5 * 'A fair jaw-cracker dwarf-language must be.' --Samwise Gamgee
7 * [p.285 of _The Lord of the Rings_, II/iii: "The Ring Goes South"]
10 /* This file contains functions for compiling a regular expression. See
11 * also regexec.c which funnily enough, contains functions for executing
12 * a regular expression.
14 * This file is also copied at build time to ext/re/re_comp.c, where
15 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
16 * This causes the main functions to be compiled under new names and with
17 * debugging support added, which makes "use re 'debug'" work.
20 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
21 * confused with the original package (see point 3 below). Thanks, Henry!
24 /* Additional note: this code is very heavily munged from Henry's version
25 * in places. In some spots I've traded clarity for efficiency, so don't
26 * blame Henry for some of the lack of readability.
29 /* The names of the functions have been changed from regcomp and
30 * regexec to pregcomp and pregexec in order to avoid conflicts
31 * with the POSIX routines of the same names.
34 #ifdef PERL_EXT_RE_BUILD
39 * pregcomp and pregexec -- regsub and regerror are not used in perl
41 * Copyright (c) 1986 by University of Toronto.
42 * Written by Henry Spencer. Not derived from licensed software.
44 * Permission is granted to anyone to use this software for any
45 * purpose on any computer system, and to redistribute it freely,
46 * subject to the following restrictions:
48 * 1. The author is not responsible for the consequences of use of
49 * this software, no matter how awful, even if they arise
52 * 2. The origin of this software must not be misrepresented, either
53 * by explicit claim or by omission.
55 * 3. Altered versions must be plainly marked as such, and must not
56 * be misrepresented as being the original software.
59 **** Alterations to Henry's code are...
61 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
62 **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
63 **** by Larry Wall and others
65 **** You may distribute under the terms of either the GNU General Public
66 **** License or the Artistic License, as specified in the README file.
69 * Beware that some of this code is subtly aware of the way operator
70 * precedence is structured in regular expressions. Serious changes in
71 * regular-expression syntax might require a total rethink.
74 #define PERL_IN_REGCOMP_C
77 #ifndef PERL_IN_XSUB_RE
82 #ifdef PERL_IN_XSUB_RE
88 #include "dquote_static.c"
95 # if defined(BUGGY_MSC6)
96 /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
97 # pragma optimize("a",off)
98 /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
99 # pragma optimize("w",on )
100 # endif /* BUGGY_MSC6 */
104 #define STATIC static
107 typedef struct RExC_state_t {
108 U32 flags; /* are we folding, multilining? */
109 char *precomp; /* uncompiled string. */
110 REGEXP *rx_sv; /* The SV that is the regexp. */
111 regexp *rx; /* perl core regexp structure */
112 regexp_internal *rxi; /* internal data for regexp object pprivate field */
113 char *start; /* Start of input for compile */
114 char *end; /* End of input for compile */
115 char *parse; /* Input-scan pointer. */
116 I32 whilem_seen; /* number of WHILEM in this expr */
117 regnode *emit_start; /* Start of emitted-code area */
118 regnode *emit_bound; /* First regnode outside of the allocated space */
119 regnode *emit; /* Code-emit pointer; ®dummy = don't = compiling */
120 I32 naughty; /* How bad is this pattern? */
121 I32 sawback; /* Did we see \1, ...? */
123 I32 size; /* Code size. */
124 I32 npar; /* Capture buffer count, (OPEN). */
125 I32 cpar; /* Capture buffer count, (CLOSE). */
126 I32 nestroot; /* root parens we are in - used by accept */
130 regnode **open_parens; /* pointers to open parens */
131 regnode **close_parens; /* pointers to close parens */
132 regnode *opend; /* END node in program */
133 I32 utf8; /* whether the pattern is utf8 or not */
134 I32 orig_utf8; /* whether the pattern was originally in utf8 */
135 /* XXX use this for future optimisation of case
136 * where pattern must be upgraded to utf8. */
137 HV *paren_names; /* Paren names */
139 regnode **recurse; /* Recurse regops */
140 I32 recurse_count; /* Number of recurse regops */
142 char *starttry; /* -Dr: where regtry was called. */
143 #define RExC_starttry (pRExC_state->starttry)
146 const char *lastparse;
148 AV *paren_name_list; /* idx -> name */
149 #define RExC_lastparse (pRExC_state->lastparse)
150 #define RExC_lastnum (pRExC_state->lastnum)
151 #define RExC_paren_name_list (pRExC_state->paren_name_list)
155 #define RExC_flags (pRExC_state->flags)
156 #define RExC_precomp (pRExC_state->precomp)
157 #define RExC_rx_sv (pRExC_state->rx_sv)
158 #define RExC_rx (pRExC_state->rx)
159 #define RExC_rxi (pRExC_state->rxi)
160 #define RExC_start (pRExC_state->start)
161 #define RExC_end (pRExC_state->end)
162 #define RExC_parse (pRExC_state->parse)
163 #define RExC_whilem_seen (pRExC_state->whilem_seen)
164 #ifdef RE_TRACK_PATTERN_OFFSETS
165 #define RExC_offsets (pRExC_state->rxi->u.offsets) /* I am not like the others */
167 #define RExC_emit (pRExC_state->emit)
168 #define RExC_emit_start (pRExC_state->emit_start)
169 #define RExC_emit_bound (pRExC_state->emit_bound)
170 #define RExC_naughty (pRExC_state->naughty)
171 #define RExC_sawback (pRExC_state->sawback)
172 #define RExC_seen (pRExC_state->seen)
173 #define RExC_size (pRExC_state->size)
174 #define RExC_npar (pRExC_state->npar)
175 #define RExC_nestroot (pRExC_state->nestroot)
176 #define RExC_extralen (pRExC_state->extralen)
177 #define RExC_seen_zerolen (pRExC_state->seen_zerolen)
178 #define RExC_seen_evals (pRExC_state->seen_evals)
179 #define RExC_utf8 (pRExC_state->utf8)
180 #define RExC_orig_utf8 (pRExC_state->orig_utf8)
181 #define RExC_open_parens (pRExC_state->open_parens)
182 #define RExC_close_parens (pRExC_state->close_parens)
183 #define RExC_opend (pRExC_state->opend)
184 #define RExC_paren_names (pRExC_state->paren_names)
185 #define RExC_recurse (pRExC_state->recurse)
186 #define RExC_recurse_count (pRExC_state->recurse_count)
189 #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
190 #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
191 ((*s) == '{' && regcurly(s)))
194 #undef SPSTART /* dratted cpp namespace... */
197 * Flags to be passed up and down.
199 #define WORST 0 /* Worst case. */
200 #define HASWIDTH 0x01 /* Known to match non-null strings. */
202 /* Simple enough to be STAR/PLUS operand, in an EXACT node must be a single
203 * character, and if utf8, must be invariant. */
205 #define SPSTART 0x04 /* Starts with * or +. */
206 #define TRYAGAIN 0x08 /* Weeded out a declaration. */
207 #define POSTPONED 0x10 /* (?1),(?&name), (??{...}) or similar */
209 #define REG_NODE_NUM(x) ((x) ? (int)((x)-RExC_emit_start) : -1)
211 /* whether trie related optimizations are enabled */
212 #if PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION
213 #define TRIE_STUDY_OPT
214 #define FULL_TRIE_STUDY
220 #define PBYTE(u8str,paren) ((U8*)(u8str))[(paren) >> 3]
221 #define PBITVAL(paren) (1 << ((paren) & 7))
222 #define PAREN_TEST(u8str,paren) ( PBYTE(u8str,paren) & PBITVAL(paren))
223 #define PAREN_SET(u8str,paren) PBYTE(u8str,paren) |= PBITVAL(paren)
224 #define PAREN_UNSET(u8str,paren) PBYTE(u8str,paren) &= (~PBITVAL(paren))
226 /* If not already in utf8, do a longjmp back to the beginning */
227 #define UTF8_LONGJMP 42 /* Choose a value not likely to ever conflict */
228 #define REQUIRE_UTF8 STMT_START { \
229 if (! UTF) JMPENV_JUMP(UTF8_LONGJMP); \
232 /* About scan_data_t.
234 During optimisation we recurse through the regexp program performing
235 various inplace (keyhole style) optimisations. In addition study_chunk
236 and scan_commit populate this data structure with information about
237 what strings MUST appear in the pattern. We look for the longest
238 string that must appear at a fixed location, and we look for the
239 longest string that may appear at a floating location. So for instance
244 Both 'FOO' and 'A' are fixed strings. Both 'B' and 'BAR' are floating
245 strings (because they follow a .* construct). study_chunk will identify
246 both FOO and BAR as being the longest fixed and floating strings respectively.
248 The strings can be composites, for instance
252 will result in a composite fixed substring 'foo'.
254 For each string some basic information is maintained:
256 - offset or min_offset
257 This is the position the string must appear at, or not before.
258 It also implicitly (when combined with minlenp) tells us how many
259 characters must match before the string we are searching for.
260 Likewise when combined with minlenp and the length of the string it
261 tells us how many characters must appear after the string we have
265 Only used for floating strings. This is the rightmost point that
266 the string can appear at. If set to I32 max it indicates that the
267 string can occur infinitely far to the right.
270 A pointer to the minimum length of the pattern that the string
271 was found inside. This is important as in the case of positive
272 lookahead or positive lookbehind we can have multiple patterns
277 The minimum length of the pattern overall is 3, the minimum length
278 of the lookahead part is 3, but the minimum length of the part that
279 will actually match is 1. So 'FOO's minimum length is 3, but the
280 minimum length for the F is 1. This is important as the minimum length
281 is used to determine offsets in front of and behind the string being
282 looked for. Since strings can be composites this is the length of the
283 pattern at the time it was commited with a scan_commit. Note that
284 the length is calculated by study_chunk, so that the minimum lengths
285 are not known until the full pattern has been compiled, thus the
286 pointer to the value.
290 In the case of lookbehind the string being searched for can be
291 offset past the start point of the final matching string.
292 If this value was just blithely removed from the min_offset it would
293 invalidate some of the calculations for how many chars must match
294 before or after (as they are derived from min_offset and minlen and
295 the length of the string being searched for).
296 When the final pattern is compiled and the data is moved from the
297 scan_data_t structure into the regexp structure the information
298 about lookbehind is factored in, with the information that would
299 have been lost precalculated in the end_shift field for the
302 The fields pos_min and pos_delta are used to store the minimum offset
303 and the delta to the maximum offset at the current point in the pattern.
307 typedef struct scan_data_t {
308 /*I32 len_min; unused */
309 /*I32 len_delta; unused */
313 I32 last_end; /* min value, <0 unless valid. */
316 SV **longest; /* Either &l_fixed, or &l_float. */
317 SV *longest_fixed; /* longest fixed string found in pattern */
318 I32 offset_fixed; /* offset where it starts */
319 I32 *minlen_fixed; /* pointer to the minlen relevent to the string */
320 I32 lookbehind_fixed; /* is the position of the string modfied by LB */
321 SV *longest_float; /* longest floating string found in pattern */
322 I32 offset_float_min; /* earliest point in string it can appear */
323 I32 offset_float_max; /* latest point in string it can appear */
324 I32 *minlen_float; /* pointer to the minlen relevent to the string */
325 I32 lookbehind_float; /* is the position of the string modified by LB */
329 struct regnode_charclass_class *start_class;
333 * Forward declarations for pregcomp()'s friends.
336 static const scan_data_t zero_scan_data =
337 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ,0};
339 #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
340 #define SF_BEFORE_SEOL 0x0001
341 #define SF_BEFORE_MEOL 0x0002
342 #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
343 #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
346 # define SF_FIX_SHIFT_EOL (0+2)
347 # define SF_FL_SHIFT_EOL (0+4)
349 # define SF_FIX_SHIFT_EOL (+2)
350 # define SF_FL_SHIFT_EOL (+4)
353 #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
354 #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
356 #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
357 #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
358 #define SF_IS_INF 0x0040
359 #define SF_HAS_PAR 0x0080
360 #define SF_IN_PAR 0x0100
361 #define SF_HAS_EVAL 0x0200
362 #define SCF_DO_SUBSTR 0x0400
363 #define SCF_DO_STCLASS_AND 0x0800
364 #define SCF_DO_STCLASS_OR 0x1000
365 #define SCF_DO_STCLASS (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
366 #define SCF_WHILEM_VISITED_POS 0x2000
368 #define SCF_TRIE_RESTUDY 0x4000 /* Do restudy? */
369 #define SCF_SEEN_ACCEPT 0x8000
371 #define UTF cBOOL(RExC_utf8)
372 #define LOC cBOOL(RExC_flags & RXf_PMf_LOCALE)
373 #define UNI_SEMANTICS cBOOL(RExC_flags & RXf_PMf_UNICODE)
374 #define FOLD cBOOL(RExC_flags & RXf_PMf_FOLD)
376 #define OOB_UNICODE 12345678
377 #define OOB_NAMEDCLASS -1
379 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
380 #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
383 /* length of regex to show in messages that don't mark a position within */
384 #define RegexLengthToShowInErrorMessages 127
387 * If MARKER[12] are adjusted, be sure to adjust the constants at the top
388 * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
389 * op/pragma/warn/regcomp.
391 #define MARKER1 "<-- HERE" /* marker as it appears in the description */
392 #define MARKER2 " <-- HERE " /* marker as it appears within the regex */
394 #define REPORT_LOCATION " in regex; marked by " MARKER1 " in m/%.*s" MARKER2 "%s/"
397 * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
398 * arg. Show regex, up to a maximum length. If it's too long, chop and add
401 #define _FAIL(code) STMT_START { \
402 const char *ellipses = ""; \
403 IV len = RExC_end - RExC_precomp; \
406 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx_sv); \
407 if (len > RegexLengthToShowInErrorMessages) { \
408 /* chop 10 shorter than the max, to ensure meaning of "..." */ \
409 len = RegexLengthToShowInErrorMessages - 10; \
415 #define FAIL(msg) _FAIL( \
416 Perl_croak(aTHX_ "%s in regex m/%.*s%s/", \
417 msg, (int)len, RExC_precomp, ellipses))
419 #define FAIL2(msg,arg) _FAIL( \
420 Perl_croak(aTHX_ msg " in regex m/%.*s%s/", \
421 arg, (int)len, RExC_precomp, ellipses))
424 * Simple_vFAIL -- like FAIL, but marks the current location in the scan
426 #define Simple_vFAIL(m) STMT_START { \
427 const IV offset = RExC_parse - RExC_precomp; \
428 Perl_croak(aTHX_ "%s" REPORT_LOCATION, \
429 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
433 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
435 #define vFAIL(m) STMT_START { \
437 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx_sv); \
442 * Like Simple_vFAIL(), but accepts two arguments.
444 #define Simple_vFAIL2(m,a1) STMT_START { \
445 const IV offset = RExC_parse - RExC_precomp; \
446 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, \
447 (int)offset, RExC_precomp, RExC_precomp + offset); \
451 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
453 #define vFAIL2(m,a1) STMT_START { \
455 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx_sv); \
456 Simple_vFAIL2(m, a1); \
461 * Like Simple_vFAIL(), but accepts three arguments.
463 #define Simple_vFAIL3(m, a1, a2) STMT_START { \
464 const IV offset = RExC_parse - RExC_precomp; \
465 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, \
466 (int)offset, RExC_precomp, RExC_precomp + offset); \
470 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
472 #define vFAIL3(m,a1,a2) STMT_START { \
474 SAVEDESTRUCTOR_X(clear_re,(void*)RExC_rx_sv); \
475 Simple_vFAIL3(m, a1, a2); \
479 * Like Simple_vFAIL(), but accepts four arguments.
