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, 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; /* Code-emit pointer; ®dummy = don't = compiling */
113 I32 naughty; /* How bad is this pattern? */
114 I32 sawback; /* Did we see \1, ...? */
116 I32 size; /* Code size. */
117 I32 npar; /* Capture buffer count, (OPEN). */
118 I32 cpar; /* Capture buffer count, (CLOSE). */
119 I32 nestroot; /* root parens we are in - used by accept */
123 regnode **open_parens; /* pointers to open parens */
124 regnode **close_parens; /* pointers to close parens */
125 regnode *opend; /* END node in program */
127 HV *charnames; /* cache of named sequences */
128 HV *paren_names; /* Paren names */
129 regnode **recurse; /* Recurse regops */
130 I32 recurse_count; /* Number of recurse regops */
132 char *starttry; /* -Dr: where regtry was called. */
133 #define RExC_starttry (pRExC_state->starttry)
136 const char *lastparse;
138 #define RExC_lastparse (pRExC_state->lastparse)
139 #define RExC_lastnum (pRExC_state->lastnum)
143 #define RExC_flags (pRExC_state->flags)
144 #define RExC_precomp (pRExC_state->precomp)
145 #define RExC_rx (pRExC_state->rx)
146 #define RExC_rxi (pRExC_state->rxi)
147 #define RExC_start (pRExC_state->start)
148 #define RExC_end (pRExC_state->end)
149 #define RExC_parse (pRExC_state->parse)
150 #define RExC_whilem_seen (pRExC_state->whilem_seen)
151 #define RExC_offsets (pRExC_state->rxi->offsets) /* I am not like the others */
152 #define RExC_emit (pRExC_state->emit)
153 #define RExC_emit_start (pRExC_state->emit_start)
154 #define RExC_naughty (pRExC_state->naughty)
155 #define RExC_sawback (pRExC_state->sawback)
156 #define RExC_seen (pRExC_state->seen)
157 #define RExC_size (pRExC_state->size)
158 #define RExC_npar (pRExC_state->npar)
159 #define RExC_nestroot (pRExC_state->nestroot)
160 #define RExC_extralen (pRExC_state->extralen)
161 #define RExC_seen_zerolen (pRExC_state->seen_zerolen)
162 #define RExC_seen_evals (pRExC_state->seen_evals)
163 #define RExC_utf8 (pRExC_state->utf8)
164 #define RExC_charnames (pRExC_state->charnames)
165 #define RExC_open_parens (pRExC_state->open_parens)
166 #define RExC_close_parens (pRExC_state->close_parens)
167 #define RExC_opend (pRExC_state->opend)
168 #define RExC_paren_names (pRExC_state->paren_names)
169 #define RExC_recurse (pRExC_state->recurse)
170 #define RExC_recurse_count (pRExC_state->recurse_count)
172 #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
173 #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
174 ((*s) == '{' && regcurly(s)))
177 #undef SPSTART /* dratted cpp namespace... */
180 * Flags to be passed up and down.
182 #define WORST 0 /* Worst case. */
183 #define HASWIDTH 0x1 /* Known to match non-null strings. */
184 #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */
185 #define SPSTART 0x4 /* Starts with * or +. */
186 #define TRYAGAIN 0x8 /* Weeded out a declaration. */
188 #define REG_NODE_NUM(x) ((x) ? (int)((x)-RExC_emit_start) : -1)
190 /* whether trie related optimizations are enabled */
191 #if PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION
192 #define TRIE_STUDY_OPT
193 #define FULL_TRIE_STUDY
199 #define PBYTE(u8str,paren) ((U8*)(u8str))[(paren) >> 3]
200 #define PBITVAL(paren) (1 << ((paren) & 7))
201 #define PAREN_TEST(u8str,paren) ( PBYTE(u8str,paren) & PBITVAL(paren))
202 #define PAREN_SET(u8str,paren) PBYTE(u8str,paren) |= PBITVAL(paren)
203 #define PAREN_UNSET(u8str,paren) PBYTE(u8str,paren) &= (~PBITVAL(paren))
206 /* About scan_data_t.
208 During optimisation we recurse through the regexp program performing
209 various inplace (keyhole style) optimisations. In addition study_chunk
210 and scan_commit populate this data structure with information about
211 what strings MUST appear in the pattern. We look for the longest
212 string that must appear for at a fixed location, and we look for the
213 longest string that may appear at a floating location. So for instance
218 Both 'FOO' and 'A' are fixed strings. Both 'B' and 'BAR' are floating
219 strings (because they follow a .* construct). study_chunk will identify
220 both FOO and BAR as being the longest fixed and floating strings respectively.
222 The strings can be composites, for instance
226 will result in a composite fixed substring 'foo'.
228 For each string some basic information is maintained:
230 - offset or min_offset
231 This is the position the string must appear at, or not before.
232 It also implicitly (when combined with minlenp) tells us how many
233 character must match before the string we are searching.
234 Likewise when combined with minlenp and the length of the string
235 tells us how many characters must appear after the string we have
239 Only used for floating strings. This is the rightmost point that
240 the string can appear at. Ifset to I32 max it indicates that the
241 string can occur infinitely far to the right.
244 A pointer to the minimum length of the pattern that the string
245 was found inside. This is important as in the case of positive
246 lookahead or positive lookbehind we can have multiple patterns
251 The minimum length of the pattern overall is 3, the minimum length
252 of the lookahead part is 3, but the minimum length of the part that
253 will actually match is 1. So 'FOO's minimum length is 3, but the
254 minimum length for the F is 1. This is important as the minimum length
255 is used to determine offsets in front of and behind the string being
256 looked for. Since strings can be composites this is the length of the
257 pattern at the time it was commited with a scan_commit. Note that
258 the length is calculated by study_chunk, so that the minimum lengths
259 are not known until the full pattern has been compiled, thus the
260 pointer to the value.
264 In the case of lookbehind the string being searched for can be
265 offset past the start point of the final matching string.
266 If this value was just blithely removed from the min_offset it would
267 invalidate some of the calculations for how many chars must match
268 before or after (as they are derived from min_offset and minlen and
269 the length of the string being searched for).
270 When the final pattern is compiled and the data is moved from the
271 scan_data_t structure into the regexp structure the information
272 about lookbehind is factored in, with the information that would
273 have been lost precalculated in the end_shift field for the
276 The fields pos_min and pos_delta are used to store the minimum offset
277 and the delta to the maximum offset at the current point in the pattern.
281 typedef struct scan_data_t {
282 /*I32 len_min; unused */
283 /*I32 len_delta; unused */
287 I32 last_end; /* min value, <0 unless valid. */
290 SV **longest; /* Either &l_fixed, or &l_float. */
291 SV *longest_fixed; /* longest fixed string found in pattern */
292 I32 offset_fixed; /* offset where it starts */
293 I32 *minlen_fixed; /* pointer to the minlen relevent to the string */
294 I32 lookbehind_fixed; /* is the position of the string modfied by LB */
295 SV *longest_float; /* longest floating string found in pattern */
296 I32 offset_float_min; /* earliest point in string it can appear */
297 I32 offset_float_max; /* latest point in string it can appear */
298 I32 *minlen_float; /* pointer to the minlen relevent to the string */
299 I32 lookbehind_float; /* is the position of the string modified by LB */
303 struct regnode_charclass_class *start_class;
307 * Forward declarations for pregcomp()'s friends.
310 static const scan_data_t zero_scan_data =
311 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ,0};
313 #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
314 #define SF_BEFORE_SEOL 0x0001
315 #define SF_BEFORE_MEOL 0x0002
316 #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
317 #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
320 # define SF_FIX_SHIFT_EOL (0+2)
321 # define SF_FL_SHIFT_EOL (0+4)
323 # define SF_FIX_SHIFT_EOL (+2)
324 # define SF_FL_SHIFT_EOL (+4)
327 #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
328 #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
330 #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
331 #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
332 #define SF_IS_INF 0x0040
333 #define SF_HAS_PAR 0x0080
334 #define SF_IN_PAR 0x0100
335 #define SF_HAS_EVAL 0x0200
336 #define SCF_DO_SUBSTR 0x0400
337 #define SCF_DO_STCLASS_AND 0x0800
338 #define SCF_DO_STCLASS_OR 0x1000
339 #define SCF_DO_STCLASS (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
340 #define SCF_WHILEM_VISITED_POS 0x2000
342 #define SCF_TRIE_RESTUDY 0x4000 /* Do restudy? */
343 #define SCF_SEEN_ACCEPT 0x8000
345 #define UTF (RExC_utf8 != 0)
346 #define LOC ((RExC_flags & RXf_PMf_LOCALE) != 0)
347 #define FOLD ((RExC_flags & RXf_PMf_FOLD) != 0)
349 #define OOB_UNICODE 12345678
350 #define OOB_NAMEDCLASS -1
352 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
353 #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
356 /* length of regex to show in messages that don't mark a position within */
357 #define RegexLengthToShowInErrorMessages 127
360 * If MARKER[12] are adjusted, be sure to adjust the constants at the top
361 * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
362 * op/pragma/warn/regcomp.
364 #define MARKER1 "<-- HERE" /* marker as it appears in the description */
365 #define MARKER2 " <-- HERE " /* marker as it appears within the regex */
367 #define REPORT_LOCATION " in regex; marked by " MARKER1 " in m/%.*s" MARKER2 "%s/"
370 * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
371 * arg. Show regex, up to a maximum length. If it's too long, chop and add
374 #define _FAIL(code) STMT_START { \
375 const char *ellipses = ""; \
376 IV len = RExC_end - RExC_precomp; \
379 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
380 if (len > RegexLengthToShowInErrorMessages) { \
381 /* chop 10 shorter than the max, to ensure meaning of "..." */ \
382 len = RegexLengthToShowInErrorMessages - 10; \
388 #define FAIL(msg) _FAIL( \
389 Perl_croak(aTHX_ "%s in regex m/%.*s%s/", \
390 msg, (int)len, RExC_precomp, ellipses))
392 #define FAIL2(msg,arg) _FAIL( \
393 Perl_croak(aTHX_ msg " in regex m/%.*s%s/", \
394 arg, (int)len, RExC_precomp, ellipses))
397 * Simple_vFAIL -- like FAIL, but marks the current location in the scan
399 #define Simple_vFAIL(m) STMT_START { \
400 const IV offset = RExC_parse - RExC_precomp; \
401 Perl_croak(aTHX_ "%s" REPORT_LOCATION, \
402 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
406 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
408 #define vFAIL(m) STMT_START { \
410 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
415 * Like Simple_vFAIL(), but accepts two arguments.
417 #define Simple_vFAIL2(m,a1) STMT_START { \
418 const IV offset = RExC_parse - RExC_precomp; \
419 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, \
420 (int)offset, RExC_precomp, RExC_precomp + offset); \
424 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
426 #define vFAIL2(m,a1) STMT_START { \
428 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
429 Simple_vFAIL2(m, a1); \
434 * Like Simple_vFAIL(), but accepts three arguments.
436 #define Simple_vFAIL3(m, a1, a2) STMT_START { \
437 const IV offset = RExC_parse - RExC_precomp; \
438 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, \
439 (int)offset, RExC_precomp, RExC_precomp + offset); \
443 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
445 #define vFAIL3(m,a1,a2) STMT_START { \
447 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx); \
448 Simple_vFAIL3(m, a1, a2); \
452 * Like Simple_vFAIL(), but accepts four arguments.
454 #define Simple_vFAIL4(m, a1, a2, a3) STMT_START { \
455 const IV offset = RExC_parse - RExC_precomp; \
456 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3, \
457 (int)offset, RExC_precomp, RExC_precomp + offset); \
460 #define vWARN(loc,m) STMT_START { \
461 const IV offset = loc - RExC_precomp; \
462 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s" REPORT_LOCATION, \
463 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
466 #define vWARNdep(loc,m) STMT_START { \
467 const IV offset = loc - RExC_precomp; \
468 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_REGEXP), \
469 "%s" REPORT_LOCATION, \
470 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
474 #define vWARN2(loc, m, a1) STMT_START { \
475 const IV offset = loc - RExC_precomp; \
476 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
477 a1, (int)offset, RExC_precomp, RExC_precomp + offset); \
480 #define vWARN3(loc, m, a1, a2) STMT_START { \
481 const IV offset = loc - RExC_precomp; \
482 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
483 a1, a2, (int)offset, RExC_precomp, RExC_precomp + offset); \
486 #define vWARN4(loc, m, a1, a2, a3) STMT_START { \
487 const IV offset = loc - RExC_precomp; \
488 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
489 a1, a2, a3, (int)offset, RExC_precomp, RExC_precomp + offset); \
492 #define vWARN5(loc, m, a1, a2, a3, a4) STMT_START { \
493 const IV offset = loc - RExC_precomp; \
494 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
495 a1, a2, a3, a4, (int)offset, RExC_precomp, RExC_precomp + offset); \
499 /* Allow for side effects in s */
500 #define REGC(c,s) STMT_START { \
501 if (!SIZE_ONLY) *(s) = (c); else (void)(s); \
504 /* Macros for recording node offsets. 20001227 mjd@plover.com
505 * Nodes are numbered 1, 2, 3, 4. Node #n's position is recorded in
506 * element 2*n-1 of the array. Element #2n holds the byte length node #n.
507 * Element 0 holds the number n.
508 * Position is 1 indexed.
511 #define Set_Node_Offset_To_R(node,byte) STMT_START { \
513 MJD_OFFSET_DEBUG(("** (%d) offset of node %d is %d.\n", \
514 __LINE__, (int)(node), (int)(byte))); \
516 Perl_croak(aTHX_ "value of node is %d in Offset macro", (int)(node)); \
518 RExC_offsets[2*(node)-1] = (byte); \
523 #define Set_Node_Offset(node,byte) \
524 Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
525 #define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
527 #define Set_Node_Length_To_R(node,len) STMT_START { \
529 MJD_OFFSET_DEBUG(("** (%d) size of node %d is %d.\n", \
530 __LINE__, (int)(node), (int)(len))); \
532 Perl_croak(aTHX_ "value of node is %d in Length macro", (int)(node)); \
534 RExC_offsets[2*(node)] = (len); \
539 #define Set_Node_Length(node,len) \
540 Set_Node_Length_To_R((node)-RExC_emit_start, len)
541 #define Set_Cur_Node_Length(len) Set_Node_Length(RExC_emit, len)
542 #define Set_Node_Cur_Length(node) \
543 Set_Node_Length(node, RExC_parse - parse_start)
545 /* Get offsets and lengths */
546 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
547 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
549 #define Set_Node_Offset_Length(node,offset,len) STMT_START { \
550 Set_Node_Offset_To_R((node)-RExC_emit_start, (offset)); \
551 Set_Node_Length_To_R((node)-RExC_emit_start, (len)); \
555 #if PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS
556 #define EXPERIMENTAL_INPLACESCAN
559 #define DEBUG_STUDYDATA(str,data,depth) \
560 DEBUG_OPTIMISE_MORE_r(if(data){ \
561 PerlIO_printf(Perl_debug_log, \
562 "%*s" str "Pos:%"IVdf"/%"IVdf \
563 " Flags: 0x%"UVXf" Whilem_c: %"IVdf" Lcp: %"IVdf" %s", \
564 (int)(depth)*2, "", \
565 (IV)((data)->pos_min), \
566 (IV)((data)->pos_delta), \
567 (UV)((data)->flags), \
568 (IV)((data)->whilem_c), \
569 (IV)((data)->last_closep ? *((data)->last_closep) : -1), \
570 is_inf ? "INF " : "" \
572 if ((data)->last_found) \
573 PerlIO_printf(Perl_debug_log, \
574 "Last:'%s' %"IVdf":%"IVdf"/%"IVdf" %sFixed:'%s' @ %"IVdf \
575 " %sFloat: '%s' @ %"IVdf"/%"IVdf"", \
576 SvPVX_const((data)->last_found), \
577 (IV)((data)->last_end), \
578 (IV)((data)->last_start_min), \
579 (IV)((data)->last_start_max), \
580 ((data)->longest && \
581 (data)->longest==&((data)->longest_fixed)) ? "*" : "", \
582 SvPVX_const((data)->longest_fixed), \
583 (IV)((data)->offset_fixed), \
584 ((data)->longest && \
585 (data)->longest==&((data)->longest_float)) ? "*" : "", \
586 SvPVX_const((data)->longest_float), \
587 (IV)((data)->offset_float_min), \
588 (IV)((data)->offset_float_max) \
590 PerlIO_printf(Perl_debug_log,"\n"); \
593 static void clear_re(pTHX_ void *r);
595 /* Mark that we cannot extend a found fixed substring at this point.
596 Update the longest found anchored substring and the longest found
597 floating substrings if needed. */
600 S_scan_commit(pTHX_ const RExC_state_t *pRExC_state, scan_data_t *data, I32 *minlenp, int is_inf)
602 const STRLEN l = CHR_SVLEN(data->last_found);
603 const STRLEN old_l = CHR_SVLEN(*data->longest);
604 GET_RE_DEBUG_FLAGS_DECL;
606 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
607 SvSetMagicSV(*data->longest, data->last_found);
608 if (*data->longest == data->longest_fixed) {
609 data->offset_fixed = l ? data->last_start_min : data->pos_min;
610 if (data->flags & SF_BEFORE_EOL)
612 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
614 data->flags &= ~SF_FIX_BEFORE_EOL;
615 data->minlen_fixed=minlenp;
616 data->lookbehind_fixed=0;
618 else { /* *data->longest == data->longest_float */
619 data->offset_float_min = l ? data->last_start_min : data->pos_min;
620 data->offset_float_max = (l
621 ? data->last_start_max
622 : data->pos_min + data->pos_delta);
623 if (is_inf || (U32)data->offset_float_max > (U32)I32_MAX)
624 data->offset_float_max = I32_MAX;
625 if (data->flags & SF_BEFORE_EOL)
627 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
629 data->flags &= ~SF_FL_BEFORE_EOL;
630 data->minlen_float=minlenp;
631 data->lookbehind_float=0;
634 SvCUR_set(data->last_found, 0);
636 SV * const sv = data->last_found;
637 if (SvUTF8(sv) && SvMAGICAL(sv)) {
638 MAGIC * const mg = mg_find(sv, PERL_MAGIC_utf8);
644 data->flags &= ~SF_BEFORE_EOL;
645 DEBUG_STUDYDATA("cl_anything: ",data,0);
648 /* Can match anything (initialization) */
650 S_cl_anything(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
652 ANYOF_CLASS_ZERO(cl);
653 ANYOF_BITMAP_SETALL(cl);
654 cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
656 cl->flags |= ANYOF_LOCALE;
659 /* Can match anything (initialization) */
661 S_cl_is_anything(const struct regnode_charclass_class *cl)
665 for (value = 0; value <= ANYOF_MAX; value += 2)
666 if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
668 if (!(cl->flags & ANYOF_UNICODE_ALL))
670 if (!ANYOF_BITMAP_TESTALLSET((const void*)cl))
675 /* Can match anything (initialization) */
677 S_cl_init(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
679 Zero(cl, 1, struct regnode_charclass_class);
681 cl_anything(pRExC_state, cl);
685 S_cl_init_zero(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
687 Zero(cl, 1, struct regnode_charclass_class);
689 cl_anything(pRExC_state, cl);
691 cl->flags |= ANYOF_LOCALE;
694 /* 'And' a given class with another one. Can create false positives */
695 /* We assume that cl is not inverted */
697 S_cl_and(struct regnode_charclass_class *cl,
698 const struct regnode_charclass_class *and_with)
701 assert(and_with->type == ANYOF);
702 if (!(and_with->flags & ANYOF_CLASS)
703 && !(cl->flags & ANYOF_CLASS)
704 && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
705 && !(and_with->flags & ANYOF_FOLD)
706 && !(cl->flags & ANYOF_FOLD)) {
709 if (and_with->flags & ANYOF_INVERT)
710 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
711 cl->bitmap[i] &= ~and_with->bitmap[i];
713 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
714 cl->bitmap[i] &= and_with->bitmap[i];
715 } /* XXXX: logic is complicated otherwise, leave it along for a moment. */
716 if (!(and_with->flags & ANYOF_EOS))
717 cl->flags &= ~ANYOF_EOS;
719 if (cl->flags & ANYOF_UNICODE_ALL && and_with->flags & ANYOF_UNICODE &&
720 !(and_with->flags & ANYOF_INVERT)) {
721 cl->flags &= ~ANYOF_UNICODE_ALL;
722 cl->flags |= ANYOF_UNICODE;
723 ARG_SET(cl, ARG(and_with));
725 if (!(and_with->flags & ANYOF_UNICODE_ALL) &&
726 !(and_with->flags & ANYOF_INVERT))
727 cl->flags &= ~ANYOF_UNICODE_ALL;
728 if (!(and_with->flags & (ANYOF_UNICODE|ANYOF_UNICODE_ALL)) &&
729 !(and_with->flags & ANYOF_INVERT))
730 cl->flags &= ~ANYOF_UNICODE;
733 /* 'OR' a given class with another one. Can create false positives */
734 /* We assume that cl is not inverted */
736 S_cl_or(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, const struct regnode_charclass_class *or_with)
738 if (or_with->flags & ANYOF_INVERT) {
740 * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
741 * <= (B1 | !B2) | (CL1 | !CL2)
742 * which is wasteful if CL2 is small, but we ignore CL2:
743 * (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1
744 * XXXX Can we handle case-fold? Unclear:
745 * (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) =
746 * (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i'))
748 if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
749 && !(or_with->flags & ANYOF_FOLD)
750 && !(cl->flags & ANYOF_FOLD) ) {
753 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
754 cl->bitmap[i] |= ~or_with->bitmap[i];
755 } /* XXXX: logic is complicated otherwise */
757 cl_anything(pRExC_state, cl);
760 /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */
761 if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
762 && (!(or_with->flags & ANYOF_FOLD)
763 || (cl->flags & ANYOF_FOLD)) ) {
766 /* OR char bitmap and class bitmap separately */
767 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
768 cl->bitmap[i] |= or_with->bitmap[i];
769 if (or_with->flags & ANYOF_CLASS) {
770 for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++)
771 cl->classflags[i] |= or_with->classflags[i];
772 cl->flags |= ANYOF_CLASS;
775 else { /* XXXX: logic is complicated, leave it along for a moment. */
776 cl_anything(pRExC_state, cl);
779 if (or_with->flags & ANYOF_EOS)
780 cl->flags |= ANYOF_EOS;
782 if (cl->flags & ANYOF_UNICODE && or_with->flags & ANYOF_UNICODE &&
783 ARG(cl) != ARG(or_with)) {
784 cl->flags |= ANYOF_UNICODE_ALL;
785 cl->flags &= ~ANYOF_UNICODE;
787 if (or_with->flags & ANYOF_UNICODE_ALL) {
788 cl->flags |= ANYOF_UNICODE_ALL;
789 cl->flags &= ~ANYOF_UNICODE;
793 #define TRIE_LIST_ITEM(state,idx) (trie->states[state].trans.list)[ idx ]
794 #define TRIE_LIST_CUR(state) ( TRIE_LIST_ITEM( state, 0 ).forid )
795 #define TRIE_LIST_LEN(state) ( TRIE_LIST_ITEM( state, 0 ).newstate )
796 #define TRIE_LIST_USED(idx) ( trie->states[state].trans.list ? (TRIE_LIST_CUR( idx ) - 1) : 0 )
801 dump_trie(trie,widecharmap,revcharmap)
802 dump_trie_interim_list(trie,widecharmap,revcharmap,next_alloc)
803 dump_trie_interim_table(trie,widecharmap,revcharmap,next_alloc)
805 These routines dump out a trie in a somewhat readable format.
806 The _interim_ variants are used for debugging the interim
807 tables that are used to generate the final compressed
808 representation which is what dump_trie expects.
810 Part of the reason for their existance is to provide a form
811 of documentation as to how the different representations function.
816 Dumps the final compressed table form of the trie to Perl_debug_log.
817 Used for debugging make_trie().
821 S_dump_trie(pTHX_ const struct _reg_trie_data *trie, HV *widecharmap,
822 AV *revcharmap, U32 depth)
825 SV *sv=sv_newmortal();
826 int colwidth= widecharmap ? 6 : 4;
827 GET_RE_DEBUG_FLAGS_DECL;
830 PerlIO_printf( Perl_debug_log, "%*sChar : %-6s%-6s%-4s ",
831 (int)depth * 2 + 2,"",
832 "Match","Base","Ofs" );
834 for( state = 0 ; state < trie->uniquecharcount ; state++ ) {
835 SV ** const tmp = av_fetch( revcharmap, state, 0);
837 PerlIO_printf( Perl_debug_log, "%*s",
839 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
840 PL_colors[0], PL_colors[1],
841 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
842 PERL_PV_ESCAPE_FIRSTCHAR
847 PerlIO_printf( Perl_debug_log, "\n%*sState|-----------------------",
848 (int)depth * 2 + 2,"");
850 for( state = 0 ; state < trie->uniquecharcount ; state++ )
851 PerlIO_printf( Perl_debug_log, "%.*s", colwidth, "--------");
852 PerlIO_printf( Perl_debug_log, "\n");
854 for( state = 1 ; state < trie->statecount ; state++ ) {
855 const U32 base = trie->states[ state ].trans.base;
857 PerlIO_printf( Perl_debug_log, "%*s#%4"UVXf"|", (int)depth * 2 + 2,"", (UV)state);
859 if ( trie->states[ state ].wordnum ) {
860 PerlIO_printf( Perl_debug_log, " W%4X", trie->states[ state ].wordnum );
862 PerlIO_printf( Perl_debug_log, "%6s", "" );
865 PerlIO_printf( Perl_debug_log, " @%4"UVXf" ", (UV)base );
870 while( ( base + ofs < trie->uniquecharcount ) ||
871 ( base + ofs - trie->uniquecharcount < trie->lasttrans
872 && trie->trans[ base + ofs - trie->uniquecharcount ].check != state))
875 PerlIO_printf( Perl_debug_log, "+%2"UVXf"[ ", (UV)ofs);
877 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
878 if ( ( base + ofs >= trie->uniquecharcount ) &&
879 ( base + ofs - trie->uniquecharcount < trie->lasttrans ) &&
880 trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
882 PerlIO_printf( Perl_debug_log, "%*"UVXf,
884 (UV)trie->trans[ base + ofs - trie->uniquecharcount ].next );
886 PerlIO_printf( Perl_debug_log, "%*s",colwidth," ." );
890 PerlIO_printf( Perl_debug_log, "]");
893 PerlIO_printf( Perl_debug_log, "\n" );
897 Dumps a fully constructed but uncompressed trie in list form.
898 List tries normally only are used for construction when the number of
899 possible chars (trie->uniquecharcount) is very high.
900 Used for debugging make_trie().
903 S_dump_trie_interim_list(pTHX_ const struct _reg_trie_data *trie,
904 HV *widecharmap, AV *revcharmap, U32 next_alloc,
908 SV *sv=sv_newmortal();
909 int colwidth= widecharmap ? 6 : 4;
910 GET_RE_DEBUG_FLAGS_DECL;
911 /* print out the table precompression. */
912 PerlIO_printf( Perl_debug_log, "%*sState :Word | Transition Data\n%*s%s",
913 (int)depth * 2 + 2,"", (int)depth * 2 + 2,"",
914 "------:-----+-----------------\n" );
916 for( state=1 ; state < next_alloc ; state ++ ) {
919 PerlIO_printf( Perl_debug_log, "%*s %4"UVXf" :",
920 (int)depth * 2 + 2,"", (UV)state );
921 if ( ! trie->states[ state ].wordnum ) {
922 PerlIO_printf( Perl_debug_log, "%5s| ","");
924 PerlIO_printf( Perl_debug_log, "W%4x| ",
925 trie->states[ state ].wordnum
928 for( charid = 1 ; charid <= TRIE_LIST_USED( state ) ; charid++ ) {
929 SV ** const tmp = av_fetch( revcharmap, TRIE_LIST_ITEM(state,charid).forid, 0);
931 PerlIO_printf( Perl_debug_log, "%*s:%3X=%4"UVXf" | ",
933 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
934 PL_colors[0], PL_colors[1],
935 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
936 PERL_PV_ESCAPE_FIRSTCHAR
938 TRIE_LIST_ITEM(state,charid).forid,
939 (UV)TRIE_LIST_ITEM(state,charid).newstate
942 PerlIO_printf(Perl_debug_log, "\n%*s| ",
943 (int)((depth * 2) + 14), "");
946 PerlIO_printf( Perl_debug_log, "\n");
951 Dumps a fully constructed but uncompressed trie in table form.
952 This is the normal DFA style state transition table, with a few
953 twists to facilitate compression later.
954 Used for debugging make_trie().
957 S_dump_trie_interim_table(pTHX_ const struct _reg_trie_data *trie,
958 HV *widecharmap, AV *revcharmap, U32 next_alloc,
963 SV *sv=sv_newmortal();
964 int colwidth= widecharmap ? 6 : 4;
965 GET_RE_DEBUG_FLAGS_DECL;
968 print out the table precompression so that we can do a visual check
969 that they are identical.
972 PerlIO_printf( Perl_debug_log, "%*sChar : ",(int)depth * 2 + 2,"" );
974 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
975 SV ** const tmp = av_fetch( revcharmap, charid, 0);
977 PerlIO_printf( Perl_debug_log, "%*s",
979 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
980 PL_colors[0], PL_colors[1],
981 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
982 PERL_PV_ESCAPE_FIRSTCHAR
988 PerlIO_printf( Perl_debug_log, "\n%*sState+-",(int)depth * 2 + 2,"" );
990 for( charid=0 ; charid < trie->uniquecharcount ; charid++ ) {
991 PerlIO_printf( Perl_debug_log, "%.*s", colwidth,"--------");
994 PerlIO_printf( Perl_debug_log, "\n" );
996 for( state=1 ; state < next_alloc ; state += trie->uniquecharcount ) {
998 PerlIO_printf( Perl_debug_log, "%*s%4"UVXf" : ",
999 (int)depth * 2 + 2,"",
1000 (UV)TRIE_NODENUM( state ) );
1002 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
1003 UV v=(UV)SAFE_TRIE_NODENUM( trie->trans[ state + charid ].next );
1005 PerlIO_printf( Perl_debug_log, "%*"UVXf, colwidth, v );
1007 PerlIO_printf( Perl_debug_log, "%*s", colwidth, "." );
1009 if ( ! trie->states[ TRIE_NODENUM( state ) ].wordnum ) {
1010 PerlIO_printf( Perl_debug_log, " (%4"UVXf")\n", (UV)trie->trans[ state ].check );
1012 PerlIO_printf( Perl_debug_log, " (%4"UVXf") W%4X\n", (UV)trie->trans[ state ].check,
1013 trie->states[ TRIE_NODENUM( state ) ].wordnum );
1020 /* make_trie(startbranch,first,last,tail,word_count,flags,depth)
1021 startbranch: the first branch in the whole branch sequence
1022 first : start branch of sequence of branch-exact nodes.
1023 May be the same as startbranch
1024 last : Thing following the last branch.
1025 May be the same as tail.
1026 tail : item following the branch sequence
1027 count : words in the sequence
1028 flags : currently the OP() type we will be building one of /EXACT(|F|Fl)/
1029 depth : indent depth
1031 Inplace optimizes a sequence of 2 or more Branch-Exact nodes into a TRIE node.
1033 A trie is an N'ary tree where the branches are determined by digital
1034 decomposition of the key. IE, at the root node you look up the 1st character and
1035 follow that branch repeat until you find the end of the branches. Nodes can be
1036 marked as "accepting" meaning they represent a complete word. Eg:
1040 would convert into the following structure. Numbers represent states, letters
1041 following numbers represent valid transitions on the letter from that state, if
1042 the number is in square brackets it represents an accepting state, otherwise it
1043 will be in parenthesis.
1045 +-h->+-e->[3]-+-r->(8)-+-s->[9]
1049 (1) +-i->(6)-+-s->[7]
1051 +-s->(3)-+-h->(4)-+-e->[5]
1053 Accept Word Mapping: 3=>1 (he),5=>2 (she), 7=>3 (his), 9=>4 (hers)
1055 This shows that when matching against the string 'hers' we will begin at state 1
1056 read 'h' and move to state 2, read 'e' and move to state 3 which is accepting,
1057 then read 'r' and go to state 8 followed by 's' which takes us to state 9 which
1058 is also accepting. Thus we know that we can match both 'he' and 'hers' with a
1059 single traverse. We store a mapping from accepting to state to which word was
1060 matched, and then when we have multiple possibilities we try to complete the
1061 rest of the regex in the order in which they occured in the alternation.
1063 The only prior NFA like behaviour that would be changed by the TRIE support is
1064 the silent ignoring of duplicate alternations which are of the form:
1066 / (DUPE|DUPE) X? (?{ ... }) Y /x
1068 Thus EVAL blocks follwing a trie may be called a different number of times with
1069 and without the optimisation. With the optimisations dupes will be silently
1070 ignored. This inconsistant behaviour of EVAL type nodes is well established as
1071 the following demonstrates:
1073 'words'=~/(word|word|word)(?{ print $1 })[xyz]/
1075 which prints out 'word' three times, but
1077 'words'=~/(word|word|word)(?{ print $1 })S/
1079 which doesnt print it out at all. This is due to other optimisations kicking in.
