5 * "A fair jaw-cracker dwarf-language must be." --Samwise Gamgee
8 /* This file contains functions for compiling a regular expression. See
9 * also regexec.c which funnily enough, contains functions for executing
10 * a regular expression.
12 * This file is also copied at build time to ext/re/re_comp.c, where
13 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
14 * This causes the main functions to be compiled under new names and with
15 * debugging support added, which makes "use re 'debug'" work.
18 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
19 * confused with the original package (see point 3 below). Thanks, Henry!
22 /* Additional note: this code is very heavily munged from Henry's version
23 * in places. In some spots I've traded clarity for efficiency, so don't
24 * blame Henry for some of the lack of readability.
27 /* The names of the functions have been changed from regcomp and
28 * regexec to pregcomp and pregexec in order to avoid conflicts
29 * with the POSIX routines of the same names.
32 #ifdef PERL_EXT_RE_BUILD
37 * pregcomp and pregexec -- regsub and regerror are not used in perl
39 * Copyright (c) 1986 by University of Toronto.
40 * Written by Henry Spencer. Not derived from licensed software.
42 * Permission is granted to anyone to use this software for any
43 * purpose on any computer system, and to redistribute it freely,
44 * subject to the following restrictions:
46 * 1. The author is not responsible for the consequences of use of
47 * this software, no matter how awful, even if they arise
50 * 2. The origin of this software must not be misrepresented, either
51 * by explicit claim or by omission.
53 * 3. Altered versions must be plainly marked as such, and must not
54 * be misrepresented as being the original software.
57 **** Alterations to Henry's code are...
59 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
60 **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007 by Larry Wall and others
62 **** You may distribute under the terms of either the GNU General Public
63 **** License or the Artistic License, as specified in the README file.
66 * Beware that some of this code is subtly aware of the way operator
67 * precedence is structured in regular expressions. Serious changes in
68 * regular-expression syntax might require a total rethink.
71 #define PERL_IN_REGCOMP_C
74 #ifndef PERL_IN_XSUB_RE
79 #ifdef PERL_IN_XSUB_RE
90 # if defined(BUGGY_MSC6)
91 /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
92 # pragma optimize("a",off)
93 /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
94 # pragma optimize("w",on )
95 # endif /* BUGGY_MSC6 */
102 typedef struct RExC_state_t {
103 U32 flags; /* are we folding, multilining? */
104 char *precomp; /* uncompiled string. */
105 regexp *rx; /* perl core regexp structure */
106 regexp_internal *rxi; /* internal data for regexp object pprivate field */
107 char *start; /* Start of input for compile */
108 char *end; /* End of input for compile */
109 char *parse; /* Input-scan pointer. */
110 I32 whilem_seen; /* number of WHILEM in this expr */
111 regnode *emit_start; /* Start of emitted-code area */
112 regnode *emit_bound; /* First regnode outside of the allocated space */
113 regnode *emit; /* Code-emit pointer; ®dummy = don't = compiling */
114 I32 naughty; /* How bad is this pattern? */
115 I32 sawback; /* Did we see \1, ...? */
117 I32 size; /* Code size. */
118 I32 npar; /* Capture buffer count, (OPEN). */
119 I32 cpar; /* Capture buffer count, (CLOSE). */
120 I32 nestroot; /* root parens we are in - used by accept */
124 regnode **open_parens; /* pointers to open parens */
125 regnode **close_parens; /* pointers to close parens */
126 regnode *opend; /* END node in program */
127 I32 utf8; /* whether the pattern is utf8 or not */
128 I32 orig_utf8; /* whether the pattern was originally in utf8 */
129 /* XXX use this for future optimisation of case
130 * where pattern must be upgraded to utf8. */
131 HV *charnames; /* cache of named sequences */
132 HV *paren_names; /* Paren names */
134 regnode **recurse; /* Recurse regops */
135 I32 recurse_count; /* Number of recurse regops */
137 char *starttry; /* -Dr: where regtry was called. */
138 #define RExC_starttry (pRExC_state->starttry)
141 const char *lastparse;
143 AV *paren_name_list; /* idx -> name */
144 #define RExC_lastparse (pRExC_state->lastparse)
145 #define RExC_lastnum (pRExC_state->lastnum)
146 #define RExC_paren_name_list (pRExC_state->paren_name_list)
150 #define RExC_flags (pRExC_state->flags)
151 #define RExC_precomp (pRExC_state->precomp)
152 #define RExC_rx (pRExC_state->rx)
153 #define RExC_rxi (pRExC_state->rxi)
154 #define RExC_start (pRExC_state->start)
155 #define RExC_end (pRExC_state->end)
156 #define RExC_parse (pRExC_state->parse)
157 #define RExC_whilem_seen (pRExC_state->whilem_seen)
158 #ifdef RE_TRACK_PATTERN_OFFSETS
159 #define RExC_offsets (pRExC_state->rxi->u.offsets) /* I am not like the others */
161 #define RExC_emit (pRExC_state->emit)
162 #define RExC_emit_start (pRExC_state->emit_start)
163 #define RExC_emit_bound (pRExC_state->emit_bound)
164 #define RExC_naughty (pRExC_state->naughty)
165 #define RExC_sawback (pRExC_state->sawback)
166 #define RExC_seen (pRExC_state->seen)
167 #define RExC_size (pRExC_state->size)
168 #define RExC_npar (pRExC_state->npar)
169 #define RExC_nestroot (pRExC_state->nestroot)
170 #define RExC_extralen (pRExC_state->extralen)
171 #define RExC_seen_zerolen (pRExC_state->seen_zerolen)
172 #define RExC_seen_evals (pRExC_state->seen_evals)
173 #define RExC_utf8 (pRExC_state->utf8)
174 #define RExC_orig_utf8 (pRExC_state->orig_utf8)
175 #define RExC_charnames (pRExC_state->charnames)
176 #define RExC_open_parens (pRExC_state->open_parens)
177 #define RExC_close_parens (pRExC_state->close_parens)
178 #define RExC_opend (pRExC_state->opend)
179 #define RExC_paren_names (pRExC_state->paren_names)
180 #define RExC_recurse (pRExC_state->recurse)
181 #define RExC_recurse_count (pRExC_state->recurse_count)
184 #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
185 #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
186 ((*s) == '{' && regcurly(s)))
189 #undef SPSTART /* dratted cpp namespace... */
192 * Flags to be passed up and down.
194 #define WORST 0 /* Worst case. */
195 #define HASWIDTH 0x01 /* Known to match non-null strings. */
196 #define SIMPLE 0x02 /* Simple enough to be STAR/PLUS operand. */
197 #define SPSTART 0x04 /* Starts with * or +. */
198 #define TRYAGAIN 0x08 /* Weeded out a declaration. */
199 #define POSTPONED 0x10 /* (?1),(?&name), (??{...}) or similar */
201 #define REG_NODE_NUM(x) ((x) ? (int)((x)-RExC_emit_start) : -1)
203 /* whether trie related optimizations are enabled */
204 #if PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION
205 #define TRIE_STUDY_OPT
206 #define FULL_TRIE_STUDY
212 #define PBYTE(u8str,paren) ((U8*)(u8str))[(paren) >> 3]
213 #define PBITVAL(paren) (1 << ((paren) & 7))
214 #define PAREN_TEST(u8str,paren) ( PBYTE(u8str,paren) & PBITVAL(paren))
215 #define PAREN_SET(u8str,paren) PBYTE(u8str,paren) |= PBITVAL(paren)
216 #define PAREN_UNSET(u8str,paren) PBYTE(u8str,paren) &= (~PBITVAL(paren))
219 /* About scan_data_t.
221 During optimisation we recurse through the regexp program performing
222 various inplace (keyhole style) optimisations. In addition study_chunk
223 and scan_commit populate this data structure with information about
224 what strings MUST appear in the pattern. We look for the longest
225 string that must appear for at a fixed location, and we look for the
226 longest string that may appear at a floating location. So for instance
231 Both 'FOO' and 'A' are fixed strings. Both 'B' and 'BAR' are floating
232 strings (because they follow a .* construct). study_chunk will identify
233 both FOO and BAR as being the longest fixed and floating strings respectively.
235 The strings can be composites, for instance
239 will result in a composite fixed substring 'foo'.
241 For each string some basic information is maintained:
243 - offset or min_offset
244 This is the position the string must appear at, or not before.
245 It also implicitly (when combined with minlenp) tells us how many
246 character must match before the string we are searching.
247 Likewise when combined with minlenp and the length of the string
248 tells us how many characters must appear after the string we have
252 Only used for floating strings. This is the rightmost point that
253 the string can appear at. Ifset to I32 max it indicates that the
254 string can occur infinitely far to the right.
257 A pointer to the minimum length of the pattern that the string
258 was found inside. This is important as in the case of positive
259 lookahead or positive lookbehind we can have multiple patterns
264 The minimum length of the pattern overall is 3, the minimum length
265 of the lookahead part is 3, but the minimum length of the part that
266 will actually match is 1. So 'FOO's minimum length is 3, but the
267 minimum length for the F is 1. This is important as the minimum length
268 is used to determine offsets in front of and behind the string being
269 looked for. Since strings can be composites this is the length of the
270 pattern at the time it was commited with a scan_commit. Note that
271 the length is calculated by study_chunk, so that the minimum lengths
272 are not known until the full pattern has been compiled, thus the
273 pointer to the value.
277 In the case of lookbehind the string being searched for can be
278 offset past the start point of the final matching string.
279 If this value was just blithely removed from the min_offset it would
280 invalidate some of the calculations for how many chars must match
281 before or after (as they are derived from min_offset and minlen and
282 the length of the string being searched for).
283 When the final pattern is compiled and the data is moved from the
284 scan_data_t structure into the regexp structure the information
285 about lookbehind is factored in, with the information that would
286 have been lost precalculated in the end_shift field for the
289 The fields pos_min and pos_delta are used to store the minimum offset
290 and the delta to the maximum offset at the current point in the pattern.
294 typedef struct scan_data_t {
295 /*I32 len_min; unused */
296 /*I32 len_delta; unused */
300 I32 last_end; /* min value, <0 unless valid. */
303 SV **longest; /* Either &l_fixed, or &l_float. */
304 SV *longest_fixed; /* longest fixed string found in pattern */
305 I32 offset_fixed; /* offset where it starts */
306 I32 *minlen_fixed; /* pointer to the minlen relevent to the string */
307 I32 lookbehind_fixed; /* is the position of the string modfied by LB */
308 SV *longest_float; /* longest floating string found in pattern */
309 I32 offset_float_min; /* earliest point in string it can appear */
310 I32 offset_float_max; /* latest point in string it can appear */
311 I32 *minlen_float; /* pointer to the minlen relevent to the string */
312 I32 lookbehind_float; /* is the position of the string modified by LB */
316 struct regnode_charclass_class *start_class;
320 * Forward declarations for pregcomp()'s friends.
323 static const scan_data_t zero_scan_data =
324 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ,0};
326 #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
327 #define SF_BEFORE_SEOL 0x0001
328 #define SF_BEFORE_MEOL 0x0002
329 #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
330 #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
333 # define SF_FIX_SHIFT_EOL (0+2)
334 # define SF_FL_SHIFT_EOL (0+4)
336 # define SF_FIX_SHIFT_EOL (+2)
337 # define SF_FL_SHIFT_EOL (+4)
340 #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
341 #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
343 #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
344 #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
345 #define SF_IS_INF 0x0040
346 #define SF_HAS_PAR 0x0080
347 #define SF_IN_PAR 0x0100
348 #define SF_HAS_EVAL 0x0200
349 #define SCF_DO_SUBSTR 0x0400
350 #define SCF_DO_STCLASS_AND 0x0800
351 #define SCF_DO_STCLASS_OR 0x1000
352 #define SCF_DO_STCLASS (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
353 #define SCF_WHILEM_VISITED_POS 0x2000
355 #define SCF_TRIE_RESTUDY 0x4000 /* Do restudy? */
356 #define SCF_SEEN_ACCEPT 0x8000
358 #define UTF (RExC_utf8 != 0)
359 #define LOC ((RExC_flags & RXf_PMf_LOCALE) != 0)
360 #define FOLD ((RExC_flags & RXf_PMf_FOLD) != 0)
362 #define OOB_UNICODE 12345678
363 #define OOB_NAMEDCLASS -1
365 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
366 #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
369 /* length of regex to show in messages that don't mark a position within */
370 #define RegexLengthToShowInErrorMessages 127
373 * If MARKER[12] are adjusted, be sure to adjust the constants at the top
374 * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
375 * op/pragma/warn/regcomp.
377 #define MARKER1 "<-- HERE" /* marker as it appears in the description */
378 #define MARKER2 " <-- HERE " /* marker as it appears within the regex */
380 #define REPORT_LOCATION " in regex; marked by " MARKER1 " in m/%.*s" MARKER2 "%s/"
383 * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
384 * arg. Show regex, up to a maximum length. If it's too long, chop and add
387 #define _FAIL(code) STMT_START { \
388 const char *ellipses = ""; \
389 IV len = RExC_end - RExC_precomp; \
392 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
393 if (len > RegexLengthToShowInErrorMessages) { \
394 /* chop 10 shorter than the max, to ensure meaning of "..." */ \
395 len = RegexLengthToShowInErrorMessages - 10; \
401 #define FAIL(msg) _FAIL( \
402 Perl_croak(aTHX_ "%s in regex m/%.*s%s/", \
403 msg, (int)len, RExC_precomp, ellipses))
405 #define FAIL2(msg,arg) _FAIL( \
406 Perl_croak(aTHX_ msg " in regex m/%.*s%s/", \
407 arg, (int)len, RExC_precomp, ellipses))
410 * Simple_vFAIL -- like FAIL, but marks the current location in the scan
412 #define Simple_vFAIL(m) STMT_START { \
413 const IV offset = RExC_parse - RExC_precomp; \
414 Perl_croak(aTHX_ "%s" REPORT_LOCATION, \
415 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
419 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
421 #define vFAIL(m) STMT_START { \
423 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
428 * Like Simple_vFAIL(), but accepts two arguments.
430 #define Simple_vFAIL2(m,a1) STMT_START { \
431 const IV offset = RExC_parse - RExC_precomp; \
432 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, \
433 (int)offset, RExC_precomp, RExC_precomp + offset); \
437 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
439 #define vFAIL2(m,a1) STMT_START { \
441 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
442 Simple_vFAIL2(m, a1); \
447 * Like Simple_vFAIL(), but accepts three arguments.
449 #define Simple_vFAIL3(m, a1, a2) STMT_START { \
450 const IV offset = RExC_parse - RExC_precomp; \
451 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, \
452 (int)offset, RExC_precomp, RExC_precomp + offset); \
456 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
458 #define vFAIL3(m,a1,a2) STMT_START { \
460 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
461 Simple_vFAIL3(m, a1, a2); \
465 * Like Simple_vFAIL(), but accepts four arguments.
467 #define Simple_vFAIL4(m, a1, a2, a3) STMT_START { \
468 const IV offset = RExC_parse - RExC_precomp; \
469 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3, \
470 (int)offset, RExC_precomp, RExC_precomp + offset); \
473 #define vWARN(loc,m) STMT_START { \
474 const IV offset = loc - RExC_precomp; \
475 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s" REPORT_LOCATION, \
476 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
479 #define vWARNdep(loc,m) STMT_START { \
480 const IV offset = loc - RExC_precomp; \
481 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_REGEXP), \
482 "%s" REPORT_LOCATION, \
483 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
487 #define vWARN2(loc, m, a1) STMT_START { \
488 const IV offset = loc - RExC_precomp; \
489 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
490 a1, (int)offset, RExC_precomp, RExC_precomp + offset); \
493 #define vWARN3(loc, m, a1, a2) STMT_START { \
494 const IV offset = loc - RExC_precomp; \
495 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
496 a1, a2, (int)offset, RExC_precomp, RExC_precomp + offset); \
499 #define vWARN4(loc, m, a1, a2, a3) STMT_START { \
500 const IV offset = loc - RExC_precomp; \
501 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
502 a1, a2, a3, (int)offset, RExC_precomp, RExC_precomp + offset); \
505 #define vWARN5(loc, m, a1, a2, a3, a4) STMT_START { \
506 const IV offset = loc - RExC_precomp; \
507 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
508 a1, a2, a3, a4, (int)offset, RExC_precomp, RExC_precomp + offset); \
512 /* Allow for side effects in s */
513 #define REGC(c,s) STMT_START { \
514 if (!SIZE_ONLY) *(s) = (c); else (void)(s); \
517 /* Macros for recording node offsets. 20001227 mjd@plover.com
518 * Nodes are numbered 1, 2, 3, 4. Node #n's position is recorded in
519 * element 2*n-1 of the array. Element #2n holds the byte length node #n.
520 * Element 0 holds the number n.
521 * Position is 1 indexed.
523 #ifndef RE_TRACK_PATTERN_OFFSETS
524 #define Set_Node_Offset_To_R(node,byte)
525 #define Set_Node_Offset(node,byte)
526 #define Set_Cur_Node_Offset
527 #define Set_Node_Length_To_R(node,len)
528 #define Set_Node_Length(node,len)
529 #define Set_Node_Cur_Length(node)
530 #define Node_Offset(n)
531 #define Node_Length(n)
532 #define Set_Node_Offset_Length(node,offset,len)
533 #define ProgLen(ri) ri->u.proglen
534 #define SetProgLen(ri,x) ri->u.proglen = x
536 #define ProgLen(ri) ri->u.offsets[0]
537 #define SetProgLen(ri,x) ri->u.offsets[0] = x
538 #define Set_Node_Offset_To_R(node,byte) STMT_START { \
540 MJD_OFFSET_DEBUG(("** (%d) offset of node %d is %d.\n", \
541 __LINE__, (int)(node), (int)(byte))); \
543 Perl_croak(aTHX_ "value of node is %d in Offset macro", (int)(node)); \
545 RExC_offsets[2*(node)-1] = (byte); \
550 #define Set_Node_Offset(node,byte) \
551 Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
552 #define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
554 #define Set_Node_Length_To_R(node,len) STMT_START { \
556 MJD_OFFSET_DEBUG(("** (%d) size of node %d is %d.\n", \
557 __LINE__, (int)(node), (int)(len))); \
559 Perl_croak(aTHX_ "value of node is %d in Length macro", (int)(node)); \
561 RExC_offsets[2*(node)] = (len); \
566 #define Set_Node_Length(node,len) \
567 Set_Node_Length_To_R((node)-RExC_emit_start, len)
568 #define Set_Cur_Node_Length(len) Set_Node_Length(RExC_emit, len)
569 #define Set_Node_Cur_Length(node) \
570 Set_Node_Length(node, RExC_parse - parse_start)
572 /* Get offsets and lengths */
573 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
574 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
576 #define Set_Node_Offset_Length(node,offset,len) STMT_START { \
577 Set_Node_Offset_To_R((node)-RExC_emit_start, (offset)); \
578 Set_Node_Length_To_R((node)-RExC_emit_start, (len)); \
582 #if PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS
583 #define EXPERIMENTAL_INPLACESCAN
584 #endif /*RE_TRACK_PATTERN_OFFSETS*/
586 #define DEBUG_STUDYDATA(str,data,depth) \
587 DEBUG_OPTIMISE_MORE_r(if(data){ \
588 PerlIO_printf(Perl_debug_log, \
589 "%*s" str "Pos:%"IVdf"/%"IVdf \
590 " Flags: 0x%"UVXf" Whilem_c: %"IVdf" Lcp: %"IVdf" %s", \
591 (int)(depth)*2, "", \
592 (IV)((data)->pos_min), \
593 (IV)((data)->pos_delta), \
594 (UV)((data)->flags), \
595 (IV)((data)->whilem_c), \
596 (IV)((data)->last_closep ? *((data)->last_closep) : -1), \
597 is_inf ? "INF " : "" \
599 if ((data)->last_found) \
600 PerlIO_printf(Perl_debug_log, \
601 "Last:'%s' %"IVdf":%"IVdf"/%"IVdf" %sFixed:'%s' @ %"IVdf \
602 " %sFloat: '%s' @ %"IVdf"/%"IVdf"", \
603 SvPVX_const((data)->last_found), \
604 (IV)((data)->last_end), \
605 (IV)((data)->last_start_min), \
606 (IV)((data)->last_start_max), \
607 ((data)->longest && \
608 (data)->longest==&((data)->longest_fixed)) ? "*" : "", \
609 SvPVX_const((data)->longest_fixed), \
610 (IV)((data)->offset_fixed), \
611 ((data)->longest && \
612 (data)->longest==&((data)->longest_float)) ? "*" : "", \
613 SvPVX_const((data)->longest_float), \
614 (IV)((data)->offset_float_min), \
615 (IV)((data)->offset_float_max) \
617 PerlIO_printf(Perl_debug_log,"\n"); \
620 static void clear_re(pTHX_ void *r);
622 /* Mark that we cannot extend a found fixed substring at this point.
623 Update the longest found anchored substring and the longest found
624 floating substrings if needed. */
627 S_scan_commit(pTHX_ const RExC_state_t *pRExC_state, scan_data_t *data, I32 *minlenp, int is_inf)
629 const STRLEN l = CHR_SVLEN(data->last_found);
630 const STRLEN old_l = CHR_SVLEN(*data->longest);
631 GET_RE_DEBUG_FLAGS_DECL;
633 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
634 SvSetMagicSV(*data->longest, data->last_found);
635 if (*data->longest == data->longest_fixed) {
636 data->offset_fixed = l ? data->last_start_min : data->pos_min;
637 if (data->flags & SF_BEFORE_EOL)
639 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
641 data->flags &= ~SF_FIX_BEFORE_EOL;
642 data->minlen_fixed=minlenp;
643 data->lookbehind_fixed=0;
645 else { /* *data->longest == data->longest_float */
646 data->offset_float_min = l ? data->last_start_min : data->pos_min;
647 data->offset_float_max = (l
648 ? data->last_start_max
649 : data->pos_min + data->pos_delta);
650 if (is_inf || (U32)data->offset_float_max > (U32)I32_MAX)
651 data->offset_float_max = I32_MAX;
652 if (data->flags & SF_BEFORE_EOL)
654 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
656 data->flags &= ~SF_FL_BEFORE_EOL;
657 data->minlen_float=minlenp;
658 data->lookbehind_float=0;
661 SvCUR_set(data->last_found, 0);
663 SV * const sv = data->last_found;
664 if (SvUTF8(sv) && SvMAGICAL(sv)) {
665 MAGIC * const mg = mg_find(sv, PERL_MAGIC_utf8);
671 data->flags &= ~SF_BEFORE_EOL;
672 DEBUG_STUDYDATA("commit: ",data,0);
675 /* Can match anything (initialization) */
677 S_cl_anything(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
679 ANYOF_CLASS_ZERO(cl);
680 ANYOF_BITMAP_SETALL(cl);
681 cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
683 cl->flags |= ANYOF_LOCALE;
686 /* Can match anything (initialization) */
688 S_cl_is_anything(const struct regnode_charclass_class *cl)
692 for (value = 0; value <= ANYOF_MAX; value += 2)
693 if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
695 if (!(cl->flags & ANYOF_UNICODE_ALL))
697 if (!ANYOF_BITMAP_TESTALLSET((const void*)cl))
702 /* Can match anything (initialization) */
704 S_cl_init(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
706 Zero(cl, 1, struct regnode_charclass_class);
708 cl_anything(pRExC_state, cl);
712 S_cl_init_zero(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
714 Zero(cl, 1, struct regnode_charclass_class);
716 cl_anything(pRExC_state, cl);
718 cl->flags |= ANYOF_LOCALE;
721 /* 'And' a given class with another one. Can create false positives */
722 /* We assume that cl is not inverted */
724 S_cl_and(struct regnode_charclass_class *cl,
725 const struct regnode_charclass_class *and_with)
728 assert(and_with->type == ANYOF);
729 if (!(and_with->flags & ANYOF_CLASS)
730 && !(cl->flags & ANYOF_CLASS)
731 && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
732 && !(and_with->flags & ANYOF_FOLD)
733 && !(cl->flags & ANYOF_FOLD)) {
736 if (and_with->flags & ANYOF_INVERT)
737 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
738 cl->bitmap[i] &= ~and_with->bitmap[i];
740 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
741 cl->bitmap[i] &= and_with->bitmap[i];
742 } /* XXXX: logic is complicated otherwise, leave it along for a moment. */
743 if (!(and_with->flags & ANYOF_EOS))
744 cl->flags &= ~ANYOF_EOS;
746 if (cl->flags & ANYOF_UNICODE_ALL && and_with->flags & ANYOF_UNICODE &&
747 !(and_with->flags & ANYOF_INVERT)) {
748 cl->flags &= ~ANYOF_UNICODE_ALL;
749 cl->flags |= ANYOF_UNICODE;
750 ARG_SET(cl, ARG(and_with));
752 if (!(and_with->flags & ANYOF_UNICODE_ALL) &&
753 !(and_with->flags & ANYOF_INVERT))
754 cl->flags &= ~ANYOF_UNICODE_ALL;
755 if (!(and_with->flags & (ANYOF_UNICODE|ANYOF_UNICODE_ALL)) &&
756 !(and_with->flags & ANYOF_INVERT))
757 cl->flags &= ~ANYOF_UNICODE;
760 /* 'OR' a given class with another one. Can create false positives */
761 /* We assume that cl is not inverted */
763 S_cl_or(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, const struct regnode_charclass_class *or_with)
765 if (or_with->flags & ANYOF_INVERT) {
767 * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
768 * <= (B1 | !B2) | (CL1 | !CL2)
769 * which is wasteful if CL2 is small, but we ignore CL2:
770 * (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1
771 * XXXX Can we handle case-fold? Unclear:
772 * (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) =
773 * (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i'))
775 if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
776 && !(or_with->flags & ANYOF_FOLD)
777 && !(cl->flags & ANYOF_FOLD) ) {
780 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
781 cl->bitmap[i] |= ~or_with->bitmap[i];
782 } /* XXXX: logic is complicated otherwise */
784 cl_anything(pRExC_state, cl);
787 /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */
788 if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
789 && (!(or_with->flags & ANYOF_FOLD)
790 || (cl->flags & ANYOF_FOLD)) ) {
793 /* OR char bitmap and class bitmap separately */
794 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
795 cl->bitmap[i] |= or_with->bitmap[i];
796 if (or_with->flags & ANYOF_CLASS) {
797 for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++)
798 cl->classflags[i] |= or_with->classflags[i];
799 cl->flags |= ANYOF_CLASS;
802 else { /* XXXX: logic is complicated, leave it along for a moment. */
803 cl_anything(pRExC_state, cl);
806 if (or_with->flags & ANYOF_EOS)
807 cl->flags |= ANYOF_EOS;
809 if (cl->flags & ANYOF_UNICODE && or_with->flags & ANYOF_UNICODE &&
810 ARG(cl) != ARG(or_with)) {
811 cl->flags |= ANYOF_UNICODE_ALL;
812 cl->flags &= ~ANYOF_UNICODE;
814 if (or_with->flags & ANYOF_UNICODE_ALL) {
815 cl->flags |= ANYOF_UNICODE_ALL;
816 cl->flags &= ~ANYOF_UNICODE;
820 #define TRIE_LIST_ITEM(state,idx) (trie->states[state].trans.list)[ idx ]
821 #define TRIE_LIST_CUR(state) ( TRIE_LIST_ITEM( state, 0 ).forid )
822 #define TRIE_LIST_LEN(state) ( TRIE_LIST_ITEM( state, 0 ).newstate )
823 #define TRIE_LIST_USED(idx) ( trie->states[state].trans.list ? (TRIE_LIST_CUR( idx ) - 1) : 0 )
828 dump_trie(trie,widecharmap,revcharmap)
829 dump_trie_interim_list(trie,widecharmap,revcharmap,next_alloc)
830 dump_trie_interim_table(trie,widecharmap,revcharmap,next_alloc)
832 These routines dump out a trie in a somewhat readable format.
833 The _interim_ variants are used for debugging the interim
834 tables that are used to generate the final compressed
835 representation which is what dump_trie expects.
837 Part of the reason for their existance is to provide a form
838 of documentation as to how the different representations function.
843 Dumps the final compressed table form of the trie to Perl_debug_log.
844 Used for debugging make_trie().
848 S_dump_trie(pTHX_ const struct _reg_trie_data *trie, HV *widecharmap,
849 AV *revcharmap, U32 depth)
852 SV *sv=sv_newmortal();
853 int colwidth= widecharmap ? 6 : 4;
854 GET_RE_DEBUG_FLAGS_DECL;
857 PerlIO_printf( Perl_debug_log, "%*sChar : %-6s%-6s%-4s ",
858 (int)depth * 2 + 2,"",
859 "Match","Base","Ofs" );
861 for( state = 0 ; state < trie->uniquecharcount ; state++ ) {
862 SV ** const tmp = av_fetch( revcharmap, state, 0);
864 PerlIO_printf( Perl_debug_log, "%*s",
866 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
867 PL_colors[0], PL_colors[1],
868 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
869 PERL_PV_ESCAPE_FIRSTCHAR
874 PerlIO_printf( Perl_debug_log, "\n%*sState|-----------------------",
875 (int)depth * 2 + 2,"");
877 for( state = 0 ; state < trie->uniquecharcount ; state++ )
878 PerlIO_printf( Perl_debug_log, "%.*s", colwidth, "--------");
879 PerlIO_printf( Perl_debug_log, "\n");
881 for( state = 1 ; state < trie->statecount ; state++ ) {
882 const U32 base = trie->states[ state ].trans.base;
884 PerlIO_printf( Perl_debug_log, "%*s#%4"UVXf"|", (int)depth * 2 + 2,"", (UV)state);
886 if ( trie->states[ state ].wordnum ) {
887 PerlIO_printf( Perl_debug_log, " W%4X", trie->states[ state ].wordnum );
889 PerlIO_printf( Perl_debug_log, "%6s", "" );
892 PerlIO_printf( Perl_debug_log, " @%4"UVXf" ", (UV)base );
897 while( ( base + ofs < trie->uniquecharcount ) ||
898 ( base + ofs - trie->uniquecharcount < trie->lasttrans
899 && trie->trans[ base + ofs - trie->uniquecharcount ].check != state))
902 PerlIO_printf( Perl_debug_log, "+%2"UVXf"[ ", (UV)ofs);
904 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
905 if ( ( base + ofs >= trie->uniquecharcount ) &&
906 ( base + ofs - trie->uniquecharcount < trie->lasttrans ) &&
907 trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
909 PerlIO_printf( Perl_debug_log, "%*"UVXf,
911 (UV)trie->trans[ base + ofs - trie->uniquecharcount ].next );
913 PerlIO_printf( Perl_debug_log, "%*s",colwidth," ." );
917 PerlIO_printf( Perl_debug_log, "]");
920 PerlIO_printf( Perl_debug_log, "\n" );
924 Dumps a fully constructed but uncompressed trie in list form.
925 List tries normally only are used for construction when the number of
926 possible chars (trie->uniquecharcount) is very high.
927 Used for debugging make_trie().
930 S_dump_trie_interim_list(pTHX_ const struct _reg_trie_data *trie,
931 HV *widecharmap, AV *revcharmap, U32 next_alloc,
935 SV *sv=sv_newmortal();
936 int colwidth= widecharmap ? 6 : 4;
937 GET_RE_DEBUG_FLAGS_DECL;
938 /* print out the table precompression. */
939 PerlIO_printf( Perl_debug_log, "%*sState :Word | Transition Data\n%*s%s",
940 (int)depth * 2 + 2,"", (int)depth * 2 + 2,"",
941 "------:-----+-----------------\n" );
943 for( state=1 ; state < next_alloc ; state ++ ) {
946 PerlIO_printf( Perl_debug_log, "%*s %4"UVXf" :",
947 (int)depth * 2 + 2,"", (UV)state );
948 if ( ! trie->states[ state ].wordnum ) {
949 PerlIO_printf( Perl_debug_log, "%5s| ","");
951 PerlIO_printf( Perl_debug_log, "W%4x| ",
952 trie->states[ state ].wordnum
955 for( charid = 1 ; charid <= TRIE_LIST_USED( state ) ; charid++ ) {
956 SV ** const tmp = av_fetch( revcharmap, TRIE_LIST_ITEM(state,charid).forid, 0);
958 PerlIO_printf( Perl_debug_log, "%*s:%3X=%4"UVXf" | ",
960 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
961 PL_colors[0], PL_colors[1],
962 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
963 PERL_PV_ESCAPE_FIRSTCHAR
965 TRIE_LIST_ITEM(state,charid).forid,
966 (UV)TRIE_LIST_ITEM(state,charid).newstate
969 PerlIO_printf(Perl_debug_log, "\n%*s| ",
970 (int)((depth * 2) + 14), "");
973 PerlIO_printf( Perl_debug_log, "\n");
978 Dumps a fully constructed but uncompressed trie in table form.
979 This is the normal DFA style state transition table, with a few
980 twists to facilitate compression later.
981 Used for debugging make_trie().
984 S_dump_trie_interim_table(pTHX_ const struct _reg_trie_data *trie,
985 HV *widecharmap, AV *revcharmap, U32 next_alloc,
990 SV *sv=sv_newmortal();
991 int colwidth= widecharmap ? 6 : 4;
992 GET_RE_DEBUG_FLAGS_DECL;
995 print out the table precompression so that we can do a visual check
996 that they are identical.
999 PerlIO_printf( Perl_debug_log, "%*sChar : ",(int)depth * 2 + 2,"" );
1001 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
1002 SV ** const tmp = av_fetch( revcharmap, charid, 0);
1004 PerlIO_printf( Perl_debug_log, "%*s",
1006 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
1007 PL_colors[0], PL_colors[1],
1008 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
1009 PERL_PV_ESCAPE_FIRSTCHAR
1015 PerlIO_printf( Perl_debug_log, "\n%*sState+-",(int)depth * 2 + 2,"" );
1017 for( charid=0 ; charid < trie->uniquecharcount ; charid++ ) {
1018 PerlIO_printf( Perl_debug_log, "%.*s", colwidth,"--------");
1021 PerlIO_printf( Perl_debug_log, "\n" );
1023 for( state=1 ; state < next_alloc ; state += trie->uniquecharcount ) {
1025 PerlIO_printf( Perl_debug_log, "%*s%4"UVXf" : ",
1026 (int)depth * 2 + 2,"",
1027 (UV)TRIE_NODENUM( state ) );
1029 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
1030 UV v=(UV)SAFE_TRIE_NODENUM( trie->trans[ state + charid ].next );
1032 PerlIO_printf( Perl_debug_log, "%*"UVXf, colwidth, v );
1034 PerlIO_printf( Perl_debug_log, "%*s", colwidth, "." );
1036 if ( ! trie->states[ TRIE_NODENUM( state ) ].wordnum ) {
1037 PerlIO_printf( Perl_debug_log, " (%4"UVXf")\n", (UV)trie->trans[ state ].check );
1039 PerlIO_printf( Perl_debug_log, " (%4"UVXf") W%4X\n", (UV)trie->trans[ state ].check,
1040 trie->states[ TRIE_NODENUM( state ) ].wordnum );
1047 /* make_trie(startbranch,first,last,tail,word_count,flags,depth)
1048 startbranch: the first branch in the whole branch sequence
1049 first : start branch of sequence of branch-exact nodes.
1050 May be the same as startbranch
1051 last : Thing following the last branch.
1052 May be the same as tail.
1053 tail : item following the branch sequence
1054 count : words in the sequence
1055 flags : currently the OP() type we will be building one of /EXACT(|F|Fl)/
1056 depth : indent depth
1058 Inplace optimizes a sequence of 2 or more Branch-Exact nodes into a TRIE node.
1060 A trie is an N'ary tree where the branches are determined by digital
1061 decomposition of the key. IE, at the root node you look up the 1st character and
1062 follow that branch repeat until you find the end of the branches. Nodes can be
1063 marked as "accepting" meaning they represent a complete word. Eg:
1067 would convert into the following structure. Numbers represent states, letters
1068 following numbers represent valid transitions on the letter from that state, if
1069 the number is in square brackets it represents an accepting state, otherwise it
1070 will be in parenthesis.
1072 +-h->+-e->[3]-+-r->(8)-+-s->[9]
1076 (1) +-i->(6)-+-s->[7]
1078 +-s->(3)-+-h->(4)-+-e->[5]
1080 Accept Word Mapping: 3=>1 (he),5=>2 (she), 7=>3 (his), 9=>4 (hers)
1082 This shows that when matching against the string 'hers' we will begin at state 1
1083 read 'h' and move to state 2, read 'e' and move to state 3 which is accepting,
1084 then read 'r' and go to state 8 followed by 's' which takes us to state 9 which
1085 is also accepting. Thus we know that we can match both 'he' and 'hers' with a
1086 single traverse. We store a mapping from accepting to state to which word was
1087 matched, and then when we have multiple possibilities we try to complete the
1088 rest of the regex in the order in which they occured in the alternation.