481 #define Simple_vFAIL4(m, a1, a2, a3) STMT_START { \
482 const IV offset = RExC_parse - RExC_precomp; \
483 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3, \
484 (int)offset, RExC_precomp, RExC_precomp + offset); \
487 #define ckWARNreg(loc,m) STMT_START { \
488 const IV offset = loc - RExC_precomp; \
489 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
490 (int)offset, RExC_precomp, RExC_precomp + offset); \
493 #define ckWARNregdep(loc,m) STMT_START { \
494 const IV offset = loc - RExC_precomp; \
495 Perl_ck_warner_d(aTHX_ packWARN2(WARN_DEPRECATED, WARN_REGEXP), \
497 (int)offset, RExC_precomp, RExC_precomp + offset); \
500 #define ckWARN2reg(loc, m, a1) STMT_START { \
501 const IV offset = loc - RExC_precomp; \
502 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
503 a1, (int)offset, RExC_precomp, RExC_precomp + offset); \
506 #define vWARN3(loc, m, a1, a2) STMT_START { \
507 const IV offset = loc - RExC_precomp; \
508 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
509 a1, a2, (int)offset, RExC_precomp, RExC_precomp + offset); \
512 #define ckWARN3reg(loc, m, a1, a2) STMT_START { \
513 const IV offset = loc - RExC_precomp; \
514 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
515 a1, a2, (int)offset, RExC_precomp, RExC_precomp + offset); \
518 #define vWARN4(loc, m, a1, a2, a3) STMT_START { \
519 const IV offset = loc - RExC_precomp; \
520 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
521 a1, a2, a3, (int)offset, RExC_precomp, RExC_precomp + offset); \
524 #define ckWARN4reg(loc, m, a1, a2, a3) STMT_START { \
525 const IV offset = loc - RExC_precomp; \
526 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
527 a1, a2, a3, (int)offset, RExC_precomp, RExC_precomp + offset); \
530 #define vWARN5(loc, m, a1, a2, a3, a4) STMT_START { \
531 const IV offset = loc - RExC_precomp; \
532 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
533 a1, a2, a3, a4, (int)offset, RExC_precomp, RExC_precomp + offset); \
537 /* Allow for side effects in s */
538 #define REGC(c,s) STMT_START { \
539 if (!SIZE_ONLY) *(s) = (c); else (void)(s); \
542 /* Macros for recording node offsets. 20001227 mjd@plover.com
543 * Nodes are numbered 1, 2, 3, 4. Node #n's position is recorded in
544 * element 2*n-1 of the array. Element #2n holds the byte length node #n.
545 * Element 0 holds the number n.
546 * Position is 1 indexed.
548 #ifndef RE_TRACK_PATTERN_OFFSETS
549 #define Set_Node_Offset_To_R(node,byte)
550 #define Set_Node_Offset(node,byte)
551 #define Set_Cur_Node_Offset
552 #define Set_Node_Length_To_R(node,len)
553 #define Set_Node_Length(node,len)
554 #define Set_Node_Cur_Length(node)
555 #define Node_Offset(n)
556 #define Node_Length(n)
557 #define Set_Node_Offset_Length(node,offset,len)
558 #define ProgLen(ri) ri->u.proglen
559 #define SetProgLen(ri,x) ri->u.proglen = x
561 #define ProgLen(ri) ri->u.offsets[0]
562 #define SetProgLen(ri,x) ri->u.offsets[0] = x
563 #define Set_Node_Offset_To_R(node,byte) STMT_START { \
565 MJD_OFFSET_DEBUG(("** (%d) offset of node %d is %d.\n", \
566 __LINE__, (int)(node), (int)(byte))); \
568 Perl_croak(aTHX_ "value of node is %d in Offset macro", (int)(node)); \
570 RExC_offsets[2*(node)-1] = (byte); \
575 #define Set_Node_Offset(node,byte) \
576 Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
577 #define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
579 #define Set_Node_Length_To_R(node,len) STMT_START { \
581 MJD_OFFSET_DEBUG(("** (%d) size of node %d is %d.\n", \
582 __LINE__, (int)(node), (int)(len))); \
584 Perl_croak(aTHX_ "value of node is %d in Length macro", (int)(node)); \
586 RExC_offsets[2*(node)] = (len); \
591 #define Set_Node_Length(node,len) \
592 Set_Node_Length_To_R((node)-RExC_emit_start, len)
593 #define Set_Cur_Node_Length(len) Set_Node_Length(RExC_emit, len)
594 #define Set_Node_Cur_Length(node) \
595 Set_Node_Length(node, RExC_parse - parse_start)
597 /* Get offsets and lengths */
598 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
599 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
601 #define Set_Node_Offset_Length(node,offset,len) STMT_START { \
602 Set_Node_Offset_To_R((node)-RExC_emit_start, (offset)); \
603 Set_Node_Length_To_R((node)-RExC_emit_start, (len)); \
607 #if PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS
608 #define EXPERIMENTAL_INPLACESCAN
609 #endif /*PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS*/
611 #define DEBUG_STUDYDATA(str,data,depth) \
612 DEBUG_OPTIMISE_MORE_r(if(data){ \
613 PerlIO_printf(Perl_debug_log, \
614 "%*s" str "Pos:%"IVdf"/%"IVdf \
615 " Flags: 0x%"UVXf" Whilem_c: %"IVdf" Lcp: %"IVdf" %s", \
616 (int)(depth)*2, "", \
617 (IV)((data)->pos_min), \
618 (IV)((data)->pos_delta), \
619 (UV)((data)->flags), \
620 (IV)((data)->whilem_c), \
621 (IV)((data)->last_closep ? *((data)->last_closep) : -1), \
622 is_inf ? "INF " : "" \
624 if ((data)->last_found) \
625 PerlIO_printf(Perl_debug_log, \
626 "Last:'%s' %"IVdf":%"IVdf"/%"IVdf" %sFixed:'%s' @ %"IVdf \
627 " %sFloat: '%s' @ %"IVdf"/%"IVdf"", \
628 SvPVX_const((data)->last_found), \
629 (IV)((data)->last_end), \
630 (IV)((data)->last_start_min), \
631 (IV)((data)->last_start_max), \
632 ((data)->longest && \
633 (data)->longest==&((data)->longest_fixed)) ? "*" : "", \
634 SvPVX_const((data)->longest_fixed), \
635 (IV)((data)->offset_fixed), \
636 ((data)->longest && \
637 (data)->longest==&((data)->longest_float)) ? "*" : "", \
638 SvPVX_const((data)->longest_float), \
639 (IV)((data)->offset_float_min), \
640 (IV)((data)->offset_float_max) \
642 PerlIO_printf(Perl_debug_log,"\n"); \
645 static void clear_re(pTHX_ void *r);
647 /* Mark that we cannot extend a found fixed substring at this point.
648 Update the longest found anchored substring and the longest found
649 floating substrings if needed. */
652 S_scan_commit(pTHX_ const RExC_state_t *pRExC_state, scan_data_t *data, I32 *minlenp, int is_inf)
654 const STRLEN l = CHR_SVLEN(data->last_found);
655 const STRLEN old_l = CHR_SVLEN(*data->longest);
656 GET_RE_DEBUG_FLAGS_DECL;
658 PERL_ARGS_ASSERT_SCAN_COMMIT;
660 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
661 SvSetMagicSV(*data->longest, data->last_found);
662 if (*data->longest == data->longest_fixed) {
663 data->offset_fixed = l ? data->last_start_min : data->pos_min;
664 if (data->flags & SF_BEFORE_EOL)
666 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
668 data->flags &= ~SF_FIX_BEFORE_EOL;
669 data->minlen_fixed=minlenp;
670 data->lookbehind_fixed=0;
672 else { /* *data->longest == data->longest_float */
673 data->offset_float_min = l ? data->last_start_min : data->pos_min;
674 data->offset_float_max = (l
675 ? data->last_start_max
676 : data->pos_min + data->pos_delta);
677 if (is_inf || (U32)data->offset_float_max > (U32)I32_MAX)
678 data->offset_float_max = I32_MAX;
679 if (data->flags & SF_BEFORE_EOL)
681 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
683 data->flags &= ~SF_FL_BEFORE_EOL;
684 data->minlen_float=minlenp;
685 data->lookbehind_float=0;
688 SvCUR_set(data->last_found, 0);
690 SV * const sv = data->last_found;
691 if (SvUTF8(sv) && SvMAGICAL(sv)) {
692 MAGIC * const mg = mg_find(sv, PERL_MAGIC_utf8);
698 data->flags &= ~SF_BEFORE_EOL;
699 DEBUG_STUDYDATA("commit: ",data,0);
702 /* Can match anything (initialization) */
704 S_cl_anything(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
706 PERL_ARGS_ASSERT_CL_ANYTHING;
708 ANYOF_CLASS_ZERO(cl);
709 ANYOF_BITMAP_SETALL(cl);
710 cl->flags = ANYOF_EOS|ANYOF_UNICODE_ALL;
712 cl->flags |= ANYOF_LOCALE;
713 cl->flags |= ANYOF_LOC_NONBITMAP_FOLD;
716 /* Can match anything (initialization) */
718 S_cl_is_anything(const struct regnode_charclass_class *cl)
722 PERL_ARGS_ASSERT_CL_IS_ANYTHING;
724 for (value = 0; value <= ANYOF_MAX; value += 2)
725 if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
727 if (!(cl->flags & ANYOF_UNICODE_ALL))
729 if (!ANYOF_BITMAP_TESTALLSET((const void*)cl))
734 /* Can match anything (initialization) */
736 S_cl_init(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
738 PERL_ARGS_ASSERT_CL_INIT;
740 Zero(cl, 1, struct regnode_charclass_class);
742 cl_anything(pRExC_state, cl);
746 S_cl_init_zero(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
748 PERL_ARGS_ASSERT_CL_INIT_ZERO;
750 Zero(cl, 1, struct regnode_charclass_class);
752 cl_anything(pRExC_state, cl);
754 cl->flags |= ANYOF_LOCALE;
757 /* 'And' a given class with another one. Can create false positives */
758 /* We assume that cl is not inverted */
760 S_cl_and(struct regnode_charclass_class *cl,
761 const struct regnode_charclass_class *and_with)
763 PERL_ARGS_ASSERT_CL_AND;
765 assert(and_with->type == ANYOF);
767 if (!(ANYOF_CLASS_TEST_ANY_SET(and_with))
768 && !(ANYOF_CLASS_TEST_ANY_SET(cl))
769 && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
770 && !(and_with->flags & ANYOF_LOC_NONBITMAP_FOLD)
771 && !(cl->flags & ANYOF_LOC_NONBITMAP_FOLD)) {
774 if (and_with->flags & ANYOF_INVERT)
775 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
776 cl->bitmap[i] &= ~and_with->bitmap[i];
778 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
779 cl->bitmap[i] &= and_with->bitmap[i];
780 } /* XXXX: logic is complicated otherwise, leave it along for a moment. */
781 if (!(and_with->flags & ANYOF_EOS))
782 cl->flags &= ~ANYOF_EOS;
784 if (!(and_with->flags & ANYOF_LOC_NONBITMAP_FOLD))
785 cl->flags &= ~ANYOF_LOC_NONBITMAP_FOLD;
787 if (cl->flags & ANYOF_UNICODE_ALL && and_with->flags & ANYOF_NONBITMAP &&
788 !(and_with->flags & ANYOF_INVERT)) {
789 cl->flags &= ~ANYOF_UNICODE_ALL;
790 cl->flags |= and_with->flags & ANYOF_NONBITMAP; /* field is 2 bits; use
793 ARG_SET(cl, ARG(and_with));
795 if (!(and_with->flags & ANYOF_UNICODE_ALL) &&
796 !(and_with->flags & ANYOF_INVERT))
797 cl->flags &= ~ANYOF_UNICODE_ALL;
798 if (!(and_with->flags & (ANYOF_NONBITMAP|ANYOF_UNICODE_ALL)) &&
799 !(and_with->flags & ANYOF_INVERT))
800 cl->flags &= ~ANYOF_NONBITMAP;
803 /* 'OR' a given class with another one. Can create false positives */
804 /* We assume that cl is not inverted */
806 S_cl_or(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, const struct regnode_charclass_class *or_with)
808 PERL_ARGS_ASSERT_CL_OR;
810 if (or_with->flags & ANYOF_INVERT) {
812 * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
813 * <= (B1 | !B2) | (CL1 | !CL2)
814 * which is wasteful if CL2 is small, but we ignore CL2:
815 * (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1
816 * XXXX Can we handle case-fold? Unclear:
817 * (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) =
818 * (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i'))
820 if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
821 && !(or_with->flags & ANYOF_LOC_NONBITMAP_FOLD)
822 && !(cl->flags & ANYOF_LOC_NONBITMAP_FOLD) ) {
825 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
826 cl->bitmap[i] |= ~or_with->bitmap[i];
827 } /* XXXX: logic is complicated otherwise */
829 cl_anything(pRExC_state, cl);
832 /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */
833 if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
834 && (!(or_with->flags & ANYOF_LOC_NONBITMAP_FOLD)
835 || (cl->flags & ANYOF_LOC_NONBITMAP_FOLD)) ) {
838 /* OR char bitmap and class bitmap separately */
839 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
840 cl->bitmap[i] |= or_with->bitmap[i];
841 if (ANYOF_CLASS_TEST_ANY_SET(or_with)) {
842 for (i = 0; i < ANYOF_CLASSBITMAP_SIZE; i++)
843 cl->classflags[i] |= or_with->classflags[i];
844 cl->flags |= ANYOF_CLASS;
847 else { /* XXXX: logic is complicated, leave it along for a moment. */
848 cl_anything(pRExC_state, cl);
851 if (or_with->flags & ANYOF_EOS)
852 cl->flags |= ANYOF_EOS;
854 if (or_with->flags & ANYOF_LOC_NONBITMAP_FOLD)
855 cl->flags |= ANYOF_LOC_NONBITMAP_FOLD;
857 /* If both nodes match something outside the bitmap, but what they match
858 * outside is not the same pointer, and hence not easily compared, give up
859 * and allow the start class to match everything outside the bitmap */
860 if (cl->flags & ANYOF_NONBITMAP && or_with->flags & ANYOF_NONBITMAP &&
861 ARG(cl) != ARG(or_with)) {
862 cl->flags |= ANYOF_UNICODE_ALL;
865 if (or_with->flags & ANYOF_UNICODE_ALL) {
866 cl->flags |= ANYOF_UNICODE_ALL;
870 #define TRIE_LIST_ITEM(state,idx) (trie->states[state].trans.list)[ idx ]
871 #define TRIE_LIST_CUR(state) ( TRIE_LIST_ITEM( state, 0 ).forid )
872 #define TRIE_LIST_LEN(state) ( TRIE_LIST_ITEM( state, 0 ).newstate )
873 #define TRIE_LIST_USED(idx) ( trie->states[state].trans.list ? (TRIE_LIST_CUR( idx ) - 1) : 0 )
878 dump_trie(trie,widecharmap,revcharmap)
879 dump_trie_interim_list(trie,widecharmap,revcharmap,next_alloc)
880 dump_trie_interim_table(trie,widecharmap,revcharmap,next_alloc)
882 These routines dump out a trie in a somewhat readable format.
883 The _interim_ variants are used for debugging the interim
884 tables that are used to generate the final compressed
885 representation which is what dump_trie expects.
887 Part of the reason for their existance is to provide a form
888 of documentation as to how the different representations function.