1081 Example of what happens on a structural level:
1083 The regexp /(ac|ad|ab)+/ will produce the folowing debug output:
1085 1: CURLYM[1] {1,32767}(18)
1096 This would be optimizable with startbranch=5, first=5, last=16, tail=16
1097 and should turn into:
1099 1: CURLYM[1] {1,32767}(18)
1101 [Words:3 Chars Stored:6 Unique Chars:4 States:5 NCP:1]
1109 Cases where tail != last would be like /(?foo|bar)baz/:
1119 which would be optimizable with startbranch=1, first=1, last=7, tail=8
1120 and would end up looking like:
1123 [Words:2 Chars Stored:6 Unique Chars:5 States:7 NCP:1]
1130 d = uvuni_to_utf8_flags(d, uv, 0);
1132 is the recommended Unicode-aware way of saying
1137 #define TRIE_STORE_REVCHAR \
1139 SV *tmp = newSVpvs(""); \
1140 if (UTF) SvUTF8_on(tmp); \
1141 Perl_sv_catpvf( aTHX_ tmp, "%c", (int)uvc ); \
1142 av_push( revcharmap, tmp ); \
1145 #define TRIE_READ_CHAR STMT_START { \
1149 if ( foldlen > 0 ) { \
1150 uvc = utf8n_to_uvuni( scan, UTF8_MAXLEN, &len, uniflags ); \
1155 uvc = utf8n_to_uvuni( (const U8*)uc, UTF8_MAXLEN, &len, uniflags);\
1156 uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \
1157 foldlen -= UNISKIP( uvc ); \
1158 scan = foldbuf + UNISKIP( uvc ); \
1161 uvc = utf8n_to_uvuni( (const U8*)uc, UTF8_MAXLEN, &len, uniflags);\
1171 #define TRIE_LIST_PUSH(state,fid,ns) STMT_START { \
1172 if ( TRIE_LIST_CUR( state ) >=TRIE_LIST_LEN( state ) ) { \
1173 U32 ging = TRIE_LIST_LEN( state ) *= 2; \
1174 Renew( trie->states[ state ].trans.list, ging, reg_trie_trans_le ); \
1176 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).forid = fid; \
1177 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).newstate = ns; \
1178 TRIE_LIST_CUR( state )++; \
1181 #define TRIE_LIST_NEW(state) STMT_START { \
1182 Newxz( trie->states[ state ].trans.list, \
1183 4, reg_trie_trans_le ); \
1184 TRIE_LIST_CUR( state ) = 1; \
1185 TRIE_LIST_LEN( state ) = 4; \
1188 #define TRIE_HANDLE_WORD(state) STMT_START { \
1189 U16 dupe= trie->states[ state ].wordnum; \
1190 regnode * const noper_next = regnext( noper ); \
1192 if (trie->wordlen) \
1193 trie->wordlen[ curword ] = wordlen; \
1195 /* store the word for dumping */ \
1197 if (OP(noper) != NOTHING) \
1198 tmp = newSVpvn(STRING(noper), STR_LEN(noper)); \
1200 tmp = newSVpvn( "", 0 ); \
1201 if ( UTF ) SvUTF8_on( tmp ); \
1202 av_push( trie_words, tmp ); \
1207 if ( noper_next < tail ) { \
1209 trie->jump = (U16 *) PerlMemShared_calloc( word_count + 1, sizeof(U16) ); \
1210 trie->jump[curword] = (U16)(noper_next - convert); \
1212 jumper = noper_next; \
1214 nextbranch= regnext(cur); \
1218 /* So it's a dupe. This means we need to maintain a */\
1219 /* linked-list from the first to the next. */\
1220 /* we only allocate the nextword buffer when there */\
1221 /* a dupe, so first time we have to do the allocation */\
1222 if (!trie->nextword) \
1223 trie->nextword = (U16 *) \
1224 PerlMemShared_calloc( word_count + 1, sizeof(U16)); \
1225 while ( trie->nextword[dupe] ) \
1226 dupe= trie->nextword[dupe]; \
1227 trie->nextword[dupe]= curword; \
1229 /* we haven't inserted this word yet. */ \
1230 trie->states[ state ].wordnum = curword; \
1235 #define TRIE_TRANS_STATE(state,base,ucharcount,charid,special) \
1236 ( ( base + charid >= ucharcount \
1237 && base + charid < ubound \
1238 && state == trie->trans[ base - ucharcount + charid ].check \
1239 && trie->trans[ base - ucharcount + charid ].next ) \
1240 ? trie->trans[ base - ucharcount + charid ].next \
1241 : ( state==1 ? special : 0 ) \
1245 #define MADE_JUMP_TRIE 2
1246 #define MADE_EXACT_TRIE 4
1249 S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *first, regnode *last, regnode *tail, U32 word_count, U32 flags, U32 depth)
1252 /* first pass, loop through and scan words */
1253 reg_trie_data *trie;
1254 HV *widecharmap = NULL;
1255 AV *revcharmap = newAV();
1257 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1262 regnode *jumper = NULL;
1263 regnode *nextbranch = NULL;
1264 regnode *convert = NULL;
1265 /* we just use folder as a flag in utf8 */
1266 const U8 * const folder = ( flags == EXACTF
1268 : ( flags == EXACTFL
1275 const U32 data_slot = add_data( pRExC_state, 4, "tuuu" );
1276 AV *trie_words = NULL;
1277 /* along with revcharmap, this only used during construction but both are
1278 * useful during debugging so we store them in the struct when debugging.
1281 const U32 data_slot = add_data( pRExC_state, 2, "tu" );
1282 STRLEN trie_charcount=0;
1284 SV *re_trie_maxbuff;
1285 GET_RE_DEBUG_FLAGS_DECL;
1287 PERL_UNUSED_ARG(depth);
1290 trie = (reg_trie_data *) PerlMemShared_calloc( 1, sizeof(reg_trie_data) );
1292 trie->startstate = 1;
1293 trie->wordcount = word_count;
1294 RExC_rxi->data->data[ data_slot ] = (void*)trie;
1295 trie->charmap = (U16 *) PerlMemShared_calloc( 256, sizeof(U16) );
1296 if (!(UTF && folder))
1297 trie->bitmap = (char *) PerlMemShared_calloc( ANYOF_BITMAP_SIZE, 1 );
1299 trie_words = newAV();
1302 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
1303 if (!SvIOK(re_trie_maxbuff)) {
1304 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
1307 PerlIO_printf( Perl_debug_log,
1308 "%*smake_trie start==%d, first==%d, last==%d, tail==%d depth=%d\n",
1309 (int)depth * 2 + 2, "",
1310 REG_NODE_NUM(startbranch),REG_NODE_NUM(first),
1311 REG_NODE_NUM(last), REG_NODE_NUM(tail),
1315 /* Find the node we are going to overwrite */
1316 if ( first == startbranch && OP( last ) != BRANCH ) {
1317 /* whole branch chain */
1320 /* branch sub-chain */
1321 convert = NEXTOPER( first );
1324 /* -- First loop and Setup --
1326 We first traverse the branches and scan each word to determine if it
1327 contains widechars, and how many unique chars there are, this is
1328 important as we have to build a table with at least as many columns as we
1331 We use an array of integers to represent the character codes 0..255
1332 (trie->charmap) and we use a an HV* to store unicode characters. We use the
1333 native representation of the character value as the key and IV's for the
1336 *TODO* If we keep track of how many times each character is used we can
1337 remap the columns so that the table compression later on is more
1338 efficient in terms of memory by ensuring most common value is in the
1339 middle and the least common are on the outside. IMO this would be better
1340 than a most to least common mapping as theres a decent chance the most
1341 common letter will share a node with the least common, meaning the node
1342 will not be compressable. With a middle is most common approach the worst
1343 case is when we have the least common nodes twice.
1347 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1348 regnode * const noper = NEXTOPER( cur );
1349 const U8 *uc = (U8*)STRING( noper );
1350 const U8 * const e = uc + STR_LEN( noper );
1352 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1353 const U8 *scan = (U8*)NULL;
1354 U32 wordlen = 0; /* required init */
1357 if (OP(noper) == NOTHING) {
1362 TRIE_BITMAP_SET(trie,*uc);
1363 if ( folder ) TRIE_BITMAP_SET(trie,folder[ *uc ]);
1365 for ( ; uc < e ; uc += len ) {
1366 TRIE_CHARCOUNT(trie)++;
1370 if ( !trie->charmap[ uvc ] ) {
1371 trie->charmap[ uvc ]=( ++trie->uniquecharcount );
1373 trie->charmap[ folder[ uvc ] ] = trie->charmap[ uvc ];
1379 widecharmap = newHV();
1381 svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 1 );
1384 Perl_croak( aTHX_ "error creating/fetching widecharmap entry for 0x%"UVXf, uvc );
1386 if ( !SvTRUE( *svpp ) ) {
1387 sv_setiv( *svpp, ++trie->uniquecharcount );
1392 if( cur == first ) {
1395 } else if (chars < trie->minlen) {
1397 } else if (chars > trie->maxlen) {
1401 } /* end first pass */
1402 DEBUG_TRIE_COMPILE_r(
1403 PerlIO_printf( Perl_debug_log, "%*sTRIE(%s): W:%d C:%d Uq:%d Min:%d Max:%d\n",
1404 (int)depth * 2 + 2,"",
1405 ( widecharmap ? "UTF8" : "NATIVE" ), (int)word_count,
1406 (int)TRIE_CHARCOUNT(trie), trie->uniquecharcount,
1407 (int)trie->minlen, (int)trie->maxlen )
1409 trie->wordlen = (U32 *) PerlMemShared_calloc( word_count, sizeof(U32) );
1412 We now know what we are dealing with in terms of unique chars and
1413 string sizes so we can calculate how much memory a naive
1414 representation using a flat table will take. If it's over a reasonable
1415 limit (as specified by ${^RE_TRIE_MAXBUF}) we use a more memory
1416 conservative but potentially much slower representation using an array
1419 At the end we convert both representations into the same compressed
1420 form that will be used in regexec.c for matching with. The latter
1421 is a form that cannot be used to construct with but has memory
1422 properties similar to the list form and access properties similar
1423 to the table form making it both suitable for fast searches and
1424 small enough that its feasable to store for the duration of a program.
1426 See the comment in the code where the compressed table is produced
1427 inplace from the flat tabe representation for an explanation of how
1428 the compression works.
1433 if ( (IV)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1) > SvIV(re_trie_maxbuff) ) {
1435 Second Pass -- Array Of Lists Representation
1437 Each state will be represented by a list of charid:state records
1438 (reg_trie_trans_le) the first such element holds the CUR and LEN
1439 points of the allocated array. (See defines above).
1441 We build the initial structure using the lists, and then convert
1442 it into the compressed table form which allows faster lookups
1443 (but cant be modified once converted).
1446 STRLEN transcount = 1;
1448 DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log,
1449 "%*sCompiling trie using list compiler\n",
1450 (int)depth * 2 + 2, ""));
1452 trie->states = (reg_trie_state *)
1453 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
1454 sizeof(reg_trie_state) );
1458 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1460 regnode * const noper = NEXTOPER( cur );
1461 U8 *uc = (U8*)STRING( noper );
1462 const U8 * const e = uc + STR_LEN( noper );
1463 U32 state = 1; /* required init */
1464 U16 charid = 0; /* sanity init */
1465 U8 *scan = (U8*)NULL; /* sanity init */
1466 STRLEN foldlen = 0; /* required init */
1467 U32 wordlen = 0; /* required init */
1468 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1470 if (OP(noper) != NOTHING) {
1471 for ( ; uc < e ; uc += len ) {
1476 charid = trie->charmap[ uvc ];
1478 SV** const svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 0);
1482 charid=(U16)SvIV( *svpp );
1485 /* charid is now 0 if we dont know the char read, or nonzero if we do */
1492 if ( !trie->states[ state ].trans.list ) {
1493 TRIE_LIST_NEW( state );
1495 for ( check = 1; check <= TRIE_LIST_USED( state ); check++ ) {
1496 if ( TRIE_LIST_ITEM( state, check ).forid == charid ) {
1497 newstate = TRIE_LIST_ITEM( state, check ).newstate;
1502 newstate = next_alloc++;
1503 TRIE_LIST_PUSH( state, charid, newstate );
1508 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %"IVdf, uvc );
1512 TRIE_HANDLE_WORD(state);
1514 } /* end second pass */
1516 /* next alloc is the NEXT state to be allocated */
1517 trie->statecount = next_alloc;
1518 trie->states = (reg_trie_state *)
1519 PerlMemShared_realloc( trie->states,
1521 * sizeof(reg_trie_state) );
1523 /* and now dump it out before we compress it */
1524 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_list(trie, widecharmap,
1525 revcharmap, next_alloc,
1529 trie->trans = (reg_trie_trans *)
1530 PerlMemShared_calloc( transcount, sizeof(reg_trie_trans) );
1537 for( state=1 ; state < next_alloc ; state ++ ) {
1541 DEBUG_TRIE_COMPILE_MORE_r(
1542 PerlIO_printf( Perl_debug_log, "tp: %d zp: %d ",tp,zp)
1546 if (trie->states[state].trans.list) {
1547 U16 minid=TRIE_LIST_ITEM( state, 1).forid;
1551 for( idx = 2 ; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
1552 const U16 forid = TRIE_LIST_ITEM( state, idx).forid;
1553 if ( forid < minid ) {
1555 } else if ( forid > maxid ) {
1559 if ( transcount < tp + maxid - minid + 1) {
1561 trie->trans = (reg_trie_trans *)
1562 PerlMemShared_realloc( trie->trans,
1564 * sizeof(reg_trie_trans) );
1565 Zero( trie->trans + (transcount / 2), transcount / 2 , reg_trie_trans );
1567 base = trie->uniquecharcount + tp - minid;
1568 if ( maxid == minid ) {
1570 for ( ; zp < tp ; zp++ ) {
1571 if ( ! trie->trans[ zp ].next ) {
1572 base = trie->uniquecharcount + zp - minid;
1573 trie->trans[ zp ].next = TRIE_LIST_ITEM( state, 1).newstate;
1574 trie->trans[ zp ].check = state;
1580 trie->trans[ tp ].next = TRIE_LIST_ITEM( state, 1).newstate;
1581 trie->trans[ tp ].check = state;
1586 for ( idx=1; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
1587 const U32 tid = base - trie->uniquecharcount + TRIE_LIST_ITEM( state, idx ).forid;
1588 trie->trans[ tid ].next = TRIE_LIST_ITEM( state, idx ).newstate;
1589 trie->trans[ tid ].check = state;
1591 tp += ( maxid - minid + 1 );
1593 Safefree(trie->states[ state ].trans.list);
1596 DEBUG_TRIE_COMPILE_MORE_r(
1597 PerlIO_printf( Perl_debug_log, " base: %d\n",base);
1600 trie->states[ state ].trans.base=base;
1602 trie->lasttrans = tp + 1;
1606 Second Pass -- Flat Table Representation.
1608 we dont use the 0 slot of either trans[] or states[] so we add 1 to each.
1609 We know that we will need Charcount+1 trans at most to store the data
1610 (one row per char at worst case) So we preallocate both structures
1611 assuming worst case.
1613 We then construct the trie using only the .next slots of the entry
1616 We use the .check field of the first entry of the node temporarily to
1617 make compression both faster and easier by keeping track of how many non
1618 zero fields are in the node.
1620 Since trans are numbered from 1 any 0 pointer in the table is a FAIL
1623 There are two terms at use here: state as a TRIE_NODEIDX() which is a
1624 number representing the first entry of the node, and state as a
1625 TRIE_NODENUM() which is the trans number. state 1 is TRIE_NODEIDX(1) and
1626 TRIE_NODENUM(1), state 2 is TRIE_NODEIDX(2) and TRIE_NODENUM(3) if there
1627 are 2 entrys per node. eg:
1635 The table is internally in the right hand, idx form. However as we also
1636 have to deal with the states array which is indexed by nodenum we have to
1637 use TRIE_NODENUM() to convert.
1640 DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log,
1641 "%*sCompiling trie using table compiler\n",
1642 (int)depth * 2 + 2, ""));
1644 trie->trans = (reg_trie_trans *)
1645 PerlMemShared_calloc( ( TRIE_CHARCOUNT(trie) + 1 )
1646 * trie->uniquecharcount + 1,
1647 sizeof(reg_trie_trans) );
1648 trie->states = (reg_trie_state *)
1649 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
1650 sizeof(reg_trie_state) );
1651 next_alloc = trie->uniquecharcount + 1;
1654 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1656 regnode * const noper = NEXTOPER( cur );
1657 const U8 *uc = (U8*)STRING( noper );
1658 const U8 * const e = uc + STR_LEN( noper );
1660 U32 state = 1; /* required init */
1662 U16 charid = 0; /* sanity init */
1663 U32 accept_state = 0; /* sanity init */
1664 U8 *scan = (U8*)NULL; /* sanity init */
1666 STRLEN foldlen = 0; /* required init */
1667 U32 wordlen = 0; /* required init */
1668 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1670 if ( OP(noper) != NOTHING ) {
1671 for ( ; uc < e ; uc += len ) {
1676 charid = trie->charmap[ uvc ];
1678 SV* const * const svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 0);
1679 charid = svpp ? (U16)SvIV(*svpp) : 0;
1683 if ( !trie->trans[ state + charid ].next ) {
1684 trie->trans[ state + charid ].next = next_alloc;
1685 trie->trans[ state ].check++;
1686 next_alloc += trie->uniquecharcount;
1688 state = trie->trans[ state + charid ].next;
1690 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %"IVdf, uvc );
1692 /* charid is now 0 if we dont know the char read, or nonzero if we do */
1695 accept_state = TRIE_NODENUM( state );
1696 TRIE_HANDLE_WORD(accept_state);
1698 } /* end second pass */
1700 /* and now dump it out before we compress it */
1701 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_table(trie, widecharmap,
1703 next_alloc, depth+1));
1707 * Inplace compress the table.*
1709 For sparse data sets the table constructed by the trie algorithm will
1710 be mostly 0/FAIL transitions or to put it another way mostly empty.
1711 (Note that leaf nodes will not contain any transitions.)
1713 This algorithm compresses the tables by eliminating most such
1714 transitions, at the cost of a modest bit of extra work during lookup:
1716 - Each states[] entry contains a .base field which indicates the
1717 index in the state[] array wheres its transition data is stored.
1719 - If .base is 0 there are no valid transitions from that node.
1721 - If .base is nonzero then charid is added to it to find an entry in
1724 -If trans[states[state].base+charid].check!=state then the
1725 transition is taken to be a 0/Fail transition. Thus if there are fail
1726 transitions at the front of the node then the .base offset will point
1727 somewhere inside the previous nodes data (or maybe even into a node
1728 even earlier), but the .check field determines if the transition is
1732 The following process inplace converts the table to the compressed
1733 table: We first do not compress the root node 1,and mark its all its
1734 .check pointers as 1 and set its .base pointer as 1 as well. This
1735 allows to do a DFA construction from the compressed table later, and
1736 ensures that any .base pointers we calculate later are greater than
1739 - We set 'pos' to indicate the first entry of the second node.
1741 - We then iterate over the columns of the node, finding the first and
1742 last used entry at l and m. We then copy l..m into pos..(pos+m-l),
1743 and set the .check pointers accordingly, and advance pos
1744 appropriately and repreat for the next node. Note that when we copy
1745 the next pointers we have to convert them from the original
1746 NODEIDX form to NODENUM form as the former is not valid post
1749 - If a node has no transitions used we mark its base as 0 and do not
1750 advance the pos pointer.
1752 - If a node only has one transition we use a second pointer into the
1753 structure to fill in allocated fail transitions from other states.
1754 This pointer is independent of the main pointer and scans forward
1755 looking for null transitions that are allocated to a state. When it
1756 finds one it writes the single transition into the "hole". If the
1757 pointer doesnt find one the single transition is appended as normal.
1759 - Once compressed we can Renew/realloc the structures to release the
1762 See "Table-Compression Methods" in sec 3.9 of the Red Dragon,
1763 specifically Fig 3.47 and the associated pseudocode.
1767 const U32 laststate = TRIE_NODENUM( next_alloc );
1770 trie->statecount = laststate;
1772 for ( state = 1 ; state < laststate ; state++ ) {
1774 const U32 stateidx = TRIE_NODEIDX( state );
1775 const U32 o_used = trie->trans[ stateidx ].check;
1776 U32 used = trie->trans[ stateidx ].check;
1777 trie->trans[ stateidx ].check = 0;
1779 for ( charid = 0 ; used && charid < trie->uniquecharcount ; charid++ ) {
1780 if ( flag || trie->trans[ stateidx + charid ].next ) {
1781 if ( trie->trans[ stateidx + charid ].next ) {
1783 for ( ; zp < pos ; zp++ ) {
1784 if ( ! trie->trans[ zp ].next ) {
1788 trie->states[ state ].trans.base = zp + trie->uniquecharcount - charid ;
1789 trie->trans[ zp ].next = SAFE_TRIE_NODENUM( trie->trans[ stateidx + charid ].next );
1790 trie->trans[ zp ].check = state;
1791 if ( ++zp > pos ) pos = zp;
1798 trie->states[ state ].trans.base = pos + trie->uniquecharcount - charid ;
1800 trie->trans[ pos ].next = SAFE_TRIE_NODENUM( trie->trans[ stateidx + charid ].next );
1801 trie->trans[ pos ].check = state;
1806 trie->lasttrans = pos + 1;
1807 trie->states = (reg_trie_state *)
1808 PerlMemShared_realloc( trie->states, laststate
1809 * sizeof(reg_trie_state) );
1810 DEBUG_TRIE_COMPILE_MORE_r(
1811 PerlIO_printf( Perl_debug_log,
1812 "%*sAlloc: %d Orig: %"IVdf" elements, Final:%"IVdf". Savings of %%%5.2f\n",
1813 (int)depth * 2 + 2,"",
1814 (int)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1 ),
1817 ( ( next_alloc - pos ) * 100 ) / (double)next_alloc );
1820 } /* end table compress */
1822 DEBUG_TRIE_COMPILE_MORE_r(
1823 PerlIO_printf(Perl_debug_log, "%*sStatecount:%"UVxf" Lasttrans:%"UVxf"\n",
1824 (int)depth * 2 + 2, "",
1825 (UV)trie->statecount,
1826 (UV)trie->lasttrans)
1828 /* resize the trans array to remove unused space */
1829 trie->trans = (reg_trie_trans *)
1830 PerlMemShared_realloc( trie->trans, trie->lasttrans
1831 * sizeof(reg_trie_trans) );
1833 /* and now dump out the compressed format */
1834 DEBUG_TRIE_COMPILE_r(dump_trie(trie, widecharmap, revcharmap, depth+1));
1836 { /* Modify the program and insert the new TRIE node*/
1837 U8 nodetype =(U8)(flags & 0xFF);
1841 regnode *optimize = NULL;
1843 U32 mjd_nodelen = 0;
1846 This means we convert either the first branch or the first Exact,
1847 depending on whether the thing following (in 'last') is a branch
1848 or not and whther first is the startbranch (ie is it a sub part of
1849 the alternation or is it the whole thing.)
1850 Assuming its a sub part we conver the EXACT otherwise we convert
1851 the whole branch sequence, including the first.
1853 /* Find the node we are going to overwrite */
1854 if ( first != startbranch || OP( last ) == BRANCH ) {
1855 /* branch sub-chain */
1856 NEXT_OFF( first ) = (U16)(last - first);
1858 mjd_offset= Node_Offset((convert));
1859 mjd_nodelen= Node_Length((convert));
1861 /* whole branch chain */
1864 const regnode *nop = NEXTOPER( convert );
1865 mjd_offset= Node_Offset((nop));
1866 mjd_nodelen= Node_Length((nop));
1871 PerlIO_printf(Perl_debug_log, "%*sMJD offset:%"UVuf" MJD length:%"UVuf"\n",
1872 (int)depth * 2 + 2, "",
1873 (UV)mjd_offset, (UV)mjd_nodelen)
1876 /* But first we check to see if there is a common prefix we can
1877 split out as an EXACT and put in front of the TRIE node. */
1878 trie->startstate= 1;
1879 if ( trie->bitmap && !widecharmap && !trie->jump ) {
1881 for ( state = 1 ; state < trie->statecount-1 ; state++ ) {
1885 const U32 base = trie->states[ state ].trans.base;
1887 if ( trie->states[state].wordnum )
1890 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
1891 if ( ( base + ofs >= trie->uniquecharcount ) &&
1892 ( base + ofs - trie->uniquecharcount < trie->lasttrans ) &&
1893 trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
1895 if ( ++count > 1 ) {
1896 SV **tmp = av_fetch( revcharmap, ofs, 0);
1897 const U8 *ch = (U8*)SvPV_nolen_const( *tmp );
1898 if ( state == 1 ) break;
1900 Zero(trie->bitmap, ANYOF_BITMAP_SIZE, char);
1902 PerlIO_printf(Perl_debug_log,
1903 "%*sNew Start State=%"UVuf" Class: [",
1904 (int)depth * 2 + 2, "",
1907 SV ** const tmp = av_fetch( revcharmap, idx, 0);
1908 const U8 * const ch = (U8*)SvPV_nolen_const( *tmp );
1910 TRIE_BITMAP_SET(trie,*ch);
1912 TRIE_BITMAP_SET(trie, folder[ *ch ]);
1914 PerlIO_printf(Perl_debug_log, (char*)ch)
1918 TRIE_BITMAP_SET(trie,*ch);
1920 TRIE_BITMAP_SET(trie,folder[ *ch ]);
1921 DEBUG_OPTIMISE_r(PerlIO_printf( Perl_debug_log,"%s", ch));
1927 SV **tmp = av_fetch( revcharmap, idx, 0);
1928 char *ch = SvPV_nolen( *tmp );
1930 SV *sv=sv_newmortal();
1931 PerlIO_printf( Perl_debug_log,
1932 "%*sPrefix State: %"UVuf" Idx:%"UVuf" Char='%s'\n",
1933 (int)depth * 2 + 2, "",
1935 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 6,
1936 PL_colors[0], PL_colors[1],
1937 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
1938 PERL_PV_ESCAPE_FIRSTCHAR
1943 OP( convert ) = nodetype;
1944 str=STRING(convert);
1955 DEBUG_OPTIMISE_r(PerlIO_printf( Perl_debug_log,"]\n"));
1961 regnode *n = convert+NODE_SZ_STR(convert);
1962 NEXT_OFF(convert) = NODE_SZ_STR(convert);
1963 trie->startstate = state;
1964 trie->minlen -= (state - 1);
1965 trie->maxlen -= (state - 1);
1967 regnode *fix = convert;
1968 U32 word = trie->wordcount;
1970 Set_Node_Offset_Length(convert, mjd_offset, state - 1);
1971 while( ++fix < n ) {
1972 Set_Node_Offset_Length(fix, 0, 0);
1975 SV ** const tmp = av_fetch( trie_words, word, 0 );
1977 if ( STR_LEN(convert) <= SvCUR(*tmp) )
1978 sv_chop(*tmp, SvPV_nolen(*tmp) + STR_LEN(convert));
1980 sv_chop(*tmp, SvPV_nolen(*tmp) + SvCUR(*tmp));
1987 NEXT_OFF(convert) = (U16)(tail - convert);
1988 DEBUG_r(optimize= n);
1994 if ( trie->maxlen ) {
1995 NEXT_OFF( convert ) = (U16)(tail - convert);
1996 ARG_SET( convert, data_slot );
1997 /* Store the offset to the first unabsorbed branch in
1998 jump[0], which is otherwise unused by the jump logic.
1999 We use this when dumping a trie and during optimisation. */
2001 trie->jump[0] = (U16)(nextbranch - convert);
2004 if ( !trie->states[trie->startstate].wordnum && trie->bitmap &&
2005 ( (char *)jumper - (char *)convert) >= (int)sizeof(struct regnode_charclass) )
2007 OP( convert ) = TRIEC;
2008 Copy(trie->bitmap, ((struct regnode_charclass *)convert)->bitmap, ANYOF_BITMAP_SIZE, char);
2009 PerlMemShared_free(trie->bitmap);
2012 OP( convert ) = TRIE;
2014 /* store the type in the flags */
2015 convert->flags = nodetype;
2019 + regarglen[ OP( convert ) ];
2021 /* XXX We really should free up the resource in trie now,
2022 as we won't use them - (which resources?) dmq */
2024 /* needed for dumping*/
2025 DEBUG_r(if (optimize) {
2026 regnode *opt = convert;
2027 while ( ++opt < optimize) {
2028 Set_Node_Offset_Length(opt,0,0);
2031 Try to clean up some of the debris left after the
2034 while( optimize < jumper ) {
2035 mjd_nodelen += Node_Length((optimize));
2036 OP( optimize ) = OPTIMIZED;
2037 Set_Node_Offset_Length(optimize,0,0);
2040 Set_Node_Offset_Length(convert,mjd_offset,mjd_nodelen);
2042 } /* end node insert */
2043 RExC_rxi->data->data[ data_slot + 1 ] = (void*)widecharmap;
2045 RExC_rxi->data->data[ data_slot + TRIE_WORDS_OFFSET ] = (void*)trie_words;
2046 RExC_rxi->data->data[ data_slot + 3 ] = (void*)revcharmap;
2048 SvREFCNT_dec(revcharmap);
2052 : trie->startstate>1
2058 S_make_trie_failtable(pTHX_ RExC_state_t *pRExC_state, regnode *source, regnode *stclass, U32 depth)
2060 /* The Trie is constructed and compressed now so we can build a fail array now if its needed
2062 This is basically the Aho-Corasick algorithm. Its from exercise 3.31 and 3.32 in the
2063 "Red Dragon" -- Compilers, principles, techniques, and tools. Aho, Sethi, Ullman 1985/88
2066 We find the fail state for each state in the trie, this state is the longest proper
2067 suffix of the current states 'word' that is also a proper prefix of another word in our
2068 trie. State 1 represents the word '' and is the thus the default fail state. This allows
2069 the DFA not to have to restart after its tried and failed a word at a given point, it
2070 simply continues as though it had been matching the other word in the first place.
2072 'abcdgu'=~/abcdefg|cdgu/
2073 When we get to 'd' we are still matching the first word, we would encounter 'g' which would
2074 fail, which would bring use to the state representing 'd' in the second word where we would
2075 try 'g' and succeed, prodceding to match 'cdgu'.
2077 /* add a fail transition */
2078 const U32 trie_offset = ARG(source);
2079 reg_trie_data *trie=(reg_trie_data *)RExC_rxi->data->data[trie_offset];
2081 const U32 ucharcount = trie->uniquecharcount;
2082 const U32 numstates = trie->statecount;
2083 const U32 ubound = trie->lasttrans + ucharcount;
2087 U32 base = trie->states[ 1 ].trans.base;
2090 const U32 data_slot = add_data( pRExC_state, 1, "T" );
2091 GET_RE_DEBUG_FLAGS_DECL;
2093 PERL_UNUSED_ARG(depth);
2097 ARG_SET( stclass, data_slot );
2098 aho = (reg_ac_data *) PerlMemShared_calloc( 1, sizeof(reg_ac_data) );
2099 RExC_rxi->data->data[ data_slot ] = (void*)aho;
2100 aho->trie=trie_offset;
2101 aho->states=(reg_trie_state *)PerlMemShared_malloc( numstates * sizeof(reg_trie_state) );
2102 Copy( trie->states, aho->states, numstates, reg_trie_state );
2103 Newxz( q, numstates, U32);
2104 aho->fail = (U32 *) PerlMemShared_calloc( numstates, sizeof(U32) );
2107 /* initialize fail[0..1] to be 1 so that we always have
2108 a valid final fail state */
2109 fail[ 0 ] = fail[ 1 ] = 1;
2111 for ( charid = 0; charid < ucharcount ; charid++ ) {
2112 const U32 newstate = TRIE_TRANS_STATE( 1, base, ucharcount, charid, 0 );
2114 q[ q_write ] = newstate;
2115 /* set to point at the root */
2116 fail[ q[ q_write++ ] ]=1;
2119 while ( q_read < q_write) {
2120 const U32 cur = q[ q_read++ % numstates ];
2121 base = trie->states[ cur ].trans.base;
2123 for ( charid = 0 ; charid < ucharcount ; charid++ ) {
2124 const U32 ch_state = TRIE_TRANS_STATE( cur, base, ucharcount, charid, 1 );
2126 U32 fail_state = cur;
2129 fail_state = fail[ fail_state ];
2130 fail_base = aho->states[ fail_state ].trans.base;
2131 } while ( !TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 ) );
2133 fail_state = TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 );
2134 fail[ ch_state ] = fail_state;
2135 if ( !aho->states[ ch_state ].wordnum && aho->states[ fail_state ].wordnum )
2137 aho->states[ ch_state ].wordnum = aho->states[ fail_state ].wordnum;
2139 q[ q_write++ % numstates] = ch_state;
2143 /* restore fail[0..1] to 0 so that we "fall out" of the AC loop
2144 when we fail in state 1, this allows us to use the
2145 charclass scan to find a valid start char. This is based on the principle
2146 that theres a good chance the string being searched contains lots of stuff
2147 that cant be a start char.
2149 fail[ 0 ] = fail[ 1 ] = 0;
2150 DEBUG_TRIE_COMPILE_r({
2151 PerlIO_printf(Perl_debug_log,
2152 "%*sStclass Failtable (%"UVuf" states): 0",
2153 (int)(depth * 2), "", (UV)numstates
2155 for( q_read=1; q_read<numstates; q_read++ ) {
2156 PerlIO_printf(Perl_debug_log, ", %"UVuf, (UV)fail[q_read]);
2158 PerlIO_printf(Perl_debug_log, "\n");
2161 /*RExC_seen |= REG_SEEN_TRIEDFA;*/
2166 * There are strange code-generation bugs caused on sparc64 by gcc-2.95.2.
2167 * These need to be revisited when a newer toolchain becomes available.
