 PL_colors[4], PL_colors[5])
7935 /* there's a previous state to backtrack to */
7937 if (st < SLAB_FIRST(PL_regmatch_slab)) {
7938 PL_regmatch_slab = PL_regmatch_slab->prev;
7939 st = SLAB_LAST(PL_regmatch_slab);
7941 PL_regmatch_state = st;
7942 locinput= st->locinput;
7944 DEBUG_STATE_pp("pop");
7946 if (yes_state == st)
7947 yes_state = st->u.yes.prev_yes_state;
7949 state_num = st->resume_state + 1; /* failure = success + 1 */
7950 goto reenter_switch;
7955 if (rex->intflags & PREGf_VERBARG_SEEN) {
7956 SV *sv_err = get_sv("REGERROR", 1);
7957 SV *sv_mrk = get_sv("REGMARK", 1);
7959 sv_commit = &PL_sv_no;
7961 sv_yes_mark = &PL_sv_yes;
7964 sv_commit = &PL_sv_yes;
7965 sv_yes_mark = &PL_sv_no;
7969 sv_setsv(sv_err, sv_commit);
7970 sv_setsv(sv_mrk, sv_yes_mark);
7974 if (last_pushed_cv) {
7977 PERL_UNUSED_VAR(SP);
7980 assert(!result || locinput - reginfo->strbeg >= 0);
7981 return result ? locinput - reginfo->strbeg : -1;
7985 - regrepeat - repeatedly match something simple, report how many
7987 * What 'simple' means is a node which can be the operand of a quantifier like
7990 * startposp - pointer a pointer to the start position. This is updated
7991 * to point to the byte following the highest successful
7993 * p - the regnode to be repeatedly matched against.
7994 * reginfo - struct holding match state, such as strend
7995 * max - maximum number of things to match.
7996 * depth - (for debugging) backtracking depth.
7999 S_regrepeat(pTHX_ regexp *prog, char **startposp, const regnode *p,
8000 regmatch_info *const reginfo, I32 max, int depth)
8002 char *scan; /* Pointer to current position in target string */
8004 char *loceol = reginfo->strend; /* local version */
8005 I32 hardcount = 0; /* How many matches so far */
8006 bool utf8_target = reginfo->is_utf8_target;
8007 unsigned int to_complement = 0; /* Invert the result? */
8009 _char_class_number classnum;
8011 PERL_UNUSED_ARG(depth);
8014 PERL_ARGS_ASSERT_REGREPEAT;
8017 if (max == REG_INFTY)
8019 else if (! utf8_target && loceol - scan > max)
8020 loceol = scan + max;
8022 /* Here, for the case of a non-UTF-8 target we have adjusted <loceol> down
8023 * to the maximum of how far we should go in it (leaving it set to the real
8024 * end, if the maximum permissible would take us beyond that). This allows
8025 * us to make the loop exit condition that we haven't gone past <loceol> to
8026 * also mean that we haven't exceeded the max permissible count, saving a
8027 * test each time through the loop. But it assumes that the OP matches a
8028 * single byte, which is true for most of the OPs below when applied to a
8029 * non-UTF-8 target. Those relatively few OPs that don't have this
8030 * characteristic will have to compensate.