1090 The only prior NFA like behaviour that would be changed by the TRIE support is
1091 the silent ignoring of duplicate alternations which are of the form:
1093 / (DUPE|DUPE) X? (?{ ... }) Y /x
1095 Thus EVAL blocks follwing a trie may be called a different number of times with
1096 and without the optimisation. With the optimisations dupes will be silently
1097 ignored. This inconsistant behaviour of EVAL type nodes is well established as
1098 the following demonstrates:
1100 'words'=~/(word|word|word)(?{ print $1 })[xyz]/
1102 which prints out 'word' three times, but
1104 'words'=~/(word|word|word)(?{ print $1 })S/
1106 which doesnt print it out at all. This is due to other optimisations kicking in.
1108 Example of what happens on a structural level:
1110 The regexp /(ac|ad|ab)+/ will produce the folowing debug output:
1112 1: CURLYM[1] {1,32767}(18)
1123 This would be optimizable with startbranch=5, first=5, last=16, tail=16
1124 and should turn into:
1126 1: CURLYM[1] {1,32767}(18)
1128 [Words:3 Chars Stored:6 Unique Chars:4 States:5 NCP:1]
1136 Cases where tail != last would be like /(?foo|bar)baz/:
1146 which would be optimizable with startbranch=1, first=1, last=7, tail=8
1147 and would end up looking like:
1150 [Words:2 Chars Stored:6 Unique Chars:5 States:7 NCP:1]
1157 d = uvuni_to_utf8_flags(d, uv, 0);
1159 is the recommended Unicode-aware way of saying
1164 #define TRIE_STORE_REVCHAR \
1167 SV *zlopp = newSV(2); \
1168 unsigned char *flrbbbbb = (unsigned char *) SvPVX(zlopp); \
1169 unsigned const char *const kapow = uvuni_to_utf8(flrbbbbb, uvc & 0xFF); \
1170 SvCUR_set(zlopp, kapow - flrbbbbb); \
1173 av_push(revcharmap, zlopp); \
1175 char ooooff = (char)uvc; \
1176 av_push(revcharmap, newSVpvn(&ooooff, 1)); \
1180 #define TRIE_READ_CHAR STMT_START { \
1184 if ( foldlen > 0 ) { \
1185 uvc = utf8n_to_uvuni( scan, UTF8_MAXLEN, &len, uniflags ); \
1190 uvc = utf8n_to_uvuni( (const U8*)uc, UTF8_MAXLEN, &len, uniflags);\
1191 uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \
1192 foldlen -= UNISKIP( uvc ); \
1193 scan = foldbuf + UNISKIP( uvc ); \
1196 uvc = utf8n_to_uvuni( (const U8*)uc, UTF8_MAXLEN, &len, uniflags);\
1206 #define TRIE_LIST_PUSH(state,fid,ns) STMT_START { \
1207 if ( TRIE_LIST_CUR( state ) >=TRIE_LIST_LEN( state ) ) { \
1208 U32 ging = TRIE_LIST_LEN( state ) *= 2; \
1209 Renew( trie->states[ state ].trans.list, ging, reg_trie_trans_le ); \
1211 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).forid = fid; \
1212 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).newstate = ns; \
1213 TRIE_LIST_CUR( state )++; \
1216 #define TRIE_LIST_NEW(state) STMT_START { \
1217 Newxz( trie->states[ state ].trans.list, \
1218 4, reg_trie_trans_le ); \
1219 TRIE_LIST_CUR( state ) = 1; \
1220 TRIE_LIST_LEN( state ) = 4; \
1223 #define TRIE_HANDLE_WORD(state) STMT_START { \
1224 U16 dupe= trie->states[ state ].wordnum; \
1225 regnode * const noper_next = regnext( noper ); \
1227 if (trie->wordlen) \
1228 trie->wordlen[ curword ] = wordlen; \
1230 /* store the word for dumping */ \
1232 if (OP(noper) != NOTHING) \
1233 tmp = newSVpvn(STRING(noper), STR_LEN(noper)); \
1235 tmp = newSVpvn( "", 0 ); \
1236 if ( UTF ) SvUTF8_on( tmp ); \
1237 av_push( trie_words, tmp ); \
1242 if ( noper_next < tail ) { \
1244 trie->jump = (U16 *) PerlMemShared_calloc( word_count + 1, sizeof(U16) ); \
1245 trie->jump[curword] = (U16)(noper_next - convert); \
1247 jumper = noper_next; \
1249 nextbranch= regnext(cur); \
1253 /* So it's a dupe. This means we need to maintain a */\
1254 /* linked-list from the first to the next. */\
1255 /* we only allocate the nextword buffer when there */\
1256 /* a dupe, so first time we have to do the allocation */\
1257 if (!trie->nextword) \
1258 trie->nextword = (U16 *) \
1259 PerlMemShared_calloc( word_count + 1, sizeof(U16)); \
1260 while ( trie->nextword[dupe] ) \
1261 dupe= trie->nextword[dupe]; \
1262 trie->nextword[dupe]= curword; \
1264 /* we haven't inserted this word yet. */ \
1265 trie->states[ state ].wordnum = curword; \
1270 #define TRIE_TRANS_STATE(state,base,ucharcount,charid,special) \
1271 ( ( base + charid >= ucharcount \
1272 && base + charid < ubound \
1273 && state == trie->trans[ base - ucharcount + charid ].check \
1274 && trie->trans[ base - ucharcount + charid ].next ) \
1275 ? trie->trans[ base - ucharcount + charid ].next \
1276 : ( state==1 ? special : 0 ) \
1280 #define MADE_JUMP_TRIE 2
1281 #define MADE_EXACT_TRIE 4
1284 S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *first, regnode *last, regnode *tail, U32 word_count, U32 flags, U32 depth)
1287 /* first pass, loop through and scan words */
1288 reg_trie_data *trie;
1289 HV *widecharmap = NULL;
1290 AV *revcharmap = newAV();
1292 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1297 regnode *jumper = NULL;
1298 regnode *nextbranch = NULL;
1299 regnode *convert = NULL;
1300 /* we just use folder as a flag in utf8 */
1301 const U8 * const folder = ( flags == EXACTF
1303 : ( flags == EXACTFL
1310 const U32 data_slot = add_data( pRExC_state, 4, "tuuu" );
1311 AV *trie_words = NULL;
1312 /* along with revcharmap, this only used during construction but both are
1313 * useful during debugging so we store them in the struct when debugging.
1316 const U32 data_slot = add_data( pRExC_state, 2, "tu" );
1317 STRLEN trie_charcount=0;
1319 SV *re_trie_maxbuff;
1320 GET_RE_DEBUG_FLAGS_DECL;
1322 PERL_UNUSED_ARG(depth);
1325 trie = (reg_trie_data *) PerlMemShared_calloc( 1, sizeof(reg_trie_data) );
1327 trie->startstate = 1;
1328 trie->wordcount = word_count;
1329 RExC_rxi->data->data[ data_slot ] = (void*)trie;
1330 trie->charmap = (U16 *) PerlMemShared_calloc( 256, sizeof(U16) );
1331 if (!(UTF && folder))
1332 trie->bitmap = (char *) PerlMemShared_calloc( ANYOF_BITMAP_SIZE, 1 );
1334 trie_words = newAV();
1337 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
1338 if (!SvIOK(re_trie_maxbuff)) {
1339 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
1342 PerlIO_printf( Perl_debug_log,
1343 "%*smake_trie start==%d, first==%d, last==%d, tail==%d depth=%d\n",
1344 (int)depth * 2 + 2, "",
1345 REG_NODE_NUM(startbranch),REG_NODE_NUM(first),
1346 REG_NODE_NUM(last), REG_NODE_NUM(tail),
1350 /* Find the node we are going to overwrite */
1351 if ( first == startbranch && OP( last ) != BRANCH ) {
1352 /* whole branch chain */
1355 /* branch sub-chain */
1356 convert = NEXTOPER( first );
1359 /* -- First loop and Setup --
1361 We first traverse the branches and scan each word to determine if it
1362 contains widechars, and how many unique chars there are, this is
1363 important as we have to build a table with at least as many columns as we
1366 We use an array of integers to represent the character codes 0..255
1367 (trie->charmap) and we use a an HV* to store Unicode characters. We use the
1368 native representation of the character value as the key and IV's for the
1371 *TODO* If we keep track of how many times each character is used we can
1372 remap the columns so that the table compression later on is more
1373 efficient in terms of memory by ensuring most common value is in the
1374 middle and the least common are on the outside. IMO this would be better
1375 than a most to least common mapping as theres a decent chance the most
1376 common letter will share a node with the least common, meaning the node
1377 will not be compressable. With a middle is most common approach the worst
1378 case is when we have the least common nodes twice.
1382 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1383 regnode * const noper = NEXTOPER( cur );
1384 const U8 *uc = (U8*)STRING( noper );
1385 const U8 * const e = uc + STR_LEN( noper );
1387 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1388 const U8 *scan = (U8*)NULL;
1389 U32 wordlen = 0; /* required init */
1391 bool set_bit = trie->bitmap ? 1 : 0; /*store the first char in the bitmap?*/
1393 if (OP(noper) == NOTHING) {
1397 if ( set_bit ) /* bitmap only alloced when !(UTF&&Folding) */
1398 TRIE_BITMAP_SET(trie,*uc); /* store the raw first byte
1399 regardless of encoding */
1401 for ( ; uc < e ; uc += len ) {
1402 TRIE_CHARCOUNT(trie)++;
1406 if ( !trie->charmap[ uvc ] ) {
1407 trie->charmap[ uvc ]=( ++trie->uniquecharcount );
1409 trie->charmap[ folder[ uvc ] ] = trie->charmap[ uvc ];
1413 /* store the codepoint in the bitmap, and if its ascii
1414 also store its folded equivelent. */
1415 TRIE_BITMAP_SET(trie,uvc);
1417 /* store the folded codepoint */
1418 if ( folder ) TRIE_BITMAP_SET(trie,folder[ uvc ]);
1421 /* store first byte of utf8 representation of
1422 codepoints in the 127 < uvc < 256 range */
1423 if (127 < uvc && uvc < 192) {
1424 TRIE_BITMAP_SET(trie,194);
1425 } else if (191 < uvc ) {
1426 TRIE_BITMAP_SET(trie,195);
1427 /* && uvc < 256 -- we know uvc is < 256 already */
1430 set_bit = 0; /* We've done our bit :-) */
1435 widecharmap = newHV();
1437 svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 1 );
1440 Perl_croak( aTHX_ "error creating/fetching widecharmap entry for 0x%"UVXf, uvc );
1442 if ( !SvTRUE( *svpp ) ) {
1443 sv_setiv( *svpp, ++trie->uniquecharcount );
1448 if( cur == first ) {
1451 } else if (chars < trie->minlen) {
1453 } else if (chars > trie->maxlen) {
1457 } /* end first pass */
1458 DEBUG_TRIE_COMPILE_r(
1459 PerlIO_printf( Perl_debug_log, "%*sTRIE(%s): W:%d C:%d Uq:%d Min:%d Max:%d\n",
1460 (int)depth * 2 + 2,"",
1461 ( widecharmap ? "UTF8" : "NATIVE" ), (int)word_count,
1462 (int)TRIE_CHARCOUNT(trie), trie->uniquecharcount,
1463 (int)trie->minlen, (int)trie->maxlen )
1465 trie->wordlen = (U32 *) PerlMemShared_calloc( word_count, sizeof(U32) );
1468 We now know what we are dealing with in terms of unique chars and
1469 string sizes so we can calculate how much memory a naive
1470 representation using a flat table will take. If it's over a reasonable
1471 limit (as specified by ${^RE_TRIE_MAXBUF}) we use a more memory
1472 conservative but potentially much slower representation using an array
1475 At the end we convert both representations into the same compressed
1476 form that will be used in regexec.c for matching with. The latter
1477 is a form that cannot be used to construct with but has memory
1478 properties similar to the list form and access properties similar
1479 to the table form making it both suitable for fast searches and
1480 small enough that its feasable to store for the duration of a program.
1482 See the comment in the code where the compressed table is produced
1483 inplace from the flat tabe representation for an explanation of how
1484 the compression works.
1489 if ( (IV)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1) > SvIV(re_trie_maxbuff) ) {
1491 Second Pass -- Array Of Lists Representation
1493 Each state will be represented by a list of charid:state records
1494 (reg_trie_trans_le) the first such element holds the CUR and LEN
1495 points of the allocated array. (See defines above).
1497 We build the initial structure using the lists, and then convert
1498 it into the compressed table form which allows faster lookups
1499 (but cant be modified once converted).
1502 STRLEN transcount = 1;
1504 DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log,
1505 "%*sCompiling trie using list compiler\n",
1506 (int)depth * 2 + 2, ""));
1508 trie->states = (reg_trie_state *)
1509 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
1510 sizeof(reg_trie_state) );
1514 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1516 regnode * const noper = NEXTOPER( cur );
1517 U8 *uc = (U8*)STRING( noper );
1518 const U8 * const e = uc + STR_LEN( noper );
1519 U32 state = 1; /* required init */
1520 U16 charid = 0; /* sanity init */
1521 U8 *scan = (U8*)NULL; /* sanity init */
1522 STRLEN foldlen = 0; /* required init */
1523 U32 wordlen = 0; /* required init */
1524 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1526 if (OP(noper) != NOTHING) {
1527 for ( ; uc < e ; uc += len ) {
1532 charid = trie->charmap[ uvc ];
1534 SV** const svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 0);
1538 charid=(U16)SvIV( *svpp );
1541 /* charid is now 0 if we dont know the char read, or nonzero if we do */
1548 if ( !trie->states[ state ].trans.list ) {
1549 TRIE_LIST_NEW( state );
1551 for ( check = 1; check <= TRIE_LIST_USED( state ); check++ ) {
1552 if ( TRIE_LIST_ITEM( state, check ).forid == charid ) {
1553 newstate = TRIE_LIST_ITEM( state, check ).newstate;
1558 newstate = next_alloc++;
1559 TRIE_LIST_PUSH( state, charid, newstate );
1564 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %"IVdf, uvc );
1568 TRIE_HANDLE_WORD(state);
1570 } /* end second pass */
1572 /* next alloc is the NEXT state to be allocated */
1573 trie->statecount = next_alloc;
1574 trie->states = (reg_trie_state *)
1575 PerlMemShared_realloc( trie->states,
1577 * sizeof(reg_trie_state) );
1579 /* and now dump it out before we compress it */
1580 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_list(trie, widecharmap,
1581 revcharmap, next_alloc,
1585 trie->trans = (reg_trie_trans *)
1586 PerlMemShared_calloc( transcount, sizeof(reg_trie_trans) );
1593 for( state=1 ; state < next_alloc ; state ++ ) {
1597 DEBUG_TRIE_COMPILE_MORE_r(
1598 PerlIO_printf( Perl_debug_log, "tp: %d zp: %d ",tp,zp)
1602 if (trie->states[state].trans.list) {
1603 U16 minid=TRIE_LIST_ITEM( state, 1).forid;
1607 for( idx = 2 ; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
1608 const U16 forid = TRIE_LIST_ITEM( state, idx).forid;
1609 if ( forid < minid ) {
1611 } else if ( forid > maxid ) {
1615 if ( transcount < tp + maxid - minid + 1) {
1617 trie->trans = (reg_trie_trans *)
1618 PerlMemShared_realloc( trie->trans,
1620 * sizeof(reg_trie_trans) );
1621 Zero( trie->trans + (transcount / 2), transcount / 2 , reg_trie_trans );
1623 base = trie->uniquecharcount + tp - minid;
1624 if ( maxid == minid ) {
1626 for ( ; zp < tp ; zp++ ) {
1627 if ( ! trie->trans[ zp ].next ) {
1628 base = trie->uniquecharcount + zp - minid;
1629 trie->trans[ zp ].next = TRIE_LIST_ITEM( state, 1).newstate;
1630 trie->trans[ zp ].check = state;
1636 trie->trans[ tp ].next = TRIE_LIST_ITEM( state, 1).newstate;
1637 trie->trans[ tp ].check = state;
1642 for ( idx=1; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
1643 const U32 tid = base - trie->uniquecharcount + TRIE_LIST_ITEM( state, idx ).forid;
1644 trie->trans[ tid ].next = TRIE_LIST_ITEM( state, idx ).newstate;
1645 trie->trans[ tid ].check = state;
1647 tp += ( maxid - minid + 1 );
1649 Safefree(trie->states[ state ].trans.list);
1652 DEBUG_TRIE_COMPILE_MORE_r(
1653 PerlIO_printf( Perl_debug_log, " base: %d\n",base);
1656 trie->states[ state ].trans.base=base;
1658 trie->lasttrans = tp + 1;
1662 Second Pass -- Flat Table Representation.
1664 we dont use the 0 slot of either trans[] or states[] so we add 1 to each.
1665 We know that we will need Charcount+1 trans at most to store the data
1666 (one row per char at worst case) So we preallocate both structures
1667 assuming worst case.
1669 We then construct the trie using only the .next slots of the entry
1672 We use the .check field of the first entry of the node temporarily to
1673 make compression both faster and easier by keeping track of how many non
1674 zero fields are in the node.
1676 Since trans are numbered from 1 any 0 pointer in the table is a FAIL
1679 There are two terms at use here: state as a TRIE_NODEIDX() which is a
1680 number representing the first entry of the node, and state as a
1681 TRIE_NODENUM() which is the trans number. state 1 is TRIE_NODEIDX(1) and
1682 TRIE_NODENUM(1), state 2 is TRIE_NODEIDX(2) and TRIE_NODENUM(3) if there
1683 are 2 entrys per node. eg:
1691 The table is internally in the right hand, idx form. However as we also
1692 have to deal with the states array which is indexed by nodenum we have to
1693 use TRIE_NODENUM() to convert.
1696 DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log,
1697 "%*sCompiling trie using table compiler\n",
1698 (int)depth * 2 + 2, ""));
1700 trie->trans = (reg_trie_trans *)
1701 PerlMemShared_calloc( ( TRIE_CHARCOUNT(trie) + 1 )
1702 * trie->uniquecharcount + 1,
1703 sizeof(reg_trie_trans) );
1704 trie->states = (reg_trie_state *)
1705 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
1706 sizeof(reg_trie_state) );
1707 next_alloc = trie->uniquecharcount + 1;
1710 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1712 regnode * const noper = NEXTOPER( cur );
1713 const U8 *uc = (U8*)STRING( noper );
1714 const U8 * const e = uc + STR_LEN( noper );
1716 U32 state = 1; /* required init */
1718 U16 charid = 0; /* sanity init */
1719 U32 accept_state = 0; /* sanity init */
1720 U8 *scan = (U8*)NULL; /* sanity init */
1722 STRLEN foldlen = 0; /* required init */
1723 U32 wordlen = 0; /* required init */
1724 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1726 if ( OP(noper) != NOTHING ) {
1727 for ( ; uc < e ; uc += len ) {
1732 charid = trie->charmap[ uvc ];
1734 SV* const * const svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 0);
1735 charid = svpp ? (U16)SvIV(*svpp) : 0;
1739 if ( !trie->trans[ state + charid ].next ) {
1740 trie->trans[ state + charid ].next = next_alloc;
1741 trie->trans[ state ].check++;
1742 next_alloc += trie->uniquecharcount;
1744 state = trie->trans[ state + charid ].next;
1746 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %"IVdf, uvc );
1748 /* charid is now 0 if we dont know the char read, or nonzero if we do */
1751 accept_state = TRIE_NODENUM( state );
1752 TRIE_HANDLE_WORD(accept_state);
1754 } /* end second pass */
1756 /* and now dump it out before we compress it */
1757 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_table(trie, widecharmap,
1759 next_alloc, depth+1));
1763 * Inplace compress the table.*
1765 For sparse data sets the table constructed by the trie algorithm will
1766 be mostly 0/FAIL transitions or to put it another way mostly empty.
1767 (Note that leaf nodes will not contain any transitions.)
1769 This algorithm compresses the tables by eliminating most such
1770 transitions, at the cost of a modest bit of extra work during lookup:
1772 - Each states[] entry contains a .base field which indicates the
1773 index in the state[] array wheres its transition data is stored.
1775 - If .base is 0 there are no valid transitions from that node.
1777 - If .base is nonzero then charid is added to it to find an entry in
1780 -If trans[states[state].base+charid].check!=state then the
1781 transition is taken to be a 0/Fail transition. Thus if there are fail
1782 transitions at the front of the node then the .base offset will point
1783 somewhere inside the previous nodes data (or maybe even into a node
1784 even earlier), but the .check field determines if the transition is
1788 The following process inplace converts the table to the compressed
1789 table: We first do not compress the root node 1,and mark its all its
1790 .check pointers as 1 and set its .base pointer as 1 as well. This
1791 allows to do a DFA construction from the compressed table later, and
1792 ensures that any .base pointers we calculate later are greater than
1795 - We set 'pos' to indicate the first entry of the second node.
1797 - We then iterate over the columns of the node, finding the first and
1798 last used entry at l and m. We then copy l..m into pos..(pos+m-l),
1799 and set the .check pointers accordingly, and advance pos
1800 appropriately and repreat for the next node. Note that when we copy
1801 the next pointers we have to convert them from the original
1802 NODEIDX form to NODENUM form as the former is not valid post
1805 - If a node has no transitions used we mark its base as 0 and do not
1806 advance the pos pointer.
1808 - If a node only has one transition we use a second pointer into the
1809 structure to fill in allocated fail transitions from other states.
1810 This pointer is independent of the main pointer and scans forward
1811 looking for null transitions that are allocated to a state. When it
1812 finds one it writes the single transition into the "hole". If the
1813 pointer doesnt find one the single transition is appended as normal.
1815 - Once compressed we can Renew/realloc the structures to release the
1818 See "Table-Compression Methods" in sec 3.9 of the Red Dragon,
1819 specifically Fig 3.47 and the associated pseudocode.
1823 const U32 laststate = TRIE_NODENUM( next_alloc );
1826 trie->statecount = laststate;
1828 for ( state = 1 ; state < laststate ; state++ ) {
1830 const U32 stateidx = TRIE_NODEIDX( state );
1831 const U32 o_used = trie->trans[ stateidx ].check;
1832 U32 used = trie->trans[ stateidx ].check;
1833 trie->trans[ stateidx ].check = 0;
1835 for ( charid = 0 ; used && charid < trie->uniquecharcount ; charid++ ) {
1836 if ( flag || trie->trans[ stateidx + charid ].next ) {
1837 if ( trie->trans[ stateidx + charid ].next ) {
1839 for ( ; zp < pos ; zp++ ) {
1840 if ( ! trie->trans[ zp ].next ) {
1844 trie->states[ state ].trans.base = zp + trie->uniquecharcount - charid ;
1845 trie->trans[ zp ].next = SAFE_TRIE_NODENUM( trie->trans[ stateidx + charid ].next );
1846 trie->trans[ zp ].check = state;
1847 if ( ++zp > pos ) pos = zp;
1854 trie->states[ state ].trans.base = pos + trie->uniquecharcount - charid ;
1856 trie->trans[ pos ].next = SAFE_TRIE_NODENUM( trie->trans[ stateidx + charid ].next );
1857 trie->trans[ pos ].check = state;
1862 trie->lasttrans = pos + 1;
1863 trie->states = (reg_trie_state *)
1864 PerlMemShared_realloc( trie->states, laststate
1865 * sizeof(reg_trie_state) );
1866 DEBUG_TRIE_COMPILE_MORE_r(
1867 PerlIO_printf( Perl_debug_log,
1868 "%*sAlloc: %d Orig: %"IVdf" elements, Final:%"IVdf". Savings of %%%5.2f\n",
1869 (int)depth * 2 + 2,"",
1870 (int)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1 ),
1873 ( ( next_alloc - pos ) * 100 ) / (double)next_alloc );
1876 } /* end table compress */
1878 DEBUG_TRIE_COMPILE_MORE_r(
1879 PerlIO_printf(Perl_debug_log, "%*sStatecount:%"UVxf" Lasttrans:%"UVxf"\n",
1880 (int)depth * 2 + 2, "",
1881 (UV)trie->statecount,
1882 (UV)trie->lasttrans)
1884 /* resize the trans array to remove unused space */
1885 trie->trans = (reg_trie_trans *)
1886 PerlMemShared_realloc( trie->trans, trie->lasttrans
1887 * sizeof(reg_trie_trans) );
1889 /* and now dump out the compressed format */
1890 DEBUG_TRIE_COMPILE_r(dump_trie(trie, widecharmap, revcharmap, depth+1));
1892 { /* Modify the program and insert the new TRIE node*/
1893 U8 nodetype =(U8)(flags & 0xFF);
1897 regnode *optimize = NULL;
1898 #ifdef RE_TRACK_PATTERN_OFFSETS
1901 U32 mjd_nodelen = 0;
1902 #endif /* RE_TRACK_PATTERN_OFFSETS */
1903 #endif /* DEBUGGING */
1905 This means we convert either the first branch or the first Exact,
1906 depending on whether the thing following (in 'last') is a branch
1907 or not and whther first is the startbranch (ie is it a sub part of
1908 the alternation or is it the whole thing.)
1909 Assuming its a sub part we conver the EXACT otherwise we convert
1910 the whole branch sequence, including the first.
1912 /* Find the node we are going to overwrite */
1913 if ( first != startbranch || OP( last ) == BRANCH ) {
1914 /* branch sub-chain */
1915 NEXT_OFF( first ) = (U16)(last - first);
1916 #ifdef RE_TRACK_PATTERN_OFFSETS
1918 mjd_offset= Node_Offset((convert));
1919 mjd_nodelen= Node_Length((convert));
1922 /* whole branch chain */
1924 #ifdef RE_TRACK_PATTERN_OFFSETS
1927 const regnode *nop = NEXTOPER( convert );
1928 mjd_offset= Node_Offset((nop));
1929 mjd_nodelen= Node_Length((nop));
1933 PerlIO_printf(Perl_debug_log, "%*sMJD offset:%"UVuf" MJD length:%"UVuf"\n",
1934 (int)depth * 2 + 2, "",
1935 (UV)mjd_offset, (UV)mjd_nodelen)
1938 /* But first we check to see if there is a common prefix we can
1939 split out as an EXACT and put in front of the TRIE node. */
1940 trie->startstate= 1;
1941 if ( trie->bitmap && !widecharmap && !trie->jump ) {
1943 for ( state = 1 ; state < trie->statecount-1 ; state++ ) {
1947 const U32 base = trie->states[ state ].trans.base;
1949 if ( trie->states[state].wordnum )
1952 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
1953 if ( ( base + ofs >= trie->uniquecharcount ) &&
1954 ( base + ofs - trie->uniquecharcount < trie->lasttrans ) &&
1955 trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
1957 if ( ++count > 1 ) {
1958 SV **tmp = av_fetch( revcharmap, ofs, 0);
1959 const U8 *ch = (U8*)SvPV_nolen_const( *tmp );
1960 if ( state == 1 ) break;
1962 Zero(trie->bitmap, ANYOF_BITMAP_SIZE, char);
1964 PerlIO_printf(Perl_debug_log,
1965 "%*sNew Start State=%"UVuf" Class: [",
1966 (int)depth * 2 + 2, "",
1969 SV ** const tmp = av_fetch( revcharmap, idx, 0);
1970 const U8 * const ch = (U8*)SvPV_nolen_const( *tmp );
1972 TRIE_BITMAP_SET(trie,*ch);
1974 TRIE_BITMAP_SET(trie, folder[ *ch ]);
1976 PerlIO_printf(Perl_debug_log, (char*)ch)
1980 TRIE_BITMAP_SET(trie,*ch);
1982 TRIE_BITMAP_SET(trie,folder[ *ch ]);
1983 DEBUG_OPTIMISE_r(PerlIO_printf( Perl_debug_log,"%s", ch));
1989 SV **tmp = av_fetch( revcharmap, idx, 0);
1991 char *ch = SvPV( *tmp, len );
1993 SV *sv=sv_newmortal();
1994 PerlIO_printf( Perl_debug_log,
1995 "%*sPrefix State: %"UVuf" Idx:%"UVuf" Char='%s'\n",
1996 (int)depth * 2 + 2, "",
1998 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 6,
1999 PL_colors[0], PL_colors[1],
2000 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
2001 PERL_PV_ESCAPE_FIRSTCHAR
2006 OP( convert ) = nodetype;
2007 str=STRING(convert);
2010 STR_LEN(convert) += len;
2016 DEBUG_OPTIMISE_r(PerlIO_printf( Perl_debug_log,"]\n"));
2022 regnode *n = convert+NODE_SZ_STR(convert);
2023 NEXT_OFF(convert) = NODE_SZ_STR(convert);
2024 trie->startstate = state;
2025 trie->minlen -= (state - 1);
2026 trie->maxlen -= (state - 1);
2028 /* At least the UNICOS C compiler choked on this
2029 * being argument to DEBUG_r(), so let's just have
2032 #ifdef PERL_EXT_RE_BUILD
2038 regnode *fix = convert;
2039 U32 word = trie->wordcount;
2041 Set_Node_Offset_Length(convert, mjd_offset, state - 1);
2042 while( ++fix < n ) {
2043 Set_Node_Offset_Length(fix, 0, 0);
2046 SV ** const tmp = av_fetch( trie_words, word, 0 );
2048 if ( STR_LEN(convert) <= SvCUR(*tmp) )
2049 sv_chop(*tmp, SvPV_nolen(*tmp) + STR_LEN(convert));
2051 sv_chop(*tmp, SvPV_nolen(*tmp) + SvCUR(*tmp));
2059 NEXT_OFF(convert) = (U16)(tail - convert);
2060 DEBUG_r(optimize= n);
2066 if ( trie->maxlen ) {
2067 NEXT_OFF( convert ) = (U16)(tail - convert);
2068 ARG_SET( convert, data_slot );
2069 /* Store the offset to the first unabsorbed branch in
2070 jump[0], which is otherwise unused by the jump logic.
2071 We use this when dumping a trie and during optimisation. */
2073 trie->jump[0] = (U16)(nextbranch - convert);
2076 if ( !trie->states[trie->startstate].wordnum && trie->bitmap &&
2077 ( (char *)jumper - (char *)convert) >= (int)sizeof(struct regnode_charclass) )
2079 OP( convert ) = TRIEC;
2080 Copy(trie->bitmap, ((struct regnode_charclass *)convert)->bitmap, ANYOF_BITMAP_SIZE, char);
2081 PerlMemShared_free(trie->bitmap);
2084 OP( convert ) = TRIE;
2086 /* store the type in the flags */
2087 convert->flags = nodetype;
2091 + regarglen[ OP( convert ) ];
2093 /* XXX We really should free up the resource in trie now,
2094 as we won't use them - (which resources?) dmq */
2096 /* needed for dumping*/
2097 DEBUG_r(if (optimize) {
2098 regnode *opt = convert;
2100 while ( ++opt < optimize) {
2101 Set_Node_Offset_Length(opt,0,0);
2104 Try to clean up some of the debris left after the
2107 while( optimize < jumper ) {
2108 mjd_nodelen += Node_Length((optimize));
2109 OP( optimize ) = OPTIMIZED;
2110 Set_Node_Offset_Length(optimize,0,0);
2113 Set_Node_Offset_Length(convert,mjd_offset,mjd_nodelen);
2115 } /* end node insert */
2116 RExC_rxi->data->data[ data_slot + 1 ] = (void*)widecharmap;
2118 RExC_rxi->data->data[ data_slot + TRIE_WORDS_OFFSET ] = (void*)trie_words;
2119 RExC_rxi->data->data[ data_slot + 3 ] = (void*)revcharmap;
2121 SvREFCNT_dec(revcharmap);
2125 : trie->startstate>1
2131 S_make_trie_failtable(pTHX_ RExC_state_t *pRExC_state, regnode *source, regnode *stclass, U32 depth)
2133 /* The Trie is constructed and compressed now so we can build a fail array now if its needed
2135 This is basically the Aho-Corasick algorithm. Its from exercise 3.31 and 3.32 in the
2136 "Red Dragon" -- Compilers, principles, techniques, and tools. Aho, Sethi, Ullman 1985/88
2139 We find the fail state for each state in the trie, this state is the longest proper
2140 suffix of the current states 'word' that is also a proper prefix of another word in our
2141 trie. State 1 represents the word '' and is the thus the default fail state. This allows
2142 the DFA not to have to restart after its tried and failed a word at a given point, it
2143 simply continues as though it had been matching the other word in the first place.
2145 'abcdgu'=~/abcdefg|cdgu/
2146 When we get to 'd' we are still matching the first word, we would encounter 'g' which would
2147 fail, which would bring use to the state representing 'd' in the second word where we would
2148 try 'g' and succeed, prodceding to match 'cdgu'.
2150 /* add a fail transition */
2151 const U32 trie_offset = ARG(source);
2152 reg_trie_data *trie=(reg_trie_data *)RExC_rxi->data->data[trie_offset];
2154 const U32 ucharcount = trie->uniquecharcount;
2155 const U32 numstates = trie->statecount;
2156 const U32 ubound = trie->lasttrans + ucharcount;
2160 U32 base = trie->states[ 1 ].trans.base;
2163 const U32 data_slot = add_data( pRExC_state, 1, "T" );
2164 GET_RE_DEBUG_FLAGS_DECL;
2166 PERL_UNUSED_ARG(depth);
2170 ARG_SET( stclass, data_slot );
2171 aho = (reg_ac_data *) PerlMemShared_calloc( 1, sizeof(reg_ac_data) );
2172 RExC_rxi->data->data[ data_slot ] = (void*)aho;
2173 aho->trie=trie_offset;
2174 aho->states=(reg_trie_state *)PerlMemShared_malloc( numstates * sizeof(reg_trie_state) );
2175 Copy( trie->states, aho->states, numstates, reg_trie_state );
2176 Newxz( q, numstates, U32);
2177 aho->fail = (U32 *) PerlMemShared_calloc( numstates, sizeof(U32) );
2180 /* initialize fail[0..1] to be 1 so that we always have
2181 a valid final fail state */
2182 fail[ 0 ] = fail[ 1 ] = 1;
2184 for ( charid = 0; charid < ucharcount ; charid++ ) {
2185 const U32 newstate = TRIE_TRANS_STATE( 1, base, ucharcount, charid, 0 );
2187 q[ q_write ] = newstate;
2188 /* set to point at the root */
2189 fail[ q[ q_write++ ] ]=1;
2192 while ( q_read < q_write) {
2193 const U32 cur = q[ q_read++ % numstates ];
2194 base = trie->states[ cur ].trans.base;
2196 for ( charid = 0 ; charid < ucharcount ; charid++ ) {
2197 const U32 ch_state = TRIE_TRANS_STATE( cur, base, ucharcount, charid, 1 );
2199 U32 fail_state = cur;
2202 fail_state = fail[ fail_state ];
2203 fail_base = aho->states[ fail_state ].trans.base;
2204 } while ( !TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 ) );
2206 fail_state = TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 );
2207 fail[ ch_state ] = fail_state;
2208 if ( !aho->states[ ch_state ].wordnum && aho->states[ fail_state ].wordnum )
2210 aho->states[ ch_state ].wordnum = aho->states[ fail_state ].wordnum;
2212 q[ q_write++ % numstates] = ch_state;
2216 /* restore fail[0..1] to 0 so that we "fall out" of the AC loop
2217 when we fail in state 1, this allows us to use the
2218 charclass scan to find a valid start char. This is based on the principle
2219 that theres a good chance the string being searched contains lots of stuff
2220 that cant be a start char.
2222 fail[ 0 ] = fail[ 1 ] = 0;
2223 DEBUG_TRIE_COMPILE_r({
2224 PerlIO_printf(Perl_debug_log,
2225 "%*sStclass Failtable (%"UVuf" states): 0",
2226 (int)(depth * 2), "", (UV)numstates
2228 for( q_read=1; q_read<numstates; q_read++ ) {
2229 PerlIO_printf(Perl_debug_log, ", %"UVuf, (UV)fail[q_read]);
2231 PerlIO_printf(Perl_debug_log, "\n");
2234 /*RExC_seen |= REG_SEEN_TRIEDFA;*/
2239 * There are strange code-generation bugs caused on sparc64 by gcc-2.95.2.
2240 * These need to be revisited when a newer toolchain becomes available.