2169 #if defined(__sparc64__) && defined(__GNUC__)
2170 # if __GNUC__ < 2 || (__GNUC__ == 2 && __GNUC_MINOR__ < 96)
2171 # undef SPARC64_GCC_WORKAROUND
2172 # define SPARC64_GCC_WORKAROUND 1
2176 #define DEBUG_PEEP(str,scan,depth) \
2177 DEBUG_OPTIMISE_r({if (scan){ \
2178 SV * const mysv=sv_newmortal(); \
2179 regnode *Next = regnext(scan); \
2180 regprop(RExC_rx, mysv, scan); \
2181 PerlIO_printf(Perl_debug_log, "%*s" str ">%3d: %s (%d)\n", \
2182 (int)depth*2, "", REG_NODE_NUM(scan), SvPV_nolen_const(mysv),\
2183 Next ? (REG_NODE_NUM(Next)) : 0 ); \
2190 #define JOIN_EXACT(scan,min,flags) \
2191 if (PL_regkind[OP(scan)] == EXACT) \
2192 join_exact(pRExC_state,(scan),(min),(flags),NULL,depth+1)
2195 S_join_exact(pTHX_ RExC_state_t *pRExC_state, regnode *scan, I32 *min, U32 flags,regnode *val, U32 depth) {
2196 /* Merge several consecutive EXACTish nodes into one. */
2197 regnode *n = regnext(scan);
2199 regnode *next = scan + NODE_SZ_STR(scan);
2203 regnode *stop = scan;
2204 GET_RE_DEBUG_FLAGS_DECL;
2206 PERL_UNUSED_ARG(depth);
2208 #ifndef EXPERIMENTAL_INPLACESCAN
2209 PERL_UNUSED_ARG(flags);
2210 PERL_UNUSED_ARG(val);
2212 DEBUG_PEEP("join",scan,depth);
2214 /* Skip NOTHING, merge EXACT*. */
2216 ( PL_regkind[OP(n)] == NOTHING ||
2217 (stringok && (OP(n) == OP(scan))))
2219 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
2221 if (OP(n) == TAIL || n > next)
2223 if (PL_regkind[OP(n)] == NOTHING) {
2224 DEBUG_PEEP("skip:",n,depth);
2225 NEXT_OFF(scan) += NEXT_OFF(n);
2226 next = n + NODE_STEP_REGNODE;
2233 else if (stringok) {
2234 const unsigned int oldl = STR_LEN(scan);
2235 regnode * const nnext = regnext(n);
2237 DEBUG_PEEP("merg",n,depth);
2240 if (oldl + STR_LEN(n) > U8_MAX)
2242 NEXT_OFF(scan) += NEXT_OFF(n);
2243 STR_LEN(scan) += STR_LEN(n);
2244 next = n + NODE_SZ_STR(n);
2245 /* Now we can overwrite *n : */
2246 Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
2254 #ifdef EXPERIMENTAL_INPLACESCAN
2255 if (flags && !NEXT_OFF(n)) {
2256 DEBUG_PEEP("atch", val, depth);
2257 if (reg_off_by_arg[OP(n)]) {
2258 ARG_SET(n, val - n);
2261 NEXT_OFF(n) = val - n;
2268 if (UTF && ( OP(scan) == EXACTF ) && ( STR_LEN(scan) >= 6 ) ) {
2270 Two problematic code points in Unicode casefolding of EXACT nodes:
2272 U+0390 - GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
2273 U+03B0 - GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
2279 U+03B9 U+0308 U+0301 0xCE 0xB9 0xCC 0x88 0xCC 0x81
2280 U+03C5 U+0308 U+0301 0xCF 0x85 0xCC 0x88 0xCC 0x81
2282 This means that in case-insensitive matching (or "loose matching",
2283 as Unicode calls it), an EXACTF of length six (the UTF-8 encoded byte
2284 length of the above casefolded versions) can match a target string
2285 of length two (the byte length of UTF-8 encoded U+0390 or U+03B0).
2286 This would rather mess up the minimum length computation.
2288 What we'll do is to look for the tail four bytes, and then peek
2289 at the preceding two bytes to see whether we need to decrease
2290 the minimum length by four (six minus two).
2292 Thanks to the design of UTF-8, there cannot be false matches:
2293 A sequence of valid UTF-8 bytes cannot be a subsequence of
2294 another valid sequence of UTF-8 bytes.
2297 char * const s0 = STRING(scan), *s, *t;
2298 char * const s1 = s0 + STR_LEN(scan) - 1;
2299 char * const s2 = s1 - 4;
2300 #ifdef EBCDIC /* RD tunifold greek 0390 and 03B0 */
2301 const char t0[] = "\xaf\x49\xaf\x42";
2303 const char t0[] = "\xcc\x88\xcc\x81";
2305 const char * const t1 = t0 + 3;
2308 s < s2 && (t = ninstr(s, s1, t0, t1));
2311 if (((U8)t[-1] == 0x68 && (U8)t[-2] == 0xB4) ||
2312 ((U8)t[-1] == 0x46 && (U8)t[-2] == 0xB5))
2314 if (((U8)t[-1] == 0xB9 && (U8)t[-2] == 0xCE) ||
2315 ((U8)t[-1] == 0x85 && (U8)t[-2] == 0xCF))
2323 n = scan + NODE_SZ_STR(scan);
2325 if (PL_regkind[OP(n)] != NOTHING || OP(n) == NOTHING) {
2332 DEBUG_OPTIMISE_r(if (merged){DEBUG_PEEP("finl",scan,depth)});
2336 /* REx optimizer. Converts nodes into quickier variants "in place".
2337 Finds fixed substrings. */
2339 /* Stops at toplevel WHILEM as well as at "last". At end *scanp is set
2340 to the position after last scanned or to NULL. */
2342 #define INIT_AND_WITHP \
2343 assert(!and_withp); \
2344 Newx(and_withp,1,struct regnode_charclass_class); \
2345 SAVEFREEPV(and_withp)
2347 /* this is a chain of data about sub patterns we are processing that
2348 need to be handled seperately/specially in study_chunk. Its so
2349 we can simulate recursion without losing state. */
2351 typedef struct scan_frame {
2352 regnode *last; /* last node to process in this frame */
2353 regnode *next; /* next node to process when last is reached */
2354 struct scan_frame *prev; /*previous frame*/
2355 I32 stop; /* what stopparen do we use */
2359 #define SCAN_COMMIT(s, data, m) scan_commit(s, data, m, is_inf)
2362 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
2363 I32 *minlenp, I32 *deltap,
2368 struct regnode_charclass_class *and_withp,
2369 U32 flags, U32 depth)
2370 /* scanp: Start here (read-write). */
2371 /* deltap: Write maxlen-minlen here. */
2372 /* last: Stop before this one. */
2373 /* data: string data about the pattern */
2374 /* stopparen: treat close N as END */
2375 /* recursed: which subroutines have we recursed into */
2376 /* and_withp: Valid if flags & SCF_DO_STCLASS_OR */
2379 I32 min = 0, pars = 0, code;
2380 regnode *scan = *scanp, *next;
2382 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
2383 int is_inf_internal = 0; /* The studied chunk is infinite */
2384 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
2385 scan_data_t data_fake;
2386 SV *re_trie_maxbuff = NULL;
2387 regnode *first_non_open = scan;
2388 I32 stopmin = I32_MAX;
2389 scan_frame *frame = NULL;
2391 GET_RE_DEBUG_FLAGS_DECL;
2394 StructCopy(&zero_scan_data, &data_fake, scan_data_t);
2398 while (first_non_open && OP(first_non_open) == OPEN)
2399 first_non_open=regnext(first_non_open);
2404 while ( scan && OP(scan) != END && scan < last ){
2405 /* Peephole optimizer: */
2406 DEBUG_STUDYDATA("Peep:", data,depth);
2407 DEBUG_PEEP("Peep",scan,depth);
2408 JOIN_EXACT(scan,&min,0);
2410 /* Follow the next-chain of the current node and optimize
2411 away all the NOTHINGs from it. */
2412 if (OP(scan) != CURLYX) {
2413 const int max = (reg_off_by_arg[OP(scan)]
2415 /* I32 may be smaller than U16 on CRAYs! */
2416 : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
2417 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
2421 /* Skip NOTHING and LONGJMP. */
2422 while ((n = regnext(n))
2423 && ((PL_regkind[OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
2424 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
2425 && off + noff < max)
2427 if (reg_off_by_arg[OP(scan)])
2430 NEXT_OFF(scan) = off;
2435 /* The principal pseudo-switch. Cannot be a switch, since we
2436 look into several different things. */
2437 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
2438 || OP(scan) == IFTHEN) {
2439 next = regnext(scan);
2441 /* demq: the op(next)==code check is to see if we have "branch-branch" AFAICT */
2443 if (OP(next) == code || code == IFTHEN) {
2444 /* NOTE - There is similar code to this block below for handling
2445 TRIE nodes on a re-study. If you change stuff here check there
2447 I32 max1 = 0, min1 = I32_MAX, num = 0;
2448 struct regnode_charclass_class accum;
2449 regnode * const startbranch=scan;
2451 if (flags & SCF_DO_SUBSTR)
2452 SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot merge strings after this. */
2453 if (flags & SCF_DO_STCLASS)
2454 cl_init_zero(pRExC_state, &accum);
2456 while (OP(scan) == code) {
2457 I32 deltanext, minnext, f = 0, fake;
2458 struct regnode_charclass_class this_class;
2461 data_fake.flags = 0;
2463 data_fake.whilem_c = data->whilem_c;
2464 data_fake.last_closep = data->last_closep;
2467 data_fake.last_closep = &fake;
2469 data_fake.pos_delta = delta;
2470 next = regnext(scan);
2471 scan = NEXTOPER(scan);
2473 scan = NEXTOPER(scan);
2474 if (flags & SCF_DO_STCLASS) {
2475 cl_init(pRExC_state, &this_class);
2476 data_fake.start_class = &this_class;
2477 f = SCF_DO_STCLASS_AND;
2479 if (flags & SCF_WHILEM_VISITED_POS)
2480 f |= SCF_WHILEM_VISITED_POS;
2482 /* we suppose the run is continuous, last=next...*/
2483 minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext,
2485 stopparen, recursed, NULL, f,depth+1);
2488 if (max1 < minnext + deltanext)
2489 max1 = minnext + deltanext;
2490 if (deltanext == I32_MAX)
2491 is_inf = is_inf_internal = 1;
2493 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
2495 if (data_fake.flags & SCF_SEEN_ACCEPT) {
2496 if ( stopmin > minnext)
2497 stopmin = min + min1;
2498 flags &= ~SCF_DO_SUBSTR;
2500 data->flags |= SCF_SEEN_ACCEPT;
2503 if (data_fake.flags & SF_HAS_EVAL)
2504 data->flags |= SF_HAS_EVAL;
2505 data->whilem_c = data_fake.whilem_c;
2507 if (flags & SCF_DO_STCLASS)
2508 cl_or(pRExC_state, &accum, &this_class);
2510 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
2512 if (flags & SCF_DO_SUBSTR) {
2513 data->pos_min += min1;
2514 data->pos_delta += max1 - min1;
2515 if (max1 != min1 || is_inf)
2516 data->longest = &(data->longest_float);
2519 delta += max1 - min1;
2520 if (flags & SCF_DO_STCLASS_OR) {
2521 cl_or(pRExC_state, data->start_class, &accum);
2523 cl_and(data->start_class, and_withp);
2524 flags &= ~SCF_DO_STCLASS;
2527 else if (flags & SCF_DO_STCLASS_AND) {
2529 cl_and(data->start_class, &accum);
2530 flags &= ~SCF_DO_STCLASS;
2533 /* Switch to OR mode: cache the old value of
2534 * data->start_class */
2536 StructCopy(data->start_class, and_withp,
2537 struct regnode_charclass_class);
2538 flags &= ~SCF_DO_STCLASS_AND;
2539 StructCopy(&accum, data->start_class,
2540 struct regnode_charclass_class);
2541 flags |= SCF_DO_STCLASS_OR;
2542 data->start_class->flags |= ANYOF_EOS;
2546 if (PERL_ENABLE_TRIE_OPTIMISATION && OP( startbranch ) == BRANCH ) {
2549 Assuming this was/is a branch we are dealing with: 'scan' now
2550 points at the item that follows the branch sequence, whatever
2551 it is. We now start at the beginning of the sequence and look
2558 which would be constructed from a pattern like /A|LIST|OF|WORDS/
2560 If we can find such a subseqence we need to turn the first
2561 element into a trie and then add the subsequent branch exact
2562 strings to the trie.
2566 1. patterns where the whole set of branch can be converted.
2568 2. patterns where only a subset can be converted.
2570 In case 1 we can replace the whole set with a single regop
2571 for the trie. In case 2 we need to keep the start and end
2574 'BRANCH EXACT; BRANCH EXACT; BRANCH X'
2575 becomes BRANCH TRIE; BRANCH X;
2577 There is an additional case, that being where there is a
2578 common prefix, which gets split out into an EXACT like node
2579 preceding the TRIE node.
2581 If x(1..n)==tail then we can do a simple trie, if not we make
2582 a "jump" trie, such that when we match the appropriate word
2583 we "jump" to the appopriate tail node. Essentailly we turn
2584 a nested if into a case structure of sorts.
2589 if (!re_trie_maxbuff) {
2590 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
2591 if (!SvIOK(re_trie_maxbuff))
2592 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
2594 if ( SvIV(re_trie_maxbuff)>=0 ) {
2596 regnode *first = (regnode *)NULL;
2597 regnode *last = (regnode *)NULL;
2598 regnode *tail = scan;
2603 SV * const mysv = sv_newmortal(); /* for dumping */
2605 /* var tail is used because there may be a TAIL
2606 regop in the way. Ie, the exacts will point to the
2607 thing following the TAIL, but the last branch will
2608 point at the TAIL. So we advance tail. If we
2609 have nested (?:) we may have to move through several
2613 while ( OP( tail ) == TAIL ) {
2614 /* this is the TAIL generated by (?:) */
2615 tail = regnext( tail );
2620 regprop(RExC_rx, mysv, tail );
2621 PerlIO_printf( Perl_debug_log, "%*s%s%s\n",
2622 (int)depth * 2 + 2, "",
2623 "Looking for TRIE'able sequences. Tail node is: ",
2624 SvPV_nolen_const( mysv )
2630 step through the branches, cur represents each
2631 branch, noper is the first thing to be matched
2632 as part of that branch and noper_next is the
2633 regnext() of that node. if noper is an EXACT
2634 and noper_next is the same as scan (our current
2635 position in the regex) then the EXACT branch is
2636 a possible optimization target. Once we have
2637 two or more consequetive such branches we can
2638 create a trie of the EXACT's contents and stich
2639 it in place. If the sequence represents all of
2640 the branches we eliminate the whole thing and
2641 replace it with a single TRIE. If it is a
2642 subsequence then we need to stitch it in. This
2643 means the first branch has to remain, and needs
2644 to be repointed at the item on the branch chain
2645 following the last branch optimized. This could
2646 be either a BRANCH, in which case the
2647 subsequence is internal, or it could be the
2648 item following the branch sequence in which
2649 case the subsequence is at the end.
2653 /* dont use tail as the end marker for this traverse */
2654 for ( cur = startbranch ; cur != scan ; cur = regnext( cur ) ) {
2655 regnode * const noper = NEXTOPER( cur );
2656 #if defined(DEBUGGING) || defined(NOJUMPTRIE)
2657 regnode * const noper_next = regnext( noper );
2661 regprop(RExC_rx, mysv, cur);
2662 PerlIO_printf( Perl_debug_log, "%*s- %s (%d)",
2663 (int)depth * 2 + 2,"", SvPV_nolen_const( mysv ), REG_NODE_NUM(cur) );
2665 regprop(RExC_rx, mysv, noper);
2666 PerlIO_printf( Perl_debug_log, " -> %s",
2667 SvPV_nolen_const(mysv));
2670 regprop(RExC_rx, mysv, noper_next );
2671 PerlIO_printf( Perl_debug_log,"\t=> %s\t",
2672 SvPV_nolen_const(mysv));
2674 PerlIO_printf( Perl_debug_log, "(First==%d,Last==%d,Cur==%d)\n",
2675 REG_NODE_NUM(first), REG_NODE_NUM(last), REG_NODE_NUM(cur) );
2677 if ( (((first && optype!=NOTHING) ? OP( noper ) == optype
2678 : PL_regkind[ OP( noper ) ] == EXACT )
2679 || OP(noper) == NOTHING )
2681 && noper_next == tail
2686 if ( !first || optype == NOTHING ) {
2687 if (!first) first = cur;
2688 optype = OP( noper );
2694 make_trie( pRExC_state,
2695 startbranch, first, cur, tail, count,
2698 if ( PL_regkind[ OP( noper ) ] == EXACT
2700 && noper_next == tail
2705 optype = OP( noper );
2715 regprop(RExC_rx, mysv, cur);
2716 PerlIO_printf( Perl_debug_log,
2717 "%*s- %s (%d) <SCAN FINISHED>\n", (int)depth * 2 + 2,
2718 "", SvPV_nolen_const( mysv ),REG_NODE_NUM(cur));
2722 made= make_trie( pRExC_state, startbranch, first, scan, tail, count, optype, depth+1 );
2723 #ifdef TRIE_STUDY_OPT
2724 if ( ((made == MADE_EXACT_TRIE &&
2725 startbranch == first)
2726 || ( first_non_open == first )) &&
2728 flags |= SCF_TRIE_RESTUDY;
2729 if ( startbranch == first
2732 RExC_seen &=~REG_TOP_LEVEL_BRANCHES;
2742 else if ( code == BRANCHJ ) { /* single branch is optimized. */
2743 scan = NEXTOPER(NEXTOPER(scan));
2744 } else /* single branch is optimized. */
2745 scan = NEXTOPER(scan);
2747 } else if (OP(scan) == SUSPEND || OP(scan) == GOSUB || OP(scan) == GOSTART) {
2748 scan_frame *newframe = NULL;
2753 if (OP(scan) != SUSPEND) {
2754 /* set the pointer */
2755 if (OP(scan) == GOSUB) {
2757 RExC_recurse[ARG2L(scan)] = scan;
2758 start = RExC_open_parens[paren-1];
2759 end = RExC_close_parens[paren-1];
2762 start = RExC_rxi->program + 1;
2766 Newxz(recursed, (((RExC_npar)>>3) +1), U8);
2767 SAVEFREEPV(recursed);
2769 if (!PAREN_TEST(recursed,paren+1)) {
2770 PAREN_SET(recursed,paren+1);
2771 Newx(newframe,1,scan_frame);
2773 if (flags & SCF_DO_SUBSTR) {
2774 SCAN_COMMIT(pRExC_state,data,minlenp);
2775 data->longest = &(data->longest_float);
2777 is_inf = is_inf_internal = 1;
2778 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
2779 cl_anything(pRExC_state, data->start_class);
2780 flags &= ~SCF_DO_STCLASS;
2783 Newx(newframe,1,scan_frame);
2786 end = regnext(scan);
2791 SAVEFREEPV(newframe);
2792 newframe->next = regnext(scan);
2793 newframe->last = last;
2794 newframe->stop = stopparen;
2795 newframe->prev = frame;
2805 else if (OP(scan) == EXACT) {
2806 I32 l = STR_LEN(scan);
2809 const U8 * const s = (U8*)STRING(scan);
2810 l = utf8_length(s, s + l);
2811 uc = utf8_to_uvchr(s, NULL);
2813 uc = *((U8*)STRING(scan));
2816 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
2817 /* The code below prefers earlier match for fixed
2818 offset, later match for variable offset. */
2819 if (data->last_end == -1) { /* Update the start info. */
2820 data->last_start_min = data->pos_min;
2821 data->last_start_max = is_inf
2822 ? I32_MAX : data->pos_min + data->pos_delta;
2824 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
2826 SvUTF8_on(data->last_found);
2828 SV * const sv = data->last_found;
2829 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
2830 mg_find(sv, PERL_MAGIC_utf8) : NULL;
2831 if (mg && mg->mg_len >= 0)
2832 mg->mg_len += utf8_length((U8*)STRING(scan),
2833 (U8*)STRING(scan)+STR_LEN(scan));
2835 data->last_end = data->pos_min + l;
2836 data->pos_min += l; /* As in the first entry. */
2837 data->flags &= ~SF_BEFORE_EOL;
2839 if (flags & SCF_DO_STCLASS_AND) {
2840 /* Check whether it is compatible with what we know already! */
2844 (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
2845 && !ANYOF_BITMAP_TEST(data->start_class, uc)
2846 && (!(data->start_class->flags & ANYOF_FOLD)
2847 || !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
2850 ANYOF_CLASS_ZERO(data->start_class);
2851 ANYOF_BITMAP_ZERO(data->start_class);
2853 ANYOF_BITMAP_SET(data->start_class, uc);
2854 data->start_class->flags &= ~ANYOF_EOS;
2856 data->start_class->flags &= ~ANYOF_UNICODE_ALL;
2858 else if (flags & SCF_DO_STCLASS_OR) {
2859 /* false positive possible if the class is case-folded */
2861 ANYOF_BITMAP_SET(data->start_class, uc);
2863 data->start_class->flags |= ANYOF_UNICODE_ALL;
2864 data->start_class->flags &= ~ANYOF_EOS;
2865 cl_and(data->start_class, and_withp);
2867 flags &= ~SCF_DO_STCLASS;
2869 else if (PL_regkind[OP(scan)] == EXACT) { /* But OP != EXACT! */
2870 I32 l = STR_LEN(scan);
2871 UV uc = *((U8*)STRING(scan));
2873 /* Search for fixed substrings supports EXACT only. */
2874 if (flags & SCF_DO_SUBSTR) {
2876 SCAN_COMMIT(pRExC_state, data, minlenp);
2879 const U8 * const s = (U8 *)STRING(scan);
2880 l = utf8_length(s, s + l);
2881 uc = utf8_to_uvchr(s, NULL);
2884 if (flags & SCF_DO_SUBSTR)
2886 if (flags & SCF_DO_STCLASS_AND) {
2887 /* Check whether it is compatible with what we know already! */
2891 (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
2892 && !ANYOF_BITMAP_TEST(data->start_class, uc)
2893 && !ANYOF_BITMAP_TEST(data->start_class, PL_fold[uc])))
2895 ANYOF_CLASS_ZERO(data->start_class);
2896 ANYOF_BITMAP_ZERO(data->start_class);
2898 ANYOF_BITMAP_SET(data->start_class, uc);
2899 data->start_class->flags &= ~ANYOF_EOS;
2900 data->start_class->flags |= ANYOF_FOLD;
2901 if (OP(scan) == EXACTFL)
2902 data->start_class->flags |= ANYOF_LOCALE;
2905 else if (flags & SCF_DO_STCLASS_OR) {
2906 if (data->start_class->flags & ANYOF_FOLD) {
2907 /* false positive possible if the class is case-folded.
2908 Assume that the locale settings are the same... */
2910 ANYOF_BITMAP_SET(data->start_class, uc);
2911 data->start_class->flags &= ~ANYOF_EOS;
2913 cl_and(data->start_class, and_withp);
2915 flags &= ~SCF_DO_STCLASS;
2917 else if (strchr((const char*)PL_varies,OP(scan))) {
2918 I32 mincount, maxcount, minnext, deltanext, fl = 0;
2919 I32 f = flags, pos_before = 0;
2920 regnode * const oscan = scan;
2921 struct regnode_charclass_class this_class;
2922 struct regnode_charclass_class *oclass = NULL;
2923 I32 next_is_eval = 0;
2925 switch (PL_regkind[OP(scan)]) {
2926 case WHILEM: /* End of (?:...)* . */
2927 scan = NEXTOPER(scan);
2930 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
2931 next = NEXTOPER(scan);
2932 if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
2934 maxcount = REG_INFTY;
2935 next = regnext(scan);
2936 scan = NEXTOPER(scan);
2940 if (flags & SCF_DO_SUBSTR)
2945 if (flags & SCF_DO_STCLASS) {
2947 maxcount = REG_INFTY;
2948 next = regnext(scan);
2949 scan = NEXTOPER(scan);
2952 is_inf = is_inf_internal = 1;
2953 scan = regnext(scan);
2954 if (flags & SCF_DO_SUBSTR) {
2955 SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot extend fixed substrings */
2956 data->longest = &(data->longest_float);
2958 goto optimize_curly_tail;
2960 if (stopparen>0 && (OP(scan)==CURLYN || OP(scan)==CURLYM)
2961 && (scan->flags == stopparen))
2966 mincount = ARG1(scan);
2967 maxcount = ARG2(scan);
2969 next = regnext(scan);
2970 if (OP(scan) == CURLYX) {
2971 I32 lp = (data ? *(data->last_closep) : 0);
2972 scan->flags = ((lp <= (I32)U8_MAX) ? (U8)lp : U8_MAX);
2974 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
2975 next_is_eval = (OP(scan) == EVAL);
2977 if (flags & SCF_DO_SUBSTR) {
2978 if (mincount == 0) SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot extend fixed substrings */
2979 pos_before = data->pos_min;
2983 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
2985 data->flags |= SF_IS_INF;
2987 if (flags & SCF_DO_STCLASS) {
2988 cl_init(pRExC_state, &this_class);
2989 oclass = data->start_class;
2990 data->start_class = &this_class;
2991 f |= SCF_DO_STCLASS_AND;
2992 f &= ~SCF_DO_STCLASS_OR;
2994 /* These are the cases when once a subexpression
2995 fails at a particular position, it cannot succeed
2996 even after backtracking at the enclosing scope.
2998 XXXX what if minimal match and we are at the
2999 initial run of {n,m}? */
3000 if ((mincount != maxcount - 1) && (maxcount != REG_INFTY))
3001 f &= ~SCF_WHILEM_VISITED_POS;
3003 /* This will finish on WHILEM, setting scan, or on NULL: */
3004 minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext,
3005 last, data, stopparen, recursed, NULL,
3007 ? (f & ~SCF_DO_SUBSTR) : f),depth+1);
3009 if (flags & SCF_DO_STCLASS)
3010 data->start_class = oclass;
3011 if (mincount == 0 || minnext == 0) {
3012 if (flags & SCF_DO_STCLASS_OR) {
3013 cl_or(pRExC_state, data->start_class, &this_class);
3015 else if (flags & SCF_DO_STCLASS_AND) {
3016 /* Switch to OR mode: cache the old value of
3017 * data->start_class */
3019 StructCopy(data->start_class, and_withp,
3020 struct regnode_charclass_class);
3021 flags &= ~SCF_DO_STCLASS_AND;
3022 StructCopy(&this_class, data->start_class,
3023 struct regnode_charclass_class);
3024 flags |= SCF_DO_STCLASS_OR;
3025 data->start_class->flags |= ANYOF_EOS;
3027 } else { /* Non-zero len */
3028 if (flags & SCF_DO_STCLASS_OR) {
3029 cl_or(pRExC_state, data->start_class, &this_class);
3030 cl_and(data->start_class, and_withp);
3032 else if (flags & SCF_DO_STCLASS_AND)
3033 cl_and(data->start_class, &this_class);
3034 flags &= ~SCF_DO_STCLASS;
3036 if (!scan) /* It was not CURLYX, but CURLY. */
3038 if ( /* ? quantifier ok, except for (?{ ... }) */
3039 (next_is_eval || !(mincount == 0 && maxcount == 1))
3040 && (minnext == 0) && (deltanext == 0)
3041 && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
3042 && maxcount <= REG_INFTY/3 /* Complement check for big count */
3043 && ckWARN(WARN_REGEXP))
3046 "Quantifier unexpected on zero-length expression");
3049 min += minnext * mincount;
3050 is_inf_internal |= ((maxcount == REG_INFTY
3051 && (minnext + deltanext) > 0)
3052 || deltanext == I32_MAX);
3053 is_inf |= is_inf_internal;
3054 delta += (minnext + deltanext) * maxcount - minnext * mincount;
3056 /* Try powerful optimization CURLYX => CURLYN. */
3057 if ( OP(oscan) == CURLYX && data
3058 && data->flags & SF_IN_PAR
3059 && !(data->flags & SF_HAS_EVAL)
3060 && !deltanext && minnext == 1 ) {
3061 /* Try to optimize to CURLYN. */
3062 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
3063 regnode * const nxt1 = nxt;
3070 if (!strchr((const char*)PL_simple,OP(nxt))
3071 && !(PL_regkind[OP(nxt)] == EXACT
3072 && STR_LEN(nxt) == 1))
3078 if (OP(nxt) != CLOSE)
3080 if (RExC_open_parens) {
3081 RExC_open_parens[ARG(nxt1)-1]=oscan; /*open->CURLYM*/
3082 RExC_close_parens[ARG(nxt1)-1]=nxt+2; /*close->while*/
3084 /* Now we know that nxt2 is the only contents: */
3085 oscan->flags = (U8)ARG(nxt);
3087 OP(nxt1) = NOTHING; /* was OPEN. */
3090 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
3091 NEXT_OFF(nxt1+ 1) = 0; /* just for consistancy. */
3092 NEXT_OFF(nxt2) = 0; /* just for consistancy with CURLY. */
3093 OP(nxt) = OPTIMIZED; /* was CLOSE. */
3094 OP(nxt + 1) = OPTIMIZED; /* was count. */
3095 NEXT_OFF(nxt+ 1) = 0; /* just for consistancy. */
3100 /* Try optimization CURLYX => CURLYM. */
3101 if ( OP(oscan) == CURLYX && data
3102 && !(data->flags & SF_HAS_PAR)
3103 && !(data->flags & SF_HAS_EVAL)
3104 && !deltanext /* atom is fixed width */
3105 && minnext != 0 /* CURLYM can't handle zero width */
3107 /* XXXX How to optimize if data == 0? */
3108 /* Optimize to a simpler form. */
3109 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
3113 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
3114 && (OP(nxt2) != WHILEM))
3116 OP(nxt2) = SUCCEED; /* Whas WHILEM */
3117 /* Need to optimize away parenths. */
3118 if (data->flags & SF_IN_PAR) {
3119 /* Set the parenth number. */
3120 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
3122 if (OP(nxt) != CLOSE)
3123 FAIL("Panic opt close");
3124 oscan->flags = (U8)ARG(nxt);
3125 if (RExC_open_parens) {
3126 RExC_open_parens[ARG(nxt1)-1]=oscan; /*open->CURLYM*/
3127 RExC_close_parens[ARG(nxt1)-1]=nxt2+1; /*close->NOTHING*/
3129 OP(nxt1) = OPTIMIZED; /* was OPEN. */
3130 OP(nxt) = OPTIMIZED; /* was CLOSE. */
3133 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
3134 OP(nxt + 1) = OPTIMIZED; /* was count. */
3135 NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
3136 NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
3139 while ( nxt1 && (OP(nxt1) != WHILEM)) {
3140 regnode *nnxt = regnext(nxt1);
3143 if (reg_off_by_arg[OP(nxt1)])
3144 ARG_SET(nxt1, nxt2 - nxt1);
3145 else if (nxt2 - nxt1 < U16_MAX)
3146 NEXT_OFF(nxt1) = nxt2 - nxt1;
3148 OP(nxt) = NOTHING; /* Cannot beautify */
3153 /* Optimize again: */
3154 study_chunk(pRExC_state, &nxt1, minlenp, &deltanext, nxt,
3155 NULL, stopparen, recursed, NULL, 0,depth+1);
3160 else if ((OP(oscan) == CURLYX)
3161 && (flags & SCF_WHILEM_VISITED_POS)
3162 /* See the comment on a similar expression above.