893 Dumps the final compressed table form of the trie to Perl_debug_log.
894 Used for debugging make_trie().
898 S_dump_trie(pTHX_ const struct _reg_trie_data *trie, HV *widecharmap,
899 AV *revcharmap, U32 depth)
902 SV *sv=sv_newmortal();
903 int colwidth= widecharmap ? 6 : 4;
905 GET_RE_DEBUG_FLAGS_DECL;
907 PERL_ARGS_ASSERT_DUMP_TRIE;
909 PerlIO_printf( Perl_debug_log, "%*sChar : %-6s%-6s%-4s ",
910 (int)depth * 2 + 2,"",
911 "Match","Base","Ofs" );
913 for( state = 0 ; state < trie->uniquecharcount ; state++ ) {
914 SV ** const tmp = av_fetch( revcharmap, state, 0);
916 PerlIO_printf( Perl_debug_log, "%*s",
918 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
919 PL_colors[0], PL_colors[1],
920 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
921 PERL_PV_ESCAPE_FIRSTCHAR
926 PerlIO_printf( Perl_debug_log, "\n%*sState|-----------------------",
927 (int)depth * 2 + 2,"");
929 for( state = 0 ; state < trie->uniquecharcount ; state++ )
930 PerlIO_printf( Perl_debug_log, "%.*s", colwidth, "--------");
931 PerlIO_printf( Perl_debug_log, "\n");
933 for( state = 1 ; state < trie->statecount ; state++ ) {
934 const U32 base = trie->states[ state ].trans.base;
936 PerlIO_printf( Perl_debug_log, "%*s#%4"UVXf"|", (int)depth * 2 + 2,"", (UV)state);
938 if ( trie->states[ state ].wordnum ) {
939 PerlIO_printf( Perl_debug_log, " W%4X", trie->states[ state ].wordnum );
941 PerlIO_printf( Perl_debug_log, "%6s", "" );
944 PerlIO_printf( Perl_debug_log, " @%4"UVXf" ", (UV)base );
949 while( ( base + ofs < trie->uniquecharcount ) ||
950 ( base + ofs - trie->uniquecharcount < trie->lasttrans
951 && trie->trans[ base + ofs - trie->uniquecharcount ].check != state))
954 PerlIO_printf( Perl_debug_log, "+%2"UVXf"[ ", (UV)ofs);
956 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
957 if ( ( base + ofs >= trie->uniquecharcount ) &&
958 ( base + ofs - trie->uniquecharcount < trie->lasttrans ) &&
959 trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
961 PerlIO_printf( Perl_debug_log, "%*"UVXf,
963 (UV)trie->trans[ base + ofs - trie->uniquecharcount ].next );
965 PerlIO_printf( Perl_debug_log, "%*s",colwidth," ." );
969 PerlIO_printf( Perl_debug_log, "]");
972 PerlIO_printf( Perl_debug_log, "\n" );
974 PerlIO_printf(Perl_debug_log, "%*sword_info N:(prev,len)=", (int)depth*2, "");
975 for (word=1; word <= trie->wordcount; word++) {
976 PerlIO_printf(Perl_debug_log, " %d:(%d,%d)",
977 (int)word, (int)(trie->wordinfo[word].prev),
978 (int)(trie->wordinfo[word].len));
980 PerlIO_printf(Perl_debug_log, "\n" );
983 Dumps a fully constructed but uncompressed trie in list form.
984 List tries normally only are used for construction when the number of
985 possible chars (trie->uniquecharcount) is very high.
986 Used for debugging make_trie().
989 S_dump_trie_interim_list(pTHX_ const struct _reg_trie_data *trie,
990 HV *widecharmap, AV *revcharmap, U32 next_alloc,
994 SV *sv=sv_newmortal();
995 int colwidth= widecharmap ? 6 : 4;
996 GET_RE_DEBUG_FLAGS_DECL;
998 PERL_ARGS_ASSERT_DUMP_TRIE_INTERIM_LIST;
1000 /* print out the table precompression. */
1001 PerlIO_printf( Perl_debug_log, "%*sState :Word | Transition Data\n%*s%s",
1002 (int)depth * 2 + 2,"", (int)depth * 2 + 2,"",
1003 "------:-----+-----------------\n" );
1005 for( state=1 ; state < next_alloc ; state ++ ) {
1008 PerlIO_printf( Perl_debug_log, "%*s %4"UVXf" :",
1009 (int)depth * 2 + 2,"", (UV)state );
1010 if ( ! trie->states[ state ].wordnum ) {
1011 PerlIO_printf( Perl_debug_log, "%5s| ","");
1013 PerlIO_printf( Perl_debug_log, "W%4x| ",
1014 trie->states[ state ].wordnum
1017 for( charid = 1 ; charid <= TRIE_LIST_USED( state ) ; charid++ ) {
1018 SV ** const tmp = av_fetch( revcharmap, TRIE_LIST_ITEM(state,charid).forid, 0);
1020 PerlIO_printf( Perl_debug_log, "%*s:%3X=%4"UVXf" | ",
1022 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
1023 PL_colors[0], PL_colors[1],
1024 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
1025 PERL_PV_ESCAPE_FIRSTCHAR
1027 TRIE_LIST_ITEM(state,charid).forid,
1028 (UV)TRIE_LIST_ITEM(state,charid).newstate
1031 PerlIO_printf(Perl_debug_log, "\n%*s| ",
1032 (int)((depth * 2) + 14), "");
1035 PerlIO_printf( Perl_debug_log, "\n");
1040 Dumps a fully constructed but uncompressed trie in table form.
1041 This is the normal DFA style state transition table, with a few
1042 twists to facilitate compression later.
1043 Used for debugging make_trie().
1046 S_dump_trie_interim_table(pTHX_ const struct _reg_trie_data *trie,
1047 HV *widecharmap, AV *revcharmap, U32 next_alloc,
1052 SV *sv=sv_newmortal();
1053 int colwidth= widecharmap ? 6 : 4;
1054 GET_RE_DEBUG_FLAGS_DECL;
1056 PERL_ARGS_ASSERT_DUMP_TRIE_INTERIM_TABLE;
1059 print out the table precompression so that we can do a visual check
1060 that they are identical.
1063 PerlIO_printf( Perl_debug_log, "%*sChar : ",(int)depth * 2 + 2,"" );
1065 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
1066 SV ** const tmp = av_fetch( revcharmap, charid, 0);
1068 PerlIO_printf( Perl_debug_log, "%*s",
1070 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
1071 PL_colors[0], PL_colors[1],
1072 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
1073 PERL_PV_ESCAPE_FIRSTCHAR
1079 PerlIO_printf( Perl_debug_log, "\n%*sState+-",(int)depth * 2 + 2,"" );
1081 for( charid=0 ; charid < trie->uniquecharcount ; charid++ ) {
1082 PerlIO_printf( Perl_debug_log, "%.*s", colwidth,"--------");
1085 PerlIO_printf( Perl_debug_log, "\n" );
1087 for( state=1 ; state < next_alloc ; state += trie->uniquecharcount ) {
1089 PerlIO_printf( Perl_debug_log, "%*s%4"UVXf" : ",
1090 (int)depth * 2 + 2,"",
1091 (UV)TRIE_NODENUM( state ) );
1093 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
1094 UV v=(UV)SAFE_TRIE_NODENUM( trie->trans[ state + charid ].next );
1096 PerlIO_printf( Perl_debug_log, "%*"UVXf, colwidth, v );
1098 PerlIO_printf( Perl_debug_log, "%*s", colwidth, "." );
1100 if ( ! trie->states[ TRIE_NODENUM( state ) ].wordnum ) {
1101 PerlIO_printf( Perl_debug_log, " (%4"UVXf")\n", (UV)trie->trans[ state ].check );
1103 PerlIO_printf( Perl_debug_log, " (%4"UVXf") W%4X\n", (UV)trie->trans[ state ].check,
1104 trie->states[ TRIE_NODENUM( state ) ].wordnum );
1112 /* make_trie(startbranch,first,last,tail,word_count,flags,depth)
1113 startbranch: the first branch in the whole branch sequence
1114 first : start branch of sequence of branch-exact nodes.
1115 May be the same as startbranch
1116 last : Thing following the last branch.
1117 May be the same as tail.
1118 tail : item following the branch sequence
1119 count : words in the sequence
1120 flags : currently the OP() type we will be building one of /EXACT(|F|Fl)/
1121 depth : indent depth
1123 Inplace optimizes a sequence of 2 or more Branch-Exact nodes into a TRIE node.
1125 A trie is an N'ary tree where the branches are determined by digital
1126 decomposition of the key. IE, at the root node you look up the 1st character and
1127 follow that branch repeat until you find the end of the branches. Nodes can be
1128 marked as "accepting" meaning they represent a complete word. Eg:
1132 would convert into the following structure. Numbers represent states, letters
1133 following numbers represent valid transitions on the letter from that state, if
1134 the number is in square brackets it represents an accepting state, otherwise it
1135 will be in parenthesis.
1137 +-h->+-e->[3]-+-r->(8)-+-s->[9]
1141 (1) +-i->(6)-+-s->[7]
1143 +-s->(3)-+-h->(4)-+-e->[5]
1145 Accept Word Mapping: 3=>1 (he),5=>2 (she), 7=>3 (his), 9=>4 (hers)
1147 This shows that when matching against the string 'hers' we will begin at state 1
1148 read 'h' and move to state 2, read 'e' and move to state 3 which is accepting,
1149 then read 'r' and go to state 8 followed by 's' which takes us to state 9 which
1150 is also accepting. Thus we know that we can match both 'he' and 'hers' with a
1151 single traverse. We store a mapping from accepting to state to which word was
1152 matched, and then when we have multiple possibilities we try to complete the
1153 rest of the regex in the order in which they occured in the alternation.
1155 The only prior NFA like behaviour that would be changed by the TRIE support is
1156 the silent ignoring of duplicate alternations which are of the form:
1158 / (DUPE|DUPE) X? (?{ ... }) Y /x
1160 Thus EVAL blocks following a trie may be called a different number of times with
1161 and without the optimisation. With the optimisations dupes will be silently
1162 ignored. This inconsistant behaviour of EVAL type nodes is well established as
1163 the following demonstrates:
1165 'words'=~/(word|word|word)(?{ print $1 })[xyz]/
1167 which prints out 'word' three times, but
1169 'words'=~/(word|word|word)(?{ print $1 })S/
1171 which doesnt print it out at all. This is due to other optimisations kicking in.
1173 Example of what happens on a structural level:
1175 The regexp /(ac|ad|ab)+/ will produce the folowing debug output:
1177 1: CURLYM[1] {1,32767}(18)
1188 This would be optimizable with startbranch=5, first=5, last=16, tail=16
1189 and should turn into:
1191 1: CURLYM[1] {1,32767}(18)
1193 [Words:3 Chars Stored:6 Unique Chars:4 States:5 NCP:1]
1201 Cases where tail != last would be like /(?foo|bar)baz/:
1211 which would be optimizable with startbranch=1, first=1, last=7, tail=8
1212 and would end up looking like:
1215 [Words:2 Chars Stored:6 Unique Chars:5 States:7 NCP:1]
1222 d = uvuni_to_utf8_flags(d, uv, 0);
1224 is the recommended Unicode-aware way of saying
1229 #define TRIE_STORE_REVCHAR \
1232 SV *zlopp = newSV(2); \
1233 unsigned char *flrbbbbb = (unsigned char *) SvPVX(zlopp); \
1234 unsigned const char *const kapow = uvuni_to_utf8(flrbbbbb, uvc & 0xFF); \
1235 SvCUR_set(zlopp, kapow - flrbbbbb); \
1238 av_push(revcharmap, zlopp); \
1240 char ooooff = (char)uvc; \
1241 av_push(revcharmap, newSVpvn(&ooooff, 1)); \
1245 #define TRIE_READ_CHAR STMT_START { \
1249 if ( foldlen > 0 ) { \
1250 uvc = utf8n_to_uvuni( scan, UTF8_MAXLEN, &len, uniflags ); \
1255 uvc = utf8n_to_uvuni( (const U8*)uc, UTF8_MAXLEN, &len, uniflags);\
1256 uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \
1257 foldlen -= UNISKIP( uvc ); \
1258 scan = foldbuf + UNISKIP( uvc ); \
1261 uvc = utf8n_to_uvuni( (const U8*)uc, UTF8_MAXLEN, &len, uniflags);\
1271 #define TRIE_LIST_PUSH(state,fid,ns) STMT_START { \
1272 if ( TRIE_LIST_CUR( state ) >=TRIE_LIST_LEN( state ) ) { \
1273 U32 ging = TRIE_LIST_LEN( state ) *= 2; \
1274 Renew( trie->states[ state ].trans.list, ging, reg_trie_trans_le ); \
1276 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).forid = fid; \
1277 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).newstate = ns; \
1278 TRIE_LIST_CUR( state )++; \
1281 #define TRIE_LIST_NEW(state) STMT_START { \
1282 Newxz( trie->states[ state ].trans.list, \
1283 4, reg_trie_trans_le ); \
1284 TRIE_LIST_CUR( state ) = 1; \
1285 TRIE_LIST_LEN( state ) = 4; \
1288 #define TRIE_HANDLE_WORD(state) STMT_START { \
1289 U16 dupe= trie->states[ state ].wordnum; \
1290 regnode * const noper_next = regnext( noper ); \
1293 /* store the word for dumping */ \
1295 if (OP(noper) != NOTHING) \
1296 tmp = newSVpvn_utf8(STRING(noper), STR_LEN(noper), UTF); \
1298 tmp = newSVpvn_utf8( "", 0, UTF ); \
1299 av_push( trie_words, tmp ); \
1303 trie->wordinfo[curword].prev = 0; \
1304 trie->wordinfo[curword].len = wordlen; \
1305 trie->wordinfo[curword].accept = state; \
1307 if ( noper_next < tail ) { \
1309 trie->jump = (U16 *) PerlMemShared_calloc( word_count + 1, sizeof(U16) ); \
1310 trie->jump[curword] = (U16)(noper_next - convert); \
1312 jumper = noper_next; \
1314 nextbranch= regnext(cur); \
1318 /* It's a dupe. Pre-insert into the wordinfo[].prev */\
1319 /* chain, so that when the bits of chain are later */\
1320 /* linked together, the dups appear in the chain */\
1321 trie->wordinfo[curword].prev = trie->wordinfo[dupe].prev; \
1322 trie->wordinfo[dupe].prev = curword; \
1324 /* we haven't inserted this word yet. */ \
1325 trie->states[ state ].wordnum = curword; \
1330 #define TRIE_TRANS_STATE(state,base,ucharcount,charid,special) \
1331 ( ( base + charid >= ucharcount \
1332 && base + charid < ubound \
1333 && state == trie->trans[ base - ucharcount + charid ].check \
1334 && trie->trans[ base - ucharcount + charid ].next ) \
1335 ? trie->trans[ base - ucharcount + charid ].next \
1336 : ( state==1 ? special : 0 ) \
1340 #define MADE_JUMP_TRIE 2
1341 #define MADE_EXACT_TRIE 4
1344 S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *first, regnode *last, regnode *tail, U32 word_count, U32 flags, U32 depth)
1347 /* first pass, loop through and scan words */
1348 reg_trie_data *trie;
1349 HV *widecharmap = NULL;
1350 AV *revcharmap = newAV();
1352 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1357 regnode *jumper = NULL;
1358 regnode *nextbranch = NULL;
1359 regnode *convert = NULL;
1360 U32 *prev_states; /* temp array mapping each state to previous one */
1361 /* we just use folder as a flag in utf8 */
1362 const U8 * folder = NULL;
1365 const U32 data_slot = add_data( pRExC_state, 4, "tuuu" );
1366 AV *trie_words = NULL;
1367 /* along with revcharmap, this only used during construction but both are
1368 * useful during debugging so we store them in the struct when debugging.