8032 * There is no adjustment for UTF-8 targets, as the number of bytes per
8033 * character varies. OPs will have to test both that the count is less
8034 * than the max permissible (using <hardcount> to keep track), and that we
8035 * are still within the bounds of the string (using <loceol>. A few OPs
8036 * match a single byte no matter what the encoding. They can omit the max
8037 * test if, for the UTF-8 case, they do the adjustment that was skipped
8040 * Thus, the code above sets things up for the common case; and exceptional
8041 * cases need extra work; the common case is to make sure <scan> doesn't
8042 * go past <loceol>, and for UTF-8 to also use <hardcount> to make sure the
8043 * count doesn't exceed the maximum permissible */
8048 while (scan < loceol && hardcount < max && *scan != '\n') {
8049 scan += UTF8SKIP(scan);
8053 while (scan < loceol && *scan != '\n')
8059 while (scan < loceol && hardcount < max) {
8060 scan += UTF8SKIP(scan);
8067 case CANY: /* Move <scan> forward <max> bytes, unless goes off end */
8068 if (utf8_target && loceol - scan > max) {
8070 /* <loceol> hadn't been adjusted in the UTF-8 case */
8078 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
8079 if (utf8_target && UTF8_IS_ABOVE_LATIN1(*scan)) {
8080 _CHECK_AND_OUTPUT_WIDE_LOCALE_UTF8_MSG(scan, loceol);
8084 assert(STR_LEN(p) == reginfo->is_utf8_pat ? UTF8SKIP(STRING(p)) : 1);
8088 /* Can use a simple loop if the pattern char to match on is invariant
8089 * under UTF-8, or both target and pattern aren't UTF-8. Note that we
8090 * can use UTF8_IS_INVARIANT() even if the pattern isn't UTF-8, as it's
8091 * true iff it doesn't matter if the argument is in UTF-8 or not */
8092 if (UTF8_IS_INVARIANT(c) || (! utf8_target && ! reginfo->is_utf8_pat)) {
8093 if (utf8_target && loceol - scan > max) {
8094 /* We didn't adjust <loceol> because is UTF-8, but ok to do so,
8095 * since here, to match at all, 1 char == 1 byte */
8096 loceol = scan + max;
8098 while (scan < loceol && UCHARAT(scan) == c) {
8102 else if (reginfo->is_utf8_pat) {
8104 STRLEN scan_char_len;
8106 /* When both target and pattern are UTF-8, we have to do
8108 while (hardcount < max
8110 && (scan_char_len = UTF8SKIP(scan)) <= STR_LEN(p)
8111 && memEQ(scan, STRING(p), scan_char_len))
8113 scan += scan_char_len;
8117 else if (! UTF8_IS_ABOVE_LATIN1(c)) {
8119 /* Target isn't utf8; convert the character in the UTF-8
8120 * pattern to non-UTF8, and do a simple loop */
8121 c = TWO_BYTE_UTF8_TO_NATIVE(c, *(STRING(p) + 1));
8122 while (scan < loceol && UCHARAT(scan) == c) {
8125 } /* else pattern char is above Latin1, can't possibly match the
8130 /* Here, the string must be utf8; pattern isn't, and <c> is
8131 * different in utf8 than not, so can't compare them directly.
8132 * Outside the loop, find the two utf8 bytes that represent c, and
8133 * then look for those in sequence in the utf8 string */
8134 U8 high = UTF8_TWO_BYTE_HI(c);
8135 U8 low = UTF8_TWO_BYTE_LO(c);
8137 while (hardcount < max
8138 && scan + 1 < loceol
8139 && UCHARAT(scan) == high
8140 && UCHARAT(scan + 1) == low)
8148 case EXACTFA_NO_TRIE: /* This node only generated for non-utf8 patterns */