3163 However, this time it not a subexpression
3164 we care about, but the expression itself. */
3165 && (maxcount == REG_INFTY)
3166 && data && ++data->whilem_c < 16) {
3167 /* This stays as CURLYX, we can put the count/of pair. */
3168 /* Find WHILEM (as in regexec.c) */
3169 regnode *nxt = oscan + NEXT_OFF(oscan);
3171 if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
3173 PREVOPER(nxt)->flags = (U8)(data->whilem_c
3174 | (RExC_whilem_seen << 4)); /* On WHILEM */
3176 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
3178 if (flags & SCF_DO_SUBSTR) {
3179 SV *last_str = NULL;
3180 int counted = mincount != 0;
3182 if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
3183 #if defined(SPARC64_GCC_WORKAROUND)
3186 const char *s = NULL;
3189 if (pos_before >= data->last_start_min)
3192 b = data->last_start_min;
3195 s = SvPV_const(data->last_found, l);
3196 old = b - data->last_start_min;
3199 I32 b = pos_before >= data->last_start_min
3200 ? pos_before : data->last_start_min;
3202 const char * const s = SvPV_const(data->last_found, l);
3203 I32 old = b - data->last_start_min;
3207 old = utf8_hop((U8*)s, old) - (U8*)s;
3210 /* Get the added string: */
3211 last_str = newSVpvn(s + old, l);
3213 SvUTF8_on(last_str);
3214 if (deltanext == 0 && pos_before == b) {
3215 /* What was added is a constant string */
3217 SvGROW(last_str, (mincount * l) + 1);
3218 repeatcpy(SvPVX(last_str) + l,
3219 SvPVX_const(last_str), l, mincount - 1);
3220 SvCUR_set(last_str, SvCUR(last_str) * mincount);
3221 /* Add additional parts. */
3222 SvCUR_set(data->last_found,
3223 SvCUR(data->last_found) - l);
3224 sv_catsv(data->last_found, last_str);
3226 SV * sv = data->last_found;
3228 SvUTF8(sv) && SvMAGICAL(sv) ?
3229 mg_find(sv, PERL_MAGIC_utf8) : NULL;
3230 if (mg && mg->mg_len >= 0)
3231 mg->mg_len += CHR_SVLEN(last_str);
3233 data->last_end += l * (mincount - 1);
3236 /* start offset must point into the last copy */
3237 data->last_start_min += minnext * (mincount - 1);
3238 data->last_start_max += is_inf ? I32_MAX
3239 : (maxcount - 1) * (minnext + data->pos_delta);
3242 /* It is counted once already... */
3243 data->pos_min += minnext * (mincount - counted);
3244 data->pos_delta += - counted * deltanext +
3245 (minnext + deltanext) * maxcount - minnext * mincount;
3246 if (mincount != maxcount) {
3247 /* Cannot extend fixed substrings found inside
3249 SCAN_COMMIT(pRExC_state,data,minlenp);
3250 if (mincount && last_str) {
3251 SV * const sv = data->last_found;
3252 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
3253 mg_find(sv, PERL_MAGIC_utf8) : NULL;
3257 sv_setsv(sv, last_str);
3258 data->last_end = data->pos_min;
3259 data->last_start_min =
3260 data->pos_min - CHR_SVLEN(last_str);
3261 data->last_start_max = is_inf
3263 : data->pos_min + data->pos_delta
3264 - CHR_SVLEN(last_str);
3266 data->longest = &(data->longest_float);
3268 SvREFCNT_dec(last_str);
3270 if (data && (fl & SF_HAS_EVAL))
3271 data->flags |= SF_HAS_EVAL;
3272 optimize_curly_tail:
3273 if (OP(oscan) != CURLYX) {
3274 while (PL_regkind[OP(next = regnext(oscan))] == NOTHING
3276 NEXT_OFF(oscan) += NEXT_OFF(next);
3279 default: /* REF and CLUMP only? */
3280 if (flags & SCF_DO_SUBSTR) {
3281 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
3282 data->longest = &(data->longest_float);
3284 is_inf = is_inf_internal = 1;
3285 if (flags & SCF_DO_STCLASS_OR)
3286 cl_anything(pRExC_state, data->start_class);
3287 flags &= ~SCF_DO_STCLASS;
3291 else if (strchr((const char*)PL_simple,OP(scan))) {
3294 if (flags & SCF_DO_SUBSTR) {
3295 SCAN_COMMIT(pRExC_state,data,minlenp);
3299 if (flags & SCF_DO_STCLASS) {
3300 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
3302 /* Some of the logic below assumes that switching
3303 locale on will only add false positives. */
3304 switch (PL_regkind[OP(scan)]) {
3308 /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */
3309 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
3310 cl_anything(pRExC_state, data->start_class);
3313 if (OP(scan) == SANY)
3315 if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */
3316 value = (ANYOF_BITMAP_TEST(data->start_class,'\n')
3317 || (data->start_class->flags & ANYOF_CLASS));
3318 cl_anything(pRExC_state, data->start_class);
3320 if (flags & SCF_DO_STCLASS_AND || !value)
3321 ANYOF_BITMAP_CLEAR(data->start_class,'\n');
3324 if (flags & SCF_DO_STCLASS_AND)
3325 cl_and(data->start_class,
3326 (struct regnode_charclass_class*)scan);
3328 cl_or(pRExC_state, data->start_class,
3329 (struct regnode_charclass_class*)scan);
3332 if (flags & SCF_DO_STCLASS_AND) {
3333 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3334 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
3335 for (value = 0; value < 256; value++)
3336 if (!isALNUM(value))
3337 ANYOF_BITMAP_CLEAR(data->start_class, value);
3341 if (data->start_class->flags & ANYOF_LOCALE)
3342 ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
3344 for (value = 0; value < 256; value++)
3346 ANYOF_BITMAP_SET(data->start_class, value);
3351 if (flags & SCF_DO_STCLASS_AND) {
3352 if (data->start_class->flags & ANYOF_LOCALE)
3353 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
3356 ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
3357 data->start_class->flags |= ANYOF_LOCALE;
3361 if (flags & SCF_DO_STCLASS_AND) {
3362 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3363 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
3364 for (value = 0; value < 256; value++)
3366 ANYOF_BITMAP_CLEAR(data->start_class, value);
3370 if (data->start_class->flags & ANYOF_LOCALE)
3371 ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
3373 for (value = 0; value < 256; value++)
3374 if (!isALNUM(value))
3375 ANYOF_BITMAP_SET(data->start_class, value);
3380 if (flags & SCF_DO_STCLASS_AND) {
3381 if (data->start_class->flags & ANYOF_LOCALE)
3382 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
3385 data->start_class->flags |= ANYOF_LOCALE;
3386 ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
3390 if (flags & SCF_DO_STCLASS_AND) {
3391 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3392 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
3393 for (value = 0; value < 256; value++)
3394 if (!isSPACE(value))
3395 ANYOF_BITMAP_CLEAR(data->start_class, value);
3399 if (data->start_class->flags & ANYOF_LOCALE)
3400 ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
3402 for (value = 0; value < 256; value++)
3404 ANYOF_BITMAP_SET(data->start_class, value);
3409 if (flags & SCF_DO_STCLASS_AND) {
3410 if (data->start_class->flags & ANYOF_LOCALE)
3411 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
3414 data->start_class->flags |= ANYOF_LOCALE;
3415 ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
3419 if (flags & SCF_DO_STCLASS_AND) {
3420 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3421 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
3422 for (value = 0; value < 256; value++)
3424 ANYOF_BITMAP_CLEAR(data->start_class, value);
3428 if (data->start_class->flags & ANYOF_LOCALE)
3429 ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
3431 for (value = 0; value < 256; value++)
3432 if (!isSPACE(value))
3433 ANYOF_BITMAP_SET(data->start_class, value);
3438 if (flags & SCF_DO_STCLASS_AND) {
3439 if (data->start_class->flags & ANYOF_LOCALE) {
3440 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
3441 for (value = 0; value < 256; value++)
3442 if (!isSPACE(value))
3443 ANYOF_BITMAP_CLEAR(data->start_class, value);
3447 data->start_class->flags |= ANYOF_LOCALE;
3448 ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
3452 if (flags & SCF_DO_STCLASS_AND) {
3453 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT);
3454 for (value = 0; value < 256; value++)
3455 if (!isDIGIT(value))
3456 ANYOF_BITMAP_CLEAR(data->start_class, value);
3459 if (data->start_class->flags & ANYOF_LOCALE)
3460 ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT);
3462 for (value = 0; value < 256; value++)
3464 ANYOF_BITMAP_SET(data->start_class, value);
3469 if (flags & SCF_DO_STCLASS_AND) {
3470 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_DIGIT);
3471 for (value = 0; value < 256; value++)
3473 ANYOF_BITMAP_CLEAR(data->start_class, value);
3476 if (data->start_class->flags & ANYOF_LOCALE)
3477 ANYOF_CLASS_SET(data->start_class,ANYOF_NDIGIT);
3479 for (value = 0; value < 256; value++)
3480 if (!isDIGIT(value))
3481 ANYOF_BITMAP_SET(data->start_class, value);
3486 if (flags & SCF_DO_STCLASS_OR)
3487 cl_and(data->start_class, and_withp);
3488 flags &= ~SCF_DO_STCLASS;
3491 else if (PL_regkind[OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
3492 data->flags |= (OP(scan) == MEOL
3496 else if ( PL_regkind[OP(scan)] == BRANCHJ
3497 /* Lookbehind, or need to calculate parens/evals/stclass: */
3498 && (scan->flags || data || (flags & SCF_DO_STCLASS))
3499 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
3500 if ( !PERL_ENABLE_POSITIVE_ASSERTION_STUDY
3501 || OP(scan) == UNLESSM )
3503 /* Negative Lookahead/lookbehind
3504 In this case we can't do fixed string optimisation.
3507 I32 deltanext, minnext, fake = 0;
3509 struct regnode_charclass_class intrnl;
3512 data_fake.flags = 0;
3514 data_fake.whilem_c = data->whilem_c;
3515 data_fake.last_closep = data->last_closep;
3518 data_fake.last_closep = &fake;
3519 data_fake.pos_delta = delta;
3520 if ( flags & SCF_DO_STCLASS && !scan->flags
3521 && OP(scan) == IFMATCH ) { /* Lookahead */
3522 cl_init(pRExC_state, &intrnl);
3523 data_fake.start_class = &intrnl;
3524 f |= SCF_DO_STCLASS_AND;
3526 if (flags & SCF_WHILEM_VISITED_POS)
3527 f |= SCF_WHILEM_VISITED_POS;
3528 next = regnext(scan);
3529 nscan = NEXTOPER(NEXTOPER(scan));
3530 minnext = study_chunk(pRExC_state, &nscan, minlenp, &deltanext,
3531 last, &data_fake, stopparen, recursed, NULL, f, depth+1);
3534 FAIL("Variable length lookbehind not implemented");
3536 else if (minnext > (I32)U8_MAX) {
3537 FAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
3539 scan->flags = (U8)minnext;
3542 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
3544 if (data_fake.flags & SF_HAS_EVAL)
3545 data->flags |= SF_HAS_EVAL;
3546 data->whilem_c = data_fake.whilem_c;
3548 if (f & SCF_DO_STCLASS_AND) {
3549 const int was = (data->start_class->flags & ANYOF_EOS);
3551 cl_and(data->start_class, &intrnl);
3553 data->start_class->flags |= ANYOF_EOS;
3556 #if PERL_ENABLE_POSITIVE_ASSERTION_STUDY
3558 /* Positive Lookahead/lookbehind
3559 In this case we can do fixed string optimisation,
3560 but we must be careful about it. Note in the case of
3561 lookbehind the positions will be offset by the minimum
3562 length of the pattern, something we won't know about
3563 until after the recurse.
3565 I32 deltanext, fake = 0;
3567 struct regnode_charclass_class intrnl;
3569 /* We use SAVEFREEPV so that when the full compile
3570 is finished perl will clean up the allocated
3571 minlens when its all done. This was we don't
3572 have to worry about freeing them when we know
3573 they wont be used, which would be a pain.
3576 Newx( minnextp, 1, I32 );
3577 SAVEFREEPV(minnextp);
3580 StructCopy(data, &data_fake, scan_data_t);
3581 if ((flags & SCF_DO_SUBSTR) && data->last_found) {
3584 SCAN_COMMIT(pRExC_state, &data_fake,minlenp);
3585 data_fake.last_found=newSVsv(data->last_found);
3589 data_fake.last_closep = &fake;
3590 data_fake.flags = 0;
3591 data_fake.pos_delta = delta;
3593 data_fake.flags |= SF_IS_INF;
3594 if ( flags & SCF_DO_STCLASS && !scan->flags
3595 && OP(scan) == IFMATCH ) { /* Lookahead */
3596 cl_init(pRExC_state, &intrnl);
3597 data_fake.start_class = &intrnl;
3598 f |= SCF_DO_STCLASS_AND;
3600 if (flags & SCF_WHILEM_VISITED_POS)
3601 f |= SCF_WHILEM_VISITED_POS;
3602 next = regnext(scan);
3603 nscan = NEXTOPER(NEXTOPER(scan));
3605 *minnextp = study_chunk(pRExC_state, &nscan, minnextp, &deltanext,
3606 last, &data_fake, stopparen, recursed, NULL, f,depth+1);
3609 FAIL("Variable length lookbehind not implemented");
3611 else if (*minnextp > (I32)U8_MAX) {
3612 FAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
3614 scan->flags = (U8)*minnextp;
3619 if (f & SCF_DO_STCLASS_AND) {
3620 const int was = (data->start_class->flags & ANYOF_EOS);
3622 cl_and(data->start_class, &intrnl);
3624 data->start_class->flags |= ANYOF_EOS;
3627 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
3629 if (data_fake.flags & SF_HAS_EVAL)
3630 data->flags |= SF_HAS_EVAL;
3631 data->whilem_c = data_fake.whilem_c;
3632 if ((flags & SCF_DO_SUBSTR) && data_fake.last_found) {
3633 if (RExC_rx->minlen<*minnextp)
3634 RExC_rx->minlen=*minnextp;
3635 SCAN_COMMIT(pRExC_state, &data_fake, minnextp);
3636 SvREFCNT_dec(data_fake.last_found);
3638 if ( data_fake.minlen_fixed != minlenp )
3640 data->offset_fixed= data_fake.offset_fixed;
3641 data->minlen_fixed= data_fake.minlen_fixed;
3642 data->lookbehind_fixed+= scan->flags;
3644 if ( data_fake.minlen_float != minlenp )
3646 data->minlen_float= data_fake.minlen_float;
3647 data->offset_float_min=data_fake.offset_float_min;
3648 data->offset_float_max=data_fake.offset_float_max;
3649 data->lookbehind_float+= scan->flags;
3658 else if (OP(scan) == OPEN) {
3659 if (stopparen != (I32)ARG(scan))
3662 else if (OP(scan) == CLOSE) {
3663 if (stopparen == (I32)ARG(scan)) {
3666 if ((I32)ARG(scan) == is_par) {
3667 next = regnext(scan);
3669 if ( next && (OP(next) != WHILEM) && next < last)
3670 is_par = 0; /* Disable optimization */
3673 *(data->last_closep) = ARG(scan);
3675 else if (OP(scan) == EVAL) {
3677 data->flags |= SF_HAS_EVAL;
3679 else if ( PL_regkind[OP(scan)] == ENDLIKE ) {
3680 if (flags & SCF_DO_SUBSTR) {
3681 SCAN_COMMIT(pRExC_state,data,minlenp);
3682 flags &= ~SCF_DO_SUBSTR;
3684 if (data && OP(scan)==ACCEPT) {
3685 data->flags |= SCF_SEEN_ACCEPT;
3690 else if (OP(scan) == LOGICAL && scan->flags == 2) /* Embedded follows */
3692 if (flags & SCF_DO_SUBSTR) {
3693 SCAN_COMMIT(pRExC_state,data,minlenp);
3694 data->longest = &(data->longest_float);
3696 is_inf = is_inf_internal = 1;
3697 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
3698 cl_anything(pRExC_state, data->start_class);
3699 flags &= ~SCF_DO_STCLASS;
3701 else if (OP(scan) == GPOS) {
3702 if (!(RExC_rx->extflags & RXf_GPOS_FLOAT) &&
3703 !(delta || is_inf || (data && data->pos_delta)))
3705 if (!(RExC_rx->extflags & RXf_ANCH) && (flags & SCF_DO_SUBSTR))
3706 RExC_rx->extflags |= RXf_ANCH_GPOS;
3707 if (RExC_rx->gofs < (U32)min)
3708 RExC_rx->gofs = min;
3710 RExC_rx->extflags |= RXf_GPOS_FLOAT;
3714 #ifdef TRIE_STUDY_OPT
3715 #ifdef FULL_TRIE_STUDY
3716 else if (PL_regkind[OP(scan)] == TRIE) {
3717 /* NOTE - There is similar code to this block above for handling
3718 BRANCH nodes on the initial study. If you change stuff here
3720 regnode *trie_node= scan;
3721 regnode *tail= regnext(scan);
3722 reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
3723 I32 max1 = 0, min1 = I32_MAX;
3724 struct regnode_charclass_class accum;
3726 if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
3727 SCAN_COMMIT(pRExC_state, data,minlenp); /* Cannot merge strings after this. */
3728 if (flags & SCF_DO_STCLASS)
3729 cl_init_zero(pRExC_state, &accum);
3735 const regnode *nextbranch= NULL;
3738 for ( word=1 ; word <= trie->wordcount ; word++)
3740 I32 deltanext=0, minnext=0, f = 0, fake;
3741 struct regnode_charclass_class this_class;
3743 data_fake.flags = 0;
3745 data_fake.whilem_c = data->whilem_c;
3746 data_fake.last_closep = data->last_closep;
3749 data_fake.last_closep = &fake;
3750 data_fake.pos_delta = delta;
3751 if (flags & SCF_DO_STCLASS) {
3752 cl_init(pRExC_state, &this_class);
3753 data_fake.start_class = &this_class;
3754 f = SCF_DO_STCLASS_AND;
3756 if (flags & SCF_WHILEM_VISITED_POS)
3757 f |= SCF_WHILEM_VISITED_POS;
3759 if (trie->jump[word]) {
3761 nextbranch = trie_node + trie->jump[0];
3762 scan= trie_node + trie->jump[word];
3763 /* We go from the jump point to the branch that follows
3764 it. Note this means we need the vestigal unused branches
3765 even though they arent otherwise used.
3767 minnext = study_chunk(pRExC_state, &scan, minlenp,
3768 &deltanext, (regnode *)nextbranch, &data_fake,
3769 stopparen, recursed, NULL, f,depth+1);
3771 if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
3772 nextbranch= regnext((regnode*)nextbranch);
3774 if (min1 > (I32)(minnext + trie->minlen))
3775 min1 = minnext + trie->minlen;
3776 if (max1 < (I32)(minnext + deltanext + trie->maxlen))
3777 max1 = minnext + deltanext + trie->maxlen;
3778 if (deltanext == I32_MAX)
3779 is_inf = is_inf_internal = 1;
3781 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
3783 if (data_fake.flags & SCF_SEEN_ACCEPT) {
3784 if ( stopmin > min + min1)
3785 stopmin = min + min1;
3786 flags &= ~SCF_DO_SUBSTR;
3788 data->flags |= SCF_SEEN_ACCEPT;
3791 if (data_fake.flags & SF_HAS_EVAL)
3792 data->flags |= SF_HAS_EVAL;
3793 data->whilem_c = data_fake.whilem_c;
3795 if (flags & SCF_DO_STCLASS)
3796 cl_or(pRExC_state, &accum, &this_class);
3799 if (flags & SCF_DO_SUBSTR) {
3800 data->pos_min += min1;
3801 data->pos_delta += max1 - min1;
3802 if (max1 != min1 || is_inf)
3803 data->longest = &(data->longest_float);
3806 delta += max1 - min1;
3807 if (flags & SCF_DO_STCLASS_OR) {
3808 cl_or(pRExC_state, data->start_class, &accum);
3810 cl_and(data->start_class, and_withp);
3811 flags &= ~SCF_DO_STCLASS;
3814 else if (flags & SCF_DO_STCLASS_AND) {
3816 cl_and(data->start_class, &accum);
3817 flags &= ~SCF_DO_STCLASS;
3820 /* Switch to OR mode: cache the old value of
3821 * data->start_class */
3823 StructCopy(data->start_class, and_withp,
3824 struct regnode_charclass_class);
3825 flags &= ~SCF_DO_STCLASS_AND;
3826 StructCopy(&accum, data->start_class,
3827 struct regnode_charclass_class);
3828 flags |= SCF_DO_STCLASS_OR;
3829 data->start_class->flags |= ANYOF_EOS;
3836 else if (PL_regkind[OP(scan)] == TRIE) {
3837 reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
3840 min += trie->minlen;
3841 delta += (trie->maxlen - trie->minlen);
3842 flags &= ~SCF_DO_STCLASS; /* xxx */
3843 if (flags & SCF_DO_SUBSTR) {
3844 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
3845 data->pos_min += trie->minlen;
3846 data->pos_delta += (trie->maxlen - trie->minlen);
3847 if (trie->maxlen != trie->minlen)
3848 data->longest = &(data->longest_float);
3850 if (trie->jump) /* no more substrings -- for now /grr*/
3851 flags &= ~SCF_DO_SUBSTR;
3853 #endif /* old or new */
3854 #endif /* TRIE_STUDY_OPT */
3855 /* Else: zero-length, ignore. */
3856 scan = regnext(scan);
3861 stopparen = frame->stop;
3862 frame = frame->prev;
3863 goto fake_study_recurse;
3868 DEBUG_STUDYDATA("pre-fin:",data,depth);
3871 *deltap = is_inf_internal ? I32_MAX : delta;
3872 if (flags & SCF_DO_SUBSTR && is_inf)
3873 data->pos_delta = I32_MAX - data->pos_min;
3874 if (is_par > (I32)U8_MAX)
3876 if (is_par && pars==1 && data) {
3877 data->flags |= SF_IN_PAR;
3878 data->flags &= ~SF_HAS_PAR;
3880 else if (pars && data) {
3881 data->flags |= SF_HAS_PAR;
3882 data->flags &= ~SF_IN_PAR;
3884 if (flags & SCF_DO_STCLASS_OR)
3885 cl_and(data->start_class, and_withp);
3886 if (flags & SCF_TRIE_RESTUDY)
3887 data->flags |= SCF_TRIE_RESTUDY;
3889 DEBUG_STUDYDATA("post-fin:",data,depth);
3891 return min < stopmin ? min : stopmin;
3895 S_add_data(RExC_state_t *pRExC_state, U32 n, const char *s)
3897 U32 count = RExC_rxi->data ? RExC_rxi->data->count : 0;
3899 Renewc(RExC_rxi->data,
3900 sizeof(*RExC_rxi->data) + sizeof(void*) * (count + n - 1),
3901 char, struct reg_data);
3903 Renew(RExC_rxi->data->what, count + n, U8);
3905 Newx(RExC_rxi->data->what, n, U8);
3906 RExC_rxi->data->count = count + n;
3907 Copy(s, RExC_rxi->data->what + count, n, U8);
3911 /*XXX: todo make this not included in a non debugging perl */
3912 #ifndef PERL_IN_XSUB_RE
3914 Perl_reginitcolors(pTHX)
3917 const char * const s = PerlEnv_getenv("PERL_RE_COLORS");
3919 char *t = savepv(s);
3923 t = strchr(t, '\t');
3929 PL_colors[i] = t = (char *)"";
3934 PL_colors[i++] = (char *)"";
3941 #ifdef TRIE_STUDY_OPT
3942 #define CHECK_RESTUDY_GOTO \
3944 (data.flags & SCF_TRIE_RESTUDY) \
3948 #define CHECK_RESTUDY_GOTO
3952 - pregcomp - compile a regular expression into internal code
3954 * We can't allocate space until we know how big the compiled form will be,
3955 * but we can't compile it (and thus know how big it is) until we've got a
3956 * place to put the code. So we cheat: we compile it twice, once with code
3957 * generation turned off and size counting turned on, and once "for real".
3958 * This also means that we don't allocate space until we are sure that the
3959 * thing really will compile successfully, and we never have to move the
3960 * code and thus invalidate pointers into it. (Note that it has to be in
3961 * one piece because free() must be able to free it all.) [NB: not true in perl]
3963 * Beware that the optimization-preparation code in here knows about some
3964 * of the structure of the compiled regexp. [I'll say.]
3969 #ifndef PERL_IN_XSUB_RE
3970 #define RE_ENGINE_PTR &PL_core_reg_engine
3972 extern const struct regexp_engine my_reg_engine;
3973 #define RE_ENGINE_PTR &my_reg_engine
3976 #ifndef PERL_IN_XSUB_RE
3978 Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
3981 HV * const table = GvHV(PL_hintgv);
3982 /* Dispatch a request to compile a regexp to correct
3985 SV **ptr= hv_fetchs(table, "regcomp", FALSE);
3986 GET_RE_DEBUG_FLAGS_DECL;
3987 if (ptr && SvIOK(*ptr) && SvIV(*ptr)) {
3988 const regexp_engine *eng=INT2PTR(regexp_engine*,SvIV(*ptr));
3990 PerlIO_printf(Perl_debug_log, "Using engine %"UVxf"\n",
3993 return CALLREGCOMP_ENG(eng, exp, xend, pm);
3996 return Perl_re_compile(aTHX_ exp, xend, pm);
4001 Perl_re_compile(pTHX_ char *exp, char *xend, PMOP *pm)
4005 register regexp_internal *ri;
4013 RExC_state_t RExC_state;
4014 RExC_state_t * const pRExC_state = &RExC_state;
4015 #ifdef TRIE_STUDY_OPT
4017 RExC_state_t copyRExC_state;
4019 GET_RE_DEBUG_FLAGS_DECL;
4020 DEBUG_r(if (!PL_colorset) reginitcolors());
4023 FAIL("NULL regexp argument");
4025 RExC_utf8 = pm->op_pmdynflags & PMdf_CMP_UTF8;
4029 SV *dsv= sv_newmortal();
4030 RE_PV_QUOTED_DECL(s, RExC_utf8,
4031 dsv, RExC_precomp, (xend - exp), 60);
4032 PerlIO_printf(Perl_debug_log, "%sCompiling REx%s %s\n",
4033 PL_colors[4],PL_colors[5],s);
4035 RExC_flags = pm->op_pmflags;
4039 RExC_seen_zerolen = *exp == '^' ? -1 : 0;
4040 RExC_seen_evals = 0;
4043 /* First pass: determine size, legality. */
4051 RExC_emit = &PL_regdummy;
4052 RExC_whilem_seen = 0;
4053 RExC_charnames = NULL;
4054 RExC_open_parens = NULL;
4055 RExC_close_parens = NULL;
4057 RExC_paren_names = NULL;
4058 RExC_recurse = NULL;
4059 RExC_recurse_count = 0;
4061 #if 0 /* REGC() is (currently) a NOP at the first pass.
4062 * Clever compilers notice this and complain. --jhi */
4063 REGC((U8)REG_MAGIC, (char*)RExC_emit);
4065 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log, "Starting first pass (sizing)\n"));
4066 if (reg(pRExC_state, 0, &flags,1) == NULL) {
4067 RExC_precomp = NULL;
4071 PerlIO_printf(Perl_debug_log,
4072 "Required size %"IVdf" nodes\n"
4073 "Starting second pass (creation)\n",
4076 RExC_lastparse=NULL;
4078 /* Small enough for pointer-storage convention?
4079 If extralen==0, this means that we will not need long jumps. */
4080 if (RExC_size >= 0x10000L && RExC_extralen)
4081 RExC_size += RExC_extralen;
4084 if (RExC_whilem_seen > 15)
4085 RExC_whilem_seen = 15;
4088 /* Make room for a sentinel value at the end of the program */
4092 /* Allocate space and zero-initialize. Note, the two step process
4093 of zeroing when in debug mode, thus anything assigned has to
4094 happen after that */
4095 Newxz(r, 1, regexp);
4096 Newxc(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode),
4097 char, regexp_internal);
4098 if ( r == NULL || ri == NULL )
4099 FAIL("Regexp out of space");
4101 /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
4102 Zero(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode), char);
4104 /* bulk initialize base fields with 0. */
4105 Zero(ri, sizeof(regexp_internal), char);
4108 /* non-zero initialization begins here */
4110 r->engine= RE_ENGINE_PTR;
4112 r->prelen = xend - exp;
4113 r->precomp = savepvn(RExC_precomp, r->prelen);
4114 r->extflags = pm->op_pmflags & RXf_PMf_COMPILETIME;
4116 r->nparens = RExC_npar - 1; /* set early to validate backrefs */
4118 if (RExC_seen & REG_SEEN_RECURSE) {
4119 Newxz(RExC_open_parens, RExC_npar,regnode *);
4120 SAVEFREEPV(RExC_open_parens);
4121 Newxz(RExC_close_parens,RExC_npar,regnode *);
4122 SAVEFREEPV(RExC_close_parens);
4125 /* Useful during FAIL. */
4126 Newxz(ri->offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
4128 ri->offsets[0] = RExC_size;
4130 DEBUG_OFFSETS_r(PerlIO_printf(Perl_debug_log,
4131 "%s %"UVuf" bytes for offset annotations.\n",
4132 ri->offsets ? "Got" : "Couldn't get",
4133 (UV)((2*RExC_size+1) * sizeof(U32))));
4138 /* Second pass: emit code. */
4139 RExC_flags = pm->op_pmflags; /* don't let top level (?i) bleed */
4144 RExC_emit_start = ri->program;
4145 RExC_emit = ri->program;
4147 /* put a sentinal on the end of the program so we can check for
4149 ri->program[RExC_size].type = 255;
4151 /* Store the count of eval-groups for security checks: */
4152 RExC_rx->seen_evals = RExC_seen_evals;
4153 REGC((U8)REG_MAGIC, (char*) RExC_emit++);
4154 if (reg(pRExC_state, 0, &flags,1) == NULL)
4157 /* XXXX To minimize changes to RE engine we always allocate
4158 3-units-long substrs field. */
4159 Newx(r->substrs, 1, struct reg_substr_data);
4160 if (RExC_recurse_count) {
4161 Newxz(RExC_recurse,RExC_recurse_count,regnode *);
4162 SAVEFREEPV(RExC_recurse);
4166 r->minlen = minlen = sawplus = sawopen = 0;
4167 Zero(r->substrs, 1, struct reg_substr_data);
4169 #ifdef TRIE_STUDY_OPT
4172 DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log,"Restudying\n"));
4174 RExC_state = copyRExC_state;
4175 if (seen & REG_TOP_LEVEL_BRANCHES)
4176 RExC_seen |= REG_TOP_LEVEL_BRANCHES;
4178 RExC_seen &= ~REG_TOP_LEVEL_BRANCHES;
4179 if (data.last_found) {
4180 SvREFCNT_dec(data.longest_fixed);
4181 SvREFCNT_dec(data.longest_float);
4182 SvREFCNT_dec(data.last_found);
4184 StructCopy(&zero_scan_data, &data, scan_data_t);
4186 StructCopy(&zero_scan_data, &data, scan_data_t);
4187 copyRExC_state = RExC_state;
4190 StructCopy(&zero_scan_data, &data, scan_data_t);
4193 /* Dig out information for optimizations. */
4194 r->extflags = pm->op_pmflags & RXf_PMf_COMPILETIME; /* Again? */
4195 pm->op_pmflags = RExC_flags;
4197 r->extflags |= RXf_UTF8; /* Unicode in it? */
4198 ri->regstclass = NULL;
4199 if (RExC_naughty >= 10) /* Probably an expensive pattern. */
4200 r->intflags |= PREGf_NAUGHTY;
4201 scan = ri->program + 1; /* First BRANCH. */
4203 /* testing for BRANCH here tells us whether there is "must appear"
4204 data in the pattern. If there is then we can use it for optimisations */
4205 if (!(RExC_seen & REG_TOP_LEVEL_BRANCHES)) { /* Only one top-level choice. */
4207 STRLEN longest_float_length, longest_fixed_length;
4208 struct regnode_charclass_class ch_class; /* pointed to by data */
4210 I32 last_close = 0; /* pointed to by data */
4213 /* Skip introductions and multiplicators >= 1. */
4214 while ((OP(first) == OPEN && (sawopen = 1)) ||
4215 /* An OR of *one* alternative - should not happen now. */
4216 (OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
4217 /* for now we can't handle lookbehind IFMATCH*/
4218 (OP(first) == IFMATCH && !first->flags) ||
4219 (OP(first) == PLUS) ||
4220 (OP(first) == MINMOD) ||
4221 /* An {n,m} with n>0 */
4222 (PL_regkind[OP(first)] == CURLY && ARG1(first) > 0) )
4225 if (OP(first) == PLUS)
4228 first += regarglen[OP(first)];
4229 if (OP(first) == IFMATCH) {
4230 first = NEXTOPER(first);
4231 first += EXTRA_STEP_2ARGS;
4232 } else /* XXX possible optimisation for /(?=)/ */
4233 first = NEXTOPER(first);
4236 /* Starting-point info. */
4238 DEBUG_PEEP("first:",first,0);
4239 /* Ignore EXACT as we deal with it later. */
4240 if (PL_regkind[OP(first)] == EXACT) {
4241 if (OP(first) == EXACT)
4242 NOOP; /* Empty, get anchored substr later. */
4243 else if ((OP(first) == EXACTF || OP(first) == EXACTFL))
4244 ri->regstclass = first;
4247 else if (PL_regkind[OP(first)] == TRIE &&
4248 ((reg_trie_data *)ri->data->data[ ARG(first) ])->minlen>0)
4251 /* this can happen only on restudy */
4252 if ( OP(first) == TRIE ) {
4253 struct regnode_1 *trieop = (struct regnode_1 *)
4254 PerlMemShared_calloc(1, sizeof(struct regnode_1));
4255 StructCopy(first,trieop,struct regnode_1);
4256 trie_op=(regnode *)trieop;
4258 struct regnode_charclass *trieop = (struct regnode_charclass *)
4259 PerlMemShared_calloc(1, sizeof(struct regnode_charclass));
4260 StructCopy(first,trieop,struct regnode_charclass);
4261 trie_op=(regnode *)trieop;
4264 make_trie_failtable(pRExC_state, (regnode *)first, trie_op, 0);
4265 ri->regstclass = trie_op;
4268 else if (strchr((const char*)PL_simple,OP(first)))
4269 ri->regstclass = first;
4270 else if (PL_regkind[OP(first)] == BOUND ||
4271 PL_regkind[OP(first)] == NBOUND)
4272 ri->regstclass = first;
4273 else if (PL_regkind[OP(first)] == BOL) {
4274 r->extflags |= (OP(first) == MBOL
4276 : (OP(first) == SBOL
4279 first = NEXTOPER(first);
4282 else if (OP(first) == GPOS) {
4283 r->extflags |= RXf_ANCH_GPOS;
4284 first = NEXTOPER(first);
4287 else if ((!sawopen || !RExC_sawback) &&
4288 (OP(first) == STAR &&
4289 PL_regkind[OP(NEXTOPER(first))] == REG_ANY) &&
4290 !(r->extflags & RXf_ANCH) && !(RExC_seen & REG_SEEN_EVAL))
4292 /* turn .* into ^.* with an implied $*=1 */
4294 (OP(NEXTOPER(first)) == REG_ANY)
4297 r->extflags |= type;
4298 r->intflags |= PREGf_IMPLICIT;
4299 first = NEXTOPER(first);
4302 if (sawplus && (!sawopen || !RExC_sawback)
4303 && !(RExC_seen & REG_SEEN_EVAL)) /* May examine pos and $& */
4304 /* x+ must match at the 1st pos of run of x's */
4305 r->intflags |= PREGf_SKIP;
4307 /* Scan is after the zeroth branch, first is atomic matcher. */
4308 #ifdef TRIE_STUDY_OPT
4311 PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
4312 (IV)(first - scan + 1))
4316 PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
4317 (IV)(first - scan + 1))
4323 * If there's something expensive in the r.e., find the
4324 * longest literal string that must appear and make it the
4325 * regmust. Resolve ties in favor of later strings, since
4326 * the regstart check works with the beginning of the r.e.
4327 * and avoiding duplication strengthens checking. Not a
4328 * strong reason, but sufficient in the absence of others.
4329 * [Now we resolve ties in favor of the earlier string if
4330 * it happens that c_offset_min has been invalidated, since the
4331 * earlier string may buy us something the later one won't.]