1371 const U32 data_slot = add_data( pRExC_state, 2, "tu" );
1372 STRLEN trie_charcount=0;
1374 SV *re_trie_maxbuff;
1375 GET_RE_DEBUG_FLAGS_DECL;
1377 PERL_ARGS_ASSERT_MAKE_TRIE;
1379 PERL_UNUSED_ARG(depth);
1383 case EXACTFU: folder = PL_fold_latin1; break;
1384 case EXACTF: folder = PL_fold; break;
1385 case EXACTFL: folder = PL_fold_locale; break;
1388 trie = (reg_trie_data *) PerlMemShared_calloc( 1, sizeof(reg_trie_data) );
1390 trie->startstate = 1;
1391 trie->wordcount = word_count;
1392 RExC_rxi->data->data[ data_slot ] = (void*)trie;
1393 trie->charmap = (U16 *) PerlMemShared_calloc( 256, sizeof(U16) );
1394 if (!(UTF && folder))
1395 trie->bitmap = (char *) PerlMemShared_calloc( ANYOF_BITMAP_SIZE, 1 );
1396 trie->wordinfo = (reg_trie_wordinfo *) PerlMemShared_calloc(
1397 trie->wordcount+1, sizeof(reg_trie_wordinfo));
1400 trie_words = newAV();
1403 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
1404 if (!SvIOK(re_trie_maxbuff)) {
1405 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
1408 PerlIO_printf( Perl_debug_log,
1409 "%*smake_trie start==%d, first==%d, last==%d, tail==%d depth=%d\n",
1410 (int)depth * 2 + 2, "",
1411 REG_NODE_NUM(startbranch),REG_NODE_NUM(first),
1412 REG_NODE_NUM(last), REG_NODE_NUM(tail),
1416 /* Find the node we are going to overwrite */
1417 if ( first == startbranch && OP( last ) != BRANCH ) {
1418 /* whole branch chain */
1421 /* branch sub-chain */
1422 convert = NEXTOPER( first );
1425 /* -- First loop and Setup --
1427 We first traverse the branches and scan each word to determine if it
1428 contains widechars, and how many unique chars there are, this is
1429 important as we have to build a table with at least as many columns as we
1432 We use an array of integers to represent the character codes 0..255
1433 (trie->charmap) and we use a an HV* to store Unicode characters. We use the
1434 native representation of the character value as the key and IV's for the
1437 *TODO* If we keep track of how many times each character is used we can
1438 remap the columns so that the table compression later on is more
1439 efficient in terms of memory by ensuring the most common value is in the
1440 middle and the least common are on the outside. IMO this would be better
1441 than a most to least common mapping as theres a decent chance the most
1442 common letter will share a node with the least common, meaning the node
1443 will not be compressable. With a middle is most common approach the worst
1444 case is when we have the least common nodes twice.
1448 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1449 regnode * const noper = NEXTOPER( cur );
1450 const U8 *uc = (U8*)STRING( noper );
1451 const U8 * const e = uc + STR_LEN( noper );
1453 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1454 const U8 *scan = (U8*)NULL;
1455 U32 wordlen = 0; /* required init */
1457 bool set_bit = trie->bitmap ? 1 : 0; /*store the first char in the bitmap?*/
1459 if (OP(noper) == NOTHING) {
1463 if ( set_bit ) /* bitmap only alloced when !(UTF&&Folding) */
1464 TRIE_BITMAP_SET(trie,*uc); /* store the raw first byte
1465 regardless of encoding */
1467 for ( ; uc < e ; uc += len ) {
1468 TRIE_CHARCOUNT(trie)++;
1472 if ( !trie->charmap[ uvc ] ) {
1473 trie->charmap[ uvc ]=( ++trie->uniquecharcount );
1475 trie->charmap[ folder[ uvc ] ] = trie->charmap[ uvc ];
1479 /* store the codepoint in the bitmap, and its folded
1481 TRIE_BITMAP_SET(trie,uvc);
1483 /* store the folded codepoint */
1484 if ( folder ) TRIE_BITMAP_SET(trie,folder[ uvc ]);
1487 /* store first byte of utf8 representation of
1488 variant codepoints */
1489 if (! UNI_IS_INVARIANT(uvc)) {
1490 TRIE_BITMAP_SET(trie, UTF8_TWO_BYTE_HI(uvc));
1493 set_bit = 0; /* We've done our bit :-) */
1498 widecharmap = newHV();
1500 svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 1 );
1503 Perl_croak( aTHX_ "error creating/fetching widecharmap entry for 0x%"UVXf, uvc );
1505 if ( !SvTRUE( *svpp ) ) {
1506 sv_setiv( *svpp, ++trie->uniquecharcount );
1511 if( cur == first ) {
1514 } else if (chars < trie->minlen) {
1516 } else if (chars > trie->maxlen) {
1520 } /* end first pass */
1521 DEBUG_TRIE_COMPILE_r(
1522 PerlIO_printf( Perl_debug_log, "%*sTRIE(%s): W:%d C:%d Uq:%d Min:%d Max:%d\n",
1523 (int)depth * 2 + 2,"",
1524 ( widecharmap ? "UTF8" : "NATIVE" ), (int)word_count,
1525 (int)TRIE_CHARCOUNT(trie), trie->uniquecharcount,
1526 (int)trie->minlen, (int)trie->maxlen )
1530 We now know what we are dealing with in terms of unique chars and
1531 string sizes so we can calculate how much memory a naive
1532 representation using a flat table will take. If it's over a reasonable
1533 limit (as specified by ${^RE_TRIE_MAXBUF}) we use a more memory
1534 conservative but potentially much slower representation using an array
1537 At the end we convert both representations into the same compressed
1538 form that will be used in regexec.c for matching with. The latter
1539 is a form that cannot be used to construct with but has memory
1540 properties similar to the list form and access properties similar
1541 to the table form making it both suitable for fast searches and
1542 small enough that its feasable to store for the duration of a program.
1544 See the comment in the code where the compressed table is produced
1545 inplace from the flat tabe representation for an explanation of how
1546 the compression works.
1551 Newx(prev_states, TRIE_CHARCOUNT(trie) + 2, U32);
1554 if ( (IV)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1) > SvIV(re_trie_maxbuff) ) {
1556 Second Pass -- Array Of Lists Representation
1558 Each state will be represented by a list of charid:state records
1559 (reg_trie_trans_le) the first such element holds the CUR and LEN
1560 points of the allocated array. (See defines above).
1562 We build the initial structure using the lists, and then convert
1563 it into the compressed table form which allows faster lookups
1564 (but cant be modified once converted).
1567 STRLEN transcount = 1;
1569 DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log,
1570 "%*sCompiling trie using list compiler\n",
1571 (int)depth * 2 + 2, ""));
1573 trie->states = (reg_trie_state *)
1574 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
1575 sizeof(reg_trie_state) );
1579 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1581 regnode * const noper = NEXTOPER( cur );
1582 U8 *uc = (U8*)STRING( noper );
1583 const U8 * const e = uc + STR_LEN( noper );
1584 U32 state = 1; /* required init */
1585 U16 charid = 0; /* sanity init */
1586 U8 *scan = (U8*)NULL; /* sanity init */
1587 STRLEN foldlen = 0; /* required init */
1588 U32 wordlen = 0; /* required init */
1589 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1591 if (OP(noper) != NOTHING) {
1592 for ( ; uc < e ; uc += len ) {
1597 charid = trie->charmap[ uvc ];
1599 SV** const svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 0);
1603 charid=(U16)SvIV( *svpp );
1606 /* charid is now 0 if we dont know the char read, or nonzero if we do */
1613 if ( !trie->states[ state ].trans.list ) {
1614 TRIE_LIST_NEW( state );
1616 for ( check = 1; check <= TRIE_LIST_USED( state ); check++ ) {
1617 if ( TRIE_LIST_ITEM( state, check ).forid == charid ) {
1618 newstate = TRIE_LIST_ITEM( state, check ).newstate;
1623 newstate = next_alloc++;
1624 prev_states[newstate] = state;
1625 TRIE_LIST_PUSH( state, charid, newstate );
1630 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %"IVdf, uvc );
1634 TRIE_HANDLE_WORD(state);
1636 } /* end second pass */
1638 /* next alloc is the NEXT state to be allocated */
1639 trie->statecount = next_alloc;
1640 trie->states = (reg_trie_state *)
1641 PerlMemShared_realloc( trie->states,
1643 * sizeof(reg_trie_state) );
1645 /* and now dump it out before we compress it */
1646 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_list(trie, widecharmap,
1647 revcharmap, next_alloc,
1651 trie->trans = (reg_trie_trans *)
1652 PerlMemShared_calloc( transcount, sizeof(reg_trie_trans) );
1659 for( state=1 ; state < next_alloc ; state ++ ) {
1663 DEBUG_TRIE_COMPILE_MORE_r(
1664 PerlIO_printf( Perl_debug_log, "tp: %d zp: %d ",tp,zp)
1668 if (trie->states[state].trans.list) {
1669 U16 minid=TRIE_LIST_ITEM( state, 1).forid;
1673 for( idx = 2 ; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
1674 const U16 forid = TRIE_LIST_ITEM( state, idx).forid;
1675 if ( forid < minid ) {
1677 } else if ( forid > maxid ) {
1681 if ( transcount < tp + maxid - minid + 1) {
1683 trie->trans = (reg_trie_trans *)
1684 PerlMemShared_realloc( trie->trans,
1686 * sizeof(reg_trie_trans) );
1687 Zero( trie->trans + (transcount / 2), transcount / 2 , reg_trie_trans );
1689 base = trie->uniquecharcount + tp - minid;
1690 if ( maxid == minid ) {
1692 for ( ; zp < tp ; zp++ ) {
1693 if ( ! trie->trans[ zp ].next ) {
1694 base = trie->uniquecharcount + zp - minid;
1695 trie->trans[ zp ].next = TRIE_LIST_ITEM( state, 1).newstate;
1696 trie->trans[ zp ].check = state;
1702 trie->trans[ tp ].next = TRIE_LIST_ITEM( state, 1).newstate;
1703 trie->trans[ tp ].check = state;
1708 for ( idx=1; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
1709 const U32 tid = base - trie->uniquecharcount + TRIE_LIST_ITEM( state, idx ).forid;
1710 trie->trans[ tid ].next = TRIE_LIST_ITEM( state, idx ).newstate;
1711 trie->trans[ tid ].check = state;
1713 tp += ( maxid - minid + 1 );
1715 Safefree(trie->states[ state ].trans.list);
1718 DEBUG_TRIE_COMPILE_MORE_r(
1719 PerlIO_printf( Perl_debug_log, " base: %d\n",base);
1722 trie->states[ state ].trans.base=base;
1724 trie->lasttrans = tp + 1;
1728 Second Pass -- Flat Table Representation.
1730 we dont use the 0 slot of either trans[] or states[] so we add 1 to each.
1731 We know that we will need Charcount+1 trans at most to store the data
1732 (one row per char at worst case) So we preallocate both structures
1733 assuming worst case.
1735 We then construct the trie using only the .next slots of the entry
1738 We use the .check field of the first entry of the node temporarily to
1739 make compression both faster and easier by keeping track of how many non
1740 zero fields are in the node.
1742 Since trans are numbered from 1 any 0 pointer in the table is a FAIL
1745 There are two terms at use here: state as a TRIE_NODEIDX() which is a
1746 number representing the first entry of the node, and state as a
1747 TRIE_NODENUM() which is the trans number. state 1 is TRIE_NODEIDX(1) and
1748 TRIE_NODENUM(1), state 2 is TRIE_NODEIDX(2) and TRIE_NODENUM(3) if there
1749 are 2 entrys per node. eg:
1757 The table is internally in the right hand, idx form. However as we also
1758 have to deal with the states array which is indexed by nodenum we have to
1759 use TRIE_NODENUM() to convert.
1762 DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log,
1763 "%*sCompiling trie using table compiler\n",
1764 (int)depth * 2 + 2, ""));
1766 trie->trans = (reg_trie_trans *)
1767 PerlMemShared_calloc( ( TRIE_CHARCOUNT(trie) + 1 )
1768 * trie->uniquecharcount + 1,
1769 sizeof(reg_trie_trans) );
1770 trie->states = (reg_trie_state *)
1771 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
1772 sizeof(reg_trie_state) );
1773 next_alloc = trie->uniquecharcount + 1;
1776 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1778 regnode * const noper = NEXTOPER( cur );
1779 const U8 *uc = (U8*)STRING( noper );
1780 const U8 * const e = uc + STR_LEN( noper );
1782 U32 state = 1; /* required init */
1784 U16 charid = 0; /* sanity init */
1785 U32 accept_state = 0; /* sanity init */
1786 U8 *scan = (U8*)NULL; /* sanity init */
1788 STRLEN foldlen = 0; /* required init */
1789 U32 wordlen = 0; /* required init */
1790 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1792 if ( OP(noper) != NOTHING ) {
1793 for ( ; uc < e ; uc += len ) {
1798 charid = trie->charmap[ uvc ];
1800 SV* const * const svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 0);
1801 charid = svpp ? (U16)SvIV(*svpp) : 0;
1805 if ( !trie->trans[ state + charid ].next ) {
1806 trie->trans[ state + charid ].next = next_alloc;
1807 trie->trans[ state ].check++;
1808 prev_states[TRIE_NODENUM(next_alloc)]
1809 = TRIE_NODENUM(state);
1810 next_alloc += trie->uniquecharcount;
1812 state = trie->trans[ state + charid ].next;
1814 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %"IVdf, uvc );
1816 /* charid is now 0 if we dont know the char read, or nonzero if we do */
1819 accept_state = TRIE_NODENUM( state );
1820 TRIE_HANDLE_WORD(accept_state);
1822 } /* end second pass */
1824 /* and now dump it out before we compress it */
1825 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_table(trie, widecharmap,
1827 next_alloc, depth+1));
1831 * Inplace compress the table.*
1833 For sparse data sets the table constructed by the trie algorithm will
1834 be mostly 0/FAIL transitions or to put it another way mostly empty.
1835 (Note that leaf nodes will not contain any transitions.)
1837 This algorithm compresses the tables by eliminating most such
1838 transitions, at the cost of a modest bit of extra work during lookup:
1840 - Each states[] entry contains a .base field which indicates the
1841 index in the state[] array wheres its transition data is stored.
1843 - If .base is 0 there are no valid transitions from that node.
1845 - If .base is nonzero then charid is added to it to find an entry in
1848 -If trans[states[state].base+charid].check!=state then the
1849 transition is taken to be a 0/Fail transition. Thus if there are fail
1850 transitions at the front of the node then the .base offset will point
1851 somewhere inside the previous nodes data (or maybe even into a node
1852 even earlier), but the .check field determines if the transition is
1856 The following process inplace converts the table to the compressed
1857 table: We first do not compress the root node 1,and mark all its
1858 .check pointers as 1 and set its .base pointer as 1 as well. This
1859 allows us to do a DFA construction from the compressed table later,
1860 and ensures that any .base pointers we calculate later are greater
1863 - We set 'pos' to indicate the first entry of the second node.
1865 - We then iterate over the columns of the node, finding the first and
1866 last used entry at l and m. We then copy l..m into pos..(pos+m-l),
1867 and set the .check pointers accordingly, and advance pos
1868 appropriately and repreat for the next node. Note that when we copy
1869 the next pointers we have to convert them from the original
1870 NODEIDX form to NODENUM form as the former is not valid post
1873 - If a node has no transitions used we mark its base as 0 and do not
1874 advance the pos pointer.
1876 - If a node only has one transition we use a second pointer into the
1877 structure to fill in allocated fail transitions from other states.
1878 This pointer is independent of the main pointer and scans forward
1879 looking for null transitions that are allocated to a state. When it
1880 finds one it writes the single transition into the "hole". If the
1881 pointer doesnt find one the single transition is appended as normal.
1883 - Once compressed we can Renew/realloc the structures to release the
1886 See "Table-Compression Methods" in sec 3.9 of the Red Dragon,
1887 specifically Fig 3.47 and the associated pseudocode.
1891 const U32 laststate = TRIE_NODENUM( next_alloc );
1894 trie->statecount = laststate;
1896 for ( state = 1 ; state < laststate ; state++ ) {
1898 const U32 stateidx = TRIE_NODEIDX( state );
1899 const U32 o_used = trie->trans[ stateidx ].check;
1900 U32 used = trie->trans[ stateidx ].check;
1901 trie->trans[ stateidx ].check = 0;
1903 for ( charid = 0 ; used && charid < trie->uniquecharcount ; charid++ ) {
1904 if ( flag || trie->trans[ stateidx + charid ].next ) {
1905 if ( trie->trans[ stateidx + charid ].next ) {
1907 for ( ; zp < pos ; zp++ ) {
1908 if ( ! trie->trans[ zp ].next ) {
1912 trie->states[ state ].trans.base = zp + trie->uniquecharcount - charid ;
1913 trie->trans[ zp ].next = SAFE_TRIE_NODENUM( trie->trans[ stateidx + charid ].next );
1914 trie->trans[ zp ].check = state;
1915 if ( ++zp > pos ) pos = zp;
1922 trie->states[ state ].trans.base = pos + trie->uniquecharcount - charid ;
1924 trie->trans[ pos ].next = SAFE_TRIE_NODENUM( trie->trans[ stateidx + charid ].next );
1925 trie->trans[ pos ].check = state;
1930 trie->lasttrans = pos + 1;
1931 trie->states = (reg_trie_state *)
1932 PerlMemShared_realloc( trie->states, laststate
1933 * sizeof(reg_trie_state) );
1934 DEBUG_TRIE_COMPILE_MORE_r(
1935 PerlIO_printf( Perl_debug_log,
1936 "%*sAlloc: %d Orig: %"IVdf" elements, Final:%"IVdf". Savings of %%%5.2f\n",
1937 (int)depth * 2 + 2,"",
1938 (int)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1 ),
1941 ( ( next_alloc - pos ) * 100 ) / (double)next_alloc );
1944 } /* end table compress */
1946 DEBUG_TRIE_COMPILE_MORE_r(
1947 PerlIO_printf(Perl_debug_log, "%*sStatecount:%"UVxf" Lasttrans:%"UVxf"\n",
1948 (int)depth * 2 + 2, "",
1949 (UV)trie->statecount,
1950 (UV)trie->lasttrans)
1952 /* resize the trans array to remove unused space */
1953 trie->trans = (reg_trie_trans *)
1954 PerlMemShared_realloc( trie->trans, trie->lasttrans
1955 * sizeof(reg_trie_trans) );
1957 { /* Modify the program and insert the new TRIE node */
1958 U8 nodetype =(U8)(flags & 0xFF);
1962 regnode *optimize = NULL;
1963 #ifdef RE_TRACK_PATTERN_OFFSETS
1966 U32 mjd_nodelen = 0;
1967 #endif /* RE_TRACK_PATTERN_OFFSETS */
1968 #endif /* DEBUGGING */
1970 This means we convert either the first branch or the first Exact,
1971 depending on whether the thing following (in 'last') is a branch
1972 or not and whther first is the startbranch (ie is it a sub part of
1973 the alternation or is it the whole thing.)
1974 Assuming its a sub part we convert the EXACT otherwise we convert
1975 the whole branch sequence, including the first.
1977 /* Find the node we are going to overwrite */
1978 if ( first != startbranch || OP( last ) == BRANCH ) {
1979 /* branch sub-chain */
1980 NEXT_OFF( first ) = (U16)(last - first);
1981 #ifdef RE_TRACK_PATTERN_OFFSETS
1983 mjd_offset= Node_Offset((convert));
1984 mjd_nodelen= Node_Length((convert));
1987 /* whole branch chain */
1989 #ifdef RE_TRACK_PATTERN_OFFSETS
1992 const regnode *nop = NEXTOPER( convert );
1993 mjd_offset= Node_Offset((nop));
1994 mjd_nodelen= Node_Length((nop));
1998 PerlIO_printf(Perl_debug_log, "%*sMJD offset:%"UVuf" MJD length:%"UVuf"\n",
1999 (int)depth * 2 + 2, "",
2000 (UV)mjd_offset, (UV)mjd_nodelen)
2003 /* But first we check to see if there is a common prefix we can
2004 split out as an EXACT and put in front of the TRIE node. */
2005 trie->startstate= 1;
2006 if ( trie->bitmap && !widecharmap && !trie->jump ) {
2008 for ( state = 1 ; state < trie->statecount-1 ; state++ ) {
2012 const U32 base = trie->states[ state ].trans.base;
2014 if ( trie->states[state].wordnum )
2017 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
2018 if ( ( base + ofs >= trie->uniquecharcount ) &&
2019 ( base + ofs - trie->uniquecharcount < trie->lasttrans ) &&
2020 trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
2022 if ( ++count > 1 ) {
2023 SV **tmp = av_fetch( revcharmap, ofs, 0);
2024 const U8 *ch = (U8*)SvPV_nolen_const( *tmp );
2025 if ( state == 1 ) break;
2027 Zero(trie->bitmap, ANYOF_BITMAP_SIZE, char);
2029 PerlIO_printf(Perl_debug_log,
2030 "%*sNew Start State=%"UVuf" Class: [",
2031 (int)depth * 2 + 2, "",
2034 SV ** const tmp = av_fetch( revcharmap, idx, 0);
2035 const U8 * const ch = (U8*)SvPV_nolen_const( *tmp );
2037 TRIE_BITMAP_SET(trie,*ch);
2039 TRIE_BITMAP_SET(trie, folder[ *ch ]);
2041 PerlIO_printf(Perl_debug_log, "%s", (char*)ch)
2045 TRIE_BITMAP_SET(trie,*ch);
2047 TRIE_BITMAP_SET(trie,folder[ *ch ]);
2048 DEBUG_OPTIMISE_r(PerlIO_printf( Perl_debug_log,"%s", ch));
2054 SV **tmp = av_fetch( revcharmap, idx, 0);
2056 char *ch = SvPV( *tmp, len );
2058 SV *sv=sv_newmortal();
2059 PerlIO_printf( Perl_debug_log,
2060 "%*sPrefix State: %"UVuf" Idx:%"UVuf" Char='%s'\n",
2061 (int)depth * 2 + 2, "",
2063 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 6,
2064 PL_colors[0], PL_colors[1],
2065 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
2066 PERL_PV_ESCAPE_FIRSTCHAR
2071 OP( convert ) = nodetype;
2072 str=STRING(convert);
2075 STR_LEN(convert) += len;
2081 DEBUG_OPTIMISE_r(PerlIO_printf( Perl_debug_log,"]\n"));
2086 trie->prefixlen = (state-1);
2088 regnode *n = convert+NODE_SZ_STR(convert);
2089 NEXT_OFF(convert) = NODE_SZ_STR(convert);
2090 trie->startstate = state;
2091 trie->minlen -= (state - 1);
2092 trie->maxlen -= (state - 1);
2094 /* At least the UNICOS C compiler choked on this
2095 * being argument to DEBUG_r(), so let's just have
2098 #ifdef PERL_EXT_RE_BUILD
2104 regnode *fix = convert;
2105 U32 word = trie->wordcount;
2107 Set_Node_Offset_Length(convert, mjd_offset, state - 1);
2108 while( ++fix < n ) {
2109 Set_Node_Offset_Length(fix, 0, 0);
2112 SV ** const tmp = av_fetch( trie_words, word, 0 );
2114 if ( STR_LEN(convert) <= SvCUR(*tmp) )
2115 sv_chop(*tmp, SvPV_nolen(*tmp) + STR_LEN(convert));
2117 sv_chop(*tmp, SvPV_nolen(*tmp) + SvCUR(*tmp));
2125 NEXT_OFF(convert) = (U16)(tail - convert);
2126 DEBUG_r(optimize= n);
2132 if ( trie->maxlen ) {
2133 NEXT_OFF( convert ) = (U16)(tail - convert);
2134 ARG_SET( convert, data_slot );
2135 /* Store the offset to the first unabsorbed branch in
2136 jump[0], which is otherwise unused by the jump logic.
2137 We use this when dumping a trie and during optimisation. */
2139 trie->jump[0] = (U16)(nextbranch - convert);
2141 /* If the start state is not accepting (meaning there is no empty string/NOTHING)
2142 * and there is a bitmap
2143 * and the first "jump target" node we found leaves enough room
2144 * then convert the TRIE node into a TRIEC node, with the bitmap
2145 * embedded inline in the opcode - this is hypothetically faster.
2147 if ( !trie->states[trie->startstate].wordnum
2149 && ( (char *)jumper - (char *)convert) >= (int)sizeof(struct regnode_charclass) )
2151 OP( convert ) = TRIEC;
2152 Copy(trie->bitmap, ((struct regnode_charclass *)convert)->bitmap, ANYOF_BITMAP_SIZE, char);
2153 PerlMemShared_free(trie->bitmap);
2156 OP( convert ) = TRIE;
2158 /* store the type in the flags */
2159 convert->flags = nodetype;
2163 + regarglen[ OP( convert ) ];
2165 /* XXX We really should free up the resource in trie now,
2166 as we won't use them - (which resources?) dmq */
2168 /* needed for dumping*/
2169 DEBUG_r(if (optimize) {
2170 regnode *opt = convert;
2172 while ( ++opt < optimize) {
2173 Set_Node_Offset_Length(opt,0,0);
2176 Try to clean up some of the debris left after the
2179 while( optimize < jumper ) {
2180 mjd_nodelen += Node_Length((optimize));
2181 OP( optimize ) = OPTIMIZED;
2182 Set_Node_Offset_Length(optimize,0,0);
2185 Set_Node_Offset_Length(convert,mjd_offset,mjd_nodelen);
2187 } /* end node insert */
2189 /* Finish populating the prev field of the wordinfo array. Walk back
2190 * from each accept state until we find another accept state, and if
2191 * so, point the first word's .prev field at the second word. If the
2192 * second already has a .prev field set, stop now. This will be the
2193 * case either if we've already processed that word's accept state,
2194 * or that state had multiple words, and the overspill words were
2195 * already linked up earlier.
2202 for (word=1; word <= trie->wordcount; word++) {
2204 if (trie->wordinfo[word].prev)
2206 state = trie->wordinfo[word].accept;
2208 state = prev_states[state];
2211 prev = trie->states[state].wordnum;
2215 trie->wordinfo[word].prev = prev;
2217 Safefree(prev_states);
2221 /* and now dump out the compressed format */
2222 DEBUG_TRIE_COMPILE_r(dump_trie(trie, widecharmap, revcharmap, depth+1));
2224 RExC_rxi->data->data[ data_slot + 1 ] = (void*)widecharmap;
2226 RExC_rxi->data->data[ data_slot + TRIE_WORDS_OFFSET ] = (void*)trie_words;
2227 RExC_rxi->data->data[ data_slot + 3 ] = (void*)revcharmap;
2229 SvREFCNT_dec(revcharmap);
2233 : trie->startstate>1
2239 S_make_trie_failtable(pTHX_ RExC_state_t *pRExC_state, regnode *source, regnode *stclass, U32 depth)
2241 /* The Trie is constructed and compressed now so we can build a fail array if it's needed
2243 This is basically the Aho-Corasick algorithm. Its from exercise 3.31 and 3.32 in the
2244 "Red Dragon" -- Compilers, principles, techniques, and tools. Aho, Sethi, Ullman 1985/88
2247 We find the fail state for each state in the trie, this state is the longest proper
2248 suffix of the current state's 'word' that is also a proper prefix of another word in our
2249 trie. State 1 represents the word '' and is thus the default fail state. This allows
2250 the DFA not to have to restart after its tried and failed a word at a given point, it
2251 simply continues as though it had been matching the other word in the first place.
2253 'abcdgu'=~/abcdefg|cdgu/
2254 When we get to 'd' we are still matching the first word, we would encounter 'g' which would
2255 fail, which would bring us to the state representing 'd' in the second word where we would
2256 try 'g' and succeed, proceeding to match 'cdgu'.
2258 /* add a fail transition */
2259 const U32 trie_offset = ARG(source);
2260 reg_trie_data *trie=(reg_trie_data *)RExC_rxi->data->data[trie_offset];
2262 const U32 ucharcount = trie->uniquecharcount;
2263 const U32 numstates = trie->statecount;
2264 const U32 ubound = trie->lasttrans + ucharcount;
2268 U32 base = trie->states[ 1 ].trans.base;
2271 const U32 data_slot = add_data( pRExC_state, 1, "T" );
2272 GET_RE_DEBUG_FLAGS_DECL;
2274 PERL_ARGS_ASSERT_MAKE_TRIE_FAILTABLE;
2276 PERL_UNUSED_ARG(depth);
2280 ARG_SET( stclass, data_slot );
2281 aho = (reg_ac_data *) PerlMemShared_calloc( 1, sizeof(reg_ac_data) );
2282 RExC_rxi->data->data[ data_slot ] = (void*)aho;
2283 aho->trie=trie_offset;
2284 aho->states=(reg_trie_state *)PerlMemShared_malloc( numstates * sizeof(reg_trie_state) );
2285 Copy( trie->states, aho->states, numstates, reg_trie_state );
2286 Newxz( q, numstates, U32);
2287 aho->fail = (U32 *) PerlMemShared_calloc( numstates, sizeof(U32) );
2290 /* initialize fail[0..1] to be 1 so that we always have
2291 a valid final fail state */
2292 fail[ 0 ] = fail[ 1 ] = 1;
2294 for ( charid = 0; charid < ucharcount ; charid++ ) {
2295 const U32 newstate = TRIE_TRANS_STATE( 1, base, ucharcount, charid, 0 );
2297 q[ q_write ] = newstate;
2298 /* set to point at the root */
2299 fail[ q[ q_write++ ] ]=1;
2302 while ( q_read < q_write) {
2303 const U32 cur = q[ q_read++ % numstates ];
2304 base = trie->states[ cur ].trans.base;
2306 for ( charid = 0 ; charid < ucharcount ; charid++ ) {
2307 const U32 ch_state = TRIE_TRANS_STATE( cur, base, ucharcount, charid, 1 );
2309 U32 fail_state = cur;
2312 fail_state = fail[ fail_state ];
2313 fail_base = aho->states[ fail_state ].trans.base;
2314 } while ( !TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 ) );
2316 fail_state = TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 );
2317 fail[ ch_state ] = fail_state;
2318 if ( !aho->states[ ch_state ].wordnum && aho->states[ fail_state ].wordnum )
2320 aho->states[ ch_state ].wordnum = aho->states[ fail_state ].wordnum;
2322 q[ q_write++ % numstates] = ch_state;
2326 /* restore fail[0..1] to 0 so that we "fall out" of the AC loop
2327 when we fail in state 1, this allows us to use the
2328 charclass scan to find a valid start char. This is based on the principle
2329 that theres a good chance the string being searched contains lots of stuff
2330 that cant be a start char.
2332 fail[ 0 ] = fail[ 1 ] = 0;
2333 DEBUG_TRIE_COMPILE_r({
2334 PerlIO_printf(Perl_debug_log,
2335 "%*sStclass Failtable (%"UVuf" states): 0",
2336 (int)(depth * 2), "", (UV)numstates
2338 for( q_read=1; q_read<numstates; q_read++ ) {
2339 PerlIO_printf(Perl_debug_log, ", %"UVuf, (UV)fail[q_read]);
2341 PerlIO_printf(Perl_debug_log, "\n");
2344 /*RExC_seen |= REG_SEEN_TRIEDFA;*/
2349 * There are strange code-generation bugs caused on sparc64 by gcc-2.95.2.
2350 * These need to be revisited when a newer toolchain becomes available.