8149 assert(! reginfo->is_utf8_pat);
8152 utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
8156 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
8157 utf8_flags = FOLDEQ_LOCALE;
8160 case EXACTF: /* This node only generated for non-utf8 patterns */
8161 assert(! reginfo->is_utf8_pat);
8166 if (! utf8_target) {
8169 utf8_flags = FOLDEQ_LOCALE | FOLDEQ_S2_ALREADY_FOLDED
8170 | FOLDEQ_S2_FOLDS_SANE;
8175 utf8_flags = reginfo->is_utf8_pat ? FOLDEQ_S2_ALREADY_FOLDED : 0;
8179 U8 c1_utf8[UTF8_MAXBYTES+1], c2_utf8[UTF8_MAXBYTES+1];
8181 assert(STR_LEN(p) == reginfo->is_utf8_pat ? UTF8SKIP(STRING(p)) : 1);
8183 if (S_setup_EXACTISH_ST_c1_c2(aTHX_ p, &c1, c1_utf8, &c2, c2_utf8,
8186 if (c1 == CHRTEST_VOID) {
8187 /* Use full Unicode fold matching */
8188 char *tmpeol = reginfo->strend;
8189 STRLEN pat_len = reginfo->is_utf8_pat ? UTF8SKIP(STRING(p)) : 1;
8190 while (hardcount < max
8191 && foldEQ_utf8_flags(scan, &tmpeol, 0, utf8_target,
8192 STRING(p), NULL, pat_len,
8193 reginfo->is_utf8_pat, utf8_flags))
8196 tmpeol = reginfo->strend;
8200 else if (utf8_target) {
8202 while (scan < loceol
8204 && memEQ(scan, c1_utf8, UTF8SKIP(scan)))
8206 scan += UTF8SKIP(scan);
8211 while (scan < loceol
8213 && (memEQ(scan, c1_utf8, UTF8SKIP(scan))
8214 || memEQ(scan, c2_utf8, UTF8SKIP(scan))))
8216 scan += UTF8SKIP(scan);
8221 else if (c1 == c2) {
8222 while (scan < loceol && UCHARAT(scan) == c1) {
8227 while (scan < loceol &&
8228 (UCHARAT(scan) == c1 || UCHARAT(scan) == c2))
8237 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
8241 while (hardcount < max
8243 && reginclass(prog, p, (U8*)scan, (U8*) loceol, utf8_target))
8245 scan += UTF8SKIP(scan);
8249 while (scan < loceol && REGINCLASS(prog, p, (U8*)scan))
8254 /* The argument (FLAGS) to all the POSIX node types is the class number */
8261 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
8262 if (! utf8_target) {
8263 while (scan < loceol && to_complement ^ cBOOL(isFOO_lc(FLAGS(p),
8269 while (hardcount < max && scan < loceol
8270 && to_complement ^ cBOOL(isFOO_utf8_lc(FLAGS(p),
8273 scan += UTF8SKIP(scan);
8286 if (utf8_target && loceol - scan > max) {
8288 /* We didn't adjust <loceol> at the beginning of this routine
8289 * because is UTF-8, but it is actually ok to do so, since here, to
8290 * match, 1 char == 1 byte. */
8291 loceol = scan + max;
8293 while (scan < loceol && _generic_isCC_A((U8) *scan, FLAGS(p))) {
8306 if (! utf8_target) {
8307 while (scan < loceol && ! _generic_isCC_A((U8) *scan, FLAGS(p))) {
8313 /* The complement of something that matches only ASCII matches all
8314 * non-ASCII, plus everything in ASCII that isn't in the class. */
8315 while (hardcount < max && scan < loceol
8316 && (! isASCII_utf8(scan)
8317 || ! _generic_isCC_A((U8) *scan, FLAGS(p))))
8319 scan += UTF8SKIP(scan);
8330 if (! utf8_target) {
8331 while (scan < loceol && to_complement
8332 ^ cBOOL(_generic_isCC((U8) *scan, FLAGS(p))))
8339 classnum = (_char_class_number) FLAGS(p);
8340 if (classnum < _FIRST_NON_SWASH_CC) {
8342 /* Here, a swash is needed for above-Latin1 code points.
8343 * Process as many Latin1 code points using the built-in rules.