4334 data.longest_fixed = newSVpvs("");
4335 data.longest_float = newSVpvs("");
4336 data.last_found = newSVpvs("");
4337 data.longest = &(data.longest_fixed);
4339 if (!ri->regstclass) {
4340 cl_init(pRExC_state, &ch_class);
4341 data.start_class = &ch_class;
4342 stclass_flag = SCF_DO_STCLASS_AND;
4343 } else /* XXXX Check for BOUND? */
4345 data.last_closep = &last_close;
4347 minlen = study_chunk(pRExC_state, &first, &minlen, &fake, scan + RExC_size, /* Up to end */
4348 &data, -1, NULL, NULL,
4349 SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag,0);
4355 if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
4356 && data.last_start_min == 0 && data.last_end > 0
4357 && !RExC_seen_zerolen
4358 && !(RExC_seen & REG_SEEN_VERBARG)
4359 && (!(RExC_seen & REG_SEEN_GPOS) || (r->extflags & RXf_ANCH_GPOS)))
4360 r->extflags |= RXf_CHECK_ALL;
4361 scan_commit(pRExC_state, &data,&minlen,0);
4362 SvREFCNT_dec(data.last_found);
4364 /* Note that code very similar to this but for anchored string
4365 follows immediately below, changes may need to be made to both.
4368 longest_float_length = CHR_SVLEN(data.longest_float);
4369 if (longest_float_length
4370 || (data.flags & SF_FL_BEFORE_EOL
4371 && (!(data.flags & SF_FL_BEFORE_MEOL)
4372 || (RExC_flags & RXf_PMf_MULTILINE))))
4376 if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */
4377 && data.offset_fixed == data.offset_float_min
4378 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
4379 goto remove_float; /* As in (a)+. */
4381 /* copy the information about the longest float from the reg_scan_data
4382 over to the program. */
4383 if (SvUTF8(data.longest_float)) {
4384 r->float_utf8 = data.longest_float;
4385 r->float_substr = NULL;
4387 r->float_substr = data.longest_float;
4388 r->float_utf8 = NULL;
4390 /* float_end_shift is how many chars that must be matched that
4391 follow this item. We calculate it ahead of time as once the
4392 lookbehind offset is added in we lose the ability to correctly
4394 ml = data.minlen_float ? *(data.minlen_float)
4395 : (I32)longest_float_length;
4396 r->float_end_shift = ml - data.offset_float_min
4397 - longest_float_length + (SvTAIL(data.longest_float) != 0)
4398 + data.lookbehind_float;
4399 r->float_min_offset = data.offset_float_min - data.lookbehind_float;
4400 r->float_max_offset = data.offset_float_max;
4401 if (data.offset_float_max < I32_MAX) /* Don't offset infinity */
4402 r->float_max_offset -= data.lookbehind_float;
4404 t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
4405 && (!(data.flags & SF_FL_BEFORE_MEOL)
4406 || (RExC_flags & RXf_PMf_MULTILINE)));
4407 fbm_compile(data.longest_float, t ? FBMcf_TAIL : 0);
4411 r->float_substr = r->float_utf8 = NULL;
4412 SvREFCNT_dec(data.longest_float);
4413 longest_float_length = 0;
4416 /* Note that code very similar to this but for floating string
4417 is immediately above, changes may need to be made to both.
4420 longest_fixed_length = CHR_SVLEN(data.longest_fixed);
4421 if (longest_fixed_length
4422 || (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
4423 && (!(data.flags & SF_FIX_BEFORE_MEOL)
4424 || (RExC_flags & RXf_PMf_MULTILINE))))
4428 /* copy the information about the longest fixed
4429 from the reg_scan_data over to the program. */
4430 if (SvUTF8(data.longest_fixed)) {
4431 r->anchored_utf8 = data.longest_fixed;
4432 r->anchored_substr = NULL;
4434 r->anchored_substr = data.longest_fixed;
4435 r->anchored_utf8 = NULL;
4437 /* fixed_end_shift is how many chars that must be matched that
4438 follow this item. We calculate it ahead of time as once the
4439 lookbehind offset is added in we lose the ability to correctly
4441 ml = data.minlen_fixed ? *(data.minlen_fixed)
4442 : (I32)longest_fixed_length;
4443 r->anchored_end_shift = ml - data.offset_fixed
4444 - longest_fixed_length + (SvTAIL(data.longest_fixed) != 0)
4445 + data.lookbehind_fixed;
4446 r->anchored_offset = data.offset_fixed - data.lookbehind_fixed;
4448 t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
4449 && (!(data.flags & SF_FIX_BEFORE_MEOL)
4450 || (RExC_flags & RXf_PMf_MULTILINE)));
4451 fbm_compile(data.longest_fixed, t ? FBMcf_TAIL : 0);
4454 r->anchored_substr = r->anchored_utf8 = NULL;
4455 SvREFCNT_dec(data.longest_fixed);
4456 longest_fixed_length = 0;
4459 && (OP(ri->regstclass) == REG_ANY || OP(ri->regstclass) == SANY))
4460 ri->regstclass = NULL;
4461 if ((!(r->anchored_substr || r->anchored_utf8) || r->anchored_offset)
4463 && !(data.start_class->flags & ANYOF_EOS)
4464 && !cl_is_anything(data.start_class))
4466 const U32 n = add_data(pRExC_state, 1, "f");
4468 Newx(RExC_rxi->data->data[n], 1,
4469 struct regnode_charclass_class);
4470 StructCopy(data.start_class,
4471 (struct regnode_charclass_class*)RExC_rxi->data->data[n],
4472 struct regnode_charclass_class);
4473 ri->regstclass = (regnode*)RExC_rxi->data->data[n];
4474 r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
4475 DEBUG_COMPILE_r({ SV *sv = sv_newmortal();
4476 regprop(r, sv, (regnode*)data.start_class);
4477 PerlIO_printf(Perl_debug_log,
4478 "synthetic stclass \"%s\".\n",
4479 SvPVX_const(sv));});
4482 /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
4483 if (longest_fixed_length > longest_float_length) {
4484 r->check_end_shift = r->anchored_end_shift;
4485 r->check_substr = r->anchored_substr;
4486 r->check_utf8 = r->anchored_utf8;
4487 r->check_offset_min = r->check_offset_max = r->anchored_offset;
4488 if (r->extflags & RXf_ANCH_SINGLE)
4489 r->extflags |= RXf_NOSCAN;
4492 r->check_end_shift = r->float_end_shift;
4493 r->check_substr = r->float_substr;
4494 r->check_utf8 = r->float_utf8;
4495 r->check_offset_min = r->float_min_offset;
4496 r->check_offset_max = r->float_max_offset;
4498 /* XXXX Currently intuiting is not compatible with ANCH_GPOS.
4499 This should be changed ASAP! */
4500 if ((r->check_substr || r->check_utf8) && !(r->extflags & RXf_ANCH_GPOS)) {
4501 r->extflags |= RXf_USE_INTUIT;
4502 if (SvTAIL(r->check_substr ? r->check_substr : r->check_utf8))
4503 r->extflags |= RXf_INTUIT_TAIL;
4505 /* XXX Unneeded? dmq (shouldn't as this is handled elsewhere)
4506 if ( (STRLEN)minlen < longest_float_length )
4507 minlen= longest_float_length;
4508 if ( (STRLEN)minlen < longest_fixed_length )
4509 minlen= longest_fixed_length;
4513 /* Several toplevels. Best we can is to set minlen. */
4515 struct regnode_charclass_class ch_class;
4518 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log, "\nMulti Top Level\n"));
4520 scan = ri->program + 1;
4521 cl_init(pRExC_state, &ch_class);
4522 data.start_class = &ch_class;
4523 data.last_closep = &last_close;
4526 minlen = study_chunk(pRExC_state, &scan, &minlen, &fake, scan + RExC_size,
4527 &data, -1, NULL, NULL, SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS,0);
4531 r->check_substr = r->check_utf8 = r->anchored_substr = r->anchored_utf8
4532 = r->float_substr = r->float_utf8 = NULL;
4533 if (!(data.start_class->flags & ANYOF_EOS)
4534 && !cl_is_anything(data.start_class))
4536 const U32 n = add_data(pRExC_state, 1, "f");
4538 Newx(RExC_rxi->data->data[n], 1,
4539 struct regnode_charclass_class);
4540 StructCopy(data.start_class,
4541 (struct regnode_charclass_class*)RExC_rxi->data->data[n],
4542 struct regnode_charclass_class);
4543 ri->regstclass = (regnode*)RExC_rxi->data->data[n];
4544 r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
4545 DEBUG_COMPILE_r({ SV* sv = sv_newmortal();
4546 regprop(r, sv, (regnode*)data.start_class);
4547 PerlIO_printf(Perl_debug_log,
4548 "synthetic stclass \"%s\".\n",
4549 SvPVX_const(sv));});
4553 /* Guard against an embedded (?=) or (?<=) with a longer minlen than
4554 the "real" pattern. */
4556 PerlIO_printf(Perl_debug_log,"minlen: %"IVdf" r->minlen:%"IVdf"\n",
4557 (IV)minlen, (IV)r->minlen);
4559 r->minlenret = minlen;
4560 if (r->minlen < minlen)
4563 if (RExC_seen & REG_SEEN_GPOS)
4564 r->extflags |= RXf_GPOS_SEEN;
4565 if (RExC_seen & REG_SEEN_LOOKBEHIND)
4566 r->extflags |= RXf_LOOKBEHIND_SEEN;
4567 if (RExC_seen & REG_SEEN_EVAL)
4568 r->extflags |= RXf_EVAL_SEEN;
4569 if (RExC_seen & REG_SEEN_CANY)
4570 r->extflags |= RXf_CANY_SEEN;
4571 if (RExC_seen & REG_SEEN_VERBARG)
4572 r->intflags |= PREGf_VERBARG_SEEN;
4573 if (RExC_seen & REG_SEEN_CUTGROUP)
4574 r->intflags |= PREGf_CUTGROUP_SEEN;
4575 if (RExC_paren_names)
4576 r->paren_names = (HV*)SvREFCNT_inc(RExC_paren_names);
4578 r->paren_names = NULL;
4580 if (RExC_recurse_count) {
4581 for ( ; RExC_recurse_count ; RExC_recurse_count-- ) {
4582 const regnode *scan = RExC_recurse[RExC_recurse_count-1];
4583 ARG2L_SET( scan, RExC_open_parens[ARG(scan)-1] - scan );
4586 Newxz(r->startp, RExC_npar, I32);
4587 Newxz(r->endp, RExC_npar, I32);
4588 /* assume we don't need to swap parens around before we match */
4591 PerlIO_printf(Perl_debug_log,"Final program:\n");
4594 DEBUG_OFFSETS_r(if (ri->offsets) {
4595 const U32 len = ri->offsets[0];
4597 GET_RE_DEBUG_FLAGS_DECL;
4598 PerlIO_printf(Perl_debug_log, "Offsets: [%"UVuf"]\n\t", (UV)ri->offsets[0]);
4599 for (i = 1; i <= len; i++) {
4600 if (ri->offsets[i*2-1] || ri->offsets[i*2])
4601 PerlIO_printf(Perl_debug_log, "%"UVuf":%"UVuf"[%"UVuf"] ",
4602 (UV)i, (UV)ri->offsets[i*2-1], (UV)ri->offsets[i*2]);
4604 PerlIO_printf(Perl_debug_log, "\n");
4609 #undef CORE_ONLY_BLOCK
4610 #undef RE_ENGINE_PTR
4612 #ifndef PERL_IN_XSUB_RE
4614 Perl_reg_named_buff_sv(pTHX_ SV* namesv)
4616 I32 parno = 0; /* no match */
4618 const REGEXP * const rx = PM_GETRE(PL_curpm);
4619 if (rx && rx->paren_names) {
4620 HE *he_str = hv_fetch_ent( rx->paren_names, namesv, 0, 0 );
4623 SV* sv_dat=HeVAL(he_str);
4624 I32 *nums=(I32*)SvPVX(sv_dat);
4625 for ( i=0; i<SvIVX(sv_dat); i++ ) {
4626 if ((I32)(rx->lastparen) >= nums[i] &&
4627 rx->endp[nums[i]] != -1)
4640 SV *sv= sv_newmortal();
4641 Perl_sv_setpvf(aTHX_ sv, "%"IVdf,(IV)parno);
4642 gv_paren= Perl_gv_fetchsv(aTHX_ sv, GV_ADD, SVt_PVGV);
4643 return GvSVn(gv_paren);
4648 /* Scans the name of a named buffer from the pattern.
4649 * If flags is REG_RSN_RETURN_NULL returns null.
4650 * If flags is REG_RSN_RETURN_NAME returns an SV* containing the name
4651 * If flags is REG_RSN_RETURN_DATA returns the data SV* corresponding
4652 * to the parsed name as looked up in the RExC_paren_names hash.
4653 * If there is an error throws a vFAIL().. type exception.
4656 #define REG_RSN_RETURN_NULL 0
4657 #define REG_RSN_RETURN_NAME 1
4658 #define REG_RSN_RETURN_DATA 2
4661 S_reg_scan_name(pTHX_ RExC_state_t *pRExC_state, U32 flags) {
4662 char *name_start = RExC_parse;
4665 while( isIDFIRST_uni(utf8n_to_uvchr((U8*)RExC_parse,
4666 RExC_end - RExC_parse, &numlen, UTF8_ALLOW_DEFAULT)))
4668 RExC_parse += numlen;
4671 while( isIDFIRST(*RExC_parse) )
4675 SV* sv_name = sv_2mortal(Perl_newSVpvn(aTHX_ name_start,
4676 (int)(RExC_parse - name_start)));
4679 if ( flags == REG_RSN_RETURN_NAME)
4681 else if (flags==REG_RSN_RETURN_DATA) {
4684 if ( ! sv_name ) /* should not happen*/
4685 Perl_croak(aTHX_ "panic: no svname in reg_scan_name");
4686 if (RExC_paren_names)
4687 he_str = hv_fetch_ent( RExC_paren_names, sv_name, 0, 0 );
4689 sv_dat = HeVAL(he_str);
4691 vFAIL("Reference to nonexistent named group");
4695 Perl_croak(aTHX_ "panic: bad flag in reg_scan_name");
4702 #define DEBUG_PARSE_MSG(funcname) DEBUG_PARSE_r({ \
4703 int rem=(int)(RExC_end - RExC_parse); \
4712 if (RExC_lastparse!=RExC_parse) \
4713 PerlIO_printf(Perl_debug_log," >%.*s%-*s", \
4716 iscut ? "..." : "<" \
4719 PerlIO_printf(Perl_debug_log,"%16s",""); \
4724 num=REG_NODE_NUM(RExC_emit); \
4725 if (RExC_lastnum!=num) \
4726 PerlIO_printf(Perl_debug_log,"|%4d",num); \
4728 PerlIO_printf(Perl_debug_log,"|%4s",""); \
4729 PerlIO_printf(Perl_debug_log,"|%*s%-4s", \
4730 (int)((depth*2)), "", \
4734 RExC_lastparse=RExC_parse; \
4739 #define DEBUG_PARSE(funcname) DEBUG_PARSE_r({ \
4740 DEBUG_PARSE_MSG((funcname)); \
4741 PerlIO_printf(Perl_debug_log,"%4s","\n"); \
4743 #define DEBUG_PARSE_FMT(funcname,fmt,args) DEBUG_PARSE_r({ \
4744 DEBUG_PARSE_MSG((funcname)); \
4745 PerlIO_printf(Perl_debug_log,fmt "\n",args); \
4748 - reg - regular expression, i.e. main body or parenthesized thing
4750 * Caller must absorb opening parenthesis.
4752 * Combining parenthesis handling with the base level of regular expression
4753 * is a trifle forced, but the need to tie the tails of the branches to what
4754 * follows makes it hard to avoid.
4756 #define REGTAIL(x,y,z) regtail((x),(y),(z),depth+1)
4758 #define REGTAIL_STUDY(x,y,z) regtail_study((x),(y),(z),depth+1)
4760 #define REGTAIL_STUDY(x,y,z) regtail((x),(y),(z),depth+1)
4763 /* this idea is borrowed from STR_WITH_LEN in handy.h */
4764 #define CHECK_WORD(s,v,l) \
4765 (((sizeof(s)-1)==(l)) && (strnEQ(start_verb, (s ""), (sizeof(s)-1))))
4768 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
4769 /* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
4772 register regnode *ret; /* Will be the head of the group. */
4773 register regnode *br;
4774 register regnode *lastbr;
4775 register regnode *ender = NULL;
4776 register I32 parno = 0;
4778 const I32 oregflags = RExC_flags;
4779 bool have_branch = 0;
4782 /* for (?g), (?gc), and (?o) warnings; warning
4783 about (?c) will warn about (?g) -- japhy */
4785 #define WASTED_O 0x01
4786 #define WASTED_G 0x02
4787 #define WASTED_C 0x04
4788 #define WASTED_GC (0x02|0x04)
4789 I32 wastedflags = 0x00;
4791 char * parse_start = RExC_parse; /* MJD */
4792 char * const oregcomp_parse = RExC_parse;
4794 GET_RE_DEBUG_FLAGS_DECL;
4795 DEBUG_PARSE("reg ");
4798 *flagp = 0; /* Tentatively. */
4801 /* Make an OPEN node, if parenthesized. */
4803 if ( *RExC_parse == '*') { /* (*VERB:ARG) */
4804 char *start_verb = RExC_parse;
4805 STRLEN verb_len = 0;
4806 char *start_arg = NULL;
4807 unsigned char op = 0;
4809 int internal_argval = 0; /* internal_argval is only useful if !argok */
4810 while ( *RExC_parse && *RExC_parse != ')' ) {
4811 if ( *RExC_parse == ':' ) {
4812 start_arg = RExC_parse + 1;
4818 verb_len = RExC_parse - start_verb;
4821 while ( *RExC_parse && *RExC_parse != ')' )
4823 if ( *RExC_parse != ')' )
4824 vFAIL("Unterminated verb pattern argument");
4825 if ( RExC_parse == start_arg )
4828 if ( *RExC_parse != ')' )
4829 vFAIL("Unterminated verb pattern");
4832 switch ( *start_verb ) {
4833 case 'A': /* (*ACCEPT) */
4834 if ( CHECK_WORD("ACCEPT",start_verb,verb_len) ) {
4836 internal_argval = RExC_nestroot;
4839 case 'C': /* (*COMMIT) */
4840 if ( CHECK_WORD("COMMIT",start_verb,verb_len) )
4843 case 'F': /* (*FAIL) */
4844 if ( verb_len==1 || CHECK_WORD("FAIL",start_verb,verb_len) ) {
4849 case ':': /* (*:NAME) */
4850 case 'M': /* (*MARK:NAME) */
4851 if ( verb_len==0 || CHECK_WORD("MARK",start_verb,verb_len) ) {
4856 case 'P': /* (*PRUNE) */
4857 if ( CHECK_WORD("PRUNE",start_verb,verb_len) )
4860 case 'S': /* (*SKIP) */
4861 if ( CHECK_WORD("SKIP",start_verb,verb_len) )
4864 case 'T': /* (*THEN) */
4865 /* [19:06] <TimToady> :: is then */
4866 if ( CHECK_WORD("THEN",start_verb,verb_len) ) {
4868 RExC_seen |= REG_SEEN_CUTGROUP;
4874 vFAIL3("Unknown verb pattern '%.*s'",
4875 verb_len, start_verb);
4878 if ( start_arg && internal_argval ) {
4879 vFAIL3("Verb pattern '%.*s' may not have an argument",
4880 verb_len, start_verb);
4881 } else if ( argok < 0 && !start_arg ) {
4882 vFAIL3("Verb pattern '%.*s' has a mandatory argument",
4883 verb_len, start_verb);
4885 ret = reganode(pRExC_state, op, internal_argval);
4886 if ( ! internal_argval && ! SIZE_ONLY ) {
4888 SV *sv = newSVpvn( start_arg, RExC_parse - start_arg);
4889 ARG(ret) = add_data( pRExC_state, 1, "S" );
4890 RExC_rxi->data->data[ARG(ret)]=(void*)sv;
4897 if (!internal_argval)
4898 RExC_seen |= REG_SEEN_VERBARG;
4899 } else if ( start_arg ) {
4900 vFAIL3("Verb pattern '%.*s' may not have an argument",
4901 verb_len, start_verb);
4903 ret = reg_node(pRExC_state, op);
4905 nextchar(pRExC_state);
4908 if (*RExC_parse == '?') { /* (?...) */
4909 U32 posflags = 0, negflags = 0;
4910 U32 *flagsp = &posflags;
4911 bool is_logical = 0;
4912 const char * const seqstart = RExC_parse;
4915 paren = *RExC_parse++;
4916 ret = NULL; /* For look-ahead/behind. */
4919 case '<': /* (?<...) */
4920 if (*RExC_parse == '!')
4922 else if (*RExC_parse != '=')
4927 case '\'': /* (?'...') */
4928 name_start= RExC_parse;
4929 svname = reg_scan_name(pRExC_state,
4930 SIZE_ONLY ? /* reverse test from the others */
4931 REG_RSN_RETURN_NAME :
4932 REG_RSN_RETURN_NULL);
4933 if (RExC_parse == name_start)
4935 if (*RExC_parse != paren)
4936 vFAIL2("Sequence (?%c... not terminated",
4937 paren=='>' ? '<' : paren);
4941 if (!svname) /* shouldnt happen */
4943 "panic: reg_scan_name returned NULL");
4944 if (!RExC_paren_names) {
4945 RExC_paren_names= newHV();
4946 sv_2mortal((SV*)RExC_paren_names);
4948 he_str = hv_fetch_ent( RExC_paren_names, svname, 1, 0 );
4950 sv_dat = HeVAL(he_str);
4952 /* croak baby croak */
4954 "panic: paren_name hash element allocation failed");
4955 } else if ( SvPOK(sv_dat) ) {
4956 IV count=SvIV(sv_dat);
4957 I32 *pv=(I32*)SvGROW(sv_dat,SvCUR(sv_dat)+sizeof(I32)+1);
4958 SvCUR_set(sv_dat,SvCUR(sv_dat)+sizeof(I32));
4959 pv[count]=RExC_npar;
4962 (void)SvUPGRADE(sv_dat,SVt_PVNV);
4963 sv_setpvn(sv_dat, (char *)&(RExC_npar), sizeof(I32));
4968 /*sv_dump(sv_dat);*/
4970 nextchar(pRExC_state);
4972 goto capturing_parens;
4974 RExC_seen |= REG_SEEN_LOOKBEHIND;
4976 case '=': /* (?=...) */
4977 case '!': /* (?!...) */
4978 RExC_seen_zerolen++;
4979 if (*RExC_parse == ')') {
4980 ret=reg_node(pRExC_state, OPFAIL);
4981 nextchar(pRExC_state);
4984 case ':': /* (?:...) */
4985 case '>': /* (?>...) */
4987 case '$': /* (?$...) */
4988 case '@': /* (?@...) */
4989 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
4991 case '#': /* (?#...) */
4992 while (*RExC_parse && *RExC_parse != ')')
4994 if (*RExC_parse != ')')
4995 FAIL("Sequence (?#... not terminated");
4996 nextchar(pRExC_state);
4999 case '0' : /* (?0) */
5000 case 'R' : /* (?R) */
5001 if (*RExC_parse != ')')
5002 FAIL("Sequence (?R) not terminated");
5003 ret = reg_node(pRExC_state, GOSTART);
5004 nextchar(pRExC_state);
5007 { /* named and numeric backreferences */
5010 case '&': /* (?&NAME) */
5011 parse_start = RExC_parse - 1;
5013 SV *sv_dat = reg_scan_name(pRExC_state,
5014 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
5015 num = sv_dat ? *((I32 *)SvPVX(sv_dat)) : 0;
5017 goto gen_recurse_regop;
5020 if (!(RExC_parse[0] >= '1' && RExC_parse[0] <= '9')) {
5022 vFAIL("Illegal pattern");
5024 goto parse_recursion;
5026 case '-': /* (?-1) */
5027 if (!(RExC_parse[0] >= '1' && RExC_parse[0] <= '9')) {
5028 RExC_parse--; /* rewind to let it be handled later */
5032 case '1': case '2': case '3': case '4': /* (?1) */
5033 case '5': case '6': case '7': case '8': case '9':
5036 num = atoi(RExC_parse);
5037 parse_start = RExC_parse - 1; /* MJD */
5038 if (*RExC_parse == '-')
5040 while (isDIGIT(*RExC_parse))
5042 if (*RExC_parse!=')')
5043 vFAIL("Expecting close bracket");
5046 if ( paren == '-' ) {
5048 Diagram of capture buffer numbering.
5049 Top line is the normal capture buffer numbers
5050 Botton line is the negative indexing as from
5054 /(a(x)y)(a(b(c(?-2)d)e)f)(g(h))/
5058 num = RExC_npar + num;
5061 vFAIL("Reference to nonexistent group");
5063 } else if ( paren == '+' ) {
5064 num = RExC_npar + num - 1;
5067 ret = reganode(pRExC_state, GOSUB, num);
5069 if (num > (I32)RExC_rx->nparens) {
5071 vFAIL("Reference to nonexistent group");
5073 ARG2L_SET( ret, RExC_recurse_count++);
5075 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
5076 "Recurse #%"UVuf" to %"IVdf"\n", (UV)ARG(ret), (IV)ARG2L(ret)));
5080 RExC_seen |= REG_SEEN_RECURSE;
5081 Set_Node_Length(ret, 1 + regarglen[OP(ret)]); /* MJD */
5082 Set_Node_Offset(ret, parse_start); /* MJD */
5084 nextchar(pRExC_state);
5086 } /* named and numeric backreferences */
5089 case 'p': /* (?p...) */
5090 if (SIZE_ONLY && ckWARN2(WARN_DEPRECATED, WARN_REGEXP))
5091 vWARNdep(RExC_parse, "(?p{}) is deprecated - use (??{})");
5093 case '?': /* (??...) */
5095 if (*RExC_parse != '{')
5097 paren = *RExC_parse++;
5099 case '{': /* (?{...}) */
5104 char *s = RExC_parse;
5106 RExC_seen_zerolen++;
5107 RExC_seen |= REG_SEEN_EVAL;
5108 while (count && (c = *RExC_parse)) {
5119 if (*RExC_parse != ')') {
5121 vFAIL("Sequence (?{...}) not terminated or not {}-balanced");
5125 OP_4tree *sop, *rop;
5126 SV * const sv = newSVpvn(s, RExC_parse - 1 - s);
5129 Perl_save_re_context(aTHX);
5130 rop = sv_compile_2op(sv, &sop, "re", &pad);
5131 sop->op_private |= OPpREFCOUNTED;
5132 /* re_dup will OpREFCNT_inc */
5133 OpREFCNT_set(sop, 1);
5136 n = add_data(pRExC_state, 3, "nop");
5137 RExC_rxi->data->data[n] = (void*)rop;
5138 RExC_rxi->data->data[n+1] = (void*)sop;
5139 RExC_rxi->data->data[n+2] = (void*)pad;
5142 else { /* First pass */
5143 if (PL_reginterp_cnt < ++RExC_seen_evals
5145 /* No compiled RE interpolated, has runtime
5146 components ===> unsafe. */
5147 FAIL("Eval-group not allowed at runtime, use re 'eval'");
5148 if (PL_tainting && PL_tainted)
5149 FAIL("Eval-group in insecure regular expression");
5150 #if PERL_VERSION > 8
5151 if (IN_PERL_COMPILETIME)
5156 nextchar(pRExC_state);
5158 ret = reg_node(pRExC_state, LOGICAL);
5161 REGTAIL(pRExC_state, ret, reganode(pRExC_state, EVAL, n));
5162 /* deal with the length of this later - MJD */
5165 ret = reganode(pRExC_state, EVAL, n);
5166 Set_Node_Length(ret, RExC_parse - parse_start + 1);
5167 Set_Node_Offset(ret, parse_start);
5170 case '(': /* (?(?{...})...) and (?(?=...)...) */
5173 if (RExC_parse[0] == '?') { /* (?(?...)) */
5174 if (RExC_parse[1] == '=' || RExC_parse[1] == '!'
5175 || RExC_parse[1] == '<'
5176 || RExC_parse[1] == '{') { /* Lookahead or eval. */
5179 ret = reg_node(pRExC_state, LOGICAL);
5182 REGTAIL(pRExC_state, ret, reg(pRExC_state, 1, &flag,depth+1));
5186 else if ( RExC_parse[0] == '<' /* (?(<NAME>)...) */
5187 || RExC_parse[0] == '\'' ) /* (?('NAME')...) */
5189 char ch = RExC_parse[0] == '<' ? '>' : '\'';
5190 char *name_start= RExC_parse++;
5192 SV *sv_dat=reg_scan_name(pRExC_state,
5193 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
5194 if (RExC_parse == name_start || *RExC_parse != ch)
5195 vFAIL2("Sequence (?(%c... not terminated",
5196 (ch == '>' ? '<' : ch));
5199 num = add_data( pRExC_state, 1, "S" );
5200 RExC_rxi->data->data[num]=(void*)sv_dat;
5201 SvREFCNT_inc(sv_dat);
5203 ret = reganode(pRExC_state,NGROUPP,num);
5204 goto insert_if_check_paren;
5206 else if (RExC_parse[0] == 'D' &&
5207 RExC_parse[1] == 'E' &&
5208 RExC_parse[2] == 'F' &&
5209 RExC_parse[3] == 'I' &&
5210 RExC_parse[4] == 'N' &&
5211 RExC_parse[5] == 'E')
5213 ret = reganode(pRExC_state,DEFINEP,0);
5216 goto insert_if_check_paren;
5218 else if (RExC_parse[0] == 'R') {
5221 if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
5222 parno = atoi(RExC_parse++);
5223 while (isDIGIT(*RExC_parse))
5225 } else if (RExC_parse[0] == '&') {
5228 sv_dat = reg_scan_name(pRExC_state,
5229 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
5230 parno = sv_dat ? *((I32 *)SvPVX(sv_dat)) : 0;
5232 ret = reganode(pRExC_state,INSUBP,parno);
5233 goto insert_if_check_paren;
5235 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
5238 parno = atoi(RExC_parse++);
5240 while (isDIGIT(*RExC_parse))
5242 ret = reganode(pRExC_state, GROUPP, parno);
5244 insert_if_check_paren:
5245 if ((c = *nextchar(pRExC_state)) != ')')
5246 vFAIL("Switch condition not recognized");
5248 REGTAIL(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
5249 br = regbranch(pRExC_state, &flags, 1,depth+1);
5251 br = reganode(pRExC_state, LONGJMP, 0);
5253 REGTAIL(pRExC_state, br, reganode(pRExC_state, LONGJMP, 0));
5254 c = *nextchar(pRExC_state);
5259 vFAIL("(?(DEFINE)....) does not allow branches");
5260 lastbr = reganode(pRExC_state, IFTHEN, 0); /* Fake one for optimizer. */
5261 regbranch(pRExC_state, &flags, 1,depth+1);
5262 REGTAIL(pRExC_state, ret, lastbr);
5265 c = *nextchar(pRExC_state);
5270 vFAIL("Switch (?(condition)... contains too many branches");
5271 ender = reg_node(pRExC_state, TAIL);
5272 REGTAIL(pRExC_state, br, ender);
5274 REGTAIL(pRExC_state, lastbr, ender);
5275 REGTAIL(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
5278 REGTAIL(pRExC_state, ret, ender);
5282 vFAIL2("Unknown switch condition (?(%.2s", RExC_parse);
5286 RExC_parse--; /* for vFAIL to print correctly */
5287 vFAIL("Sequence (? incomplete");
5291 parse_flags: /* (?i) */
5292 while (*RExC_parse && strchr("iogcmsx", *RExC_parse)) {
5293 /* (?g), (?gc) and (?o) are useless here
5294 and must be globally applied -- japhy */
5296 if (*RExC_parse == 'o' || *RExC_parse == 'g') {
5297 if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
5298 const I32 wflagbit = *RExC_parse == 'o' ? WASTED_O : WASTED_G;
5299 if (! (wastedflags & wflagbit) ) {
5300 wastedflags |= wflagbit;
5303 "Useless (%s%c) - %suse /%c modifier",
5304 flagsp == &negflags ? "?-" : "?",
5306 flagsp == &negflags ? "don't " : "",
5312 else if (*RExC_parse == 'c') {
5313 if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
5314 if (! (wastedflags & WASTED_C) ) {
5315 wastedflags |= WASTED_GC;
5318 "Useless (%sc) - %suse /gc modifier",
5319 flagsp == &negflags ? "?-" : "?",
5320 flagsp == &negflags ? "don't " : ""
5325 else { pmflag(flagsp, *RExC_parse); }
5329 if (*RExC_parse == '-') {
5331 wastedflags = 0; /* reset so (?g-c) warns twice */
5335 RExC_flags |= posflags;
5336 RExC_flags &= ~negflags;
5337 if (*RExC_parse == ':') {
5343 if (*RExC_parse != ')') {
5345 vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
5347 nextchar(pRExC_state);
5357 ret = reganode(pRExC_state, OPEN, parno);
5360 RExC_nestroot = parno;
5361 if (RExC_seen & REG_SEEN_RECURSE) {
5362 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
5363 "Setting open paren #%"IVdf" to %d\n",
5364 (IV)parno, REG_NODE_NUM(ret)));
5365 RExC_open_parens[parno-1]= ret;
5368 Set_Node_Length(ret, 1); /* MJD */
5369 Set_Node_Offset(ret, RExC_parse); /* MJD */
5376 /* Pick up the branches, linking them together. */
5377 parse_start = RExC_parse; /* MJD */
5378 br = regbranch(pRExC_state, &flags, 1,depth+1);
5379 /* branch_len = (paren != 0); */
5383 if (*RExC_parse == '|') {
5384 if (!SIZE_ONLY && RExC_extralen) {
5385 reginsert(pRExC_state, BRANCHJ, br, depth+1);
5388 reginsert(pRExC_state, BRANCH, br, depth+1);
5389 Set_Node_Length(br, paren != 0);
5390 Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
5394 RExC_extralen += 1; /* For BRANCHJ-BRANCH. */
5396 else if (paren == ':') {
5397 *flagp |= flags&SIMPLE;
5399 if (is_open) { /* Starts with OPEN. */
5400 REGTAIL(pRExC_state, ret, br); /* OPEN -> first. */
5402 else if (paren != '?') /* Not Conditional */
5404 *flagp |= flags & (SPSTART | HASWIDTH);
5406 while (*RExC_parse == '|') {
5407 if (!SIZE_ONLY && RExC_extralen) {
5408 ender = reganode(pRExC_state, LONGJMP,0);
5409 REGTAIL(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
5412 RExC_extralen += 2; /* Account for LONGJMP. */
5413 nextchar(pRExC_state);
5414 br = regbranch(pRExC_state, &flags, 0, depth+1);
5418 REGTAIL(pRExC_state, lastbr, br); /* BRANCH -> BRANCH. */
5422 *flagp |= flags&SPSTART;
5425 if (have_branch || paren != ':') {
5426 /* Make a closing node, and hook it on the end. */
5429 ender = reg_node(pRExC_state, TAIL);
5432 ender = reganode(pRExC_state, CLOSE, parno);
5433 if (!SIZE_ONLY && RExC_seen & REG_SEEN_RECURSE) {
5434 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
5435 "Setting close paren #%"IVdf" to %d\n",
5436 (IV)parno, REG_NODE_NUM(ender)));
5437 RExC_close_parens[parno-1]= ender;
5438 if (RExC_nestroot == parno)
5441 Set_Node_Offset(ender,RExC_parse+1); /* MJD */
5442 Set_Node_Length(ender,1); /* MJD */
5448 *flagp &= ~HASWIDTH;
5451 ender = reg_node(pRExC_state, SUCCEED);
5454 ender = reg_node(pRExC_state, END);
5456 assert(!RExC_opend); /* there can only be one! */
5461 REGTAIL(pRExC_state, lastbr, ender);
5463 if (have_branch && !SIZE_ONLY) {
5465 RExC_seen |= REG_TOP_LEVEL_BRANCHES;
5467 /* Hook the tails of the branches to the closing node. */
5468 for (br = ret; br; br = regnext(br)) {
5469 const U8 op = PL_regkind[OP(br)];
5471 REGTAIL_STUDY(pRExC_state, NEXTOPER(br), ender);
5473 else if (op == BRANCHJ) {
5474 REGTAIL_STUDY(pRExC_state, NEXTOPER(NEXTOPER(br)), ender);
5482 static const char parens[] = "=!<,>";
5484 if (paren && (p = strchr(parens, paren))) {
5485 U8 node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
5486 int flag = (p - parens) > 1;
5489 node = SUSPEND, flag = 0;
5490 reginsert(pRExC_state, node,ret, depth+1);
5491 Set_Node_Cur_Length(ret);
5492 Set_Node_Offset(ret, parse_start + 1);
5494 REGTAIL_STUDY(pRExC_state, ret, reg_node(pRExC_state, TAIL));
5498 /* Check for proper termination. */
5500 RExC_flags = oregflags;
5501 if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
5502 RExC_parse = oregcomp_parse;
5503 vFAIL("Unmatched (");
5506 else if (!paren && RExC_parse < RExC_end) {
5507 if (*RExC_parse == ')') {
5509 vFAIL("Unmatched )");
5512 FAIL("Junk on end of regexp"); /* "Can't happen". */
5520 - regbranch - one alternative of an | operator
5522 * Implements the concatenation operator.