2352 #if defined(__sparc64__) && defined(__GNUC__)
2353 # if __GNUC__ < 2 || (__GNUC__ == 2 && __GNUC_MINOR__ < 96)
2354 # undef SPARC64_GCC_WORKAROUND
2355 # define SPARC64_GCC_WORKAROUND 1
2359 #define DEBUG_PEEP(str,scan,depth) \
2360 DEBUG_OPTIMISE_r({if (scan){ \
2361 SV * const mysv=sv_newmortal(); \
2362 regnode *Next = regnext(scan); \
2363 regprop(RExC_rx, mysv, scan); \
2364 PerlIO_printf(Perl_debug_log, "%*s" str ">%3d: %s (%d)\n", \
2365 (int)depth*2, "", REG_NODE_NUM(scan), SvPV_nolen_const(mysv),\
2366 Next ? (REG_NODE_NUM(Next)) : 0 ); \
2373 #define JOIN_EXACT(scan,min,flags) \
2374 if (PL_regkind[OP(scan)] == EXACT) \
2375 join_exact(pRExC_state,(scan),(min),(flags),NULL,depth+1)
2378 S_join_exact(pTHX_ RExC_state_t *pRExC_state, regnode *scan, I32 *min, U32 flags,regnode *val, U32 depth) {
2379 /* Merge several consecutive EXACTish nodes into one. */
2380 regnode *n = regnext(scan);
2382 regnode *next = scan + NODE_SZ_STR(scan);
2386 regnode *stop = scan;
2387 GET_RE_DEBUG_FLAGS_DECL;
2389 PERL_UNUSED_ARG(depth);
2392 PERL_ARGS_ASSERT_JOIN_EXACT;
2393 #ifndef EXPERIMENTAL_INPLACESCAN
2394 PERL_UNUSED_ARG(flags);
2395 PERL_UNUSED_ARG(val);
2397 DEBUG_PEEP("join",scan,depth);
2399 /* Skip NOTHING, merge EXACT*. */
2401 ( PL_regkind[OP(n)] == NOTHING ||
2402 (stringok && (OP(n) == OP(scan))))
2404 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
2406 if (OP(n) == TAIL || n > next)
2408 if (PL_regkind[OP(n)] == NOTHING) {
2409 DEBUG_PEEP("skip:",n,depth);
2410 NEXT_OFF(scan) += NEXT_OFF(n);
2411 next = n + NODE_STEP_REGNODE;
2418 else if (stringok) {
2419 const unsigned int oldl = STR_LEN(scan);
2420 regnode * const nnext = regnext(n);
2422 DEBUG_PEEP("merg",n,depth);
2425 if (oldl + STR_LEN(n) > U8_MAX)
2427 NEXT_OFF(scan) += NEXT_OFF(n);
2428 STR_LEN(scan) += STR_LEN(n);
2429 next = n + NODE_SZ_STR(n);
2430 /* Now we can overwrite *n : */
2431 Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
2439 #ifdef EXPERIMENTAL_INPLACESCAN
2440 if (flags && !NEXT_OFF(n)) {
2441 DEBUG_PEEP("atch", val, depth);
2442 if (reg_off_by_arg[OP(n)]) {
2443 ARG_SET(n, val - n);
2446 NEXT_OFF(n) = val - n;
2452 #define GREEK_SMALL_LETTER_IOTA_WITH_DIALYTIKA_AND_TONOS 0x0390
2453 #define IOTA_D_T GREEK_SMALL_LETTER_IOTA_WITH_DIALYTIKA_AND_TONOS
2454 #define GREEK_SMALL_LETTER_UPSILON_WITH_DIALYTIKA_AND_TONOS 0x03B0
2455 #define UPSILON_D_T GREEK_SMALL_LETTER_UPSILON_WITH_DIALYTIKA_AND_TONOS
2458 && ( OP(scan) == EXACTF || OP(scan) == EXACTFU)
2459 && ( STR_LEN(scan) >= 6 ) )
2462 Two problematic code points in Unicode casefolding of EXACT nodes:
2464 U+0390 - GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
2465 U+03B0 - GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
2471 U+03B9 U+0308 U+0301 0xCE 0xB9 0xCC 0x88 0xCC 0x81
2472 U+03C5 U+0308 U+0301 0xCF 0x85 0xCC 0x88 0xCC 0x81
2474 This means that in case-insensitive matching (or "loose matching",
2475 as Unicode calls it), an EXACTF of length six (the UTF-8 encoded byte
2476 length of the above casefolded versions) can match a target string
2477 of length two (the byte length of UTF-8 encoded U+0390 or U+03B0).
2478 This would rather mess up the minimum length computation.
2480 What we'll do is to look for the tail four bytes, and then peek
2481 at the preceding two bytes to see whether we need to decrease
2482 the minimum length by four (six minus two).
2484 Thanks to the design of UTF-8, there cannot be false matches:
2485 A sequence of valid UTF-8 bytes cannot be a subsequence of
2486 another valid sequence of UTF-8 bytes.
2489 char * const s0 = STRING(scan), *s, *t;
2490 char * const s1 = s0 + STR_LEN(scan) - 1;
2491 char * const s2 = s1 - 4;
2492 #ifdef EBCDIC /* RD tunifold greek 0390 and 03B0 */
2493 const char t0[] = "\xaf\x49\xaf\x42";
2495 const char t0[] = "\xcc\x88\xcc\x81";
2497 const char * const t1 = t0 + 3;
2500 s < s2 && (t = ninstr(s, s1, t0, t1));
2503 if (((U8)t[-1] == 0x68 && (U8)t[-2] == 0xB4) ||
2504 ((U8)t[-1] == 0x46 && (U8)t[-2] == 0xB5))
2506 if (((U8)t[-1] == 0xB9 && (U8)t[-2] == 0xCE) ||
2507 ((U8)t[-1] == 0x85 && (U8)t[-2] == 0xCF))
2515 n = scan + NODE_SZ_STR(scan);
2517 if (PL_regkind[OP(n)] != NOTHING || OP(n) == NOTHING) {
2524 DEBUG_OPTIMISE_r(if (merged){DEBUG_PEEP("finl",scan,depth)});
2528 /* REx optimizer. Converts nodes into quickier variants "in place".
2529 Finds fixed substrings. */
2531 /* Stops at toplevel WHILEM as well as at "last". At end *scanp is set
2532 to the position after last scanned or to NULL. */
2534 #define INIT_AND_WITHP \
2535 assert(!and_withp); \
2536 Newx(and_withp,1,struct regnode_charclass_class); \
2537 SAVEFREEPV(and_withp)
2539 /* this is a chain of data about sub patterns we are processing that
2540 need to be handled seperately/specially in study_chunk. Its so
2541 we can simulate recursion without losing state. */
2543 typedef struct scan_frame {
2544 regnode *last; /* last node to process in this frame */
2545 regnode *next; /* next node to process when last is reached */
2546 struct scan_frame *prev; /*previous frame*/
2547 I32 stop; /* what stopparen do we use */
2551 #define SCAN_COMMIT(s, data, m) scan_commit(s, data, m, is_inf)
2553 #define CASE_SYNST_FNC(nAmE) \
2555 if (flags & SCF_DO_STCLASS_AND) { \
2556 for (value = 0; value < 256; value++) \
2557 if (!is_ ## nAmE ## _cp(value)) \
2558 ANYOF_BITMAP_CLEAR(data->start_class, value); \
2561 for (value = 0; value < 256; value++) \
2562 if (is_ ## nAmE ## _cp(value)) \
2563 ANYOF_BITMAP_SET(data->start_class, value); \
2567 if (flags & SCF_DO_STCLASS_AND) { \
2568 for (value = 0; value < 256; value++) \
2569 if (is_ ## nAmE ## _cp(value)) \
2570 ANYOF_BITMAP_CLEAR(data->start_class, value); \
2573 for (value = 0; value < 256; value++) \
2574 if (!is_ ## nAmE ## _cp(value)) \
2575 ANYOF_BITMAP_SET(data->start_class, value); \
2582 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
2583 I32 *minlenp, I32 *deltap,
2588 struct regnode_charclass_class *and_withp,
2589 U32 flags, U32 depth)
2590 /* scanp: Start here (read-write). */
2591 /* deltap: Write maxlen-minlen here. */
2592 /* last: Stop before this one. */
2593 /* data: string data about the pattern */
2594 /* stopparen: treat close N as END */
2595 /* recursed: which subroutines have we recursed into */
2596 /* and_withp: Valid if flags & SCF_DO_STCLASS_OR */
2599 I32 min = 0, pars = 0, code;
2600 regnode *scan = *scanp, *next;
2602 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
2603 int is_inf_internal = 0; /* The studied chunk is infinite */
2604 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
2605 scan_data_t data_fake;
2606 SV *re_trie_maxbuff = NULL;
2607 regnode *first_non_open = scan;
2608 I32 stopmin = I32_MAX;
2609 scan_frame *frame = NULL;
2610 GET_RE_DEBUG_FLAGS_DECL;
2612 PERL_ARGS_ASSERT_STUDY_CHUNK;
2615 StructCopy(&zero_scan_data, &data_fake, scan_data_t);
2619 while (first_non_open && OP(first_non_open) == OPEN)
2620 first_non_open=regnext(first_non_open);
2625 while ( scan && OP(scan) != END && scan < last ){
2626 /* Peephole optimizer: */
2627 DEBUG_STUDYDATA("Peep:", data,depth);
2628 DEBUG_PEEP("Peep",scan,depth);
2629 JOIN_EXACT(scan,&min,0);
2631 /* Follow the next-chain of the current node and optimize
2632 away all the NOTHINGs from it. */
2633 if (OP(scan) != CURLYX) {
2634 const int max = (reg_off_by_arg[OP(scan)]
2636 /* I32 may be smaller than U16 on CRAYs! */
2637 : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
2638 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
2642 /* Skip NOTHING and LONGJMP. */
2643 while ((n = regnext(n))
2644 && ((PL_regkind[OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
2645 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
2646 && off + noff < max)
2648 if (reg_off_by_arg[OP(scan)])
2651 NEXT_OFF(scan) = off;
2656 /* The principal pseudo-switch. Cannot be a switch, since we
2657 look into several different things. */
2658 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
2659 || OP(scan) == IFTHEN) {
2660 next = regnext(scan);
2662 /* demq: the op(next)==code check is to see if we have "branch-branch" AFAICT */
2664 if (OP(next) == code || code == IFTHEN) {
2665 /* NOTE - There is similar code to this block below for handling
2666 TRIE nodes on a re-study. If you change stuff here check there
2668 I32 max1 = 0, min1 = I32_MAX, num = 0;
2669 struct regnode_charclass_class accum;
2670 regnode * const startbranch=scan;
2672 if (flags & SCF_DO_SUBSTR)
2673 SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot merge strings after this. */
2674 if (flags & SCF_DO_STCLASS)
2675 cl_init_zero(pRExC_state, &accum);
2677 while (OP(scan) == code) {
2678 I32 deltanext, minnext, f = 0, fake;
2679 struct regnode_charclass_class this_class;
2682 data_fake.flags = 0;
2684 data_fake.whilem_c = data->whilem_c;
2685 data_fake.last_closep = data->last_closep;
2688 data_fake.last_closep = &fake;
2690 data_fake.pos_delta = delta;
2691 next = regnext(scan);
2692 scan = NEXTOPER(scan);
2694 scan = NEXTOPER(scan);
2695 if (flags & SCF_DO_STCLASS) {
2696 cl_init(pRExC_state, &this_class);
2697 data_fake.start_class = &this_class;
2698 f = SCF_DO_STCLASS_AND;
2700 if (flags & SCF_WHILEM_VISITED_POS)
2701 f |= SCF_WHILEM_VISITED_POS;
2703 /* we suppose the run is continuous, last=next...*/
2704 minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext,
2706 stopparen, recursed, NULL, f,depth+1);
2709 if (max1 < minnext + deltanext)
2710 max1 = minnext + deltanext;
2711 if (deltanext == I32_MAX)
2712 is_inf = is_inf_internal = 1;
2714 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
2716 if (data_fake.flags & SCF_SEEN_ACCEPT) {
2717 if ( stopmin > minnext)
2718 stopmin = min + min1;
2719 flags &= ~SCF_DO_SUBSTR;
2721 data->flags |= SCF_SEEN_ACCEPT;
2724 if (data_fake.flags & SF_HAS_EVAL)
2725 data->flags |= SF_HAS_EVAL;
2726 data->whilem_c = data_fake.whilem_c;
2728 if (flags & SCF_DO_STCLASS)
2729 cl_or(pRExC_state, &accum, &this_class);
2731 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
2733 if (flags & SCF_DO_SUBSTR) {
2734 data->pos_min += min1;
2735 data->pos_delta += max1 - min1;
2736 if (max1 != min1 || is_inf)
2737 data->longest = &(data->longest_float);
2740 delta += max1 - min1;
2741 if (flags & SCF_DO_STCLASS_OR) {
2742 cl_or(pRExC_state, data->start_class, &accum);
2744 cl_and(data->start_class, and_withp);
2745 flags &= ~SCF_DO_STCLASS;
2748 else if (flags & SCF_DO_STCLASS_AND) {
2750 cl_and(data->start_class, &accum);
2751 flags &= ~SCF_DO_STCLASS;
2754 /* Switch to OR mode: cache the old value of
2755 * data->start_class */
2757 StructCopy(data->start_class, and_withp,
2758 struct regnode_charclass_class);
2759 flags &= ~SCF_DO_STCLASS_AND;
2760 StructCopy(&accum, data->start_class,
2761 struct regnode_charclass_class);
2762 flags |= SCF_DO_STCLASS_OR;
2763 data->start_class->flags |= ANYOF_EOS;
2767 if (PERL_ENABLE_TRIE_OPTIMISATION && OP( startbranch ) == BRANCH ) {
2770 Assuming this was/is a branch we are dealing with: 'scan' now
2771 points at the item that follows the branch sequence, whatever
2772 it is. We now start at the beginning of the sequence and look
2779 which would be constructed from a pattern like /A|LIST|OF|WORDS/
2781 If we can find such a subseqence we need to turn the first
2782 element into a trie and then add the subsequent branch exact
2783 strings to the trie.
2787 1. patterns where the whole set of branches can be converted.
2789 2. patterns where only a subset can be converted.
2791 In case 1 we can replace the whole set with a single regop
2792 for the trie. In case 2 we need to keep the start and end
2795 'BRANCH EXACT; BRANCH EXACT; BRANCH X'
2796 becomes BRANCH TRIE; BRANCH X;
2798 There is an additional case, that being where there is a
2799 common prefix, which gets split out into an EXACT like node
2800 preceding the TRIE node.
2802 If x(1..n)==tail then we can do a simple trie, if not we make
2803 a "jump" trie, such that when we match the appropriate word
2804 we "jump" to the appopriate tail node. Essentailly we turn
2805 a nested if into a case structure of sorts.
2810 if (!re_trie_maxbuff) {
2811 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
2812 if (!SvIOK(re_trie_maxbuff))
2813 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
2815 if ( SvIV(re_trie_maxbuff)>=0 ) {
2817 regnode *first = (regnode *)NULL;
2818 regnode *last = (regnode *)NULL;
2819 regnode *tail = scan;
2824 SV * const mysv = sv_newmortal(); /* for dumping */
2826 /* var tail is used because there may be a TAIL
2827 regop in the way. Ie, the exacts will point to the
2828 thing following the TAIL, but the last branch will
2829 point at the TAIL. So we advance tail. If we
2830 have nested (?:) we may have to move through several
2834 while ( OP( tail ) == TAIL ) {
2835 /* this is the TAIL generated by (?:) */
2836 tail = regnext( tail );
2841 regprop(RExC_rx, mysv, tail );
2842 PerlIO_printf( Perl_debug_log, "%*s%s%s\n",
2843 (int)depth * 2 + 2, "",
2844 "Looking for TRIE'able sequences. Tail node is: ",
2845 SvPV_nolen_const( mysv )
2851 step through the branches, cur represents each
2852 branch, noper is the first thing to be matched
2853 as part of that branch and noper_next is the
2854 regnext() of that node. if noper is an EXACT
2855 and noper_next is the same as scan (our current
2856 position in the regex) then the EXACT branch is
2857 a possible optimization target. Once we have
2858 two or more consequetive such branches we can
2859 create a trie of the EXACT's contents and stich
2860 it in place. If the sequence represents all of
2861 the branches we eliminate the whole thing and
2862 replace it with a single TRIE. If it is a
2863 subsequence then we need to stitch it in. This
2864 means the first branch has to remain, and needs
2865 to be repointed at the item on the branch chain
2866 following the last branch optimized. This could
2867 be either a BRANCH, in which case the
2868 subsequence is internal, or it could be the
2869 item following the branch sequence in which
2870 case the subsequence is at the end.
2874 /* dont use tail as the end marker for this traverse */
2875 for ( cur = startbranch ; cur != scan ; cur = regnext( cur ) ) {
2876 regnode * const noper = NEXTOPER( cur );
2877 #if defined(DEBUGGING) || defined(NOJUMPTRIE)
2878 regnode * const noper_next = regnext( noper );
2882 regprop(RExC_rx, mysv, cur);
2883 PerlIO_printf( Perl_debug_log, "%*s- %s (%d)",
2884 (int)depth * 2 + 2,"", SvPV_nolen_const( mysv ), REG_NODE_NUM(cur) );
2886 regprop(RExC_rx, mysv, noper);
2887 PerlIO_printf( Perl_debug_log, " -> %s",
2888 SvPV_nolen_const(mysv));
2891 regprop(RExC_rx, mysv, noper_next );
2892 PerlIO_printf( Perl_debug_log,"\t=> %s\t",
2893 SvPV_nolen_const(mysv));
2895 PerlIO_printf( Perl_debug_log, "(First==%d,Last==%d,Cur==%d)\n",
2896 REG_NODE_NUM(first), REG_NODE_NUM(last), REG_NODE_NUM(cur) );
2898 if ( (((first && optype!=NOTHING) ? OP( noper ) == optype
2899 : PL_regkind[ OP( noper ) ] == EXACT )
2900 || OP(noper) == NOTHING )
2902 && noper_next == tail
2907 if ( !first || optype == NOTHING ) {
2908 if (!first) first = cur;
2909 optype = OP( noper );
2915 Currently we do not believe that the trie logic can
2916 handle case insensitive matching properly when the
2917 pattern is not unicode (thus forcing unicode semantics).
2919 If/when this is fixed the following define can be swapped
2920 in below to fully enable trie logic.
2922 #define TRIE_TYPE_IS_SAFE 1
2925 #define TRIE_TYPE_IS_SAFE (UTF || optype==EXACT)
2927 if ( last && TRIE_TYPE_IS_SAFE ) {
2928 make_trie( pRExC_state,
2929 startbranch, first, cur, tail, count,
2932 if ( PL_regkind[ OP( noper ) ] == EXACT
2934 && noper_next == tail
2939 optype = OP( noper );
2949 regprop(RExC_rx, mysv, cur);
2950 PerlIO_printf( Perl_debug_log,
2951 "%*s- %s (%d) <SCAN FINISHED>\n", (int)depth * 2 + 2,
2952 "", SvPV_nolen_const( mysv ),REG_NODE_NUM(cur));
2956 if ( last && TRIE_TYPE_IS_SAFE ) {
2957 made= make_trie( pRExC_state, startbranch, first, scan, tail, count, optype, depth+1 );
2958 #ifdef TRIE_STUDY_OPT
2959 if ( ((made == MADE_EXACT_TRIE &&
2960 startbranch == first)
2961 || ( first_non_open == first )) &&
2963 flags |= SCF_TRIE_RESTUDY;
2964 if ( startbranch == first
2967 RExC_seen &=~REG_TOP_LEVEL_BRANCHES;
2977 else if ( code == BRANCHJ ) { /* single branch is optimized. */
2978 scan = NEXTOPER(NEXTOPER(scan));
2979 } else /* single branch is optimized. */
2980 scan = NEXTOPER(scan);
2982 } else if (OP(scan) == SUSPEND || OP(scan) == GOSUB || OP(scan) == GOSTART) {
2983 scan_frame *newframe = NULL;
2988 if (OP(scan) != SUSPEND) {
2989 /* set the pointer */
2990 if (OP(scan) == GOSUB) {
2992 RExC_recurse[ARG2L(scan)] = scan;
2993 start = RExC_open_parens[paren-1];
2994 end = RExC_close_parens[paren-1];
2997 start = RExC_rxi->program + 1;
3001 Newxz(recursed, (((RExC_npar)>>3) +1), U8);
3002 SAVEFREEPV(recursed);
3004 if (!PAREN_TEST(recursed,paren+1)) {
3005 PAREN_SET(recursed,paren+1);
3006 Newx(newframe,1,scan_frame);
3008 if (flags & SCF_DO_SUBSTR) {
3009 SCAN_COMMIT(pRExC_state,data,minlenp);
3010 data->longest = &(data->longest_float);
3012 is_inf = is_inf_internal = 1;
3013 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
3014 cl_anything(pRExC_state, data->start_class);
3015 flags &= ~SCF_DO_STCLASS;
3018 Newx(newframe,1,scan_frame);
3021 end = regnext(scan);
3026 SAVEFREEPV(newframe);
3027 newframe->next = regnext(scan);
3028 newframe->last = last;
3029 newframe->stop = stopparen;
3030 newframe->prev = frame;
3040 else if (OP(scan) == EXACT) {
3041 I32 l = STR_LEN(scan);
3044 const U8 * const s = (U8*)STRING(scan);
3045 l = utf8_length(s, s + l);
3046 uc = utf8_to_uvchr(s, NULL);
3048 uc = *((U8*)STRING(scan));
3051 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
3052 /* The code below prefers earlier match for fixed
3053 offset, later match for variable offset. */
3054 if (data->last_end == -1) { /* Update the start info. */
3055 data->last_start_min = data->pos_min;
3056 data->last_start_max = is_inf
3057 ? I32_MAX : data->pos_min + data->pos_delta;
3059 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
3061 SvUTF8_on(data->last_found);
3063 SV * const sv = data->last_found;
3064 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
3065 mg_find(sv, PERL_MAGIC_utf8) : NULL;
3066 if (mg && mg->mg_len >= 0)
3067 mg->mg_len += utf8_length((U8*)STRING(scan),
3068 (U8*)STRING(scan)+STR_LEN(scan));
3070 data->last_end = data->pos_min + l;
3071 data->pos_min += l; /* As in the first entry. */
3072 data->flags &= ~SF_BEFORE_EOL;
3074 if (flags & SCF_DO_STCLASS_AND) {
3075 /* Check whether it is compatible with what we know already! */
3079 /* If compatibile, we or it in below. It is compatible if is
3080 * in the bitmp and either 1) its bit or its fold is set, or 2)
3081 * it's for a locale. Even if there isn't unicode semantics
3082 * here, at runtime there may be because of matching against a
3083 * utf8 string, so accept a possible false positive for
3084 * latin1-range folds */
3086 (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
3087 && !ANYOF_BITMAP_TEST(data->start_class, uc)
3088 && (!(data->start_class->flags & ANYOF_LOC_NONBITMAP_FOLD)
3089 || !ANYOF_BITMAP_TEST(data->start_class, PL_fold_latin1[uc])))
3092 ANYOF_CLASS_ZERO(data->start_class);
3093 ANYOF_BITMAP_ZERO(data->start_class);
3095 ANYOF_BITMAP_SET(data->start_class, uc);
3096 data->start_class->flags &= ~ANYOF_EOS;
3098 data->start_class->flags &= ~ANYOF_UNICODE_ALL;
3100 else if (flags & SCF_DO_STCLASS_OR) {
3101 /* false positive possible if the class is case-folded */
3103 ANYOF_BITMAP_SET(data->start_class, uc);
3105 data->start_class->flags |= ANYOF_UNICODE_ALL;
3106 data->start_class->flags &= ~ANYOF_EOS;
3107 cl_and(data->start_class, and_withp);
3109 flags &= ~SCF_DO_STCLASS;
3111 else if (PL_regkind[OP(scan)] == EXACT) { /* But OP != EXACT! */
3112 I32 l = STR_LEN(scan);
3113 UV uc = *((U8*)STRING(scan));
3115 /* Search for fixed substrings supports EXACT only. */
3116 if (flags & SCF_DO_SUBSTR) {
3118 SCAN_COMMIT(pRExC_state, data, minlenp);
3121 const U8 * const s = (U8 *)STRING(scan);
3122 l = utf8_length(s, s + l);
3123 uc = utf8_to_uvchr(s, NULL);
3126 if (flags & SCF_DO_SUBSTR)
3128 if (flags & SCF_DO_STCLASS_AND) {
3129 /* Check whether it is compatible with what we know already! */
3132 (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
3133 && !ANYOF_BITMAP_TEST(data->start_class, uc)
3134 && !ANYOF_BITMAP_TEST(data->start_class, PL_fold_latin1[uc])))
3138 ANYOF_CLASS_ZERO(data->start_class);
3139 ANYOF_BITMAP_ZERO(data->start_class);
3141 ANYOF_BITMAP_SET(data->start_class, uc);
3142 data->start_class->flags &= ~ANYOF_EOS;
3143 data->start_class->flags |= ANYOF_LOC_NONBITMAP_FOLD;
3144 if (OP(scan) == EXACTFL) {
3145 data->start_class->flags |= ANYOF_LOCALE;
3149 /* Also set the other member of the fold pair. In case
3150 * that unicode semantics is called for at runtime, use
3151 * the full latin1 fold. (Can't do this for locale,
3152 * because not known until runtime */
3153 ANYOF_BITMAP_SET(data->start_class, PL_fold_latin1[uc]);
3157 else if (flags & SCF_DO_STCLASS_OR) {
3158 if (data->start_class->flags & ANYOF_LOC_NONBITMAP_FOLD) {
3159 /* false positive possible if the class is case-folded.
3160 Assume that the locale settings are the same... */
3162 ANYOF_BITMAP_SET(data->start_class, uc);
3163 if (OP(scan) != EXACTFL) {
3165 /* And set the other member of the fold pair, but
3166 * can't do that in locale because not known until
3168 ANYOF_BITMAP_SET(data->start_class,
3169 PL_fold_latin1[uc]);
3172 data->start_class->flags &= ~ANYOF_EOS;
3174 cl_and(data->start_class, and_withp);
3176 flags &= ~SCF_DO_STCLASS;
3178 else if (REGNODE_VARIES(OP(scan))) {
3179 I32 mincount, maxcount, minnext, deltanext, fl = 0;
3180 I32 f = flags, pos_before = 0;
3181 regnode * const oscan = scan;
3182 struct regnode_charclass_class this_class;
3183 struct regnode_charclass_class *oclass = NULL;
3184 I32 next_is_eval = 0;
3186 switch (PL_regkind[OP(scan)]) {
3187 case WHILEM: /* End of (?:...)* . */
3188 scan = NEXTOPER(scan);
3191 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
3192 next = NEXTOPER(scan);
3193 if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
3195 maxcount = REG_INFTY;
3196 next = regnext(scan);
3197 scan = NEXTOPER(scan);
3201 if (flags & SCF_DO_SUBSTR)
3206 if (flags & SCF_DO_STCLASS) {
3208 maxcount = REG_INFTY;
3209 next = regnext(scan);
3210 scan = NEXTOPER(scan);
3213 is_inf = is_inf_internal = 1;
3214 scan = regnext(scan);
3215 if (flags & SCF_DO_SUBSTR) {
3216 SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot extend fixed substrings */
3217 data->longest = &(data->longest_float);
3219 goto optimize_curly_tail;
3221 if (stopparen>0 && (OP(scan)==CURLYN || OP(scan)==CURLYM)
3222 && (scan->flags == stopparen))
3227 mincount = ARG1(scan);
3228 maxcount = ARG2(scan);
3230 next = regnext(scan);
3231 if (OP(scan) == CURLYX) {
3232 I32 lp = (data ? *(data->last_closep) : 0);
3233 scan->flags = ((lp <= (I32)U8_MAX) ? (U8)lp : U8_MAX);
3235 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
3236 next_is_eval = (OP(scan) == EVAL);
3238 if (flags & SCF_DO_SUBSTR) {
3239 if (mincount == 0) SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot extend fixed substrings */
3240 pos_before = data->pos_min;
3244 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
3246 data->flags |= SF_IS_INF;
3248 if (flags & SCF_DO_STCLASS) {
3249 cl_init(pRExC_state, &this_class);
3250 oclass = data->start_class;
3251 data->start_class = &this_class;
3252 f |= SCF_DO_STCLASS_AND;
3253 f &= ~SCF_DO_STCLASS_OR;
3255 /* Exclude from super-linear cache processing any {n,m}
3256 regops for which the combination of input pos and regex
3257 pos is not enough information to determine if a match
3260 For example, in the regex /foo(bar\s*){4,8}baz/ with the
3261 regex pos at the \s*, the prospects for a match depend not
3262 only on the input position but also on how many (bar\s*)
3263 repeats into the {4,8} we are. */
3264 if ((mincount > 1) || (maxcount > 1 && maxcount != REG_INFTY))
3265 f &= ~SCF_WHILEM_VISITED_POS;
3267 /* This will finish on WHILEM, setting scan, or on NULL: */
3268 minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext,
3269 last, data, stopparen, recursed, NULL,
3271 ? (f & ~SCF_DO_SUBSTR) : f),depth+1);
3273 if (flags & SCF_DO_STCLASS)
3274 data->start_class = oclass;
3275 if (mincount == 0 || minnext == 0) {
3276 if (flags & SCF_DO_STCLASS_OR) {
3277 cl_or(pRExC_state, data->start_class, &this_class);
3279 else if (flags & SCF_DO_STCLASS_AND) {
3280 /* Switch to OR mode: cache the old value of
3281 * data->start_class */
3283 StructCopy(data->start_class, and_withp,
3284 struct regnode_charclass_class);
3285 flags &= ~SCF_DO_STCLASS_AND;
3286 StructCopy(&this_class, data->start_class,
3287 struct regnode_charclass_class);
3288 flags |= SCF_DO_STCLASS_OR;
3289 data->start_class->flags |= ANYOF_EOS;
3291 } else { /* Non-zero len */
3292 if (flags & SCF_DO_STCLASS_OR) {
3293 cl_or(pRExC_state, data->start_class, &this_class);
3294 cl_and(data->start_class, and_withp);
3296 else if (flags & SCF_DO_STCLASS_AND)
3297 cl_and(data->start_class, &this_class);
3298 flags &= ~SCF_DO_STCLASS;
3300 if (!scan) /* It was not CURLYX, but CURLY. */
3302 if ( /* ? quantifier ok, except for (?{ ... }) */
3303 (next_is_eval || !(mincount == 0 && maxcount == 1))
3304 && (minnext == 0) && (deltanext == 0)
3305 && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
3306 && maxcount <= REG_INFTY/3) /* Complement check for big count */
3308 ckWARNreg(RExC_parse,
3309 "Quantifier unexpected on zero-length expression");
3312 min += minnext * mincount;
3313 is_inf_internal |= ((maxcount == REG_INFTY
3314 && (minnext + deltanext) > 0)
3315 || deltanext == I32_MAX);
3316 is_inf |= is_inf_internal;
3317 delta += (minnext + deltanext) * maxcount - minnext * mincount;
3319 /* Try powerful optimization CURLYX => CURLYN. */
3320 if ( OP(oscan) == CURLYX && data
3321 && data->flags & SF_IN_PAR
3322 && !(data->flags & SF_HAS_EVAL)
3323 && !deltanext && minnext == 1 ) {
3324 /* Try to optimize to CURLYN. */
3325 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
3326 regnode * const nxt1 = nxt;
3333 if (!REGNODE_SIMPLE(OP(nxt))
3334 && !(PL_regkind[OP(nxt)] == EXACT
3335 && STR_LEN(nxt) == 1))
3341 if (OP(nxt) != CLOSE)
3343 if (RExC_open_parens) {
3344 RExC_open_parens[ARG(nxt1)-1]=oscan; /*open->CURLYM*/
3345 RExC_close_parens[ARG(nxt1)-1]=nxt+2; /*close->while*/
3347 /* Now we know that nxt2 is the only contents: */
3348 oscan->flags = (U8)ARG(nxt);
3350 OP(nxt1) = NOTHING; /* was OPEN. */
3353 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
3354 NEXT_OFF(nxt1+ 1) = 0; /* just for consistency. */
3355 NEXT_OFF(nxt2) = 0; /* just for consistency with CURLY. */
3356 OP(nxt) = OPTIMIZED; /* was CLOSE. */
3357 OP(nxt + 1) = OPTIMIZED; /* was count. */
3358 NEXT_OFF(nxt+ 1) = 0; /* just for consistency. */
3363 /* Try optimization CURLYX => CURLYM. */
3364 if ( OP(oscan) == CURLYX && data
3365 && !(data->flags & SF_HAS_PAR)
3366 && !(data->flags & SF_HAS_EVAL)
3367 && !deltanext /* atom is fixed width */
3368 && minnext != 0 /* CURLYM can't handle zero width */
3370 /* XXXX How to optimize if data == 0? */
3371 /* Optimize to a simpler form. */
3372 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
3376 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
3377 && (OP(nxt2) != WHILEM))
3379 OP(nxt2) = SUCCEED; /* Whas WHILEM */
3380 /* Need to optimize away parenths. */
3381 if ((data->flags & SF_IN_PAR) && OP(nxt) == CLOSE) {
3382 /* Set the parenth number. */
3383 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
3385 oscan->flags = (U8)ARG(nxt);
3386 if (RExC_open_parens) {
3387 RExC_open_parens[ARG(nxt1)-1]=oscan; /*open->CURLYM*/
3388 RExC_close_parens[ARG(nxt1)-1]=nxt2+1; /*close->NOTHING*/
3390 OP(nxt1) = OPTIMIZED; /* was OPEN. */
3391 OP(nxt) = OPTIMIZED; /* was CLOSE. */
3394 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
3395 OP(nxt + 1) = OPTIMIZED; /* was count. */
3396 NEXT_OFF(nxt1 + 1) = 0; /* just for consistancy. */
3397 NEXT_OFF(nxt + 1) = 0; /* just for consistancy. */
3400 while ( nxt1 && (OP(nxt1) != WHILEM)) {
3401 regnode *nnxt = regnext(nxt1);
3403 if (reg_off_by_arg[OP(nxt1)])
3404 ARG_SET(nxt1, nxt2 - nxt1);
3405 else if (nxt2 - nxt1 < U16_MAX)
3406 NEXT_OFF(nxt1) = nxt2 - nxt1;
3408 OP(nxt) = NOTHING; /* Cannot beautify */
3413 /* Optimize again: */
3414 study_chunk(pRExC_state, &nxt1, minlenp, &deltanext, nxt,
3415 NULL, stopparen, recursed, NULL, 0,depth+1);
3420 else if ((OP(oscan) == CURLYX)
3421 && (flags & SCF_WHILEM_VISITED_POS)
3422 /* See the comment on a similar expression above.