2242 #if defined(__sparc64__) && defined(__GNUC__)
2243 # if __GNUC__ < 2 || (__GNUC__ == 2 && __GNUC_MINOR__ < 96)
2244 # undef SPARC64_GCC_WORKAROUND
2245 # define SPARC64_GCC_WORKAROUND 1
2249 #define DEBUG_PEEP(str,scan,depth) \
2250 DEBUG_OPTIMISE_r({if (scan){ \
2251 SV * const mysv=sv_newmortal(); \
2252 regnode *Next = regnext(scan); \
2253 regprop(RExC_rx, mysv, scan); \
2254 PerlIO_printf(Perl_debug_log, "%*s" str ">%3d: %s (%d)\n", \
2255 (int)depth*2, "", REG_NODE_NUM(scan), SvPV_nolen_const(mysv),\
2256 Next ? (REG_NODE_NUM(Next)) : 0 ); \
2263 #define JOIN_EXACT(scan,min,flags) \
2264 if (PL_regkind[OP(scan)] == EXACT) \
2265 join_exact(pRExC_state,(scan),(min),(flags),NULL,depth+1)
2268 S_join_exact(pTHX_ RExC_state_t *pRExC_state, regnode *scan, I32 *min, U32 flags,regnode *val, U32 depth) {
2269 /* Merge several consecutive EXACTish nodes into one. */
2270 regnode *n = regnext(scan);
2272 regnode *next = scan + NODE_SZ_STR(scan);
2276 regnode *stop = scan;
2277 GET_RE_DEBUG_FLAGS_DECL;
2279 PERL_UNUSED_ARG(depth);
2281 #ifndef EXPERIMENTAL_INPLACESCAN
2282 PERL_UNUSED_ARG(flags);
2283 PERL_UNUSED_ARG(val);
2285 DEBUG_PEEP("join",scan,depth);
2287 /* Skip NOTHING, merge EXACT*. */
2289 ( PL_regkind[OP(n)] == NOTHING ||
2290 (stringok && (OP(n) == OP(scan))))
2292 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
2294 if (OP(n) == TAIL || n > next)
2296 if (PL_regkind[OP(n)] == NOTHING) {
2297 DEBUG_PEEP("skip:",n,depth);
2298 NEXT_OFF(scan) += NEXT_OFF(n);
2299 next = n + NODE_STEP_REGNODE;
2306 else if (stringok) {
2307 const unsigned int oldl = STR_LEN(scan);
2308 regnode * const nnext = regnext(n);
2310 DEBUG_PEEP("merg",n,depth);
2313 if (oldl + STR_LEN(n) > U8_MAX)
2315 NEXT_OFF(scan) += NEXT_OFF(n);
2316 STR_LEN(scan) += STR_LEN(n);
2317 next = n + NODE_SZ_STR(n);
2318 /* Now we can overwrite *n : */
2319 Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
2327 #ifdef EXPERIMENTAL_INPLACESCAN
2328 if (flags && !NEXT_OFF(n)) {
2329 DEBUG_PEEP("atch", val, depth);
2330 if (reg_off_by_arg[OP(n)]) {
2331 ARG_SET(n, val - n);
2334 NEXT_OFF(n) = val - n;
2341 if (UTF && ( OP(scan) == EXACTF ) && ( STR_LEN(scan) >= 6 ) ) {
2343 Two problematic code points in Unicode casefolding of EXACT nodes:
2345 U+0390 - GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
2346 U+03B0 - GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
2352 U+03B9 U+0308 U+0301 0xCE 0xB9 0xCC 0x88 0xCC 0x81
2353 U+03C5 U+0308 U+0301 0xCF 0x85 0xCC 0x88 0xCC 0x81
2355 This means that in case-insensitive matching (or "loose matching",
2356 as Unicode calls it), an EXACTF of length six (the UTF-8 encoded byte
2357 length of the above casefolded versions) can match a target string
2358 of length two (the byte length of UTF-8 encoded U+0390 or U+03B0).
2359 This would rather mess up the minimum length computation.
2361 What we'll do is to look for the tail four bytes, and then peek
2362 at the preceding two bytes to see whether we need to decrease
2363 the minimum length by four (six minus two).
2365 Thanks to the design of UTF-8, there cannot be false matches:
2366 A sequence of valid UTF-8 bytes cannot be a subsequence of
2367 another valid sequence of UTF-8 bytes.
2370 char * const s0 = STRING(scan), *s, *t;
2371 char * const s1 = s0 + STR_LEN(scan) - 1;
2372 char * const s2 = s1 - 4;
2373 #ifdef EBCDIC /* RD tunifold greek 0390 and 03B0 */
2374 const char t0[] = "\xaf\x49\xaf\x42";
2376 const char t0[] = "\xcc\x88\xcc\x81";
2378 const char * const t1 = t0 + 3;
2381 s < s2 && (t = ninstr(s, s1, t0, t1));
2384 if (((U8)t[-1] == 0x68 && (U8)t[-2] == 0xB4) ||
2385 ((U8)t[-1] == 0x46 && (U8)t[-2] == 0xB5))
2387 if (((U8)t[-1] == 0xB9 && (U8)t[-2] == 0xCE) ||
2388 ((U8)t[-1] == 0x85 && (U8)t[-2] == 0xCF))
2396 n = scan + NODE_SZ_STR(scan);
2398 if (PL_regkind[OP(n)] != NOTHING || OP(n) == NOTHING) {
2405 DEBUG_OPTIMISE_r(if (merged){DEBUG_PEEP("finl",scan,depth)});
2409 /* REx optimizer. Converts nodes into quickier variants "in place".
2410 Finds fixed substrings. */
2412 /* Stops at toplevel WHILEM as well as at "last". At end *scanp is set
2413 to the position after last scanned or to NULL. */
2415 #define INIT_AND_WITHP \
2416 assert(!and_withp); \
2417 Newx(and_withp,1,struct regnode_charclass_class); \
2418 SAVEFREEPV(and_withp)
2420 /* this is a chain of data about sub patterns we are processing that
2421 need to be handled seperately/specially in study_chunk. Its so
2422 we can simulate recursion without losing state. */
2424 typedef struct scan_frame {
2425 regnode *last; /* last node to process in this frame */
2426 regnode *next; /* next node to process when last is reached */
2427 struct scan_frame *prev; /*previous frame*/
2428 I32 stop; /* what stopparen do we use */
2432 #define SCAN_COMMIT(s, data, m) scan_commit(s, data, m, is_inf)
2434 #define CASE_SYNST_FNC(nAmE) \
2436 if (flags & SCF_DO_STCLASS_AND) { \
2437 for (value = 0; value < 256; value++) \
2438 if (!is_ ## nAmE ## _cp(value)) \
2439 ANYOF_BITMAP_CLEAR(data->start_class, value); \
2442 for (value = 0; value < 256; value++) \
2443 if (is_ ## nAmE ## _cp(value)) \
2444 ANYOF_BITMAP_SET(data->start_class, value); \
2448 if (flags & SCF_DO_STCLASS_AND) { \
2449 for (value = 0; value < 256; value++) \
2450 if (is_ ## nAmE ## _cp(value)) \
2451 ANYOF_BITMAP_CLEAR(data->start_class, value); \
2454 for (value = 0; value < 256; value++) \
2455 if (!is_ ## nAmE ## _cp(value)) \
2456 ANYOF_BITMAP_SET(data->start_class, value); \
2463 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
2464 I32 *minlenp, I32 *deltap,
2469 struct regnode_charclass_class *and_withp,
2470 U32 flags, U32 depth)
2471 /* scanp: Start here (read-write). */
2472 /* deltap: Write maxlen-minlen here. */
2473 /* last: Stop before this one. */
2474 /* data: string data about the pattern */
2475 /* stopparen: treat close N as END */
2476 /* recursed: which subroutines have we recursed into */
2477 /* and_withp: Valid if flags & SCF_DO_STCLASS_OR */
2480 I32 min = 0, pars = 0, code;
2481 regnode *scan = *scanp, *next;
2483 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
2484 int is_inf_internal = 0; /* The studied chunk is infinite */
2485 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
2486 scan_data_t data_fake;
2487 SV *re_trie_maxbuff = NULL;
2488 regnode *first_non_open = scan;
2489 I32 stopmin = I32_MAX;
2490 scan_frame *frame = NULL;
2492 GET_RE_DEBUG_FLAGS_DECL;
2495 StructCopy(&zero_scan_data, &data_fake, scan_data_t);
2499 while (first_non_open && OP(first_non_open) == OPEN)
2500 first_non_open=regnext(first_non_open);
2505 while ( scan && OP(scan) != END && scan < last ){
2506 /* Peephole optimizer: */
2507 DEBUG_STUDYDATA("Peep:", data,depth);
2508 DEBUG_PEEP("Peep",scan,depth);
2509 JOIN_EXACT(scan,&min,0);
2511 /* Follow the next-chain of the current node and optimize
2512 away all the NOTHINGs from it. */
2513 if (OP(scan) != CURLYX) {
2514 const int max = (reg_off_by_arg[OP(scan)]
2516 /* I32 may be smaller than U16 on CRAYs! */
2517 : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
2518 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
2522 /* Skip NOTHING and LONGJMP. */
2523 while ((n = regnext(n))
2524 && ((PL_regkind[OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
2525 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
2526 && off + noff < max)
2528 if (reg_off_by_arg[OP(scan)])
2531 NEXT_OFF(scan) = off;
2536 /* The principal pseudo-switch. Cannot be a switch, since we
2537 look into several different things. */
2538 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
2539 || OP(scan) == IFTHEN) {
2540 next = regnext(scan);
2542 /* demq: the op(next)==code check is to see if we have "branch-branch" AFAICT */
2544 if (OP(next) == code || code == IFTHEN) {
2545 /* NOTE - There is similar code to this block below for handling
2546 TRIE nodes on a re-study. If you change stuff here check there
2548 I32 max1 = 0, min1 = I32_MAX, num = 0;
2549 struct regnode_charclass_class accum;
2550 regnode * const startbranch=scan;
2552 if (flags & SCF_DO_SUBSTR)
2553 SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot merge strings after this. */
2554 if (flags & SCF_DO_STCLASS)
2555 cl_init_zero(pRExC_state, &accum);
2557 while (OP(scan) == code) {
2558 I32 deltanext, minnext, f = 0, fake;
2559 struct regnode_charclass_class this_class;
2562 data_fake.flags = 0;
2564 data_fake.whilem_c = data->whilem_c;
2565 data_fake.last_closep = data->last_closep;
2568 data_fake.last_closep = &fake;
2570 data_fake.pos_delta = delta;
2571 next = regnext(scan);
2572 scan = NEXTOPER(scan);
2574 scan = NEXTOPER(scan);
2575 if (flags & SCF_DO_STCLASS) {
2576 cl_init(pRExC_state, &this_class);
2577 data_fake.start_class = &this_class;
2578 f = SCF_DO_STCLASS_AND;
2580 if (flags & SCF_WHILEM_VISITED_POS)
2581 f |= SCF_WHILEM_VISITED_POS;
2583 /* we suppose the run is continuous, last=next...*/
2584 minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext,
2586 stopparen, recursed, NULL, f,depth+1);
2589 if (max1 < minnext + deltanext)
2590 max1 = minnext + deltanext;
2591 if (deltanext == I32_MAX)
2592 is_inf = is_inf_internal = 1;
2594 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
2596 if (data_fake.flags & SCF_SEEN_ACCEPT) {
2597 if ( stopmin > minnext)
2598 stopmin = min + min1;
2599 flags &= ~SCF_DO_SUBSTR;
2601 data->flags |= SCF_SEEN_ACCEPT;
2604 if (data_fake.flags & SF_HAS_EVAL)
2605 data->flags |= SF_HAS_EVAL;
2606 data->whilem_c = data_fake.whilem_c;
2608 if (flags & SCF_DO_STCLASS)
2609 cl_or(pRExC_state, &accum, &this_class);
2611 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
2613 if (flags & SCF_DO_SUBSTR) {
2614 data->pos_min += min1;
2615 data->pos_delta += max1 - min1;
2616 if (max1 != min1 || is_inf)
2617 data->longest = &(data->longest_float);
2620 delta += max1 - min1;
2621 if (flags & SCF_DO_STCLASS_OR) {
2622 cl_or(pRExC_state, data->start_class, &accum);
2624 cl_and(data->start_class, and_withp);
2625 flags &= ~SCF_DO_STCLASS;
2628 else if (flags & SCF_DO_STCLASS_AND) {
2630 cl_and(data->start_class, &accum);
2631 flags &= ~SCF_DO_STCLASS;
2634 /* Switch to OR mode: cache the old value of
2635 * data->start_class */
2637 StructCopy(data->start_class, and_withp,
2638 struct regnode_charclass_class);
2639 flags &= ~SCF_DO_STCLASS_AND;
2640 StructCopy(&accum, data->start_class,
2641 struct regnode_charclass_class);
2642 flags |= SCF_DO_STCLASS_OR;
2643 data->start_class->flags |= ANYOF_EOS;
2647 if (PERL_ENABLE_TRIE_OPTIMISATION && OP( startbranch ) == BRANCH ) {
2650 Assuming this was/is a branch we are dealing with: 'scan' now
2651 points at the item that follows the branch sequence, whatever
2652 it is. We now start at the beginning of the sequence and look
2659 which would be constructed from a pattern like /A|LIST|OF|WORDS/
2661 If we can find such a subseqence we need to turn the first
2662 element into a trie and then add the subsequent branch exact
2663 strings to the trie.
2667 1. patterns where the whole set of branch can be converted.
2669 2. patterns where only a subset can be converted.
2671 In case 1 we can replace the whole set with a single regop
2672 for the trie. In case 2 we need to keep the start and end
2675 'BRANCH EXACT; BRANCH EXACT; BRANCH X'
2676 becomes BRANCH TRIE; BRANCH X;
2678 There is an additional case, that being where there is a
2679 common prefix, which gets split out into an EXACT like node
2680 preceding the TRIE node.
2682 If x(1..n)==tail then we can do a simple trie, if not we make
2683 a "jump" trie, such that when we match the appropriate word
2684 we "jump" to the appopriate tail node. Essentailly we turn
2685 a nested if into a case structure of sorts.
2690 if (!re_trie_maxbuff) {
2691 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
2692 if (!SvIOK(re_trie_maxbuff))
2693 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
2695 if ( SvIV(re_trie_maxbuff)>=0 ) {
2697 regnode *first = (regnode *)NULL;
2698 regnode *last = (regnode *)NULL;
2699 regnode *tail = scan;
2704 SV * const mysv = sv_newmortal(); /* for dumping */
2706 /* var tail is used because there may be a TAIL
2707 regop in the way. Ie, the exacts will point to the
2708 thing following the TAIL, but the last branch will
2709 point at the TAIL. So we advance tail. If we
2710 have nested (?:) we may have to move through several
2714 while ( OP( tail ) == TAIL ) {
2715 /* this is the TAIL generated by (?:) */
2716 tail = regnext( tail );
2721 regprop(RExC_rx, mysv, tail );
2722 PerlIO_printf( Perl_debug_log, "%*s%s%s\n",
2723 (int)depth * 2 + 2, "",
2724 "Looking for TRIE'able sequences. Tail node is: ",
2725 SvPV_nolen_const( mysv )
2731 step through the branches, cur represents each
2732 branch, noper is the first thing to be matched
2733 as part of that branch and noper_next is the
2734 regnext() of that node. if noper is an EXACT
2735 and noper_next is the same as scan (our current
2736 position in the regex) then the EXACT branch is
2737 a possible optimization target. Once we have
2738 two or more consequetive such branches we can
2739 create a trie of the EXACT's contents and stich
2740 it in place. If the sequence represents all of
2741 the branches we eliminate the whole thing and
2742 replace it with a single TRIE. If it is a
2743 subsequence then we need to stitch it in. This
2744 means the first branch has to remain, and needs
2745 to be repointed at the item on the branch chain
2746 following the last branch optimized. This could
2747 be either a BRANCH, in which case the
2748 subsequence is internal, or it could be the
2749 item following the branch sequence in which
2750 case the subsequence is at the end.
2754 /* dont use tail as the end marker for this traverse */
2755 for ( cur = startbranch ; cur != scan ; cur = regnext( cur ) ) {
2756 regnode * const noper = NEXTOPER( cur );
2757 #if defined(DEBUGGING) || defined(NOJUMPTRIE)
2758 regnode * const noper_next = regnext( noper );
2762 regprop(RExC_rx, mysv, cur);
2763 PerlIO_printf( Perl_debug_log, "%*s- %s (%d)",
2764 (int)depth * 2 + 2,"", SvPV_nolen_const( mysv ), REG_NODE_NUM(cur) );
2766 regprop(RExC_rx, mysv, noper);
2767 PerlIO_printf( Perl_debug_log, " -> %s",
2768 SvPV_nolen_const(mysv));
2771 regprop(RExC_rx, mysv, noper_next );
2772 PerlIO_printf( Perl_debug_log,"\t=> %s\t",
2773 SvPV_nolen_const(mysv));
2775 PerlIO_printf( Perl_debug_log, "(First==%d,Last==%d,Cur==%d)\n",
2776 REG_NODE_NUM(first), REG_NODE_NUM(last), REG_NODE_NUM(cur) );
2778 if ( (((first && optype!=NOTHING) ? OP( noper ) == optype
2779 : PL_regkind[ OP( noper ) ] == EXACT )
2780 || OP(noper) == NOTHING )
2782 && noper_next == tail
2787 if ( !first || optype == NOTHING ) {
2788 if (!first) first = cur;
2789 optype = OP( noper );
2795 make_trie( pRExC_state,
2796 startbranch, first, cur, tail, count,
2799 if ( PL_regkind[ OP( noper ) ] == EXACT
2801 && noper_next == tail
2806 optype = OP( noper );
2816 regprop(RExC_rx, mysv, cur);
2817 PerlIO_printf( Perl_debug_log,
2818 "%*s- %s (%d) <SCAN FINISHED>\n", (int)depth * 2 + 2,
2819 "", SvPV_nolen_const( mysv ),REG_NODE_NUM(cur));
2823 made= make_trie( pRExC_state, startbranch, first, scan, tail, count, optype, depth+1 );
2824 #ifdef TRIE_STUDY_OPT
2825 if ( ((made == MADE_EXACT_TRIE &&
2826 startbranch == first)
2827 || ( first_non_open == first )) &&
2829 flags |= SCF_TRIE_RESTUDY;
2830 if ( startbranch == first
2833 RExC_seen &=~REG_TOP_LEVEL_BRANCHES;
2843 else if ( code == BRANCHJ ) { /* single branch is optimized. */
2844 scan = NEXTOPER(NEXTOPER(scan));
2845 } else /* single branch is optimized. */
2846 scan = NEXTOPER(scan);
2848 } else if (OP(scan) == SUSPEND || OP(scan) == GOSUB || OP(scan) == GOSTART) {
2849 scan_frame *newframe = NULL;
2854 if (OP(scan) != SUSPEND) {
2855 /* set the pointer */
2856 if (OP(scan) == GOSUB) {
2858 RExC_recurse[ARG2L(scan)] = scan;
2859 start = RExC_open_parens[paren-1];
2860 end = RExC_close_parens[paren-1];
2863 start = RExC_rxi->program + 1;
2867 Newxz(recursed, (((RExC_npar)>>3) +1), U8);
2868 SAVEFREEPV(recursed);
2870 if (!PAREN_TEST(recursed,paren+1)) {
2871 PAREN_SET(recursed,paren+1);
2872 Newx(newframe,1,scan_frame);
2874 if (flags & SCF_DO_SUBSTR) {
2875 SCAN_COMMIT(pRExC_state,data,minlenp);
2876 data->longest = &(data->longest_float);
2878 is_inf = is_inf_internal = 1;
2879 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
2880 cl_anything(pRExC_state, data->start_class);
2881 flags &= ~SCF_DO_STCLASS;
2884 Newx(newframe,1,scan_frame);
2887 end = regnext(scan);
2892 SAVEFREEPV(newframe);
2893 newframe->next = regnext(scan);
2894 newframe->last = last;
2895 newframe->stop = stopparen;
2896 newframe->prev = frame;
2906 else if (OP(scan) == EXACT) {
2907 I32 l = STR_LEN(scan);
2910 const U8 * const s = (U8*)STRING(scan);
2911 l = utf8_length(s, s + l);
2912 uc = utf8_to_uvchr(s, NULL);
2914 uc = *((U8*)STRING(scan));
2917 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
2918 /* The code below prefers earlier match for fixed
2919 offset, later match for variable offset. */
2920 if (data->last_end == -1) { /* Update the start info. */
2921 data->last_start_min = data->pos_min;
2922 data->last_start_max = is_inf
2923 ? I32_MAX : data->pos_min + data->pos_delta;
2925 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
2927 SvUTF8_on(data->last_found);
2929 SV * const sv = data->last_found;
2930 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
2931 mg_find(sv, PERL_MAGIC_utf8) : NULL;
2932 if (mg && mg->mg_len >= 0)
2933 mg->mg_len += utf8_length((U8*)STRING(scan),
2934 (U8*)STRING(scan)+STR_LEN(scan));
2936 data->last_end = data->pos_min + l;
2937 data->pos_min += l; /* As in the first entry. */
2938 data->flags &= ~SF_BEFORE_EOL;
2940 if (flags & SCF_DO_STCLASS_AND) {
2941 /* Check whether it is compatible with what we know already! */
2945 (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
2946 && !ANYOF_BITMAP_TEST(data->start_class, uc)
2947 && (!(data->start_class->flags & ANYOF_FOLD)
2948 || !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
2951 ANYOF_CLASS_ZERO(data->start_class);
2952 ANYOF_BITMAP_ZERO(data->start_class);
2954 ANYOF_BITMAP_SET(data->start_class, uc);
2955 data->start_class->flags &= ~ANYOF_EOS;
2957 data->start_class->flags &= ~ANYOF_UNICODE_ALL;
2959 else if (flags & SCF_DO_STCLASS_OR) {
2960 /* false positive possible if the class is case-folded */
2962 ANYOF_BITMAP_SET(data->start_class, uc);
2964 data->start_class->flags |= ANYOF_UNICODE_ALL;
2965 data->start_class->flags &= ~ANYOF_EOS;
2966 cl_and(data->start_class, and_withp);
2968 flags &= ~SCF_DO_STCLASS;
2970 else if (PL_regkind[OP(scan)] == EXACT) { /* But OP != EXACT! */
2971 I32 l = STR_LEN(scan);
2972 UV uc = *((U8*)STRING(scan));
2974 /* Search for fixed substrings supports EXACT only. */
2975 if (flags & SCF_DO_SUBSTR) {
2977 SCAN_COMMIT(pRExC_state, data, minlenp);
2980 const U8 * const s = (U8 *)STRING(scan);
2981 l = utf8_length(s, s + l);
2982 uc = utf8_to_uvchr(s, NULL);
2985 if (flags & SCF_DO_SUBSTR)
2987 if (flags & SCF_DO_STCLASS_AND) {
2988 /* Check whether it is compatible with what we know already! */
2992 (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
2993 && !ANYOF_BITMAP_TEST(data->start_class, uc)
2994 && !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
2996 ANYOF_CLASS_ZERO(data->start_class);
2997 ANYOF_BITMAP_ZERO(data->start_class);
2999 ANYOF_BITMAP_SET(data->start_class, uc);
3000 data->start_class->flags &= ~ANYOF_EOS;
3001 data->start_class->flags |= ANYOF_FOLD;
3002 if (OP(scan) == EXACTFL)
3003 data->start_class->flags |= ANYOF_LOCALE;
3006 else if (flags & SCF_DO_STCLASS_OR) {
3007 if (data->start_class->flags & ANYOF_FOLD) {
3008 /* false positive possible if the class is case-folded.
3009 Assume that the locale settings are the same... */
3011 ANYOF_BITMAP_SET(data->start_class, uc);
3012 data->start_class->flags &= ~ANYOF_EOS;
3014 cl_and(data->start_class, and_withp);
3016 flags &= ~SCF_DO_STCLASS;
3018 else if (strchr((const char*)PL_varies,OP(scan))) {
3019 I32 mincount, maxcount, minnext, deltanext, fl = 0;
3020 I32 f = flags, pos_before = 0;
3021 regnode * const oscan = scan;
3022 struct regnode_charclass_class this_class;
3023 struct regnode_charclass_class *oclass = NULL;
3024 I32 next_is_eval = 0;
3026 switch (PL_regkind[OP(scan)]) {
3027 case WHILEM: /* End of (?:...)* . */
3028 scan = NEXTOPER(scan);
3031 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
3032 next = NEXTOPER(scan);
3033 if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
3035 maxcount = REG_INFTY;
3036 next = regnext(scan);
3037 scan = NEXTOPER(scan);
3041 if (flags & SCF_DO_SUBSTR)
3046 if (flags & SCF_DO_STCLASS) {
3048 maxcount = REG_INFTY;
3049 next = regnext(scan);
3050 scan = NEXTOPER(scan);
3053 is_inf = is_inf_internal = 1;
3054 scan = regnext(scan);
3055 if (flags & SCF_DO_SUBSTR) {
3056 SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot extend fixed substrings */
3057 data->longest = &(data->longest_float);
3059 goto optimize_curly_tail;
3061 if (stopparen>0 && (OP(scan)==CURLYN || OP(scan)==CURLYM)
3062 && (scan->flags == stopparen))
3067 mincount = ARG1(scan);
3068 maxcount = ARG2(scan);
3070 next = regnext(scan);
3071 if (OP(scan) == CURLYX) {
3072 I32 lp = (data ? *(data->last_closep) : 0);
3073 scan->flags = ((lp <= (I32)U8_MAX) ? (U8)lp : U8_MAX);
3075 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
3076 next_is_eval = (OP(scan) == EVAL);
3078 if (flags & SCF_DO_SUBSTR) {
3079 if (mincount == 0) SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot extend fixed substrings */
3080 pos_before = data->pos_min;
3084 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
3086 data->flags |= SF_IS_INF;
3088 if (flags & SCF_DO_STCLASS) {
3089 cl_init(pRExC_state, &this_class);
3090 oclass = data->start_class;
3091 data->start_class = &this_class;
3092 f |= SCF_DO_STCLASS_AND;
3093 f &= ~SCF_DO_STCLASS_OR;
3095 /* These are the cases when once a subexpression
3096 fails at a particular position, it cannot succeed
3097 even after backtracking at the enclosing scope.
3099 XXXX what if minimal match and we are at the
3100 initial run of {n,m}? */
3101 if ((mincount != maxcount - 1) && (maxcount != REG_INFTY))
3102 f &= ~SCF_WHILEM_VISITED_POS;
3104 /* This will finish on WHILEM, setting scan, or on NULL: */
3105 minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext,
3106 last, data, stopparen, recursed, NULL,
3108 ? (f & ~SCF_DO_SUBSTR) : f),depth+1);
3110 if (flags & SCF_DO_STCLASS)
3111 data->start_class = oclass;
3112 if (mincount == 0 || minnext == 0) {
3113 if (flags & SCF_DO_STCLASS_OR) {
3114 cl_or(pRExC_state, data->start_class, &this_class);
3116 else if (flags & SCF_DO_STCLASS_AND) {
3117 /* Switch to OR mode: cache the old value of
3118 * data->start_class */
3120 StructCopy(data->start_class, and_withp,
3121 struct regnode_charclass_class);
3122 flags &= ~SCF_DO_STCLASS_AND;
3123 StructCopy(&this_class, data->start_class,
3124 struct regnode_charclass_class);
3125 flags |= SCF_DO_STCLASS_OR;
3126 data->start_class->flags |= ANYOF_EOS;
3128 } else { /* Non-zero len */
3129 if (flags & SCF_DO_STCLASS_OR) {
3130 cl_or(pRExC_state, data->start_class, &this_class);
3131 cl_and(data->start_class, and_withp);
3133 else if (flags & SCF_DO_STCLASS_AND)
3134 cl_and(data->start_class, &this_class);
3135 flags &= ~SCF_DO_STCLASS;
3137 if (!scan) /* It was not CURLYX, but CURLY. */
3139 if ( /* ? quantifier ok, except for (?{ ... }) */
3140 (next_is_eval || !(mincount == 0 && maxcount == 1))
3141 && (minnext == 0) && (deltanext == 0)
3142 && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
3143 && maxcount <= REG_INFTY/3 /* Complement check for big count */
3144 && ckWARN(WARN_REGEXP))
3147 "Quantifier unexpected on zero-length expression");
3150 min += minnext * mincount;
3151 is_inf_internal |= ((maxcount == REG_INFTY
3152 && (minnext + deltanext) > 0)
3153 || deltanext == I32_MAX);
3154 is_inf |= is_inf_internal;
3155 delta += (minnext + deltanext) * maxcount - minnext * mincount;
3157 /* Try powerful optimization CURLYX => CURLYN. */
3158 if ( OP(oscan) == CURLYX && data
3159 && data->flags & SF_IN_PAR
3160 && !(data->flags & SF_HAS_EVAL)
3161 && !deltanext && minnext == 1 ) {
3162 /* Try to optimize to CURLYN. */
3163 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
3164 regnode * const nxt1 = nxt;
3171 if (!strchr((const char*)PL_simple,OP(nxt))
3172 && !(PL_regkind[OP(nxt)] == EXACT
3173 && STR_LEN(nxt) == 1))
3179 if (OP(nxt) != CLOSE)
3181 if (RExC_open_parens) {
3182 RExC_open_parens[ARG(nxt1)-1]=oscan; /*open->CURLYM*/
3183 RExC_close_parens[ARG(nxt1)-1]=nxt+2; /*close->while*/
3185 /* Now we know that nxt2 is the only contents: */
3186 oscan->flags = (U8)ARG(nxt);
3188 OP(nxt1) = NOTHING; /* was OPEN. */
3191 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
3192 NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
3193 NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
3194 OP(nxt) = OPTIMIZED; /* was CLOSE. */
3195 OP(nxt + 1) = OPTIMIZED; /* was count. */
3196 NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
3201 /* Try optimization CURLYX => CURLYM. */
3202 if ( OP(oscan) == CURLYX && data
3203 && !(data->flags & SF_HAS_PAR)
3204 && !(data->flags & SF_HAS_EVAL)
3205 && !deltanext /* atom is fixed width */
3206 && minnext != 0 /* CURLYM can't handle zero width */
3208 /* XXXX How to optimize if data == 0? */
3209 /* Optimize to a simpler form. */
3210 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
3214 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
3215 && (OP(nxt2) != WHILEM))
3217 OP(nxt2) = SUCCEED; /* Whas WHILEM */
3218 /* Need to optimize away parenths. */
3219 if (data->flags & SF_IN_PAR) {
3220 /* Set the parenth number. */
3221 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
3223 if (OP(nxt) != CLOSE)
3224 FAIL("Panic opt close");
3225 oscan->flags = (U8)ARG(nxt);
3226 if (RExC_open_parens) {
3227 RExC_open_parens[ARG(nxt1)-1]=oscan; /*open->CURLYM*/
3228 RExC_close_parens[ARG(nxt1)-1]=nxt2+1; /*close->NOTHING*/
3230 OP(nxt1) = OPTIMIZED; /* was OPEN. */
3231 OP(nxt) = OPTIMIZED; /* was CLOSE. */
3234 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
3235 OP(nxt + 1) = OPTIMIZED; /* was count. */
3236 NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
3237 NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
3240 while ( nxt1 && (OP(nxt1) != WHILEM)) {
3241 regnode *nnxt = regnext(nxt1);
3244 if (reg_off_by_arg[OP(nxt1)])
3245 ARG_SET(nxt1, nxt2 - nxt1);
3246 else if (nxt2 - nxt1 < U16_MAX)
3247 NEXT_OFF(nxt1) = nxt2 - nxt1;
3249 OP(nxt) = NOTHING; /* Cannot beautify */
3254 /* Optimize again: */
3255 study_chunk(pRExC_state, &nxt1, minlenp, &deltanext, nxt,
3256 NULL, stopparen, recursed, NULL, 0,depth+1);
3261 else if ((OP(oscan) == CURLYX)
3262 && (flags & SCF_WHILEM_VISITED_POS)
3263 /* See the comment on a similar expression above.
3264 However, this time it not a subexpression
3265 we care about, but the expression itself. */
3266 && (maxcount == REG_INFTY)
3267 && data && ++data->whilem_c < 16) {
3268 /* This stays as CURLYX, we can put the count/of pair. */
3269 /* Find WHILEM (as in regexec.c) */
3270 regnode *nxt = oscan + NEXT_OFF(oscan);
3272 if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
3274 PREVOPER(nxt)->flags = (U8)(data->whilem_c
3275 | (RExC_whilem_seen << 4)); /* On WHILEM */
3277 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
3279 if (flags & SCF_DO_SUBSTR) {
3280 SV *last_str = NULL;
3281 int counted = mincount != 0;
3283 if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
3284 #if defined(SPARC64_GCC_WORKAROUND)
3287 const char *s = NULL;
3290 if (pos_before >= data->last_start_min)
3293 b = data->last_start_min;
3296 s = SvPV_const(data->last_found, l);
3297 old = b - data->last_start_min;
3300 I32 b = pos_before >= data->last_start_min
3301 ? pos_before : data->last_start_min;
3303 const char * const s = SvPV_const(data->last_found, l);
3304 I32 old = b - data->last_start_min;
3308 old = utf8_hop((U8*)s, old) - (U8*)s;
3311 /* Get the added string: */
3312 last_str = newSVpvn(s + old, l);
3314 SvUTF8_on(last_str);
3315 if (deltanext == 0 && pos_before == b) {
3316 /* What was added is a constant string */
3318 SvGROW(last_str, (mincount * l) + 1);
3319 repeatcpy(SvPVX(last_str) + l,
3320 SvPVX_const(last_str), l, mincount - 1);
3321 SvCUR_set(last_str, SvCUR(last_str) * mincount);
3322 /* Add additional parts. */
3323 SvCUR_set(data->last_found,
3324 SvCUR(data->last_found) - l);
3325 sv_catsv(data->last_found, last_str);
3327 SV * sv = data->last_found;
3329 SvUTF8(sv) && SvMAGICAL(sv) ?
3330 mg_find(sv, PERL_MAGIC_utf8) : NULL;
3331 if (mg && mg->mg_len >= 0)
3332 mg->mg_len += CHR_SVLEN(last_str) - l;
3334 data->last_end += l * (mincount - 1);
3337 /* start offset must point into the last copy */
3338 data->last_start_min += minnext * (mincount - 1);
3339 data->last_start_max += is_inf ? I32_MAX
3340 : (maxcount - 1) * (minnext + data->pos_delta);
3343 /* It is counted once already... */
3344 data->pos_min += minnext * (mincount - counted);
3345 data->pos_delta += - counted * deltanext +
3346 (minnext + deltanext) * maxcount - minnext * mincount;
3347 if (mincount != maxcount) {
3348 /* Cannot extend fixed substrings found inside
3350 SCAN_COMMIT(pRExC_state,data,minlenp);
3351 if (mincount && last_str) {
3352 SV * const sv = data->last_found;
3353 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
3354 mg_find(sv, PERL_MAGIC_utf8) : NULL;
3358 sv_setsv(sv, last_str);
3359 data->last_end = data->pos_min;
3360 data->last_start_min =
3361 data->pos_min - CHR_SVLEN(last_str);
3362 data->last_start_max = is_inf
3364 : data->pos_min + data->pos_delta
3365 - CHR_SVLEN(last_str);
3367 data->longest = &(data->longest_float);
3369 SvREFCNT_dec(last_str);
3371 if (data && (fl & SF_HAS_EVAL))
3372 data->flags |= SF_HAS_EVAL;
3373 optimize_curly_tail:
3374 if (OP(oscan) != CURLYX) {
3375 while (PL_regkind[OP(next = regnext(oscan))] == NOTHING
3377 NEXT_OFF(oscan) += NEXT_OFF(next);
3380 default: /* REF and CLUMP only? */
3381 if (flags & SCF_DO_SUBSTR) {
3382 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
3383 data->longest = &(data->longest_float);
3385 is_inf = is_inf_internal = 1;
3386 if (flags & SCF_DO_STCLASS_OR)
3387 cl_anything(pRExC_state, data->start_class);
3388 flags &= ~SCF_DO_STCLASS;
3392 else if (OP(scan) == LNBREAK) {
3393 if (flags & SCF_DO_STCLASS) {
3395 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
3396 if (flags & SCF_DO_STCLASS_AND) {
3397 for (value = 0; value < 256; value++)
3398 if (!is_VERTWS_cp(value))
3399 ANYOF_BITMAP_CLEAR(data->start_class, value);
3402 for (value = 0; value < 256; value++)
3403 if (is_VERTWS_cp(value))
3404 ANYOF_BITMAP_SET(data->start_class, value);
3406 if (flags & SCF_DO_STCLASS_OR)
3407 cl_and(data->start_class, and_withp);
3408 flags &= ~SCF_DO_STCLASS;
3412 if (flags & SCF_DO_SUBSTR) {
3413 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
3415 data->pos_delta += 1;
3416 data->longest = &(data->longest_float);
3420 else if (OP(scan) == FOLDCHAR) {
3421 int d = ARG(scan)==0xDF ? 1 : 2;
3422 flags &= ~SCF_DO_STCLASS;
3425 if (flags & SCF_DO_SUBSTR) {
3426 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
3428 data->pos_delta += d;
3429 data->longest = &(data->longest_float);
3432 else if (strchr((const char*)PL_simple,OP(scan))) {
3435 if (flags & SCF_DO_SUBSTR) {
3436 SCAN_COMMIT(pRExC_state,data,minlenp);
3440 if (flags & SCF_DO_STCLASS) {
3441 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
3443 /* Some of the logic below assumes that switching
3444 locale on will only add false positives. */
3445 switch (PL_regkind[OP(scan)]) {
3449 /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */
3450 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
3451 cl_anything(pRExC_state, data->start_class);
3454 if (OP(scan) == SANY)
3456 if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */
3457 value = (ANYOF_BITMAP_TEST(data->start_class,'\n')
3458 || (data->start_class->flags & ANYOF_CLASS));
3459 cl_anything(pRExC_state, data->start_class);
3461 if (flags & SCF_DO_STCLASS_AND || !value)
3462 ANYOF_BITMAP_CLEAR(data->start_class,'\n');
3465 if (flags & SCF_DO_STCLASS_AND)
3466 cl_and(data->start_class,
3467 (struct regnode_charclass_class*)scan);
3469 cl_or(pRExC_state, data->start_class,
3470 (struct regnode_charclass_class*)scan);
3473 if (flags & SCF_DO_STCLASS_AND) {
3474 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3475 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
3476 for (value = 0; value < 256; value++)
3477 if (!isALNUM(value))
3478 ANYOF_BITMAP_CLEAR(data->start_class, value);
3482 if (data->start_class->flags & ANYOF_LOCALE)
3483 ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
3485 for (value = 0; value < 256; value++)
3487 ANYOF_BITMAP_SET(data->start_class, value);
3492 if (flags & SCF_DO_STCLASS_AND) {
3493 if (data->start_class->flags & ANYOF_LOCALE)
3494 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
3497 ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
3498 data->start_class->flags |= ANYOF_LOCALE;
3502 if (flags & SCF_DO_STCLASS_AND) {
3503 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3504 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
3505 for (value = 0; value < 256; value++)
3507 ANYOF_BITMAP_CLEAR(data->start_class, value);
3511 if (data->start_class->flags & ANYOF_LOCALE)
3512 ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
3514 for (value = 0; value < 256; value++)
3515 if (!isALNUM(value))
3516 ANYOF_BITMAP_SET(data->start_class, value);
3521 if (flags & SCF_DO_STCLASS_AND) {
3522 if (data->start_class->flags & ANYOF_LOCALE)
3523 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
3526 data->start_class->flags |= ANYOF_LOCALE;
3527 ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
3531 if (flags & SCF_DO_STCLASS_AND) {
3532 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3533 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
3534 for (value = 0; value < 256; value++)
3535 if (!isSPACE(value))
3536 ANYOF_BITMAP_CLEAR(data->start_class, value);
3540 if (data->start_class->flags & ANYOF_LOCALE)
3541 ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
3543 for (value = 0; value < 256; value++)
3545 ANYOF_BITMAP_SET(data->start_class, value);
3550 if (flags & SCF_DO_STCLASS_AND) {
3551 if (data->start_class->flags & ANYOF_LOCALE)
3552 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
3555 data->start_class->flags |= ANYOF_LOCALE;
3556 ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
3560 if (flags & SCF_DO_STCLASS_AND) {
3561 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3562 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
3563 for (value = 0; value < 256; value++)
3565 ANYOF_BITMAP_CLEAR(data->start_class, value);
3569 if (data->start_class->flags & ANYOF_LOCALE)
3570 ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
3572 for (value = 0; value < 256; value++)
3573 if (!isSPACE(value))
3574 ANYOF_BITMAP_SET(data->start_class, value);
3579 if (flags & SCF_DO_STCLASS_AND) {
3580 if (data->start_class->flags & ANYOF_LOCALE) {
3581 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
3582 for (value = 0; value < 256; value++)
3583 if (!isSPACE(value))
3584 ANYOF_BITMAP_CLEAR(data->start_class, value);
3588 data->start_class->flags |= ANYOF_LOCALE;
3589 ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
3593 if (flags & SCF_DO_STCLASS_AND) {
3594 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT);
3595 for (value = 0; value < 256; value++)
3596 if (!isDIGIT(value))
3597 ANYOF_BITMAP_CLEAR(data->start_class, value);
3600 if (data->start_class->flags & ANYOF_LOCALE)
3601 ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT);
3603 for (value = 0; value < 256; value++)
3605 ANYOF_BITMAP_SET(data->start_class, value);
3610 if (flags & SCF_DO_STCLASS_AND) {
3611 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_DIGIT);
3612 for (value = 0; value < 256; value++)
3614 ANYOF_BITMAP_CLEAR(data->start_class, value);
3617 if (data->start_class->flags & ANYOF_LOCALE)
3618 ANYOF_CLASS_SET(data->start_class,ANYOF_NDIGIT);
3620 for (value = 0; value < 256; value++)
3621 if (!isDIGIT(value))
3622 ANYOF_BITMAP_SET(data->start_class, value);
3626 CASE_SYNST_FNC(VERTWS);
3627 CASE_SYNST_FNC(HORIZWS);
3630 if (flags & SCF_DO_STCLASS_OR)
3631 cl_and(data->start_class, and_withp);
3632 flags &= ~SCF_DO_STCLASS;
3635 else if (PL_regkind[OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
3636 data->flags |= (OP(scan) == MEOL
3640 else if ( PL_regkind[OP(scan)] == BRANCHJ
3641 /* Lookbehind, or need to calculate parens/evals/stclass: */
3642 && (scan->flags || data || (flags & SCF_DO_STCLASS))
3643 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
3644 if ( !PERL_ENABLE_POSITIVE_ASSERTION_STUDY
3645 || OP(scan) == UNLESSM )
3647 /* Negative Lookahead/lookbehind
3648 In this case we can't do fixed string optimisation.
3651 I32 deltanext, minnext, fake = 0;
3653 struct regnode_charclass_class intrnl;
3656 data_fake.flags = 0;
3658 data_fake.whilem_c = data->whilem_c;
3659 data_fake.last_closep = data->last_closep;
3662 data_fake.last_closep = &fake;
3663 data_fake.pos_delta = delta;
3664 if ( flags & SCF_DO_STCLASS && !scan->flags
3665 && OP(scan) == IFMATCH ) { /* Lookahead */
3666 cl_init(pRExC_state, &intrnl);
3667 data_fake.start_class = &intrnl;
3668 f |= SCF_DO_STCLASS_AND;
3670 if (flags & SCF_WHILEM_VISITED_POS)
3671 f |= SCF_WHILEM_VISITED_POS;
3672 next = regnext(scan);
3673 nscan = NEXTOPER(NEXTOPER(scan));
3674 minnext = study_chunk(pRExC_state, &nscan, minlenp, &deltanext,
3675 last, &data_fake, stopparen, recursed, NULL, f, depth+1);
3678 FAIL("Variable length lookbehind not implemented");
3680 else if (minnext > (I32)U8_MAX) {
3681 FAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
3683 scan->flags = (U8)minnext;
3686 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
3688 if (data_fake.flags & SF_HAS_EVAL)
3689 data->flags |= SF_HAS_EVAL;
3690 data->whilem_c = data_fake.whilem_c;
3692 if (f & SCF_DO_STCLASS_AND) {
3693 const int was = (data->start_class->flags & ANYOF_EOS);
3695 cl_and(data->start_class, &intrnl);
3697 data->start_class->flags |= ANYOF_EOS;
3700 #if PERL_ENABLE_POSITIVE_ASSERTION_STUDY
3702 /* Positive Lookahead/lookbehind
3703 In this case we can do fixed string optimisation,
3704 but we must be careful about it. Note in the case of
3705 lookbehind the positions will be offset by the minimum
3706 length of the pattern, something we won't know about
3707 until after the recurse.
3709 I32 deltanext, fake = 0;
3711 struct regnode_charclass_class intrnl;
3713 /* We use SAVEFREEPV so that when the full compile
3714 is finished perl will clean up the allocated
3715 minlens when its all done. This was we don't
3716 have to worry about freeing them when we know
3717 they wont be used, which would be a pain.
3720 Newx( minnextp, 1, I32 );
3721 SAVEFREEPV(minnextp);
3724 StructCopy(data, &data_fake, scan_data_t);
3725 if ((flags & SCF_DO_SUBSTR) && data->last_found) {
3728 SCAN_COMMIT(pRExC_state, &data_fake,minlenp);
3729 data_fake.last_found=newSVsv(data->last_found);
3733 data_fake.last_closep = &fake;
3734 data_fake.flags = 0;
3735 data_fake.pos_delta = delta;
3737 data_fake.flags |= SF_IS_INF;
3738 if ( flags & SCF_DO_STCLASS && !scan->flags
3739 && OP(scan) == IFMATCH ) { /* Lookahead */
3740 cl_init(pRExC_state, &intrnl);
3741 data_fake.start_class = &intrnl;
3742 f |= SCF_DO_STCLASS_AND;
3744 if (flags & SCF_WHILEM_VISITED_POS)
3745 f |= SCF_WHILEM_VISITED_POS;
3746 next = regnext(scan);
3747 nscan = NEXTOPER(NEXTOPER(scan));
3749 *minnextp = study_chunk(pRExC_state, &nscan, minnextp, &deltanext,
3750 last, &data_fake, stopparen, recursed, NULL, f,depth+1);
3753 FAIL("Variable length lookbehind not implemented");
3755 else if (*minnextp > (I32)U8_MAX) {
3756 FAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
3758 scan->flags = (U8)*minnextp;
3763 if (f & SCF_DO_STCLASS_AND) {
3764 const int was = (data->start_class->flags & ANYOF_EOS);
3766 cl_and(data->start_class, &intrnl);
3768 data->start_class->flags |= ANYOF_EOS;
3771 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
3773 if (data_fake.flags & SF_HAS_EVAL)
3774 data->flags |= SF_HAS_EVAL;
3775 data->whilem_c = data_fake.whilem_c;
3776 if ((flags & SCF_DO_SUBSTR) && data_fake.last_found) {
3777 if (RExC_rx->minlen<*minnextp)
3778 RExC_rx->minlen=*minnextp;
3779 SCAN_COMMIT(pRExC_state, &data_fake, minnextp);
3780 SvREFCNT_dec(data_fake.last_found);
3782 if ( data_fake.minlen_fixed != minlenp )
3784 data->offset_fixed= data_fake.offset_fixed;
3785 data->minlen_fixed= data_fake.minlen_fixed;
3786 data->lookbehind_fixed+= scan->flags;
3788 if ( data_fake.minlen_float != minlenp )
3790 data->minlen_float= data_fake.minlen_float;
3791 data->offset_float_min=data_fake.offset_float_min;
3792 data->offset_float_max=data_fake.offset_float_max;
3793 data->lookbehind_float+= scan->flags;
3802 else if (OP(scan) == OPEN) {
3803 if (stopparen != (I32)ARG(scan))
3806 else if (OP(scan) == CLOSE) {
3807 if (stopparen == (I32)ARG(scan)) {
3810 if ((I32)ARG(scan) == is_par) {
3811 next = regnext(scan);
3813 if ( next && (OP(next) != WHILEM) && next < last)
3814 is_par = 0; /* Disable optimization */
3817 *(data->last_closep) = ARG(scan);
3819 else if (OP(scan) == EVAL) {
3821 data->flags |= SF_HAS_EVAL;
3823 else if ( PL_regkind[OP(scan)] == ENDLIKE ) {
3824 if (flags & SCF_DO_SUBSTR) {
3825 SCAN_COMMIT(pRExC_state,data,minlenp);
3826 flags &= ~SCF_DO_SUBSTR;
3828 if (data && OP(scan)==ACCEPT) {
3829 data->flags |= SCF_SEEN_ACCEPT;
3834 else if (OP(scan) == LOGICAL && scan->flags == 2) /* Embedded follows */
3836 if (flags & SCF_DO_SUBSTR) {
3837 SCAN_COMMIT(pRExC_state,data,minlenp);
3838 data->longest = &(data->longest_float);
3840 is_inf = is_inf_internal = 1;
3841 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
3842 cl_anything(pRExC_state, data->start_class);
3843 flags &= ~SCF_DO_STCLASS;
3845 else if (OP(scan) == GPOS) {
3846 if (!(RExC_rx->extflags & RXf_GPOS_FLOAT) &&
3847 !(delta || is_inf || (data && data->pos_delta)))
3849 if (!(RExC_rx->extflags & RXf_ANCH) && (flags & SCF_DO_SUBSTR))
3850 RExC_rx->extflags |= RXf_ANCH_GPOS;
3851 if (RExC_rx->gofs < (U32)min)
3852 RExC_rx->gofs = min;
3854 RExC_rx->extflags |= RXf_GPOS_FLOAT;
3858 #ifdef TRIE_STUDY_OPT
3859 #ifdef FULL_TRIE_STUDY
3860 else if (PL_regkind[OP(scan)] == TRIE) {
3861 /* NOTE - There is similar code to this block above for handling
3862 BRANCH nodes on the initial study. If you change stuff here
3864 regnode *trie_node= scan;
3865 regnode *tail= regnext(scan);
3866 reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
3867 I32 max1 = 0, min1 = I32_MAX;
3868 struct regnode_charclass_class accum;
3870 if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
3871 SCAN_COMMIT(pRExC_state, data,minlenp); /* Cannot merge strings after this. */
3872 if (flags & SCF_DO_STCLASS)
3873 cl_init_zero(pRExC_state, &accum);
3879 const regnode *nextbranch= NULL;
3882 for ( word=1 ; word <= trie->wordcount ; word++)
3884 I32 deltanext=0, minnext=0, f = 0, fake;
3885 struct regnode_charclass_class this_class;
3887 data_fake.flags = 0;
3889 data_fake.whilem_c = data->whilem_c;
3890 data_fake.last_closep = data->last_closep;
3893 data_fake.last_closep = &fake;
3894 data_fake.pos_delta = delta;
3895 if (flags & SCF_DO_STCLASS) {
3896 cl_init(pRExC_state, &this_class);
3897 data_fake.start_class = &this_class;
3898 f = SCF_DO_STCLASS_AND;
3900 if (flags & SCF_WHILEM_VISITED_POS)
3901 f |= SCF_WHILEM_VISITED_POS;
3903 if (trie->jump[word]) {
3905 nextbranch = trie_node + trie->jump[0];
3906 scan= trie_node + trie->jump[word];
3907 /* We go from the jump point to the branch that follows
3908 it. Note this means we need the vestigal unused branches
3909 even though they arent otherwise used.
3911 minnext = study_chunk(pRExC_state, &scan, minlenp,
3912 &deltanext, (regnode *)nextbranch, &data_fake,
3913 stopparen, recursed, NULL, f,depth+1);
3915 if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
3916 nextbranch= regnext((regnode*)nextbranch);
3918 if (min1 > (I32)(minnext + trie->minlen))
3919 min1 = minnext + trie->minlen;
3920 if (max1 < (I32)(minnext + deltanext + trie->maxlen))
3921 max1 = minnext + deltanext + trie->maxlen;
3922 if (deltanext == I32_MAX)
3923 is_inf = is_inf_internal = 1;
3925 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
3927 if (data_fake.flags & SCF_SEEN_ACCEPT) {
3928 if ( stopmin > min + min1)
3929 stopmin = min + min1;
3930 flags &= ~SCF_DO_SUBSTR;
3932 data->flags |= SCF_SEEN_ACCEPT;
3935 if (data_fake.flags & SF_HAS_EVAL)
3936 data->flags |= SF_HAS_EVAL;
3937 data->whilem_c = data_fake.whilem_c;
3939 if (flags & SCF_DO_STCLASS)
3940 cl_or(pRExC_state, &accum, &this_class);
3943 if (flags & SCF_DO_SUBSTR) {
3944 data->pos_min += min1;
3945 data->pos_delta += max1 - min1;
3946 if (max1 != min1 || is_inf)
3947 data->longest = &(data->longest_float);
3950 delta += max1 - min1;
3951 if (flags & SCF_DO_STCLASS_OR) {
3952 cl_or(pRExC_state, data->start_class, &accum);
3954 cl_and(data->start_class, and_withp);
3955 flags &= ~SCF_DO_STCLASS;
3958 else if (flags & SCF_DO_STCLASS_AND) {
3960 cl_and(data->start_class, &accum);
3961 flags &= ~SCF_DO_STCLASS;
3964 /* Switch to OR mode: cache the old value of
3965 * data->start_class */
3967 StructCopy(data->start_class, and_withp,
3968 struct regnode_charclass_class);
3969 flags &= ~SCF_DO_STCLASS_AND;
3970 StructCopy(&accum, data->start_class,
3971 struct regnode_charclass_class);
3972 flags |= SCF_DO_STCLASS_OR;
3973 data->start_class->flags |= ANYOF_EOS;
3980 else if (PL_regkind[OP(scan)] == TRIE) {
3981 reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
3984 min += trie->minlen;
3985 delta += (trie->maxlen - trie->minlen);
3986 flags &= ~SCF_DO_STCLASS; /* xxx */
3987 if (flags & SCF_DO_SUBSTR) {
3988 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
3989 data->pos_min += trie->minlen;
3990 data->pos_delta += (trie->maxlen - trie->minlen);
3991 if (trie->maxlen != trie->minlen)
3992 data->longest = &(data->longest_float);
3994 if (trie->jump) /* no more substrings -- for now /grr*/
3995 flags &= ~SCF_DO_SUBSTR;
3997 #endif /* old or new */
3998 #endif /* TRIE_STUDY_OPT */
4000 /* Else: zero-length, ignore. */
4001 scan = regnext(scan);
4006 stopparen = frame->stop;
4007 frame = frame->prev;
4008 goto fake_study_recurse;
4013 DEBUG_STUDYDATA("pre-fin:",data,depth);
4016 *deltap = is_inf_internal ? I32_MAX : delta;
4017 if (flags & SCF_DO_SUBSTR && is_inf)
4018 data->pos_delta = I32_MAX - data->pos_min;
4019 if (is_par > (I32)U8_MAX)
4021 if (is_par && pars==1 && data) {
4022 data->flags |= SF_IN_PAR;
4023 data->flags &= ~SF_HAS_PAR;
4025 else if (pars && data) {
4026 data->flags |= SF_HAS_PAR;
4027 data->flags &= ~SF_IN_PAR;
4029 if (flags & SCF_DO_STCLASS_OR)
4030 cl_and(data->start_class, and_withp);
4031 if (flags & SCF_TRIE_RESTUDY)
4032 data->flags |= SCF_TRIE_RESTUDY;
4034 DEBUG_STUDYDATA("post-fin:",data,depth);
4036 return min < stopmin ? min : stopmin;
4040 S_add_data(RExC_state_t *pRExC_state, U32 n, const char *s)
4042 U32 count = RExC_rxi->data ? RExC_rxi->data->count : 0;
4044 Renewc(RExC_rxi->data,
4045 sizeof(*RExC_rxi->data) + sizeof(void*) * (count + n - 1),
4046 char, struct reg_data);
4048 Renew(RExC_rxi->data->what, count + n, U8);
4050 Newx(RExC_rxi->data->what, n, U8);
4051 RExC_rxi->data->count = count + n;
4052 Copy(s, RExC_rxi->data->what + count, n, U8);
4056 /*XXX: todo make this not included in a non debugging perl */
4057 #ifndef PERL_IN_XSUB_RE
4059 Perl_reginitcolors(pTHX)
4062 const char * const s = PerlEnv_getenv("PERL_RE_COLORS");
4064 char *t = savepv(s);
4068 t = strchr(t, '\t');
4074 PL_colors[i] = t = (char *)"";
4079 PL_colors[i++] = (char *)"";
4086 #ifdef TRIE_STUDY_OPT
4087 #define CHECK_RESTUDY_GOTO \
4089 (data.flags & SCF_TRIE_RESTUDY) \
4093 #define CHECK_RESTUDY_GOTO
4097 - pregcomp - compile a regular expression into internal code
4099 * We can't allocate space until we know how big the compiled form will be,
4100 * but we can't compile it (and thus know how big it is) until we've got a
4101 * place to put the code. So we cheat: we compile it twice, once with code
4102 * generation turned off and size counting turned on, and once "for real".
4103 * This also means that we don't allocate space until we are sure that the
4104 * thing really will compile successfully, and we never have to move the
4105 * code and thus invalidate pointers into it. (Note that it has to be in
4106 * one piece because free() must be able to free it all.) [NB: not true in perl]
4108 * Beware that the optimization-preparation code in here knows about some
4109 * of the structure of the compiled regexp. [I'll say.]
4114 #ifndef PERL_IN_XSUB_RE
4115 #define RE_ENGINE_PTR &PL_core_reg_engine
4117 extern const struct regexp_engine my_reg_engine;
4118 #define RE_ENGINE_PTR &my_reg_engine
4121 #ifndef PERL_IN_XSUB_RE
4123 Perl_pregcomp(pTHX_ const SV * const pattern, const U32 flags)
4126 HV * const table = GvHV(PL_hintgv);
4127 /* Dispatch a request to compile a regexp to correct
4130 SV **ptr= hv_fetchs(table, "regcomp", FALSE);
4131 GET_RE_DEBUG_FLAGS_DECL;
4132 if (ptr && SvIOK(*ptr) && SvIV(*ptr)) {
4133 const regexp_engine *eng=INT2PTR(regexp_engine*,SvIV(*ptr));
4135 PerlIO_printf(Perl_debug_log, "Using engine %"UVxf"\n",
4138 return CALLREGCOMP_ENG(eng, pattern, flags);
4141 return Perl_re_compile(aTHX_ pattern, flags);
4146 Perl_re_compile(pTHX_ const SV * const pattern, const U32 pm_flags)
4150 register regexp_internal *ri;
4152 char* exp = SvPV((SV*)pattern, plen);
4153 char* xend = exp + plen;
4160 RExC_state_t RExC_state;
4161 RExC_state_t * const pRExC_state = &RExC_state;
4162 #ifdef TRIE_STUDY_OPT
4164 RExC_state_t copyRExC_state;
4166 GET_RE_DEBUG_FLAGS_DECL;
4167 DEBUG_r(if (!PL_colorset) reginitcolors());
4169 RExC_utf8 = RExC_orig_utf8 = pm_flags & RXf_UTF8;
4172 SV *dsv= sv_newmortal();
4173 RE_PV_QUOTED_DECL(s, RExC_utf8,
4174 dsv, exp, plen, 60);
4175 PerlIO_printf(Perl_debug_log, "%sCompiling REx%s %s\n",
4176 PL_colors[4],PL_colors[5],s);
4181 RExC_flags = pm_flags;
4185 RExC_seen_zerolen = *exp == '^' ? -1 : 0;
4186 RExC_seen_evals = 0;
4189 /* First pass: determine size, legality. */
4197 RExC_emit = &PL_regdummy;
4198 RExC_whilem_seen = 0;
4199 RExC_charnames = NULL;
4200 RExC_open_parens = NULL;
4201 RExC_close_parens = NULL;
4203 RExC_paren_names = NULL;
4205 RExC_paren_name_list = NULL;
4207 RExC_recurse = NULL;
4208 RExC_recurse_count = 0;
4210 #if 0 /* REGC() is (currently) a NOP at the first pass.
4211 * Clever compilers notice this and complain. --jhi */
4212 REGC((U8)REG_MAGIC, (char*)RExC_emit);
4214 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log, "Starting first pass (sizing)\n"));
4215 if (reg(pRExC_state, 0, &flags,1) == NULL) {
4216 RExC_precomp = NULL;
4219 if (RExC_utf8 && !RExC_orig_utf8) {
4220 /* It's possible to write a regexp in ascii that represents Unicode
4221 codepoints outside of the byte range, such as via \x{100}. If we
4222 detect such a sequence we have to convert the entire pattern to utf8
4223 and then recompile, as our sizing calculation will have been based
4224 on 1 byte == 1 character, but we will need to use utf8 to encode
4225 at least some part of the pattern, and therefore must convert the whole
4227 XXX: somehow figure out how to make this less expensive...
4230 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log,
4231 "UTF8 mismatch! Converting to utf8 for resizing and compile\n"));
4232 exp = (char*)Perl_bytes_to_utf8(aTHX_ (U8*)exp, &len);
4234 RExC_orig_utf8 = RExC_utf8;
4236 goto redo_first_pass;
4239 PerlIO_printf(Perl_debug_log,
4240 "Required size %"IVdf" nodes\n"
4241 "Starting second pass (creation)\n",
4244 RExC_lastparse=NULL;
4246 /* Small enough for pointer-storage convention?
4247 If extralen==0, this means that we will not need long jumps. */
4248 if (RExC_size >= 0x10000L && RExC_extralen)
4249 RExC_size += RExC_extralen;
4252 if (RExC_whilem_seen > 15)
4253 RExC_whilem_seen = 15;
4255 /* Allocate space and zero-initialize. Note, the two step process
4256 of zeroing when in debug mode, thus anything assigned has to
4257 happen after that */
4258 Newxz(r, 1, regexp);
4259 Newxc(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode),
4260 char, regexp_internal);
4261 if ( r == NULL || ri == NULL )
4262 FAIL("Regexp out of space");
4264 /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
4265 Zero(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode), char);
4267 /* bulk initialize base fields with 0. */
4268 Zero(ri, sizeof(regexp_internal), char);
4271 /* non-zero initialization begins here */
4273 r->engine= RE_ENGINE_PTR;
4276 r->extflags = pm_flags;
4278 bool has_p = ((r->extflags & RXf_PMf_KEEPCOPY) == RXf_PMf_KEEPCOPY);
4279 bool has_minus = ((r->extflags & RXf_PMf_STD_PMMOD) != RXf_PMf_STD_PMMOD);
4280 bool has_runon = ((RExC_seen & REG_SEEN_RUN_ON_COMMENT)==REG_SEEN_RUN_ON_COMMENT);
4281 U16 reganch = (U16)((r->extflags & RXf_PMf_STD_PMMOD) >> 12);
4282 const char *fptr = STD_PAT_MODS; /*"msix"*/
4284 r->wraplen = r->prelen + has_minus + has_p + has_runon
4285 + (sizeof(STD_PAT_MODS) - 1)
4286 + (sizeof("(?:)") - 1);
4288 Newx(r->wrapped, r->wraplen + 1, char );
4292 *p++ = KEEPCOPY_PAT_MOD; /*'p'*/
4294 char *r = p + (sizeof(STD_PAT_MODS) - 1) + has_minus - 1;
4295 char *colon = r + 1;
4298 while((ch = *fptr++)) {
4312 Copy(RExC_precomp, p, r->prelen, char);
4322 r->nparens = RExC_npar - 1; /* set early to validate backrefs */
4324 if (RExC_seen & REG_SEEN_RECURSE) {
4325 Newxz(RExC_open_parens, RExC_npar,regnode *);
4326 SAVEFREEPV(RExC_open_parens);
4327 Newxz(RExC_close_parens,RExC_npar,regnode *);
4328 SAVEFREEPV(RExC_close_parens);
4331 /* Useful during FAIL. */
4332 #ifdef RE_TRACK_PATTERN_OFFSETS
4333 Newxz(ri->u.offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
4334 DEBUG_OFFSETS_r(PerlIO_printf(Perl_debug_log,
4335 "%s %"UVuf" bytes for offset annotations.\n",
4336 ri->u.offsets ? "Got" : "Couldn't get",
4337 (UV)((2*RExC_size+1) * sizeof(U32))));
4339 SetProgLen(ri,RExC_size);
4343 /* Second pass: emit code. */
4344 RExC_flags = pm_flags; /* don't let top level (?i) bleed */
4349 RExC_emit_start = ri->program;
4350 RExC_emit = ri->program;
4351 RExC_emit_bound = ri->program + RExC_size + 1;
4353 /* Store the count of eval-groups for security checks: */
4354 RExC_rx->seen_evals = RExC_seen_evals;
4355 REGC((U8)REG_MAGIC, (char*) RExC_emit++);
4356 if (reg(pRExC_state, 0, &flags,1) == NULL) {
4360 /* XXXX To minimize changes to RE engine we always allocate
4361 3-units-long substrs field. */
4362 Newx(r->substrs, 1, struct reg_substr_data);
4363 if (RExC_recurse_count) {
4364 Newxz(RExC_recurse,RExC_recurse_count,regnode *);
4365 SAVEFREEPV(RExC_recurse);
4369 r->minlen = minlen = sawplus = sawopen = 0;
4370 Zero(r->substrs, 1, struct reg_substr_data);
4372 #ifdef TRIE_STUDY_OPT
4375 DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log,"Restudying\n"));
4377 RExC_state = copyRExC_state;
4378 if (seen & REG_TOP_LEVEL_BRANCHES)
4379 RExC_seen |= REG_TOP_LEVEL_BRANCHES;
4381 RExC_seen &= ~REG_TOP_LEVEL_BRANCHES;
4382 if (data.last_found) {
4383 SvREFCNT_dec(data.longest_fixed);
4384 SvREFCNT_dec(data.longest_float);
4385 SvREFCNT_dec(data.last_found);
4387 StructCopy(&zero_scan_data, &data, scan_data_t);
4389 StructCopy(&zero_scan_data, &data, scan_data_t);
4390 copyRExC_state = RExC_state;
4393 StructCopy(&zero_scan_data, &data, scan_data_t);
4396 /* Dig out information for optimizations. */
4397 r->extflags = RExC_flags; /* was pm_op */
4398 /*dmq: removed as part of de-PMOP: pm->op_pmflags = RExC_flags; */
4401 r->extflags |= RXf_UTF8; /* Unicode in it? */
4402 ri->regstclass = NULL;
4403 if (RExC_naughty >= 10) /* Probably an expensive pattern. */
4404 r->intflags |= PREGf_NAUGHTY;
4405 scan = ri->program + 1; /* First BRANCH. */
4407 /* testing for BRANCH here tells us whether there is "must appear"
4408 data in the pattern. If there is then we can use it for optimisations */
4409 if (!(RExC_seen & REG_TOP_LEVEL_BRANCHES)) { /* Only one top-level choice. */
4411 STRLEN longest_float_length, longest_fixed_length;
4412 struct regnode_charclass_class ch_class; /* pointed to by data */
4414 I32 last_close = 0; /* pointed to by data */
4415 regnode *first= scan;
4416 regnode *first_next= regnext(first);
4418 /* Skip introductions and multiplicators >= 1. */
4419 while ((OP(first) == OPEN && (sawopen = 1)) ||
4420 /* An OR of *one* alternative - should not happen now. */
4421 (OP(first) == BRANCH && OP(first_next) != BRANCH) ||
4422 /* for now we can't handle lookbehind IFMATCH*/
4423 (OP(first) == IFMATCH && !first->flags) ||
4424 (OP(first) == PLUS) ||
4425 (OP(first) == MINMOD) ||
4426 /* An {n,m} with n>0 */
4427 (PL_regkind[OP(first)] == CURLY && ARG1(first) > 0) ||
4428 (OP(first) == NOTHING && PL_regkind[OP(first_next)] != END ))
4431 if (OP(first) == PLUS)
4434 first += regarglen[OP(first)];
4435 if (OP(first) == IFMATCH) {
4436 first = NEXTOPER(first);
4437 first += EXTRA_STEP_2ARGS;
4438 } else /* XXX possible optimisation for /(?=)/ */
4439 first = NEXTOPER(first);
4440 first_next= regnext(first);
4443 /* Starting-point info. */
4445 DEBUG_PEEP("first:",first,0);
4446 /* Ignore EXACT as we deal with it later. */
4447 if (PL_regkind[OP(first)] == EXACT) {
4448 if (OP(first) == EXACT)
4449 NOOP; /* Empty, get anchored substr later. */
4450 else if ((OP(first) == EXACTF || OP(first) == EXACTFL))
4451 ri->regstclass = first;
4454 else if (PL_regkind[OP(first)] == TRIE &&
4455 ((reg_trie_data *)ri->data->data[ ARG(first) ])->minlen>0)
4458 /* this can happen only on restudy */
4459 if ( OP(first) == TRIE ) {
4460 struct regnode_1 *trieop = (struct regnode_1 *)
4461 PerlMemShared_calloc(1, sizeof(struct regnode_1));
4462 StructCopy(first,trieop,struct regnode_1);
4463 trie_op=(regnode *)trieop;
4465 struct regnode_charclass *trieop = (struct regnode_charclass *)
4466 PerlMemShared_calloc(1, sizeof(struct regnode_charclass));
4467 StructCopy(first,trieop,struct regnode_charclass);
4468 trie_op=(regnode *)trieop;
4471 make_trie_failtable(pRExC_state, (regnode *)first, trie_op, 0);
4472 ri->regstclass = trie_op;
4475 else if (strchr((const char*)PL_simple,OP(first)))
4476 ri->regstclass = first;
4477 else if (PL_regkind[OP(first)] == BOUND ||
4478 PL_regkind[OP(first)] == NBOUND)
4479 ri->regstclass = first;
4480 else if (PL_regkind[OP(first)] == BOL) {
4481 r->extflags |= (OP(first) == MBOL
4483 : (OP(first) == SBOL
4486 first = NEXTOPER(first);
4489 else if (OP(first) == GPOS) {
4490 r->extflags |= RXf_ANCH_GPOS;
4491 first = NEXTOPER(first);
4494 else if ((!sawopen || !RExC_sawback) &&
4495 (OP(first) == STAR &&
4496 PL_regkind[OP(NEXTOPER(first))] == REG_ANY) &&
4497 !(r->extflags & RXf_ANCH) && !(RExC_seen & REG_SEEN_EVAL))
4499 /* turn .* into ^.* with an implied $*=1 */
4501 (OP(NEXTOPER(first)) == REG_ANY)
4504 r->extflags |= type;
4505 r->intflags |= PREGf_IMPLICIT;
4506 first = NEXTOPER(first);
4509 if (sawplus && (!sawopen || !RExC_sawback)
4510 && !(RExC_seen & REG_SEEN_EVAL)) /* May examine pos and $& */
4511 /* x+ must match at the 1st pos of run of x's */
4512 r->intflags |= PREGf_SKIP;
4514 /* Scan is after the zeroth branch, first is atomic matcher. */
4515 #ifdef TRIE_STUDY_OPT
4518 PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
4519 (IV)(first - scan + 1))
4523 PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
4524 (IV)(first - scan + 1))
4530 * If there's something expensive in the r.e., find the
4531 * longest literal string that must appear and make it the
4532 * regmust. Resolve ties in favor of later strings, since
4533 * the regstart check works with the beginning of the r.e.
4534 * and avoiding duplication strengthens checking. Not a
4535 * strong reason, but sufficient in the absence of others.
4536 * [Now we resolve ties in favor of the earlier string if
4537 * it happens that c_offset_min has been invalidated, since the
4538 * earlier string may buy us something the later one won't.]
4541 data.longest_fixed = newSVpvs("");
4542 data.longest_float = newSVpvs("");
4543 data.last_found = newSVpvs("");
4544 data.longest = &(data.longest_fixed);
4546 if (!ri->regstclass) {
4547 cl_init(pRExC_state, &ch_class);
4548 data.start_class = &ch_class;
4549 stclass_flag = SCF_DO_STCLASS_AND;
4550 } else /* XXXX Check for BOUND? */
4552 data.last_closep = &last_close;
4554 minlen = study_chunk(pRExC_state, &first, &minlen, &fake, scan + RExC_size, /* Up to end */
4555 &data, -1, NULL, NULL,
4556 SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag,0);
4562 if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
4563 && data.last_start_min == 0 && data.last_end > 0
4564 && !RExC_seen_zerolen
4565 && !(RExC_seen & REG_SEEN_VERBARG)
4566 && (!(RExC_seen & REG_SEEN_GPOS) || (r->extflags & RXf_ANCH_GPOS)))
4567 r->extflags |= RXf_CHECK_ALL;
4568 scan_commit(pRExC_state, &data,&minlen,0);
4569 SvREFCNT_dec(data.last_found);
4571 /* Note that code very similar to this but for anchored string
4572 follows immediately below, changes may need to be made to both.
4575 longest_float_length = CHR_SVLEN(data.longest_float);
4576 if (longest_float_length
4577 || (data.flags & SF_FL_BEFORE_EOL
4578 && (!(data.flags & SF_FL_BEFORE_MEOL)
4579 || (RExC_flags & RXf_PMf_MULTILINE))))
4583 if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */
4584 && data.offset_fixed == data.offset_float_min
4585 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
4586 goto remove_float; /* As in (a)+. */
4588 /* copy the information about the longest float from the reg_scan_data
4589 over to the program. */
4590 if (SvUTF8(data.longest_float)) {
4591 r->float_utf8 = data.longest_float;
4592 r->float_substr = NULL;
4594 r->float_substr = data.longest_float;
4595 r->float_utf8 = NULL;
4597 /* float_end_shift is how many chars that must be matched that
4598 follow this item. We calculate it ahead of time as once the
4599 lookbehind offset is added in we lose the ability to correctly
4601 ml = data.minlen_float ? *(data.minlen_float)
4602 : (I32)longest_float_length;
4603 r->float_end_shift = ml - data.offset_float_min
4604 - longest_float_length + (SvTAIL(data.longest_float) != 0)
4605 + data.lookbehind_float;
4606 r->float_min_offset = data.offset_float_min - data.lookbehind_float;
4607 r->float_max_offset = data.offset_float_max;
4608 if (data.offset_float_max < I32_MAX) /* Don't offset infinity */
4609 r->float_max_offset -= data.lookbehind_float;
4611 t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
4612 && (!(data.flags & SF_FL_BEFORE_MEOL)
4613 || (RExC_flags & RXf_PMf_MULTILINE)));
4614 fbm_compile(data.longest_float, t ? FBMcf_TAIL : 0);
4618 r->float_substr = r->float_utf8 = NULL;
4619 SvREFCNT_dec(data.longest_float);
4620 longest_float_length = 0;
4623 /* Note that code very similar to this but for floating string
4624 is immediately above, changes may need to be made to both.