5525 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first, U32 depth)
5528 register regnode *ret;
5529 register regnode *chain = NULL;
5530 register regnode *latest;
5531 I32 flags = 0, c = 0;
5532 GET_RE_DEBUG_FLAGS_DECL;
5533 DEBUG_PARSE("brnc");
5537 if (!SIZE_ONLY && RExC_extralen)
5538 ret = reganode(pRExC_state, BRANCHJ,0);
5540 ret = reg_node(pRExC_state, BRANCH);
5541 Set_Node_Length(ret, 1);
5545 if (!first && SIZE_ONLY)
5546 RExC_extralen += 1; /* BRANCHJ */
5548 *flagp = WORST; /* Tentatively. */
5551 nextchar(pRExC_state);
5552 while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
5554 latest = regpiece(pRExC_state, &flags,depth+1);
5555 if (latest == NULL) {
5556 if (flags & TRYAGAIN)
5560 else if (ret == NULL)
5562 *flagp |= flags&HASWIDTH;
5563 if (chain == NULL) /* First piece. */
5564 *flagp |= flags&SPSTART;
5567 REGTAIL(pRExC_state, chain, latest);
5572 if (chain == NULL) { /* Loop ran zero times. */
5573 chain = reg_node(pRExC_state, NOTHING);
5578 *flagp |= flags&SIMPLE;
5585 - regpiece - something followed by possible [*+?]
5587 * Note that the branching code sequences used for ? and the general cases
5588 * of * and + are somewhat optimized: they use the same NOTHING node as
5589 * both the endmarker for their branch list and the body of the last branch.
5590 * It might seem that this node could be dispensed with entirely, but the
5591 * endmarker role is not redundant.
5594 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
5597 register regnode *ret;
5599 register char *next;
5601 const char * const origparse = RExC_parse;
5603 I32 max = REG_INFTY;
5605 const char *maxpos = NULL;
5606 GET_RE_DEBUG_FLAGS_DECL;
5607 DEBUG_PARSE("piec");
5609 ret = regatom(pRExC_state, &flags,depth+1);
5611 if (flags & TRYAGAIN)
5618 if (op == '{' && regcurly(RExC_parse)) {
5620 parse_start = RExC_parse; /* MJD */
5621 next = RExC_parse + 1;
5622 while (isDIGIT(*next) || *next == ',') {
5631 if (*next == '}') { /* got one */
5635 min = atoi(RExC_parse);
5639 maxpos = RExC_parse;
5641 if (!max && *maxpos != '0')
5642 max = REG_INFTY; /* meaning "infinity" */
5643 else if (max >= REG_INFTY)
5644 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
5646 nextchar(pRExC_state);
5649 if ((flags&SIMPLE)) {
5650 RExC_naughty += 2 + RExC_naughty / 2;
5651 reginsert(pRExC_state, CURLY, ret, depth+1);
5652 Set_Node_Offset(ret, parse_start+1); /* MJD */
5653 Set_Node_Cur_Length(ret);
5656 regnode * const w = reg_node(pRExC_state, WHILEM);
5659 REGTAIL(pRExC_state, ret, w);
5660 if (!SIZE_ONLY && RExC_extralen) {
5661 reginsert(pRExC_state, LONGJMP,ret, depth+1);
5662 reginsert(pRExC_state, NOTHING,ret, depth+1);
5663 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
5665 reginsert(pRExC_state, CURLYX,ret, depth+1);
5667 Set_Node_Offset(ret, parse_start+1);
5668 Set_Node_Length(ret,
5669 op == '{' ? (RExC_parse - parse_start) : 1);
5671 if (!SIZE_ONLY && RExC_extralen)
5672 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
5673 REGTAIL(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
5675 RExC_whilem_seen++, RExC_extralen += 3;
5676 RExC_naughty += 4 + RExC_naughty; /* compound interest */
5684 if (max && max < min)
5685 vFAIL("Can't do {n,m} with n > m");
5687 ARG1_SET(ret, (U16)min);
5688 ARG2_SET(ret, (U16)max);
5700 #if 0 /* Now runtime fix should be reliable. */
5702 /* if this is reinstated, don't forget to put this back into perldiag:
5704 =item Regexp *+ operand could be empty at {#} in regex m/%s/
5706 (F) The part of the regexp subject to either the * or + quantifier
5707 could match an empty string. The {#} shows in the regular
5708 expression about where the problem was discovered.
5712 if (!(flags&HASWIDTH) && op != '?')
5713 vFAIL("Regexp *+ operand could be empty");
5716 parse_start = RExC_parse;
5717 nextchar(pRExC_state);
5719 *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
5721 if (op == '*' && (flags&SIMPLE)) {
5722 reginsert(pRExC_state, STAR, ret, depth+1);
5726 else if (op == '*') {
5730 else if (op == '+' && (flags&SIMPLE)) {
5731 reginsert(pRExC_state, PLUS, ret, depth+1);
5735 else if (op == '+') {
5739 else if (op == '?') {
5744 if (!SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3 && ckWARN(WARN_REGEXP)) {
5746 "%.*s matches null string many times",
5747 (int)(RExC_parse >= origparse ? RExC_parse - origparse : 0),
5751 if (RExC_parse < RExC_end && *RExC_parse == '?') {
5752 nextchar(pRExC_state);
5753 reginsert(pRExC_state, MINMOD, ret, depth+1);
5754 REGTAIL(pRExC_state, ret, ret + NODE_STEP_REGNODE);
5756 #ifndef REG_ALLOW_MINMOD_SUSPEND
5759 if (RExC_parse < RExC_end && *RExC_parse == '+') {
5761 nextchar(pRExC_state);
5762 ender = reg_node(pRExC_state, SUCCEED);
5763 REGTAIL(pRExC_state, ret, ender);
5764 reginsert(pRExC_state, SUSPEND, ret, depth+1);
5766 ender = reg_node(pRExC_state, TAIL);
5767 REGTAIL(pRExC_state, ret, ender);
5771 if (RExC_parse < RExC_end && ISMULT2(RExC_parse)) {
5773 vFAIL("Nested quantifiers");
5780 /* reg_namedseq(pRExC_state,UVp)
5782 This is expected to be called by a parser routine that has
5783 recognized'\N' and needs to handle the rest. RExC_parse is
5784 expected to point at the first char following the N at the time
5787 If valuep is non-null then it is assumed that we are parsing inside
5788 of a charclass definition and the first codepoint in the resolved
5789 string is returned via *valuep and the routine will return NULL.
5790 In this mode if a multichar string is returned from the charnames
5791 handler a warning will be issued, and only the first char in the
5792 sequence will be examined. If the string returned is zero length
5793 then the value of *valuep is undefined and NON-NULL will
5794 be returned to indicate failure. (This will NOT be a valid pointer
5797 If value is null then it is assumed that we are parsing normal text
5798 and inserts a new EXACT node into the program containing the resolved
5799 string and returns a pointer to the new node. If the string is
5800 zerolength a NOTHING node is emitted.
5802 On success RExC_parse is set to the char following the endbrace.
5803 Parsing failures will generate a fatal errorvia vFAIL(...)
5805 NOTE: We cache all results from the charnames handler locally in
5806 the RExC_charnames hash (created on first use) to prevent a charnames
5807 handler from playing silly-buggers and returning a short string and
5808 then a long string for a given pattern. Since the regexp program
5809 size is calculated during an initial parse this would result
5810 in a buffer overrun so we cache to prevent the charname result from
5811 changing during the course of the parse.
5815 S_reg_namedseq(pTHX_ RExC_state_t *pRExC_state, UV *valuep)
5817 char * name; /* start of the content of the name */
5818 char * endbrace; /* endbrace following the name */
5821 STRLEN len; /* this has various purposes throughout the code */
5822 bool cached = 0; /* if this is true then we shouldn't refcount dev sv_str */
5823 regnode *ret = NULL;
5825 if (*RExC_parse != '{') {
5826 vFAIL("Missing braces on \\N{}");
5828 name = RExC_parse+1;
5829 endbrace = strchr(RExC_parse, '}');
5832 vFAIL("Missing right brace on \\N{}");
5834 RExC_parse = endbrace + 1;
5837 /* RExC_parse points at the beginning brace,
5838 endbrace points at the last */
5839 if ( name[0]=='U' && name[1]=='+' ) {
5840 /* its a "unicode hex" notation {U+89AB} */
5841 I32 fl = PERL_SCAN_ALLOW_UNDERSCORES
5842 | PERL_SCAN_DISALLOW_PREFIX
5843 | (SIZE_ONLY ? PERL_SCAN_SILENT_ILLDIGIT : 0);
5845 len = (STRLEN)(endbrace - name - 2);
5846 cp = grok_hex(name + 2, &len, &fl, NULL);
5847 if ( len != (STRLEN)(endbrace - name - 2) ) {
5856 sv_str= Perl_newSVpvf_nocontext("%c",(int)cp);
5858 /* fetch the charnames handler for this scope */
5859 HV * const table = GvHV(PL_hintgv);
5861 hv_fetchs(table, "charnames", FALSE) :
5863 SV *cv= cvp ? *cvp : NULL;
5866 /* create an SV with the name as argument */
5867 sv_name = newSVpvn(name, endbrace - name);
5869 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5870 vFAIL2("Constant(\\N{%s}) unknown: "
5871 "(possibly a missing \"use charnames ...\")",
5874 if (!cvp || !SvOK(*cvp)) { /* when $^H{charnames} = undef; */
5875 vFAIL2("Constant(\\N{%s}): "
5876 "$^H{charnames} is not defined",SvPVX(sv_name));
5881 if (!RExC_charnames) {
5882 /* make sure our cache is allocated */
5883 RExC_charnames = newHV();
5884 sv_2mortal((SV*)RExC_charnames);
5886 /* see if we have looked this one up before */
5887 he_str = hv_fetch_ent( RExC_charnames, sv_name, 0, 0 );
5889 sv_str = HeVAL(he_str);
5902 count= call_sv(cv, G_SCALAR);
5904 if (count == 1) { /* XXXX is this right? dmq */
5906 SvREFCNT_inc_simple_void(sv_str);
5914 if ( !sv_str || !SvOK(sv_str) ) {
5915 vFAIL2("Constant(\\N{%s}): Call to &{$^H{charnames}} "
5916 "did not return a defined value",SvPVX(sv_name));
5918 if (hv_store_ent( RExC_charnames, sv_name, sv_str, 0))
5923 char *p = SvPV(sv_str, len);
5926 if ( SvUTF8(sv_str) ) {
5927 *valuep = utf8_to_uvchr((U8*)p, &numlen);
5931 We have to turn on utf8 for high bit chars otherwise
5932 we get failures with
5934 "ss" =~ /[\N{LATIN SMALL LETTER SHARP S}]/i
5935 "SS" =~ /[\N{LATIN SMALL LETTER SHARP S}]/i
5937 This is different from what \x{} would do with the same
5938 codepoint, where the condition is > 0xFF.
5945 /* warn if we havent used the whole string? */
5947 if (numlen<len && SIZE_ONLY && ckWARN(WARN_REGEXP)) {
5949 "Ignoring excess chars from \\N{%s} in character class",
5953 } else if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
5955 "Ignoring zero length \\N{%s} in character class",
5960 SvREFCNT_dec(sv_name);
5962 SvREFCNT_dec(sv_str);
5963 return len ? NULL : (regnode *)&len;
5964 } else if(SvCUR(sv_str)) {
5969 char * parse_start = name-3; /* needed for the offsets */
5970 GET_RE_DEBUG_FLAGS_DECL; /* needed for the offsets */
5972 ret = reg_node(pRExC_state,
5973 (U8)(FOLD ? (LOC ? EXACTFL : EXACTF) : EXACT));
5976 if ( RExC_utf8 && !SvUTF8(sv_str) ) {
5977 sv_utf8_upgrade(sv_str);
5978 } else if ( !RExC_utf8 && SvUTF8(sv_str) ) {
5982 p = SvPV(sv_str, len);
5984 /* len is the length written, charlen is the size the char read */
5985 for ( len = 0; p < pend; p += charlen ) {
5987 UV uvc = utf8_to_uvchr((U8*)p, &charlen);
5989 STRLEN foldlen,numlen;
5990 U8 tmpbuf[UTF8_MAXBYTES_CASE+1], *foldbuf;
5991 uvc = toFOLD_uni(uvc, tmpbuf, &foldlen);
5992 /* Emit all the Unicode characters. */
5994 for (foldbuf = tmpbuf;
5998 uvc = utf8_to_uvchr(foldbuf, &numlen);
6000 const STRLEN unilen = reguni(pRExC_state, uvc, s);
6003 /* In EBCDIC the numlen
6004 * and unilen can differ. */
6006 if (numlen >= foldlen)
6010 break; /* "Can't happen." */
6013 const STRLEN unilen = reguni(pRExC_state, uvc, s);
6025 RExC_size += STR_SZ(len);
6028 RExC_emit += STR_SZ(len);
6030 Set_Node_Cur_Length(ret); /* MJD */
6032 nextchar(pRExC_state);
6034 ret = reg_node(pRExC_state,NOTHING);
6037 SvREFCNT_dec(sv_str);
6040 SvREFCNT_dec(sv_name);
6050 * It returns the code point in utf8 for the value in *encp.
6051 * value: a code value in the source encoding
6052 * encp: a pointer to an Encode object
6054 * If the result from Encode is not a single character,
6055 * it returns U+FFFD (Replacement character) and sets *encp to NULL.
6058 S_reg_recode(pTHX_ const char value, SV **encp)
6061 SV * const sv = sv_2mortal(newSVpvn(&value, numlen));
6062 const char * const s = encp && *encp ? sv_recode_to_utf8(sv, *encp)
6064 const STRLEN newlen = SvCUR(sv);
6065 UV uv = UNICODE_REPLACEMENT;
6069 ? utf8n_to_uvchr((U8*)s, newlen, &numlen, UTF8_ALLOW_DEFAULT)
6072 if (!newlen || numlen != newlen) {
6073 uv = UNICODE_REPLACEMENT;
6082 - regatom - the lowest level
6084 * Optimization: gobbles an entire sequence of ordinary characters so that
6085 * it can turn them into a single node, which is smaller to store and
6086 * faster to run. Backslashed characters are exceptions, each becoming a
6087 * separate node; the code is simpler that way and it's not worth fixing.
6089 * [Yes, it is worth fixing, some scripts can run twice the speed.]
6090 * [It looks like its ok, as in S_study_chunk we merge adjacent EXACT nodes]
6093 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
6096 register regnode *ret = NULL;
6098 char *parse_start = RExC_parse;
6099 GET_RE_DEBUG_FLAGS_DECL;
6100 DEBUG_PARSE("atom");
6101 *flagp = WORST; /* Tentatively. */
6104 switch (*RExC_parse) {
6106 RExC_seen_zerolen++;
6107 nextchar(pRExC_state);
6108 if (RExC_flags & RXf_PMf_MULTILINE)
6109 ret = reg_node(pRExC_state, MBOL);
6110 else if (RExC_flags & RXf_PMf_SINGLELINE)
6111 ret = reg_node(pRExC_state, SBOL);
6113 ret = reg_node(pRExC_state, BOL);
6114 Set_Node_Length(ret, 1); /* MJD */
6117 nextchar(pRExC_state);
6119 RExC_seen_zerolen++;
6120 if (RExC_flags & RXf_PMf_MULTILINE)
6121 ret = reg_node(pRExC_state, MEOL);
6122 else if (RExC_flags & RXf_PMf_SINGLELINE)
6123 ret = reg_node(pRExC_state, SEOL);
6125 ret = reg_node(pRExC_state, EOL);
6126 Set_Node_Length(ret, 1); /* MJD */
6129 nextchar(pRExC_state);
6130 if (RExC_flags & RXf_PMf_SINGLELINE)
6131 ret = reg_node(pRExC_state, SANY);
6133 ret = reg_node(pRExC_state, REG_ANY);
6134 *flagp |= HASWIDTH|SIMPLE;
6136 Set_Node_Length(ret, 1); /* MJD */
6140 char * const oregcomp_parse = ++RExC_parse;
6141 ret = regclass(pRExC_state,depth+1);
6142 if (*RExC_parse != ']') {
6143 RExC_parse = oregcomp_parse;
6144 vFAIL("Unmatched [");
6146 nextchar(pRExC_state);
6147 *flagp |= HASWIDTH|SIMPLE;
6148 Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
6152 nextchar(pRExC_state);
6153 ret = reg(pRExC_state, 1, &flags,depth+1);
6155 if (flags & TRYAGAIN) {
6156 if (RExC_parse == RExC_end) {
6157 /* Make parent create an empty node if needed. */
6165 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
6169 if (flags & TRYAGAIN) {
6173 vFAIL("Internal urp");
6174 /* Supposed to be caught earlier. */
6177 if (!regcurly(RExC_parse)) {
6186 vFAIL("Quantifier follows nothing");
6189 switch (*++RExC_parse) {
6191 RExC_seen_zerolen++;
6192 ret = reg_node(pRExC_state, SBOL);
6194 nextchar(pRExC_state);
6195 Set_Node_Length(ret, 2); /* MJD */
6198 ret = reg_node(pRExC_state, GPOS);
6199 RExC_seen |= REG_SEEN_GPOS;
6201 nextchar(pRExC_state);
6202 Set_Node_Length(ret, 2); /* MJD */
6205 ret = reg_node(pRExC_state, SEOL);
6207 RExC_seen_zerolen++; /* Do not optimize RE away */
6208 nextchar(pRExC_state);
6211 ret = reg_node(pRExC_state, EOS);
6213 RExC_seen_zerolen++; /* Do not optimize RE away */
6214 nextchar(pRExC_state);
6215 Set_Node_Length(ret, 2); /* MJD */
6218 ret = reg_node(pRExC_state, CANY);
6219 RExC_seen |= REG_SEEN_CANY;
6220 *flagp |= HASWIDTH|SIMPLE;
6221 nextchar(pRExC_state);
6222 Set_Node_Length(ret, 2); /* MJD */
6225 ret = reg_node(pRExC_state, CLUMP);
6227 nextchar(pRExC_state);
6228 Set_Node_Length(ret, 2); /* MJD */
6231 ret = reg_node(pRExC_state, (U8)(LOC ? ALNUML : ALNUM));
6232 *flagp |= HASWIDTH|SIMPLE;
6233 nextchar(pRExC_state);
6234 Set_Node_Length(ret, 2); /* MJD */
6237 ret = reg_node(pRExC_state, (U8)(LOC ? NALNUML : NALNUM));
6238 *flagp |= HASWIDTH|SIMPLE;
6239 nextchar(pRExC_state);
6240 Set_Node_Length(ret, 2); /* MJD */
6243 RExC_seen_zerolen++;
6244 RExC_seen |= REG_SEEN_LOOKBEHIND;
6245 ret = reg_node(pRExC_state, (U8)(LOC ? BOUNDL : BOUND));
6247 nextchar(pRExC_state);
6248 Set_Node_Length(ret, 2); /* MJD */
6251 RExC_seen_zerolen++;
6252 RExC_seen |= REG_SEEN_LOOKBEHIND;
6253 ret = reg_node(pRExC_state, (U8)(LOC ? NBOUNDL : NBOUND));
6255 nextchar(pRExC_state);
6256 Set_Node_Length(ret, 2); /* MJD */
6259 ret = reg_node(pRExC_state, (U8)(LOC ? SPACEL : SPACE));
6260 *flagp |= HASWIDTH|SIMPLE;
6261 nextchar(pRExC_state);
6262 Set_Node_Length(ret, 2); /* MJD */
6265 ret = reg_node(pRExC_state, (U8)(LOC ? NSPACEL : NSPACE));
6266 *flagp |= HASWIDTH|SIMPLE;
6267 nextchar(pRExC_state);
6268 Set_Node_Length(ret, 2); /* MJD */
6271 ret = reg_node(pRExC_state, DIGIT);
6272 *flagp |= HASWIDTH|SIMPLE;
6273 nextchar(pRExC_state);
6274 Set_Node_Length(ret, 2); /* MJD */
6277 ret = reg_node(pRExC_state, NDIGIT);
6278 *flagp |= HASWIDTH|SIMPLE;
6279 nextchar(pRExC_state);
6280 Set_Node_Length(ret, 2); /* MJD */
6285 char* const oldregxend = RExC_end;
6286 char* parse_start = RExC_parse - 2;
6288 if (RExC_parse[1] == '{') {
6289 /* a lovely hack--pretend we saw [\pX] instead */
6290 RExC_end = strchr(RExC_parse, '}');
6292 const U8 c = (U8)*RExC_parse;
6294 RExC_end = oldregxend;
6295 vFAIL2("Missing right brace on \\%c{}", c);
6300 RExC_end = RExC_parse + 2;
6301 if (RExC_end > oldregxend)
6302 RExC_end = oldregxend;
6306 ret = regclass(pRExC_state,depth+1);
6308 RExC_end = oldregxend;
6311 Set_Node_Offset(ret, parse_start + 2);
6312 Set_Node_Cur_Length(ret);
6313 nextchar(pRExC_state);
6314 *flagp |= HASWIDTH|SIMPLE;
6318 /* Handle \N{NAME} here and not below because it can be
6319 multicharacter. join_exact() will join them up later on.
6320 Also this makes sure that things like /\N{BLAH}+/ and
6321 \N{BLAH} being multi char Just Happen. dmq*/
6323 ret= reg_namedseq(pRExC_state, NULL);
6325 case 'k': /* Handle \k<NAME> and \k'NAME' */
6327 char ch= RExC_parse[1];
6328 if (ch != '<' && ch != '\'') {
6330 vWARN( RExC_parse + 1,
6331 "Possible broken named back reference treated as literal k");
6335 char* name_start = (RExC_parse += 2);
6337 SV *sv_dat = reg_scan_name(pRExC_state,
6338 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
6339 ch= (ch == '<') ? '>' : '\'';
6341 if (RExC_parse == name_start || *RExC_parse != ch)
6342 vFAIL2("Sequence \\k%c... not terminated",
6343 (ch == '>' ? '<' : ch));
6346 ret = reganode(pRExC_state,
6347 (U8)(FOLD ? (LOC ? NREFFL : NREFF) : NREF),
6353 num = add_data( pRExC_state, 1, "S" );
6355 RExC_rxi->data->data[num]=(void*)sv_dat;
6356 SvREFCNT_inc(sv_dat);
6358 /* override incorrect value set in reganode MJD */
6359 Set_Node_Offset(ret, parse_start+1);
6360 Set_Node_Cur_Length(ret); /* MJD */
6361 nextchar(pRExC_state);
6377 case '1': case '2': case '3': case '4':
6378 case '5': case '6': case '7': case '8': case '9':
6381 bool isg = *RExC_parse == 'g';
6386 if (*RExC_parse == '{') {
6390 if (*RExC_parse == '-') {
6395 num = atoi(RExC_parse);
6397 num = RExC_npar - num;
6399 vFAIL("Reference to nonexistent or unclosed group");
6401 if (!isg && num > 9 && num >= RExC_npar)
6404 char * const parse_start = RExC_parse - 1; /* MJD */
6405 while (isDIGIT(*RExC_parse))
6408 if (*RExC_parse != '}')
6409 vFAIL("Unterminated \\g{...} pattern");
6413 if (num > (I32)RExC_rx->nparens)
6414 vFAIL("Reference to nonexistent group");
6417 ret = reganode(pRExC_state,
6418 (U8)(FOLD ? (LOC ? REFFL : REFF) : REF),
6422 /* override incorrect value set in reganode MJD */
6423 Set_Node_Offset(ret, parse_start+1);
6424 Set_Node_Cur_Length(ret); /* MJD */
6426 nextchar(pRExC_state);
6431 if (RExC_parse >= RExC_end)
6432 FAIL("Trailing \\");
6435 /* Do not generate "unrecognized" warnings here, we fall
6436 back into the quick-grab loop below */
6443 if (RExC_flags & RXf_PMf_EXTENDED) {
6444 while (RExC_parse < RExC_end && *RExC_parse != '\n')
6446 if (RExC_parse < RExC_end)
6452 register STRLEN len;
6457 U8 tmpbuf[UTF8_MAXBYTES_CASE+1], *foldbuf;
6459 parse_start = RExC_parse - 1;
6465 ret = reg_node(pRExC_state,
6466 (U8)(FOLD ? (LOC ? EXACTFL : EXACTF) : EXACT));
6468 for (len = 0, p = RExC_parse - 1;
6469 len < 127 && p < RExC_end;
6472 char * const oldp = p;
6474 if (RExC_flags & RXf_PMf_EXTENDED)
6475 p = regwhite(p, RExC_end);
6526 ender = ASCII_TO_NATIVE('\033');
6530 ender = ASCII_TO_NATIVE('\007');
6535 char* const e = strchr(p, '}');
6539 vFAIL("Missing right brace on \\x{}");
6542 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
6543 | PERL_SCAN_DISALLOW_PREFIX;
6544 STRLEN numlen = e - p - 1;
6545 ender = grok_hex(p + 1, &numlen, &flags, NULL);
6552 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
6554 ender = grok_hex(p, &numlen, &flags, NULL);
6557 if (PL_encoding && ender < 0x100)
6558 goto recode_encoding;
6562 ender = UCHARAT(p++);
6563 ender = toCTRL(ender);
6565 case '0': case '1': case '2': case '3':case '4':
6566 case '5': case '6': case '7': case '8':case '9':
6568 (isDIGIT(p[1]) && atoi(p) >= RExC_npar) ) {
6571 ender = grok_oct(p, &numlen, &flags, NULL);
6578 if (PL_encoding && ender < 0x100)
6579 goto recode_encoding;
6583 SV* enc = PL_encoding;
6584 ender = reg_recode((const char)(U8)ender, &enc);
6585 if (!enc && SIZE_ONLY && ckWARN(WARN_REGEXP))
6586 vWARN(p, "Invalid escape in the specified encoding");
6592 FAIL("Trailing \\");
6595 if (!SIZE_ONLY&& isALPHA(*p) && ckWARN(WARN_REGEXP))
6596 vWARN2(p + 1, "Unrecognized escape \\%c passed through", UCHARAT(p));
6597 goto normal_default;
6602 if (UTF8_IS_START(*p) && UTF) {
6604 ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
6605 &numlen, UTF8_ALLOW_DEFAULT);
6612 if (RExC_flags & RXf_PMf_EXTENDED)
6613 p = regwhite(p, RExC_end);
6615 /* Prime the casefolded buffer. */
6616 ender = toFOLD_uni(ender, tmpbuf, &foldlen);
6618 if (ISMULT2(p)) { /* Back off on ?+*. */
6623 /* Emit all the Unicode characters. */
6625 for (foldbuf = tmpbuf;
6627 foldlen -= numlen) {
6628 ender = utf8_to_uvchr(foldbuf, &numlen);
6630 const STRLEN unilen = reguni(pRExC_state, ender, s);
6633 /* In EBCDIC the numlen
6634 * and unilen can differ. */
6636 if (numlen >= foldlen)
6640 break; /* "Can't happen." */
6644 const STRLEN unilen = reguni(pRExC_state, ender, s);
6653 REGC((char)ender, s++);
6659 /* Emit all the Unicode characters. */
6661 for (foldbuf = tmpbuf;
6663 foldlen -= numlen) {
6664 ender = utf8_to_uvchr(foldbuf, &numlen);
6666 const STRLEN unilen = reguni(pRExC_state, ender, s);
6669 /* In EBCDIC the numlen
6670 * and unilen can differ. */
6672 if (numlen >= foldlen)
6680 const STRLEN unilen = reguni(pRExC_state, ender, s);
6689 REGC((char)ender, s++);
6693 Set_Node_Cur_Length(ret); /* MJD */
6694 nextchar(pRExC_state);
6696 /* len is STRLEN which is unsigned, need to copy to signed */
6699 vFAIL("Internal disaster");
6703 if (len == 1 && UNI_IS_INVARIANT(ender))
6707 RExC_size += STR_SZ(len);
6710 RExC_emit += STR_SZ(len);
6720 S_regwhite(char *p, const char *e)
6725 else if (*p == '#') {
6728 } while (p < e && *p != '\n');
6736 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
6737 Character classes ([:foo:]) can also be negated ([:^foo:]).
6738 Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
6739 Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
6740 but trigger failures because they are currently unimplemented. */
6742 #define POSIXCC_DONE(c) ((c) == ':')
6743 #define POSIXCC_NOTYET(c) ((c) == '=' || (c) == '.')