8344 * Go to another loop to finish processing upon encountering
8345 * the first Latin1 code point. We could do that in this loop
8346 * as well, but the other way saves having to test if the swash
8347 * has been loaded every time through the loop: extra space to
8349 while (hardcount < max && scan < loceol) {
8350 if (UTF8_IS_INVARIANT(*scan)) {
8351 if (! (to_complement ^ cBOOL(_generic_isCC((U8) *scan,
8358 else if (UTF8_IS_DOWNGRADEABLE_START(*scan)) {
8359 if (! (to_complement
8360 ^ cBOOL(_generic_isCC(TWO_BYTE_UTF8_TO_NATIVE(*scan,
8369 goto found_above_latin1;
8376 /* For these character classes, the knowledge of how to handle
8377 * every code point is compiled in to Perl via a macro. This
8378 * code is written for making the loops as tight as possible.
8379 * It could be refactored to save space instead */
8381 case _CC_ENUM_SPACE:
8382 while (hardcount < max
8384 && (to_complement ^ cBOOL(isSPACE_utf8(scan))))
8386 scan += UTF8SKIP(scan);
8390 case _CC_ENUM_BLANK:
8391 while (hardcount < max
8393 && (to_complement ^ cBOOL(isBLANK_utf8(scan))))
8395 scan += UTF8SKIP(scan);
8399 case _CC_ENUM_XDIGIT:
8400 while (hardcount < max
8402 && (to_complement ^ cBOOL(isXDIGIT_utf8(scan))))
8404 scan += UTF8SKIP(scan);
8408 case _CC_ENUM_VERTSPACE:
8409 while (hardcount < max
8411 && (to_complement ^ cBOOL(isVERTWS_utf8(scan))))
8413 scan += UTF8SKIP(scan);
8417 case _CC_ENUM_CNTRL:
8418 while (hardcount < max
8420 && (to_complement ^ cBOOL(isCNTRL_utf8(scan))))
8422 scan += UTF8SKIP(scan);
8427 Perl_croak(aTHX_ "panic: regrepeat() node %d='%s' has an unexpected character class '%d'", OP(p), PL_reg_name[OP(p)], classnum);
8433 found_above_latin1: /* Continuation of POSIXU and NPOSIXU */
8435 /* Load the swash if not already present */
8436 if (! PL_utf8_swash_ptrs[classnum]) {
8437 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
8438 PL_utf8_swash_ptrs[classnum] = _core_swash_init(
8442 PL_XPosix_ptrs[classnum], &flags);
8445 while (hardcount < max && scan < loceol
8446 && to_complement ^ cBOOL(_generic_utf8(
8449 swash_fetch(PL_utf8_swash_ptrs[classnum],
8453 scan += UTF8SKIP(scan);
8460 while (hardcount < max && scan < loceol &&
8461 (c=is_LNBREAK_utf8_safe(scan, loceol))) {
8466 /* LNBREAK can match one or two latin chars, which is ok, but we
8467 * have to use hardcount in this situation, and throw away the
8468 * adjustment to <loceol> done before the switch statement */
8469 loceol = reginfo->strend;
8470 while (scan < loceol && (c=is_LNBREAK_latin1_safe(scan, loceol))) {
8479 _CHECK_AND_WARN_PROBLEMATIC_LOCALE;
8493 /* These are all 0 width, so match right here or not at all. */
8497 Perl_croak(aTHX_ "panic: regrepeat() called with unrecognized node type %d='%s'", OP(p), PL_reg_name[OP(p)]);
8499 NOT_REACHED; /* NOTREACHED */
8506 c = scan - *startposp;
8510 GET_RE_DEBUG_FLAGS_DECL;
8512 SV * const prop = sv_newmortal();
8513 regprop(prog, prop, p, reginfo, NULL);
8514 PerlIO_printf(Perl_debug_log,
8515 "%*s %s can match %"IVdf" times out of %"IVdf"...\n",
8516 REPORT_CODE_OFF + depth*2, "", SvPVX_const(prop),(IV)c,(IV)max);
8524 #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION)
8526 - regclass_swash - prepare the utf8 swash. Wraps the shared core version to
8527 create a copy so that changes the caller makes won't change the shared one.
8528 If <altsvp> is non-null, will return NULL in it, for back-compat.
8531 Perl_regclass_swash(pTHX_ const regexp *prog, const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
8533 PERL_ARGS_ASSERT_REGCLASS_SWASH;
8539 return newSVsv(_get_regclass_nonbitmap_data(prog, node, doinit, listsvp, NULL, NULL));
8542 #endif /* !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION) */
8545 - reginclass - determine if a character falls into a character class
8547 n is the ANYOF-type regnode
8548 p is the target string
8549 p_end points to one byte beyond the end of the target string
8550 utf8_target tells whether p is in UTF-8.
8552 Returns true if matched; false otherwise.
8554 Note that this can be a synthetic start class, a combination of various
8555 nodes, so things you think might be mutually exclusive, such as locale,
8556 aren't. It can match both locale and non-locale
8561 S_reginclass(pTHX_ regexp * const prog, const regnode * const n, const U8* const p, const U8* const p_end, const bool utf8_target)
8564 const char flags = ANYOF_FLAGS(n);
8568 PERL_ARGS_ASSERT_REGINCLASS;
8570 /* If c is not already the code point, get it. Note that
8571 * UTF8_IS_INVARIANT() works even if not in UTF-8 */
8572 if (! UTF8_IS_INVARIANT(c) && utf8_target) {
8574 c = utf8n_to_uvchr(p, p_end - p, &c_len,
8575 (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV)
8576 | UTF8_ALLOW_FFFF | UTF8_CHECK_ONLY);
8577 /* see [perl #37836] for UTF8_ALLOW_ANYUV; [perl #38293] for
8578 * UTF8_ALLOW_FFFF */
8579 if (c_len == (STRLEN)-1)
8580 Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)");
8581 if (c > 255 && OP(n) == ANYOFL && ! is_ANYOF_SYNTHETIC(n)) {
8582 _CHECK_AND_OUTPUT_WIDE_LOCALE_CP_MSG(c);
8586 /* If this character is potentially in the bitmap, check it */
8587 if (c < NUM_ANYOF_CODE_POINTS) {
8588 if (ANYOF_BITMAP_TEST(n, c))
8590 else if ((flags & ANYOF_MATCHES_ALL_NON_UTF8_NON_ASCII)
8596 else if (flags & ANYOF_LOCALE_FLAGS) {
8597 if ((flags & ANYOF_LOC_FOLD)
8599 && ANYOF_BITMAP_TEST(n, PL_fold_locale[c]))
8603 else if (ANYOF_POSIXL_TEST_ANY_SET(n)
8607 /* The data structure is arranged so bits 0, 2, 4, ... are set
8608 * if the class includes the Posix character class given by
8609 * bit/2; and 1, 3, 5, ... are set if the class includes the
8610 * complemented Posix class given by int(bit/2). So we loop
8611 * through the bits, each time changing whether we complement
8612 * the result or not. Suppose for the sake of illustration
8613 * that bits 0-3 mean respectively, \w, \W, \s, \S. If bit 0
8614 * is set, it means there is a match for this ANYOF node if the
8615 * character is in the class given by the expression (0 / 2 = 0
8616 * = \w). If it is in that class, isFOO_lc() will return 1,
8617 * and since 'to_complement' is 0, the result will stay TRUE,
8618 * and we exit the loop. Suppose instead that bit 0 is 0, but
8619 * bit 1 is 1. That means there is a match if the character
8620 * matches \W. We won't bother to call isFOO_lc() on bit 0,
8621 * but will on bit 1. On the second iteration 'to_complement'
8622 * will be 1, so the exclusive or will reverse things, so we
8623 * are testing for \W. On the third iteration, 'to_complement'
8624 * will be 0, and we would be testing for \s; the fourth
8625 * iteration would test for \S, etc.