4627 longest_fixed_length = CHR_SVLEN(data.longest_fixed);
4628 if (longest_fixed_length
4629 || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
4630 && (!(data.flags & SF_FIX_BEFORE_MEOL)
4631 || (RExC_flags & RXf_PMf_MULTILINE))))
4635 /* copy the information about the longest fixed
4636 from the reg_scan_data over to the program. */
4637 if (SvUTF8(data.longest_fixed)) {
4638 r->anchored_utf8 = data.longest_fixed;
4639 r->anchored_substr = NULL;
4641 r->anchored_substr = data.longest_fixed;
4642 r->anchored_utf8 = NULL;
4644 /* fixed_end_shift is how many chars that must be matched that
4645 follow this item. We calculate it ahead of time as once the
4646 lookbehind offset is added in we lose the ability to correctly
4648 ml = data.minlen_fixed ? *(data.minlen_fixed)
4649 : (I32)longest_fixed_length;
4650 r->anchored_end_shift = ml - data.offset_fixed
4651 - longest_fixed_length + (SvTAIL(data.longest_fixed) != 0)
4652 + data.lookbehind_fixed;
4653 r->anchored_offset = data.offset_fixed - data.lookbehind_fixed;
4655 t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
4656 && (!(data.flags & SF_FIX_BEFORE_MEOL)
4657 || (RExC_flags & RXf_PMf_MULTILINE)));
4658 fbm_compile(data.longest_fixed, t ? FBMcf_TAIL : 0);
4661 r->anchored_substr = r->anchored_utf8 = NULL;
4662 SvREFCNT_dec(data.longest_fixed);
4663 longest_fixed_length = 0;
4666 && (OP(ri->regstclass) == REG_ANY || OP(ri->regstclass) == SANY))
4667 ri->regstclass = NULL;
4668 if ((!(r->anchored_substr || r->anchored_utf8) || r->anchored_offset)
4670 && !(data.start_class->flags & ANYOF_EOS)
4671 && !cl_is_anything(data.start_class))
4673 const U32 n = add_data(pRExC_state, 1, "f");
4675 Newx(RExC_rxi->data->data[n], 1,
4676 struct regnode_charclass_class);
4677 StructCopy(data.start_class,
4678 (struct regnode_charclass_class*)RExC_rxi->data->data[n],
4679 struct regnode_charclass_class);
4680 ri->regstclass = (regnode*)RExC_rxi->data->data[n];
4681 r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
4682 DEBUG_COMPILE_r({ SV *sv = sv_newmortal();
4683 regprop(r, sv, (regnode*)data.start_class);
4684 PerlIO_printf(Perl_debug_log,
4685 "synthetic stclass \"%s\".\n",
4686 SvPVX_const(sv));});
4689 /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
4690 if (longest_fixed_length > longest_float_length) {
4691 r->check_end_shift = r->anchored_end_shift;
4692 r->check_substr = r->anchored_substr;
4693 r->check_utf8 = r->anchored_utf8;
4694 r->check_offset_min = r->check_offset_max = r->anchored_offset;
4695 if (r->extflags & RXf_ANCH_SINGLE)
4696 r->extflags |= RXf_NOSCAN;
4699 r->check_end_shift = r->float_end_shift;
4700 r->check_substr = r->float_substr;
4701 r->check_utf8 = r->float_utf8;
4702 r->check_offset_min = r->float_min_offset;
4703 r->check_offset_max = r->float_max_offset;
4705 /* XXXX Currently intuiting is not compatible with ANCH_GPOS.
4706 This should be changed ASAP! */
4707 if ((r->check_substr || r->check_utf8) && !(r->extflags & RXf_ANCH_GPOS)) {
4708 r->extflags |= RXf_USE_INTUIT;
4709 if (SvTAIL(r->check_substr ? r->check_substr : r->check_utf8))
4710 r->extflags |= RXf_INTUIT_TAIL;
4712 /* XXX Unneeded? dmq (shouldn't as this is handled elsewhere)
4713 if ( (STRLEN)minlen < longest_float_length )
4714 minlen= longest_float_length;
4715 if ( (STRLEN)minlen < longest_fixed_length )
4716 minlen= longest_fixed_length;
4720 /* Several toplevels. Best we can is to set minlen. */
4722 struct regnode_charclass_class ch_class;
4725 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log, "\nMulti Top Level\n"));
4727 scan = ri->program + 1;
4728 cl_init(pRExC_state, &ch_class);
4729 data.start_class = &ch_class;
4730 data.last_closep = &last_close;
4733 minlen = study_chunk(pRExC_state, &scan, &minlen, &fake, scan + RExC_size,
4734 &data, -1, NULL, NULL, SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS,0);
4738 r->check_substr = r->check_utf8 = r->anchored_substr = r->anchored_utf8
4739 = r->float_substr = r->float_utf8 = NULL;
4740 if (!(data.start_class->flags & ANYOF_EOS)
4741 && !cl_is_anything(data.start_class))
4743 const U32 n = add_data(pRExC_state, 1, "f");
4745 Newx(RExC_rxi->data->data[n], 1,
4746 struct regnode_charclass_class);
4747 StructCopy(data.start_class,
4748 (struct regnode_charclass_class*)RExC_rxi->data->data[n],
4749 struct regnode_charclass_class);
4750 ri->regstclass = (regnode*)RExC_rxi->data->data[n];
4751 r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
4752 DEBUG_COMPILE_r({ SV* sv = sv_newmortal();
4753 regprop(r, sv, (regnode*)data.start_class);
4754 PerlIO_printf(Perl_debug_log,
4755 "synthetic stclass \"%s\".\n",
4756 SvPVX_const(sv));});
4760 /* Guard against an embedded (?=) or (?<=) with a longer minlen than
4761 the "real" pattern. */
4763 PerlIO_printf(Perl_debug_log,"minlen: %"IVdf" r->minlen:%"IVdf"\n",
4764 (IV)minlen, (IV)r->minlen);
4766 r->minlenret = minlen;
4767 if (r->minlen < minlen)
4770 if (RExC_seen & REG_SEEN_GPOS)
4771 r->extflags |= RXf_GPOS_SEEN;
4772 if (RExC_seen & REG_SEEN_LOOKBEHIND)
4773 r->extflags |= RXf_LOOKBEHIND_SEEN;
4774 if (RExC_seen & REG_SEEN_EVAL)
4775 r->extflags |= RXf_EVAL_SEEN;
4776 if (RExC_seen & REG_SEEN_CANY)
4777 r->extflags |= RXf_CANY_SEEN;
4778 if (RExC_seen & REG_SEEN_VERBARG)
4779 r->intflags |= PREGf_VERBARG_SEEN;
4780 if (RExC_seen & REG_SEEN_CUTGROUP)
4781 r->intflags |= PREGf_CUTGROUP_SEEN;
4782 if (RExC_paren_names)
4783 r->paren_names = (HV*)SvREFCNT_inc(RExC_paren_names);
4785 r->paren_names = NULL;
4787 #ifdef STUPID_PATTERN_CHECKS
4789 r->extflags |= RXf_NULL;
4790 if (r->extflags & RXf_SPLIT && r->prelen == 1 && r->precomp[0] == ' ')
4791 /* XXX: this should happen BEFORE we compile */
4792 r->extflags |= (RXf_SKIPWHITE|RXf_WHITE);
4793 else if (r->prelen == 3 && memEQ("\\s+", r->precomp, 3))
4794 r->extflags |= RXf_WHITE;
4795 else if (r->prelen == 1 && r->precomp[0] == '^')
4796 r->extflags |= RXf_START_ONLY;
4798 if (r->extflags & RXf_SPLIT && r->prelen == 1 && r->precomp[0] == ' ')
4799 /* XXX: this should happen BEFORE we compile */
4800 r->extflags |= (RXf_SKIPWHITE|RXf_WHITE);
4802 regnode *first = ri->program + 1;
4804 U8 nop = OP(NEXTOPER(first));
4806 if (PL_regkind[fop] == NOTHING && nop == END)
4807 r->extflags |= RXf_NULL;
4808 else if (PL_regkind[fop] == BOL && nop == END)
4809 r->extflags |= RXf_START_ONLY;
4810 else if (fop == PLUS && nop ==SPACE && OP(regnext(first))==END)
4811 r->extflags |= RXf_WHITE;
4815 if (RExC_paren_names) {
4816 ri->name_list_idx = add_data( pRExC_state, 1, "p" );
4817 ri->data->data[ri->name_list_idx] = (void*)SvREFCNT_inc(RExC_paren_name_list);
4820 ri->name_list_idx = 0;
4822 if (RExC_recurse_count) {
4823 for ( ; RExC_recurse_count ; RExC_recurse_count-- ) {
4824 const regnode *scan = RExC_recurse[RExC_recurse_count-1];
4825 ARG2L_SET( scan, RExC_open_parens[ARG(scan)-1] - scan );
4828 Newxz(r->offs, RExC_npar, regexp_paren_pair);
4829 /* assume we don't need to swap parens around before we match */
4832 PerlIO_printf(Perl_debug_log,"Final program:\n");
4835 #ifdef RE_TRACK_PATTERN_OFFSETS
4836 DEBUG_OFFSETS_r(if (ri->u.offsets) {
4837 const U32 len = ri->u.offsets[0];
4839 GET_RE_DEBUG_FLAGS_DECL;
4840 PerlIO_printf(Perl_debug_log, "Offsets: [%"UVuf"]\n\t", (UV)ri->u.offsets[0]);
4841 for (i = 1; i <= len; i++) {
4842 if (ri->u.offsets[i*2-1] || ri->u.offsets[i*2])
4843 PerlIO_printf(Perl_debug_log, "%"UVuf":%"UVuf"[%"UVuf"] ",
4844 (UV)i, (UV)ri->u.offsets[i*2-1], (UV)ri->u.offsets[i*2]);
4846 PerlIO_printf(Perl_debug_log, "\n");
4852 #undef RE_ENGINE_PTR
4856 Perl_reg_named_buff(pTHX_ REGEXP * const rx, SV * const key, SV * const value,
4859 PERL_UNUSED_ARG(value);
4861 if (flags & RXapif_FETCH) {
4862 return reg_named_buff_fetch(rx, key, flags);
4863 } else if (flags & (RXapif_STORE | RXapif_DELETE | RXapif_CLEAR)) {
4864 Perl_croak(aTHX_ PL_no_modify);
4866 } else if (flags & RXapif_EXISTS) {
4867 return reg_named_buff_exists(rx, key, flags)
4870 } else if (flags & RXapif_REGNAMES) {
4871 return reg_named_buff_all(rx, flags);
4872 } else if (flags & (RXapif_SCALAR | RXapif_REGNAMES_COUNT)) {
4873 return reg_named_buff_scalar(rx, flags);
4875 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff", (int)flags);
4881 Perl_reg_named_buff_iter(pTHX_ REGEXP * const rx, const SV * const lastkey,
4884 PERL_UNUSED_ARG(lastkey);
4886 if (flags & RXapif_FIRSTKEY)
4887 return reg_named_buff_firstkey(rx, flags);
4888 else if (flags & RXapif_NEXTKEY)
4889 return reg_named_buff_nextkey(rx, flags);
4891 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_iter", (int)flags);
4897 Perl_reg_named_buff_fetch(pTHX_ REGEXP * const rx, SV * const namesv, const U32 flags)
4899 AV *retarray = NULL;
4901 if (flags & RXapif_ALL)
4904 if (rx && rx->paren_names) {
4905 HE *he_str = hv_fetch_ent( rx->paren_names, namesv, 0, 0 );
4908 SV* sv_dat=HeVAL(he_str);
4909 I32 *nums=(I32*)SvPVX(sv_dat);
4910 for ( i=0; i<SvIVX(sv_dat); i++ ) {
4911 if ((I32)(rx->nparens) >= nums[i]
4912 && rx->offs[nums[i]].start != -1
4913 && rx->offs[nums[i]].end != -1)
4916 CALLREG_NUMBUF_FETCH(rx,nums[i],ret);
4920 ret = newSVsv(&PL_sv_undef);
4923 SvREFCNT_inc_simple_void(ret);
4924 av_push(retarray, ret);
4928 return newRV((SV*)retarray);
4935 Perl_reg_named_buff_exists(pTHX_ REGEXP * const rx, SV * const key,
4938 if (rx && rx->paren_names) {
4939 if (flags & RXapif_ALL) {
4940 return hv_exists_ent(rx->paren_names, key, 0);
4942 SV *sv = CALLREG_NAMED_BUFF_FETCH(rx, key, flags);
4956 Perl_reg_named_buff_firstkey(pTHX_ REGEXP * const rx, const U32 flags)
4958 if ( rx && rx->paren_names ) {
4959 (void)hv_iterinit(rx->paren_names);
4961 return CALLREG_NAMED_BUFF_NEXTKEY(rx, NULL, flags & ~RXapif_FIRSTKEY);
4968 Perl_reg_named_buff_nextkey(pTHX_ REGEXP * const rx, const U32 flags)
4970 if (rx && rx->paren_names) {
4971 HV *hv = rx->paren_names;
4973 while ( (temphe = hv_iternext_flags(hv,0)) ) {
4976 SV* sv_dat = HeVAL(temphe);
4977 I32 *nums = (I32*)SvPVX(sv_dat);
4978 for ( i = 0; i < SvIVX(sv_dat); i++ ) {
4979 if ((I32)(rx->lastcloseparen) >= nums[i] &&
4980 rx->offs[nums[i]].start != -1 &&
4981 rx->offs[nums[i]].end != -1)
4987 if (parno || flags & RXapif_ALL) {
4989 char *pv = HePV(temphe, len);
4990 return newSVpvn(pv,len);
4998 Perl_reg_named_buff_scalar(pTHX_ REGEXP * const rx, const U32 flags)
5004 if (rx && rx->paren_names) {
5005 if (flags & (RXapif_ALL | RXapif_REGNAMES_COUNT)) {
5006 return newSViv(HvTOTALKEYS(rx->paren_names));
5007 } else if (flags & RXapif_ONE) {
5008 ret = CALLREG_NAMED_BUFF_ALL(rx, (flags | RXapif_REGNAMES));
5009 av = (AV*)SvRV(ret);
5010 length = av_len(av);
5011 return newSViv(length + 1);
5013 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_scalar", (int)flags);
5017 return &PL_sv_undef;
5021 Perl_reg_named_buff_all(pTHX_ REGEXP * const rx, const U32 flags)
5025 if (rx && rx->paren_names) {
5026 HV *hv= rx->paren_names;
5028 (void)hv_iterinit(hv);
5029 while ( (temphe = hv_iternext_flags(hv,0)) ) {
5032 SV* sv_dat = HeVAL(temphe);
5033 I32 *nums = (I32*)SvPVX(sv_dat);
5034 for ( i = 0; i < SvIVX(sv_dat); i++ ) {
5035 if ((I32)(rx->lastcloseparen) >= nums[i] &&
5036 rx->offs[nums[i]].start != -1 &&
5037 rx->offs[nums[i]].end != -1)
5043 if (parno || flags & RXapif_ALL) {
5045 char *pv = HePV(temphe, len);
5046 av_push(av, newSVpvn(pv,len));
5051 return newRV((SV*)av);
5055 Perl_reg_numbered_buff_fetch(pTHX_ REGEXP * const rx, const I32 paren, SV * const sv)
5062 sv_setsv(sv,&PL_sv_undef);
5066 if (paren == RX_BUFF_IDX_PREMATCH && rx->offs[0].start != -1) {
5068 i = rx->offs[0].start;
5072 if (paren == RX_BUFF_IDX_POSTMATCH && rx->offs[0].end != -1) {
5074 s = rx->subbeg + rx->offs[0].end;
5075 i = rx->sublen - rx->offs[0].end;
5078 if ( 0 <= paren && paren <= (I32)rx->nparens &&
5079 (s1 = rx->offs[paren].start) != -1 &&
5080 (t1 = rx->offs[paren].end) != -1)
5084 s = rx->subbeg + s1;
5086 sv_setsv(sv,&PL_sv_undef);
5089 assert(rx->sublen >= (s - rx->subbeg) + i );
5091 const int oldtainted = PL_tainted;
5093 sv_setpvn(sv, s, i);
5094 PL_tainted = oldtainted;
5095 if ( (rx->extflags & RXf_CANY_SEEN)
5096 ? (RX_MATCH_UTF8(rx)
5097 && (!i || is_utf8_string((U8*)s, i)))
5098 : (RX_MATCH_UTF8(rx)) )
5105 if (RX_MATCH_TAINTED(rx)) {
5106 if (SvTYPE(sv) >= SVt_PVMG) {
5107 MAGIC* const mg = SvMAGIC(sv);
5110 SvMAGIC_set(sv, mg->mg_moremagic);
5112 if ((mgt = SvMAGIC(sv))) {
5113 mg->mg_moremagic = mgt;
5114 SvMAGIC_set(sv, mg);
5124 sv_setsv(sv,&PL_sv_undef);
5130 Perl_reg_numbered_buff_store(pTHX_ REGEXP * const rx, const I32 paren,
5131 SV const * const value)
5133 PERL_UNUSED_ARG(rx);
5134 PERL_UNUSED_ARG(paren);
5135 PERL_UNUSED_ARG(value);
5138 Perl_croak(aTHX_ PL_no_modify);
5142 Perl_reg_numbered_buff_length(pTHX_ REGEXP * const rx, const SV * const sv,
5148 /* Some of this code was originally in C<Perl_magic_len> in F<mg.c> */
5150 /* $` / ${^PREMATCH} */
5151 case RX_BUFF_IDX_PREMATCH:
5152 if (rx->offs[0].start != -1) {
5153 i = rx->offs[0].start;
5161 /* $' / ${^POSTMATCH} */
5162 case RX_BUFF_IDX_POSTMATCH:
5163 if (rx->offs[0].end != -1) {
5164 i = rx->sublen - rx->offs[0].end;
5166 s1 = rx->offs[0].end;
5172 /* $& / ${^MATCH}, $1, $2, ... */
5174 if (paren <= (I32)rx->nparens &&
5175 (s1 = rx->offs[paren].start) != -1 &&
5176 (t1 = rx->offs[paren].end) != -1)
5181 if (ckWARN(WARN_UNINITIALIZED))
5182 report_uninit((SV*)sv);
5187 if (i > 0 && RX_MATCH_UTF8(rx)) {
5188 const char * const s = rx->subbeg + s1;
5193 if (is_utf8_string_loclen((U8*)s, i, &ep, &el))
5200 Perl_reg_qr_package(pTHX_ REGEXP * const rx)
5202 PERL_UNUSED_ARG(rx);
5203 return newSVpvs("Regexp");
5206 /* Scans the name of a named buffer from the pattern.
5207 * If flags is REG_RSN_RETURN_NULL returns null.
5208 * If flags is REG_RSN_RETURN_NAME returns an SV* containing the name
5209 * If flags is REG_RSN_RETURN_DATA returns the data SV* corresponding
5210 * to the parsed name as looked up in the RExC_paren_names hash.
5211 * If there is an error throws a vFAIL().. type exception.
5214 #define REG_RSN_RETURN_NULL 0
5215 #define REG_RSN_RETURN_NAME 1
5216 #define REG_RSN_RETURN_DATA 2
5219 S_reg_scan_name(pTHX_ RExC_state_t *pRExC_state, U32 flags) {
5220 char *name_start = RExC_parse;
5222 if (isIDFIRST_lazy_if(RExC_parse, UTF)) {
5223 /* skip IDFIRST by using do...while */
5226 RExC_parse += UTF8SKIP(RExC_parse);
5227 } while (isALNUM_utf8((U8*)RExC_parse));
5231 } while (isALNUM(*RExC_parse));
5235 SV* sv_name = sv_2mortal(Perl_newSVpvn(aTHX_ name_start,
5236 (int)(RExC_parse - name_start)));
5239 if ( flags == REG_RSN_RETURN_NAME)
5241 else if (flags==REG_RSN_RETURN_DATA) {
5244 if ( ! sv_name ) /* should not happen*/
5245 Perl_croak(aTHX_ "panic: no svname in reg_scan_name");
5246 if (RExC_paren_names)
5247 he_str = hv_fetch_ent( RExC_paren_names, sv_name, 0, 0 );
5249 sv_dat = HeVAL(he_str);
5251 vFAIL("Reference to nonexistent named group");
5255 Perl_croak(aTHX_ "panic: bad flag in reg_scan_name");
5262 #define DEBUG_PARSE_MSG(funcname) DEBUG_PARSE_r({ \
5263 int rem=(int)(RExC_end - RExC_parse); \
5272 if (RExC_lastparse!=RExC_parse) \
5273 PerlIO_printf(Perl_debug_log," >%.*s%-*s", \
5276 iscut ? "..." : "<" \
5279 PerlIO_printf(Perl_debug_log,"%16s",""); \
5282 num = RExC_size + 1; \
5284 num=REG_NODE_NUM(RExC_emit); \
5285 if (RExC_lastnum!=num) \
5286 PerlIO_printf(Perl_debug_log,"|%4d",num); \
5288 PerlIO_printf(Perl_debug_log,"|%4s",""); \
5289 PerlIO_printf(Perl_debug_log,"|%*s%-4s", \
5290 (int)((depth*2)), "", \
5294 RExC_lastparse=RExC_parse; \
5299 #define DEBUG_PARSE(funcname) DEBUG_PARSE_r({ \
5300 DEBUG_PARSE_MSG((funcname)); \
5301 PerlIO_printf(Perl_debug_log,"%4s","\n"); \
5303 #define DEBUG_PARSE_FMT(funcname,fmt,args) DEBUG_PARSE_r({ \
5304 DEBUG_PARSE_MSG((funcname)); \
5305 PerlIO_printf(Perl_debug_log,fmt "\n",args); \
5308 - reg - regular expression, i.e. main body or parenthesized thing
5310 * Caller must absorb opening parenthesis.
5312 * Combining parenthesis handling with the base level of regular expression
5313 * is a trifle forced, but the need to tie the tails of the branches to what
5314 * follows makes it hard to avoid.
5316 #define REGTAIL(x,y,z) regtail((x),(y),(z),depth+1)
5318 #define REGTAIL_STUDY(x,y,z) regtail_study((x),(y),(z),depth+1)
5320 #define REGTAIL_STUDY(x,y,z) regtail((x),(y),(z),depth+1)
5324 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
5325 /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
5328 register regnode *ret; /* Will be the head of the group. */
5329 register regnode *br;
5330 register regnode *lastbr;
5331 register regnode *ender = NULL;
5332 register I32 parno = 0;
5334 U32 oregflags = RExC_flags;
5335 bool have_branch = 0;
5337 I32 freeze_paren = 0;
5338 I32 after_freeze = 0;
5340 /* for (?g), (?gc), and (?o) warnings; warning
5341 about (?c) will warn about (?g) -- japhy */
5343 #define WASTED_O 0x01
5344 #define WASTED_G 0x02
5345 #define WASTED_C 0x04
5346 #define WASTED_GC (0x02|0x04)
5347 I32 wastedflags = 0x00;
5349 char * parse_start = RExC_parse; /* MJD */
5350 char * const oregcomp_parse = RExC_parse;
5352 GET_RE_DEBUG_FLAGS_DECL;
5353 DEBUG_PARSE("reg ");
5355 *flagp = 0; /* Tentatively. */
5358 /* Make an OPEN node, if parenthesized. */
5360 if ( *RExC_parse == '*') { /* (*VERB:ARG) */
5361 char *start_verb = RExC_parse;
5362 STRLEN verb_len = 0;
5363 char *start_arg = NULL;
5364 unsigned char op = 0;
5366 int internal_argval = 0; /* internal_argval is only useful if !argok */
5367 while ( *RExC_parse && *RExC_parse != ')' ) {
5368 if ( *RExC_parse == ':' ) {
5369 start_arg = RExC_parse + 1;
5375 verb_len = RExC_parse - start_verb;
5378 while ( *RExC_parse && *RExC_parse != ')' )
5380 if ( *RExC_parse != ')' )
5381 vFAIL("Unterminated verb pattern argument");
5382 if ( RExC_parse == start_arg )
5385 if ( *RExC_parse != ')' )
5386 vFAIL("Unterminated verb pattern");
5389 switch ( *start_verb ) {
5390 case 'A': /* (*ACCEPT) */
5391 if ( memEQs(start_verb,verb_len,"ACCEPT") ) {
5393 internal_argval = RExC_nestroot;
5396 case 'C': /* (*COMMIT) */
5397 if ( memEQs(start_verb,verb_len,"COMMIT") )
5400 case 'F': /* (*FAIL) */
5401 if ( verb_len==1 || memEQs(start_verb,verb_len,"FAIL") ) {
5406 case ':': /* (*:NAME) */
5407 case 'M': /* (*MARK:NAME) */
5408 if ( verb_len==0 || memEQs(start_verb,verb_len,"MARK") ) {
5413 case 'P': /* (*PRUNE) */
5414 if ( memEQs(start_verb,verb_len,"PRUNE") )
5417 case 'S': /* (*SKIP) */
5418 if ( memEQs(start_verb,verb_len,"SKIP") )
5421 case 'T': /* (*THEN) */
5422 /* [19:06] <TimToady> :: is then */
5423 if ( memEQs(start_verb,verb_len,"THEN") ) {
5425 RExC_seen |= REG_SEEN_CUTGROUP;
5431 vFAIL3("Unknown verb pattern '%.*s'",
5432 verb_len, start_verb);
5435 if ( start_arg && internal_argval ) {
5436 vFAIL3("Verb pattern '%.*s' may not have an argument",
5437 verb_len, start_verb);
5438 } else if ( argok < 0 && !start_arg ) {
5439 vFAIL3("Verb pattern '%.*s' has a mandatory argument",
5440 verb_len, start_verb);
5442 ret = reganode(pRExC_state, op, internal_argval);
5443 if ( ! internal_argval && ! SIZE_ONLY ) {
5445 SV *sv = newSVpvn( start_arg, RExC_parse - start_arg);
5446 ARG(ret) = add_data( pRExC_state, 1, "S" );
5447 RExC_rxi->data->data[ARG(ret)]=(void*)sv;
5454 if (!internal_argval)
5455 RExC_seen |= REG_SEEN_VERBARG;
5456 } else if ( start_arg ) {
5457 vFAIL3("Verb pattern '%.*s' may not have an argument",
5458 verb_len, start_verb);
5460 ret = reg_node(pRExC_state, op);
5462 nextchar(pRExC_state);
5465 if (*RExC_parse == '?') { /* (?...) */
5466 bool is_logical = 0;
5467 const char * const seqstart = RExC_parse;
5470 paren = *RExC_parse++;
5471 ret = NULL; /* For look-ahead/behind. */
5474 case 'P': /* (?P...) variants for those used to PCRE/Python */
5475 paren = *RExC_parse++;
5476 if ( paren == '<') /* (?P<...>) named capture */
5478 else if (paren == '>') { /* (?P>name) named recursion */
5479 goto named_recursion;
5481 else if (paren == '=') { /* (?P=...) named backref */
5482 /* this pretty much dupes the code for \k<NAME> in regatom(), if
5483 you change this make sure you change that */
5484 char* name_start = RExC_parse;
5486 SV *sv_dat = reg_scan_name(pRExC_state,
5487 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
5488 if (RExC_parse == name_start || *RExC_parse != ')')
5489 vFAIL2("Sequence %.3s... not terminated",parse_start);
5492 num = add_data( pRExC_state, 1, "S" );
5493 RExC_rxi->data->data[num]=(void*)sv_dat;
5494 SvREFCNT_inc_simple_void(sv_dat);
5497 ret = reganode(pRExC_state,
5498 (U8)(FOLD ? (LOC ? NREFFL : NREFF) : NREF),
5502 Set_Node_Offset(ret, parse_start+1);
5503 Set_Node_Cur_Length(ret); /* MJD */
5505 nextchar(pRExC_state);
5509 vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
5511 case '<': /* (?<...) */
5512 if (*RExC_parse == '!')
5514 else if (*RExC_parse != '=')
5520 case '\'': /* (?'...') */
5521 name_start= RExC_parse;
5522 svname = reg_scan_name(pRExC_state,
5523 SIZE_ONLY ? /* reverse test from the others */
5524 REG_RSN_RETURN_NAME :
5525 REG_RSN_RETURN_NULL);
5526 if (RExC_parse == name_start) {
5528 vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
5531 if (*RExC_parse != paren)
5532 vFAIL2("Sequence (?%c... not terminated",
5533 paren=='>' ? '<' : paren);
5537 if (!svname) /* shouldnt happen */
5539 "panic: reg_scan_name returned NULL");
5540 if (!RExC_paren_names) {
5541 RExC_paren_names= newHV();
5542 sv_2mortal((SV*)RExC_paren_names);
5544 RExC_paren_name_list= newAV();
5545 sv_2mortal((SV*)RExC_paren_name_list);
5548 he_str = hv_fetch_ent( RExC_paren_names, svname, 1, 0 );
5550 sv_dat = HeVAL(he_str);
5552 /* croak baby croak */
5554 "panic: paren_name hash element allocation failed");
5555 } else if ( SvPOK(sv_dat) ) {
5556 /* (?|...) can mean we have dupes so scan to check
5557 its already been stored. Maybe a flag indicating
5558 we are inside such a construct would be useful,
5559 but the arrays are likely to be quite small, so
5560 for now we punt -- dmq */
5561 IV count = SvIV(sv_dat);
5562 I32 *pv = (I32*)SvPVX(sv_dat);
5564 for ( i = 0 ; i < count ; i++ ) {
5565 if ( pv[i] == RExC_npar ) {
5571 pv = (I32*)SvGROW(sv_dat, SvCUR(sv_dat) + sizeof(I32)+1);
5572 SvCUR_set(sv_dat, SvCUR(sv_dat) + sizeof(I32));
5573 pv[count] = RExC_npar;
5577 (void)SvUPGRADE(sv_dat,SVt_PVNV);
5578 sv_setpvn(sv_dat, (char *)&(RExC_npar), sizeof(I32));
5583 if (!av_store(RExC_paren_name_list, RExC_npar, SvREFCNT_inc(svname)))
5584 SvREFCNT_dec(svname);
5587 /*sv_dump(sv_dat);*/
5589 nextchar(pRExC_state);
5591 goto capturing_parens;
5593 RExC_seen |= REG_SEEN_LOOKBEHIND;
5595 case '=': /* (?=...) */
5596 case '!': /* (?!...) */
5597 RExC_seen_zerolen++;
5598 if (*RExC_parse == ')') {
5599 ret=reg_node(pRExC_state, OPFAIL);
5600 nextchar(pRExC_state);
5604 case '|': /* (?|...) */
5605 /* branch reset, behave like a (?:...) except that
5606 buffers in alternations share the same numbers */
5608 after_freeze = freeze_paren = RExC_npar;
5610 case ':': /* (?:...) */
5611 case '>': /* (?>...) */
5613 case '$': /* (?$...) */
5614 case '@': /* (?@...) */
5615 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
5617 case '#': /* (?#...) */
5618 while (*RExC_parse && *RExC_parse != ')')
5620 if (*RExC_parse != ')')
5621 FAIL("Sequence (?#... not terminated");
5622 nextchar(pRExC_state);
5625 case '0' : /* (?0) */
5626 case 'R' : /* (?R) */
5627 if (*RExC_parse != ')')
5628 FAIL("Sequence (?R) not terminated");
5629 ret = reg_node(pRExC_state, GOSTART);
5630 *flagp |= POSTPONED;
5631 nextchar(pRExC_state);
5634 { /* named and numeric backreferences */
5636 case '&': /* (?&NAME) */
5637 parse_start = RExC_parse - 1;
5640 SV *sv_dat = reg_scan_name(pRExC_state,
5641 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
5642 num = sv_dat ? *((I32 *)SvPVX(sv_dat)) : 0;
5644 goto gen_recurse_regop;
5647 if (!(RExC_parse[0] >= '1' && RExC_parse[0] <= '9')) {
5649 vFAIL("Illegal pattern");
5651 goto parse_recursion;
5653 case '-': /* (?-1) */
5654 if (!(RExC_parse[0] >= '1' && RExC_parse[0] <= '9')) {
5655 RExC_parse--; /* rewind to let it be handled later */
5659 case '1': case '2': case '3': case '4': /* (?1) */
5660 case '5': case '6': case '7': case '8': case '9':
5663 num = atoi(RExC_parse);
5664 parse_start = RExC_parse - 1; /* MJD */
5665 if (*RExC_parse == '-')
5667 while (isDIGIT(*RExC_parse))
5669 if (*RExC_parse!=')')
5670 vFAIL("Expecting close bracket");
5673 if ( paren == '-' ) {
5675 Diagram of capture buffer numbering.
5676 Top line is the normal capture buffer numbers
5677 Botton line is the negative indexing as from
5681 /(a(x)y)(a(b(c(?-2)d)e)f)(g(h))/
5685 num = RExC_npar + num;
5688 vFAIL("Reference to nonexistent group");
5690 } else if ( paren == '+' ) {
5691 num = RExC_npar + num - 1;
5694 ret = reganode(pRExC_state, GOSUB, num);
5696 if (num > (I32)RExC_rx->nparens) {
5698 vFAIL("Reference to nonexistent group");
5700 ARG2L_SET( ret, RExC_recurse_count++);
5702 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
5703 "Recurse #%"UVuf" to %"IVdf"\n", (UV)ARG(ret), (IV)ARG2L(ret)));
5707 RExC_seen |= REG_SEEN_RECURSE;
5708 Set_Node_Length(ret, 1 + regarglen[OP(ret)]); /* MJD */
5709 Set_Node_Offset(ret, parse_start); /* MJD */
5711 *flagp |= POSTPONED;
5712 nextchar(pRExC_state);
5714 } /* named and numeric backreferences */
5717 case '?': /* (??...) */
5719 if (*RExC_parse != '{') {
5721 vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
5724 *flagp |= POSTPONED;
5725 paren = *RExC_parse++;
5727 case '{': /* (?{...}) */
5732 char *s = RExC_parse;
5734 RExC_seen_zerolen++;
5735 RExC_seen |= REG_SEEN_EVAL;
5736 while (count && (c = *RExC_parse)) {
5747 if (*RExC_parse != ')') {
5749 vFAIL("Sequence (?{...}) not terminated or not {}-balanced");
5753 OP_4tree *sop, *rop;
5754 SV * const sv = newSVpvn(s, RExC_parse - 1 - s);
5757 Perl_save_re_context(aTHX);
5758 rop = sv_compile_2op(sv, &sop, "re", &pad);
5759 sop->op_private |= OPpREFCOUNTED;
5760 /* re_dup will OpREFCNT_inc */
5761 OpREFCNT_set(sop, 1);
5764 n = add_data(pRExC_state, 3, "nop");
5765 RExC_rxi->data->data[n] = (void*)rop;
5766 RExC_rxi->data->data[n+1] = (void*)sop;
5767 RExC_rxi->data->data[n+2] = (void*)pad;
5770 else { /* First pass */
5771 if (PL_reginterp_cnt < ++RExC_seen_evals
5773 /* No compiled RE interpolated, has runtime
5774 components ===> unsafe. */
5775 FAIL("Eval-group not allowed at runtime, use re 'eval'");
5776 if (PL_tainting && PL_tainted)
5777 FAIL("Eval-group in insecure regular expression");
5778 #if PERL_VERSION > 8
5779 if (IN_PERL_COMPILETIME)
5784 nextchar(pRExC_state);
5786 ret = reg_node(pRExC_state, LOGICAL);
5789 REGTAIL(pRExC_state, ret, reganode(pRExC_state, EVAL, n));
5790 /* deal with the length of this later - MJD */
5793 ret = reganode(pRExC_state, EVAL, n);
5794 Set_Node_Length(ret, RExC_parse - parse_start + 1);
5795 Set_Node_Offset(ret, parse_start);
5798 case '(': /* (?(?{...})...) and (?(?=...)...) */
5801 if (RExC_parse[0] == '?') { /* (?(?...)) */
5802 if (RExC_parse[1] == '=' || RExC_parse[1] == '!'
5803 || RExC_parse[1] == '<'
5804 || RExC_parse[1] == '{') { /* Lookahead or eval. */
5807 ret = reg_node(pRExC_state, LOGICAL);
5810 REGTAIL(pRExC_state, ret, reg(pRExC_state, 1, &flag,depth+1));
5814 else if ( RExC_parse[0] == '<' /* (?(<NAME>)...) */
5815 || RExC_parse[0] == '\'' ) /* (?('NAME')...) */
5817 char ch = RExC_parse[0] == '<' ? '>' : '\'';
5818 char *name_start= RExC_parse++;
5820 SV *sv_dat=reg_scan_name(pRExC_state,
5821 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
5822 if (RExC_parse == name_start || *RExC_parse != ch)
5823 vFAIL2("Sequence (?(%c... not terminated",
5824 (ch == '>' ? '<' : ch));
5827 num = add_data( pRExC_state, 1, "S" );
5828 RExC_rxi->data->data[num]=(void*)sv_dat;
5829 SvREFCNT_inc_simple_void(sv_dat);
5831 ret = reganode(pRExC_state,NGROUPP,num);
5832 goto insert_if_check_paren;
5834 else if (RExC_parse[0] == 'D' &&
5835 RExC_parse[1] == 'E' &&
5836 RExC_parse[2] == 'F' &&
5837 RExC_parse[3] == 'I' &&
5838 RExC_parse[4] == 'N' &&
5839 RExC_parse[5] == 'E')
5841 ret = reganode(pRExC_state,DEFINEP,0);
5844 goto insert_if_check_paren;
5846 else if (RExC_parse[0] == 'R') {
5849 if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
5850 parno = atoi(RExC_parse++);
5851 while (isDIGIT(*RExC_parse))
5853 } else if (RExC_parse[0] == '&') {
5856 sv_dat = reg_scan_name(pRExC_state,
5857 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
5858 parno = sv_dat ? *((I32 *)SvPVX(sv_dat)) : 0;
5860 ret = reganode(pRExC_state,INSUBP,parno);
5861 goto insert_if_check_paren;
5863 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
5866 parno = atoi(RExC_parse++);
5868 while (isDIGIT(*RExC_parse))
5870 ret = reganode(pRExC_state, GROUPP, parno);
5872 insert_if_check_paren:
5873 if ((c = *nextchar(pRExC_state)) != ')')
5874 vFAIL("Switch condition not recognized");
5876 REGTAIL(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
5877 br = regbranch(pRExC_state, &flags, 1,depth+1);
5879 br = reganode(pRExC_state, LONGJMP, 0);
5881 REGTAIL(pRExC_state, br, reganode(pRExC_state, LONGJMP, 0));
5882 c = *nextchar(pRExC_state);
5887 vFAIL("(?(DEFINE)....) does not allow branches");
5888 lastbr = reganode(pRExC_state, IFTHEN, 0); /* Fake one for optimizer. */
5889 regbranch(pRExC_state, &flags, 1,depth+1);
5890 REGTAIL(pRExC_state, ret, lastbr);
5893 c = *nextchar(pRExC_state);
5898 vFAIL("Switch (?(condition)... contains too many branches");
5899 ender = reg_node(pRExC_state, TAIL);
5900 REGTAIL(pRExC_state, br, ender);
5902 REGTAIL(pRExC_state, lastbr, ender);
5903 REGTAIL(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
5906 REGTAIL(pRExC_state, ret, ender);
5907 RExC_size++; /* XXX WHY do we need this?!!
5908 For large programs it seems to be required
5909 but I can't figure out why. -- dmq*/
5913 vFAIL2("Unknown switch condition (?(%.2s", RExC_parse);
5917 RExC_parse--; /* for vFAIL to print correctly */
5918 vFAIL("Sequence (? incomplete");
5922 parse_flags: /* (?i) */
5924 U32 posflags = 0, negflags = 0;
5925 U32 *flagsp = &posflags;
5927 while (*RExC_parse) {
5928 /* && strchr("iogcmsx", *RExC_parse) */
5929 /* (?g), (?gc) and (?o) are useless here
5930 and must be globally applied -- japhy */
5931 switch (*RExC_parse) {
5932 CASE_STD_PMMOD_FLAGS_PARSE_SET(flagsp);
5933 case ONCE_PAT_MOD: /* 'o' */
5934 case GLOBAL_PAT_MOD: /* 'g' */
5935 if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
5936 const I32 wflagbit = *RExC_parse == 'o' ? WASTED_O : WASTED_G;
5937 if (! (wastedflags & wflagbit) ) {
5938 wastedflags |= wflagbit;
5941 "Useless (%s%c) - %suse /%c modifier",
5942 flagsp == &negflags ? "?-" : "?",
5944 flagsp == &negflags ? "don't " : "",
5951 case CONTINUE_PAT_MOD: /* 'c' */
5952 if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
5953 if (! (wastedflags & WASTED_C) ) {
5954 wastedflags |= WASTED_GC;
5957 "Useless (%sc) - %suse /gc modifier",
5958 flagsp == &negflags ? "?-" : "?",
5959 flagsp == &negflags ? "don't " : ""
5964 case KEEPCOPY_PAT_MOD: /* 'p' */
5965 if (flagsp == &negflags) {
5966 if (SIZE_ONLY && ckWARN(WARN_REGEXP))
5967 vWARN(RExC_parse + 1,"Useless use of (?-p)");
5969 *flagsp |= RXf_PMf_KEEPCOPY;
5973 if (flagsp == &negflags) {
5975 vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
5979 wastedflags = 0; /* reset so (?g-c) warns twice */
5985 RExC_flags |= posflags;
5986 RExC_flags &= ~negflags;
5988 oregflags |= posflags;
5989 oregflags &= ~negflags;
5991 nextchar(pRExC_state);
6002 vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
6007 }} /* one for the default block, one for the switch */
6014 ret = reganode(pRExC_state, OPEN, parno);
6017 RExC_nestroot = parno;
6018 if (RExC_seen & REG_SEEN_RECURSE
6019 && !RExC_open_parens[parno-1])
6021 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
6022 "Setting open paren #%"IVdf" to %d\n",
6023 (IV)parno, REG_NODE_NUM(ret)));
6024 RExC_open_parens[parno-1]= ret;
6027 Set_Node_Length(ret, 1); /* MJD */
6028 Set_Node_Offset(ret, RExC_parse); /* MJD */
6036 /* Pick up the branches, linking them together. */
6037 parse_start = RExC_parse; /* MJD */
6038 br = regbranch(pRExC_state, &flags, 1,depth+1);
6039 /* branch_len = (paren != 0); */
6043 if (*RExC_parse == '|') {
6044 if (!SIZE_ONLY && RExC_extralen) {
6045 reginsert(pRExC_state, BRANCHJ, br, depth+1);
6048 reginsert(pRExC_state, BRANCH, br, depth+1);
6049 Set_Node_Length(br, paren != 0);
6050 Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
6054 RExC_extralen += 1; /* For BRANCHJ-BRANCH. */
6056 else if (paren == ':') {
6057 *flagp |= flags&SIMPLE;
6059 if (is_open) { /* Starts with OPEN. */
6060 REGTAIL(pRExC_state, ret, br); /* OPEN -> first. */
6062 else if (paren != '?') /* Not Conditional */
6064 *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
6066 while (*RExC_parse == '|') {
6067 if (!SIZE_ONLY && RExC_extralen) {
6068 ender = reganode(pRExC_state, LONGJMP,0);
6069 REGTAIL(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
6072 RExC_extralen += 2; /* Account for LONGJMP. */
6073 nextchar(pRExC_state);
6075 if (RExC_npar > after_freeze)
6076 after_freeze = RExC_npar;
6077 RExC_npar = freeze_paren;
6079 br = regbranch(pRExC_state, &flags, 0, depth+1);
6083 REGTAIL(pRExC_state, lastbr, br); /* BRANCH -> BRANCH. */
6085 *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
6088 if (have_branch || paren != ':') {
6089 /* Make a closing node, and hook it on the end. */
6092 ender = reg_node(pRExC_state, TAIL);
6095 ender = reganode(pRExC_state, CLOSE, parno);
6096 if (!SIZE_ONLY && RExC_seen & REG_SEEN_RECURSE) {
6097 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
6098 "Setting close paren #%"IVdf" to %d\n",
6099 (IV)parno, REG_NODE_NUM(ender)));
6100 RExC_close_parens[parno-1]= ender;
6101 if (RExC_nestroot == parno)
6104 Set_Node_Offset(ender,RExC_parse+1); /* MJD */
6105 Set_Node_Length(ender,1); /* MJD */
6111 *flagp &= ~HASWIDTH;
6114 ender = reg_node(pRExC_state, SUCCEED);
6117 ender = reg_node(pRExC_state, END);
6119 assert(!RExC_opend); /* there can only be one! */
6124 REGTAIL(pRExC_state, lastbr, ender);
6126 if (have_branch && !SIZE_ONLY) {
6128 RExC_seen |= REG_TOP_LEVEL_BRANCHES;
6130 /* Hook the tails of the branches to the closing node. */
6131 for (br = ret; br; br = regnext(br)) {
6132 const U8 op = PL_regkind[OP(br)];
6134 REGTAIL_STUDY(pRExC_state, NEXTOPER(br), ender);
6136 else if (op == BRANCHJ) {
6137 REGTAIL_STUDY(pRExC_state, NEXTOPER(NEXTOPER(br)), ender);
6145 static const char parens[] = "=!<,>";
6147 if (paren && (p = strchr(parens, paren))) {
6148 U8 node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
6149 int flag = (p - parens) > 1;
6152 node = SUSPEND, flag = 0;
6153 reginsert(pRExC_state, node,ret, depth+1);
6154 Set_Node_Cur_Length(ret);
6155 Set_Node_Offset(ret, parse_start + 1);
6157 REGTAIL_STUDY(pRExC_state, ret, reg_node(pRExC_state, TAIL));
6161 /* Check for proper termination. */
6163 RExC_flags = oregflags;
6164 if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
6165 RExC_parse = oregcomp_parse;
6166 vFAIL("Unmatched (");
6169 else if (!paren && RExC_parse < RExC_end) {
6170 if (*RExC_parse == ')') {
6172 vFAIL("Unmatched )");
6175 FAIL("Junk on end of regexp"); /* "Can't happen". */
6179 RExC_npar = after_freeze;
6184 - regbranch - one alternative of an | operator
6186 * Implements the concatenation operator.
6189 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first, U32 depth)
6192 register regnode *ret;
6193 register regnode *chain = NULL;
6194 register regnode *latest;
6195 I32 flags = 0, c = 0;
6196 GET_RE_DEBUG_FLAGS_DECL;
6197 DEBUG_PARSE("brnc");
6202 if (!SIZE_ONLY && RExC_extralen)
6203 ret = reganode(pRExC_state, BRANCHJ,0);
6205 ret = reg_node(pRExC_state, BRANCH);
6206 Set_Node_Length(ret, 1);
6210 if (!first && SIZE_ONLY)
6211 RExC_extralen += 1; /* BRANCHJ */
6213 *flagp = WORST; /* Tentatively. */
6216 nextchar(pRExC_state);
6217 while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
6219 latest = regpiece(pRExC_state, &flags,depth+1);
6220 if (latest == NULL) {
6221 if (flags & TRYAGAIN)
6225 else if (ret == NULL)
6227 *flagp |= flags&(HASWIDTH|POSTPONED);
6228 if (chain == NULL) /* First piece. */
6229 *flagp |= flags&SPSTART;
6232 REGTAIL(pRExC_state, chain, latest);
6237 if (chain == NULL) { /* Loop ran zero times. */
6238 chain = reg_node(pRExC_state, NOTHING);
6243 *flagp |= flags&SIMPLE;
6250 - regpiece - something followed by possible [*+?]
6252 * Note that the branching code sequences used for ? and the general cases
6253 * of * and + are somewhat optimized: they use the same NOTHING node as
6254 * both the endmarker for their branch list and the body of the last branch.
6255 * It might seem that this node could be dispensed with entirely, but the
6256 * endmarker role is not redundant.
6259 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
6262 register regnode *ret;
6264 register char *next;
6266 const char * const origparse = RExC_parse;
6268 I32 max = REG_INFTY;
6270 const char *maxpos = NULL;
6271 GET_RE_DEBUG_FLAGS_DECL;
6272 DEBUG_PARSE("piec");
6274 ret = regatom(pRExC_state, &flags,depth+1);
6276 if (flags & TRYAGAIN)
6283 if (op == '{' && regcurly(RExC_parse)) {
6285 parse_start = RExC_parse; /* MJD */
6286 next = RExC_parse + 1;
6287 while (isDIGIT(*next) || *next == ',') {
6296 if (*next == '}') { /* got one */
6300 min = atoi(RExC_parse);
6304 maxpos = RExC_parse;
6306 if (!max && *maxpos != '0')
6307 max = REG_INFTY; /* meaning "infinity" */
6308 else if (max >= REG_INFTY)
6309 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
6311 nextchar(pRExC_state);
6314 if ((flags&SIMPLE)) {
6315 RExC_naughty += 2 + RExC_naughty / 2;
6316 reginsert(pRExC_state, CURLY, ret, depth+1);
6317 Set_Node_Offset(ret, parse_start+1); /* MJD */
6318 Set_Node_Cur_Length(ret);
6321 regnode * const w = reg_node(pRExC_state, WHILEM);
6324 REGTAIL(pRExC_state, ret, w);
6325 if (!SIZE_ONLY && RExC_extralen) {
6326 reginsert(pRExC_state, LONGJMP,ret, depth+1);
6327 reginsert(pRExC_state, NOTHING,ret, depth+1);
6328 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
6330 reginsert(pRExC_state, CURLYX,ret, depth+1);
6332 Set_Node_Offset(ret, parse_start+1);
6333 Set_Node_Length(ret,
6334 op == '{' ? (RExC_parse - parse_start) : 1);
6336 if (!SIZE_ONLY && RExC_extralen)
6337 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
6338 REGTAIL(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
6340 RExC_whilem_seen++, RExC_extralen += 3;
6341 RExC_naughty += 4 + RExC_naughty; /* compound interest */
6349 if (max && max < min)
6350 vFAIL("Can't do {n,m} with n > m");
6352 ARG1_SET(ret, (U16)min);
6353 ARG2_SET(ret, (U16)max);
6365 #if 0 /* Now runtime fix should be reliable. */
6367 /* if this is reinstated, don't forget to put this back into perldiag:
6369 =item Regexp *+ operand could be empty at {#} in regex m/%s/
6371 (F) The part of the regexp subject to either the * or + quantifier
6372 could match an empty string. The {#} shows in the regular
6373 expression about where the problem was discovered.
6377 if (!(flags&HASWIDTH) && op != '?')
6378 vFAIL("Regexp *+ operand could be empty");
6381 parse_start = RExC_parse;
6382 nextchar(pRExC_state);
6384 *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
6386 if (op == '*' && (flags&SIMPLE)) {
6387 reginsert(pRExC_state, STAR, ret, depth+1);
6391 else if (op == '*') {
6395 else if (op == '+' && (flags&SIMPLE)) {
6396 reginsert(pRExC_state, PLUS, ret, depth+1);
6400 else if (op == '+') {
6404 else if (op == '?') {
6409 if (!SIZE_ONLY && !(flags&(HASWIDTH|POSTPONED)) && max > REG_INFTY/3 && ckWARN(WARN_REGEXP)) {
6411 "%.*s matches null string many times",
6412 (int)(RExC_parse >= origparse ? RExC_parse - origparse : 0),
6416 if (RExC_parse < RExC_end && *RExC_parse == '?') {
6417 nextchar(pRExC_state);
6418 reginsert(pRExC_state, MINMOD, ret, depth+1);
6419 REGTAIL(pRExC_state, ret, ret + NODE_STEP_REGNODE);
6421 #ifndef REG_ALLOW_MINMOD_SUSPEND
6424 if (RExC_parse < RExC_end && *RExC_parse == '+') {
6426 nextchar(pRExC_state);
6427 ender = reg_node(pRExC_state, SUCCEED);
6428 REGTAIL(pRExC_state, ret, ender);
6429 reginsert(pRExC_state, SUSPEND, ret, depth+1);
6431 ender = reg_node(pRExC_state, TAIL);
6432 REGTAIL(pRExC_state, ret, ender);
6436 if (RExC_parse < RExC_end && ISMULT2(RExC_parse)) {
6438 vFAIL("Nested quantifiers");
6445 /* reg_namedseq(pRExC_state,UVp)
6447 This is expected to be called by a parser routine that has
6448 recognized'\N' and needs to handle the rest. RExC_parse is
6449 expected to point at the first char following the N at the time
6452 If valuep is non-null then it is assumed that we are parsing inside
6453 of a charclass definition and the first codepoint in the resolved
6454 string is returned via *valuep and the routine will return NULL.
6455 In this mode if a multichar string is returned from the charnames
6456 handler a warning will be issued, and only the first char in the
6457 sequence will be examined. If the string returned is zero length
6458 then the value of *valuep is undefined and NON-NULL will
6459 be returned to indicate failure. (This will NOT be a valid pointer
6462 If value is null then it is assumed that we are parsing normal text
6463 and inserts a new EXACT node into the program containing the resolved
6464 string and returns a pointer to the new node. If the string is
6465 zerolength a NOTHING node is emitted.
6467 On success RExC_parse is set to the char following the endbrace.
6468 Parsing failures will generate a fatal errorvia vFAIL(...)
6470 NOTE: We cache all results from the charnames handler locally in
6471 the RExC_charnames hash (created on first use) to prevent a charnames
6472 handler from playing silly-buggers and returning a short string and
6473 then a long string for a given pattern. Since the regexp program
6474 size is calculated during an initial parse this would result
6475 in a buffer overrun so we cache to prevent the charname result from
6476 changing during the course of the parse.
6480 S_reg_namedseq(pTHX_ RExC_state_t *pRExC_state, UV *valuep)
6482 char * name; /* start of the content of the name */
6483 char * endbrace; /* endbrace following the name */
6486 STRLEN len; /* this has various purposes throughout the code */
6487 bool cached = 0; /* if this is true then we shouldn't refcount dev sv_str */
6488 regnode *ret = NULL;
6490 if (*RExC_parse != '{') {
6491 vFAIL("Missing braces on \\N{}");
6493 name = RExC_parse+1;
6494 endbrace = strchr(RExC_parse, '}');
6497 vFAIL("Missing right brace on \\N{}");
6499 RExC_parse = endbrace + 1;
6502 /* RExC_parse points at the beginning brace,
6503 endbrace points at the last */
6504 if ( name[0]=='U' && name[1]=='+' ) {
6505 /* its a "Unicode hex" notation {U+89AB} */
6506 I32 fl = PERL_SCAN_ALLOW_UNDERSCORES
6507 | PERL_SCAN_DISALLOW_PREFIX
6508 | (SIZE_ONLY ? PERL_SCAN_SILENT_ILLDIGIT : 0);
6511 len = (STRLEN)(endbrace - name - 2);
6512 cp = grok_hex(name + 2, &len, &fl, NULL);
6513 if ( len != (STRLEN)(endbrace - name - 2) ) {
6523 sv_str= newSVpvn(&string, 1);
6525 /* fetch the charnames handler for this scope */
6526 HV * const table = GvHV(PL_hintgv);
6528 hv_fetchs(table, "charnames", FALSE) :
6530 SV *cv= cvp ? *cvp : NULL;
6533 /* create an SV with the name as argument */
6534 sv_name = newSVpvn(name, endbrace - name);
6536 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
6537 vFAIL2("Constant(\\N{%s}) unknown: "
6538 "(possibly a missing \"use charnames ...\")",
6541 if (!cvp || !SvOK(*cvp)) { /* when $^H{charnames} = undef; */
6542 vFAIL2("Constant(\\N{%s}): "
6543 "$^H{charnames} is not defined",SvPVX(sv_name));
6548 if (!RExC_charnames) {
6549 /* make sure our cache is allocated */
6550 RExC_charnames = newHV();
6551 sv_2mortal((SV*)RExC_charnames);
6553 /* see if we have looked this one up before */
6554 he_str = hv_fetch_ent( RExC_charnames, sv_name, 0, 0 );
6556 sv_str = HeVAL(he_str);
6569 count= call_sv(cv, G_SCALAR);
6571 if (count == 1) { /* XXXX is this right? dmq */
6573 SvREFCNT_inc_simple_void(sv_str);
6581 if ( !sv_str || !SvOK(sv_str) ) {
6582 vFAIL2("Constant(\\N{%s}): Call to &{$^H{charnames}} "
6583 "did not return a defined value",SvPVX(sv_name));
6585 if (hv_store_ent( RExC_charnames, sv_name, sv_str, 0))
6590 char *p = SvPV(sv_str, len);
6593 if ( SvUTF8(sv_str) ) {
6594 *valuep = utf8_to_uvchr((U8*)p, &numlen);
6598 We have to turn on utf8 for high bit chars otherwise
6599 we get failures with
6601 "ss" =~ /[\N{LATIN SMALL LETTER SHARP S}]/i
6602 "SS" =~ /[\N{LATIN SMALL LETTER SHARP S}]/i
6604 This is different from what \x{} would do with the same
6605 codepoint, where the condition is > 0xFF.
6612 /* warn if we havent used the whole string? */
6614 if (numlen<len && SIZE_ONLY && ckWARN(WARN_REGEXP)) {
6616 "Ignoring excess chars from \\N{%s} in character class",
6620 } else if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
6622 "Ignoring zero length \\N{%s} in character class",
6627 SvREFCNT_dec(sv_name);
6629 SvREFCNT_dec(sv_str);
6630 return len ? NULL : (regnode *)&len;
6631 } else if(SvCUR(sv_str)) {
6637 char * parse_start = name-3; /* needed for the offsets */
6639 GET_RE_DEBUG_FLAGS_DECL; /* needed for the offsets */
6641 ret = reg_node(pRExC_state,
6642 (U8)(FOLD ? (LOC ? EXACTFL : EXACTF) : EXACT));
6645 if ( RExC_utf8 && !SvUTF8(sv_str) ) {
6646 sv_utf8_upgrade(sv_str);
6647 } else if ( !RExC_utf8 && SvUTF8(sv_str) ) {
6651 p = SvPV(sv_str, len);
6653 /* len is the length written, charlen is the size the char read */
6654 for ( len = 0; p < pend; p += charlen ) {
6656 UV uvc = utf8_to_uvchr((U8*)p, &charlen);
6658 STRLEN foldlen,numlen;
6659 U8 tmpbuf[UTF8_MAXBYTES_CASE+1], *foldbuf;
6660 uvc = toFOLD_uni(uvc, tmpbuf, &foldlen);
6661 /* Emit all the Unicode characters. */
6663 for (foldbuf = tmpbuf;
6667 uvc = utf8_to_uvchr(foldbuf, &numlen);
6669 const STRLEN unilen = reguni(pRExC_state, uvc, s);
6672 /* In EBCDIC the numlen
6673 * and unilen can differ. */
6675 if (numlen >= foldlen)
6679 break; /* "Can't happen." */
6682 const STRLEN unilen = reguni(pRExC_state, uvc, s);
6694 RExC_size += STR_SZ(len);
6697 RExC_emit += STR_SZ(len);
6699 Set_Node_Cur_Length(ret); /* MJD */
6701 nextchar(pRExC_state);
6703 ret = reg_node(pRExC_state,NOTHING);
6706 SvREFCNT_dec(sv_str);
6709 SvREFCNT_dec(sv_name);
6719 * It returns the code point in utf8 for the value in *encp.
6720 * value: a code value in the source encoding
6721 * encp: a pointer to an Encode object
6723 * If the result from Encode is not a single character,
6724 * it returns U+FFFD (Replacement character) and sets *encp to NULL.
6727 S_reg_recode(pTHX_ const char value, SV **encp)
6730 SV * const sv = sv_2mortal(newSVpvn(&value, numlen));
6731 const char * const s = *encp ? sv_recode_to_utf8(sv, *encp) : SvPVX(sv);
6732 const STRLEN newlen = SvCUR(sv);
6733 UV uv = UNICODE_REPLACEMENT;
6737 ? utf8n_to_uvchr((U8*)s, newlen, &numlen, UTF8_ALLOW_DEFAULT)
6740 if (!newlen || numlen != newlen) {
6741 uv = UNICODE_REPLACEMENT;
6749 - regatom - the lowest level
6751 Try to identify anything special at the start of the pattern. If there
6752 is, then handle it as required. This may involve generating a single regop,
6753 such as for an assertion; or it may involve recursing, such as to
6754 handle a () structure.
6756 If the string doesn't start with something special then we gobble up
6757 as much literal text as we can.
6759 Once we have been able to handle whatever type of thing started the
6760 sequence, we return.
6762 Note: we have to be careful with escapes, as they can be both literal
6763 and special, and in the case of \10 and friends can either, depending
6764 on context. Specifically there are two seperate switches for handling
6765 escape sequences, with the one for handling literal escapes requiring
6766 a dummy entry for all of the special escapes that are actually handled
6771 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
6774 register regnode *ret = NULL;
6776 char *parse_start = RExC_parse;
6777 GET_RE_DEBUG_FLAGS_DECL;
6778 DEBUG_PARSE("atom");
6779 *flagp = WORST; /* Tentatively. */
6783 switch ((U8)*RExC_parse) {
6785 RExC_seen_zerolen++;
6786 nextchar(pRExC_state);
6787 if (RExC_flags & RXf_PMf_MULTILINE)
6788 ret = reg_node(pRExC_state, MBOL);
6789 else if (RExC_flags & RXf_PMf_SINGLELINE)
6790 ret = reg_node(pRExC_state, SBOL);
6792 ret = reg_node(pRExC_state, BOL);
6793 Set_Node_Length(ret, 1); /* MJD */
6796 nextchar(pRExC_state);
6798 RExC_seen_zerolen++;
6799 if (RExC_flags & RXf_PMf_MULTILINE)
6800 ret = reg_node(pRExC_state, MEOL);
6801 else if (RExC_flags & RXf_PMf_SINGLELINE)
6802 ret = reg_node(pRExC_state, SEOL);
6804 ret = reg_node(pRExC_state, EOL);
6805 Set_Node_Length(ret, 1); /* MJD */
6808 nextchar(pRExC_state);
6809 if (RExC_flags & RXf_PMf_SINGLELINE)
6810 ret = reg_node(pRExC_state, SANY);
6812 ret = reg_node(pRExC_state, REG_ANY);
6813 *flagp |= HASWIDTH|SIMPLE;
6815 Set_Node_Length(ret, 1); /* MJD */
6819 char * const oregcomp_parse = ++RExC_parse;
6820 ret = regclass(pRExC_state,depth+1);
6821 if (*RExC_parse != ']') {
6822 RExC_parse = oregcomp_parse;
6823 vFAIL("Unmatched [");
6825 nextchar(pRExC_state);
6826 *flagp |= HASWIDTH|SIMPLE;
6827 Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
6831 nextchar(pRExC_state);
6832 ret = reg(pRExC_state, 1, &flags,depth+1);
6834 if (flags & TRYAGAIN) {
6835 if (RExC_parse == RExC_end) {
6836 /* Make parent create an empty node if needed. */
6844 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
6848 if (flags & TRYAGAIN) {
6852 vFAIL("Internal urp");
6853 /* Supposed to be caught earlier. */
6856 if (!regcurly(RExC_parse)) {
6865 vFAIL("Quantifier follows nothing");
6872 len=0; /* silence a spurious compiler warning */
6873 if ((cp = what_len_TRICKYFOLD_safe(RExC_parse,RExC_end,UTF,len))) {
6874 *flagp |= HASWIDTH; /* could be SIMPLE too, but needs a handler in regexec.regrepeat */
6875 RExC_parse+=len-1; /* we get one from nextchar() as well. :-( */
6876 ret = reganode(pRExC_state, FOLDCHAR, cp);
6877 Set_Node_Length(ret, 1); /* MJD */
6878 nextchar(pRExC_state); /* kill whitespace under /x */
6886 This switch handles escape sequences that resolve to some kind
6887 of special regop and not to literal text. Escape sequnces that
6888 resolve to literal text are handled below in the switch marked
6891 Every entry in this switch *must* have a corresponding entry
6892 in the literal escape switch. However, the opposite is not
6893 required, as the default for this switch is to jump to the
6894 literal text handling code.
6896 switch (*++RExC_parse) {
6897 /* Special Escapes */
6899 RExC_seen_zerolen++;
6900 ret = reg_node(pRExC_state, SBOL);
6902 goto finish_meta_pat;
6904 ret = reg_node(pRExC_state, GPOS);
6905 RExC_seen |= REG_SEEN_GPOS;
6907 goto finish_meta_pat;
6909 RExC_seen_zerolen++;
6910 ret = reg_node(pRExC_state, KEEPS);
6912 /* XXX:dmq : disabling in-place substitution seems to
6913 * be necessary here to avoid cases of memory corruption, as
6914 * with: C<$_="x" x 80; s/x\K/y/> -- rgs
6916 RExC_seen |= REG_SEEN_LOOKBEHIND;
6917 goto finish_meta_pat;
6919 ret = reg_node(pRExC_state, SEOL);
6921 RExC_seen_zerolen++; /* Do not optimize RE away */
6922 goto finish_meta_pat;
6924 ret = reg_node(pRExC_state, EOS);
6926 RExC_seen_zerolen++; /* Do not optimize RE away */
6927 goto finish_meta_pat;
6929 ret = reg_node(pRExC_state, CANY);
6930 RExC_seen |= REG_SEEN_CANY;
6931 *flagp |= HASWIDTH|SIMPLE;
6932 goto finish_meta_pat;
6934 ret = reg_node(pRExC_state, CLUMP);
6936 goto finish_meta_pat;
6938 ret = reg_node(pRExC_state, (U8)(LOC ? ALNUML : ALNUM));
6939 *flagp |= HASWIDTH|SIMPLE;
6940 goto finish_meta_pat;
6942 ret = reg_node(pRExC_state, (U8)(LOC ? NALNUML : NALNUM));
6943 *flagp |= HASWIDTH|SIMPLE;
6944 goto finish_meta_pat;
6946 RExC_seen_zerolen++;
6947 RExC_seen |= REG_SEEN_LOOKBEHIND;
6948 ret = reg_node(pRExC_state, (U8)(LOC ? BOUNDL : BOUND));
6950 goto finish_meta_pat;
6952 RExC_seen_zerolen++;
6953 RExC_seen |= REG_SEEN_LOOKBEHIND;
6954 ret = reg_node(pRExC_state, (U8)(LOC ? NBOUNDL : NBOUND));
6956 goto finish_meta_pat;
6958 ret = reg_node(pRExC_state, (U8)(LOC ? SPACEL : SPACE));
6959 *flagp |= HASWIDTH|SIMPLE;
6960 goto finish_meta_pat;
6962 ret = reg_node(pRExC_state, (U8)(LOC ? NSPACEL : NSPACE));
6963 *flagp |= HASWIDTH|SIMPLE;
6964 goto finish_meta_pat;
6966 ret = reg_node(pRExC_state, DIGIT);
6967 *flagp |= HASWIDTH|SIMPLE;
6968 goto finish_meta_pat;
6970 ret = reg_node(pRExC_state, NDIGIT);
6971 *flagp |= HASWIDTH|SIMPLE;
6972 goto finish_meta_pat;
6974 ret = reg_node(pRExC_state, LNBREAK);
6975 *flagp |= HASWIDTH|SIMPLE;
6976 goto finish_meta_pat;
6978 ret = reg_node(pRExC_state, HORIZWS);
6979 *flagp |= HASWIDTH|SIMPLE;
6980 goto finish_meta_pat;
6982 ret = reg_node(pRExC_state, NHORIZWS);
6983 *flagp |= HASWIDTH|SIMPLE;
6984 goto finish_meta_pat;
6986 ret = reg_node(pRExC_state, VERTWS);
6987 *flagp |= HASWIDTH|SIMPLE;
6988 goto finish_meta_pat;
6990 ret = reg_node(pRExC_state, NVERTWS);
6991 *flagp |= HASWIDTH|SIMPLE;
6993 nextchar(pRExC_state);
6994 Set_Node_Length(ret, 2); /* MJD */
6999 char* const oldregxend = RExC_end;
7001 char* parse_start = RExC_parse - 2;
7004 if (RExC_parse[1] == '{') {
7005 /* a lovely hack--pretend we saw [\pX] instead */
7006 RExC_end = strchr(RExC_parse, '}');
7008 const U8 c = (U8)*RExC_parse;
7010 RExC_end = oldregxend;
7011 vFAIL2("Missing right brace on \\%c{}", c);
7016 RExC_end = RExC_parse + 2;
7017 if (RExC_end > oldregxend)
7018 RExC_end = oldregxend;
7022 ret = regclass(pRExC_state,depth+1);
7024 RExC_end = oldregxend;
7027 Set_Node_Offset(ret, parse_start + 2);
7028 Set_Node_Cur_Length(ret);
7029 nextchar(pRExC_state);
7030 *flagp |= HASWIDTH|SIMPLE;
7034 /* Handle \N{NAME} here and not below because it can be
7035 multicharacter. join_exact() will join them up later on.
7036 Also this makes sure that things like /\N{BLAH}+/ and
7037 \N{BLAH} being multi char Just Happen. dmq*/
7039 ret= reg_namedseq(pRExC_state, NULL);
7041 case 'k': /* Handle \k<NAME> and \k'NAME' */
7044 char ch= RExC_parse[1];
7045 if (ch != '<' && ch != '\'' && ch != '{') {
7047 vFAIL2("Sequence %.2s... not terminated",parse_start);
7049 /* this pretty much dupes the code for (?P=...) in reg(), if
7050 you change this make sure you change that */
7051 char* name_start = (RExC_parse += 2);
7053 SV *sv_dat = reg_scan_name(pRExC_state,
7054 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
7055 ch= (ch == '<') ? '>' : (ch == '{') ? '}' : '\'';
7056 if (RExC_parse == name_start || *RExC_parse != ch)
7057 vFAIL2("Sequence %.3s... not terminated",parse_start);
7060 num = add_data( pRExC_state, 1, "S" );
7061 RExC_rxi->data->data[num]=(void*)sv_dat;
7062 SvREFCNT_inc_simple_void(sv_dat);
7066 ret = reganode(pRExC_state,
7067 (U8)(FOLD ? (LOC ? NREFFL : NREFF) : NREF),
7071 /* override incorrect value set in reganode MJD */
7072 Set_Node_Offset(ret, parse_start+1);
7073 Set_Node_Cur_Length(ret); /* MJD */
7074 nextchar(pRExC_state);
7080 case '1': case '2': case '3': case '4':
7081 case '5': case '6': case '7': case '8': case '9':
7084 bool isg = *RExC_parse == 'g';
7089 if (*RExC_parse == '{') {
7093 if (*RExC_parse == '-') {
7097 if (hasbrace && !isDIGIT(*RExC_parse)) {
7098 if (isrel) RExC_parse--;
7100 goto parse_named_seq;
7102 num = atoi(RExC_parse);
7103 if (isg && num == 0)
7104 vFAIL("Reference to invalid group 0");
7106 num = RExC_npar - num;
7108 vFAIL("Reference to nonexistent or unclosed group");
7110 if (!isg && num > 9 && num >= RExC_npar)
7113 char * const parse_start = RExC_parse - 1; /* MJD */
7114 while (isDIGIT(*RExC_parse))
7116 if (parse_start == RExC_parse - 1)
7117 vFAIL("Unterminated \\g... pattern");
7119 if (*RExC_parse != '}')
7120 vFAIL("Unterminated \\g{...} pattern");
7124 if (num > (I32)RExC_rx->nparens)
7125 vFAIL("Reference to nonexistent group");
7128 ret = reganode(pRExC_state,
7129 (U8)(FOLD ? (LOC ? REFFL : REFF) : REF),
7133 /* override incorrect value set in reganode MJD */
7134 Set_Node_Offset(ret, parse_start+1);
7135 Set_Node_Cur_Length(ret); /* MJD */
7137 nextchar(pRExC_state);
7142 if (RExC_parse >= RExC_end)
7143 FAIL("Trailing \\");
7146 /* Do not generate "unrecognized" warnings here, we fall
7147 back into the quick-grab loop below */
7154 if (RExC_flags & RXf_PMf_EXTENDED) {
7155 if ( reg_skipcomment( pRExC_state ) )
7162 register STRLEN len;
7167 U8 tmpbuf[UTF8_MAXBYTES_CASE+1], *foldbuf;
7169 parse_start = RExC_parse - 1;
7175 ret = reg_node(pRExC_state,
7176 (U8)(FOLD ? (LOC ? EXACTFL : EXACTF) : EXACT));
7178 for (len = 0, p = RExC_parse - 1;
7179 len < 127 && p < RExC_end;
7182 char * const oldp = p;
7184 if (RExC_flags & RXf_PMf_EXTENDED)
7185 p = regwhite( pRExC_state, p );
7190 if (LOC || !FOLD || !is_TRICKYFOLD_safe(p,RExC_end,UTF))
7191 goto normal_default;
7201 /* Literal Escapes Switch
7203 This switch is meant to handle escape sequences that
7204 resolve to a literal character.
7206 Every escape sequence that represents something
7207 else, like an assertion or a char class, is handled
7208 in the switch marked 'Special Escapes' above in this
7209 routine, but also has an entry here as anything that
7210 isn't explicitly mentioned here will be treated as
7211 an unescaped equivalent literal.
7215 /* These are all the special escapes. */
7216 case 'A': /* Start assertion */
7217 case 'b': case 'B': /* Word-boundary assertion*/
7218 case 'C': /* Single char !DANGEROUS! */
7219 case 'd': case 'D': /* digit class */
7220 case 'g': case 'G': /* generic-backref, pos assertion */
7221 case 'h': case 'H': /* HORIZWS */
7222 case 'k': case 'K': /* named backref, keep marker */
7223 case 'N': /* named char sequence */
7224 case 'p': case 'P': /* Unicode property */
7225 case 'R': /* LNBREAK */
7226 case 's': case 'S': /* space class */
7227 case 'v': case 'V': /* VERTWS */
7228 case 'w': case 'W': /* word class */
7229 case 'X': /* eXtended Unicode "combining character sequence" */
7230 case 'z': case 'Z': /* End of line/string assertion */
7234 /* Anything after here is an escape that resolves to a
7235 literal. (Except digits, which may or may not)
7254 ender = ASCII_TO_NATIVE('\033');
7258 ender = ASCII_TO_NATIVE('\007');
7263 char* const e = strchr(p, '}');
7267 vFAIL("Missing right brace on \\x{}");
7270 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
7271 | PERL_SCAN_DISALLOW_PREFIX;
7272 STRLEN numlen = e - p - 1;
7273 ender = grok_hex(p + 1, &numlen, &flags, NULL);
7280 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
7282 ender = grok_hex(p, &numlen, &flags, NULL);
7285 if (PL_encoding && ender < 0x100)
7286 goto recode_encoding;
7290 ender = UCHARAT(p++);
7291 ender = toCTRL(ender);
7293 case '0': case '1': case '2': case '3':case '4':
7294 case '5': case '6': case '7': case '8':case '9':
7296 (isDIGIT(p[1]) && atoi(p) >= RExC_npar) ) {
7299 ender = grok_oct(p, &numlen, &flags, NULL);
7306 if (PL_encoding && ender < 0x100)
7307 goto recode_encoding;
7311 SV* enc = PL_encoding;
7312 ender = reg_recode((const char)(U8)ender, &enc);
7313 if (!enc && SIZE_ONLY && ckWARN(WARN_REGEXP))
7314 vWARN(p, "Invalid escape in the specified encoding");
7320 FAIL("Trailing \\");
7323 if (!SIZE_ONLY&& isALPHA(*p) && ckWARN(WARN_REGEXP))
7324 vWARN2(p + 1, "Unrecognized escape \\%c passed through", UCHARAT(p));
7325 goto normal_default;
7330 if (UTF8_IS_START(*p) && UTF) {
7332 ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
7333 &numlen, UTF8_ALLOW_DEFAULT);
7340 if ( RExC_flags & RXf_PMf_EXTENDED)
7341 p = regwhite( pRExC_state, p );
7343 /* Prime the casefolded buffer. */
7344 ender = toFOLD_uni(ender, tmpbuf, &foldlen);
7346 if (p < RExC_end && ISMULT2(p)) { /* Back off on ?+*. */
7351 /* Emit all the Unicode characters. */
7353 for (foldbuf = tmpbuf;
7355 foldlen -= numlen) {
7356 ender = utf8_to_uvchr(foldbuf, &numlen);
7358 const STRLEN unilen = reguni(pRExC_state, ender, s);
7361 /* In EBCDIC the numlen
7362 * and unilen can differ. */
7364 if (numlen >= foldlen)
7368 break; /* "Can't happen." */
7372 const STRLEN unilen = reguni(pRExC_state, ender, s);
7381 REGC((char)ender, s++);
7387 /* Emit all the Unicode characters. */
7389 for (foldbuf = tmpbuf;
7391 foldlen -= numlen) {
7392 ender = utf8_to_uvchr(foldbuf, &numlen);
7394 const STRLEN unilen = reguni(pRExC_state, ender, s);
7397 /* In EBCDIC the numlen
7398 * and unilen can differ. */
7400 if (numlen >= foldlen)
7408 const STRLEN unilen = reguni(pRExC_state, ender, s);
7417 REGC((char)ender, s++);
7421 Set_Node_Cur_Length(ret); /* MJD */
7422 nextchar(pRExC_state);
7424 /* len is STRLEN which is unsigned, need to copy to signed */
7427 vFAIL("Internal disaster");
7431 if (len == 1 && UNI_IS_INVARIANT(ender))
7435 RExC_size += STR_SZ(len);
7438 RExC_emit += STR_SZ(len);
7448 S_regwhite( RExC_state_t *pRExC_state, char *p )
7450 const char *e = RExC_end;
7454 else if (*p == '#') {
7463 RExC_seen |= REG_SEEN_RUN_ON_COMMENT;
7471 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
7472 Character classes ([:foo:]) can also be negated ([:^foo:]).
7473 Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
7474 Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
7475 but trigger failures because they are currently unimplemented. */
7477 #define POSIXCC_DONE(c) ((c) == ':')
7478 #define POSIXCC_NOTYET(c) ((c) == '=' || (c) == '.')
7479 #define POSIXCC(c) (POSIXCC_DONE(c) || POSIXCC_NOTYET(c))
7482 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value)
7485 I32 namedclass = OOB_NAMEDCLASS;
7487 if (value == '[' && RExC_parse + 1 < RExC_end &&
7488 /* I smell either [: or [= or [. -- POSIX has been here, right? */
7489 POSIXCC(UCHARAT(RExC_parse))) {
7490 const char c = UCHARAT(RExC_parse);
7491 char* const s = RExC_parse++;
7493 while (RExC_parse < RExC_end && UCHARAT(RExC_parse) != c)
7495 if (RExC_parse == RExC_end)
7496 /* Grandfather lone [:, [=, [. */
7499 const char* const t = RExC_parse++; /* skip over the c */
7502 if (UCHARAT(RExC_parse) == ']') {
7503 const char *posixcc = s + 1;
7504 RExC_parse++; /* skip over the ending ] */
7507 const I32 complement = *posixcc == '^' ? *posixcc++ : 0;
7508 const I32 skip = t - posixcc;
7510 /* Initially switch on the length of the name. */
7513 if (memEQ(posixcc, "word", 4)) /* this is not POSIX, this is the Perl \w */
7514 namedclass = complement ? ANYOF_NALNUM : ANYOF_ALNUM;
7517 /* Names all of length 5. */
7518 /* alnum alpha ascii blank cntrl digit graph lower
7519 print punct space upper */
7520 /* Offset 4 gives the best switch position. */
7521 switch (posixcc[4]) {
7523 if (memEQ(posixcc, "alph", 4)) /* alpha */
7524 namedclass = complement ? ANYOF_NALPHA : ANYOF_ALPHA;
7527 if (memEQ(posixcc, "spac", 4)) /* space */
7528 namedclass = complement ? ANYOF_NPSXSPC : ANYOF_PSXSPC;
7531 if (memEQ(posixcc, "grap", 4)) /* graph */
7532 namedclass = complement ? ANYOF_NGRAPH : ANYOF_GRAPH;
7535 if (memEQ(posixcc, "asci", 4)) /* ascii */
7536 namedclass = complement ? ANYOF_NASCII : ANYOF_ASCII;
7539 if (memEQ(posixcc, "blan", 4)) /* blank */
7540 namedclass = complement ? ANYOF_NBLANK : ANYOF_BLANK;
7543 if (memEQ(posixcc, "cntr", 4)) /* cntrl */
7544 namedclass = complement ? ANYOF_NCNTRL : ANYOF_CNTRL;
7547 if (memEQ(posixcc, "alnu", 4)) /* alnum */
7548 namedclass = complement ? ANYOF_NALNUMC : ANYOF_ALNUMC;
7551 if (memEQ(posixcc, "lowe", 4)) /* lower */
7552 namedclass = complement ? ANYOF_NLOWER : ANYOF_LOWER;
7553 else if (memEQ(posixcc, "uppe", 4)) /* upper */
7554 namedclass = complement ? ANYOF_NUPPER : ANYOF_UPPER;
7557 if (memEQ(posixcc, "digi", 4)) /* digit */
7558 namedclass = complement ? ANYOF_NDIGIT : ANYOF_DIGIT;
7559 else if (memEQ(posixcc, "prin", 4)) /* print */
7560 namedclass = complement ? ANYOF_NPRINT : ANYOF_PRINT;
7561 else if (memEQ(posixcc, "punc", 4)) /* punct */
7562 namedclass = complement ? ANYOF_NPUNCT : ANYOF_PUNCT;
7567 if (memEQ(posixcc, "xdigit", 6))
7568 namedclass = complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT;
7572 if (namedclass == OOB_NAMEDCLASS)
7573 Simple_vFAIL3("POSIX class [:%.*s:] unknown",
7575 assert (posixcc[skip] == ':');
7576 assert (posixcc[skip+1] == ']');
7577 } else if (!SIZE_ONLY) {
7578 /* [[=foo=]] and [[.foo.]] are still future. */
7580 /* adjust RExC_parse so the warning shows after
7582 while (UCHARAT(RExC_parse) && UCHARAT(RExC_parse) != ']')
7584 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
7587 /* Maternal grandfather:
7588 * "[:" ending in ":" but not in ":]" */
7598 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
7601 if (POSIXCC(UCHARAT(RExC_parse))) {
7602 const char *s = RExC_parse;
7603 const char c = *s++;
7607 if (*s && c == *s && s[1] == ']') {
7608 if (ckWARN(WARN_REGEXP))
7610 "POSIX syntax [%c %c] belongs inside character classes",
7613 /* [[=foo=]] and [[.foo.]] are still future. */
7614 if (POSIXCC_NOTYET(c)) {
7615 /* adjust RExC_parse so the error shows after
7617 while (UCHARAT(RExC_parse) && UCHARAT(RExC_parse++) != ']')
7619 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
7626 #define _C_C_T_(NAME,TEST,WORD) \
7629 ANYOF_CLASS_SET(ret, ANYOF_##NAME); \
7631 for (value = 0; value < 256; value++) \
7633 ANYOF_BITMAP_SET(ret, value); \
7638 case ANYOF_N##NAME: \
7640 ANYOF_CLASS_SET(ret, ANYOF_N##NAME); \
7642 for (value = 0; value < 256; value++) \
7644 ANYOF_BITMAP_SET(ret, value); \
7650 #define _C_C_T_NOLOC_(NAME,TEST,WORD) \
7652 for (value = 0; value < 256; value++) \
7654 ANYOF_BITMAP_SET(ret, value); \
7658 case ANYOF_N##NAME: \
7659 for (value = 0; value < 256; value++) \
7661 ANYOF_BITMAP_SET(ret, value); \
7667 parse a class specification and produce either an ANYOF node that
7668 matches the pattern or if the pattern matches a single char only and
7669 that char is < 256 and we are case insensitive then we produce an
7674 S_regclass(pTHX_ RExC_state_t *pRExC_state, U32 depth)
7677 register UV nextvalue;
7678 register IV prevvalue = OOB_UNICODE;
7679 register IV range = 0;
7680 UV value = 0; /* XXX:dmq: needs to be referenceable (unfortunately) */
7681 register regnode *ret;
7684 char *rangebegin = NULL;
7685 bool need_class = 0;
7688 bool optimize_invert = TRUE;
7689 AV* unicode_alternate = NULL;
7691 UV literal_endpoint = 0;
7693 UV stored = 0; /* number of chars stored in the class */
7695 regnode * const orig_emit = RExC_emit; /* Save the original RExC_emit in
7696 case we need to change the emitted regop to an EXACT. */
7697 const char * orig_parse = RExC_parse;
7698 GET_RE_DEBUG_FLAGS_DECL;
7700 PERL_UNUSED_ARG(depth);
7703 DEBUG_PARSE("clas");
7705 /* Assume we are going to generate an ANYOF node. */
7706 ret = reganode(pRExC_state, ANYOF, 0);
7709 ANYOF_FLAGS(ret) = 0;
7711 if (UCHARAT(RExC_parse) == '^') { /* Complement of range. */
7715 ANYOF_FLAGS(ret) |= ANYOF_INVERT;
7719 RExC_size += ANYOF_SKIP;
7720 listsv = &PL_sv_undef; /* For code scanners: listsv always non-NULL. */
7723 RExC_emit += ANYOF_SKIP;
7725 ANYOF_FLAGS(ret) |= ANYOF_FOLD;
7727 ANYOF_FLAGS(ret) |= ANYOF_LOCALE;
7728 ANYOF_BITMAP_ZERO(ret);
7729 listsv = newSVpvs("# comment\n");
7732 nextvalue = RExC_parse < RExC_end ? UCHARAT(RExC_parse) : 0;
7734 if (!SIZE_ONLY && POSIXCC(nextvalue))
7735 checkposixcc(pRExC_state);
7737 /* allow 1st char to be ] (allowing it to be - is dealt with later) */
7738 if (UCHARAT(RExC_parse) == ']')
7742 while (RExC_parse < RExC_end && UCHARAT(RExC_parse) != ']') {
7746 namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
7749 rangebegin = RExC_parse;
7751 value = utf8n_to_uvchr((U8*)RExC_parse,
7752 RExC_end - RExC_parse,
7753 &numlen, UTF8_ALLOW_DEFAULT);
7754 RExC_parse += numlen;
7757 value = UCHARAT(RExC_parse++);
7759 nextvalue = RExC_parse < RExC_end ? UCHARAT(RExC_parse) : 0;
7760 if (value == '[' && POSIXCC(nextvalue))
7761 namedclass = regpposixcc(pRExC_state, value);
7762 else if (value == '\\') {
7764 value = utf8n_to_uvchr((U8*)RExC_parse,
7765 RExC_end - RExC_parse,
7766 &numlen, UTF8_ALLOW_DEFAULT);
7767 RExC_parse += numlen;
7770 value = UCHARAT(RExC_parse++);
7771 /* Some compilers cannot handle switching on 64-bit integer
7772 * values, therefore value cannot be an UV. Yes, this will
7773 * be a problem later if we want switch on Unicode.
7774 * A similar issue a little bit later when switching on
7775 * namedclass. --jhi */
7776 switch ((I32)value) {
7777 case 'w': namedclass = ANYOF_ALNUM; break;
7778 case 'W': namedclass = ANYOF_NALNUM; break;
7779 case 's': namedclass = ANYOF_SPACE; break;
7780 case 'S': namedclass = ANYOF_NSPACE; break;
7781 case 'd': namedclass = ANYOF_DIGIT; break;
7782 case 'D': namedclass = ANYOF_NDIGIT; break;
7783 case 'v': namedclass = ANYOF_VERTWS; break;
7784 case 'V': namedclass = ANYOF_NVERTWS; break;
7785 case 'h': namedclass = ANYOF_HORIZWS; break;
7786 case 'H': namedclass = ANYOF_NHORIZWS; break;
7787 case 'N': /* Handle \N{NAME} in class */
7789 /* We only pay attention to the first char of
7790 multichar strings being returned. I kinda wonder
7791 if this makes sense as it does change the behaviour
7792 from earlier versions, OTOH that behaviour was broken
7794 UV v; /* value is register so we cant & it /grrr */
7795 if (reg_namedseq(pRExC_state, &v)) {
7805 if (RExC_parse >= RExC_end)
7806 vFAIL2("Empty \\%c{}", (U8)value);
7807 if (*RExC_parse == '{') {
7808 const U8 c = (U8)value;
7809 e = strchr(RExC_parse++, '}');
7811 vFAIL2("Missing right brace on \\%c{}", c);
7812 while (isSPACE(UCHARAT(RExC_parse)))
7814 if (e == RExC_parse)
7815 vFAIL2("Empty \\%c{}", c);
7817 while (isSPACE(UCHARAT(RExC_parse + n - 1)))
7825 if (UCHARAT(RExC_parse) == '^') {
7828 value = value == 'p' ? 'P' : 'p'; /* toggle */
7829 while (isSPACE(UCHARAT(RExC_parse))) {
7834 Perl_sv_catpvf(aTHX_ listsv, "%cutf8::%.*s\n",
7835 (value=='p' ? '+' : '!'), (int)n, RExC_parse);
7838 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
7839 namedclass = ANYOF_MAX; /* no official name, but it's named */
7842 case 'n': value = '\n'; break;
7843 case 'r': value = '\r'; break;
7844 case 't': value = '\t'; break;
7845 case 'f': value = '\f'; break;
7846 case 'b': value = '\b'; break;
7847 case 'e': value = ASCII_TO_NATIVE('\033');break;
7848 case 'a': value = ASCII_TO_NATIVE('\007');break;
7850 if (*RExC_parse == '{') {
7851 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
7852 | PERL_SCAN_DISALLOW_PREFIX;
7853 char * const e = strchr(RExC_parse++, '}');
7855 vFAIL("Missing right brace on \\x{}");
7857 numlen = e - RExC_parse;
7858 value = grok_hex(RExC_parse, &numlen, &flags, NULL);
7862 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
7864 value = grok_hex(RExC_parse, &numlen, &flags, NULL);
7865 RExC_parse += numlen;
7867 if (PL_encoding && value < 0x100)
7868 goto recode_encoding;
7871 value = UCHARAT(RExC_parse++);
7872 value = toCTRL(value);
7874 case '0': case '1': case '2': case '3': case '4':
7875 case '5': case '6': case '7': case '8': case '9':
7879 value = grok_oct(--RExC_parse, &numlen, &flags, NULL);
7880 RExC_parse += numlen;
7881 if (PL_encoding && value < 0x100)
7882 goto recode_encoding;
7887 SV* enc = PL_encoding;
7888 value = reg_recode((const char)(U8)value, &enc);
7889 if (!enc && SIZE_ONLY && ckWARN(WARN_REGEXP))
7891 "Invalid escape in the specified encoding");
7895 if (!SIZE_ONLY && isALPHA(value) && ckWARN(WARN_REGEXP))
7897 "Unrecognized escape \\%c in character class passed through",
7901 } /* end of \blah */
7907 if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
7909 if (!SIZE_ONLY && !need_class)
7910 ANYOF_CLASS_ZERO(ret);
7914 /* a bad range like a-\d, a-[:digit:] ? */
7917 if (ckWARN(WARN_REGEXP)) {
7919 RExC_parse >= rangebegin ?
7920 RExC_parse - rangebegin : 0;
7922 "False [] range \"%*.*s\"",
7925 if (prevvalue < 256) {
7926 ANYOF_BITMAP_SET(ret, prevvalue);
7927 ANYOF_BITMAP_SET(ret, '-');
7930 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
7931 Perl_sv_catpvf(aTHX_ listsv,
7932 "%04"UVxf"\n%04"UVxf"\n", (UV)prevvalue, (UV) '-');
7936 range = 0; /* this was not a true range */
7942 const char *what = NULL;
7945 if (namedclass > OOB_NAMEDCLASS)
7946 optimize_invert = FALSE;
7947 /* Possible truncation here but in some 64-bit environments
7948 * the compiler gets heartburn about switch on 64-bit values.
7949 * A similar issue a little earlier when switching on value.
7951 switch ((I32)namedclass) {
7952 case _C_C_T_(ALNUM, isALNUM(value), "Word");
7953 case _C_C_T_(ALNUMC, isALNUMC(value), "Alnum");
7954 case _C_C_T_(ALPHA, isALPHA(value), "Alpha");
7955 case _C_C_T_(BLANK, isBLANK(value), "Blank");
7956 case _C_C_T_(CNTRL, isCNTRL(value), "Cntrl");
7957 case _C_C_T_(GRAPH, isGRAPH(value), "Graph");
7958 case _C_C_T_(LOWER, isLOWER(value), "Lower");
7959 case _C_C_T_(PRINT, isPRINT(value), "Print");
7960 case _C_C_T_(PSXSPC, isPSXSPC(value), "Space");
7961 case _C_C_T_(PUNCT, isPUNCT(value), "Punct");
7962 case _C_C_T_(SPACE, isSPACE(value), "SpacePerl");
7963 case _C_C_T_(UPPER, isUPPER(value), "Upper");
7964 case _C_C_T_(XDIGIT, isXDIGIT(value), "XDigit");
7965 case _C_C_T_NOLOC_(VERTWS, is_VERTWS_latin1(&value), "VertSpace");
7966 case _C_C_T_NOLOC_(HORIZWS, is_HORIZWS_latin1(&value), "HorizSpace");
7969 ANYOF_CLASS_SET(ret, ANYOF_ASCII);
7972 for (value = 0; value < 128; value++)
7973 ANYOF_BITMAP_SET(ret, value);
7975 for (value = 0; value < 256; value++) {
7977 ANYOF_BITMAP_SET(ret, value);
7986 ANYOF_CLASS_SET(ret, ANYOF_NASCII);
7989 for (value = 128; value < 256; value++)
7990 ANYOF_BITMAP_SET(ret, value);
7992 for (value = 0; value < 256; value++) {
7993 if (!isASCII(value))
7994 ANYOF_BITMAP_SET(ret, value);
8003 ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
8005 /* consecutive digits assumed */
8006 for (value = '0'; value <= '9'; value++)
8007 ANYOF_BITMAP_SET(ret, value);
8014 ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
8016 /* consecutive digits assumed */
8017 for (value = 0; value < '0'; value++)
8018 ANYOF_BITMAP_SET(ret, value);
8019 for (value = '9' + 1; value < 256; value++)
8020 ANYOF_BITMAP_SET(ret, value);
8026 /* this is to handle \p and \P */
8029 vFAIL("Invalid [::] class");
8033 /* Strings such as "+utf8::isWord\n" */
8034 Perl_sv_catpvf(aTHX_ listsv, "%cutf8::Is%s\n", yesno, what);
8037 ANYOF_FLAGS(ret) |= ANYOF_CLASS;
8040 } /* end of namedclass \blah */
8043 if (prevvalue > (IV)value) /* b-a */ {
8044 const int w = RExC_parse - rangebegin;
8045 Simple_vFAIL4("Invalid [] range \"%*.*s\"", w, w, rangebegin);
8046 range = 0; /* not a valid range */
8050 prevvalue = value; /* save the beginning of the range */
8051 if (*RExC_parse == '-' && RExC_parse+1 < RExC_end &&
8052 RExC_parse[1] != ']') {
8055 /* a bad range like \w-, [:word:]- ? */
8056 if (namedclass > OOB_NAMEDCLASS) {
8057 if (ckWARN(WARN_REGEXP)) {
8059 RExC_parse >= rangebegin ?
8060 RExC_parse - rangebegin : 0;
8062 "False [] range \"%*.*s\"",
8066 ANYOF_BITMAP_SET(ret, '-');
8068 range = 1; /* yeah, it's a range! */
8069 continue; /* but do it the next time */
8073 /* now is the next time */
8074 /*stored += (value - prevvalue + 1);*/
8076 if (prevvalue < 256) {
8077 const IV ceilvalue = value < 256 ? value : 255;
8080 /* In EBCDIC [\x89-\x91] should include
8081 * the \x8e but [i-j] should not. */
8082 if (literal_endpoint == 2 &&
8083 ((isLOWER(prevvalue) && isLOWER(ceilvalue)) ||
8084 (isUPPER(prevvalue) && isUPPER(ceilvalue))))
8086 if (isLOWER(prevvalue)) {
8087 for (i = prevvalue; i <= ceilvalue; i++)
8088 if (isLOWER(i) && !ANYOF_BITMAP_TEST(ret,i)) {
8090 ANYOF_BITMAP_SET(ret, i);
8093 for (i = prevvalue; i <= ceilvalue; i++)
8094 if (isUPPER(i) && !ANYOF_BITMAP_TEST(ret,i)) {
8096 ANYOF_BITMAP_SET(ret, i);
8102 for (i = prevvalue; i <= ceilvalue; i++) {
8103 if (!ANYOF_BITMAP_TEST(ret,i)) {
8105 ANYOF_BITMAP_SET(ret, i);
8109 if (value > 255 || UTF) {
8110 const UV prevnatvalue = NATIVE_TO_UNI(prevvalue);
8111 const UV natvalue = NATIVE_TO_UNI(value);
8112 stored+=2; /* can't optimize this class */
8113 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
8114 if (prevnatvalue < natvalue) { /* what about > ? */
8115 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n",
8116 prevnatvalue, natvalue);
8118 else if (prevnatvalue == natvalue) {
8119 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n", natvalue);
8121 U8 foldbuf[UTF8_MAXBYTES_CASE+1];
8123 const UV f = to_uni_fold(natvalue, foldbuf, &foldlen);
8125 #ifdef EBCDIC /* RD t/uni/fold ff and 6b */
8126 if (RExC_precomp[0] == ':' &&
8127 RExC_precomp[1] == '[' &&
8128 (f == 0xDF || f == 0x92)) {
8129 f = NATIVE_TO_UNI(f);
8132 /* If folding and foldable and a single
8133 * character, insert also the folded version
8134 * to the charclass. */
8136 #ifdef EBCDIC /* RD tunifold ligatures s,t fb05, fb06 */
8137 if ((RExC_precomp[0] == ':' &&
8138 RExC_precomp[1] == '[' &&
8140 (value == 0xFB05 || value == 0xFB06))) ?
8141 foldlen == ((STRLEN)UNISKIP(f) - 1) :
8142 foldlen == (STRLEN)UNISKIP(f) )
8144 if (foldlen == (STRLEN)UNISKIP(f))
8146 Perl_sv_catpvf(aTHX_ listsv,
8149 /* Any multicharacter foldings
8150 * require the following transform:
8151 * [ABCDEF] -> (?:[ABCabcDEFd]|pq|rst)
8152 * where E folds into "pq" and F folds
8153 * into "rst", all other characters
8154 * fold to single characters. We save
8155 * away these multicharacter foldings,
8156 * to be later saved as part of the
8157 * additional "s" data. */
8160 if (!unicode_alternate)
8161 unicode_alternate = newAV();
8162 sv = newSVpvn((char*)foldbuf, foldlen);
8164 av_push(unicode_alternate, sv);
8168 /* If folding and the value is one of the Greek
8169 * sigmas insert a few more sigmas to make the
8170 * folding rules of the sigmas to work right.
8171 * Note that not all the possible combinations
8172 * are handled here: some of them are handled
8173 * by the standard folding rules, and some of
8174 * them (literal or EXACTF cases) are handled
8175 * during runtime in regexec.c:S_find_byclass(). */
8176 if (value == UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) {
8177 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
8178 (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA);
8179 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
8180 (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA);
8182 else if (value == UNICODE_GREEK_CAPITAL_LETTER_SIGMA)
8183 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
8184 (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA);
8189 literal_endpoint = 0;
8193 range = 0; /* this range (if it was one) is done now */
8197 ANYOF_FLAGS(ret) |= ANYOF_LARGE;
8199 RExC_size += ANYOF_CLASS_ADD_SKIP;
8201 RExC_emit += ANYOF_CLASS_ADD_SKIP;
8207 /****** !SIZE_ONLY AFTER HERE *********/
8209 if( stored == 1 && (value < 128 || (value < 256 && !UTF))
8210 && !( ANYOF_FLAGS(ret) & ( ANYOF_FLAGS_ALL ^ ANYOF_FOLD ) )
8212 /* optimize single char class to an EXACT node
8213 but *only* when its not a UTF/high char */
8214 const char * cur_parse= RExC_parse;
8215 RExC_emit = (regnode *)orig_emit;
8216 RExC_parse = (char *)orig_parse;
8217 ret = reg_node(pRExC_state,
8218 (U8)((ANYOF_FLAGS(ret) & ANYOF_FOLD) ? EXACTF : EXACT));
8219 RExC_parse = (char *)cur_parse;
8220 *STRING(ret)= (char)value;
8222 RExC_emit += STR_SZ(1);
8225 /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
8226 if ( /* If the only flag is folding (plus possibly inversion). */
8227 ((ANYOF_FLAGS(ret) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD)
8229 for (value = 0; value < 256; ++value) {
8230 if (ANYOF_BITMAP_TEST(ret, value)) {
8231 UV fold = PL_fold[value];
8234 ANYOF_BITMAP_SET(ret, fold);
8237 ANYOF_FLAGS(ret) &= ~ANYOF_FOLD;
8240 /* optimize inverted simple patterns (e.g. [^a-z]) */
8241 if (optimize_invert &&
8242 /* If the only flag is inversion. */
8243 (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) {
8244 for (value = 0; value < ANYOF_BITMAP_SIZE; ++value)
8245 ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL;
8246 ANYOF_FLAGS(ret) = ANYOF_UNICODE_ALL;
8249 AV * const av = newAV();
8251 /* The 0th element stores the character class description
8252 * in its textual form: used later (regexec.c:Perl_regclass_swash())
8253 * to initialize the appropriate swash (which gets stored in
8254 * the 1st element), and also useful for dumping the regnode.
8255 * The 2nd element stores the multicharacter foldings,
8256 * used later (regexec.c:S_reginclass()). */
8257 av_store(av, 0, listsv);
8258 av_store(av, 1, NULL);
8259 av_store(av, 2, (SV*)unicode_alternate);
8260 rv = newRV_noinc((SV*)av);
8261 n = add_data(pRExC_state, 1, "s");
8262 RExC_rxi->data->data[n] = (void*)rv;
8270 /* reg_skipcomment()
8272 Absorbs an /x style # comments from the input stream.
8273 Returns true if there is more text remaining in the stream.
8274 Will set the REG_SEEN_RUN_ON_COMMENT flag if the comment
8275 terminates the pattern without including a newline.
8277 Note its the callers responsibility to ensure that we are
8283 S_reg_skipcomment(pTHX_ RExC_state_t *pRExC_state)
8286 while (RExC_parse < RExC_end)
8287 if (*RExC_parse++ == '\n') {
8292 /* we ran off the end of the pattern without ending
8293 the comment, so we have to add an \n when wrapping */
8294 RExC_seen |= REG_SEEN_RUN_ON_COMMENT;
8302 Advance that parse position, and optionall absorbs
8303 "whitespace" from the inputstream.
8305 Without /x "whitespace" means (?#...) style comments only,
8306 with /x this means (?#...) and # comments and whitespace proper.
8308 Returns the RExC_parse point from BEFORE the scan occurs.
8310 This is the /x friendly way of saying RExC_parse++.
8314 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
8316 char* const retval = RExC_parse++;
8319 if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
8320 RExC_parse[2] == '#') {
8321 while (*RExC_parse != ')') {
8322 if (RExC_parse == RExC_end)
8323 FAIL("Sequence (?#... not terminated");
8329 if (RExC_flags & RXf_PMf_EXTENDED) {
8330 if (isSPACE(*RExC_parse)) {
8334 else if (*RExC_parse == '#') {
8335 if ( reg_skipcomment( pRExC_state ) )
8344 - reg_node - emit a node
8346 STATIC regnode * /* Location. */
8347 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
8350 register regnode *ptr;
8351 regnode * const ret = RExC_emit;
8352 GET_RE_DEBUG_FLAGS_DECL;
8355 SIZE_ALIGN(RExC_size);
8359 if (RExC_emit >= RExC_emit_bound)
8360 Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d", op);
8362 NODE_ALIGN_FILL(ret);
8364 FILL_ADVANCE_NODE(ptr, op);
8365 #ifdef RE_TRACK_PATTERN_OFFSETS
8366 if (RExC_offsets) { /* MJD */
8367 MJD_OFFSET_DEBUG(("%s:%d: (op %s) %s %"UVuf" (len %"UVuf") (max %"UVuf").\n",
8368 "reg_node", __LINE__,
8370 (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0]
8371 ? "Overwriting end of array!\n" : "OK",
8372 (UV)(RExC_emit - RExC_emit_start),
8373 (UV)(RExC_parse - RExC_start),
8374 (UV)RExC_offsets[0]));
8375 Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
8383 - reganode - emit a node with an argument
8385 STATIC regnode * /* Location. */
8386 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
8389 register regnode *ptr;
8390 regnode * const ret = RExC_emit;
8391 GET_RE_DEBUG_FLAGS_DECL;
8394 SIZE_ALIGN(RExC_size);
8399 assert(2==regarglen[op]+1);
8401 Anything larger than this has to allocate the extra amount.
8402 If we changed this to be:
8404 RExC_size += (1 + regarglen[op]);
8406 then it wouldn't matter. Its not clear what side effect
8407 might come from that so its not done so far.
8412 if (RExC_emit >= RExC_emit_bound)
8413 Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d", op);
8415 NODE_ALIGN_FILL(ret);
8417 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
8418 #ifdef RE_TRACK_PATTERN_OFFSETS
8419 if (RExC_offsets) { /* MJD */
8420 MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n",
8424 (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0] ?
8425 "Overwriting end of array!\n" : "OK",
8426 (UV)(RExC_emit - RExC_emit_start),
8427 (UV)(RExC_parse - RExC_start),
8428 (UV)RExC_offsets[0]));
8429 Set_Cur_Node_Offset;
8437 - reguni - emit (if appropriate) a Unicode character
8440 S_reguni(pTHX_ const RExC_state_t *pRExC_state, UV uv, char* s)
8443 return SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
8447 - reginsert - insert an operator in front of already-emitted operand
8449 * Means relocating the operand.
8452 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd, U32 depth)
8455 register regnode *src;
8456 register regnode *dst;
8457 register regnode *place;
8458 const int offset = regarglen[(U8)op];
8459 const int size = NODE_STEP_REGNODE + offset;
8460 GET_RE_DEBUG_FLAGS_DECL;
8461 PERL_UNUSED_ARG(depth);
8462 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
8463 DEBUG_PARSE_FMT("inst"," - %s",PL_reg_name[op]);
8472 if (RExC_open_parens) {
8474 /*DEBUG_PARSE_FMT("inst"," - %"IVdf, (IV)RExC_npar);*/
8475 for ( paren=0 ; paren < RExC_npar ; paren++ ) {
8476 if ( RExC_open_parens[paren] >= opnd ) {
8477 /*DEBUG_PARSE_FMT("open"," - %d",size);*/
8478 RExC_open_parens[paren] += size;
8480 /*DEBUG_PARSE_FMT("open"," - %s","ok");*/
8482 if ( RExC_close_parens[paren] >= opnd ) {
8483 /*DEBUG_PARSE_FMT("close"," - %d",size);*/
8484 RExC_close_parens[paren] += size;
8486 /*DEBUG_PARSE_FMT("close"," - %s","ok");*/
8491 while (src > opnd) {
8492 StructCopy(--src, --dst, regnode);
8493 #ifdef RE_TRACK_PATTERN_OFFSETS
8494 if (RExC_offsets) { /* MJD 20010112 */
8495 MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s copy %"UVuf" -> %"UVuf" (max %"UVuf").\n",
8499 (UV)(dst - RExC_emit_start) > RExC_offsets[0]
8500 ? "Overwriting end of array!\n" : "OK",
8501 (UV)(src - RExC_emit_start),
8502 (UV)(dst - RExC_emit_start),
8503 (UV)RExC_offsets[0]));
8504 Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
8505 Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
8511 place = opnd; /* Op node, where operand used to be. */
8512 #ifdef RE_TRACK_PATTERN_OFFSETS
8513 if (RExC_offsets) { /* MJD */
8514 MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n",
8518 (UV)(place - RExC_emit_start) > RExC_offsets[0]
8519 ? "Overwriting end of array!\n" : "OK",
8520 (UV)(place - RExC_emit_start),
8521 (UV)(RExC_parse - RExC_start),
8522 (UV)RExC_offsets[0]));
8523 Set_Node_Offset(place, RExC_parse);
8524 Set_Node_Length(place, 1);
8527 src = NEXTOPER(place);
8528 FILL_ADVANCE_NODE(place, op);
8529 Zero(src, offset, regnode);
8533 - regtail - set the next-pointer at the end of a node chain of p to val.
8534 - SEE ALSO: regtail_study
8536 /* TODO: All three parms should be const */
8538 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,U32 depth)
8541 register regnode *scan;
8542 GET_RE_DEBUG_FLAGS_DECL;
8544 PERL_UNUSED_ARG(depth);
8550 /* Find last node. */
8553 regnode * const temp = regnext(scan);
8555 SV * const mysv=sv_newmortal();
8556 DEBUG_PARSE_MSG((scan==p ? "tail" : ""));
8557 regprop(RExC_rx, mysv, scan);
8558 PerlIO_printf(Perl_debug_log, "~ %s (%d) %s %s\n",
8559 SvPV_nolen_const(mysv), REG_NODE_NUM(scan),
8560 (temp == NULL ? "->" : ""),
8561 (temp == NULL ? PL_reg_name[OP(val)] : "")
8569 if (reg_off_by_arg[OP(scan)]) {
8570 ARG_SET(scan, val - scan);
8573 NEXT_OFF(scan) = val - scan;
8579 - regtail_study - set the next-pointer at the end of a node chain of p to val.
8580 - Look for optimizable sequences at the same time.
8581 - currently only looks for EXACT chains.
8583 This is expermental code. The idea is to use this routine to perform
8584 in place optimizations on branches and groups as they are constructed,
8585 with the long term intention of removing optimization from study_chunk so
8586 that it is purely analytical.
8588 Currently only used when in DEBUG mode. The macro REGTAIL_STUDY() is used
8589 to control which is which.
8592 /* TODO: All four parms should be const */
8595 S_regtail_study(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,U32 depth)
8598 register regnode *scan;
8600 #ifdef EXPERIMENTAL_INPLACESCAN
8604 GET_RE_DEBUG_FLAGS_DECL;
8610 /* Find last node. */
8614 regnode * const temp = regnext(scan);
8615 #ifdef EXPERIMENTAL_INPLACESCAN
8616 if (PL_regkind[OP(scan)] == EXACT)
8617 if (join_exact(pRExC_state,scan,&min,1,val,depth+1))
8625 if( exact == PSEUDO )
8627 else if ( exact != OP(scan) )
8636 SV * const mysv=sv_newmortal();
8637 DEBUG_PARSE_MSG((scan==p ? "tsdy" : ""));
8638 regprop(RExC_rx, mysv, scan);
8639 PerlIO_printf(Perl_debug_log, "~ %s (%d) -> %s\n",
8640 SvPV_nolen_const(mysv),
8642 PL_reg_name[exact]);
8649 SV * const mysv_val=sv_newmortal();
8650 DEBUG_PARSE_MSG("");
8651 regprop(RExC_rx, mysv_val, val);
8652 PerlIO_printf(Perl_debug_log, "~ attach to %s (%"IVdf") offset to %"IVdf"\n",
8653 SvPV_nolen_const(mysv_val),
8654 (IV)REG_NODE_NUM(val),
8658 if (reg_off_by_arg[OP(scan)]) {
8659 ARG_SET(scan, val - scan);
8662 NEXT_OFF(scan) = val - scan;
8670 - regcurly - a little FSA that accepts {\d+,?\d*}
8673 S_regcurly(register const char *s)
8692 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
8696 S_regdump_extflags(pTHX_ const char *lead, const U32 flags) {
8699 for (bit=0; bit<32; bit++) {
8700 if (flags & (1<<bit)) {
8702 PerlIO_printf(Perl_debug_log, "%s",lead);
8703 PerlIO_printf(Perl_debug_log, "%s ",PL_reg_extflags_name[bit]);
8708 PerlIO_printf(Perl_debug_log, "\n");
8710 PerlIO_printf(Perl_debug_log, "%s[none-set]\n",lead);
8716 Perl_regdump(pTHX_ const regexp *r)
8720 SV * const sv = sv_newmortal();
8721 SV *dsv= sv_newmortal();
8723 GET_RE_DEBUG_FLAGS_DECL;
8725 (void)dumpuntil(r, ri->program, ri->program + 1, NULL, NULL, sv, 0, 0);
8727 /* Header fields of interest. */
8728 if (r->anchored_substr) {
8729 RE_PV_QUOTED_DECL(s, 0, dsv, SvPVX_const(r->anchored_substr),
8730 RE_SV_DUMPLEN(r->anchored_substr), 30);
8731 PerlIO_printf(Perl_debug_log,
8732 "anchored %s%s at %"IVdf" ",
8733 s, RE_SV_TAIL(r->anchored_substr),
8734 (IV)r->anchored_offset);
8735 } else if (r->anchored_utf8) {
8736 RE_PV_QUOTED_DECL(s, 1, dsv, SvPVX_const(r->anchored_utf8),
8737 RE_SV_DUMPLEN(r->anchored_utf8), 30);
8738 PerlIO_printf(Perl_debug_log,
8739 "anchored utf8 %s%s at %"IVdf" ",
8740 s, RE_SV_TAIL(r->anchored_utf8),
8741 (IV)r->anchored_offset);
8743 if (r->float_substr) {
8744 RE_PV_QUOTED_DECL(s, 0, dsv, SvPVX_const(r->float_substr),
8745 RE_SV_DUMPLEN(r->float_substr), 30);
8746 PerlIO_printf(Perl_debug_log,
8747 "floating %s%s at %"IVdf"..%"UVuf" ",
8748 s, RE_SV_TAIL(r->float_substr),
8749 (IV)r->float_min_offset, (UV)r->float_max_offset);
8750 } else if (r->float_utf8) {
8751 RE_PV_QUOTED_DECL(s, 1, dsv, SvPVX_const(r->float_utf8),
8752 RE_SV_DUMPLEN(r->float_utf8), 30);
8753 PerlIO_printf(Perl_debug_log,
8754 "floating utf8 %s%s at %"IVdf"..%"UVuf" ",
8755 s, RE_SV_TAIL(r->float_utf8),
8756 (IV)r->float_min_offset, (UV)r->float_max_offset);
8758 if (r->check_substr || r->check_utf8)
8759 PerlIO_printf(Perl_debug_log,
8761 (r->check_substr == r->float_substr
8762 && r->check_utf8 == r->float_utf8
8763 ? "(checking floating" : "(checking anchored"));
8764 if (r->extflags & RXf_NOSCAN)
8765 PerlIO_printf(Perl_debug_log, " noscan");
8766 if (r->extflags & RXf_CHECK_ALL)
8767 PerlIO_printf(Perl_debug_log, " isall");
8768 if (r->check_substr || r->check_utf8)
8769 PerlIO_printf(Perl_debug_log, ") ");
8771 if (ri->regstclass) {
8772 regprop(r, sv, ri->regstclass);
8773 PerlIO_printf(Perl_debug_log, "stclass %s ", SvPVX_const(sv));
8775 if (r->extflags & RXf_ANCH) {
8776 PerlIO_printf(Perl_debug_log, "anchored");
8777 if (r->extflags & RXf_ANCH_BOL)
8778 PerlIO_printf(Perl_debug_log, "(BOL)");
8779 if (r->extflags & RXf_ANCH_MBOL)
8780 PerlIO_printf(Perl_debug_log, "(MBOL)");
8781 if (r->extflags & RXf_ANCH_SBOL)
8782 PerlIO_printf(Perl_debug_log, "(SBOL)");
8783 if (r->extflags & RXf_ANCH_GPOS)
8784 PerlIO_printf(Perl_debug_log, "(GPOS)");
8785 PerlIO_putc(Perl_debug_log, ' ');
8787 if (r->extflags & RXf_GPOS_SEEN)
8788 PerlIO_printf(Perl_debug_log, "GPOS:%"UVuf" ", (UV)r->gofs);
8789 if (r->intflags & PREGf_SKIP)
8790 PerlIO_printf(Perl_debug_log, "plus ");
8791 if (r->intflags & PREGf_IMPLICIT)
8792 PerlIO_printf(Perl_debug_log, "implicit ");
8793 PerlIO_printf(Perl_debug_log, "minlen %"IVdf" ", (IV)r->minlen);
8794 if (r->extflags & RXf_EVAL_SEEN)
8795 PerlIO_printf(Perl_debug_log, "with eval ");
8796 PerlIO_printf(Perl_debug_log, "\n");
8797 DEBUG_FLAGS_r(regdump_extflags("r->extflags: ",r->extflags));
8799 PERL_UNUSED_CONTEXT;
8801 #endif /* DEBUGGING */
8805 - regprop - printable representation of opcode
8808 Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
8813 RXi_GET_DECL(prog,progi);
8814 GET_RE_DEBUG_FLAGS_DECL;
8817 sv_setpvn(sv, "", 0);
8819 if (OP(o) > REGNODE_MAX) /* regnode.type is unsigned */
8820 /* It would be nice to FAIL() here, but this may be called from
8821 regexec.c, and it would be hard to supply pRExC_state. */
8822 Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d", (int)OP(o), (int)REGNODE_MAX);
8823 sv_catpv(sv, PL_reg_name[OP(o)]); /* Take off const! */
8825 k = PL_regkind[OP(o)];
8829 /* Using is_utf8_string() (via PERL_PV_UNI_DETECT)
8830 * is a crude hack but it may be the best for now since
8831 * we have no flag "this EXACTish node was UTF-8"
8833 pv_pretty(sv, STRING(o), STR_LEN(o), 60, PL_colors[0], PL_colors[1],
8834 PERL_PV_ESCAPE_UNI_DETECT |
8835 PERL_PV_PRETTY_ELLIPSES |
8836 PERL_PV_PRETTY_LTGT |
8837 PERL_PV_PRETTY_NOCLEAR
8839 } else if (k == TRIE) {
8840 /* print the details of the trie in dumpuntil instead, as
8841 * progi->data isn't available here */
8842 const char op = OP(o);
8843 const U32 n = ARG(o);
8844 const reg_ac_data * const ac = IS_TRIE_AC(op) ?
8845 (reg_ac_data *)progi->data->data[n] :
8847 const reg_trie_data * const trie
8848 = (reg_trie_data*)progi->data->data[!IS_TRIE_AC(op) ? n : ac->trie];
8850 Perl_sv_catpvf(aTHX_ sv, "-%s",PL_reg_name[o->flags]);
8851 DEBUG_TRIE_COMPILE_r(
8852 Perl_sv_catpvf(aTHX_ sv,
8853 "<S:%"UVuf"/%"IVdf" W:%"UVuf" L:%"UVuf"/%"UVuf" C:%"UVuf"/%"UVuf">",
8854 (UV)trie->startstate,
8855 (IV)trie->statecount-1, /* -1 because of the unused 0 element */
8856 (UV)trie->wordcount,
8859 (UV)TRIE_CHARCOUNT(trie),
8860 (UV)trie->uniquecharcount
8863 if ( IS_ANYOF_TRIE(op) || trie->bitmap ) {
8865 int rangestart = -1;
8866 U8* bitmap = IS_ANYOF_TRIE(op) ? (U8*)ANYOF_BITMAP(o) : (U8*)TRIE_BITMAP(trie);
8868 for (i = 0; i <= 256; i++) {
8869 if (i < 256 && BITMAP_TEST(bitmap,i)) {
8870 if (rangestart == -1)
8872 } else if (rangestart != -1) {
8873 if (i <= rangestart + 3)
8874 for (; rangestart < i; rangestart++)
8875 put_byte(sv, rangestart);
8877 put_byte(sv, rangestart);
8879 put_byte(sv, i - 1);
8887 } else if (k == CURLY) {
8888 if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
8889 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
8890 Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
8892 else if (k == WHILEM && o->flags) /* Ordinal/of */
8893 Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
8894 else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP || OP(o)==ACCEPT) {
8895 Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o)); /* Parenth number */
8896 if ( prog->paren_names ) {
8897 if ( k != REF || OP(o) < NREF) {
8898 AV *list= (AV *)progi->data->data[progi->name_list_idx];
8899 SV **name= av_fetch(list, ARG(o), 0 );
8901 Perl_sv_catpvf(aTHX_ sv, " '%"SVf"'", SVfARG(*name));
8904 AV *list= (AV *)progi->data->data[ progi->name_list_idx ];
8905 SV *sv_dat=(SV*)progi->data->data[ ARG( o ) ];
8906 I32 *nums=(I32*)SvPVX(sv_dat);
8907 SV **name= av_fetch(list, nums[0], 0 );
8910 for ( n=0; n<SvIVX(sv_dat); n++ ) {
8911 Perl_sv_catpvf(aTHX_ sv, "%s%"IVdf,
8912 (n ? "," : ""), (IV)nums[n]);
8914 Perl_sv_catpvf(aTHX_ sv, " '%"SVf"'", SVfARG(*name));
8918 } else if (k == GOSUB)
8919 Perl_sv_catpvf(aTHX_ sv, "%d[%+d]", (int)ARG(o),(int)ARG2L(o)); /* Paren and offset */
8920 else if (k == VERB) {
8922 Perl_sv_catpvf(aTHX_ sv, ":%"SVf,
8923 SVfARG((SV*)progi->data->data[ ARG( o ) ]));
8924 } else if (k == LOGICAL)
8925 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* 2: embedded, otherwise 1 */
8926 else if (k == FOLDCHAR)
8927 Perl_sv_catpvf(aTHX_ sv, "[0x%"UVXf"]", PTR2UV(ARG(o)) );
8928 else if (k == ANYOF) {
8929 int i, rangestart = -1;
8930 const U8 flags = ANYOF_FLAGS(o);
8932 /* Should be synchronized with * ANYOF_ #xdefines in regcomp.h */
8933 static const char * const anyofs[] = {
8966 if (flags & ANYOF_LOCALE)
8967 sv_catpvs(sv, "{loc}");
8968 if (flags & ANYOF_FOLD)
8969 sv_catpvs(sv, "{i}");
8970 Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
8971 if (flags & ANYOF_INVERT)
8973 for (i = 0; i <= 256; i++) {
8974 if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
8975 if (rangestart == -1)
8977 } else if (rangestart != -1) {
8978 if (i <= rangestart + 3)
8979 for (; rangestart < i; rangestart++)
8980 put_byte(sv, rangestart);
8982 put_byte(sv, rangestart);
8984 put_byte(sv, i - 1);
8990 if (o->flags & ANYOF_CLASS)
8991 for (i = 0; i < (int)(sizeof(anyofs)/sizeof(char*)); i++)
8992 if (ANYOF_CLASS_TEST(o,i))
8993 sv_catpv(sv, anyofs[i]);
8995 if (flags & ANYOF_UNICODE)
8996 sv_catpvs(sv, "{unicode}");
8997 else if (flags & ANYOF_UNICODE_ALL)
8998 sv_catpvs(sv, "{unicode_all}");
9002 SV * const sw = regclass_swash(prog, o, FALSE, &lv, 0);
9006 U8 s[UTF8_MAXBYTES_CASE+1];
9008 for (i = 0; i <= 256; i++) { /* just the first 256 */
9009 uvchr_to_utf8(s, i);
9011 if (i < 256 && swash_fetch(sw, s, TRUE)) {
9012 if (rangestart == -1)
9014 } else if (rangestart != -1) {
9015 if (i <= rangestart + 3)
9016 for (; rangestart < i; rangestart++) {
9017 const U8 * const e = uvchr_to_utf8(s,rangestart);
9019 for(p = s; p < e; p++)
9023 const U8 *e = uvchr_to_utf8(s,rangestart);
9025 for (p = s; p < e; p++)
9028 e = uvchr_to_utf8(s, i-1);
9029 for (p = s; p < e; p++)
9036 sv_catpvs(sv, "..."); /* et cetera */
9040 char *s = savesvpv(lv);
9041 char * const origs = s;
9043 while (*s && *s != '\n')
9047 const char * const t = ++s;
9065 Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
9067 else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
9068 Perl_sv_catpvf(aTHX_ sv, "[%d]", -(o->flags));
9070 PERL_UNUSED_CONTEXT;
9071 PERL_UNUSED_ARG(sv);
9073 PERL_UNUSED_ARG(prog);
9074 #endif /* DEBUGGING */
9078 Perl_re_intuit_string(pTHX_ REGEXP * const prog)
9079 { /* Assume that RE_INTUIT is set */
9081 GET_RE_DEBUG_FLAGS_DECL;
9082 PERL_UNUSED_CONTEXT;
9086 const char * const s = SvPV_nolen_const(prog->check_substr
9087 ? prog->check_substr : prog->check_utf8);
9089 if (!PL_colorset) reginitcolors();
9090 PerlIO_printf(Perl_debug_log,
9091 "%sUsing REx %ssubstr:%s \"%s%.60s%s%s\"\n",
9093 prog->check_substr ? "" : "utf8 ",
9094 PL_colors[5],PL_colors[0],
9097 (strlen(s) > 60 ? "..." : ""));
9100 return prog->check_substr ? prog->check_substr : prog->check_utf8;
9106 handles refcounting and freeing the perl core regexp structure. When
9107 it is necessary to actually free the structure the first thing it
9108 does is call the 'free' method of the regexp_engine associated to to
9109 the regexp, allowing the handling of the void *pprivate; member
9110 first. (This routine is not overridable by extensions, which is why
9111 the extensions free is called first.)
9113 See regdupe and regdupe_internal if you change anything here.
9115 #ifndef PERL_IN_XSUB_RE
9117 Perl_pregfree(pTHX_ struct regexp *r)
9120 GET_RE_DEBUG_FLAGS_DECL;
9122 if (!r || (--r->refcnt > 0))
9125 ReREFCNT_dec(r->mother_re);
9127 CALLREGFREE_PVT(r); /* free the private data */
9129 SvREFCNT_dec(r->paren_names);
9130 Safefree(r->wrapped);
9133 if (r->anchored_substr)
9134 SvREFCNT_dec(r->anchored_substr);
9135 if (r->anchored_utf8)
9136 SvREFCNT_dec(r->anchored_utf8);
9137 if (r->float_substr)
9138 SvREFCNT_dec(r->float_substr);
9140 SvREFCNT_dec(r->float_utf8);
9141 Safefree(r->substrs);
9143 RX_MATCH_COPY_FREE(r);
9144 #ifdef PERL_OLD_COPY_ON_WRITE
9146 SvREFCNT_dec(r->saved_copy);
9155 This is a hacky workaround to the structural issue of match results
9156 being stored in the regexp structure which is in turn stored in
9157 PL_curpm/PL_reg_curpm. The problem is that due to qr// the pattern
9158 could be PL_curpm in multiple contexts, and could require multiple
9159 result sets being associated with the pattern simultaneously, such
9160 as when doing a recursive match with (??{$qr})
9162 The solution is to make a lightweight copy of the regexp structure
9163 when a qr// is returned from the code executed by (??{$qr}) this
9164 lightweight copy doesnt actually own any of its data except for
9165 the starp/end and the actual regexp structure itself.
9171 Perl_reg_temp_copy (pTHX_ struct regexp *r) {
9173 register const I32 npar = r->nparens+1;
9174 (void)ReREFCNT_inc(r);
9175 Newx(ret, 1, regexp);
9176 StructCopy(r, ret, regexp);
9177 Newx(ret->offs, npar, regexp_paren_pair);
9178 Copy(r->offs, ret->offs, npar, regexp_paren_pair);
9181 Newx(ret->substrs, 1, struct reg_substr_data);
9182 StructCopy(r->substrs, ret->substrs, struct reg_substr_data);
9184 SvREFCNT_inc_void(ret->anchored_substr);
9185 SvREFCNT_inc_void(ret->anchored_utf8);
9186 SvREFCNT_inc_void(ret->float_substr);
9187 SvREFCNT_inc_void(ret->float_utf8);
9189 /* check_substr and check_utf8, if non-NULL, point to either their
9190 anchored or float namesakes, and don't hold a second reference. */
9192 RX_MATCH_COPIED_off(ret);
9193 #ifdef PERL_OLD_COPY_ON_WRITE
9194 ret->saved_copy = NULL;
9203 /* regfree_internal()
9205 Free the private data in a regexp. This is overloadable by
9206 extensions. Perl takes care of the regexp structure in pregfree(),
9207 this covers the *pprivate pointer which technically perldoesnt
9208 know about, however of course we have to handle the
9209 regexp_internal structure when no extension is in use.
9211 Note this is called before freeing anything in the regexp
9216 Perl_regfree_internal(pTHX_ REGEXP * const r)
9220 GET_RE_DEBUG_FLAGS_DECL;
9226 SV *dsv= sv_newmortal();
9227 RE_PV_QUOTED_DECL(s, (r->extflags & RXf_UTF8),
9228 dsv, r->precomp, r->prelen, 60);
9229 PerlIO_printf(Perl_debug_log,"%sFreeing REx:%s %s\n",
9230 PL_colors[4],PL_colors[5],s);
9233 #ifdef RE_TRACK_PATTERN_OFFSETS
9235 Safefree(ri->u.offsets); /* 20010421 MJD */
9238 int n = ri->data->count;
9239 PAD* new_comppad = NULL;
9244 /* If you add a ->what type here, update the comment in regcomp.h */
9245 switch (ri->data->what[n]) {
9249 SvREFCNT_dec((SV*)ri->data->data[n]);
9252 Safefree(ri->data->data[n]);
9255 new_comppad = (AV*)ri->data->data[n];
9258 if (new_comppad == NULL)
9259 Perl_croak(aTHX_ "panic: pregfree comppad");
9260 PAD_SAVE_LOCAL(old_comppad,
9261 /* Watch out for global destruction's random ordering. */
9262 (SvTYPE(new_comppad) == SVt_PVAV) ? new_comppad : NULL
9265 refcnt = OpREFCNT_dec((OP_4tree*)ri->data->data[n]);
9268 op_free((OP_4tree*)ri->data->data[n]);
9270 PAD_RESTORE_LOCAL(old_comppad);
9271 SvREFCNT_dec((SV*)new_comppad);
9277 { /* Aho Corasick add-on structure for a trie node.
9278 Used in stclass optimization only */
9280 reg_ac_data *aho=(reg_ac_data*)ri->data->data[n];
9282 refcount = --aho->refcount;
9285 PerlMemShared_free(aho->states);
9286 PerlMemShared_free(aho->fail);
9287 /* do this last!!!! */
9288 PerlMemShared_free(ri->data->data[n]);
9289 PerlMemShared_free(ri->regstclass);
9295 /* trie structure. */
9297 reg_trie_data *trie=(reg_trie_data*)ri->data->data[n];
9299 refcount = --trie->refcount;
9302 PerlMemShared_free(trie->charmap);
9303 PerlMemShared_free(trie->states);
9304 PerlMemShared_free(trie->trans);
9306 PerlMemShared_free(trie->bitmap);
9308 PerlMemShared_free(trie->wordlen);
9310 PerlMemShared_free(trie->jump);
9312 PerlMemShared_free(trie->nextword);
9313 /* do this last!!!! */
9314 PerlMemShared_free(ri->data->data[n]);
9319 Perl_croak(aTHX_ "panic: regfree data code '%c'", ri->data->what[n]);
9322 Safefree(ri->data->what);
9329 #define sv_dup_inc(s,t) SvREFCNT_inc(sv_dup(s,t))
9330 #define av_dup_inc(s,t) (AV*)SvREFCNT_inc(sv_dup((SV*)s,t))
9331 #define hv_dup_inc(s,t) (HV*)SvREFCNT_inc(sv_dup((SV*)s,t))
9332 #define SAVEPVN(p,n) ((p) ? savepvn(p,n) : NULL)
9335 re_dup - duplicate a regexp.
9337 This routine is expected to clone a given regexp structure. It is not
9338 compiler under USE_ITHREADS.
9340 After all of the core data stored in struct regexp is duplicated
9341 the regexp_engine.dupe method is used to copy any private data
9342 stored in the *pprivate pointer. This allows extensions to handle
9343 any duplication it needs to do.
9345 See pregfree() and regfree_internal() if you change anything here.
9347 #if defined(USE_ITHREADS)
9348 #ifndef PERL_IN_XSUB_RE
9350 Perl_re_dup(pTHX_ const regexp *r, CLONE_PARAMS *param)
9357 return (REGEXP *)NULL;
9359 if ((ret = (REGEXP *)ptr_table_fetch(PL_ptr_table, r)))
9363 npar = r->nparens+1;
9364 Newx(ret, 1, regexp);
9365 StructCopy(r, ret, regexp);
9366 Newx(ret->offs, npar, regexp_paren_pair);
9367 Copy(r->offs, ret->offs, npar, regexp_paren_pair);
9369 /* no need to copy these */
9370 Newx(ret->swap, npar, regexp_paren_pair);
9374 /* Do it this way to avoid reading from *r after the StructCopy().
9375 That way, if any of the sv_dup_inc()s dislodge *r from the L1
9376 cache, it doesn't matter. */
9377 const bool anchored = r->check_substr == r->anchored_substr;
9378 Newx(ret->substrs, 1, struct reg_substr_data);
9379 StructCopy(r->substrs, ret->substrs, struct reg_substr_data);
9381 ret->anchored_substr = sv_dup_inc(ret->anchored_substr, param);
9382 ret->anchored_utf8 = sv_dup_inc(ret->anchored_utf8, param);
9383 ret->float_substr = sv_dup_inc(ret->float_substr, param);
9384 ret->float_utf8 = sv_dup_inc(ret->float_utf8, param);
9386 /* check_substr and check_utf8, if non-NULL, point to either their
9387 anchored or float namesakes, and don't hold a second reference. */
9389 if (ret->check_substr) {
9391 assert(r->check_utf8 == r->anchored_utf8);
9392 ret->check_substr = ret->anchored_substr;
9393 ret->check_utf8 = ret->anchored_utf8;
9395 assert(r->check_substr == r->float_substr);
9396 assert(r->check_utf8 == r->float_utf8);
9397 ret->check_substr = ret->float_substr;
9398 ret->check_utf8 = ret->float_utf8;
9403 ret->wrapped = SAVEPVN(ret->wrapped, ret->wraplen+1);
9404 ret->precomp = ret->wrapped + (ret->precomp - ret->wrapped);
9405 ret->paren_names = hv_dup_inc(ret->paren_names, param);
9408 RXi_SET(ret,CALLREGDUPE_PVT(ret,param));
9410 if (RX_MATCH_COPIED(ret))
9411 ret->subbeg = SAVEPVN(ret->subbeg, ret->sublen);
9414 #ifdef PERL_OLD_COPY_ON_WRITE
9415 ret->saved_copy = NULL;
9418 ret->mother_re = NULL;
9420 ret->seen_evals = 0;
9422 ptr_table_store(PL_ptr_table, r, ret);
9425 #endif /* PERL_IN_XSUB_RE */
9430 This is the internal complement to regdupe() which is used to copy
9431 the structure pointed to by the *pprivate pointer in the regexp.
9432 This is the core version of the extension overridable cloning hook.
9433 The regexp structure being duplicated will be copied by perl prior
9434 to this and will be provided as the regexp *r argument, however
9435 with the /old/ structures pprivate pointer value. Thus this routine
9436 may override any copying normally done by perl.
9438 It returns a pointer to the new regexp_internal structure.
9442 Perl_regdupe_internal(pTHX_ REGEXP * const r, CLONE_PARAMS *param)
9445 regexp_internal *reti;
9449 npar = r->nparens+1;
9452 Newxc(reti, sizeof(regexp_internal) + (len+1)*sizeof(regnode), char, regexp_internal);
9453 Copy(ri->program, reti->program, len+1, regnode);
9456 reti->regstclass = NULL;
9460 const int count = ri->data->count;
9463 Newxc(d, sizeof(struct reg_data) + count*sizeof(void *),
9464 char, struct reg_data);
9465 Newx(d->what, count, U8);
9468 for (i = 0; i < count; i++) {
9469 d->what[i] = ri->data->what[i];
9470 switch (d->what[i]) {
9471 /* legal options are one of: sSfpontTu
9472 see also regcomp.h and pregfree() */
9475 case 'p': /* actually an AV, but the dup function is identical. */
9476 case 'u': /* actually an HV, but the dup function is identical. */
9477 d->data[i] = sv_dup_inc((SV *)ri->data->data[i], param);
9480 /* This is cheating. */
9481 Newx(d->data[i], 1, struct regnode_charclass_class);
9482 StructCopy(ri->data->data[i], d->data[i],
9483 struct regnode_charclass_class);
9484 reti->regstclass = (regnode*)d->data[i];
9487 /* Compiled op trees are readonly and in shared memory,
9488 and can thus be shared without duplication. */
9490 d->data[i] = (void*)OpREFCNT_inc((OP*)ri->data->data[i]);
9494 /* Trie stclasses are readonly and can thus be shared
9495 * without duplication. We free the stclass in pregfree
9496 * when the corresponding reg_ac_data struct is freed.
9498 reti->regstclass= ri->regstclass;
9502 ((reg_trie_data*)ri->data->data[i])->refcount++;
9506 d->data[i] = ri->data->data[i];
9509 Perl_croak(aTHX_ "panic: re_dup unknown data code '%c'", ri->data->what[i]);
9518 reti->name_list_idx = ri->name_list_idx;
9520 #ifdef RE_TRACK_PATTERN_OFFSETS
9521 if (ri->u.offsets) {
9522 Newx(reti->u.offsets, 2*len+1, U32);
9523 Copy(ri->u.offsets, reti->u.offsets, 2*len+1, U32);
9526 SetProgLen(reti,len);
9532 #endif /* USE_ITHREADS */
9537 converts a regexp embedded in a MAGIC struct to its stringified form,
9538 caching the converted form in the struct and returns the cached
9541 If lp is nonnull then it is used to return the length of the
9544 If flags is nonnull and the returned string contains UTF8 then
9545 (*flags & 1) will be true.
9547 If haseval is nonnull then it is used to return whether the pattern
9550 Normally called via macro:
9552 CALLREG_STRINGIFY(mg,&len,&utf8);
9556 CALLREG_AS_STR(mg,&lp,&flags,&haseval)
9558 See sv_2pv_flags() in sv.c for an example of internal usage.
9561 #ifndef PERL_IN_XSUB_RE
9564 Perl_reg_stringify(pTHX_ MAGIC *mg, STRLEN *lp, U32 *flags, I32 *haseval ) {
9566 const regexp * const re = (regexp *)mg->mg_obj;
9568 *haseval = re->seen_evals;
9570 *flags = ((re->extflags & RXf_UTF8) ? 1 : 0);
9577 - regnext - dig the "next" pointer out of a node
9580 Perl_regnext(pTHX_ register regnode *p)
9583 register I32 offset;
9588 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
9597 S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
9600 STRLEN l1 = strlen(pat1);
9601 STRLEN l2 = strlen(pat2);
9604 const char *message;
9610 Copy(pat1, buf, l1 , char);
9611 Copy(pat2, buf + l1, l2 , char);
9612 buf[l1 + l2] = '\n';
9613 buf[l1 + l2 + 1] = '\0';
9615 /* ANSI variant takes additional second argument */
9616 va_start(args, pat2);
9620 msv = vmess(buf, &args);
9622 message = SvPV_const(msv,l1);
9625 Copy(message, buf, l1 , char);
9626 buf[l1-1] = '\0'; /* Overwrite \n */
9627 Perl_croak(aTHX_ "%s", buf);
9630 /* XXX Here's a total kludge. But we need to re-enter for swash routines. */
9632 #ifndef PERL_IN_XSUB_RE
9634 Perl_save_re_context(pTHX)
9638 struct re_save_state *state;
9640 SAVEVPTR(PL_curcop);
9641 SSGROW(SAVESTACK_ALLOC_FOR_RE_SAVE_STATE + 1);
9643 state = (struct re_save_state *)(PL_savestack + PL_savestack_ix);
9644 PL_savestack_ix += SAVESTACK_ALLOC_FOR_RE_SAVE_STATE;
9645 SSPUSHINT(SAVEt_RE_STATE);
9647 Copy(&PL_reg_state, state, 1, struct re_save_state);
9649 PL_reg_start_tmp = 0;
9650 PL_reg_start_tmpl = 0;
9651 PL_reg_oldsaved = NULL;
9652 PL_reg_oldsavedlen = 0;
9654 PL_reg_leftiter = 0;
9655 PL_reg_poscache = NULL;
9656 PL_reg_poscache_size = 0;
9657 #ifdef PERL_OLD_COPY_ON_WRITE
9661 /* Save $1..$n (#18107: UTF-8 s/(\w+)/uc($1)/e); AMS 20021106. */
9663 const REGEXP * const rx = PM_GETRE(PL_curpm);
9666 for (i = 1; i <= rx->nparens; i++) {
9667 char digits[TYPE_CHARS(long)];
9668 const STRLEN len = my_snprintf(digits, sizeof(digits), "%lu", (long)i);
9669 GV *const *const gvp
9670 = (GV**)hv_fetch(PL_defstash, digits, len, 0);
9673 GV * const gv = *gvp;
9674 if (SvTYPE(gv) == SVt_PVGV && GvSV(gv))
9684 clear_re(pTHX_ void *r)
9687 ReREFCNT_dec((regexp *)r);
9693 S_put_byte(pTHX_ SV *sv, int c)
9695 /* Our definition of isPRINT() ignores locales, so only bytes that are
9696 not part of UTF-8 are considered printable. I assume that the same
9697 holds for UTF-EBCDIC.
9698 Also, code point 255 is not printable in either (it's E0 in EBCDIC,
9699 which Wikipedia says:
9701 EO, or Eight Ones, is an 8-bit EBCDIC character code represented as all
9702 ones (binary 1111 1111, hexadecimal FF). It is similar, but not
9703 identical, to the ASCII delete (DEL) or rubout control character.
9704 ) So the old condition can be simplified to !isPRINT(c) */
9706 Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
9708 const char string = c;
9709 if (c == '-' || c == ']' || c == '\\' || c == '^')
9710 sv_catpvs(sv, "\\");
9711 sv_catpvn(sv, &string, 1);
9716 #define CLEAR_OPTSTART \
9717 if (optstart) STMT_START { \
9718 DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log, " (%"IVdf" nodes)\n", (IV)(node - optstart))); \
9722 #define DUMPUNTIL(b,e) CLEAR_OPTSTART; node=dumpuntil(r,start,(b),(e),last,sv,indent+1,depth+1);
9724 STATIC const regnode *
9725 S_dumpuntil(pTHX_ const regexp *r, const regnode *start, const regnode *node,
9726 const regnode *last, const regnode *plast,
9727 SV* sv, I32 indent, U32 depth)
9730 register U8 op = PSEUDO; /* Arbitrary non-END op. */
9731 register const regnode *next;
9732 const regnode *optstart= NULL;
9735 GET_RE_DEBUG_FLAGS_DECL;
9737 #ifdef DEBUG_DUMPUNTIL
9738 PerlIO_printf(Perl_debug_log, "--- %d : %d - %d - %d\n",indent,node-start,
9739 last ? last-start : 0,plast ? plast-start : 0);
9742 if (plast && plast < last)
9745 while (PL_regkind[op] != END && (!last || node < last)) {
9746 /* While that wasn't END last time... */
9749 if (op == CLOSE || op == WHILEM)
9751 next = regnext((regnode *)node);
9754 if (OP(node) == OPTIMIZED) {
9755 if (!optstart && RE_DEBUG_FLAG(RE_DEBUG_COMPILE_OPTIMISE))
9762 regprop(r, sv, node);
9763 PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
9764 (int)(2*indent + 1), "", SvPVX_const(sv));
9766 if (OP(node) != OPTIMIZED) {
9767 if (next == NULL) /* Next ptr. */
9768 PerlIO_printf(Perl_debug_log, " (0)");
9769 else if (PL_regkind[(U8)op] == BRANCH && PL_regkind[OP(next)] != BRANCH )
9770 PerlIO_printf(Perl_debug_log, " (FAIL)");
9772 PerlIO_printf(Perl_debug_log, " (%"IVdf")", (IV)(next - start));
9773 (void)PerlIO_putc(Perl_debug_log, '\n');
9777 if (PL_regkind[(U8)op] == BRANCHJ) {
9780 register const regnode *nnode = (OP(next) == LONGJMP
9781 ? regnext((regnode *)next)
9783 if (last && nnode > last)
9785 DUMPUNTIL(NEXTOPER(NEXTOPER(node)), nnode);
9788 else if (PL_regkind[(U8)op] == BRANCH) {
9790 DUMPUNTIL(NEXTOPER(node), next);
9792 else if ( PL_regkind[(U8)op] == TRIE ) {
9793 const regnode *this_trie = node;
9794 const char op = OP(node);
9795 const U32 n = ARG(node);
9796 const reg_ac_data * const ac = op>=AHOCORASICK ?
9797 (reg_ac_data *)ri->data->data[n] :
9799 const reg_trie_data * const trie =
9800 (reg_trie_data*)ri->data->data[op<AHOCORASICK ? n : ac->trie];
9802 AV *const trie_words = (AV *) ri->data->data[n + TRIE_WORDS_OFFSET];
9804 const regnode *nextbranch= NULL;
9806 sv_setpvn(sv, "", 0);
9807 for (word_idx= 0; word_idx < (I32)trie->wordcount; word_idx++) {
9808 SV ** const elem_ptr = av_fetch(trie_words,word_idx,0);
9810 PerlIO_printf(Perl_debug_log, "%*s%s ",
9811 (int)(2*(indent+3)), "",
9812 elem_ptr ? pv_pretty(sv, SvPV_nolen_const(*elem_ptr), SvCUR(*elem_ptr), 60,
9813 PL_colors[0], PL_colors[1],
9814 (SvUTF8(*elem_ptr) ? PERL_PV_ESCAPE_UNI : 0) |
9815 PERL_PV_PRETTY_ELLIPSES |
9821 U16 dist= trie->jump[word_idx+1];
9822 PerlIO_printf(Perl_debug_log, "(%"UVuf")\n",
9823 (UV)((dist ? this_trie + dist : next) - start));
9826 nextbranch= this_trie + trie->jump[0];
9827 DUMPUNTIL(this_trie + dist, nextbranch);
9829 if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
9830 nextbranch= regnext((regnode *)nextbranch);
9832 PerlIO_printf(Perl_debug_log, "\n");
9835 if (last && next > last)
9840 else if ( op == CURLY ) { /* "next" might be very big: optimizer */
9841 DUMPUNTIL(NEXTOPER(node) + EXTRA_STEP_2ARGS,
9842 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1);
9844 else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
9846 DUMPUNTIL(NEXTOPER(node) + EXTRA_STEP_2ARGS, next);
9848 else if ( op == PLUS || op == STAR) {
9849 DUMPUNTIL(NEXTOPER(node), NEXTOPER(node) + 1);
9851 else if (op == ANYOF) {
9852 /* arglen 1 + class block */
9853 node += 1 + ((ANYOF_FLAGS(node) & ANYOF_LARGE)
9854 ? ANYOF_CLASS_SKIP : ANYOF_SKIP);
9855 node = NEXTOPER(node);
9857 else if (PL_regkind[(U8)op] == EXACT) {
9858 /* Literal string, where present. */
9859 node += NODE_SZ_STR(node) - 1;
9860 node = NEXTOPER(node);
9863 node = NEXTOPER(node);
9864 node += regarglen[(U8)op];
9866 if (op == CURLYX || op == OPEN)
9870 #ifdef DEBUG_DUMPUNTIL
9871 PerlIO_printf(Perl_debug_log, "--- %d\n", (int)indent);
9876 #endif /* DEBUGGING */
9880 * c-indentation-style: bsd
9882 * indent-tabs-mode: t
9885 * ex: set ts=8 sts=4 sw=4 noet:
projects / p5sagit/p5-mst-13.2.git / blob