3423 However, this time it's not a subexpression
3424 we care about, but the expression itself. */
3425 && (maxcount == REG_INFTY)
3426 && data && ++data->whilem_c < 16) {
3427 /* This stays as CURLYX, we can put the count/of pair. */
3428 /* Find WHILEM (as in regexec.c) */
3429 regnode *nxt = oscan + NEXT_OFF(oscan);
3431 if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
3433 PREVOPER(nxt)->flags = (U8)(data->whilem_c
3434 | (RExC_whilem_seen << 4)); /* On WHILEM */
3436 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
3438 if (flags & SCF_DO_SUBSTR) {
3439 SV *last_str = NULL;
3440 int counted = mincount != 0;
3442 if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
3443 #if defined(SPARC64_GCC_WORKAROUND)
3446 const char *s = NULL;
3449 if (pos_before >= data->last_start_min)
3452 b = data->last_start_min;
3455 s = SvPV_const(data->last_found, l);
3456 old = b - data->last_start_min;
3459 I32 b = pos_before >= data->last_start_min
3460 ? pos_before : data->last_start_min;
3462 const char * const s = SvPV_const(data->last_found, l);
3463 I32 old = b - data->last_start_min;
3467 old = utf8_hop((U8*)s, old) - (U8*)s;
3469 /* Get the added string: */
3470 last_str = newSVpvn_utf8(s + old, l, UTF);
3471 if (deltanext == 0 && pos_before == b) {
3472 /* What was added is a constant string */
3474 SvGROW(last_str, (mincount * l) + 1);
3475 repeatcpy(SvPVX(last_str) + l,
3476 SvPVX_const(last_str), l, mincount - 1);
3477 SvCUR_set(last_str, SvCUR(last_str) * mincount);
3478 /* Add additional parts. */
3479 SvCUR_set(data->last_found,
3480 SvCUR(data->last_found) - l);
3481 sv_catsv(data->last_found, last_str);
3483 SV * sv = data->last_found;
3485 SvUTF8(sv) && SvMAGICAL(sv) ?
3486 mg_find(sv, PERL_MAGIC_utf8) : NULL;
3487 if (mg && mg->mg_len >= 0)
3488 mg->mg_len += CHR_SVLEN(last_str) - l;
3490 data->last_end += l * (mincount - 1);
3493 /* start offset must point into the last copy */
3494 data->last_start_min += minnext * (mincount - 1);
3495 data->last_start_max += is_inf ? I32_MAX
3496 : (maxcount - 1) * (minnext + data->pos_delta);
3499 /* It is counted once already... */
3500 data->pos_min += minnext * (mincount - counted);
3501 data->pos_delta += - counted * deltanext +
3502 (minnext + deltanext) * maxcount - minnext * mincount;
3503 if (mincount != maxcount) {
3504 /* Cannot extend fixed substrings found inside
3506 SCAN_COMMIT(pRExC_state,data,minlenp);
3507 if (mincount && last_str) {
3508 SV * const sv = data->last_found;
3509 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
3510 mg_find(sv, PERL_MAGIC_utf8) : NULL;
3514 sv_setsv(sv, last_str);
3515 data->last_end = data->pos_min;
3516 data->last_start_min =
3517 data->pos_min - CHR_SVLEN(last_str);
3518 data->last_start_max = is_inf
3520 : data->pos_min + data->pos_delta
3521 - CHR_SVLEN(last_str);
3523 data->longest = &(data->longest_float);
3525 SvREFCNT_dec(last_str);
3527 if (data && (fl & SF_HAS_EVAL))
3528 data->flags |= SF_HAS_EVAL;
3529 optimize_curly_tail:
3530 if (OP(oscan) != CURLYX) {
3531 while (PL_regkind[OP(next = regnext(oscan))] == NOTHING
3533 NEXT_OFF(oscan) += NEXT_OFF(next);
3536 default: /* REF and CLUMP only? */
3537 if (flags & SCF_DO_SUBSTR) {
3538 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
3539 data->longest = &(data->longest_float);
3541 is_inf = is_inf_internal = 1;
3542 if (flags & SCF_DO_STCLASS_OR)
3543 cl_anything(pRExC_state, data->start_class);
3544 flags &= ~SCF_DO_STCLASS;
3548 else if (OP(scan) == LNBREAK) {
3549 if (flags & SCF_DO_STCLASS) {
3551 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
3552 if (flags & SCF_DO_STCLASS_AND) {
3553 for (value = 0; value < 256; value++)
3554 if (!is_VERTWS_cp(value))
3555 ANYOF_BITMAP_CLEAR(data->start_class, value);
3558 for (value = 0; value < 256; value++)
3559 if (is_VERTWS_cp(value))
3560 ANYOF_BITMAP_SET(data->start_class, value);
3562 if (flags & SCF_DO_STCLASS_OR)
3563 cl_and(data->start_class, and_withp);
3564 flags &= ~SCF_DO_STCLASS;
3568 if (flags & SCF_DO_SUBSTR) {
3569 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
3571 data->pos_delta += 1;
3572 data->longest = &(data->longest_float);
3575 else if (OP(scan) == FOLDCHAR) {
3576 int d = ARG(scan) == LATIN_SMALL_LETTER_SHARP_S ? 1 : 2;
3577 flags &= ~SCF_DO_STCLASS;
3580 if (flags & SCF_DO_SUBSTR) {
3581 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
3583 data->pos_delta += d;
3584 data->longest = &(data->longest_float);
3587 else if (REGNODE_SIMPLE(OP(scan))) {
3590 if (flags & SCF_DO_SUBSTR) {
3591 SCAN_COMMIT(pRExC_state,data,minlenp);
3595 if (flags & SCF_DO_STCLASS) {
3596 data->start_class->flags &= ~ANYOF_EOS; /* No match on empty */
3598 /* Some of the logic below assumes that switching
3599 locale on will only add false positives. */
3600 switch (PL_regkind[OP(scan)]) {
3604 /* Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan)); */
3605 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
3606 cl_anything(pRExC_state, data->start_class);
3609 if (OP(scan) == SANY)
3611 if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */
3612 value = (ANYOF_BITMAP_TEST(data->start_class,'\n')
3613 || ANYOF_CLASS_TEST_ANY_SET(data->start_class));
3614 cl_anything(pRExC_state, data->start_class);
3616 if (flags & SCF_DO_STCLASS_AND || !value)
3617 ANYOF_BITMAP_CLEAR(data->start_class,'\n');
3620 if (flags & SCF_DO_STCLASS_AND)
3621 cl_and(data->start_class,
3622 (struct regnode_charclass_class*)scan);
3624 cl_or(pRExC_state, data->start_class,
3625 (struct regnode_charclass_class*)scan);
3628 if (flags & SCF_DO_STCLASS_AND) {
3629 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3630 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
3631 if (FLAGS(scan) & USE_UNI) {
3632 for (value = 0; value < 256; value++) {
3633 if (!isWORDCHAR_L1(value)) {
3634 ANYOF_BITMAP_CLEAR(data->start_class, value);
3638 for (value = 0; value < 256; value++) {
3639 if (!isALNUM(value)) {
3640 ANYOF_BITMAP_CLEAR(data->start_class, value);
3647 if (data->start_class->flags & ANYOF_LOCALE)
3648 ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
3649 else if (FLAGS(scan) & USE_UNI) {
3650 for (value = 0; value < 256; value++) {
3651 if (isWORDCHAR_L1(value)) {
3652 ANYOF_BITMAP_SET(data->start_class, value);
3656 for (value = 0; value < 256; value++) {
3657 if (isALNUM(value)) {
3658 ANYOF_BITMAP_SET(data->start_class, value);
3665 if (flags & SCF_DO_STCLASS_AND) {
3666 if (data->start_class->flags & ANYOF_LOCALE)
3667 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NALNUM);
3670 ANYOF_CLASS_SET(data->start_class,ANYOF_ALNUM);
3671 data->start_class->flags |= ANYOF_LOCALE;
3675 if (flags & SCF_DO_STCLASS_AND) {
3676 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3677 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
3678 if (FLAGS(scan) & USE_UNI) {
3679 for (value = 0; value < 256; value++) {
3680 if (isWORDCHAR_L1(value)) {
3681 ANYOF_BITMAP_CLEAR(data->start_class, value);
3685 for (value = 0; value < 256; value++) {
3686 if (isALNUM(value)) {
3687 ANYOF_BITMAP_CLEAR(data->start_class, value);
3694 if (data->start_class->flags & ANYOF_LOCALE)
3695 ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
3697 for (value = 0; value < 256; value++)
3698 if (!isALNUM(value))
3699 ANYOF_BITMAP_SET(data->start_class, value);
3704 if (flags & SCF_DO_STCLASS_AND) {
3705 if (data->start_class->flags & ANYOF_LOCALE)
3706 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_ALNUM);
3709 data->start_class->flags |= ANYOF_LOCALE;
3710 ANYOF_CLASS_SET(data->start_class,ANYOF_NALNUM);
3714 if (flags & SCF_DO_STCLASS_AND) {
3715 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3716 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
3717 if (FLAGS(scan) & USE_UNI) {
3718 for (value = 0; value < 256; value++) {
3719 if (!isSPACE_L1(value)) {
3720 ANYOF_BITMAP_CLEAR(data->start_class, value);
3724 for (value = 0; value < 256; value++) {
3725 if (!isSPACE(value)) {
3726 ANYOF_BITMAP_CLEAR(data->start_class, value);
3733 if (data->start_class->flags & ANYOF_LOCALE) {
3734 ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
3736 else if (FLAGS(scan) & USE_UNI) {
3737 for (value = 0; value < 256; value++) {
3738 if (isSPACE_L1(value)) {
3739 ANYOF_BITMAP_SET(data->start_class, value);
3743 for (value = 0; value < 256; value++) {
3744 if (isSPACE(value)) {
3745 ANYOF_BITMAP_SET(data->start_class, value);
3752 if (flags & SCF_DO_STCLASS_AND) {
3753 if (data->start_class->flags & ANYOF_LOCALE)
3754 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NSPACE);
3757 data->start_class->flags |= ANYOF_LOCALE;
3758 ANYOF_CLASS_SET(data->start_class,ANYOF_SPACE);
3762 if (flags & SCF_DO_STCLASS_AND) {
3763 if (!(data->start_class->flags & ANYOF_LOCALE)) {
3764 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
3765 if (FLAGS(scan) & USE_UNI) {
3766 for (value = 0; value < 256; value++) {
3767 if (isSPACE_L1(value)) {
3768 ANYOF_BITMAP_CLEAR(data->start_class, value);
3772 for (value = 0; value < 256; value++) {
3773 if (isSPACE(value)) {
3774 ANYOF_BITMAP_CLEAR(data->start_class, value);
3781 if (data->start_class->flags & ANYOF_LOCALE)
3782 ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
3783 else if (FLAGS(scan) & USE_UNI) {
3784 for (value = 0; value < 256; value++) {
3785 if (!isSPACE_L1(value)) {
3786 ANYOF_BITMAP_SET(data->start_class, value);
3791 for (value = 0; value < 256; value++) {
3792 if (!isSPACE(value)) {
3793 ANYOF_BITMAP_SET(data->start_class, value);
3800 if (flags & SCF_DO_STCLASS_AND) {
3801 if (data->start_class->flags & ANYOF_LOCALE) {
3802 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_SPACE);
3803 for (value = 0; value < 256; value++)
3804 if (!isSPACE(value))
3805 ANYOF_BITMAP_CLEAR(data->start_class, value);
3809 data->start_class->flags |= ANYOF_LOCALE;
3810 ANYOF_CLASS_SET(data->start_class,ANYOF_NSPACE);
3814 if (flags & SCF_DO_STCLASS_AND) {
3815 ANYOF_CLASS_CLEAR(data->start_class,ANYOF_NDIGIT);
3816 for (value = 0; value < 256; value++)
3817 if (!isDIGIT(value))
3818 ANYOF_BITMAP_CLEAR(data->start_class, value);
3821 if (data->start_class->flags & ANYOF_LOCALE)
3822 ANYOF_CLASS_SET(data->start_class,ANYOF_DIGIT);
3824 for (value = 0; value < 256; value++)
3826 ANYOF_BITMAP_SET(data->start_class, value);
https://perl5.git.perl.org / perl5.git / blob