8627 * Note that this code assumes that all the classes are closed
8628 * under folding. For example, if a character matches \w, then
8629 * its fold does too; and vice versa. This should be true for
8630 * any well-behaved locale for all the currently defined Posix
8631 * classes, except for :lower: and :upper:, which are handled
8632 * by the pseudo-class :cased: which matches if either of the
8633 * other two does. To get rid of this assumption, an outer
8634 * loop could be used below to iterate over both the source
8635 * character, and its fold (if different) */
8638 int to_complement = 0;
8640 while (count < ANYOF_MAX) {
8641 if (ANYOF_POSIXL_TEST(n, count)
8642 && to_complement ^ cBOOL(isFOO_lc(count/2, (U8) c)))
8655 /* If the bitmap didn't (or couldn't) match, and something outside the
8656 * bitmap could match, try that. */
8658 if (c >= NUM_ANYOF_CODE_POINTS
8659 && (flags & ANYOF_MATCHES_ALL_ABOVE_BITMAP))
8661 match = TRUE; /* Everything above the bitmap matches */
8663 else if ((flags & ANYOF_HAS_NONBITMAP_NON_UTF8_MATCHES)
8664 || (utf8_target && (flags & ANYOF_HAS_UTF8_NONBITMAP_MATCHES))
8665 || ((flags & ANYOF_LOC_FOLD)
8666 && IN_UTF8_CTYPE_LOCALE
8667 && ARG(n) != ANYOF_ONLY_HAS_BITMAP))
8669 SV* only_utf8_locale = NULL;
8670 SV * const sw = _get_regclass_nonbitmap_data(prog, n, TRUE, 0,
8671 &only_utf8_locale, NULL);
8677 } else { /* Convert to utf8 */
8678 utf8_p = utf8_buffer;
8679 append_utf8_from_native_byte(*p, &utf8_p);
8680 utf8_p = utf8_buffer;
8683 if (swash_fetch(sw, utf8_p, TRUE)) {
8687 if (! match && only_utf8_locale && IN_UTF8_CTYPE_LOCALE) {
8688 match = _invlist_contains_cp(only_utf8_locale, c);
8692 if (UNICODE_IS_SUPER(c)
8693 && (flags & ANYOF_WARN_SUPER)
8694 && ckWARN_d(WARN_NON_UNICODE))
8696 Perl_warner(aTHX_ packWARN(WARN_NON_UNICODE),
8697 "Matched non-Unicode code point 0x%04"UVXf" against Unicode property; may not be portable", c);
8701 #if ANYOF_INVERT != 1
8702 /* Depending on compiler optimization cBOOL takes time, so if don't have to
8704 # error ANYOF_INVERT needs to be set to 1, or guarded with cBOOL below,
8707 /* The xor complements the return if to invert: 1^1 = 0, 1^0 = 1 */
8708 return (flags & ANYOF_INVERT) ^ match;
8712 S_reghop3(U8 *s, SSize_t off, const U8* lim)
8714 /* return the position 'off' UTF-8 characters away from 's', forward if
8715 * 'off' >= 0, backwards if negative. But don't go outside of position
8716 * 'lim', which better be < s if off < 0 */
8718 PERL_ARGS_ASSERT_REGHOP3;
8721 while (off-- && s < lim) {
8722 /* XXX could check well-formedness here */
8727 while (off++ && s > lim) {
8729 if (UTF8_IS_CONTINUED(*s)) {
8730 while (s > lim && UTF8_IS_CONTINUATION(*s))
8733 /* XXX could check well-formedness here */
8740 S_reghop4(U8 *s, SSize_t off, const U8* llim, const U8* rlim)
8742 PERL_ARGS_ASSERT_REGHOP4;
8745 while (off-- && s < rlim) {
8746 /* XXX could check well-formedness here */
8751 while (off++ && s > llim) {
8753 if (UTF8_IS_CONTINUED(*s)) {
8754 while (s > llim && UTF8_IS_CONTINUATION(*s))
8757 /* XXX could check well-formedness here */
8763 /* like reghop3, but returns NULL on overrun, rather than returning last
8767 S_reghopmaybe3(U8* s, SSize_t off, const U8* lim)
8769 PERL_ARGS_ASSERT_REGHOPMAYBE3;
8772 while (off-- && s < lim) {
8773 /* XXX could check well-formedness here */
8780 while (off++ && s > lim) {
8782 if (UTF8_IS_CONTINUED(*s)) {
8783 while (s > lim && UTF8_IS_CONTINUATION(*s))
8786 /* XXX could check well-formedness here */
8795 /* when executing a regex that may have (?{}), extra stuff needs setting
8796 up that will be visible to the called code, even before the current
8797 match has finished. In particular:
8799 * $_ is localised to the SV currently being matched;
8800 * pos($_) is created if necessary, ready to be updated on each call-out
8802 * a fake PMOP is created that can be set to PL_curpm (normally PL_curpm
8803 isn't set until the current pattern is successfully finished), so that
8804 $1 etc of the match-so-far can be seen;
8805 * save the old values of subbeg etc of the current regex, and set then
8806 to the current string (again, this is normally only done at the end
8811 S_setup_eval_state(pTHX_ regmatch_info *const reginfo)
8814 regexp *const rex = ReANY(reginfo->prog);
8815 regmatch_info_aux_eval *eval_state = reginfo->info_aux_eval;
8817 eval_state->rex = rex;
8820 /* Make $_ available to executed code. */
8821 if (reginfo->sv != DEFSV) {
8823 DEFSV_set(reginfo->sv);
8826 if (!(mg = mg_find_mglob(reginfo->sv))) {
8827 /* prepare for quick setting of pos */
8828 mg = sv_magicext_mglob(reginfo->sv);
8831 eval_state->pos_magic = mg;
8832 eval_state->pos = mg->mg_len;
8833 eval_state->pos_flags = mg->mg_flags;
8836 eval_state->pos_magic = NULL;
8838 if (!PL_reg_curpm) {
8839 /* PL_reg_curpm is a fake PMOP that we can attach the current
8840 * regex to and point PL_curpm at, so that $1 et al are visible
8841 * within a /(?{})/. It's just allocated once per interpreter the
8842 * first time its needed */
8843 Newxz(PL_reg_curpm, 1, PMOP);
8846 SV* const repointer = &PL_sv_undef;
8847 /* this regexp is also owned by the new PL_reg_curpm, which
8848 will try to free it. */
8849 av_push(PL_regex_padav, repointer);
8850 PL_reg_curpm->op_pmoffset = av_tindex(PL_regex_padav);
8851 PL_regex_pad = AvARRAY(PL_regex_padav);
8855 SET_reg_curpm(reginfo->prog);
8856 eval_state->curpm = PL_curpm;
8857 PL_curpm = PL_reg_curpm;
8858 if (RXp_MATCH_COPIED(rex)) {
8859 /* Here is a serious problem: we cannot rewrite subbeg,
8860 since it may be needed if this match fails. Thus
8861 $` inside (?{}) could fail... */
8862 eval_state->subbeg = rex->subbeg;
8863 eval_state->sublen = rex->sublen;
8864 eval_state->suboffset = rex->suboffset;
8865 eval_state->subcoffset = rex->subcoffset;
8867 eval_state->saved_copy = rex->saved_copy;
8869 RXp_MATCH_COPIED_off(rex);
8872 eval_state->subbeg = NULL;
8873 rex->subbeg = (char *)reginfo->strbeg;
8875 rex->subcoffset = 0;
8876 rex->sublen = reginfo->strend - reginfo->strbeg;
8880 /* destructor to clear up regmatch_info_aux and regmatch_info_aux_eval */
8883 S_cleanup_regmatch_info_aux(pTHX_ void *arg)
8885 regmatch_info_aux *aux = (regmatch_info_aux *) arg;
8886 regmatch_info_aux_eval *eval_state = aux->info_aux_eval;
8889 Safefree(aux->poscache);
8893 /* undo the effects of S_setup_eval_state() */
8895 if (eval_state->subbeg) {
8896 regexp * const rex = eval_state->rex;
8897 rex->subbeg = eval_state->subbeg;
8898 rex->sublen = eval_state->sublen;
8899 rex->suboffset = eval_state->suboffset;
8900 rex->subcoffset = eval_state->subcoffset;
8902 rex->saved_copy = eval_state->saved_copy;
8904 RXp_MATCH_COPIED_on(rex);
8906 if (eval_state->pos_magic)
8908 eval_state->pos_magic->mg_len = eval_state->pos;
8909 eval_state->pos_magic->mg_flags =
8910 (eval_state->pos_magic->mg_flags & ~MGf_BYTES)
8911 | (eval_state->pos_flags & MGf_BYTES);
8914 PL_curpm = eval_state->curpm;
8917 PL_regmatch_state = aux->old_regmatch_state;
8918 PL_regmatch_slab = aux->old_regmatch_slab;
8920 /* free all slabs above current one - this must be the last action
8921 * of this function, as aux and eval_state are allocated within
8922 * slabs and may be freed here */
8924 s = PL_regmatch_slab->next;
8926 PL_regmatch_slab->next = NULL;
8928 regmatch_slab * const osl = s;
8937 S_to_utf8_substr(pTHX_ regexp *prog)
8939 /* Converts substr fields in prog from bytes to UTF-8, calling fbm_compile
8940 * on the converted value */
8944 PERL_ARGS_ASSERT_TO_UTF8_SUBSTR;
8947 if (prog->substrs->data[i].substr
8948 && !prog->substrs->data[i].utf8_substr) {
8949 SV* const sv = newSVsv(prog->substrs->data[i].substr);
8950 prog->substrs->data[i].utf8_substr = sv;
8951 sv_utf8_upgrade(sv);
8952 if (SvVALID(prog->substrs->data[i].substr)) {
8953 if (SvTAIL(prog->substrs->data[i].substr)) {
8954 /* Trim the trailing \n that fbm_compile added last
8956 SvCUR_set(sv, SvCUR(sv) - 1);
8957 /* Whilst this makes the SV technically "invalid" (as its
8958 buffer is no longer followed by "\0") when fbm_compile()
8959 adds the "\n" back, a "\0" is restored. */
8960 fbm_compile(sv, FBMcf_TAIL);
8964 if (prog->substrs->data[i].substr == prog->check_substr)
8965 prog->check_utf8 = sv;
8971 S_to_byte_substr(pTHX_ regexp *prog)
8973 /* Converts substr fields in prog from UTF-8 to bytes, calling fbm_compile
8974 * on the converted value; returns FALSE if can't be converted. */
8978 PERL_ARGS_ASSERT_TO_BYTE_SUBSTR;
8981 if (prog->substrs->data[i].utf8_substr
8982 && !prog->substrs->data[i].substr) {
8983 SV* sv = newSVsv(prog->substrs->data[i].utf8_substr);
8984 if (! sv_utf8_downgrade(sv, TRUE)) {
8987 if (SvVALID(prog->substrs->data[i].utf8_substr)) {
8988 if (SvTAIL(prog->substrs->data[i].utf8_substr)) {
8989 /* Trim the trailing \n that fbm_compile added last
8991 SvCUR_set(sv, SvCUR(sv) - 1);
8992 fbm_compile(sv, FBMcf_TAIL);
8996 prog->substrs->data[i].substr = sv;
8997 if (prog->substrs->data[i].utf8_substr == prog->check_utf8)
8998 prog->check_substr = sv;
9007 * c-indentation-style: bsd
9009 * indent-tabs-mode: nil
9012 * ex: set ts=8 sts=4 sw=4 et:
https://perl5.git.perl.org / perl5.git / blob