6744 #define POSIXCC(c) (POSIXCC_DONE(c) || POSIXCC_NOTYET(c))
6747 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value)
6750 I32 namedclass = OOB_NAMEDCLASS;
6752 if (value == '[' && RExC_parse + 1 < RExC_end &&
6753 /* I smell either [: or [= or [. -- POSIX has been here, right? */
6754 POSIXCC(UCHARAT(RExC_parse))) {
6755 const char c = UCHARAT(RExC_parse);
6756 char* const s = RExC_parse++;
6758 while (RExC_parse < RExC_end && UCHARAT(RExC_parse) != c)
6760 if (RExC_parse == RExC_end)
6761 /* Grandfather lone [:, [=, [. */
6764 const char* const t = RExC_parse++; /* skip over the c */
6767 if (UCHARAT(RExC_parse) == ']') {
6768 const char *posixcc = s + 1;
6769 RExC_parse++; /* skip over the ending ] */
6772 const I32 complement = *posixcc == '^' ? *posixcc++ : 0;
6773 const I32 skip = t - posixcc;
6775 /* Initially switch on the length of the name. */
6778 if (memEQ(posixcc, "word", 4)) /* this is not POSIX, this is the Perl \w */
6779 namedclass = complement ? ANYOF_NALNUM : ANYOF_ALNUM;
6782 /* Names all of length 5. */
6783 /* alnum alpha ascii blank cntrl digit graph lower
6784 print punct space upper */
6785 /* Offset 4 gives the best switch position. */
6786 switch (posixcc[4]) {
6788 if (memEQ(posixcc, "alph", 4)) /* alpha */
6789 namedclass = complement ? ANYOF_NALPHA : ANYOF_ALPHA;
6792 if (memEQ(posixcc, "spac", 4)) /* space */
6793 namedclass = complement ? ANYOF_NPSXSPC : ANYOF_PSXSPC;
6796 if (memEQ(posixcc, "grap", 4)) /* graph */
6797 namedclass = complement ? ANYOF_NGRAPH : ANYOF_GRAPH;
6800 if (memEQ(posixcc, "asci", 4)) /* ascii */
6801 namedclass = complement ? ANYOF_NASCII : ANYOF_ASCII;
6804 if (memEQ(posixcc, "blan", 4)) /* blank */
6805 namedclass = complement ? ANYOF_NBLANK : ANYOF_BLANK;
6808 if (memEQ(posixcc, "cntr", 4)) /* cntrl */
6809 namedclass = complement ? ANYOF_NCNTRL : ANYOF_CNTRL;
6812 if (memEQ(posixcc, "alnu", 4)) /* alnum */
6813 namedclass = complement ? ANYOF_NALNUMC : ANYOF_ALNUMC;
6816 if (memEQ(posixcc, "lowe", 4)) /* lower */
6817 namedclass = complement ? ANYOF_NLOWER : ANYOF_LOWER;
6818 else if (memEQ(posixcc, "uppe", 4)) /* upper */
6819 namedclass = complement ? ANYOF_NUPPER : ANYOF_UPPER;
6822 if (memEQ(posixcc, "digi", 4)) /* digit */
6823 namedclass = complement ? ANYOF_NDIGIT : ANYOF_DIGIT;
6824 else if (memEQ(posixcc, "prin", 4)) /* print */
6825 namedclass = complement ? ANYOF_NPRINT : ANYOF_PRINT;
6826 else if (memEQ(posixcc, "punc", 4)) /* punct */
6827 namedclass = complement ? ANYOF_NPUNCT : ANYOF_PUNCT;
6832 if (memEQ(posixcc, "xdigit", 6))
6833 namedclass = complement ? ANYOF_NXDIGIT : ANYOF_XDIGIT;
6837 if (namedclass == OOB_NAMEDCLASS)
6838 Simple_vFAIL3("POSIX class [:%.*s:] unknown",
6840 assert (posixcc[skip] == ':');
6841 assert (posixcc[skip+1] == ']');
6842 } else if (!SIZE_ONLY) {
6843 /* [[=foo=]] and [[.foo.]] are still future. */
6845 /* adjust RExC_parse so the warning shows after
6847 while (UCHARAT(RExC_parse) && UCHARAT(RExC_parse) != ']')
6849 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
6852 /* Maternal grandfather:
6853 * "[:" ending in ":" but not in ":]" */
6863 S_checkposixcc(pTHX_ RExC_state_t *pRExC_state)
6866 if (POSIXCC(UCHARAT(RExC_parse))) {
6867 const char *s = RExC_parse;
6868 const char c = *s++;
6872 if (*s && c == *s && s[1] == ']') {
6873 if (ckWARN(WARN_REGEXP))
6875 "POSIX syntax [%c %c] belongs inside character classes",
6878 /* [[=foo=]] and [[.foo.]] are still future. */
6879 if (POSIXCC_NOTYET(c)) {
6880 /* adjust RExC_parse so the error shows after
6882 while (UCHARAT(RExC_parse) && UCHARAT(RExC_parse++) != ']')
6884 Simple_vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
6892 parse a class specification and produce either an ANYOF node that
6893 matches the pattern. If the pattern matches a single char only and
6894 that char is < 256 then we produce an EXACT node instead.
6897 S_regclass(pTHX_ RExC_state_t *pRExC_state, U32 depth)
6900 register UV value = 0;
6901 register UV nextvalue;
6902 register IV prevvalue = OOB_UNICODE;
6903 register IV range = 0;
6904 register regnode *ret;
6907 char *rangebegin = NULL;
6908 bool need_class = 0;
6911 bool optimize_invert = TRUE;
6912 AV* unicode_alternate = NULL;
6914 UV literal_endpoint = 0;
6916 UV stored = 0; /* number of chars stored in the class */
6918 regnode * const orig_emit = RExC_emit; /* Save the original RExC_emit in
6919 case we need to change the emitted regop to an EXACT. */
6920 const char * orig_parse = RExC_parse;
6921 GET_RE_DEBUG_FLAGS_DECL;
6923 PERL_UNUSED_ARG(depth);
6926 DEBUG_PARSE("clas");
6928 /* Assume we are going to generate an ANYOF node. */
6929 ret = reganode(pRExC_state, ANYOF, 0);
6932 ANYOF_FLAGS(ret) = 0;
6934 if (UCHARAT(RExC_parse) == '^') { /* Complement of range. */
6938 ANYOF_FLAGS(ret) |= ANYOF_INVERT;
6942 RExC_size += ANYOF_SKIP;
6943 listsv = &PL_sv_undef; /* For code scanners: listsv always non-NULL. */
6946 RExC_emit += ANYOF_SKIP;
6948 ANYOF_FLAGS(ret) |= ANYOF_FOLD;
6950 ANYOF_FLAGS(ret) |= ANYOF_LOCALE;
6951 ANYOF_BITMAP_ZERO(ret);
6952 listsv = newSVpvs("# comment\n");
6955 nextvalue = RExC_parse < RExC_end ? UCHARAT(RExC_parse) : 0;
6957 if (!SIZE_ONLY && POSIXCC(nextvalue))
6958 checkposixcc(pRExC_state);
6960 /* allow 1st char to be ] (allowing it to be - is dealt with later) */
6961 if (UCHARAT(RExC_parse) == ']')
6965 while (RExC_parse < RExC_end && UCHARAT(RExC_parse) != ']') {
6969 namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
6972 rangebegin = RExC_parse;
6974 value = utf8n_to_uvchr((U8*)RExC_parse,
6975 RExC_end - RExC_parse,
6976 &numlen, UTF8_ALLOW_DEFAULT);
6977 RExC_parse += numlen;
6980 value = UCHARAT(RExC_parse++);
6982 nextvalue = RExC_parse < RExC_end ? UCHARAT(RExC_parse) : 0;
6983 if (value == '[' && POSIXCC(nextvalue))
6984 namedclass = regpposixcc(pRExC_state, value);
6985 else if (value == '\\') {
6987 value = utf8n_to_uvchr((U8*)RExC_parse,
6988 RExC_end - RExC_parse,
6989 &numlen, UTF8_ALLOW_DEFAULT);
6990 RExC_parse += numlen;
6993 value = UCHARAT(RExC_parse++);
6994 /* Some compilers cannot handle switching on 64-bit integer
6995 * values, therefore value cannot be an UV. Yes, this will
6996 * be a problem later if we want switch on Unicode.
6997 * A similar issue a little bit later when switching on
6998 * namedclass. --jhi */
6999 switch ((I32)value) {
7000 case 'w': namedclass = ANYOF_ALNUM; break;
7001 case 'W': namedclass = ANYOF_NALNUM; break;
7002 case 's': namedclass = ANYOF_SPACE; break;
7003 case 'S': namedclass = ANYOF_NSPACE; break;
7004 case 'd': namedclass = ANYOF_DIGIT; break;
7005 case 'D': namedclass = ANYOF_NDIGIT; break;
7006 case 'N': /* Handle \N{NAME} in class */
7008 /* We only pay attention to the first char of
7009 multichar strings being returned. I kinda wonder
7010 if this makes sense as it does change the behaviour
7011 from earlier versions, OTOH that behaviour was broken
7013 UV v; /* value is register so we cant & it /grrr */
7014 if (reg_namedseq(pRExC_state, &v)) {
7024 if (RExC_parse >= RExC_end)
7025 vFAIL2("Empty \\%c{}", (U8)value);
7026 if (*RExC_parse == '{') {
7027 const U8 c = (U8)value;
7028 e = strchr(RExC_parse++, '}');
7030 vFAIL2("Missing right brace on \\%c{}", c);
7031 while (isSPACE(UCHARAT(RExC_parse)))
7033 if (e == RExC_parse)
7034 vFAIL2("Empty \\%c{}", c);
7036 while (isSPACE(UCHARAT(RExC_parse + n - 1)))
7044 if (UCHARAT(RExC_parse) == '^') {
7047 value = value == 'p' ? 'P' : 'p'; /* toggle */
7048 while (isSPACE(UCHARAT(RExC_parse))) {
7053 Perl_sv_catpvf(aTHX_ listsv, "%cutf8::%.*s\n",
7054 (value=='p' ? '+' : '!'), (int)n, RExC_parse);
7057 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
7058 namedclass = ANYOF_MAX; /* no official name, but it's named */
7061 case 'n': value = '\n'; break;
7062 case 'r': value = '\r'; break;
7063 case 't': value = '\t'; break;
7064 case 'f': value = '\f'; break;
7065 case 'b': value = '\b'; break;
7066 case 'e': value = ASCII_TO_NATIVE('\033');break;
7067 case 'a': value = ASCII_TO_NATIVE('\007');break;
7069 if (*RExC_parse == '{') {
7070 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
7071 | PERL_SCAN_DISALLOW_PREFIX;
7072 char * const e = strchr(RExC_parse++, '}');
7074 vFAIL("Missing right brace on \\x{}");
7076 numlen = e - RExC_parse;
7077 value = grok_hex(RExC_parse, &numlen, &flags, NULL);
7081 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
7083 value = grok_hex(RExC_parse, &numlen, &flags, NULL);
7084 RExC_parse += numlen;
7086 if (PL_encoding && value < 0x100)
7087 goto recode_encoding;
7090 value = UCHARAT(RExC_parse++);
7091 value = toCTRL(value);
7093 case '0': case '1': case '2': case '3': case '4':
7094 case '5': case '6': case '7': case '8': case '9':
7098 value = grok_oct(--RExC_parse, &numlen, &flags, NULL);
7099 RExC_parse += numlen;
7100 if (PL_encoding && value < 0x100)
7101 goto recode_encoding;
7106 SV* enc = PL_encoding;
7107 value = reg_recode((const char)(U8)value, &enc);
7108 if (!enc && SIZE_ONLY && ckWARN(WARN_REGEXP))
7110 "Invalid escape in the specified encoding");
7114 if (!SIZE_ONLY && isALPHA(value) && ckWARN(WARN_REGEXP))
7116 "Unrecognized escape \\%c in character class passed through",
7120 } /* end of \blah */
7126 if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
7128 if (!SIZE_ONLY && !need_class)
7129 ANYOF_CLASS_ZERO(ret);
7133 /* a bad range like a-\d, a-[:digit:] ? */
7136 if (ckWARN(WARN_REGEXP)) {
7138 RExC_parse >= rangebegin ?
7139 RExC_parse - rangebegin : 0;
7141 "False [] range \"%*.*s\"",
7144 if (prevvalue < 256) {
7145 ANYOF_BITMAP_SET(ret, prevvalue);
7146 ANYOF_BITMAP_SET(ret, '-');
7149 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
7150 Perl_sv_catpvf(aTHX_ listsv,
7151 "%04"UVxf"\n%04"UVxf"\n", (UV)prevvalue, (UV) '-');
7155 range = 0; /* this was not a true range */
7159 const char *what = NULL;
7162 if (namedclass > OOB_NAMEDCLASS)
7163 optimize_invert = FALSE;
7164 /* Possible truncation here but in some 64-bit environments
7165 * the compiler gets heartburn about switch on 64-bit values.
7166 * A similar issue a little earlier when switching on value.
7168 switch ((I32)namedclass) {
7171 ANYOF_CLASS_SET(ret, ANYOF_ALNUM);
7173 for (value = 0; value < 256; value++)
7175 ANYOF_BITMAP_SET(ret, value);
7182 ANYOF_CLASS_SET(ret, ANYOF_NALNUM);
7184 for (value = 0; value < 256; value++)
7185 if (!isALNUM(value))
7186 ANYOF_BITMAP_SET(ret, value);
7193 ANYOF_CLASS_SET(ret, ANYOF_ALNUMC);
7195 for (value = 0; value < 256; value++)
7196 if (isALNUMC(value))
7197 ANYOF_BITMAP_SET(ret, value);
7204 ANYOF_CLASS_SET(ret, ANYOF_NALNUMC);
7206 for (value = 0; value < 256; value++)
7207 if (!isALNUMC(value))
7208 ANYOF_BITMAP_SET(ret, value);
7215 ANYOF_CLASS_SET(ret, ANYOF_ALPHA);
7217 for (value = 0; value < 256; value++)
7219 ANYOF_BITMAP_SET(ret, value);
7226 ANYOF_CLASS_SET(ret, ANYOF_NALPHA);
7228 for (value = 0; value < 256; value++)
7229 if (!isALPHA(value))
7230 ANYOF_BITMAP_SET(ret, value);
7237 ANYOF_CLASS_SET(ret, ANYOF_ASCII);
7240 for (value = 0; value < 128; value++)
7241 ANYOF_BITMAP_SET(ret, value);
7243 for (value = 0; value < 256; value++) {
7245 ANYOF_BITMAP_SET(ret, value);
7254 ANYOF_CLASS_SET(ret, ANYOF_NASCII);
7257 for (value = 128; value < 256; value++)
7258 ANYOF_BITMAP_SET(ret, value);
7260 for (value = 0; value < 256; value++) {
7261 if (!isASCII(value))
7262 ANYOF_BITMAP_SET(ret, value);
7271 ANYOF_CLASS_SET(ret, ANYOF_BLANK);
7273 for (value = 0; value < 256; value++)
7275 ANYOF_BITMAP_SET(ret, value);
7282 ANYOF_CLASS_SET(ret, ANYOF_NBLANK);
7284 for (value = 0; value < 256; value++)
7285 if (!isBLANK(value))
7286 ANYOF_BITMAP_SET(ret, value);
7293 ANYOF_CLASS_SET(ret, ANYOF_CNTRL);
7295 for (value = 0; value < 256; value++)
7297 ANYOF_BITMAP_SET(ret, value);
7304 ANYOF_CLASS_SET(ret, ANYOF_NCNTRL);
7306 for (value = 0; value < 256; value++)
7307 if (!isCNTRL(value))
7308 ANYOF_BITMAP_SET(ret, value);
7315 ANYOF_CLASS_SET(ret, ANYOF_DIGIT);
7317 /* consecutive digits assumed */
7318 for (value = '0'; value <= '9'; value++)
7319 ANYOF_BITMAP_SET(ret, value);
7326 ANYOF_CLASS_SET(ret, ANYOF_NDIGIT);
7328 /* consecutive digits assumed */
7329 for (value = 0; value < '0'; value++)
7330 ANYOF_BITMAP_SET(ret, value);
7331 for (value = '9' + 1; value < 256; value++)
7332 ANYOF_BITMAP_SET(ret, value);
7339 ANYOF_CLASS_SET(ret, ANYOF_GRAPH);
7341 for (value = 0; value < 256; value++)
7343 ANYOF_BITMAP_SET(ret, value);
7350 ANYOF_CLASS_SET(ret, ANYOF_NGRAPH);
7352 for (value = 0; value < 256; value++)
7353 if (!isGRAPH(value))
7354 ANYOF_BITMAP_SET(ret, value);
7361 ANYOF_CLASS_SET(ret, ANYOF_LOWER);
7363 for (value = 0; value < 256; value++)
7365 ANYOF_BITMAP_SET(ret, value);
7372 ANYOF_CLASS_SET(ret, ANYOF_NLOWER);
7374 for (value = 0; value < 256; value++)
7375 if (!isLOWER(value))
7376 ANYOF_BITMAP_SET(ret, value);
7383 ANYOF_CLASS_SET(ret, ANYOF_PRINT);
7385 for (value = 0; value < 256; value++)
7387 ANYOF_BITMAP_SET(ret, value);
7394 ANYOF_CLASS_SET(ret, ANYOF_NPRINT);
7396 for (value = 0; value < 256; value++)
7397 if (!isPRINT(value))
7398 ANYOF_BITMAP_SET(ret, value);
7405 ANYOF_CLASS_SET(ret, ANYOF_PSXSPC);
7407 for (value = 0; value < 256; value++)
7408 if (isPSXSPC(value))
7409 ANYOF_BITMAP_SET(ret, value);
7416 ANYOF_CLASS_SET(ret, ANYOF_NPSXSPC);
7418 for (value = 0; value < 256; value++)
7419 if (!isPSXSPC(value))
7420 ANYOF_BITMAP_SET(ret, value);
7427 ANYOF_CLASS_SET(ret, ANYOF_PUNCT);
7429 for (value = 0; value < 256; value++)
7431 ANYOF_BITMAP_SET(ret, value);
7438 ANYOF_CLASS_SET(ret, ANYOF_NPUNCT);
7440 for (value = 0; value < 256; value++)
7441 if (!isPUNCT(value))
7442 ANYOF_BITMAP_SET(ret, value);
7449 ANYOF_CLASS_SET(ret, ANYOF_SPACE);
7451 for (value = 0; value < 256; value++)
7453 ANYOF_BITMAP_SET(ret, value);
7460 ANYOF_CLASS_SET(ret, ANYOF_NSPACE);
7462 for (value = 0; value < 256; value++)
7463 if (!isSPACE(value))
7464 ANYOF_BITMAP_SET(ret, value);
7471 ANYOF_CLASS_SET(ret, ANYOF_UPPER);
7473 for (value = 0; value < 256; value++)
7475 ANYOF_BITMAP_SET(ret, value);
7482 ANYOF_CLASS_SET(ret, ANYOF_NUPPER);
7484 for (value = 0; value < 256; value++)
7485 if (!isUPPER(value))
7486 ANYOF_BITMAP_SET(ret, value);
7493 ANYOF_CLASS_SET(ret, ANYOF_XDIGIT);
7495 for (value = 0; value < 256; value++)
7496 if (isXDIGIT(value))
7497 ANYOF_BITMAP_SET(ret, value);
7504 ANYOF_CLASS_SET(ret, ANYOF_NXDIGIT);
7506 for (value = 0; value < 256; value++)
7507 if (!isXDIGIT(value))
7508 ANYOF_BITMAP_SET(ret, value);
7514 /* this is to handle \p and \P */
7517 vFAIL("Invalid [::] class");
7521 /* Strings such as "+utf8::isWord\n" */
7522 Perl_sv_catpvf(aTHX_ listsv, "%cutf8::Is%s\n", yesno, what);
7525 ANYOF_FLAGS(ret) |= ANYOF_CLASS;
7528 } /* end of namedclass \blah */
7531 if (prevvalue > (IV)value) /* b-a */ {
7532 const int w = RExC_parse - rangebegin;
7533 Simple_vFAIL4("Invalid [] range \"%*.*s\"", w, w, rangebegin);
7534 range = 0; /* not a valid range */
7538 prevvalue = value; /* save the beginning of the range */
7539 if (*RExC_parse == '-' && RExC_parse+1 < RExC_end &&
7540 RExC_parse[1] != ']') {
7543 /* a bad range like \w-, [:word:]- ? */
7544 if (namedclass > OOB_NAMEDCLASS) {
7545 if (ckWARN(WARN_REGEXP)) {
7547 RExC_parse >= rangebegin ?
7548 RExC_parse - rangebegin : 0;
7550 "False [] range \"%*.*s\"",
7554 ANYOF_BITMAP_SET(ret, '-');
7556 range = 1; /* yeah, it's a range! */
7557 continue; /* but do it the next time */
7561 /* now is the next time */
7562 /*stored += (value - prevvalue + 1);*/
7564 if (prevvalue < 256) {
7565 const IV ceilvalue = value < 256 ? value : 255;
7568 /* In EBCDIC [\x89-\x91] should include
7569 * the \x8e but [i-j] should not. */
7570 if (literal_endpoint == 2 &&
7571 ((isLOWER(prevvalue) && isLOWER(ceilvalue)) ||
7572 (isUPPER(prevvalue) && isUPPER(ceilvalue))))
7574 if (isLOWER(prevvalue)) {
7575 for (i = prevvalue; i <= ceilvalue; i++)
7577 ANYOF_BITMAP_SET(ret, i);
7579 for (i = prevvalue; i <= ceilvalue; i++)
7581 ANYOF_BITMAP_SET(ret, i);
7586 for (i = prevvalue; i <= ceilvalue; i++) {
7587 if (!ANYOF_BITMAP_TEST(ret,i)) {
7589 ANYOF_BITMAP_SET(ret, i);
7593 if (value > 255 || UTF) {
7594 const UV prevnatvalue = NATIVE_TO_UNI(prevvalue);
7595 const UV natvalue = NATIVE_TO_UNI(value);
7596 stored+=2; /* can't optimize this class */
7597 ANYOF_FLAGS(ret) |= ANYOF_UNICODE;
7598 if (prevnatvalue < natvalue) { /* what about > ? */
7599 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\t%04"UVxf"\n",
7600 prevnatvalue, natvalue);
7602 else if (prevnatvalue == natvalue) {
7603 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n", natvalue);
7605 U8 foldbuf[UTF8_MAXBYTES_CASE+1];
7607 const UV f = to_uni_fold(natvalue, foldbuf, &foldlen);
7609 #ifdef EBCDIC /* RD t/uni/fold ff and 6b */
7610 if (RExC_precomp[0] == ':' &&
7611 RExC_precomp[1] == '[' &&
7612 (f == 0xDF || f == 0x92)) {
7613 f = NATIVE_TO_UNI(f);
7616 /* If folding and foldable and a single
7617 * character, insert also the folded version
7618 * to the charclass. */
7620 #ifdef EBCDIC /* RD tunifold ligatures s,t fb05, fb06 */
7621 if ((RExC_precomp[0] == ':' &&
7622 RExC_precomp[1] == '[' &&
7624 (value == 0xFB05 || value == 0xFB06))) ?
7625 foldlen == ((STRLEN)UNISKIP(f) - 1) :
7626 foldlen == (STRLEN)UNISKIP(f) )
7628 if (foldlen == (STRLEN)UNISKIP(f))
7630 Perl_sv_catpvf(aTHX_ listsv,
7633 /* Any multicharacter foldings
7634 * require the following transform:
7635 * [ABCDEF] -> (?:[ABCabcDEFd]|pq|rst)
7636 * where E folds into "pq" and F folds
7637 * into "rst", all other characters
7638 * fold to single characters. We save
7639 * away these multicharacter foldings,
7640 * to be later saved as part of the
7641 * additional "s" data. */
7644 if (!unicode_alternate)
7645 unicode_alternate = newAV();
7646 sv = newSVpvn((char*)foldbuf, foldlen);
7648 av_push(unicode_alternate, sv);
7652 /* If folding and the value is one of the Greek
7653 * sigmas insert a few more sigmas to make the
7654 * folding rules of the sigmas to work right.
7655 * Note that not all the possible combinations
7656 * are handled here: some of them are handled
7657 * by the standard folding rules, and some of
7658 * them (literal or EXACTF cases) are handled
7659 * during runtime in regexec.c:S_find_byclass(). */
7660 if (value == UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) {
7661 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
7662 (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA);
7663 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
7664 (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA);
7666 else if (value == UNICODE_GREEK_CAPITAL_LETTER_SIGMA)
7667 Perl_sv_catpvf(aTHX_ listsv, "%04"UVxf"\n",
7668 (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA);
7673 literal_endpoint = 0;
7677 range = 0; /* this range (if it was one) is done now */
7681 ANYOF_FLAGS(ret) |= ANYOF_LARGE;
7683 RExC_size += ANYOF_CLASS_ADD_SKIP;
7685 RExC_emit += ANYOF_CLASS_ADD_SKIP;
7691 /****** !SIZE_ONLY AFTER HERE *********/
7693 if( stored == 1 && value < 256
7694 && !( ANYOF_FLAGS(ret) & ( ANYOF_FLAGS_ALL ^ ANYOF_FOLD ) )
7696 /* optimize single char class to an EXACT node
7697 but *only* when its not a UTF/high char */
7698 const char * cur_parse= RExC_parse;
7699 RExC_emit = (regnode *)orig_emit;
7700 RExC_parse = (char *)orig_parse;
7701 ret = reg_node(pRExC_state,
7702 (U8)((ANYOF_FLAGS(ret) & ANYOF_FOLD) ? EXACTF : EXACT));
7703 RExC_parse = (char *)cur_parse;
7704 *STRING(ret)= (char)value;
7706 RExC_emit += STR_SZ(1);
7709 /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
7710 if ( /* If the only flag is folding (plus possibly inversion). */
7711 ((ANYOF_FLAGS(ret) & (ANYOF_FLAGS_ALL ^ ANYOF_INVERT)) == ANYOF_FOLD)
7713 for (value = 0; value < 256; ++value) {
7714 if (ANYOF_BITMAP_TEST(ret, value)) {
7715 UV fold = PL_fold[value];
7718 ANYOF_BITMAP_SET(ret, fold);
7721 ANYOF_FLAGS(ret) &= ~ANYOF_FOLD;
7724 /* optimize inverted simple patterns (e.g. [^a-z]) */
7725 if (optimize_invert &&
7726 /* If the only flag is inversion. */
7727 (ANYOF_FLAGS(ret) & ANYOF_FLAGS_ALL) == ANYOF_INVERT) {
7728 for (value = 0; value < ANYOF_BITMAP_SIZE; ++value)
7729 ANYOF_BITMAP(ret)[value] ^= ANYOF_FLAGS_ALL;
7730 ANYOF_FLAGS(ret) = ANYOF_UNICODE_ALL;
7733 AV * const av = newAV();
7735 /* The 0th element stores the character class description
7736 * in its textual form: used later (regexec.c:Perl_regclass_swash())
7737 * to initialize the appropriate swash (which gets stored in
7738 * the 1st element), and also useful for dumping the regnode.
7739 * The 2nd element stores the multicharacter foldings,
7740 * used later (regexec.c:S_reginclass()). */
7741 av_store(av, 0, listsv);
7742 av_store(av, 1, NULL);
7743 av_store(av, 2, (SV*)unicode_alternate);
7744 rv = newRV_noinc((SV*)av);
7745 n = add_data(pRExC_state, 1, "s");
7746 RExC_rxi->data->data[n] = (void*)rv;
7753 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
7755 char* const retval = RExC_parse++;
7758 if (*RExC_parse == '(' && RExC_parse[1] == '?' &&
7759 RExC_parse[2] == '#') {
7760 while (*RExC_parse != ')') {
7761 if (RExC_parse == RExC_end)
7762 FAIL("Sequence (?#... not terminated");
7768 if (RExC_flags & RXf_PMf_EXTENDED) {
7769 if (isSPACE(*RExC_parse)) {
7773 else if (*RExC_parse == '#') {
7774 while (RExC_parse < RExC_end)
7775 if (*RExC_parse++ == '\n') break;
7784 - reg_node - emit a node
7786 STATIC regnode * /* Location. */
7787 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
7790 register regnode *ptr;
7791 regnode * const ret = RExC_emit;
7792 GET_RE_DEBUG_FLAGS_DECL;
7795 SIZE_ALIGN(RExC_size);
7800 if (OP(RExC_emit) == 255)
7801 Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %s: %d ",
7802 reg_name[op], OP(RExC_emit));
7804 NODE_ALIGN_FILL(ret);
7806 FILL_ADVANCE_NODE(ptr, op);
7807 if (RExC_offsets) { /* MJD */
7808 MJD_OFFSET_DEBUG(("%s:%d: (op %s) %s %"UVuf" (len %"UVuf") (max %"UVuf").\n",
7809 "reg_node", __LINE__,
7811 (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0]
7812 ? "Overwriting end of array!\n" : "OK",
7813 (UV)(RExC_emit - RExC_emit_start),
7814 (UV)(RExC_parse - RExC_start),
7815 (UV)RExC_offsets[0]));
7816 Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
7824 - reganode - emit a node with an argument
7826 STATIC regnode * /* Location. */
7827 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
7830 register regnode *ptr;
7831 regnode * const ret = RExC_emit;
7832 GET_RE_DEBUG_FLAGS_DECL;
7835 SIZE_ALIGN(RExC_size);
7840 assert(2==regarglen[op]+1);
7842 Anything larger than this has to allocate the extra amount.
7843 If we changed this to be:
7845 RExC_size += (1 + regarglen[op]);
7847 then it wouldn't matter. Its not clear what side effect
7848 might come from that so its not done so far.
7854 if (OP(RExC_emit) == 255)
7855 Perl_croak(aTHX_ "panic: reganode overwriting end of allocated program space");
7857 NODE_ALIGN_FILL(ret);
7859 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
7860 if (RExC_offsets) { /* MJD */
7861 MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n",
7865 (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0] ?
7866 "Overwriting end of array!\n" : "OK",
7867 (UV)(RExC_emit - RExC_emit_start),
7868 (UV)(RExC_parse - RExC_start),
7869 (UV)RExC_offsets[0]));
7870 Set_Cur_Node_Offset;
7878 - reguni - emit (if appropriate) a Unicode character
7881 S_reguni(pTHX_ const RExC_state_t *pRExC_state, UV uv, char* s)
7884 return SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
7888 - reginsert - insert an operator in front of already-emitted operand
7890 * Means relocating the operand.
7893 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd, U32 depth)
7896 register regnode *src;
7897 register regnode *dst;
7898 register regnode *place;
7899 const int offset = regarglen[(U8)op];
7900 const int size = NODE_STEP_REGNODE + offset;
7901 GET_RE_DEBUG_FLAGS_DECL;
7902 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
7903 DEBUG_PARSE_FMT("inst"," - %s",reg_name[op]);
7912 if (RExC_open_parens) {
7914 DEBUG_PARSE_FMT("inst"," - %"IVdf, (IV)RExC_npar);
7915 for ( paren=0 ; paren < RExC_npar ; paren++ ) {
7916 if ( RExC_open_parens[paren] >= opnd ) {
7917 DEBUG_PARSE_FMT("open"," - %d",size);
7918 RExC_open_parens[paren] += size;
7920 DEBUG_PARSE_FMT("open"," - %s","ok");
7922 if ( RExC_close_parens[paren] >= opnd ) {
7923 DEBUG_PARSE_FMT("close"," - %d",size);
7924 RExC_close_parens[paren] += size;
7926 DEBUG_PARSE_FMT("close"," - %s","ok");
7931 while (src > opnd) {
7932 StructCopy(--src, --dst, regnode);
7933 if (RExC_offsets) { /* MJD 20010112 */
7934 MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s copy %"UVuf" -> %"UVuf" (max %"UVuf").\n",
7938 (UV)(dst - RExC_emit_start) > RExC_offsets[0]
7939 ? "Overwriting end of array!\n" : "OK",
7940 (UV)(src - RExC_emit_start),
7941 (UV)(dst - RExC_emit_start),
7942 (UV)RExC_offsets[0]));
7943 Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
7944 Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
7949 place = opnd; /* Op node, where operand used to be. */
7950 if (RExC_offsets) { /* MJD */
7951 MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n",
7955 (UV)(place - RExC_emit_start) > RExC_offsets[0]
7956 ? "Overwriting end of array!\n" : "OK",
7957 (UV)(place - RExC_emit_start),
7958 (UV)(RExC_parse - RExC_start),
7959 (UV)RExC_offsets[0]));
7960 Set_Node_Offset(place, RExC_parse);
7961 Set_Node_Length(place, 1);
7963 src = NEXTOPER(place);
7964 FILL_ADVANCE_NODE(place, op);
7965 Zero(src, offset, regnode);
7969 - regtail - set the next-pointer at the end of a node chain of p to val.
7970 - SEE ALSO: regtail_study
7972 /* TODO: All three parms should be const */
7974 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,U32 depth)
7977 register regnode *scan;
7978 GET_RE_DEBUG_FLAGS_DECL;
7980 PERL_UNUSED_ARG(depth);
7986 /* Find last node. */
7989 regnode * const temp = regnext(scan);
7991 SV * const mysv=sv_newmortal();
7992 DEBUG_PARSE_MSG((scan==p ? "tail" : ""));
7993 regprop(RExC_rx, mysv, scan);
7994 PerlIO_printf(Perl_debug_log, "~ %s (%d) %s %s\n",
7995 SvPV_nolen_const(mysv), REG_NODE_NUM(scan),
7996 (temp == NULL ? "->" : ""),
7997 (temp == NULL ? reg_name[OP(val)] : "")
8005 if (reg_off_by_arg[OP(scan)]) {
8006 ARG_SET(scan, val - scan);
8009 NEXT_OFF(scan) = val - scan;
8015 - regtail_study - set the next-pointer at the end of a node chain of p to val.
8016 - Look for optimizable sequences at the same time.
8017 - currently only looks for EXACT chains.
8019 This is expermental code. The idea is to use this routine to perform
8020 in place optimizations on branches and groups as they are constructed,
8021 with the long term intention of removing optimization from study_chunk so
8022 that it is purely analytical.
8024 Currently only used when in DEBUG mode. The macro REGTAIL_STUDY() is used
8025 to control which is which.
8028 /* TODO: All four parms should be const */
8031 S_regtail_study(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,U32 depth)
8034 register regnode *scan;
8036 #ifdef EXPERIMENTAL_INPLACESCAN
8040 GET_RE_DEBUG_FLAGS_DECL;
8046 /* Find last node. */
8050 regnode * const temp = regnext(scan);
8051 #ifdef EXPERIMENTAL_INPLACESCAN
8052 if (PL_regkind[OP(scan)] == EXACT)
8053 if (join_exact(pRExC_state,scan,&min,1,val,depth+1))
8061 if( exact == PSEUDO )
8063 else if ( exact != OP(scan) )
8072 SV * const mysv=sv_newmortal();
8073 DEBUG_PARSE_MSG((scan==p ? "tsdy" : ""));
8074 regprop(RExC_rx, mysv, scan);
8075 PerlIO_printf(Perl_debug_log, "~ %s (%d) -> %s\n",
8076 SvPV_nolen_const(mysv),
8085 SV * const mysv_val=sv_newmortal();
8086 DEBUG_PARSE_MSG("");
8087 regprop(RExC_rx, mysv_val, val);
8088 PerlIO_printf(Perl_debug_log, "~ attach to %s (%"IVdf") offset to %"IVdf"\n",
8089 SvPV_nolen_const(mysv_val),
8090 (IV)REG_NODE_NUM(val),
8094 if (reg_off_by_arg[OP(scan)]) {
8095 ARG_SET(scan, val - scan);
8098 NEXT_OFF(scan) = val - scan;
8106 - regcurly - a little FSA that accepts {\d+,?\d*}
8109 S_regcurly(register const char *s)
8128 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
8131 Perl_regdump(pTHX_ const regexp *r)
8135 SV * const sv = sv_newmortal();
8136 SV *dsv= sv_newmortal();
8139 (void)dumpuntil(r, ri->program, ri->program + 1, NULL, NULL, sv, 0, 0);
8141 /* Header fields of interest. */
8142 if (r->anchored_substr) {
8143 RE_PV_QUOTED_DECL(s, 0, dsv, SvPVX_const(r->anchored_substr),
8144 RE_SV_DUMPLEN(r->anchored_substr), 30);
8145 PerlIO_printf(Perl_debug_log,
8146 "anchored %s%s at %"IVdf" ",
8147 s, RE_SV_TAIL(r->anchored_substr),
8148 (IV)r->anchored_offset);
8149 } else if (r->anchored_utf8) {
8150 RE_PV_QUOTED_DECL(s, 1, dsv, SvPVX_const(r->anchored_utf8),
8151 RE_SV_DUMPLEN(r->anchored_utf8), 30);
8152 PerlIO_printf(Perl_debug_log,
8153 "anchored utf8 %s%s at %"IVdf" ",
8154 s, RE_SV_TAIL(r->anchored_utf8),
8155 (IV)r->anchored_offset);
8157 if (r->float_substr) {
8158 RE_PV_QUOTED_DECL(s, 0, dsv, SvPVX_const(r->float_substr),
8159 RE_SV_DUMPLEN(r->float_substr), 30);
8160 PerlIO_printf(Perl_debug_log,
8161 "floating %s%s at %"IVdf"..%"UVuf" ",
8162 s, RE_SV_TAIL(r->float_substr),
8163 (IV)r->float_min_offset, (UV)r->float_max_offset);
8164 } else if (r->float_utf8) {
8165 RE_PV_QUOTED_DECL(s, 1, dsv, SvPVX_const(r->float_utf8),
8166 RE_SV_DUMPLEN(r->float_utf8), 30);
8167 PerlIO_printf(Perl_debug_log,
8168 "floating utf8 %s%s at %"IVdf"..%"UVuf" ",
8169 s, RE_SV_TAIL(r->float_utf8),
8170 (IV)r->float_min_offset, (UV)r->float_max_offset);
8172 if (r->check_substr || r->check_utf8)
8173 PerlIO_printf(Perl_debug_log,
8175 (r->check_substr == r->float_substr
8176 && r->check_utf8 == r->float_utf8
8177 ? "(checking floating" : "(checking anchored"));
8178 if (r->extflags & RXf_NOSCAN)
8179 PerlIO_printf(Perl_debug_log, " noscan");
8180 if (r->extflags & RXf_CHECK_ALL)
8181 PerlIO_printf(Perl_debug_log, " isall");
8182 if (r->check_substr || r->check_utf8)
8183 PerlIO_printf(Perl_debug_log, ") ");
8185 if (ri->regstclass) {
8186 regprop(r, sv, ri->regstclass);
8187 PerlIO_printf(Perl_debug_log, "stclass %s ", SvPVX_const(sv));
8189 if (r->extflags & RXf_ANCH) {
8190 PerlIO_printf(Perl_debug_log, "anchored");
8191 if (r->extflags & RXf_ANCH_BOL)
8192 PerlIO_printf(Perl_debug_log, "(BOL)");
8193 if (r->extflags & RXf_ANCH_MBOL)
8194 PerlIO_printf(Perl_debug_log, "(MBOL)");
8195 if (r->extflags & RXf_ANCH_SBOL)
8196 PerlIO_printf(Perl_debug_log, "(SBOL)");
8197 if (r->extflags & RXf_ANCH_GPOS)
8198 PerlIO_printf(Perl_debug_log, "(GPOS)");
8199 PerlIO_putc(Perl_debug_log, ' ');
8201 if (r->extflags & RXf_GPOS_SEEN)
8202 PerlIO_printf(Perl_debug_log, "GPOS:%"UVuf" ", (UV)r->gofs);
8203 if (r->intflags & PREGf_SKIP)
8204 PerlIO_printf(Perl_debug_log, "plus ");
8205 if (r->intflags & PREGf_IMPLICIT)
8206 PerlIO_printf(Perl_debug_log, "implicit ");
8207 PerlIO_printf(Perl_debug_log, "minlen %"IVdf" ", (IV)r->minlen);
8208 if (r->extflags & RXf_EVAL_SEEN)
8209 PerlIO_printf(Perl_debug_log, "with eval ");
8210 PerlIO_printf(Perl_debug_log, "\n");
8212 PERL_UNUSED_CONTEXT;
8214 #endif /* DEBUGGING */
8218 - regprop - printable representation of opcode
8221 Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
8226 RXi_GET_DECL(prog,progi);
8227 GET_RE_DEBUG_FLAGS_DECL;
8230 sv_setpvn(sv, "", 0);
8232 if (OP(o) > REGNODE_MAX) /* regnode.type is unsigned */
8233 /* It would be nice to FAIL() here, but this may be called from
8234 regexec.c, and it would be hard to supply pRExC_state. */
8235 Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d", (int)OP(o), (int)REGNODE_MAX);
8236 sv_catpv(sv, reg_name[OP(o)]); /* Take off const! */
8238 k = PL_regkind[OP(o)];
8241 SV * const dsv = sv_2mortal(newSVpvs(""));
8242 /* Using is_utf8_string() (via PERL_PV_UNI_DETECT)
8243 * is a crude hack but it may be the best for now since
8244 * we have no flag "this EXACTish node was UTF-8"
8246 const char * const s =
8247 pv_pretty(dsv, STRING(o), STR_LEN(o), 60,
8248 PL_colors[0], PL_colors[1],
8249 PERL_PV_ESCAPE_UNI_DETECT |
8250 PERL_PV_PRETTY_ELIPSES |
8253 Perl_sv_catpvf(aTHX_ sv, " %s", s );
8254 } else if (k == TRIE) {
8255 /* print the details of the trie in dumpuntil instead, as
8256 * progi->data isn't available here */
8257 const char op = OP(o);
8258 const I32 n = ARG(o);
8259 const reg_ac_data * const ac = IS_TRIE_AC(op) ?
8260 (reg_ac_data *)progi->data->data[n] :
8262 const reg_trie_data * const trie
8263 = (reg_trie_data*)progi->data->data[!IS_TRIE_AC(op) ? n : ac->trie];
8265 Perl_sv_catpvf(aTHX_ sv, "-%s",reg_name[o->flags]);
8266 DEBUG_TRIE_COMPILE_r(
8267 Perl_sv_catpvf(aTHX_ sv,
8268 "<S:%"UVuf"/%"IVdf" W:%"UVuf" L:%"UVuf"/%"UVuf" C:%"UVuf"/%"UVuf">",
8269 (UV)trie->startstate,
8270 (IV)trie->statecount-1, /* -1 because of the unused 0 element */
8271 (UV)trie->wordcount,
8274 (UV)TRIE_CHARCOUNT(trie),
8275 (UV)trie->uniquecharcount
8278 if ( IS_ANYOF_TRIE(op) || trie->bitmap ) {
8280 int rangestart = -1;
8281 U8* bitmap = IS_ANYOF_TRIE(op) ? (U8*)ANYOF_BITMAP(o) : (U8*)TRIE_BITMAP(trie);
8282 Perl_sv_catpvf(aTHX_ sv, "[");
8283 for (i = 0; i <= 256; i++) {
8284 if (i < 256 && BITMAP_TEST(bitmap,i)) {
8285 if (rangestart == -1)
8287 } else if (rangestart != -1) {
8288 if (i <= rangestart + 3)
8289 for (; rangestart < i; rangestart++)
8290 put_byte(sv, rangestart);
8292 put_byte(sv, rangestart);
8294 put_byte(sv, i - 1);
8299 Perl_sv_catpvf(aTHX_ sv, "]");
8302 } else if (k == CURLY) {
8303 if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
8304 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
8305 Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
8307 else if (k == WHILEM && o->flags) /* Ordinal/of */
8308 Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
8309 else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP || OP(o)==ACCEPT)
8310 Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o)); /* Parenth number */
8311 else if (k == GOSUB)
8312 Perl_sv_catpvf(aTHX_ sv, "%d[%+d]", (int)ARG(o),(int)ARG2L(o)); /* Paren and offset */
8313 else if (k == VERB) {
8315 Perl_sv_catpvf(aTHX_ sv, ":%"SVf,
8316 (SV*)progi->data->data[ ARG( o ) ]);
8317 } else if (k == LOGICAL)
8318 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* 2: embedded, otherwise 1 */
8319 else if (k == ANYOF) {
8320 int i, rangestart = -1;
8321 const U8 flags = ANYOF_FLAGS(o);
8323 /* Should be synchronized with * ANYOF_ #xdefines in regcomp.h */
8324 static const char * const anyofs[] = {
8357 if (flags & ANYOF_LOCALE)
8358 sv_catpvs(sv, "{loc}");
8359 if (flags & ANYOF_FOLD)
8360 sv_catpvs(sv, "{i}");
8361 Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
8362 if (flags & ANYOF_INVERT)
8364 for (i = 0; i <= 256; i++) {
8365 if (i < 256 && ANYOF_BITMAP_TEST(o,i)) {
8366 if (rangestart == -1)
8368 } else if (rangestart != -1) {
8369 if (i <= rangestart + 3)
8370 for (; rangestart < i; rangestart++)
8371 put_byte(sv, rangestart);
8373 put_byte(sv, rangestart);
8375 put_byte(sv, i - 1);
8381 if (o->flags & ANYOF_CLASS)
8382 for (i = 0; i < (int)(sizeof(anyofs)/sizeof(char*)); i++)
8383 if (ANYOF_CLASS_TEST(o,i))
8384 sv_catpv(sv, anyofs[i]);
8386 if (flags & ANYOF_UNICODE)
8387 sv_catpvs(sv, "{unicode}");
8388 else if (flags & ANYOF_UNICODE_ALL)
8389 sv_catpvs(sv, "{unicode_all}");
8393 SV * const sw = regclass_swash(prog, o, FALSE, &lv, 0);
8397 U8 s[UTF8_MAXBYTES_CASE+1];
8399 for (i = 0; i <= 256; i++) { /* just the first 256 */
8400 uvchr_to_utf8(s, i);
8402 if (i < 256 && swash_fetch(sw, s, TRUE)) {
8403 if (rangestart == -1)
8405 } else if (rangestart != -1) {
8406 if (i <= rangestart + 3)
8407 for (; rangestart < i; rangestart++) {
8408 const U8 * const e = uvchr_to_utf8(s,rangestart);
8410 for(p = s; p < e; p++)
8414 const U8 *e = uvchr_to_utf8(s,rangestart);
8416 for (p = s; p < e; p++)
8419 e = uvchr_to_utf8(s, i-1);
8420 for (p = s; p < e; p++)
8427 sv_catpvs(sv, "..."); /* et cetera */
8431 char *s = savesvpv(lv);
8432 char * const origs = s;
8434 while (*s && *s != '\n')
8438 const char * const t = ++s;
8456 Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
8458 else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
8459 Perl_sv_catpvf(aTHX_ sv, "[%d]", -(o->flags));
8461 PERL_UNUSED_CONTEXT;
8462 PERL_UNUSED_ARG(sv);
8464 PERL_UNUSED_ARG(prog);
8465 #endif /* DEBUGGING */
8469 Perl_re_intuit_string(pTHX_ regexp *prog)
8470 { /* Assume that RE_INTUIT is set */
8472 GET_RE_DEBUG_FLAGS_DECL;
8473 PERL_UNUSED_CONTEXT;
8477 const char * const s = SvPV_nolen_const(prog->check_substr
8478 ? prog->check_substr : prog->check_utf8);
8480 if (!PL_colorset) reginitcolors();
8481 PerlIO_printf(Perl_debug_log,
8482 "%sUsing REx %ssubstr:%s \"%s%.60s%s%s\"\n",
8484 prog->check_substr ? "" : "utf8 ",
8485 PL_colors[5],PL_colors[0],
8488 (strlen(s) > 60 ? "..." : ""));
8491 return prog->check_substr ? prog->check_substr : prog->check_utf8;
8497 handles refcounting and freeing the perl core regexp structure. When
8498 it is necessary to actually free the structure the first thing it
8499 does is call the 'free' method of the regexp_engine associated to to
8500 the regexp, allowing the handling of the void *pprivate; member
8501 first. (This routine is not overridable by extensions, which is why
8502 the extensions free is called first.)
8504 See regdupe and regdupe_internal if you change anything here.
8506 #ifndef PERL_IN_XSUB_RE
8508 Perl_pregfree(pTHX_ struct regexp *r)
8511 GET_RE_DEBUG_FLAGS_DECL;
8513 if (!r || (--r->refcnt > 0))
8516 CALLREGFREE_PVT(r); /* free the private data */
8518 /* gcov results gave these as non-null 100% of the time, so there's no
8519 optimisation in checking them before calling Safefree */
8520 Safefree(r->precomp);
8521 RX_MATCH_COPY_FREE(r);
8522 #ifdef PERL_OLD_COPY_ON_WRITE
8524 SvREFCNT_dec(r->saved_copy);
8527 if (r->anchored_substr)
8528 SvREFCNT_dec(r->anchored_substr);
8529 if (r->anchored_utf8)
8530 SvREFCNT_dec(r->anchored_utf8);
8531 if (r->float_substr)
8532 SvREFCNT_dec(r->float_substr);
8534 SvREFCNT_dec(r->float_utf8);
8535 Safefree(r->substrs);
8538 SvREFCNT_dec(r->paren_names);
8540 Safefree(r->startp);
8546 /* regfree_internal()
8548 Free the private data in a regexp. This is overloadable by
8549 extensions. Perl takes care of the regexp structure in pregfree(),
8550 this covers the *pprivate pointer which technically perldoesnt
8551 know about, however of course we have to handle the
8552 regexp_internal structure when no extension is in use.
8554 Note this is called before freeing anything in the regexp
8559 Perl_regfree_internal(pTHX_ struct regexp *r)
8563 GET_RE_DEBUG_FLAGS_DECL;
8569 SV *dsv= sv_newmortal();
8570 RE_PV_QUOTED_DECL(s, (r->extflags & RXf_UTF8),
8571 dsv, r->precomp, r->prelen, 60);
8572 PerlIO_printf(Perl_debug_log,"%sFreeing REx:%s %s\n",
8573 PL_colors[4],PL_colors[5],s);
8577 Safefree(ri->offsets); /* 20010421 MJD */
8579 int n = ri->data->count;
8580 PAD* new_comppad = NULL;
8585 /* If you add a ->what type here, update the comment in regcomp.h */
8586 switch (ri->data->what[n]) {
8590 SvREFCNT_dec((SV*)ri->data->data[n]);
8593 Safefree(ri->data->data[n]);
8596 new_comppad = (AV*)ri->data->data[n];
8599 if (new_comppad == NULL)
8600 Perl_croak(aTHX_ "panic: pregfree comppad");
8601 PAD_SAVE_LOCAL(old_comppad,
8602 /* Watch out for global destruction's random ordering. */
8603 (SvTYPE(new_comppad) == SVt_PVAV) ? new_comppad : NULL
8606 refcnt = OpREFCNT_dec((OP_4tree*)ri->data->data[n]);
8609 op_free((OP_4tree*)ri->data->data[n]);
8611 PAD_RESTORE_LOCAL(old_comppad);
8612 SvREFCNT_dec((SV*)new_comppad);
8618 { /* Aho Corasick add-on structure for a trie node.
8619 Used in stclass optimization only */
8621 reg_ac_data *aho=(reg_ac_data*)ri->data->data[n];
8623 refcount = --aho->refcount;
8626 PerlMemShared_free(aho->states);
8627 PerlMemShared_free(aho->fail);
8628 /* do this last!!!! */
8629 PerlMemShared_free(ri->data->data[n]);
8630 PerlMemShared_free(ri->regstclass);
8636 /* trie structure. */
8638 reg_trie_data *trie=(reg_trie_data*)ri->data->data[n];
8640 refcount = --trie->refcount;
8643 PerlMemShared_free(trie->charmap);
8644 PerlMemShared_free(trie->states);
8645 PerlMemShared_free(trie->trans);
8647 PerlMemShared_free(trie->bitmap);
8649 PerlMemShared_free(trie->wordlen);
8651 PerlMemShared_free(trie->jump);
8653 PerlMemShared_free(trie->nextword);
8654 /* do this last!!!! */
8655 PerlMemShared_free(ri->data->data[n]);
8660 Perl_croak(aTHX_ "panic: regfree data code '%c'", ri->data->what[n]);
8663 Safefree(ri->data->what);
8667 Safefree(ri->swap->startp);
8668 Safefree(ri->swap->endp);
8674 #define sv_dup_inc(s,t) SvREFCNT_inc(sv_dup(s,t))
8675 #define av_dup_inc(s,t) (AV*)SvREFCNT_inc(sv_dup((SV*)s,t))
8676 #define hv_dup_inc(s,t) (HV*)SvREFCNT_inc(sv_dup((SV*)s,t))
8677 #define SAVEPVN(p,n) ((p) ? savepvn(p,n) : NULL)
8680 regdupe - duplicate a regexp.
8682 This routine is called by sv.c's re_dup and is expected to clone a
8683 given regexp structure. It is a no-op when not under USE_ITHREADS.
8684 (Originally this *was* re_dup() for change history see sv.c)
8686 After all of the core data stored in struct regexp is duplicated
8687 the regexp_engine.dupe method is used to copy any private data
8688 stored in the *pprivate pointer. This allows extensions to handle
8689 any duplication it needs to do.
8691 See pregfree() and regfree_internal() if you change anything here.
8693 #if defined(USE_ITHREADS)
8694 #ifndef PERL_IN_XSUB_RE
8696 Perl_re_dup(pTHX_ const regexp *r, CLONE_PARAMS *param)
8701 struct reg_substr_datum *s;
8704 return (REGEXP *)NULL;
8706 if ((ret = (REGEXP *)ptr_table_fetch(PL_ptr_table, r)))
8710 npar = r->nparens+1;
8711 Newxz(ret, 1, regexp);
8712 Newx(ret->startp, npar, I32);
8713 Copy(r->startp, ret->startp, npar, I32);
8714 Newx(ret->endp, npar, I32);
8715 Copy(r->endp, ret->endp, npar, I32);
8718 Newx(ret->substrs, 1, struct reg_substr_data);
8719 for (s = ret->substrs->data, i = 0; i < 3; i++, s++) {
8720 s->min_offset = r->substrs->data[i].min_offset;
8721 s->max_offset = r->substrs->data[i].max_offset;
8722 s->end_shift = r->substrs->data[i].end_shift;
8723 s->substr = sv_dup_inc(r->substrs->data[i].substr, param);
8724 s->utf8_substr = sv_dup_inc(r->substrs->data[i].utf8_substr, param);
8727 ret->substrs = NULL;
8729 ret->precomp = SAVEPVN(r->precomp, r->prelen);
8730 ret->refcnt = r->refcnt;
8731 ret->minlen = r->minlen;
8732 ret->minlenret = r->minlenret;
8733 ret->prelen = r->prelen;
8734 ret->nparens = r->nparens;
8735 ret->lastparen = r->lastparen;
8736 ret->lastcloseparen = r->lastcloseparen;
8737 ret->intflags = r->intflags;
8738 ret->extflags = r->extflags;
8740 ret->sublen = r->sublen;
8742 ret->engine = r->engine;
8744 ret->paren_names = hv_dup_inc(r->paren_names, param);
8746 if (RX_MATCH_COPIED(ret))
8747 ret->subbeg = SAVEPVN(r->subbeg, r->sublen);
8750 #ifdef PERL_OLD_COPY_ON_WRITE
8751 ret->saved_copy = NULL;
8754 ret->pprivate = r->pprivate;
8756 RXi_SET(ret,CALLREGDUPE_PVT(ret,param));
8758 ptr_table_store(PL_ptr_table, r, ret);
8761 #endif /* PERL_IN_XSUB_RE */
8766 This is the internal complement to regdupe() which is used to copy
8767 the structure pointed to by the *pprivate pointer in the regexp.
8768 This is the core version of the extension overridable cloning hook.
8769 The regexp structure being duplicated will be copied by perl prior
8770 to this and will be provided as the regexp *r argument, however
8771 with the /old/ structures pprivate pointer value. Thus this routine
8772 may override any copying normally done by perl.
8774 It returns a pointer to the new regexp_internal structure.
8778 Perl_regdupe_internal(pTHX_ const regexp *r, CLONE_PARAMS *param)
8781 regexp_internal *reti;
8785 npar = r->nparens+1;
8786 len = ri->offsets[0];
8788 Newxc(reti, sizeof(regexp_internal) + (len+1)*sizeof(regnode), char, regexp_internal);
8789 Copy(ri->program, reti->program, len+1, regnode);
8792 Newx(reti->swap, 1, regexp_paren_ofs);
8793 /* no need to copy these */
8794 Newx(reti->swap->startp, npar, I32);
8795 Newx(reti->swap->endp, npar, I32);
8801 reti->regstclass = NULL;
8804 const int count = ri->data->count;
8807 Newxc(d, sizeof(struct reg_data) + count*sizeof(void *),
8808 char, struct reg_data);
8809 Newx(d->what, count, U8);
8812 for (i = 0; i < count; i++) {
8813 d->what[i] = ri->data->what[i];
8814 switch (d->what[i]) {
8815 /* legal options are one of: sSfpontTu
8816 see also regcomp.h and pregfree() */
8819 case 'p': /* actually an AV, but the dup function is identical. */
8820 case 'u': /* actually an HV, but the dup function is identical. */
8821 d->data[i] = sv_dup_inc((SV *)ri->data->data[i], param);
8824 /* This is cheating. */
8825 Newx(d->data[i], 1, struct regnode_charclass_class);
8826 StructCopy(ri->data->data[i], d->data[i],
8827 struct regnode_charclass_class);
8828 reti->regstclass = (regnode*)d->data[i];
8831 /* Compiled op trees are readonly and in shared memory,
8832 and can thus be shared without duplication. */
8834 d->data[i] = (void*)OpREFCNT_inc((OP*)ri->data->data[i]);
8838 /* Trie stclasses are readonly and can thus be shared
8839 * without duplication. We free the stclass in pregfree
8840 * when the corresponding reg_ac_data struct is freed.
8842 reti->regstclass= ri->regstclass;
8846 ((reg_trie_data*)ri->data->data[i])->refcount++;
8850 d->data[i] = ri->data->data[i];
8853 Perl_croak(aTHX_ "panic: re_dup unknown data code '%c'", ri->data->what[i]);
8862 Newx(reti->offsets, 2*len+1, U32);
8863 Copy(ri->offsets, reti->offsets, 2*len+1, U32);
8868 #endif /* USE_ITHREADS */
8873 converts a regexp embedded in a MAGIC struct to its stringified form,
8874 caching the converted form in the struct and returns the cached
8877 If lp is nonnull then it is used to return the length of the
8880 If flags is nonnull and the returned string contains UTF8 then
8881 (*flags & 1) will be true.
8883 If haseval is nonnull then it is used to return whether the pattern
8886 Normally called via macro:
8888 CALLREG_STRINGIFY(mg,&len,&utf8);
8892 CALLREG_AS_STR(mg,&lp,&flags,&haseval)
8894 See sv_2pv_flags() in sv.c for an example of internal usage.
8897 #ifndef PERL_IN_XSUB_RE
8899 Perl_reg_stringify(pTHX_ MAGIC *mg, STRLEN *lp, U32 *flags, I32 *haseval ) {
8901 const regexp * const re = (regexp *)mg->mg_obj;
8904 const char *fptr = "msix";
8909 bool need_newline = 0;
8910 U16 reganch = (U16)((re->extflags & RXf_PMf_COMPILETIME) >> 12);
8912 while((ch = *fptr++)) {
8914 reflags[left++] = ch;
8917 reflags[right--] = ch;
8922 reflags[left] = '-';
8926 mg->mg_len = re->prelen + 4 + left;
8928 * If /x was used, we have to worry about a regex ending with a
8929 * comment later being embedded within another regex. If so, we don't
8930 * want this regex's "commentization" to leak out to the right part of
8931 * the enclosing regex, we must cap it with a newline.
8933 * So, if /x was used, we scan backwards from the end of the regex. If
8934 * we find a '#' before we find a newline, we need to add a newline
8935 * ourself. If we find a '\n' first (or if we don't find '#' or '\n'),
8936 * we don't need to add anything. -jfriedl
8938 if (PMf_EXTENDED & re->extflags) {
8939 const char *endptr = re->precomp + re->prelen;
8940 while (endptr >= re->precomp) {
8941 const char c = *(endptr--);
8943 break; /* don't need another */
8945 /* we end while in a comment, so we need a newline */
8946 mg->mg_len++; /* save space for it */
8947 need_newline = 1; /* note to add it */
8953 Newx(mg->mg_ptr, mg->mg_len + 1 + left, char);
8954 mg->mg_ptr[0] = '(';
8955 mg->mg_ptr[1] = '?';
8956 Copy(reflags, mg->mg_ptr+2, left, char);
8957 *(mg->mg_ptr+left+2) = ':';
8958 Copy(re->precomp, mg->mg_ptr+3+left, re->prelen, char);
8960 mg->mg_ptr[mg->mg_len - 2] = '\n';
8961 mg->mg_ptr[mg->mg_len - 1] = ')';
8962 mg->mg_ptr[mg->mg_len] = 0;
8965 *haseval = re->seen_evals;
8967 *flags = ((re->extflags & RXf_UTF8) ? 1 : 0);
8975 - regnext - dig the "next" pointer out of a node
8978 Perl_regnext(pTHX_ register regnode *p)
8981 register I32 offset;
8986 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
8995 S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
8998 STRLEN l1 = strlen(pat1);
8999 STRLEN l2 = strlen(pat2);
9002 const char *message;
9008 Copy(pat1, buf, l1 , char);
9009 Copy(pat2, buf + l1, l2 , char);
9010 buf[l1 + l2] = '\n';
9011 buf[l1 + l2 + 1] = '\0';
9013 /* ANSI variant takes additional second argument */
9014 va_start(args, pat2);
9018 msv = vmess(buf, &args);
9020 message = SvPV_const(msv,l1);
9023 Copy(message, buf, l1 , char);
9024 buf[l1-1] = '\0'; /* Overwrite \n */
9025 Perl_croak(aTHX_ "%s", buf);
9028 /* XXX Here's a total kludge. But we need to re-enter for swash routines. */
9030 #ifndef PERL_IN_XSUB_RE
9032 Perl_save_re_context(pTHX)
9036 struct re_save_state *state;
9038 SAVEVPTR(PL_curcop);
9039 SSGROW(SAVESTACK_ALLOC_FOR_RE_SAVE_STATE + 1);
9041 state = (struct re_save_state *)(PL_savestack + PL_savestack_ix);
9042 PL_savestack_ix += SAVESTACK_ALLOC_FOR_RE_SAVE_STATE;
9043 SSPUSHINT(SAVEt_RE_STATE);
9045 Copy(&PL_reg_state, state, 1, struct re_save_state);
9047 PL_reg_start_tmp = 0;
9048 PL_reg_start_tmpl = 0;
9049 PL_reg_oldsaved = NULL;
9050 PL_reg_oldsavedlen = 0;
9052 PL_reg_leftiter = 0;
9053 PL_reg_poscache = NULL;
9054 PL_reg_poscache_size = 0;
9055 #ifdef PERL_OLD_COPY_ON_WRITE
9059 /* Save $1..$n (#18107: UTF-8 s/(\w+)/uc($1)/e); AMS 20021106. */
9061 const REGEXP * const rx = PM_GETRE(PL_curpm);
9064 for (i = 1; i <= rx->nparens; i++) {
9065 char digits[TYPE_CHARS(long)];
9066 const STRLEN len = my_snprintf(digits, sizeof(digits), "%lu", (long)i);
9067 GV *const *const gvp
9068 = (GV**)hv_fetch(PL_defstash, digits, len, 0);
9071 GV * const gv = *gvp;
9072 if (SvTYPE(gv) == SVt_PVGV && GvSV(gv))
9082 clear_re(pTHX_ void *r)
9085 ReREFCNT_dec((regexp *)r);
9091 S_put_byte(pTHX_ SV *sv, int c)
9093 if (isCNTRL(c) || c == 255 || !isPRINT(c))
9094 Perl_sv_catpvf(aTHX_ sv, "\\%o", c);
9095 else if (c == '-' || c == ']' || c == '\\' || c == '^')
9096 Perl_sv_catpvf(aTHX_ sv, "\\%c", c);
9098 Perl_sv_catpvf(aTHX_ sv, "%c", c);
9102 #define CLEAR_OPTSTART \
9103 if (optstart) STMT_START { \
9104 DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log, " (%"IVdf" nodes)\n", (IV)(node - optstart))); \
9108 #define DUMPUNTIL(b,e) CLEAR_OPTSTART; node=dumpuntil(r,start,(b),(e),last,sv,indent+1,depth+1);
9110 STATIC const regnode *
9111 S_dumpuntil(pTHX_ const regexp *r, const regnode *start, const regnode *node,
9112 const regnode *last, const regnode *plast,
9113 SV* sv, I32 indent, U32 depth)
9116 register U8 op = PSEUDO; /* Arbitrary non-END op. */
9117 register const regnode *next;
9118 const regnode *optstart= NULL;
9120 GET_RE_DEBUG_FLAGS_DECL;
9122 #ifdef DEBUG_DUMPUNTIL
9123 PerlIO_printf(Perl_debug_log, "--- %d : %d - %d - %d\n",indent,node-start,
9124 last ? last-start : 0,plast ? plast-start : 0);
9127 if (plast && plast < last)
9130 while (PL_regkind[op] != END && (!last || node < last)) {
9131 /* While that wasn't END last time... */
9135 if (op == CLOSE || op == WHILEM)
9137 next = regnext((regnode *)node);
9140 if (OP(node) == OPTIMIZED) {
9141 if (!optstart && RE_DEBUG_FLAG(RE_DEBUG_COMPILE_OPTIMISE))
9148 regprop(r, sv, node);
9149 PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
9150 (int)(2*indent + 1), "", SvPVX_const(sv));
9152 if (OP(node) != OPTIMIZED) {
9153 if (next == NULL) /* Next ptr. */
9154 PerlIO_printf(Perl_debug_log, "(0)");
9155 else if (PL_regkind[(U8)op] == BRANCH && PL_regkind[OP(next)] != BRANCH )
9156 PerlIO_printf(Perl_debug_log, "(FAIL)");
9158 PerlIO_printf(Perl_debug_log, "(%"IVdf")", (IV)(next - start));
9160 /*if (PL_regkind[(U8)op] != TRIE)*/
9161 (void)PerlIO_putc(Perl_debug_log, '\n');
9165 if (PL_regkind[(U8)op] == BRANCHJ) {
9168 register const regnode *nnode = (OP(next) == LONGJMP
9169 ? regnext((regnode *)next)
9171 if (last && nnode > last)
9173 DUMPUNTIL(NEXTOPER(NEXTOPER(node)), nnode);
9176 else if (PL_regkind[(U8)op] == BRANCH) {
9178 DUMPUNTIL(NEXTOPER(node), next);
9180 else if ( PL_regkind[(U8)op] == TRIE ) {
9181 const regnode *this_trie = node;
9182 const char op = OP(node);
9183 const I32 n = ARG(node);
9184 const reg_ac_data * const ac = op>=AHOCORASICK ?
9185 (reg_ac_data *)ri->data->data[n] :
9187 const reg_trie_data * const trie =
9188 (reg_trie_data*)ri->data->data[op<AHOCORASICK ? n : ac->trie];
9190 AV *const trie_words = (AV *) ri->data->data[n + TRIE_WORDS_OFFSET];
9192 const regnode *nextbranch= NULL;
9194 sv_setpvn(sv, "", 0);
9195 for (word_idx= 0; word_idx < (I32)trie->wordcount; word_idx++) {
9196 SV ** const elem_ptr = av_fetch(trie_words,word_idx,0);
9198 PerlIO_printf(Perl_debug_log, "%*s%s ",
9199 (int)(2*(indent+3)), "",
9200 elem_ptr ? pv_pretty(sv, SvPV_nolen_const(*elem_ptr), SvCUR(*elem_ptr), 60,
9201 PL_colors[0], PL_colors[1],
9202 (SvUTF8(*elem_ptr) ? PERL_PV_ESCAPE_UNI : 0) |
9203 PERL_PV_PRETTY_ELIPSES |
9209 U16 dist= trie->jump[word_idx+1];
9210 PerlIO_printf(Perl_debug_log, "(%"UVuf")\n",
9211 (UV)((dist ? this_trie + dist : next) - start));
9214 nextbranch= this_trie + trie->jump[0];
9215 DUMPUNTIL(this_trie + dist, nextbranch);
9217 if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
9218 nextbranch= regnext((regnode *)nextbranch);
9220 PerlIO_printf(Perl_debug_log, "\n");
9223 if (last && next > last)
9228 else if ( op == CURLY ) { /* "next" might be very big: optimizer */
9229 DUMPUNTIL(NEXTOPER(node) + EXTRA_STEP_2ARGS,
9230 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1);
9232 else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
9234 DUMPUNTIL(NEXTOPER(node) + EXTRA_STEP_2ARGS, next);
9236 else if ( op == PLUS || op == STAR) {
9237 DUMPUNTIL(NEXTOPER(node), NEXTOPER(node) + 1);
9239 else if (op == ANYOF) {
9240 /* arglen 1 + class block */
9241 node += 1 + ((ANYOF_FLAGS(node) & ANYOF_LARGE)
9242 ? ANYOF_CLASS_SKIP : ANYOF_SKIP);
9243 node = NEXTOPER(node);
9245 else if (PL_regkind[(U8)op] == EXACT) {
9246 /* Literal string, where present. */
9247 node += NODE_SZ_STR(node) - 1;
9248 node = NEXTOPER(node);
9251 node = NEXTOPER(node);
9252 node += regarglen[(U8)op];
9254 if (op == CURLYX || op == OPEN)
9258 #ifdef DEBUG_DUMPUNTIL
9259 PerlIO_printf(Perl_debug_log, "--- %d\n", (int)indent);
9264 #endif /* DEBUGGING */
9268 * c-indentation-style: bsd
9270 * indent-tabs-mode: t
9273 * ex: set ts=8 sts=4 sw=4 noet: