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
84 EXTERN_C const struct regexp_engine my_reg_engine;
89 #include "dquote_inline.h"
90 #include "invlist_inline.h"
91 #include "unicode_constants.h"
93 #define HAS_NONLATIN1_FOLD_CLOSURE(i) \
94 _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(i)
95 #define HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(i) \
96 _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(i)
97 #define IS_NON_FINAL_FOLD(c) _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c)
98 #define IS_IN_SOME_FOLD_L1(c) _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c)
101 #define STATIC static
104 /* this is a chain of data about sub patterns we are processing that
105 need to be handled separately/specially in study_chunk. Its so
106 we can simulate recursion without losing state. */
108 typedef struct scan_frame {
109 regnode *last_regnode; /* last node to process in this frame */
110 regnode *next_regnode; /* next node to process when last is reached */
111 U32 prev_recursed_depth;
112 I32 stopparen; /* what stopparen do we use */
113 U32 is_top_frame; /* what flags do we use? */
115 struct scan_frame *this_prev_frame; /* this previous frame */
116 struct scan_frame *prev_frame; /* previous frame */
117 struct scan_frame *next_frame; /* next frame */
120 /* Certain characters are output as a sequence with the first being a
122 #define isBACKSLASHED_PUNCT(c) \
123 ((c) == '-' || (c) == ']' || (c) == '\\' || (c) == '^')
126 struct RExC_state_t {
127 U32 flags; /* RXf_* are we folding, multilining? */
128 U32 pm_flags; /* PMf_* stuff from the calling PMOP */
129 char *precomp; /* uncompiled string. */
130 char *precomp_end; /* pointer to end of uncompiled string. */
131 REGEXP *rx_sv; /* The SV that is the regexp. */
132 regexp *rx; /* perl core regexp structure */
133 regexp_internal *rxi; /* internal data for regexp object
135 char *start; /* Start of input for compile */
136 char *end; /* End of input for compile */
137 char *parse; /* Input-scan pointer. */
138 char *adjusted_start; /* 'start', adjusted. See code use */
139 STRLEN precomp_adj; /* an offset beyond precomp. See code use */
140 SSize_t whilem_seen; /* number of WHILEM in this expr */
141 regnode *emit_start; /* Start of emitted-code area */
142 regnode *emit_bound; /* First regnode outside of the
144 regnode *emit; /* Code-emit pointer; if = &emit_dummy,
145 implies compiling, so don't emit */
146 regnode_ssc emit_dummy; /* placeholder for emit to point to;
147 large enough for the largest
148 non-EXACTish node, so can use it as
150 I32 naughty; /* How bad is this pattern? */
151 I32 sawback; /* Did we see \1, ...? */
153 SSize_t size; /* Code size. */
154 I32 npar; /* Capture buffer count, (OPEN) plus
155 one. ("par" 0 is the whole
157 I32 nestroot; /* root parens we are in - used by
161 regnode **open_parens; /* pointers to open parens */
162 regnode **close_parens; /* pointers to close parens */
163 regnode *end_op; /* END node in program */
164 I32 utf8; /* whether the pattern is utf8 or not */
165 I32 orig_utf8; /* whether the pattern was originally in utf8 */
166 /* XXX use this for future optimisation of case
167 * where pattern must be upgraded to utf8. */
168 I32 uni_semantics; /* If a d charset modifier should use unicode
169 rules, even if the pattern is not in
171 HV *paren_names; /* Paren names */
173 regnode **recurse; /* Recurse regops */
174 I32 recurse_count; /* Number of recurse regops we have generated */
175 U8 *study_chunk_recursed; /* bitmap of which subs we have moved
177 U32 study_chunk_recursed_bytes; /* bytes in bitmap */
180 I32 override_recoding;
182 I32 recode_x_to_native;
184 I32 in_multi_char_class;
185 struct reg_code_blocks *code_blocks;/* positions of literal (?{})
187 int code_index; /* next code_blocks[] slot */
188 SSize_t maxlen; /* mininum possible number of chars in string to match */
189 scan_frame *frame_head;
190 scan_frame *frame_last;
193 #ifdef ADD_TO_REGEXEC
194 char *starttry; /* -Dr: where regtry was called. */
195 #define RExC_starttry (pRExC_state->starttry)
197 SV *runtime_code_qr; /* qr with the runtime code blocks */
199 const char *lastparse;
201 AV *paren_name_list; /* idx -> name */
202 U32 study_chunk_recursed_count;
205 #define RExC_lastparse (pRExC_state->lastparse)
206 #define RExC_lastnum (pRExC_state->lastnum)
207 #define RExC_paren_name_list (pRExC_state->paren_name_list)
208 #define RExC_study_chunk_recursed_count (pRExC_state->study_chunk_recursed_count)
209 #define RExC_mysv (pRExC_state->mysv1)
210 #define RExC_mysv1 (pRExC_state->mysv1)
211 #define RExC_mysv2 (pRExC_state->mysv2)
214 bool seen_unfolded_sharp_s;
219 #define RExC_flags (pRExC_state->flags)
220 #define RExC_pm_flags (pRExC_state->pm_flags)
221 #define RExC_precomp (pRExC_state->precomp)
222 #define RExC_precomp_adj (pRExC_state->precomp_adj)
223 #define RExC_adjusted_start (pRExC_state->adjusted_start)
224 #define RExC_precomp_end (pRExC_state->precomp_end)
225 #define RExC_rx_sv (pRExC_state->rx_sv)
226 #define RExC_rx (pRExC_state->rx)
227 #define RExC_rxi (pRExC_state->rxi)
228 #define RExC_start (pRExC_state->start)
229 #define RExC_end (pRExC_state->end)
230 #define RExC_parse (pRExC_state->parse)
231 #define RExC_whilem_seen (pRExC_state->whilem_seen)
233 /* Set during the sizing pass when there is a LATIN SMALL LETTER SHARP S in any
234 * EXACTF node, hence was parsed under /di rules. If later in the parse,
235 * something forces the pattern into using /ui rules, the sharp s should be
236 * folded into the sequence 'ss', which takes up more space than previously
237 * calculated. This means that the sizing pass needs to be restarted. (The
238 * node also becomes an EXACTFU_SS.) For all other characters, an EXACTF node
239 * that gets converted to /ui (and EXACTFU) occupies the same amount of space,
240 * so there is no need to resize [perl #125990]. */
241 #define RExC_seen_unfolded_sharp_s (pRExC_state->seen_unfolded_sharp_s)
243 #ifdef RE_TRACK_PATTERN_OFFSETS
244 #define RExC_offsets (pRExC_state->rxi->u.offsets) /* I am not like the
247 #define RExC_emit (pRExC_state->emit)
248 #define RExC_emit_dummy (pRExC_state->emit_dummy)
249 #define RExC_emit_start (pRExC_state->emit_start)
250 #define RExC_emit_bound (pRExC_state->emit_bound)
251 #define RExC_sawback (pRExC_state->sawback)
252 #define RExC_seen (pRExC_state->seen)
253 #define RExC_size (pRExC_state->size)
254 #define RExC_maxlen (pRExC_state->maxlen)
255 #define RExC_npar (pRExC_state->npar)
256 #define RExC_nestroot (pRExC_state->nestroot)
257 #define RExC_extralen (pRExC_state->extralen)
258 #define RExC_seen_zerolen (pRExC_state->seen_zerolen)
259 #define RExC_utf8 (pRExC_state->utf8)
260 #define RExC_uni_semantics (pRExC_state->uni_semantics)
261 #define RExC_orig_utf8 (pRExC_state->orig_utf8)
262 #define RExC_open_parens (pRExC_state->open_parens)
263 #define RExC_close_parens (pRExC_state->close_parens)
264 #define RExC_end_op (pRExC_state->end_op)
265 #define RExC_paren_names (pRExC_state->paren_names)
266 #define RExC_recurse (pRExC_state->recurse)
267 #define RExC_recurse_count (pRExC_state->recurse_count)
268 #define RExC_study_chunk_recursed (pRExC_state->study_chunk_recursed)
269 #define RExC_study_chunk_recursed_bytes \
270 (pRExC_state->study_chunk_recursed_bytes)
271 #define RExC_in_lookbehind (pRExC_state->in_lookbehind)
272 #define RExC_contains_locale (pRExC_state->contains_locale)
274 # define RExC_recode_x_to_native (pRExC_state->recode_x_to_native)
276 #define RExC_in_multi_char_class (pRExC_state->in_multi_char_class)
277 #define RExC_frame_head (pRExC_state->frame_head)
278 #define RExC_frame_last (pRExC_state->frame_last)
279 #define RExC_frame_count (pRExC_state->frame_count)
280 #define RExC_strict (pRExC_state->strict)
281 #define RExC_study_started (pRExC_state->study_started)
282 #define RExC_warn_text (pRExC_state->warn_text)
284 /* Heuristic check on the complexity of the pattern: if TOO_NAUGHTY, we set
285 * a flag to disable back-off on the fixed/floating substrings - if it's
286 * a high complexity pattern we assume the benefit of avoiding a full match
287 * is worth the cost of checking for the substrings even if they rarely help.
289 #define RExC_naughty (pRExC_state->naughty)
290 #define TOO_NAUGHTY (10)
291 #define MARK_NAUGHTY(add) \
292 if (RExC_naughty < TOO_NAUGHTY) \
293 RExC_naughty += (add)
294 #define MARK_NAUGHTY_EXP(exp, add) \
295 if (RExC_naughty < TOO_NAUGHTY) \
296 RExC_naughty += RExC_naughty / (exp) + (add)
298 #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
299 #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
300 ((*s) == '{' && regcurly(s)))
303 * Flags to be passed up and down.
305 #define WORST 0 /* Worst case. */
306 #define HASWIDTH 0x01 /* Known to match non-null strings. */
308 /* Simple enough to be STAR/PLUS operand; in an EXACTish node must be a single
309 * character. (There needs to be a case: in the switch statement in regexec.c
310 * for any node marked SIMPLE.) Note that this is not the same thing as
313 #define SPSTART 0x04 /* Starts with * or + */
314 #define POSTPONED 0x08 /* (?1),(?&name), (??{...}) or similar */
315 #define TRYAGAIN 0x10 /* Weeded out a declaration. */
316 #define RESTART_PASS1 0x20 /* Need to restart sizing pass */
317 #define NEED_UTF8 0x40 /* In conjunction with RESTART_PASS1, need to
318 calcuate sizes as UTF-8 */
320 #define REG_NODE_NUM(x) ((x) ? (int)((x)-RExC_emit_start) : -1)
322 /* whether trie related optimizations are enabled */
323 #if PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION
324 #define TRIE_STUDY_OPT
325 #define FULL_TRIE_STUDY
331 #define PBYTE(u8str,paren) ((U8*)(u8str))[(paren) >> 3]
332 #define PBITVAL(paren) (1 << ((paren) & 7))
333 #define PAREN_TEST(u8str,paren) ( PBYTE(u8str,paren) & PBITVAL(paren))
334 #define PAREN_SET(u8str,paren) PBYTE(u8str,paren) |= PBITVAL(paren)
335 #define PAREN_UNSET(u8str,paren) PBYTE(u8str,paren) &= (~PBITVAL(paren))
337 #define REQUIRE_UTF8(flagp) STMT_START { \
340 *flagp = RESTART_PASS1|NEED_UTF8; \
345 /* Change from /d into /u rules, and restart the parse if we've already seen
346 * something whose size would increase as a result, by setting *flagp and
347 * returning 'restart_retval'. RExC_uni_semantics is a flag that indicates
348 * we've change to /u during the parse. */
349 #define REQUIRE_UNI_RULES(flagp, restart_retval) \
351 if (DEPENDS_SEMANTICS) { \
353 set_regex_charset(&RExC_flags, REGEX_UNICODE_CHARSET); \
354 RExC_uni_semantics = 1; \
355 if (RExC_seen_unfolded_sharp_s) { \
356 *flagp |= RESTART_PASS1; \
357 return restart_retval; \
362 /* This converts the named class defined in regcomp.h to its equivalent class
363 * number defined in handy.h. */
364 #define namedclass_to_classnum(class) ((int) ((class) / 2))
365 #define classnum_to_namedclass(classnum) ((classnum) * 2)
367 #define _invlist_union_complement_2nd(a, b, output) \
368 _invlist_union_maybe_complement_2nd(a, b, TRUE, output)
369 #define _invlist_intersection_complement_2nd(a, b, output) \
370 _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output)
372 /* About scan_data_t.
374 During optimisation we recurse through the regexp program performing
375 various inplace (keyhole style) optimisations. In addition study_chunk
376 and scan_commit populate this data structure with information about
377 what strings MUST appear in the pattern. We look for the longest
378 string that must appear at a fixed location, and we look for the
379 longest string that may appear at a floating location. So for instance
384 Both 'FOO' and 'A' are fixed strings. Both 'B' and 'BAR' are floating
385 strings (because they follow a .* construct). study_chunk will identify
386 both FOO and BAR as being the longest fixed and floating strings respectively.
388 The strings can be composites, for instance
392 will result in a composite fixed substring 'foo'.
394 For each string some basic information is maintained:
396 - offset or min_offset
397 This is the position the string must appear at, or not before.
398 It also implicitly (when combined with minlenp) tells us how many
399 characters must match before the string we are searching for.
400 Likewise when combined with minlenp and the length of the string it
401 tells us how many characters must appear after the string we have
405 Only used for floating strings. This is the rightmost point that
406 the string can appear at. If set to SSize_t_MAX it indicates that the
407 string can occur infinitely far to the right.
410 A pointer to the minimum number of characters of the pattern that the
411 string was found inside. This is important as in the case of positive
412 lookahead or positive lookbehind we can have multiple patterns
417 The minimum length of the pattern overall is 3, the minimum length
418 of the lookahead part is 3, but the minimum length of the part that
419 will actually match is 1. So 'FOO's minimum length is 3, but the
420 minimum length for the F is 1. This is important as the minimum length
421 is used to determine offsets in front of and behind the string being
422 looked for. Since strings can be composites this is the length of the
423 pattern at the time it was committed with a scan_commit. Note that
424 the length is calculated by study_chunk, so that the minimum lengths
425 are not known until the full pattern has been compiled, thus the
426 pointer to the value.
430 In the case of lookbehind the string being searched for can be
431 offset past the start point of the final matching string.
432 If this value was just blithely removed from the min_offset it would
433 invalidate some of the calculations for how many chars must match
434 before or after (as they are derived from min_offset and minlen and
435 the length of the string being searched for).
436 When the final pattern is compiled and the data is moved from the
437 scan_data_t structure into the regexp structure the information
438 about lookbehind is factored in, with the information that would
439 have been lost precalculated in the end_shift field for the
442 The fields pos_min and pos_delta are used to store the minimum offset
443 and the delta to the maximum offset at the current point in the pattern.
447 typedef struct scan_data_t {
448 /*I32 len_min; unused */
449 /*I32 len_delta; unused */
453 SSize_t last_end; /* min value, <0 unless valid. */
454 SSize_t last_start_min;
455 SSize_t last_start_max;
456 SV **longest; /* Either &l_fixed, or &l_float. */
457 SV *longest_fixed; /* longest fixed string found in pattern */
458 SSize_t offset_fixed; /* offset where it starts */
459 SSize_t *minlen_fixed; /* pointer to the minlen relevant to the string */
460 I32 lookbehind_fixed; /* is the position of the string modfied by LB */
461 SV *longest_float; /* longest floating string found in pattern */
462 SSize_t offset_float_min; /* earliest point in string it can appear */
463 SSize_t offset_float_max; /* latest point in string it can appear */
464 SSize_t *minlen_float; /* pointer to the minlen relevant to the string */
465 SSize_t lookbehind_float; /* is the pos of the string modified by LB */
468 SSize_t *last_closep;
469 regnode_ssc *start_class;
473 * Forward declarations for pregcomp()'s friends.
476 static const scan_data_t zero_scan_data =
477 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ,0};
479 #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
480 #define SF_BEFORE_SEOL 0x0001
481 #define SF_BEFORE_MEOL 0x0002
482 #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
483 #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
485 #define SF_FIX_SHIFT_EOL (+2)
486 #define SF_FL_SHIFT_EOL (+4)
488 #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
489 #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
491 #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
492 #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
493 #define SF_IS_INF 0x0040
494 #define SF_HAS_PAR 0x0080
495 #define SF_IN_PAR 0x0100
496 #define SF_HAS_EVAL 0x0200
499 /* SCF_DO_SUBSTR is the flag that tells the regexp analyzer to track the
500 * longest substring in the pattern. When it is not set the optimiser keeps
501 * track of position, but does not keep track of the actual strings seen,
503 * So for instance /foo/ will be parsed with SCF_DO_SUBSTR being true, but
506 * Similarly, /foo.*(blah|erm|huh).*fnorble/ will have "foo" and "fnorble"
507 * parsed with SCF_DO_SUBSTR on, but while processing the (...) it will be
508 * turned off because of the alternation (BRANCH). */
509 #define SCF_DO_SUBSTR 0x0400
511 #define SCF_DO_STCLASS_AND 0x0800
512 #define SCF_DO_STCLASS_OR 0x1000
513 #define SCF_DO_STCLASS (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
514 #define SCF_WHILEM_VISITED_POS 0x2000
516 #define SCF_TRIE_RESTUDY 0x4000 /* Do restudy? */
517 #define SCF_SEEN_ACCEPT 0x8000
518 #define SCF_TRIE_DOING_RESTUDY 0x10000
519 #define SCF_IN_DEFINE 0x20000
524 #define UTF cBOOL(RExC_utf8)
526 /* The enums for all these are ordered so things work out correctly */
527 #define LOC (get_regex_charset(RExC_flags) == REGEX_LOCALE_CHARSET)
528 #define DEPENDS_SEMANTICS (get_regex_charset(RExC_flags) \
529 == REGEX_DEPENDS_CHARSET)
530 #define UNI_SEMANTICS (get_regex_charset(RExC_flags) == REGEX_UNICODE_CHARSET)
531 #define AT_LEAST_UNI_SEMANTICS (get_regex_charset(RExC_flags) \
532 >= REGEX_UNICODE_CHARSET)
533 #define ASCII_RESTRICTED (get_regex_charset(RExC_flags) \
534 == REGEX_ASCII_RESTRICTED_CHARSET)
535 #define AT_LEAST_ASCII_RESTRICTED (get_regex_charset(RExC_flags) \
536 >= REGEX_ASCII_RESTRICTED_CHARSET)
537 #define ASCII_FOLD_RESTRICTED (get_regex_charset(RExC_flags) \
538 == REGEX_ASCII_MORE_RESTRICTED_CHARSET)
540 #define FOLD cBOOL(RExC_flags & RXf_PMf_FOLD)
542 /* For programs that want to be strictly Unicode compatible by dying if any
543 * attempt is made to match a non-Unicode code point against a Unicode
545 #define ALWAYS_WARN_SUPER ckDEAD(packWARN(WARN_NON_UNICODE))
547 #define OOB_NAMEDCLASS -1
549 /* There is no code point that is out-of-bounds, so this is problematic. But
550 * its only current use is to initialize a variable that is always set before
552 #define OOB_UNICODE 0xDEADBEEF
554 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
557 /* length of regex to show in messages that don't mark a position within */
558 #define RegexLengthToShowInErrorMessages 127
561 * If MARKER[12] are adjusted, be sure to adjust the constants at the top
562 * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
563 * op/pragma/warn/regcomp.
565 #define MARKER1 "<-- HERE" /* marker as it appears in the description */
566 #define MARKER2 " <-- HERE " /* marker as it appears within the regex */
568 #define REPORT_LOCATION " in regex; marked by " MARKER1 \
569 " in m/%" UTF8f MARKER2 "%" UTF8f "/"
571 /* The code in this file in places uses one level of recursion with parsing
572 * rebased to an alternate string constructed by us in memory. This can take
573 * the form of something that is completely different from the input, or
574 * something that uses the input as part of the alternate. In the first case,
575 * there should be no possibility of an error, as we are in complete control of
576 * the alternate string. But in the second case we don't control the input
577 * portion, so there may be errors in that. Here's an example:
579 * is handled specially because \x{df} folds to a sequence of more than one
580 * character, 'ss'. What is done is to create and parse an alternate string,
581 * which looks like this:
582 * /(?:\x{DF}|[abc\x{DF}def])/ui
583 * where it uses the input unchanged in the middle of something it constructs,
584 * which is a branch for the DF outside the character class, and clustering
585 * parens around the whole thing. (It knows enough to skip the DF inside the
586 * class while in this substitute parse.) 'abc' and 'def' may have errors that
587 * need to be reported. The general situation looks like this:
590 * Input: ----------------------------------------------------
591 * Constructed: ---------------------------------------------------
594 * The input string sI..eI is the input pattern. The string sC..EC is the
595 * constructed substitute parse string. The portions sC..tC and eC..EC are
596 * constructed by us. The portion tC..eC is an exact duplicate of the input
597 * pattern tI..eI. In the diagram, these are vertically aligned. Suppose that
598 * while parsing, we find an error at xC. We want to display a message showing
599 * the real input string. Thus we need to find the point xI in it which
600 * corresponds to xC. xC >= tC, since the portion of the string sC..tC has
601 * been constructed by us, and so shouldn't have errors. We get:
603 * xI = sI + (tI - sI) + (xC - tC)
605 * and, the offset into sI is:
607 * (xI - sI) = (tI - sI) + (xC - tC)
609 * When the substitute is constructed, we save (tI -sI) as RExC_precomp_adj,
610 * and we save tC as RExC_adjusted_start.
612 * During normal processing of the input pattern, everything points to that,
613 * with RExC_precomp_adj set to 0, and RExC_adjusted_start set to sI.
616 #define tI_sI RExC_precomp_adj
617 #define tC RExC_adjusted_start
618 #define sC RExC_precomp
619 #define xI_offset(xC) ((IV) (tI_sI + (xC - tC)))
620 #define xI(xC) (sC + xI_offset(xC))
621 #define eC RExC_precomp_end
623 #define REPORT_LOCATION_ARGS(xC) \
625 (xI(xC) > eC) /* Don't run off end */ \
626 ? eC - sC /* Length before the <--HERE */ \
628 sC), /* The input pattern printed up to the <--HERE */ \
630 (xI(xC) > eC) ? 0 : eC - xI(xC), /* Length after <--HERE */ \
631 (xI(xC) > eC) ? eC : xI(xC)) /* pattern after <--HERE */
633 /* Used to point after bad bytes for an error message, but avoid skipping
634 * past a nul byte. */
635 #define SKIP_IF_CHAR(s) (!*(s) ? 0 : UTF ? UTF8SKIP(s) : 1)
638 * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
639 * arg. Show regex, up to a maximum length. If it's too long, chop and add
642 #define _FAIL(code) STMT_START { \
643 const char *ellipses = ""; \
644 IV len = RExC_precomp_end - RExC_precomp; \
647 SAVEFREESV(RExC_rx_sv); \
648 if (len > RegexLengthToShowInErrorMessages) { \
649 /* chop 10 shorter than the max, to ensure meaning of "..." */ \
650 len = RegexLengthToShowInErrorMessages - 10; \
656 #define FAIL(msg) _FAIL( \
657 Perl_croak(aTHX_ "%s in regex m/%" UTF8f "%s/", \
658 msg, UTF8fARG(UTF, len, RExC_precomp), ellipses))
660 #define FAIL2(msg,arg) _FAIL( \
661 Perl_croak(aTHX_ msg " in regex m/%" UTF8f "%s/", \
662 arg, UTF8fARG(UTF, len, RExC_precomp), ellipses))
665 * Simple_vFAIL -- like FAIL, but marks the current location in the scan
667 #define Simple_vFAIL(m) STMT_START { \
668 Perl_croak(aTHX_ "%s" REPORT_LOCATION, \
669 m, REPORT_LOCATION_ARGS(RExC_parse)); \
673 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
675 #define vFAIL(m) STMT_START { \
677 SAVEFREESV(RExC_rx_sv); \
682 * Like Simple_vFAIL(), but accepts two arguments.
684 #define Simple_vFAIL2(m,a1) STMT_START { \
685 S_re_croak2(aTHX_ UTF, m, REPORT_LOCATION, a1, \
686 REPORT_LOCATION_ARGS(RExC_parse)); \
690 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
692 #define vFAIL2(m,a1) STMT_START { \
694 SAVEFREESV(RExC_rx_sv); \
695 Simple_vFAIL2(m, a1); \
700 * Like Simple_vFAIL(), but accepts three arguments.
702 #define Simple_vFAIL3(m, a1, a2) STMT_START { \
703 S_re_croak2(aTHX_ UTF, m, REPORT_LOCATION, a1, a2, \
704 REPORT_LOCATION_ARGS(RExC_parse)); \
708 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
710 #define vFAIL3(m,a1,a2) STMT_START { \
712 SAVEFREESV(RExC_rx_sv); \
713 Simple_vFAIL3(m, a1, a2); \
717 * Like Simple_vFAIL(), but accepts four arguments.
719 #define Simple_vFAIL4(m, a1, a2, a3) STMT_START { \
720 S_re_croak2(aTHX_ UTF, m, REPORT_LOCATION, a1, a2, a3, \
721 REPORT_LOCATION_ARGS(RExC_parse)); \
724 #define vFAIL4(m,a1,a2,a3) STMT_START { \
726 SAVEFREESV(RExC_rx_sv); \
727 Simple_vFAIL4(m, a1, a2, a3); \
730 /* A specialized version of vFAIL2 that works with UTF8f */
731 #define vFAIL2utf8f(m, a1) STMT_START { \
733 SAVEFREESV(RExC_rx_sv); \
734 S_re_croak2(aTHX_ UTF, m, REPORT_LOCATION, a1, \
735 REPORT_LOCATION_ARGS(RExC_parse)); \
738 #define vFAIL3utf8f(m, a1, a2) STMT_START { \
740 SAVEFREESV(RExC_rx_sv); \
741 S_re_croak2(aTHX_ UTF, m, REPORT_LOCATION, a1, a2, \
742 REPORT_LOCATION_ARGS(RExC_parse)); \
745 /* These have asserts in them because of [perl #122671] Many warnings in
746 * regcomp.c can occur twice. If they get output in pass1 and later in that
747 * pass, the pattern has to be converted to UTF-8 and the pass restarted, they
748 * would get output again. So they should be output in pass2, and these
749 * asserts make sure new warnings follow that paradigm. */
751 /* m is not necessarily a "literal string", in this macro */
752 #define reg_warn_non_literal_string(loc, m) STMT_START { \
753 __ASSERT_(PASS2) Perl_warner(aTHX_ packWARN(WARN_REGEXP), \
754 "%s" REPORT_LOCATION, \
755 m, REPORT_LOCATION_ARGS(loc)); \
758 #define ckWARNreg(loc,m) STMT_START { \
759 __ASSERT_(PASS2) Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), \
761 REPORT_LOCATION_ARGS(loc)); \
764 #define vWARN(loc, m) STMT_START { \
765 __ASSERT_(PASS2) Perl_warner(aTHX_ packWARN(WARN_REGEXP), \
767 REPORT_LOCATION_ARGS(loc)); \
770 #define vWARN_dep(loc, m) STMT_START { \
771 __ASSERT_(PASS2) Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), \
773 REPORT_LOCATION_ARGS(loc)); \
776 #define ckWARNdep(loc,m) STMT_START { \
777 __ASSERT_(PASS2) Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
779 REPORT_LOCATION_ARGS(loc)); \
782 #define ckWARNregdep(loc,m) STMT_START { \
783 __ASSERT_(PASS2) Perl_ck_warner_d(aTHX_ packWARN2(WARN_DEPRECATED, \
786 REPORT_LOCATION_ARGS(loc)); \
789 #define ckWARN2reg_d(loc,m, a1) STMT_START { \
790 __ASSERT_(PASS2) Perl_ck_warner_d(aTHX_ packWARN(WARN_REGEXP), \
792 a1, REPORT_LOCATION_ARGS(loc)); \
795 #define ckWARN2reg(loc, m, a1) STMT_START { \
796 __ASSERT_(PASS2) Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), \
798 a1, REPORT_LOCATION_ARGS(loc)); \
801 #define vWARN3(loc, m, a1, a2) STMT_START { \
802 __ASSERT_(PASS2) Perl_warner(aTHX_ packWARN(WARN_REGEXP), \
804 a1, a2, REPORT_LOCATION_ARGS(loc)); \
807 #define ckWARN3reg(loc, m, a1, a2) STMT_START { \
808 __ASSERT_(PASS2) Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), \
811 REPORT_LOCATION_ARGS(loc)); \
814 #define vWARN4(loc, m, a1, a2, a3) STMT_START { \
815 __ASSERT_(PASS2) Perl_warner(aTHX_ packWARN(WARN_REGEXP), \
818 REPORT_LOCATION_ARGS(loc)); \
821 #define ckWARN4reg(loc, m, a1, a2, a3) STMT_START { \
822 __ASSERT_(PASS2) Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), \
825 REPORT_LOCATION_ARGS(loc)); \
828 #define vWARN5(loc, m, a1, a2, a3, a4) STMT_START { \
829 __ASSERT_(PASS2) Perl_warner(aTHX_ packWARN(WARN_REGEXP), \
832 REPORT_LOCATION_ARGS(loc)); \
835 /* Macros for recording node offsets. 20001227 mjd@plover.com
836 * Nodes are numbered 1, 2, 3, 4. Node #n's position is recorded in
837 * element 2*n-1 of the array. Element #2n holds the byte length node #n.
838 * Element 0 holds the number n.
839 * Position is 1 indexed.
841 #ifndef RE_TRACK_PATTERN_OFFSETS
842 #define Set_Node_Offset_To_R(node,byte)
843 #define Set_Node_Offset(node,byte)
844 #define Set_Cur_Node_Offset
845 #define Set_Node_Length_To_R(node,len)
846 #define Set_Node_Length(node,len)
847 #define Set_Node_Cur_Length(node,start)
848 #define Node_Offset(n)
849 #define Node_Length(n)
850 #define Set_Node_Offset_Length(node,offset,len)
851 #define ProgLen(ri) ri->u.proglen
852 #define SetProgLen(ri,x) ri->u.proglen = x
854 #define ProgLen(ri) ri->u.offsets[0]
855 #define SetProgLen(ri,x) ri->u.offsets[0] = x
856 #define Set_Node_Offset_To_R(node,byte) STMT_START { \
858 MJD_OFFSET_DEBUG(("** (%d) offset of node %d is %d.\n", \
859 __LINE__, (int)(node), (int)(byte))); \
861 Perl_croak(aTHX_ "value of node is %d in Offset macro", \
864 RExC_offsets[2*(node)-1] = (byte); \
869 #define Set_Node_Offset(node,byte) \
870 Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
871 #define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
873 #define Set_Node_Length_To_R(node,len) STMT_START { \
875 MJD_OFFSET_DEBUG(("** (%d) size of node %d is %d.\n", \
876 __LINE__, (int)(node), (int)(len))); \
878 Perl_croak(aTHX_ "value of node is %d in Length macro", \
881 RExC_offsets[2*(node)] = (len); \
886 #define Set_Node_Length(node,len) \
887 Set_Node_Length_To_R((node)-RExC_emit_start, len)
888 #define Set_Node_Cur_Length(node, start) \
889 Set_Node_Length(node, RExC_parse - start)
891 /* Get offsets and lengths */
892 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
893 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
895 #define Set_Node_Offset_Length(node,offset,len) STMT_START { \
896 Set_Node_Offset_To_R((node)-RExC_emit_start, (offset)); \
897 Set_Node_Length_To_R((node)-RExC_emit_start, (len)); \
901 #if PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS
902 #define EXPERIMENTAL_INPLACESCAN
903 #endif /*PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS*/
907 Perl_re_printf(pTHX_ const char *fmt, ...)
911 PerlIO *f= Perl_debug_log;
912 PERL_ARGS_ASSERT_RE_PRINTF;
914 result = PerlIO_vprintf(f, fmt, ap);
920 Perl_re_indentf(pTHX_ const char *fmt, U32 depth, ...)
924 PerlIO *f= Perl_debug_log;
925 PERL_ARGS_ASSERT_RE_INDENTF;
927 PerlIO_printf(f, "%*s", ( (int)depth % 20 ) * 2, "");
928 result = PerlIO_vprintf(f, fmt, ap);
932 #endif /* DEBUGGING */
934 #define DEBUG_RExC_seen() \
935 DEBUG_OPTIMISE_MORE_r({ \
936 Perl_re_printf( aTHX_ "RExC_seen: "); \
938 if (RExC_seen & REG_ZERO_LEN_SEEN) \
939 Perl_re_printf( aTHX_ "REG_ZERO_LEN_SEEN "); \
941 if (RExC_seen & REG_LOOKBEHIND_SEEN) \
942 Perl_re_printf( aTHX_ "REG_LOOKBEHIND_SEEN "); \
944 if (RExC_seen & REG_GPOS_SEEN) \
945 Perl_re_printf( aTHX_ "REG_GPOS_SEEN "); \
947 if (RExC_seen & REG_RECURSE_SEEN) \
948 Perl_re_printf( aTHX_ "REG_RECURSE_SEEN "); \
950 if (RExC_seen & REG_TOP_LEVEL_BRANCHES_SEEN) \
951 Perl_re_printf( aTHX_ "REG_TOP_LEVEL_BRANCHES_SEEN "); \
953 if (RExC_seen & REG_VERBARG_SEEN) \
954 Perl_re_printf( aTHX_ "REG_VERBARG_SEEN "); \
956 if (RExC_seen & REG_CUTGROUP_SEEN) \
957 Perl_re_printf( aTHX_ "REG_CUTGROUP_SEEN "); \
959 if (RExC_seen & REG_RUN_ON_COMMENT_SEEN) \
960 Perl_re_printf( aTHX_ "REG_RUN_ON_COMMENT_SEEN "); \
962 if (RExC_seen & REG_UNFOLDED_MULTI_SEEN) \
963 Perl_re_printf( aTHX_ "REG_UNFOLDED_MULTI_SEEN "); \
965 if (RExC_seen & REG_UNBOUNDED_QUANTIFIER_SEEN) \
966 Perl_re_printf( aTHX_ "REG_UNBOUNDED_QUANTIFIER_SEEN "); \
968 Perl_re_printf( aTHX_ "\n"); \
971 #define DEBUG_SHOW_STUDY_FLAG(flags,flag) \
972 if ((flags) & flag) Perl_re_printf( aTHX_ "%s ", #flag)
974 #define DEBUG_SHOW_STUDY_FLAGS(flags,open_str,close_str) \
976 Perl_re_printf( aTHX_ "%s", open_str); \
977 DEBUG_SHOW_STUDY_FLAG(flags,SF_FL_BEFORE_SEOL); \
978 DEBUG_SHOW_STUDY_FLAG(flags,SF_FL_BEFORE_MEOL); \
979 DEBUG_SHOW_STUDY_FLAG(flags,SF_IS_INF); \
980 DEBUG_SHOW_STUDY_FLAG(flags,SF_HAS_PAR); \
981 DEBUG_SHOW_STUDY_FLAG(flags,SF_IN_PAR); \
982 DEBUG_SHOW_STUDY_FLAG(flags,SF_HAS_EVAL); \
983 DEBUG_SHOW_STUDY_FLAG(flags,SCF_DO_SUBSTR); \
984 DEBUG_SHOW_STUDY_FLAG(flags,SCF_DO_STCLASS_AND); \
985 DEBUG_SHOW_STUDY_FLAG(flags,SCF_DO_STCLASS_OR); \
986 DEBUG_SHOW_STUDY_FLAG(flags,SCF_DO_STCLASS); \
987 DEBUG_SHOW_STUDY_FLAG(flags,SCF_WHILEM_VISITED_POS); \
988 DEBUG_SHOW_STUDY_FLAG(flags,SCF_TRIE_RESTUDY); \
989 DEBUG_SHOW_STUDY_FLAG(flags,SCF_SEEN_ACCEPT); \
990 DEBUG_SHOW_STUDY_FLAG(flags,SCF_TRIE_DOING_RESTUDY); \
991 DEBUG_SHOW_STUDY_FLAG(flags,SCF_IN_DEFINE); \
992 Perl_re_printf( aTHX_ "%s", close_str); \
996 #define DEBUG_STUDYDATA(str,data,depth) \
997 DEBUG_OPTIMISE_MORE_r(if(data){ \
998 Perl_re_indentf( aTHX_ "" str "Pos:%" IVdf "/%" IVdf \
999 " Flags: 0x%" UVXf, \
1001 (IV)((data)->pos_min), \
1002 (IV)((data)->pos_delta), \
1003 (UV)((data)->flags) \
1005 DEBUG_SHOW_STUDY_FLAGS((data)->flags," [ ","]"); \
1006 Perl_re_printf( aTHX_ \
1007 " Whilem_c: %" IVdf " Lcp: %" IVdf " %s", \
1008 (IV)((data)->whilem_c), \
1009 (IV)((data)->last_closep ? *((data)->last_closep) : -1), \
1010 is_inf ? "INF " : "" \
1012 if ((data)->last_found) \
1013 Perl_re_printf( aTHX_ \
1014 "Last:'%s' %" IVdf ":%" IVdf "/%" IVdf \
1015 " %sFixed:'%s' @ %" IVdf \
1016 " %sFloat: '%s' @ %" IVdf "/%" IVdf, \
1017 SvPVX_const((data)->last_found), \
1018 (IV)((data)->last_end), \
1019 (IV)((data)->last_start_min), \
1020 (IV)((data)->last_start_max), \
1021 ((data)->longest && \
1022 (data)->longest==&((data)->longest_fixed)) ? "*" : "", \
1023 SvPVX_const((data)->longest_fixed), \
1024 (IV)((data)->offset_fixed), \
1025 ((data)->longest && \
1026 (data)->longest==&((data)->longest_float)) ? "*" : "", \
1027 SvPVX_const((data)->longest_float), \
1028 (IV)((data)->offset_float_min), \
1029 (IV)((data)->offset_float_max) \
1031 Perl_re_printf( aTHX_ "\n"); \
1035 /* =========================================================
1036 * BEGIN edit_distance stuff.
1038 * This calculates how many single character changes of any type are needed to
1039 * transform a string into another one. It is taken from version 3.1 of
1041 * https://metacpan.org/pod/Text::Levenshtein::Damerau::XS
1044 /* Our unsorted dictionary linked list. */
1045 /* Note we use UVs, not chars. */
1050 struct dictionary* next;
1052 typedef struct dictionary item;
1055 PERL_STATIC_INLINE item*
1056 push(UV key,item* curr)
1059 Newxz(head, 1, item);
1067 PERL_STATIC_INLINE item*
1068 find(item* head, UV key)
1070 item* iterator = head;
1072 if (iterator->key == key){
1075 iterator = iterator->next;
1081 PERL_STATIC_INLINE item*
1082 uniquePush(item* head,UV key)
1084 item* iterator = head;
1087 if (iterator->key == key) {
1090 iterator = iterator->next;
1093 return push(key,head);
1096 PERL_STATIC_INLINE void
1097 dict_free(item* head)
1099 item* iterator = head;
1102 item* temp = iterator;
1103 iterator = iterator->next;
1110 /* End of Dictionary Stuff */
1112 /* All calculations/work are done here */
1114 S_edit_distance(const UV* src,
1116 const STRLEN x, /* length of src[] */
1117 const STRLEN y, /* length of tgt[] */
1118 const SSize_t maxDistance
1122 UV swapCount,swapScore,targetCharCount,i,j;
1124 UV score_ceil = x + y;
1126 PERL_ARGS_ASSERT_EDIT_DISTANCE;
1128 /* intialize matrix start values */
1129 Newxz(scores, ( (x + 2) * (y + 2)), UV);
1130 scores[0] = score_ceil;
1131 scores[1 * (y + 2) + 0] = score_ceil;
1132 scores[0 * (y + 2) + 1] = score_ceil;
1133 scores[1 * (y + 2) + 1] = 0;
1134 head = uniquePush(uniquePush(head,src[0]),tgt[0]);
1139 for (i=1;i<=x;i++) {
1141 head = uniquePush(head,src[i]);
1142 scores[(i+1) * (y + 2) + 1] = i;
1143 scores[(i+1) * (y + 2) + 0] = score_ceil;
1146 for (j=1;j<=y;j++) {
1149 head = uniquePush(head,tgt[j]);
1150 scores[1 * (y + 2) + (j + 1)] = j;
1151 scores[0 * (y + 2) + (j + 1)] = score_ceil;
1154 targetCharCount = find(head,tgt[j-1])->value;
1155 swapScore = scores[targetCharCount * (y + 2) + swapCount] + i - targetCharCount - 1 + j - swapCount;
1157 if (src[i-1] != tgt[j-1]){
1158 scores[(i+1) * (y + 2) + (j + 1)] = MIN(swapScore,(MIN(scores[i * (y + 2) + j], MIN(scores[(i+1) * (y + 2) + j], scores[i * (y + 2) + (j + 1)])) + 1));
1162 scores[(i+1) * (y + 2) + (j + 1)] = MIN(scores[i * (y + 2) + j], swapScore);
1166 find(head,src[i-1])->value = i;
1170 IV score = scores[(x+1) * (y + 2) + (y + 1)];
1173 return (maxDistance != 0 && maxDistance < score)?(-1):score;
1177 /* END of edit_distance() stuff
1178 * ========================================================= */
1180 /* is c a control character for which we have a mnemonic? */
1181 #define isMNEMONIC_CNTRL(c) _IS_MNEMONIC_CNTRL_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c)
1184 S_cntrl_to_mnemonic(const U8 c)
1186 /* Returns the mnemonic string that represents character 'c', if one
1187 * exists; NULL otherwise. The only ones that exist for the purposes of
1188 * this routine are a few control characters */
1191 case '\a': return "\\a";
1192 case '\b': return "\\b";
1193 case ESC_NATIVE: return "\\e";
1194 case '\f': return "\\f";
1195 case '\n': return "\\n";
1196 case '\r': return "\\r";
1197 case '\t': return "\\t";
1203 /* Mark that we cannot extend a found fixed substring at this point.
1204 Update the longest found anchored substring and the longest found
1205 floating substrings if needed. */
1208 S_scan_commit(pTHX_ const RExC_state_t *pRExC_state, scan_data_t *data,
1209 SSize_t *minlenp, int is_inf)
1211 const STRLEN l = CHR_SVLEN(data->last_found);
1212 const STRLEN old_l = CHR_SVLEN(*data->longest);
1213 GET_RE_DEBUG_FLAGS_DECL;
1215 PERL_ARGS_ASSERT_SCAN_COMMIT;
1217 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
1218 SvSetMagicSV(*data->longest, data->last_found);
1219 if (*data->longest == data->longest_fixed) {
1220 data->offset_fixed = l ? data->last_start_min : data->pos_min;
1221 if (data->flags & SF_BEFORE_EOL)
1223 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
1225 data->flags &= ~SF_FIX_BEFORE_EOL;
1226 data->minlen_fixed=minlenp;
1227 data->lookbehind_fixed=0;
1229 else { /* *data->longest == data->longest_float */
1230 data->offset_float_min = l ? data->last_start_min : data->pos_min;
1231 data->offset_float_max = (l
1232 ? data->last_start_max
1233 : (data->pos_delta > SSize_t_MAX - data->pos_min
1235 : data->pos_min + data->pos_delta));
1237 || (STRLEN)data->offset_float_max > (STRLEN)SSize_t_MAX)
1238 data->offset_float_max = SSize_t_MAX;
1239 if (data->flags & SF_BEFORE_EOL)
1241 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
1243 data->flags &= ~SF_FL_BEFORE_EOL;
1244 data->minlen_float=minlenp;
1245 data->lookbehind_float=0;
1248 SvCUR_set(data->last_found, 0);
1250 SV * const sv = data->last_found;
1251 if (SvUTF8(sv) && SvMAGICAL(sv)) {
1252 MAGIC * const mg = mg_find(sv, PERL_MAGIC_utf8);
1257 data->last_end = -1;
1258 data->flags &= ~SF_BEFORE_EOL;
1259 DEBUG_STUDYDATA("commit: ",data,0);
1262 /* An SSC is just a regnode_charclass_posix with an extra field: the inversion
1263 * list that describes which code points it matches */
1266 S_ssc_anything(pTHX_ regnode_ssc *ssc)
1268 /* Set the SSC 'ssc' to match an empty string or any code point */
1270 PERL_ARGS_ASSERT_SSC_ANYTHING;
1272 assert(is_ANYOF_SYNTHETIC(ssc));
1274 /* mortalize so won't leak */
1275 ssc->invlist = sv_2mortal(_add_range_to_invlist(NULL, 0, UV_MAX));
1276 ANYOF_FLAGS(ssc) |= SSC_MATCHES_EMPTY_STRING; /* Plus matches empty */
1280 S_ssc_is_anything(const regnode_ssc *ssc)
1282 /* Returns TRUE if the SSC 'ssc' can match the empty string and any code
1283 * point; FALSE otherwise. Thus, this is used to see if using 'ssc' buys
1284 * us anything: if the function returns TRUE, 'ssc' hasn't been restricted
1285 * in any way, so there's no point in using it */
1290 PERL_ARGS_ASSERT_SSC_IS_ANYTHING;
1292 assert(is_ANYOF_SYNTHETIC(ssc));
1294 if (! (ANYOF_FLAGS(ssc) & SSC_MATCHES_EMPTY_STRING)) {
1298 /* See if the list consists solely of the range 0 - Infinity */
1299 invlist_iterinit(ssc->invlist);
1300 ret = invlist_iternext(ssc->invlist, &start, &end)
1304 invlist_iterfinish(ssc->invlist);
1310 /* If e.g., both \w and \W are set, matches everything */
1311 if (ANYOF_POSIXL_SSC_TEST_ANY_SET(ssc)) {
1313 for (i = 0; i < ANYOF_POSIXL_MAX; i += 2) {
1314 if (ANYOF_POSIXL_TEST(ssc, i) && ANYOF_POSIXL_TEST(ssc, i+1)) {
1324 S_ssc_init(pTHX_ const RExC_state_t *pRExC_state, regnode_ssc *ssc)
1326 /* Initializes the SSC 'ssc'. This includes setting it to match an empty
1327 * string, any code point, or any posix class under locale */
1329 PERL_ARGS_ASSERT_SSC_INIT;
1331 Zero(ssc, 1, regnode_ssc);
1332 set_ANYOF_SYNTHETIC(ssc);
1333 ARG_SET(ssc, ANYOF_ONLY_HAS_BITMAP);
1336 /* If any portion of the regex is to operate under locale rules that aren't
1337 * fully known at compile time, initialization includes it. The reason
1338 * this isn't done for all regexes is that the optimizer was written under
1339 * the assumption that locale was all-or-nothing. Given the complexity and
1340 * lack of documentation in the optimizer, and that there are inadequate
1341 * test cases for locale, many parts of it may not work properly, it is
1342 * safest to avoid locale unless necessary. */
1343 if (RExC_contains_locale) {
1344 ANYOF_POSIXL_SETALL(ssc);
1347 ANYOF_POSIXL_ZERO(ssc);
1352 S_ssc_is_cp_posixl_init(const RExC_state_t *pRExC_state,
1353 const regnode_ssc *ssc)
1355 /* Returns TRUE if the SSC 'ssc' is in its initial state with regard only
1356 * to the list of code points matched, and locale posix classes; hence does
1357 * not check its flags) */
1362 PERL_ARGS_ASSERT_SSC_IS_CP_POSIXL_INIT;
1364 assert(is_ANYOF_SYNTHETIC(ssc));
1366 invlist_iterinit(ssc->invlist);
1367 ret = invlist_iternext(ssc->invlist, &start, &end)
1371 invlist_iterfinish(ssc->invlist);
1377 if (RExC_contains_locale && ! ANYOF_POSIXL_SSC_TEST_ALL_SET(ssc)) {
1385 S_get_ANYOF_cp_list_for_ssc(pTHX_ const RExC_state_t *pRExC_state,
1386 const regnode_charclass* const node)
1388 /* Returns a mortal inversion list defining which code points are matched
1389 * by 'node', which is of type ANYOF. Handles complementing the result if
1390 * appropriate. If some code points aren't knowable at this time, the
1391 * returned list must, and will, contain every code point that is a
1395 SV* only_utf8_locale_invlist = NULL;
1397 const U32 n = ARG(node);
1398 bool new_node_has_latin1 = FALSE;
1400 PERL_ARGS_ASSERT_GET_ANYOF_CP_LIST_FOR_SSC;
1402 /* Look at the data structure created by S_set_ANYOF_arg() */
1403 if (n != ANYOF_ONLY_HAS_BITMAP) {
1404 SV * const rv = MUTABLE_SV(RExC_rxi->data->data[n]);
1405 AV * const av = MUTABLE_AV(SvRV(rv));
1406 SV **const ary = AvARRAY(av);
1407 assert(RExC_rxi->data->what[n] == 's');
1409 if (ary[1] && ary[1] != &PL_sv_undef) { /* Has compile-time swash */
1410 invlist = sv_2mortal(invlist_clone(_get_swash_invlist(ary[1])));
1412 else if (ary[0] && ary[0] != &PL_sv_undef) {
1414 /* Here, no compile-time swash, and there are things that won't be
1415 * known until runtime -- we have to assume it could be anything */
1416 invlist = sv_2mortal(_new_invlist(1));
1417 return _add_range_to_invlist(invlist, 0, UV_MAX);
1419 else if (ary[3] && ary[3] != &PL_sv_undef) {
1421 /* Here no compile-time swash, and no run-time only data. Use the
1422 * node's inversion list */
1423 invlist = sv_2mortal(invlist_clone(ary[3]));
1426 /* Get the code points valid only under UTF-8 locales */
1427 if ((ANYOF_FLAGS(node) & ANYOFL_FOLD)
1428 && ary[2] && ary[2] != &PL_sv_undef)
1430 only_utf8_locale_invlist = ary[2];
1435 invlist = sv_2mortal(_new_invlist(0));
1438 /* An ANYOF node contains a bitmap for the first NUM_ANYOF_CODE_POINTS
1439 * code points, and an inversion list for the others, but if there are code
1440 * points that should match only conditionally on the target string being
1441 * UTF-8, those are placed in the inversion list, and not the bitmap.
1442 * Since there are circumstances under which they could match, they are
1443 * included in the SSC. But if the ANYOF node is to be inverted, we have
1444 * to exclude them here, so that when we invert below, the end result
1445 * actually does include them. (Think about "\xe0" =~ /[^\xc0]/di;). We
1446 * have to do this here before we add the unconditionally matched code
1448 if (ANYOF_FLAGS(node) & ANYOF_INVERT) {
1449 _invlist_intersection_complement_2nd(invlist,
1454 /* Add in the points from the bit map */
1455 for (i = 0; i < NUM_ANYOF_CODE_POINTS; i++) {
1456 if (ANYOF_BITMAP_TEST(node, i)) {
1457 unsigned int start = i++;
1459 for (; i < NUM_ANYOF_CODE_POINTS && ANYOF_BITMAP_TEST(node, i); ++i) {
1462 invlist = _add_range_to_invlist(invlist, start, i-1);
1463 new_node_has_latin1 = TRUE;
1467 /* If this can match all upper Latin1 code points, have to add them
1468 * as well. But don't add them if inverting, as when that gets done below,
1469 * it would exclude all these characters, including the ones it shouldn't
1470 * that were added just above */
1471 if (! (ANYOF_FLAGS(node) & ANYOF_INVERT) && OP(node) == ANYOFD
1472 && (ANYOF_FLAGS(node) & ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER))
1474 _invlist_union(invlist, PL_UpperLatin1, &invlist);
1477 /* Similarly for these */
1478 if (ANYOF_FLAGS(node) & ANYOF_MATCHES_ALL_ABOVE_BITMAP) {
1479 _invlist_union_complement_2nd(invlist, PL_InBitmap, &invlist);
1482 if (ANYOF_FLAGS(node) & ANYOF_INVERT) {
1483 _invlist_invert(invlist);
1485 else if (new_node_has_latin1 && ANYOF_FLAGS(node) & ANYOFL_FOLD) {
1487 /* Under /li, any 0-255 could fold to any other 0-255, depending on the
1488 * locale. We can skip this if there are no 0-255 at all. */
1489 _invlist_union(invlist, PL_Latin1, &invlist);
1492 /* Similarly add the UTF-8 locale possible matches. These have to be
1493 * deferred until after the non-UTF-8 locale ones are taken care of just
1494 * above, or it leads to wrong results under ANYOF_INVERT */
1495 if (only_utf8_locale_invlist) {
1496 _invlist_union_maybe_complement_2nd(invlist,
1497 only_utf8_locale_invlist,
1498 ANYOF_FLAGS(node) & ANYOF_INVERT,
1505 /* These two functions currently do the exact same thing */
1506 #define ssc_init_zero ssc_init
1508 #define ssc_add_cp(ssc, cp) ssc_add_range((ssc), (cp), (cp))
1509 #define ssc_match_all_cp(ssc) ssc_add_range(ssc, 0, UV_MAX)
1511 /* 'AND' a given class with another one. Can create false positives. 'ssc'
1512 * should not be inverted. 'and_with->flags & ANYOF_MATCHES_POSIXL' should be
1513 * 0 if 'and_with' is a regnode_charclass instead of a regnode_ssc. */
1516 S_ssc_and(pTHX_ const RExC_state_t *pRExC_state, regnode_ssc *ssc,
1517 const regnode_charclass *and_with)
1519 /* Accumulate into SSC 'ssc' its 'AND' with 'and_with', which is either
1520 * another SSC or a regular ANYOF class. Can create false positives. */
1525 PERL_ARGS_ASSERT_SSC_AND;
1527 assert(is_ANYOF_SYNTHETIC(ssc));
1529 /* 'and_with' is used as-is if it too is an SSC; otherwise have to extract
1530 * the code point inversion list and just the relevant flags */
1531 if (is_ANYOF_SYNTHETIC(and_with)) {
1532 anded_cp_list = ((regnode_ssc *)and_with)->invlist;
1533 anded_flags = ANYOF_FLAGS(and_with);
1535 /* XXX This is a kludge around what appears to be deficiencies in the
1536 * optimizer. If we make S_ssc_anything() add in the WARN_SUPER flag,
1537 * there are paths through the optimizer where it doesn't get weeded
1538 * out when it should. And if we don't make some extra provision for
1539 * it like the code just below, it doesn't get added when it should.
1540 * This solution is to add it only when AND'ing, which is here, and
1541 * only when what is being AND'ed is the pristine, original node
1542 * matching anything. Thus it is like adding it to ssc_anything() but
1543 * only when the result is to be AND'ed. Probably the same solution
1544 * could be adopted for the same problem we have with /l matching,
1545 * which is solved differently in S_ssc_init(), and that would lead to
1546 * fewer false positives than that solution has. But if this solution
1547 * creates bugs, the consequences are only that a warning isn't raised
1548 * that should be; while the consequences for having /l bugs is
1549 * incorrect matches */
1550 if (ssc_is_anything((regnode_ssc *)and_with)) {
1551 anded_flags |= ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER;
1555 anded_cp_list = get_ANYOF_cp_list_for_ssc(pRExC_state, and_with);
1556 if (OP(and_with) == ANYOFD) {
1557 anded_flags = ANYOF_FLAGS(and_with) & ANYOF_COMMON_FLAGS;
1560 anded_flags = ANYOF_FLAGS(and_with)
1561 &( ANYOF_COMMON_FLAGS
1562 |ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER
1563 |ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP);
1564 if (ANYOFL_UTF8_LOCALE_REQD(ANYOF_FLAGS(and_with))) {
1566 ANYOFL_SHARED_UTF8_LOCALE_fold_HAS_MATCHES_nonfold_REQD;
1571 ANYOF_FLAGS(ssc) &= anded_flags;
1573 /* Below, C1 is the list of code points in 'ssc'; P1, its posix classes.
1574 * C2 is the list of code points in 'and-with'; P2, its posix classes.
1575 * 'and_with' may be inverted. When not inverted, we have the situation of
1577 * (C1 | P1) & (C2 | P2)
1578 * = (C1 & (C2 | P2)) | (P1 & (C2 | P2))
1579 * = ((C1 & C2) | (C1 & P2)) | ((P1 & C2) | (P1 & P2))
1580 * <= ((C1 & C2) | P2)) | ( P1 | (P1 & P2))
1581 * <= ((C1 & C2) | P1 | P2)
1582 * Alternatively, the last few steps could be:
1583 * = ((C1 & C2) | (C1 & P2)) | ((P1 & C2) | (P1 & P2))
1584 * <= ((C1 & C2) | C1 ) | ( C2 | (P1 & P2))
1585 * <= (C1 | C2 | (P1 & P2))
1586 * We favor the second approach if either P1 or P2 is non-empty. This is
1587 * because these components are a barrier to doing optimizations, as what
1588 * they match cannot be known until the moment of matching as they are
1589 * dependent on the current locale, 'AND"ing them likely will reduce or
1591 * But we can do better if we know that C1,P1 are in their initial state (a
1592 * frequent occurrence), each matching everything:
1593 * (<everything>) & (C2 | P2) = C2 | P2
1594 * Similarly, if C2,P2 are in their initial state (again a frequent
1595 * occurrence), the result is a no-op
1596 * (C1 | P1) & (<everything>) = C1 | P1
1599 * (C1 | P1) & ~(C2 | P2) = (C1 | P1) & (~C2 & ~P2)
1600 * = (C1 & (~C2 & ~P2)) | (P1 & (~C2 & ~P2))
1601 * <= (C1 & ~C2) | (P1 & ~P2)
1604 if ((ANYOF_FLAGS(and_with) & ANYOF_INVERT)
1605 && ! is_ANYOF_SYNTHETIC(and_with))
1609 ssc_intersection(ssc,
1611 FALSE /* Has already been inverted */
1614 /* If either P1 or P2 is empty, the intersection will be also; can skip
1616 if (! (ANYOF_FLAGS(and_with) & ANYOF_MATCHES_POSIXL)) {
1617 ANYOF_POSIXL_ZERO(ssc);
1619 else if (ANYOF_POSIXL_SSC_TEST_ANY_SET(ssc)) {
1621 /* Note that the Posix class component P from 'and_with' actually
1623 * P = Pa | Pb | ... | Pn
1624 * where each component is one posix class, such as in [\w\s].
1626 * ~P = ~(Pa | Pb | ... | Pn)
1627 * = ~Pa & ~Pb & ... & ~Pn
1628 * <= ~Pa | ~Pb | ... | ~Pn
1629 * The last is something we can easily calculate, but unfortunately
1630 * is likely to have many false positives. We could do better
1631 * in some (but certainly not all) instances if two classes in
1632 * P have known relationships. For example
1633 * :lower: <= :alpha: <= :alnum: <= \w <= :graph: <= :print:
1635 * :lower: & :print: = :lower:
1636 * And similarly for classes that must be disjoint. For example,
1637 * since \s and \w can have no elements in common based on rules in
1638 * the POSIX standard,
1639 * \w & ^\S = nothing
1640 * Unfortunately, some vendor locales do not meet the Posix
1641 * standard, in particular almost everything by Microsoft.
1642 * The loop below just changes e.g., \w into \W and vice versa */
1644 regnode_charclass_posixl temp;
1645 int add = 1; /* To calculate the index of the complement */
1647 ANYOF_POSIXL_ZERO(&temp);
1648 for (i = 0; i < ANYOF_MAX; i++) {
1650 || ! ANYOF_POSIXL_TEST((regnode_charclass_posixl*) and_with, i)
1651 || ! ANYOF_POSIXL_TEST((regnode_charclass_posixl*) and_with, i + 1));
1653 if (ANYOF_POSIXL_TEST((regnode_charclass_posixl*) and_with, i)) {
1654 ANYOF_POSIXL_SET(&temp, i + add);
1656 add = 0 - add; /* 1 goes to -1; -1 goes to 1 */
1658 ANYOF_POSIXL_AND(&temp, ssc);
1660 } /* else ssc already has no posixes */
1661 } /* else: Not inverted. This routine is a no-op if 'and_with' is an SSC
1662 in its initial state */
1663 else if (! is_ANYOF_SYNTHETIC(and_with)
1664 || ! ssc_is_cp_posixl_init(pRExC_state, (regnode_ssc *)and_with))
1666 /* But if 'ssc' is in its initial state, the result is just 'and_with';
1667 * copy it over 'ssc' */
1668 if (ssc_is_cp_posixl_init(pRExC_state, ssc)) {
1669 if (is_ANYOF_SYNTHETIC(and_with)) {
1670 StructCopy(and_with, ssc, regnode_ssc);
1673 ssc->invlist = anded_cp_list;
1674 ANYOF_POSIXL_ZERO(ssc);
1675 if (ANYOF_FLAGS(and_with) & ANYOF_MATCHES_POSIXL) {
1676 ANYOF_POSIXL_OR((regnode_charclass_posixl*) and_with, ssc);
1680 else if (ANYOF_POSIXL_SSC_TEST_ANY_SET(ssc)
1681 || (ANYOF_FLAGS(and_with) & ANYOF_MATCHES_POSIXL))
1683 /* One or the other of P1, P2 is non-empty. */
1684 if (ANYOF_FLAGS(and_with) & ANYOF_MATCHES_POSIXL) {
1685 ANYOF_POSIXL_AND((regnode_charclass_posixl*) and_with, ssc);
1687 ssc_union(ssc, anded_cp_list, FALSE);
1689 else { /* P1 = P2 = empty */
1690 ssc_intersection(ssc, anded_cp_list, FALSE);
1696 S_ssc_or(pTHX_ const RExC_state_t *pRExC_state, regnode_ssc *ssc,
1697 const regnode_charclass *or_with)
1699 /* Accumulate into SSC 'ssc' its 'OR' with 'or_with', which is either
1700 * another SSC or a regular ANYOF class. Can create false positives if
1701 * 'or_with' is to be inverted. */
1706 PERL_ARGS_ASSERT_SSC_OR;
1708 assert(is_ANYOF_SYNTHETIC(ssc));
1710 /* 'or_with' is used as-is if it too is an SSC; otherwise have to extract
1711 * the code point inversion list and just the relevant flags */
1712 if (is_ANYOF_SYNTHETIC(or_with)) {
1713 ored_cp_list = ((regnode_ssc*) or_with)->invlist;
1714 ored_flags = ANYOF_FLAGS(or_with);
1717 ored_cp_list = get_ANYOF_cp_list_for_ssc(pRExC_state, or_with);
1718 ored_flags = ANYOF_FLAGS(or_with) & ANYOF_COMMON_FLAGS;
1719 if (OP(or_with) != ANYOFD) {
1721 |= ANYOF_FLAGS(or_with)
1722 & ( ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER
1723 |ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP);
1724 if (ANYOFL_UTF8_LOCALE_REQD(ANYOF_FLAGS(or_with))) {
1726 ANYOFL_SHARED_UTF8_LOCALE_fold_HAS_MATCHES_nonfold_REQD;
1731 ANYOF_FLAGS(ssc) |= ored_flags;
1733 /* Below, C1 is the list of code points in 'ssc'; P1, its posix classes.
1734 * C2 is the list of code points in 'or-with'; P2, its posix classes.
1735 * 'or_with' may be inverted. When not inverted, we have the simple
1736 * situation of computing:
1737 * (C1 | P1) | (C2 | P2) = (C1 | C2) | (P1 | P2)
1738 * If P1|P2 yields a situation with both a class and its complement are
1739 * set, like having both \w and \W, this matches all code points, and we
1740 * can delete these from the P component of the ssc going forward. XXX We
1741 * might be able to delete all the P components, but I (khw) am not certain
1742 * about this, and it is better to be safe.
1745 * (C1 | P1) | ~(C2 | P2) = (C1 | P1) | (~C2 & ~P2)
1746 * <= (C1 | P1) | ~C2
1747 * <= (C1 | ~C2) | P1
1748 * (which results in actually simpler code than the non-inverted case)
1751 if ((ANYOF_FLAGS(or_with) & ANYOF_INVERT)
1752 && ! is_ANYOF_SYNTHETIC(or_with))
1754 /* We ignore P2, leaving P1 going forward */
1755 } /* else Not inverted */
1756 else if (ANYOF_FLAGS(or_with) & ANYOF_MATCHES_POSIXL) {
1757 ANYOF_POSIXL_OR((regnode_charclass_posixl*)or_with, ssc);
1758 if (ANYOF_POSIXL_SSC_TEST_ANY_SET(ssc)) {
1760 for (i = 0; i < ANYOF_MAX; i += 2) {
1761 if (ANYOF_POSIXL_TEST(ssc, i) && ANYOF_POSIXL_TEST(ssc, i + 1))
1763 ssc_match_all_cp(ssc);
1764 ANYOF_POSIXL_CLEAR(ssc, i);
1765 ANYOF_POSIXL_CLEAR(ssc, i+1);
1773 FALSE /* Already has been inverted */
1777 PERL_STATIC_INLINE void
1778 S_ssc_union(pTHX_ regnode_ssc *ssc, SV* const invlist, const bool invert2nd)
1780 PERL_ARGS_ASSERT_SSC_UNION;
1782 assert(is_ANYOF_SYNTHETIC(ssc));
1784 _invlist_union_maybe_complement_2nd(ssc->invlist,
1790 PERL_STATIC_INLINE void
1791 S_ssc_intersection(pTHX_ regnode_ssc *ssc,
1793 const bool invert2nd)
1795 PERL_ARGS_ASSERT_SSC_INTERSECTION;
1797 assert(is_ANYOF_SYNTHETIC(ssc));
1799 _invlist_intersection_maybe_complement_2nd(ssc->invlist,
1805 PERL_STATIC_INLINE void
1806 S_ssc_add_range(pTHX_ regnode_ssc *ssc, const UV start, const UV end)
1808 PERL_ARGS_ASSERT_SSC_ADD_RANGE;
1810 assert(is_ANYOF_SYNTHETIC(ssc));
1812 ssc->invlist = _add_range_to_invlist(ssc->invlist, start, end);
1815 PERL_STATIC_INLINE void
1816 S_ssc_cp_and(pTHX_ regnode_ssc *ssc, const UV cp)
1818 /* AND just the single code point 'cp' into the SSC 'ssc' */
1820 SV* cp_list = _new_invlist(2);
1822 PERL_ARGS_ASSERT_SSC_CP_AND;
1824 assert(is_ANYOF_SYNTHETIC(ssc));
1826 cp_list = add_cp_to_invlist(cp_list, cp);
1827 ssc_intersection(ssc, cp_list,
1828 FALSE /* Not inverted */
1830 SvREFCNT_dec_NN(cp_list);
1833 PERL_STATIC_INLINE void
1834 S_ssc_clear_locale(regnode_ssc *ssc)
1836 /* Set the SSC 'ssc' to not match any locale things */
1837 PERL_ARGS_ASSERT_SSC_CLEAR_LOCALE;
1839 assert(is_ANYOF_SYNTHETIC(ssc));
1841 ANYOF_POSIXL_ZERO(ssc);
1842 ANYOF_FLAGS(ssc) &= ~ANYOF_LOCALE_FLAGS;
1845 #define NON_OTHER_COUNT NON_OTHER_COUNT_FOR_USE_ONLY_BY_REGCOMP_DOT_C
1848 S_is_ssc_worth_it(const RExC_state_t * pRExC_state, const regnode_ssc * ssc)
1850 /* The synthetic start class is used to hopefully quickly winnow down
1851 * places where a pattern could start a match in the target string. If it
1852 * doesn't really narrow things down that much, there isn't much point to
1853 * having the overhead of using it. This function uses some very crude
1854 * heuristics to decide if to use the ssc or not.
1856 * It returns TRUE if 'ssc' rules out more than half what it considers to
1857 * be the "likely" possible matches, but of course it doesn't know what the
1858 * actual things being matched are going to be; these are only guesses
1860 * For /l matches, it assumes that the only likely matches are going to be
1861 * in the 0-255 range, uniformly distributed, so half of that is 127
1862 * For /a and /d matches, it assumes that the likely matches will be just
1863 * the ASCII range, so half of that is 63
1864 * For /u and there isn't anything matching above the Latin1 range, it
1865 * assumes that that is the only range likely to be matched, and uses
1866 * half that as the cut-off: 127. If anything matches above Latin1,
1867 * it assumes that all of Unicode could match (uniformly), except for
1868 * non-Unicode code points and things in the General Category "Other"
1869 * (unassigned, private use, surrogates, controls and formats). This
1870 * is a much large number. */
1872 U32 count = 0; /* Running total of number of code points matched by
1874 UV start, end; /* Start and end points of current range in inversion
1876 const U32 max_code_points = (LOC)
1878 : (( ! UNI_SEMANTICS
1879 || invlist_highest(ssc->invlist) < 256)
1882 const U32 max_match = max_code_points / 2;
1884 PERL_ARGS_ASSERT_IS_SSC_WORTH_IT;
1886 invlist_iterinit(ssc->invlist);
1887 while (invlist_iternext(ssc->invlist, &start, &end)) {
1888 if (start >= max_code_points) {
1891 end = MIN(end, max_code_points - 1);
1892 count += end - start + 1;
1893 if (count >= max_match) {
1894 invlist_iterfinish(ssc->invlist);
1904 S_ssc_finalize(pTHX_ RExC_state_t *pRExC_state, regnode_ssc *ssc)
1906 /* The inversion list in the SSC is marked mortal; now we need a more
1907 * permanent copy, which is stored the same way that is done in a regular
1908 * ANYOF node, with the first NUM_ANYOF_CODE_POINTS code points in a bit
1911 SV* invlist = invlist_clone(ssc->invlist);
1913 PERL_ARGS_ASSERT_SSC_FINALIZE;
1915 assert(is_ANYOF_SYNTHETIC(ssc));
1917 /* The code in this file assumes that all but these flags aren't relevant
1918 * to the SSC, except SSC_MATCHES_EMPTY_STRING, which should be cleared
1919 * by the time we reach here */
1920 assert(! (ANYOF_FLAGS(ssc)
1921 & ~( ANYOF_COMMON_FLAGS
1922 |ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER
1923 |ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP)));
1925 populate_ANYOF_from_invlist( (regnode *) ssc, &invlist);
1927 set_ANYOF_arg(pRExC_state, (regnode *) ssc, invlist,
1928 NULL, NULL, NULL, FALSE);
1930 /* Make sure is clone-safe */
1931 ssc->invlist = NULL;
1933 if (ANYOF_POSIXL_SSC_TEST_ANY_SET(ssc)) {
1934 ANYOF_FLAGS(ssc) |= ANYOF_MATCHES_POSIXL;
1937 if (RExC_contains_locale) {
1941 assert(! (ANYOF_FLAGS(ssc) & ANYOF_LOCALE_FLAGS) || RExC_contains_locale);
1944 #define TRIE_LIST_ITEM(state,idx) (trie->states[state].trans.list)[ idx ]
1945 #define TRIE_LIST_CUR(state) ( TRIE_LIST_ITEM( state, 0 ).forid )
1946 #define TRIE_LIST_LEN(state) ( TRIE_LIST_ITEM( state, 0 ).newstate )
1947 #define TRIE_LIST_USED(idx) ( trie->states[state].trans.list \
1948 ? (TRIE_LIST_CUR( idx ) - 1) \
1954 dump_trie(trie,widecharmap,revcharmap)
1955 dump_trie_interim_list(trie,widecharmap,revcharmap,next_alloc)
1956 dump_trie_interim_table(trie,widecharmap,revcharmap,next_alloc)
1958 These routines dump out a trie in a somewhat readable format.
1959 The _interim_ variants are used for debugging the interim
1960 tables that are used to generate the final compressed
1961 representation which is what dump_trie expects.
1963 Part of the reason for their existence is to provide a form
1964 of documentation as to how the different representations function.
1969 Dumps the final compressed table form of the trie to Perl_debug_log.
1970 Used for debugging make_trie().
1974 S_dump_trie(pTHX_ const struct _reg_trie_data *trie, HV *widecharmap,
1975 AV *revcharmap, U32 depth)
1978 SV *sv=sv_newmortal();
1979 int colwidth= widecharmap ? 6 : 4;
1981 GET_RE_DEBUG_FLAGS_DECL;
1983 PERL_ARGS_ASSERT_DUMP_TRIE;
1985 Perl_re_indentf( aTHX_ "Char : %-6s%-6s%-4s ",
1986 depth+1, "Match","Base","Ofs" );
1988 for( state = 0 ; state < trie->uniquecharcount ; state++ ) {
1989 SV ** const tmp = av_fetch( revcharmap, state, 0);
1991 Perl_re_printf( aTHX_ "%*s",
1993 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
1994 PL_colors[0], PL_colors[1],
1995 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
1996 PERL_PV_ESCAPE_FIRSTCHAR
2001 Perl_re_printf( aTHX_ "\n");
2002 Perl_re_indentf( aTHX_ "State|-----------------------", depth+1);
2004 for( state = 0 ; state < trie->uniquecharcount ; state++ )
2005 Perl_re_printf( aTHX_ "%.*s", colwidth, "--------");
2006 Perl_re_printf( aTHX_ "\n");
2008 for( state = 1 ; state < trie->statecount ; state++ ) {
2009 const U32 base = trie->states[ state ].trans.base;
2011 Perl_re_indentf( aTHX_ "#%4" UVXf "|", depth+1, (UV)state);
2013 if ( trie->states[ state ].wordnum ) {
2014 Perl_re_printf( aTHX_ " W%4X", trie->states[ state ].wordnum );
2016 Perl_re_printf( aTHX_ "%6s", "" );
2019 Perl_re_printf( aTHX_ " @%4" UVXf " ", (UV)base );
2024 while( ( base + ofs < trie->uniquecharcount ) ||
2025 ( base + ofs - trie->uniquecharcount < trie->lasttrans
2026 && trie->trans[ base + ofs - trie->uniquecharcount ].check
2030 Perl_re_printf( aTHX_ "+%2" UVXf "[ ", (UV)ofs);
2032 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
2033 if ( ( base + ofs >= trie->uniquecharcount )
2034 && ( base + ofs - trie->uniquecharcount
2036 && trie->trans[ base + ofs
2037 - trie->uniquecharcount ].check == state )
2039 Perl_re_printf( aTHX_ "%*" UVXf, colwidth,
2040 (UV)trie->trans[ base + ofs - trie->uniquecharcount ].next
2043 Perl_re_printf( aTHX_ "%*s",colwidth," ." );
2047 Perl_re_printf( aTHX_ "]");
2050 Perl_re_printf( aTHX_ "\n" );
2052 Perl_re_indentf( aTHX_ "word_info N:(prev,len)=",
2054 for (word=1; word <= trie->wordcount; word++) {
2055 Perl_re_printf( aTHX_ " %d:(%d,%d)",
2056 (int)word, (int)(trie->wordinfo[word].prev),
2057 (int)(trie->wordinfo[word].len));
2059 Perl_re_printf( aTHX_ "\n" );
2062 Dumps a fully constructed but uncompressed trie in list form.
2063 List tries normally only are used for construction when the number of
2064 possible chars (trie->uniquecharcount) is very high.
2065 Used for debugging make_trie().
2068 S_dump_trie_interim_list(pTHX_ const struct _reg_trie_data *trie,
2069 HV *widecharmap, AV *revcharmap, U32 next_alloc,
2073 SV *sv=sv_newmortal();
2074 int colwidth= widecharmap ? 6 : 4;
2075 GET_RE_DEBUG_FLAGS_DECL;
2077 PERL_ARGS_ASSERT_DUMP_TRIE_INTERIM_LIST;
2079 /* print out the table precompression. */
2080 Perl_re_indentf( aTHX_ "State :Word | Transition Data\n",
2082 Perl_re_indentf( aTHX_ "%s",
2083 depth+1, "------:-----+-----------------\n" );
2085 for( state=1 ; state < next_alloc ; state ++ ) {
2088 Perl_re_indentf( aTHX_ " %4" UVXf " :",
2089 depth+1, (UV)state );
2090 if ( ! trie->states[ state ].wordnum ) {
2091 Perl_re_printf( aTHX_ "%5s| ","");
2093 Perl_re_printf( aTHX_ "W%4x| ",
2094 trie->states[ state ].wordnum
2097 for( charid = 1 ; charid <= TRIE_LIST_USED( state ) ; charid++ ) {
2098 SV ** const tmp = av_fetch( revcharmap,
2099 TRIE_LIST_ITEM(state,charid).forid, 0);
2101 Perl_re_printf( aTHX_ "%*s:%3X=%4" UVXf " | ",
2103 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp),
2105 PL_colors[0], PL_colors[1],
2106 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)
2107 | PERL_PV_ESCAPE_FIRSTCHAR
2109 TRIE_LIST_ITEM(state,charid).forid,
2110 (UV)TRIE_LIST_ITEM(state,charid).newstate
2113 Perl_re_printf( aTHX_ "\n%*s| ",
2114 (int)((depth * 2) + 14), "");
2117 Perl_re_printf( aTHX_ "\n");
2122 Dumps a fully constructed but uncompressed trie in table form.
2123 This is the normal DFA style state transition table, with a few
2124 twists to facilitate compression later.
2125 Used for debugging make_trie().
2128 S_dump_trie_interim_table(pTHX_ const struct _reg_trie_data *trie,
2129 HV *widecharmap, AV *revcharmap, U32 next_alloc,
2134 SV *sv=sv_newmortal();
2135 int colwidth= widecharmap ? 6 : 4;
2136 GET_RE_DEBUG_FLAGS_DECL;
2138 PERL_ARGS_ASSERT_DUMP_TRIE_INTERIM_TABLE;
2141 print out the table precompression so that we can do a visual check
2142 that they are identical.
2145 Perl_re_indentf( aTHX_ "Char : ", depth+1 );
2147 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
2148 SV ** const tmp = av_fetch( revcharmap, charid, 0);
2150 Perl_re_printf( aTHX_ "%*s",
2152 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
2153 PL_colors[0], PL_colors[1],
2154 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
2155 PERL_PV_ESCAPE_FIRSTCHAR
2161 Perl_re_printf( aTHX_ "\n");
2162 Perl_re_indentf( aTHX_ "State+-", depth+1 );
2164 for( charid=0 ; charid < trie->uniquecharcount ; charid++ ) {
2165 Perl_re_printf( aTHX_ "%.*s", colwidth,"--------");
2168 Perl_re_printf( aTHX_ "\n" );
2170 for( state=1 ; state < next_alloc ; state += trie->uniquecharcount ) {
2172 Perl_re_indentf( aTHX_ "%4" UVXf " : ",
2174 (UV)TRIE_NODENUM( state ) );
2176 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
2177 UV v=(UV)SAFE_TRIE_NODENUM( trie->trans[ state + charid ].next );
2179 Perl_re_printf( aTHX_ "%*" UVXf, colwidth, v );
2181 Perl_re_printf( aTHX_ "%*s", colwidth, "." );
2183 if ( ! trie->states[ TRIE_NODENUM( state ) ].wordnum ) {
2184 Perl_re_printf( aTHX_ " (%4" UVXf ")\n",
2185 (UV)trie->trans[ state ].check );
2187 Perl_re_printf( aTHX_ " (%4" UVXf ") W%4X\n",
2188 (UV)trie->trans[ state ].check,
2189 trie->states[ TRIE_NODENUM( state ) ].wordnum );
2197 /* make_trie(startbranch,first,last,tail,word_count,flags,depth)
2198 startbranch: the first branch in the whole branch sequence
2199 first : start branch of sequence of branch-exact nodes.
2200 May be the same as startbranch
2201 last : Thing following the last branch.
2202 May be the same as tail.
2203 tail : item following the branch sequence
2204 count : words in the sequence
2205 flags : currently the OP() type we will be building one of /EXACT(|F|FA|FU|FU_SS|L|FLU8)/
2206 depth : indent depth
2208 Inplace optimizes a sequence of 2 or more Branch-Exact nodes into a TRIE node.
2210 A trie is an N'ary tree where the branches are determined by digital
2211 decomposition of the key. IE, at the root node you look up the 1st character and
2212 follow that branch repeat until you find the end of the branches. Nodes can be
2213 marked as "accepting" meaning they represent a complete word. Eg:
2217 would convert into the following structure. Numbers represent states, letters
2218 following numbers represent valid transitions on the letter from that state, if
2219 the number is in square brackets it represents an accepting state, otherwise it
2220 will be in parenthesis.
2222 +-h->+-e->[3]-+-r->(8)-+-s->[9]
2226 (1) +-i->(6)-+-s->[7]
2228 +-s->(3)-+-h->(4)-+-e->[5]
2230 Accept Word Mapping: 3=>1 (he),5=>2 (she), 7=>3 (his), 9=>4 (hers)
2232 This shows that when matching against the string 'hers' we will begin at state 1
2233 read 'h' and move to state 2, read 'e' and move to state 3 which is accepting,
2234 then read 'r' and go to state 8 followed by 's' which takes us to state 9 which
2235 is also accepting. Thus we know that we can match both 'he' and 'hers' with a
2236 single traverse. We store a mapping from accepting to state to which word was
2237 matched, and then when we have multiple possibilities we try to complete the
2238 rest of the regex in the order in which they occurred in the alternation.
2240 The only prior NFA like behaviour that would be changed by the TRIE support is
2241 the silent ignoring of duplicate alternations which are of the form:
2243 / (DUPE|DUPE) X? (?{ ... }) Y /x
2245 Thus EVAL blocks following a trie may be called a different number of times with
2246 and without the optimisation. With the optimisations dupes will be silently
2247 ignored. This inconsistent behaviour of EVAL type nodes is well established as
2248 the following demonstrates:
2250 'words'=~/(word|word|word)(?{ print $1 })[xyz]/
2252 which prints out 'word' three times, but
2254 'words'=~/(word|word|word)(?{ print $1 })S/
2256 which doesnt print it out at all. This is due to other optimisations kicking in.
2258 Example of what happens on a structural level:
2260 The regexp /(ac|ad|ab)+/ will produce the following debug output:
2262 1: CURLYM[1] {1,32767}(18)
2273 This would be optimizable with startbranch=5, first=5, last=16, tail=16
2274 and should turn into:
2276 1: CURLYM[1] {1,32767}(18)
2278 [Words:3 Chars Stored:6 Unique Chars:4 States:5 NCP:1]
2286 Cases where tail != last would be like /(?foo|bar)baz/:
2296 which would be optimizable with startbranch=1, first=1, last=7, tail=8
2297 and would end up looking like:
2300 [Words:2 Chars Stored:6 Unique Chars:5 States:7 NCP:1]
2307 d = uvchr_to_utf8_flags(d, uv, 0);
2309 is the recommended Unicode-aware way of saying
2314 #define TRIE_STORE_REVCHAR(val) \
2317 SV *zlopp = newSV(UTF8_MAXBYTES); \
2318 unsigned char *flrbbbbb = (unsigned char *) SvPVX(zlopp); \
2319 unsigned const char *const kapow = uvchr_to_utf8(flrbbbbb, val); \
2320 SvCUR_set(zlopp, kapow - flrbbbbb); \
2323 av_push(revcharmap, zlopp); \
2325 char ooooff = (char)val; \
2326 av_push(revcharmap, newSVpvn(&ooooff, 1)); \
2330 /* This gets the next character from the input, folding it if not already
2332 #define TRIE_READ_CHAR STMT_START { \
2335 /* if it is UTF then it is either already folded, or does not need \
2337 uvc = valid_utf8_to_uvchr( (const U8*) uc, &len); \
2339 else if (folder == PL_fold_latin1) { \
2340 /* This folder implies Unicode rules, which in the range expressible \
2341 * by not UTF is the lower case, with the two exceptions, one of \
2342 * which should have been taken care of before calling this */ \
2343 assert(*uc != LATIN_SMALL_LETTER_SHARP_S); \
2344 uvc = toLOWER_L1(*uc); \
2345 if (UNLIKELY(uvc == MICRO_SIGN)) uvc = GREEK_SMALL_LETTER_MU; \
2348 /* raw data, will be folded later if needed */ \
2356 #define TRIE_LIST_PUSH(state,fid,ns) STMT_START { \
2357 if ( TRIE_LIST_CUR( state ) >=TRIE_LIST_LEN( state ) ) { \
2358 U32 ging = TRIE_LIST_LEN( state ) * 2; \
2359 Renew( trie->states[ state ].trans.list, ging, reg_trie_trans_le ); \
2360 TRIE_LIST_LEN( state ) = ging; \
2362 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).forid = fid; \
2363 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).newstate = ns; \
2364 TRIE_LIST_CUR( state )++; \
2367 #define TRIE_LIST_NEW(state) STMT_START { \
2368 Newxz( trie->states[ state ].trans.list, \
2369 4, reg_trie_trans_le ); \
2370 TRIE_LIST_CUR( state ) = 1; \
2371 TRIE_LIST_LEN( state ) = 4; \
2374 #define TRIE_HANDLE_WORD(state) STMT_START { \
2375 U16 dupe= trie->states[ state ].wordnum; \
2376 regnode * const noper_next = regnext( noper ); \
2379 /* store the word for dumping */ \
2381 if (OP(noper) != NOTHING) \
2382 tmp = newSVpvn_utf8(STRING(noper), STR_LEN(noper), UTF); \
2384 tmp = newSVpvn_utf8( "", 0, UTF ); \
2385 av_push( trie_words, tmp ); \
2389 trie->wordinfo[curword].prev = 0; \
2390 trie->wordinfo[curword].len = wordlen; \
2391 trie->wordinfo[curword].accept = state; \
2393 if ( noper_next < tail ) { \
2395 trie->jump = (U16 *) PerlMemShared_calloc( word_count + 1, \
2397 trie->jump[curword] = (U16)(noper_next - convert); \
2399 jumper = noper_next; \
2401 nextbranch= regnext(cur); \
2405 /* It's a dupe. Pre-insert into the wordinfo[].prev */\
2406 /* chain, so that when the bits of chain are later */\
2407 /* linked together, the dups appear in the chain */\
2408 trie->wordinfo[curword].prev = trie->wordinfo[dupe].prev; \
2409 trie->wordinfo[dupe].prev = curword; \
2411 /* we haven't inserted this word yet. */ \
2412 trie->states[ state ].wordnum = curword; \
2417 #define TRIE_TRANS_STATE(state,base,ucharcount,charid,special) \
2418 ( ( base + charid >= ucharcount \
2419 && base + charid < ubound \
2420 && state == trie->trans[ base - ucharcount + charid ].check \
2421 && trie->trans[ base - ucharcount + charid ].next ) \
2422 ? trie->trans[ base - ucharcount + charid ].next \
2423 : ( state==1 ? special : 0 ) \
2426 #define TRIE_BITMAP_SET_FOLDED(trie, uvc, folder) \
2428 TRIE_BITMAP_SET(trie, uvc); \
2429 /* store the folded codepoint */ \
2431 TRIE_BITMAP_SET(trie, folder[(U8) uvc ]); \
2434 /* store first byte of utf8 representation of */ \
2435 /* variant codepoints */ \
2436 if (! UVCHR_IS_INVARIANT(uvc)) { \
2437 TRIE_BITMAP_SET(trie, UTF8_TWO_BYTE_HI(uvc)); \
2442 #define MADE_JUMP_TRIE 2
2443 #define MADE_EXACT_TRIE 4
2446 S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch,
2447 regnode *first, regnode *last, regnode *tail,
2448 U32 word_count, U32 flags, U32 depth)
2450 /* first pass, loop through and scan words */
2451 reg_trie_data *trie;
2452 HV *widecharmap = NULL;
2453 AV *revcharmap = newAV();
2459 regnode *jumper = NULL;
2460 regnode *nextbranch = NULL;
2461 regnode *convert = NULL;
2462 U32 *prev_states; /* temp array mapping each state to previous one */
2463 /* we just use folder as a flag in utf8 */
2464 const U8 * folder = NULL;
2467 const U32 data_slot = add_data( pRExC_state, STR_WITH_LEN("tuuu"));
2468 AV *trie_words = NULL;
2469 /* along with revcharmap, this only used during construction but both are
2470 * useful during debugging so we store them in the struct when debugging.
2473 const U32 data_slot = add_data( pRExC_state, STR_WITH_LEN("tu"));
2474 STRLEN trie_charcount=0;
2476 SV *re_trie_maxbuff;
2477 GET_RE_DEBUG_FLAGS_DECL;
2479 PERL_ARGS_ASSERT_MAKE_TRIE;
2481 PERL_UNUSED_ARG(depth);
2485 case EXACT: case EXACTL: break;
2489 case EXACTFLU8: folder = PL_fold_latin1; break;
2490 case EXACTF: folder = PL_fold; break;
2491 default: Perl_croak( aTHX_ "panic! In trie construction, unknown node type %u %s", (unsigned) flags, PL_reg_name[flags] );
2494 trie = (reg_trie_data *) PerlMemShared_calloc( 1, sizeof(reg_trie_data) );
2496 trie->startstate = 1;
2497 trie->wordcount = word_count;
2498 RExC_rxi->data->data[ data_slot ] = (void*)trie;
2499 trie->charmap = (U16 *) PerlMemShared_calloc( 256, sizeof(U16) );
2500 if (flags == EXACT || flags == EXACTL)
2501 trie->bitmap = (char *) PerlMemShared_calloc( ANYOF_BITMAP_SIZE, 1 );
2502 trie->wordinfo = (reg_trie_wordinfo *) PerlMemShared_calloc(
2503 trie->wordcount+1, sizeof(reg_trie_wordinfo));
2506 trie_words = newAV();
2509 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
2510 assert(re_trie_maxbuff);
2511 if (!SvIOK(re_trie_maxbuff)) {
2512 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
2514 DEBUG_TRIE_COMPILE_r({
2515 Perl_re_indentf( aTHX_
2516 "make_trie start==%d, first==%d, last==%d, tail==%d depth=%d\n",
2518 REG_NODE_NUM(startbranch),REG_NODE_NUM(first),
2519 REG_NODE_NUM(last), REG_NODE_NUM(tail), (int)depth);
2522 /* Find the node we are going to overwrite */
2523 if ( first == startbranch && OP( last ) != BRANCH ) {
2524 /* whole branch chain */
2527 /* branch sub-chain */
2528 convert = NEXTOPER( first );
2531 /* -- First loop and Setup --
2533 We first traverse the branches and scan each word to determine if it
2534 contains widechars, and how many unique chars there are, this is
2535 important as we have to build a table with at least as many columns as we
2538 We use an array of integers to represent the character codes 0..255
2539 (trie->charmap) and we use a an HV* to store Unicode characters. We use
2540 the native representation of the character value as the key and IV's for
2543 *TODO* If we keep track of how many times each character is used we can
2544 remap the columns so that the table compression later on is more
2545 efficient in terms of memory by ensuring the most common value is in the
2546 middle and the least common are on the outside. IMO this would be better
2547 than a most to least common mapping as theres a decent chance the most
2548 common letter will share a node with the least common, meaning the node
2549 will not be compressible. With a middle is most common approach the worst
2550 case is when we have the least common nodes twice.
2554 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
2555 regnode *noper = NEXTOPER( cur );
2559 U32 wordlen = 0; /* required init */
2560 STRLEN minchars = 0;
2561 STRLEN maxchars = 0;
2562 bool set_bit = trie->bitmap ? 1 : 0; /*store the first char in the
2565 if (OP(noper) == NOTHING) {
2566 /* skip past a NOTHING at the start of an alternation
2567 * eg, /(?:)a|(?:b)/ should be the same as /a|b/
2569 regnode *noper_next= regnext(noper);
2570 if (noper_next < tail)
2574 if ( noper < tail &&
2576 OP(noper) == flags ||
2579 OP(noper) == EXACTFU_SS
2583 uc= (U8*)STRING(noper);
2584 e= uc + STR_LEN(noper);
2591 if ( set_bit ) { /* bitmap only alloced when !(UTF&&Folding) */
2592 TRIE_BITMAP_SET(trie,*uc); /* store the raw first byte
2593 regardless of encoding */
2594 if (OP( noper ) == EXACTFU_SS) {
2595 /* false positives are ok, so just set this */
2596 TRIE_BITMAP_SET(trie, LATIN_SMALL_LETTER_SHARP_S);
2600 for ( ; uc < e ; uc += len ) { /* Look at each char in the current
2602 TRIE_CHARCOUNT(trie)++;
2605 /* TRIE_READ_CHAR returns the current character, or its fold if /i
2606 * is in effect. Under /i, this character can match itself, or
2607 * anything that folds to it. If not under /i, it can match just
2608 * itself. Most folds are 1-1, for example k, K, and KELVIN SIGN
2609 * all fold to k, and all are single characters. But some folds
2610 * expand to more than one character, so for example LATIN SMALL
2611 * LIGATURE FFI folds to the three character sequence 'ffi'. If
2612 * the string beginning at 'uc' is 'ffi', it could be matched by
2613 * three characters, or just by the one ligature character. (It
2614 * could also be matched by two characters: LATIN SMALL LIGATURE FF
2615 * followed by 'i', or by 'f' followed by LATIN SMALL LIGATURE FI).
2616 * (Of course 'I' and/or 'F' instead of 'i' and 'f' can also
2617 * match.) The trie needs to know the minimum and maximum number
2618 * of characters that could match so that it can use size alone to
2619 * quickly reject many match attempts. The max is simple: it is
2620 * the number of folded characters in this branch (since a fold is
2621 * never shorter than what folds to it. */
2625 /* And the min is equal to the max if not under /i (indicated by
2626 * 'folder' being NULL), or there are no multi-character folds. If
2627 * there is a multi-character fold, the min is incremented just
2628 * once, for the character that folds to the sequence. Each
2629 * character in the sequence needs to be added to the list below of
2630 * characters in the trie, but we count only the first towards the
2631 * min number of characters needed. This is done through the
2632 * variable 'foldlen', which is returned by the macros that look
2633 * for these sequences as the number of bytes the sequence
2634 * occupies. Each time through the loop, we decrement 'foldlen' by
2635 * how many bytes the current char occupies. Only when it reaches
2636 * 0 do we increment 'minchars' or look for another multi-character
2638 if (folder == NULL) {
2641 else if (foldlen > 0) {
2642 foldlen -= (UTF) ? UTF8SKIP(uc) : 1;
2647 /* See if *uc is the beginning of a multi-character fold. If
2648 * so, we decrement the length remaining to look at, to account
2649 * for the current character this iteration. (We can use 'uc'
2650 * instead of the fold returned by TRIE_READ_CHAR because for
2651 * non-UTF, the latin1_safe macro is smart enough to account
2652 * for all the unfolded characters, and because for UTF, the
2653 * string will already have been folded earlier in the
2654 * compilation process */
2656 if ((foldlen = is_MULTI_CHAR_FOLD_utf8_safe(uc, e))) {
2657 foldlen -= UTF8SKIP(uc);
2660 else if ((foldlen = is_MULTI_CHAR_FOLD_latin1_safe(uc, e))) {
2665 /* The current character (and any potential folds) should be added
2666 * to the possible matching characters for this position in this
2670 U8 folded= folder[ (U8) uvc ];
2671 if ( !trie->charmap[ folded ] ) {
2672 trie->charmap[ folded ]=( ++trie->uniquecharcount );
2673 TRIE_STORE_REVCHAR( folded );
2676 if ( !trie->charmap[ uvc ] ) {
2677 trie->charmap[ uvc ]=( ++trie->uniquecharcount );
2678 TRIE_STORE_REVCHAR( uvc );
2681 /* store the codepoint in the bitmap, and its folded
2683 TRIE_BITMAP_SET_FOLDED(trie, uvc, folder);
2684 set_bit = 0; /* We've done our bit :-) */
2688 /* XXX We could come up with the list of code points that fold
2689 * to this using PL_utf8_foldclosures, except not for
2690 * multi-char folds, as there may be multiple combinations
2691 * there that could work, which needs to wait until runtime to
2692 * resolve (The comment about LIGATURE FFI above is such an
2697 widecharmap = newHV();
2699 svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 1 );
2702 Perl_croak( aTHX_ "error creating/fetching widecharmap entry for 0x%" UVXf, uvc );
2704 if ( !SvTRUE( *svpp ) ) {
2705 sv_setiv( *svpp, ++trie->uniquecharcount );
2706 TRIE_STORE_REVCHAR(uvc);
2709 } /* end loop through characters in this branch of the trie */
2711 /* We take the min and max for this branch and combine to find the min
2712 * and max for all branches processed so far */
2713 if( cur == first ) {
2714 trie->minlen = minchars;
2715 trie->maxlen = maxchars;
2716 } else if (minchars < trie->minlen) {
2717 trie->minlen = minchars;
2718 } else if (maxchars > trie->maxlen) {
2719 trie->maxlen = maxchars;
2721 } /* end first pass */
2722 DEBUG_TRIE_COMPILE_r(
2723 Perl_re_indentf( aTHX_
2724 "TRIE(%s): W:%d C:%d Uq:%d Min:%d Max:%d\n",
2726 ( widecharmap ? "UTF8" : "NATIVE" ), (int)word_count,
2727 (int)TRIE_CHARCOUNT(trie), trie->uniquecharcount,
2728 (int)trie->minlen, (int)trie->maxlen )
2732 We now know what we are dealing with in terms of unique chars and
2733 string sizes so we can calculate how much memory a naive
2734 representation using a flat table will take. If it's over a reasonable
2735 limit (as specified by ${^RE_TRIE_MAXBUF}) we use a more memory
2736 conservative but potentially much slower representation using an array
2739 At the end we convert both representations into the same compressed
2740 form that will be used in regexec.c for matching with. The latter
2741 is a form that cannot be used to construct with but has memory
2742 properties similar to the list form and access properties similar
2743 to the table form making it both suitable for fast searches and
2744 small enough that its feasable to store for the duration of a program.
2746 See the comment in the code where the compressed table is produced
2747 inplace from the flat tabe representation for an explanation of how
2748 the compression works.
2753 Newx(prev_states, TRIE_CHARCOUNT(trie) + 2, U32);
2756 if ( (IV)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1)
2757 > SvIV(re_trie_maxbuff) )
2760 Second Pass -- Array Of Lists Representation
2762 Each state will be represented by a list of charid:state records
2763 (reg_trie_trans_le) the first such element holds the CUR and LEN
2764 points of the allocated array. (See defines above).
2766 We build the initial structure using the lists, and then convert
2767 it into the compressed table form which allows faster lookups
2768 (but cant be modified once converted).
2771 STRLEN transcount = 1;
2773 DEBUG_TRIE_COMPILE_MORE_r( Perl_re_indentf( aTHX_ "Compiling trie using list compiler\n",
2776 trie->states = (reg_trie_state *)
2777 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
2778 sizeof(reg_trie_state) );
2782 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
2784 regnode *noper = NEXTOPER( cur );
2785 U32 state = 1; /* required init */
2786 U16 charid = 0; /* sanity init */
2787 U32 wordlen = 0; /* required init */
2789 if (OP(noper) == NOTHING) {
2790 regnode *noper_next= regnext(noper);
2791 if (noper_next < tail)
2795 if ( noper < tail && ( OP(noper) == flags || ( flags == EXACTFU && OP(noper) == EXACTFU_SS ) ) ) {
2796 const U8 *uc= (U8*)STRING(noper);
2797 const U8 *e= uc + STR_LEN(noper);
2799 for ( ; uc < e ; uc += len ) {
2804 charid = trie->charmap[ uvc ];
2806 SV** const svpp = hv_fetch( widecharmap,
2813 charid=(U16)SvIV( *svpp );
2816 /* charid is now 0 if we dont know the char read, or
2817 * nonzero if we do */
2824 if ( !trie->states[ state ].trans.list ) {
2825 TRIE_LIST_NEW( state );
2828 check <= TRIE_LIST_USED( state );
2831 if ( TRIE_LIST_ITEM( state, check ).forid
2834 newstate = TRIE_LIST_ITEM( state, check ).newstate;
2839 newstate = next_alloc++;
2840 prev_states[newstate] = state;
2841 TRIE_LIST_PUSH( state, charid, newstate );
2846 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %" IVdf, uvc );
2850 TRIE_HANDLE_WORD(state);
2852 } /* end second pass */
2854 /* next alloc is the NEXT state to be allocated */
2855 trie->statecount = next_alloc;
2856 trie->states = (reg_trie_state *)
2857 PerlMemShared_realloc( trie->states,
2859 * sizeof(reg_trie_state) );
2861 /* and now dump it out before we compress it */
2862 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_list(trie, widecharmap,
2863 revcharmap, next_alloc,
2867 trie->trans = (reg_trie_trans *)
2868 PerlMemShared_calloc( transcount, sizeof(reg_trie_trans) );
2875 for( state=1 ; state < next_alloc ; state ++ ) {
2879 DEBUG_TRIE_COMPILE_MORE_r(
2880 Perl_re_printf( aTHX_ "tp: %d zp: %d ",tp,zp)
2884 if (trie->states[state].trans.list) {
2885 U16 minid=TRIE_LIST_ITEM( state, 1).forid;
2889 for( idx = 2 ; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
2890 const U16 forid = TRIE_LIST_ITEM( state, idx).forid;
2891 if ( forid < minid ) {
2893 } else if ( forid > maxid ) {
2897 if ( transcount < tp + maxid - minid + 1) {
2899 trie->trans = (reg_trie_trans *)
2900 PerlMemShared_realloc( trie->trans,
2902 * sizeof(reg_trie_trans) );
2903 Zero( trie->trans + (transcount / 2),
2907 base = trie->uniquecharcount + tp - minid;
2908 if ( maxid == minid ) {
2910 for ( ; zp < tp ; zp++ ) {
2911 if ( ! trie->trans[ zp ].next ) {
2912 base = trie->uniquecharcount + zp - minid;
2913 trie->trans[ zp ].next = TRIE_LIST_ITEM( state,
2915 trie->trans[ zp ].check = state;
2921 trie->trans[ tp ].next = TRIE_LIST_ITEM( state,
2923 trie->trans[ tp ].check = state;
2928 for ( idx=1; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
2929 const U32 tid = base
2930 - trie->uniquecharcount
2931 + TRIE_LIST_ITEM( state, idx ).forid;
2932 trie->trans[ tid ].next = TRIE_LIST_ITEM( state,
2934 trie->trans[ tid ].check = state;
2936 tp += ( maxid - minid + 1 );
2938 Safefree(trie->states[ state ].trans.list);
2941 DEBUG_TRIE_COMPILE_MORE_r(
2942 Perl_re_printf( aTHX_ " base: %d\n",base);
2945 trie->states[ state ].trans.base=base;
2947 trie->lasttrans = tp + 1;
2951 Second Pass -- Flat Table Representation.
2953 we dont use the 0 slot of either trans[] or states[] so we add 1 to
2954 each. We know that we will need Charcount+1 trans at most to store
2955 the data (one row per char at worst case) So we preallocate both
2956 structures assuming worst case.
2958 We then construct the trie using only the .next slots of the entry
2961 We use the .check field of the first entry of the node temporarily
2962 to make compression both faster and easier by keeping track of how
2963 many non zero fields are in the node.
2965 Since trans are numbered from 1 any 0 pointer in the table is a FAIL
2968 There are two terms at use here: state as a TRIE_NODEIDX() which is
2969 a number representing the first entry of the node, and state as a
2970 TRIE_NODENUM() which is the trans number. state 1 is TRIE_NODEIDX(1)
2971 and TRIE_NODENUM(1), state 2 is TRIE_NODEIDX(2) and TRIE_NODENUM(3)
2972 if there are 2 entrys per node. eg:
2980 The table is internally in the right hand, idx form. However as we
2981 also have to deal with the states array which is indexed by nodenum
2982 we have to use TRIE_NODENUM() to convert.
2985 DEBUG_TRIE_COMPILE_MORE_r( Perl_re_indentf( aTHX_ "Compiling trie using table compiler\n",
2988 trie->trans = (reg_trie_trans *)
2989 PerlMemShared_calloc( ( TRIE_CHARCOUNT(trie) + 1 )
2990 * trie->uniquecharcount + 1,
2991 sizeof(reg_trie_trans) );
2992 trie->states = (reg_trie_state *)
2993 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
2994 sizeof(reg_trie_state) );
2995 next_alloc = trie->uniquecharcount + 1;
2998 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
3000 regnode *noper = NEXTOPER( cur );
3002 U32 state = 1; /* required init */
3004 U16 charid = 0; /* sanity init */
3005 U32 accept_state = 0; /* sanity init */
3007 U32 wordlen = 0; /* required init */
3009 if (OP(noper) == NOTHING) {
3010 regnode *noper_next= regnext(noper);
3011 if (noper_next < tail)
3015 if ( noper < tail && ( OP(noper) == flags || ( flags == EXACTFU && OP(noper) == EXACTFU_SS ) ) ) {
3016 const U8 *uc= (U8*)STRING(noper);
3017 const U8 *e= uc + STR_LEN(noper);
3019 for ( ; uc < e ; uc += len ) {
3024 charid = trie->charmap[ uvc ];
3026 SV* const * const svpp = hv_fetch( widecharmap,
3030 charid = svpp ? (U16)SvIV(*svpp) : 0;
3034 if ( !trie->trans[ state + charid ].next ) {
3035 trie->trans[ state + charid ].next = next_alloc;
3036 trie->trans[ state ].check++;
3037 prev_states[TRIE_NODENUM(next_alloc)]
3038 = TRIE_NODENUM(state);
3039 next_alloc += trie->uniquecharcount;
3041 state = trie->trans[ state + charid ].next;
3043 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %" IVdf, uvc );
3045 /* charid is now 0 if we dont know the char read, or
3046 * nonzero if we do */
3049 accept_state = TRIE_NODENUM( state );
3050 TRIE_HANDLE_WORD(accept_state);
3052 } /* end second pass */
3054 /* and now dump it out before we compress it */
3055 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_table(trie, widecharmap,
3057 next_alloc, depth+1));
3061 * Inplace compress the table.*
3063 For sparse data sets the table constructed by the trie algorithm will
3064 be mostly 0/FAIL transitions or to put it another way mostly empty.
3065 (Note that leaf nodes will not contain any transitions.)
3067 This algorithm compresses the tables by eliminating most such
3068 transitions, at the cost of a modest bit of extra work during lookup:
3070 - Each states[] entry contains a .base field which indicates the
3071 index in the state[] array wheres its transition data is stored.
3073 - If .base is 0 there are no valid transitions from that node.
3075 - If .base is nonzero then charid is added to it to find an entry in
3078 -If trans[states[state].base+charid].check!=state then the
3079 transition is taken to be a 0/Fail transition. Thus if there are fail
3080 transitions at the front of the node then the .base offset will point
3081 somewhere inside the previous nodes data (or maybe even into a node
3082 even earlier), but the .check field determines if the transition is
3086 The following process inplace converts the table to the compressed
3087 table: We first do not compress the root node 1,and mark all its
3088 .check pointers as 1 and set its .base pointer as 1 as well. This
3089 allows us to do a DFA construction from the compressed table later,
3090 and ensures that any .base pointers we calculate later are greater
3093 - We set 'pos' to indicate the first entry of the second node.
3095 - We then iterate over the columns of the node, finding the first and
3096 last used entry at l and m. We then copy l..m into pos..(pos+m-l),
3097 and set the .check pointers accordingly, and advance pos
3098 appropriately and repreat for the next node. Note that when we copy
3099 the next pointers we have to convert them from the original
3100 NODEIDX form to NODENUM form as the former is not valid post
3103 - If a node has no transitions used we mark its base as 0 and do not
3104 advance the pos pointer.
3106 - If a node only has one transition we use a second pointer into the
3107 structure to fill in allocated fail transitions from other states.
3108 This pointer is independent of the main pointer and scans forward
3109 looking for null transitions that are allocated to a state. When it
3110 finds one it writes the single transition into the "hole". If the
3111 pointer doesnt find one the single transition is appended as normal.
3113 - Once compressed we can Renew/realloc the structures to release the
3116 See "Table-Compression Methods" in sec 3.9 of the Red Dragon,
3117 specifically Fig 3.47 and the associated pseudocode.
3121 const U32 laststate = TRIE_NODENUM( next_alloc );
3124 trie->statecount = laststate;
3126 for ( state = 1 ; state < laststate ; state++ ) {
3128 const U32 stateidx = TRIE_NODEIDX( state );
3129 const U32 o_used = trie->trans[ stateidx ].check;
3130 U32 used = trie->trans[ stateidx ].check;
3131 trie->trans[ stateidx ].check = 0;
3134 used && charid < trie->uniquecharcount;
3137 if ( flag || trie->trans[ stateidx + charid ].next ) {
3138 if ( trie->trans[ stateidx + charid ].next ) {
3140 for ( ; zp < pos ; zp++ ) {
3141 if ( ! trie->trans[ zp ].next ) {
3145 trie->states[ state ].trans.base
3147 + trie->uniquecharcount
3149 trie->trans[ zp ].next
3150 = SAFE_TRIE_NODENUM( trie->trans[ stateidx
3152 trie->trans[ zp ].check = state;
3153 if ( ++zp > pos ) pos = zp;
3160 trie->states[ state ].trans.base
3161 = pos + trie->uniquecharcount - charid ;
3163 trie->trans[ pos ].next
3164 = SAFE_TRIE_NODENUM(
3165 trie->trans[ stateidx + charid ].next );
3166 trie->trans[ pos ].check = state;
3171 trie->lasttrans = pos + 1;
3172 trie->states = (reg_trie_state *)
3173 PerlMemShared_realloc( trie->states, laststate
3174 * sizeof(reg_trie_state) );
3175 DEBUG_TRIE_COMPILE_MORE_r(
3176 Perl_re_indentf( aTHX_ "Alloc: %d Orig: %" IVdf " elements, Final:%" IVdf ". Savings of %%%5.2f\n",
3178 (int)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount
3182 ( ( next_alloc - pos ) * 100 ) / (double)next_alloc );
3185 } /* end table compress */
3187 DEBUG_TRIE_COMPILE_MORE_r(
3188 Perl_re_indentf( aTHX_ "Statecount:%" UVxf " Lasttrans:%" UVxf "\n",
3190 (UV)trie->statecount,
3191 (UV)trie->lasttrans)
3193 /* resize the trans array to remove unused space */
3194 trie->trans = (reg_trie_trans *)
3195 PerlMemShared_realloc( trie->trans, trie->lasttrans
3196 * sizeof(reg_trie_trans) );
3198 { /* Modify the program and insert the new TRIE node */
3199 U8 nodetype =(U8)(flags & 0xFF);
3203 regnode *optimize = NULL;
3204 #ifdef RE_TRACK_PATTERN_OFFSETS
3207 U32 mjd_nodelen = 0;
3208 #endif /* RE_TRACK_PATTERN_OFFSETS */
3209 #endif /* DEBUGGING */
3211 This means we convert either the first branch or the first Exact,
3212 depending on whether the thing following (in 'last') is a branch
3213 or not and whther first is the startbranch (ie is it a sub part of
3214 the alternation or is it the whole thing.)
3215 Assuming its a sub part we convert the EXACT otherwise we convert
3216 the whole branch sequence, including the first.
3218 /* Find the node we are going to overwrite */
3219 if ( first != startbranch || OP( last ) == BRANCH ) {
3220 /* branch sub-chain */
3221 NEXT_OFF( first ) = (U16)(last - first);
3222 #ifdef RE_TRACK_PATTERN_OFFSETS
3224 mjd_offset= Node_Offset((convert));
3225 mjd_nodelen= Node_Length((convert));
3228 /* whole branch chain */
3230 #ifdef RE_TRACK_PATTERN_OFFSETS
3233 const regnode *nop = NEXTOPER( convert );
3234 mjd_offset= Node_Offset((nop));
3235 mjd_nodelen= Node_Length((nop));
3239 Perl_re_indentf( aTHX_ "MJD offset:%" UVuf " MJD length:%" UVuf "\n",
3241 (UV)mjd_offset, (UV)mjd_nodelen)
3244 /* But first we check to see if there is a common prefix we can
3245 split out as an EXACT and put in front of the TRIE node. */
3246 trie->startstate= 1;
3247 if ( trie->bitmap && !widecharmap && !trie->jump ) {
3248 /* we want to find the first state that has more than
3249 * one transition, if that state is not the first state
3250 * then we have a common prefix which we can remove.
3253 for ( state = 1 ; state < trie->statecount-1 ; state++ ) {
3255 I32 first_ofs = -1; /* keeps track of the ofs of the first
3256 transition, -1 means none */
3258 const U32 base = trie->states[ state ].trans.base;
3260 /* does this state terminate an alternation? */
3261 if ( trie->states[state].wordnum )
3264 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
3265 if ( ( base + ofs >= trie->uniquecharcount ) &&
3266 ( base + ofs - trie->uniquecharcount < trie->lasttrans ) &&
3267 trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
3269 if ( ++count > 1 ) {
3270 /* we have more than one transition */
3273 /* if this is the first state there is no common prefix
3274 * to extract, so we can exit */
3275 if ( state == 1 ) break;
3276 tmp = av_fetch( revcharmap, ofs, 0);
3277 ch = (U8*)SvPV_nolen_const( *tmp );
3279 /* if we are on count 2 then we need to initialize the
3280 * bitmap, and store the previous char if there was one
3283 /* clear the bitmap */
3284 Zero(trie->bitmap, ANYOF_BITMAP_SIZE, char);
3286 Perl_re_indentf( aTHX_ "New Start State=%" UVuf " Class: [",
3289 if (first_ofs >= 0) {
3290 SV ** const tmp = av_fetch( revcharmap, first_ofs, 0);
3291 const U8 * const ch = (U8*)SvPV_nolen_const( *tmp );
3293 TRIE_BITMAP_SET_FOLDED(trie,*ch,folder);
3295 Perl_re_printf( aTHX_ "%s", (char*)ch)
3299 /* store the current firstchar in the bitmap */
3300 TRIE_BITMAP_SET_FOLDED(trie,*ch,folder);
3301 DEBUG_OPTIMISE_r(Perl_re_printf( aTHX_ "%s", ch));
3307 /* This state has only one transition, its transition is part
3308 * of a common prefix - we need to concatenate the char it
3309 * represents to what we have so far. */
3310 SV **tmp = av_fetch( revcharmap, first_ofs, 0);
3312 char *ch = SvPV( *tmp, len );
3314 SV *sv=sv_newmortal();
3315 Perl_re_indentf( aTHX_ "Prefix State: %" UVuf " Ofs:%" UVuf " Char='%s'\n",
3317 (UV)state, (UV)first_ofs,
3318 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 6,
3319 PL_colors[0], PL_colors[1],
3320 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
3321 PERL_PV_ESCAPE_FIRSTCHAR
3326 OP( convert ) = nodetype;
3327 str=STRING(convert);
3330 STR_LEN(convert) += len;
3336 DEBUG_OPTIMISE_r(Perl_re_printf( aTHX_ "]\n"));
3341 trie->prefixlen = (state-1);
3343 regnode *n = convert+NODE_SZ_STR(convert);
3344 NEXT_OFF(convert) = NODE_SZ_STR(convert);
3345 trie->startstate = state;
3346 trie->minlen -= (state - 1);
3347 trie->maxlen -= (state - 1);
3349 /* At least the UNICOS C compiler choked on this
3350 * being argument to DEBUG_r(), so let's just have
3353 #ifdef PERL_EXT_RE_BUILD
3359 regnode *fix = convert;
3360 U32 word = trie->wordcount;
3362 Set_Node_Offset_Length(convert, mjd_offset, state - 1);
3363 while( ++fix < n ) {
3364 Set_Node_Offset_Length(fix, 0, 0);
3367 SV ** const tmp = av_fetch( trie_words, word, 0 );
3369 if ( STR_LEN(convert) <= SvCUR(*tmp) )
3370 sv_chop(*tmp, SvPV_nolen(*tmp) + STR_LEN(convert));
3372 sv_chop(*tmp, SvPV_nolen(*tmp) + SvCUR(*tmp));
3380 NEXT_OFF(convert) = (U16)(tail - convert);
3381 DEBUG_r(optimize= n);
3387 if ( trie->maxlen ) {
3388 NEXT_OFF( convert ) = (U16)(tail - convert);
3389 ARG_SET( convert, data_slot );
3390 /* Store the offset to the first unabsorbed branch in
3391 jump[0], which is otherwise unused by the jump logic.
3392 We use this when dumping a trie and during optimisation. */
3394 trie->jump[0] = (U16)(nextbranch - convert);
3396 /* If the start state is not accepting (meaning there is no empty string/NOTHING)
3397 * and there is a bitmap
3398 * and the first "jump target" node we found leaves enough room
3399 * then convert the TRIE node into a TRIEC node, with the bitmap
3400 * embedded inline in the opcode - this is hypothetically faster.
3402 if ( !trie->states[trie->startstate].wordnum
3404 && ( (char *)jumper - (char *)convert) >= (int)sizeof(struct regnode_charclass) )
3406 OP( convert ) = TRIEC;
3407 Copy(trie->bitmap, ((struct regnode_charclass *)convert)->bitmap, ANYOF_BITMAP_SIZE, char);
3408 PerlMemShared_free(trie->bitmap);
3411 OP( convert ) = TRIE;
3413 /* store the type in the flags */
3414 convert->flags = nodetype;
3418 + regarglen[ OP( convert ) ];
3420 /* XXX We really should free up the resource in trie now,
3421 as we won't use them - (which resources?) dmq */
3423 /* needed for dumping*/
3424 DEBUG_r(if (optimize) {
3425 regnode *opt = convert;
3427 while ( ++opt < optimize) {
3428 Set_Node_Offset_Length(opt,0,0);
3431 Try to clean up some of the debris left after the
3434 while( optimize < jumper ) {
3435 mjd_nodelen += Node_Length((optimize));
3436 OP( optimize ) = OPTIMIZED;
3437 Set_Node_Offset_Length(optimize,0,0);
3440 Set_Node_Offset_Length(convert,mjd_offset,mjd_nodelen);
3442 } /* end node insert */
3444 /* Finish populating the prev field of the wordinfo array. Walk back
3445 * from each accept state until we find another accept state, and if
3446 * so, point the first word's .prev field at the second word. If the
3447 * second already has a .prev field set, stop now. This will be the
3448 * case either if we've already processed that word's accept state,
3449 * or that state had multiple words, and the overspill words were
3450 * already linked up earlier.
3457 for (word=1; word <= trie->wordcount; word++) {
3459 if (trie->wordinfo[word].prev)
3461 state = trie->wordinfo[word].accept;
3463 state = prev_states[state];
3466 prev = trie->states[state].wordnum;
3470 trie->wordinfo[word].prev = prev;
3472 Safefree(prev_states);
3476 /* and now dump out the compressed format */
3477 DEBUG_TRIE_COMPILE_r(dump_trie(trie, widecharmap, revcharmap, depth+1));
3479 RExC_rxi->data->data[ data_slot + 1 ] = (void*)widecharmap;
3481 RExC_rxi->data->data[ data_slot + TRIE_WORDS_OFFSET ] = (void*)trie_words;
3482 RExC_rxi->data->data[ data_slot + 3 ] = (void*)revcharmap;
3484 SvREFCNT_dec_NN(revcharmap);
3488 : trie->startstate>1
3494 S_construct_ahocorasick_from_trie(pTHX_ RExC_state_t *pRExC_state, regnode *source, U32 depth)
3496 /* The Trie is constructed and compressed now so we can build a fail array if
3499 This is basically the Aho-Corasick algorithm. Its from exercise 3.31 and
3501 "Red Dragon" -- Compilers, principles, techniques, and tools. Aho, Sethi,
3505 We find the fail state for each state in the trie, this state is the longest
3506 proper suffix of the current state's 'word' that is also a proper prefix of
3507 another word in our trie. State 1 represents the word '' and is thus the
3508 default fail state. This allows the DFA not to have to restart after its
3509 tried and failed a word at a given point, it simply continues as though it
3510 had been matching the other word in the first place.
3512 'abcdgu'=~/abcdefg|cdgu/
3513 When we get to 'd' we are still matching the first word, we would encounter
3514 'g' which would fail, which would bring us to the state representing 'd' in
3515 the second word where we would try 'g' and succeed, proceeding to match
3518 /* add a fail transition */
3519 const U32 trie_offset = ARG(source);
3520 reg_trie_data *trie=(reg_trie_data *)RExC_rxi->data->data[trie_offset];
3522 const U32 ucharcount = trie->uniquecharcount;
3523 const U32 numstates = trie->statecount;
3524 const U32 ubound = trie->lasttrans + ucharcount;
3528 U32 base = trie->states[ 1 ].trans.base;
3531 const U32 data_slot = add_data( pRExC_state, STR_WITH_LEN("T"));
3533 GET_RE_DEBUG_FLAGS_DECL;
3535 PERL_ARGS_ASSERT_CONSTRUCT_AHOCORASICK_FROM_TRIE;
3536 PERL_UNUSED_CONTEXT;
3538 PERL_UNUSED_ARG(depth);
3541 if ( OP(source) == TRIE ) {
3542 struct regnode_1 *op = (struct regnode_1 *)
3543 PerlMemShared_calloc(1, sizeof(struct regnode_1));
3544 StructCopy(source,op,struct regnode_1);
3545 stclass = (regnode *)op;
3547 struct regnode_charclass *op = (struct regnode_charclass *)
3548 PerlMemShared_calloc(1, sizeof(struct regnode_charclass));
3549 StructCopy(source,op,struct regnode_charclass);
3550 stclass = (regnode *)op;
3552 OP(stclass)+=2; /* convert the TRIE type to its AHO-CORASICK equivalent */
3554 ARG_SET( stclass, data_slot );
3555 aho = (reg_ac_data *) PerlMemShared_calloc( 1, sizeof(reg_ac_data) );
3556 RExC_rxi->data->data[ data_slot ] = (void*)aho;
3557 aho->trie=trie_offset;
3558 aho->states=(reg_trie_state *)PerlMemShared_malloc( numstates * sizeof(reg_trie_state) );
3559 Copy( trie->states, aho->states, numstates, reg_trie_state );
3560 Newxz( q, numstates, U32);
3561 aho->fail = (U32 *) PerlMemShared_calloc( numstates, sizeof(U32) );
3564 /* initialize fail[0..1] to be 1 so that we always have
3565 a valid final fail state */
3566 fail[ 0 ] = fail[ 1 ] = 1;
3568 for ( charid = 0; charid < ucharcount ; charid++ ) {
3569 const U32 newstate = TRIE_TRANS_STATE( 1, base, ucharcount, charid, 0 );
3571 q[ q_write ] = newstate;
3572 /* set to point at the root */
3573 fail[ q[ q_write++ ] ]=1;
3576 while ( q_read < q_write) {
3577 const U32 cur = q[ q_read++ % numstates ];
3578 base = trie->states[ cur ].trans.base;
3580 for ( charid = 0 ; charid < ucharcount ; charid++ ) {
3581 const U32 ch_state = TRIE_TRANS_STATE( cur, base, ucharcount, charid, 1 );
3583 U32 fail_state = cur;
3586 fail_state = fail[ fail_state ];
3587 fail_base = aho->states[ fail_state ].trans.base;
3588 } while ( !TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 ) );
3590 fail_state = TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 );
3591 fail[ ch_state ] = fail_state;
3592 if ( !aho->states[ ch_state ].wordnum && aho->states[ fail_state ].wordnum )
3594 aho->states[ ch_state ].wordnum = aho->states[ fail_state ].wordnum;
3596 q[ q_write++ % numstates] = ch_state;
3600 /* restore fail[0..1] to 0 so that we "fall out" of the AC loop
3601 when we fail in state 1, this allows us to use the
3602 charclass scan to find a valid start char. This is based on the principle
3603 that theres a good chance the string being searched contains lots of stuff
3604 that cant be a start char.
3606 fail[ 0 ] = fail[ 1 ] = 0;
3607 DEBUG_TRIE_COMPILE_r({
3608 Perl_re_indentf( aTHX_ "Stclass Failtable (%" UVuf " states): 0",
3609 depth, (UV)numstates
3611 for( q_read=1; q_read<numstates; q_read++ ) {
3612 Perl_re_printf( aTHX_ ", %" UVuf, (UV)fail[q_read]);
3614 Perl_re_printf( aTHX_ "\n");
3617 /*RExC_seen |= REG_TRIEDFA_SEEN;*/
3622 #define DEBUG_PEEP(str,scan,depth) \
3623 DEBUG_OPTIMISE_r({if (scan){ \
3624 regnode *Next = regnext(scan); \
3625 regprop(RExC_rx, RExC_mysv, scan, NULL, pRExC_state);\
3626 Perl_re_indentf( aTHX_ "" str ">%3d: %s (%d)", \
3627 depth, REG_NODE_NUM(scan), SvPV_nolen_const(RExC_mysv),\
3628 Next ? (REG_NODE_NUM(Next)) : 0 );\
3629 DEBUG_SHOW_STUDY_FLAGS(flags," [ ","]");\
3630 Perl_re_printf( aTHX_ "\n"); \
3633 /* The below joins as many adjacent EXACTish nodes as possible into a single
3634 * one. The regop may be changed if the node(s) contain certain sequences that
3635 * require special handling. The joining is only done if:
3636 * 1) there is room in the current conglomerated node to entirely contain the
3638 * 2) they are the exact same node type
3640 * The adjacent nodes actually may be separated by NOTHING-kind nodes, and
3641 * these get optimized out
3643 * XXX khw thinks this should be enhanced to fill EXACT (at least) nodes as full
3644 * as possible, even if that means splitting an existing node so that its first
3645 * part is moved to the preceeding node. This would maximise the efficiency of
3646 * memEQ during matching. Elsewhere in this file, khw proposes splitting
3647 * EXACTFish nodes into portions that don't change under folding vs those that
3648 * do. Those portions that don't change may be the only things in the pattern that
3649 * could be used to find fixed and floating strings.
3651 * If a node is to match under /i (folded), the number of characters it matches
3652 * can be different than its character length if it contains a multi-character
3653 * fold. *min_subtract is set to the total delta number of characters of the
3656 * And *unfolded_multi_char is set to indicate whether or not the node contains
3657 * an unfolded multi-char fold. This happens when whether the fold is valid or
3658 * not won't be known until runtime; namely for EXACTF nodes that contain LATIN
3659 * SMALL LETTER SHARP S, as only if the target string being matched against
3660 * turns out to be UTF-8 is that fold valid; and also for EXACTFL nodes whose
3661 * folding rules depend on the locale in force at runtime. (Multi-char folds
3662 * whose components are all above the Latin1 range are not run-time locale
3663 * dependent, and have already been folded by the time this function is
3666 * This is as good a place as any to discuss the design of handling these
3667 * multi-character fold sequences. It's been wrong in Perl for a very long
3668 * time. There are three code points in Unicode whose multi-character folds
3669 * were long ago discovered to mess things up. The previous designs for
3670 * dealing with these involved assigning a special node for them. This
3671 * approach doesn't always work, as evidenced by this example:
3672 * "\xDFs" =~ /s\xDF/ui # Used to fail before these patches
3673 * Both sides fold to "sss", but if the pattern is parsed to create a node that
3674 * would match just the \xDF, it won't be able to handle the case where a
3675 * successful match would have to cross the node's boundary. The new approach
3676 * that hopefully generally solves the problem generates an EXACTFU_SS node
3677 * that is "sss" in this case.
3679 * It turns out that there are problems with all multi-character folds, and not
3680 * just these three. Now the code is general, for all such cases. The
3681 * approach taken is:
3682 * 1) This routine examines each EXACTFish node that could contain multi-
3683 * character folded sequences. Since a single character can fold into
3684 * such a sequence, the minimum match length for this node is less than
3685 * the number of characters in the node. This routine returns in
3686 * *min_subtract how many characters to subtract from the the actual
3687 * length of the string to get a real minimum match length; it is 0 if
3688 * there are no multi-char foldeds. This delta is used by the caller to
3689 * adjust the min length of the match, and the delta between min and max,
3690 * so that the optimizer doesn't reject these possibilities based on size
3692 * 2) For the sequence involving the Sharp s (\xDF), the node type EXACTFU_SS
3693 * is used for an EXACTFU node that contains at least one "ss" sequence in
3694 * it. For non-UTF-8 patterns and strings, this is the only case where
3695 * there is a possible fold length change. That means that a regular
3696 * EXACTFU node without UTF-8 involvement doesn't have to concern itself
3697 * with length changes, and so can be processed faster. regexec.c takes
3698 * advantage of this. Generally, an EXACTFish node that is in UTF-8 is
3699 * pre-folded by regcomp.c (except EXACTFL, some of whose folds aren't
3700 * known until runtime). This saves effort in regex matching. However,
3701 * the pre-folding isn't done for non-UTF8 patterns because the fold of
3702 * the MICRO SIGN requires UTF-8, and we don't want to slow things down by
3703 * forcing the pattern into UTF8 unless necessary. Also what EXACTF (and,
3704 * again, EXACTFL) nodes fold to isn't known until runtime. The fold
3705 * possibilities for the non-UTF8 patterns are quite simple, except for
3706 * the sharp s. All the ones that don't involve a UTF-8 target string are
3707 * members of a fold-pair, and arrays are set up for all of them so that
3708 * the other member of the pair can be found quickly. Code elsewhere in
3709 * this file makes sure that in EXACTFU nodes, the sharp s gets folded to
3710 * 'ss', even if the pattern isn't UTF-8. This avoids the issues
3711 * described in the next item.
3712 * 3) A problem remains for unfolded multi-char folds. (These occur when the
3713 * validity of the fold won't be known until runtime, and so must remain
3714 * unfolded for now. This happens for the sharp s in EXACTF and EXACTFA
3715 * nodes when the pattern isn't in UTF-8. (Note, BTW, that there cannot
3716 * be an EXACTF node with a UTF-8 pattern.) They also occur for various
3717 * folds in EXACTFL nodes, regardless of the UTF-ness of the pattern.)
3718 * The reason this is a problem is that the optimizer part of regexec.c
3719 * (probably unwittingly, in Perl_regexec_flags()) makes an assumption
3720 * that a character in the pattern corresponds to at most a single
3721 * character in the target string. (And I do mean character, and not byte
3722 * here, unlike other parts of the documentation that have never been
3723 * updated to account for multibyte Unicode.) sharp s in EXACTF and
3724 * EXACTFL nodes can match the two character string 'ss'; in EXACTFA nodes
3725 * it can match "\x{17F}\x{17F}". These, along with other ones in EXACTFL
3726 * nodes, violate the assumption, and they are the only instances where it
3727 * is violated. I'm reluctant to try to change the assumption, as the
3728 * code involved is impenetrable to me (khw), so instead the code here
3729 * punts. This routine examines EXACTFL nodes, and (when the pattern
3730 * isn't UTF-8) EXACTF and EXACTFA for such unfolded folds, and returns a
3731 * boolean indicating whether or not the node contains such a fold. When
3732 * it is true, the caller sets a flag that later causes the optimizer in
3733 * this file to not set values for the floating and fixed string lengths,
3734 * and thus avoids the optimizer code in regexec.c that makes the invalid
3735 * assumption. Thus, there is no optimization based on string lengths for
3736 * EXACTFL nodes that contain these few folds, nor for non-UTF8-pattern
3737 * EXACTF and EXACTFA nodes that contain the sharp s. (The reason the
3738 * assumption is wrong only in these cases is that all other non-UTF-8
3739 * folds are 1-1; and, for UTF-8 patterns, we pre-fold all other folds to
3740 * their expanded versions. (Again, we can't prefold sharp s to 'ss' in
3741 * EXACTF nodes because we don't know at compile time if it actually
3742 * matches 'ss' or not. For EXACTF nodes it will match iff the target
3743 * string is in UTF-8. This is in contrast to EXACTFU nodes, where it
3744 * always matches; and EXACTFA where it never does. In an EXACTFA node in
3745 * a UTF-8 pattern, sharp s is folded to "\x{17F}\x{17F}, avoiding the
3746 * problem; but in a non-UTF8 pattern, folding it to that above-Latin1
3747 * string would require the pattern to be forced into UTF-8, the overhead
3748 * of which we want to avoid. Similarly the unfolded multi-char folds in
3749 * EXACTFL nodes will match iff the locale at the time of match is a UTF-8
3752 * Similarly, the code that generates tries doesn't currently handle
3753 * not-already-folded multi-char folds, and it looks like a pain to change
3754 * that. Therefore, trie generation of EXACTFA nodes with the sharp s
3755 * doesn't work. Instead, such an EXACTFA is turned into a new regnode,
3756 * EXACTFA_NO_TRIE, which the trie code knows not to handle. Most people
3757 * using /iaa matching will be doing so almost entirely with ASCII
3758 * strings, so this should rarely be encountered in practice */
3760 #define JOIN_EXACT(scan,min_subtract,unfolded_multi_char, flags) \
3761 if (PL_regkind[OP(scan)] == EXACT) \
3762 join_exact(pRExC_state,(scan),(min_subtract),unfolded_multi_char, (flags),NULL,depth+1)
3765 S_join_exact(pTHX_ RExC_state_t *pRExC_state, regnode *scan,
3766 UV *min_subtract, bool *unfolded_multi_char,
3767 U32 flags,regnode *val, U32 depth)
3769 /* Merge several consecutive EXACTish nodes into one. */
3770 regnode *n = regnext(scan);
3772 regnode *next = scan + NODE_SZ_STR(scan);
3776 regnode *stop = scan;
3777 GET_RE_DEBUG_FLAGS_DECL;
3779 PERL_UNUSED_ARG(depth);
3782 PERL_ARGS_ASSERT_JOIN_EXACT;
3783 #ifndef EXPERIMENTAL_INPLACESCAN
3784 PERL_UNUSED_ARG(flags);
3785 PERL_UNUSED_ARG(val);
3787 DEBUG_PEEP("join",scan,depth);
3789 /* Look through the subsequent nodes in the chain. Skip NOTHING, merge
3790 * EXACT ones that are mergeable to the current one. */
3792 && (PL_regkind[OP(n)] == NOTHING
3793 || (stringok && OP(n) == OP(scan)))
3795 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX)
3798 if (OP(n) == TAIL || n > next)
3800 if (PL_regkind[OP(n)] == NOTHING) {
3801 DEBUG_PEEP("skip:",n,depth);
3802 NEXT_OFF(scan) += NEXT_OFF(n);
3803 next = n + NODE_STEP_REGNODE;
3810 else if (stringok) {
3811 const unsigned int oldl = STR_LEN(scan);
3812 regnode * const nnext = regnext(n);
3814 /* XXX I (khw) kind of doubt that this works on platforms (should
3815 * Perl ever run on one) where U8_MAX is above 255 because of lots
3816 * of other assumptions */
3817 /* Don't join if the sum can't fit into a single node */
3818 if (oldl + STR_LEN(n) > U8_MAX)
3821 DEBUG_PEEP("merg",n,depth);
3824 NEXT_OFF(scan) += NEXT_OFF(n);
3825 STR_LEN(scan) += STR_LEN(n);
3826 next = n + NODE_SZ_STR(n);
3827 /* Now we can overwrite *n : */
3828 Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
3836 #ifdef EXPERIMENTAL_INPLACESCAN
3837 if (flags && !NEXT_OFF(n)) {
3838 DEBUG_PEEP("atch", val, depth);
3839 if (reg_off_by_arg[OP(n)]) {
3840 ARG_SET(n, val - n);
3843 NEXT_OFF(n) = val - n;
3851 *unfolded_multi_char = FALSE;
3853 /* Here, all the adjacent mergeable EXACTish nodes have been merged. We
3854 * can now analyze for sequences of problematic code points. (Prior to
3855 * this final joining, sequences could have been split over boundaries, and
3856 * hence missed). The sequences only happen in folding, hence for any
3857 * non-EXACT EXACTish node */
3858 if (OP(scan) != EXACT && OP(scan) != EXACTL) {
3859 U8* s0 = (U8*) STRING(scan);
3861 U8* s_end = s0 + STR_LEN(scan);
3863 int total_count_delta = 0; /* Total delta number of characters that
3864 multi-char folds expand to */
3866 /* One pass is made over the node's string looking for all the
3867 * possibilities. To avoid some tests in the loop, there are two main
3868 * cases, for UTF-8 patterns (which can't have EXACTF nodes) and
3873 if (OP(scan) == EXACTFL) {
3876 /* An EXACTFL node would already have been changed to another
3877 * node type unless there is at least one character in it that
3878 * is problematic; likely a character whose fold definition
3879 * won't be known until runtime, and so has yet to be folded.
3880 * For all but the UTF-8 locale, folds are 1-1 in length, but
3881 * to handle the UTF-8 case, we need to create a temporary
3882 * folded copy using UTF-8 locale rules in order to analyze it.
3883 * This is because our macros that look to see if a sequence is
3884 * a multi-char fold assume everything is folded (otherwise the
3885 * tests in those macros would be too complicated and slow).
3886 * Note that here, the non-problematic folds will have already
3887 * been done, so we can just copy such characters. We actually
3888 * don't completely fold the EXACTFL string. We skip the
3889 * unfolded multi-char folds, as that would just create work
3890 * below to figure out the size they already are */
3892 Newx(folded, UTF8_MAX_FOLD_CHAR_EXPAND * STR_LEN(scan) + 1, U8);
3895 STRLEN s_len = UTF8SKIP(s);
3896 if (! is_PROBLEMATIC_LOCALE_FOLD_utf8(s)) {
3897 Copy(s, d, s_len, U8);
3900 else if (is_FOLDS_TO_MULTI_utf8(s)) {
3901 *unfolded_multi_char = TRUE;
3902 Copy(s, d, s_len, U8);
3905 else if (isASCII(*s)) {
3906 *(d++) = toFOLD(*s);
3910 _toFOLD_utf8_flags(s, s_end, d, &len, FOLD_FLAGS_FULL);
3916 /* Point the remainder of the routine to look at our temporary
3920 } /* End of creating folded copy of EXACTFL string */
3922 /* Examine the string for a multi-character fold sequence. UTF-8
3923 * patterns have all characters pre-folded by the time this code is
3925 while (s < s_end - 1) /* Can stop 1 before the end, as minimum
3926 length sequence we are looking for is 2 */
3928 int count = 0; /* How many characters in a multi-char fold */
3929 int len = is_MULTI_CHAR_FOLD_utf8_safe(s, s_end);
3930 if (! len) { /* Not a multi-char fold: get next char */
3935 /* Nodes with 'ss' require special handling, except for
3936 * EXACTFA-ish for which there is no multi-char fold to this */
3937 if (len == 2 && *s == 's' && *(s+1) == 's'
3938 && OP(scan) != EXACTFA
3939 && OP(scan) != EXACTFA_NO_TRIE)
3942 if (OP(scan) != EXACTFL) {
3943 OP(scan) = EXACTFU_SS;
3947 else { /* Here is a generic multi-char fold. */
3948 U8* multi_end = s + len;
3950 /* Count how many characters are in it. In the case of
3951 * /aa, no folds which contain ASCII code points are
3952 * allowed, so check for those, and skip if found. */
3953 if (OP(scan) != EXACTFA && OP(scan) != EXACTFA_NO_TRIE) {
3954 count = utf8_length(s, multi_end);
3958 while (s < multi_end) {
3961 goto next_iteration;
3971 /* The delta is how long the sequence is minus 1 (1 is how long
3972 * the character that folds to the sequence is) */
3973 total_count_delta += count - 1;
3977 /* We created a temporary folded copy of the string in EXACTFL
3978 * nodes. Therefore we need to be sure it doesn't go below zero,
3979 * as the real string could be shorter */
3980 if (OP(scan) == EXACTFL) {
3981 int total_chars = utf8_length((U8*) STRING(scan),
3982 (U8*) STRING(scan) + STR_LEN(scan));
3983 if (total_count_delta > total_chars) {
3984 total_count_delta = total_chars;
3988 *min_subtract += total_count_delta;
3991 else if (OP(scan) == EXACTFA) {
3993 /* Non-UTF-8 pattern, EXACTFA node. There can't be a multi-char
3994 * fold to the ASCII range (and there are no existing ones in the
3995 * upper latin1 range). But, as outlined in the comments preceding
3996 * this function, we need to flag any occurrences of the sharp s.
3997 * This character forbids trie formation (because of added
3999 #if UNICODE_MAJOR_VERSION > 3 /* no multifolds in early Unicode */ \
4000 || (UNICODE_MAJOR_VERSION == 3 && ( UNICODE_DOT_VERSION > 0) \
4001 || UNICODE_DOT_DOT_VERSION > 0)
4003 if (*s == LATIN_SMALL_LETTER_SHARP_S) {
4004 OP(scan) = EXACTFA_NO_TRIE;
4005 *unfolded_multi_char = TRUE;
4013 /* Non-UTF-8 pattern, not EXACTFA node. Look for the multi-char
4014 * folds that are all Latin1. As explained in the comments
4015 * preceding this function, we look also for the sharp s in EXACTF
4016 * and EXACTFL nodes; it can be in the final position. Otherwise
4017 * we can stop looking 1 byte earlier because have to find at least
4018 * two characters for a multi-fold */
4019 const U8* upper = (OP(scan) == EXACTF || OP(scan) == EXACTFL)
4024 int len = is_MULTI_CHAR_FOLD_latin1_safe(s, s_end);
4025 if (! len) { /* Not a multi-char fold. */
4026 if (*s == LATIN_SMALL_LETTER_SHARP_S
4027 && (OP(scan) == EXACTF || OP(scan) == EXACTFL))
4029 *unfolded_multi_char = TRUE;
4036 && isALPHA_FOLD_EQ(*s, 's')
4037 && isALPHA_FOLD_EQ(*(s+1), 's'))
4040 /* EXACTF nodes need to know that the minimum length
4041 * changed so that a sharp s in the string can match this
4042 * ss in the pattern, but they remain EXACTF nodes, as they
4043 * won't match this unless the target string is is UTF-8,
4044 * which we don't know until runtime. EXACTFL nodes can't
4045 * transform into EXACTFU nodes */
4046 if (OP(scan) != EXACTF && OP(scan) != EXACTFL) {
4047 OP(scan) = EXACTFU_SS;
4051 *min_subtract += len - 1;
4059 /* Allow dumping but overwriting the collection of skipped
4060 * ops and/or strings with fake optimized ops */
4061 n = scan + NODE_SZ_STR(scan);
4069 DEBUG_OPTIMISE_r(if (merged){DEBUG_PEEP("finl",scan,depth)});
4073 /* REx optimizer. Converts nodes into quicker variants "in place".
4074 Finds fixed substrings. */
4076 /* Stops at toplevel WHILEM as well as at "last". At end *scanp is set
4077 to the position after last scanned or to NULL. */
4079 #define INIT_AND_WITHP \
4080 assert(!and_withp); \
4081 Newx(and_withp,1, regnode_ssc); \
4082 SAVEFREEPV(and_withp)
4086 S_unwind_scan_frames(pTHX_ const void *p)
4088 scan_frame *f= (scan_frame *)p;
4090 scan_frame *n= f->next_frame;
4098 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
4099 SSize_t *minlenp, SSize_t *deltap,
4104 regnode_ssc *and_withp,
4105 U32 flags, U32 depth)
4106 /* scanp: Start here (read-write). */
4107 /* deltap: Write maxlen-minlen here. */
4108 /* last: Stop before this one. */
4109 /* data: string data about the pattern */
4110 /* stopparen: treat close N as END */
4111 /* recursed: which subroutines have we recursed into */
4112 /* and_withp: Valid if flags & SCF_DO_STCLASS_OR */
4114 /* There must be at least this number of characters to match */
4117 regnode *scan = *scanp, *next;
4119 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
4120 int is_inf_internal = 0; /* The studied chunk is infinite */
4121 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
4122 scan_data_t data_fake;
4123 SV *re_trie_maxbuff = NULL;
4124 regnode *first_non_open = scan;
4125 SSize_t stopmin = SSize_t_MAX;
4126 scan_frame *frame = NULL;
4127 GET_RE_DEBUG_FLAGS_DECL;
4129 PERL_ARGS_ASSERT_STUDY_CHUNK;
4130 RExC_study_started= 1;
4134 while (first_non_open && OP(first_non_open) == OPEN)
4135 first_non_open=regnext(first_non_open);
4141 RExC_study_chunk_recursed_count++;
4143 DEBUG_OPTIMISE_MORE_r(
4145 Perl_re_indentf( aTHX_ "study_chunk stopparen=%ld recursed_count=%lu depth=%lu recursed_depth=%lu scan=%p last=%p",
4146 depth, (long)stopparen,
4147 (unsigned long)RExC_study_chunk_recursed_count,
4148 (unsigned long)depth, (unsigned long)recursed_depth,
4151 if (recursed_depth) {
4154 for ( j = 0 ; j < recursed_depth ; j++ ) {
4155 for ( i = 0 ; i < (U32)RExC_npar ; i++ ) {
4157 PAREN_TEST(RExC_study_chunk_recursed +
4158 ( j * RExC_study_chunk_recursed_bytes), i )
4161 !PAREN_TEST(RExC_study_chunk_recursed +
4162 (( j - 1 ) * RExC_study_chunk_recursed_bytes), i)
4165 Perl_re_printf( aTHX_ " %d",(int)i);
4169 if ( j + 1 < recursed_depth ) {
4170 Perl_re_printf( aTHX_ ",");
4174 Perl_re_printf( aTHX_ "\n");
4177 while ( scan && OP(scan) != END && scan < last ){
4178 UV min_subtract = 0; /* How mmany chars to subtract from the minimum
4179 node length to get a real minimum (because
4180 the folded version may be shorter) */
4181 bool unfolded_multi_char = FALSE;
4182 /* Peephole optimizer: */
4183 DEBUG_STUDYDATA("Peep:", data, depth);
4184 DEBUG_PEEP("Peep", scan, depth);
4187 /* The reason we do this here is that we need to deal with things like
4188 * /(?:f)(?:o)(?:o)/ which cant be dealt with by the normal EXACT
4189 * parsing code, as each (?:..) is handled by a different invocation of
4192 JOIN_EXACT(scan,&min_subtract, &unfolded_multi_char, 0);
4194 /* Follow the next-chain of the current node and optimize
4195 away all the NOTHINGs from it. */
4196 if (OP(scan) != CURLYX) {
4197 const int max = (reg_off_by_arg[OP(scan)]
4199 /* I32 may be smaller than U16 on CRAYs! */
4200 : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
4201 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
4205 /* Skip NOTHING and LONGJMP. */
4206 while ((n = regnext(n))
4207 && ((PL_regkind[OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
4208 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
4209 && off + noff < max)
4211 if (reg_off_by_arg[OP(scan)])
4214 NEXT_OFF(scan) = off;
4217 /* The principal pseudo-switch. Cannot be a switch, since we
4218 look into several different things. */
4219 if ( OP(scan) == DEFINEP ) {
4221 SSize_t deltanext = 0;
4222 SSize_t fake_last_close = 0;
4223 I32 f = SCF_IN_DEFINE;
4225 StructCopy(&zero_scan_data, &data_fake, scan_data_t);
4226 scan = regnext(scan);
4227 assert( OP(scan) == IFTHEN );
4228 DEBUG_PEEP("expect IFTHEN", scan, depth);
4230 data_fake.last_closep= &fake_last_close;
4232 next = regnext(scan);
4233 scan = NEXTOPER(NEXTOPER(scan));
4234 DEBUG_PEEP("scan", scan, depth);
4235 DEBUG_PEEP("next", next, depth);
4237 /* we suppose the run is continuous, last=next...
4238 * NOTE we dont use the return here! */
4239 (void)study_chunk(pRExC_state, &scan, &minlen,
4240 &deltanext, next, &data_fake, stopparen,
4241 recursed_depth, NULL, f, depth+1);
4246 OP(scan) == BRANCH ||
4247 OP(scan) == BRANCHJ ||
4250 next = regnext(scan);
4253 /* The op(next)==code check below is to see if we
4254 * have "BRANCH-BRANCH", "BRANCHJ-BRANCHJ", "IFTHEN-IFTHEN"
4255 * IFTHEN is special as it might not appear in pairs.
4256 * Not sure whether BRANCH-BRANCHJ is possible, regardless
4257 * we dont handle it cleanly. */
4258 if (OP(next) == code || code == IFTHEN) {
4259 /* NOTE - There is similar code to this block below for
4260 * handling TRIE nodes on a re-study. If you change stuff here
4261 * check there too. */
4262 SSize_t max1 = 0, min1 = SSize_t_MAX, num = 0;
4264 regnode * const startbranch=scan;
4266 if (flags & SCF_DO_SUBSTR) {
4267 /* Cannot merge strings after this. */
4268 scan_commit(pRExC_state, data, minlenp, is_inf);
4271 if (flags & SCF_DO_STCLASS)
4272 ssc_init_zero(pRExC_state, &accum);
4274 while (OP(scan) == code) {
4275 SSize_t deltanext, minnext, fake;
4277 regnode_ssc this_class;
4279 DEBUG_PEEP("Branch", scan, depth);
4282 StructCopy(&zero_scan_data, &data_fake, scan_data_t);
4284 data_fake.whilem_c = data->whilem_c;
4285 data_fake.last_closep = data->last_closep;
4288 data_fake.last_closep = &fake;
4290 data_fake.pos_delta = delta;
4291 next = regnext(scan);
4293 scan = NEXTOPER(scan); /* everything */
4294 if (code != BRANCH) /* everything but BRANCH */
4295 scan = NEXTOPER(scan);
4297 if (flags & SCF_DO_STCLASS) {
4298 ssc_init(pRExC_state, &this_class);
4299 data_fake.start_class = &this_class;
4300 f = SCF_DO_STCLASS_AND;
4302 if (flags & SCF_WHILEM_VISITED_POS)
4303 f |= SCF_WHILEM_VISITED_POS;
4305 /* we suppose the run is continuous, last=next...*/
4306 minnext = study_chunk(pRExC_state, &scan, minlenp,
4307 &deltanext, next, &data_fake, stopparen,
4308 recursed_depth, NULL, f,depth+1);
4312 if (deltanext == SSize_t_MAX) {
4313 is_inf = is_inf_internal = 1;
4315 } else if (max1 < minnext + deltanext)
4316 max1 = minnext + deltanext;
4318 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
4320 if (data_fake.flags & SCF_SEEN_ACCEPT) {
4321 if ( stopmin > minnext)
4322 stopmin = min + min1;
4323 flags &= ~SCF_DO_SUBSTR;
4325 data->flags |= SCF_SEEN_ACCEPT;
4328 if (data_fake.flags & SF_HAS_EVAL)
4329 data->flags |= SF_HAS_EVAL;
4330 data->whilem_c = data_fake.whilem_c;
4332 if (flags & SCF_DO_STCLASS)
4333 ssc_or(pRExC_state, &accum, (regnode_charclass*)&this_class);
4335 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
4337 if (flags & SCF_DO_SUBSTR) {
4338 data->pos_min += min1;
4339 if (data->pos_delta >= SSize_t_MAX - (max1 - min1))
4340 data->pos_delta = SSize_t_MAX;
4342 data->pos_delta += max1 - min1;
4343 if (max1 != min1 || is_inf)
4344 data->longest = &(data->longest_float);
4347 if (delta == SSize_t_MAX
4348 || SSize_t_MAX - delta - (max1 - min1) < 0)
4349 delta = SSize_t_MAX;
4351 delta += max1 - min1;
4352 if (flags & SCF_DO_STCLASS_OR) {
4353 ssc_or(pRExC_state, data->start_class, (regnode_charclass*) &accum);
4355 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
4356 flags &= ~SCF_DO_STCLASS;
4359 else if (flags & SCF_DO_STCLASS_AND) {
4361 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) &accum);
4362 flags &= ~SCF_DO_STCLASS;
4365 /* Switch to OR mode: cache the old value of
4366 * data->start_class */
4368 StructCopy(data->start_class, and_withp, regnode_ssc);
4369 flags &= ~SCF_DO_STCLASS_AND;
4370 StructCopy(&accum, data->start_class, regnode_ssc);
4371 flags |= SCF_DO_STCLASS_OR;
4375 if (PERL_ENABLE_TRIE_OPTIMISATION &&
4376 OP( startbranch ) == BRANCH )
4380 Assuming this was/is a branch we are dealing with: 'scan'
4381 now points at the item that follows the branch sequence,
4382 whatever it is. We now start at the beginning of the
4383 sequence and look for subsequences of
4389 which would be constructed from a pattern like
4392 If we can find such a subsequence we need to turn the first
4393 element into a trie and then add the subsequent branch exact
4394 strings to the trie.
4398 1. patterns where the whole set of branches can be
4401 2. patterns where only a subset can be converted.
4403 In case 1 we can replace the whole set with a single regop
4404 for the trie. In case 2 we need to keep the start and end
4407 'BRANCH EXACT; BRANCH EXACT; BRANCH X'
4408 becomes BRANCH TRIE; BRANCH X;
4410 There is an additional case, that being where there is a
4411 common prefix, which gets split out into an EXACT like node
4412 preceding the TRIE node.
4414 If x(1..n)==tail then we can do a simple trie, if not we make
4415 a "jump" trie, such that when we match the appropriate word
4416 we "jump" to the appropriate tail node. Essentially we turn
4417 a nested if into a case structure of sorts.
4422 if (!re_trie_maxbuff) {
4423 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
4424 if (!SvIOK(re_trie_maxbuff))
4425 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
4427 if ( SvIV(re_trie_maxbuff)>=0 ) {
4429 regnode *first = (regnode *)NULL;
4430 regnode *last = (regnode *)NULL;
4431 regnode *tail = scan;
4435 /* var tail is used because there may be a TAIL
4436 regop in the way. Ie, the exacts will point to the
4437 thing following the TAIL, but the last branch will
4438 point at the TAIL. So we advance tail. If we
4439 have nested (?:) we may have to move through several
4443 while ( OP( tail ) == TAIL ) {
4444 /* this is the TAIL generated by (?:) */
4445 tail = regnext( tail );
4449 DEBUG_TRIE_COMPILE_r({
4450 regprop(RExC_rx, RExC_mysv, tail, NULL, pRExC_state);
4451 Perl_re_indentf( aTHX_ "%s %" UVuf ":%s\n",
4453 "Looking for TRIE'able sequences. Tail node is ",
4454 (UV)(tail - RExC_emit_start),
4455 SvPV_nolen_const( RExC_mysv )
4461 Step through the branches
4462 cur represents each branch,
4463 noper is the first thing to be matched as part
4465 noper_next is the regnext() of that node.
4467 We normally handle a case like this
4468 /FOO[xyz]|BAR[pqr]/ via a "jump trie" but we also
4469 support building with NOJUMPTRIE, which restricts
4470 the trie logic to structures like /FOO|BAR/.
4472 If noper is a trieable nodetype then the branch is
4473 a possible optimization target. If we are building
4474 under NOJUMPTRIE then we require that noper_next is
4475 the same as scan (our current position in the regex
4478 Once we have two or more consecutive such branches
4479 we can create a trie of the EXACT's contents and
4480 stitch it in place into the program.
4482 If the sequence represents all of the branches in
4483 the alternation we replace the entire thing with a
4486 Otherwise when it is a subsequence we need to
4487 stitch it in place and replace only the relevant
4488 branches. This means the first branch has to remain
4489 as it is used by the alternation logic, and its
4490 next pointer, and needs to be repointed at the item
4491 on the branch chain following the last branch we
4492 have optimized away.
4494 This could be either a BRANCH, in which case the
4495 subsequence is internal, or it could be the item
4496 following the branch sequence in which case the
4497 subsequence is at the end (which does not
4498 necessarily mean the first node is the start of the
4501 TRIE_TYPE(X) is a define which maps the optype to a
4505 ----------------+-----------
4509 EXACTFU_SS | EXACTFU
4512 EXACTFLU8 | EXACTFLU8
4516 #define TRIE_TYPE(X) ( ( NOTHING == (X) ) \
4518 : ( EXACT == (X) ) \
4520 : ( EXACTFU == (X) || EXACTFU_SS == (X) ) \
4522 : ( EXACTFA == (X) ) \
4524 : ( EXACTL == (X) ) \
4526 : ( EXACTFLU8 == (X) ) \
4530 /* dont use tail as the end marker for this traverse */
4531 for ( cur = startbranch ; cur != scan ; cur = regnext( cur ) ) {
4532 regnode * const noper = NEXTOPER( cur );
4533 U8 noper_type = OP( noper );
4534 U8 noper_trietype = TRIE_TYPE( noper_type );
4535 #if defined(DEBUGGING) || defined(NOJUMPTRIE)
4536 regnode * const noper_next = regnext( noper );
4537 U8 noper_next_type = (noper_next && noper_next < tail) ? OP(noper_next) : 0;
4538 U8 noper_next_trietype = (noper_next && noper_next < tail) ? TRIE_TYPE( noper_next_type ) :0;
4541 DEBUG_TRIE_COMPILE_r({
4542 regprop(RExC_rx, RExC_mysv, cur, NULL, pRExC_state);
4543 Perl_re_indentf( aTHX_ "- %d:%s (%d)",
4545 REG_NODE_NUM(cur), SvPV_nolen_const( RExC_mysv ), REG_NODE_NUM(cur) );
4547 regprop(RExC_rx, RExC_mysv, noper, NULL, pRExC_state);
4548 Perl_re_printf( aTHX_ " -> %d:%s",
4549 REG_NODE_NUM(noper), SvPV_nolen_const(RExC_mysv));
4552 regprop(RExC_rx, RExC_mysv, noper_next, NULL, pRExC_state);
4553 Perl_re_printf( aTHX_ "\t=> %d:%s\t",
4554 REG_NODE_NUM(noper_next), SvPV_nolen_const(RExC_mysv));
4556 Perl_re_printf( aTHX_ "(First==%d,Last==%d,Cur==%d,tt==%s,ntt==%s,nntt==%s)\n",
4557 REG_NODE_NUM(first), REG_NODE_NUM(last), REG_NODE_NUM(cur),
4558 PL_reg_name[trietype], PL_reg_name[noper_trietype], PL_reg_name[noper_next_trietype]
4562 /* Is noper a trieable nodetype that can be merged
4563 * with the current trie (if there is one)? */
4567 ( noper_trietype == NOTHING )
4568 || ( trietype == NOTHING )
4569 || ( trietype == noper_trietype )
4572 && noper_next >= tail
4576 /* Handle mergable triable node Either we are
4577 * the first node in a new trieable sequence,
4578 * in which case we do some bookkeeping,
4579 * otherwise we update the end pointer. */
4582 if ( noper_trietype == NOTHING ) {
4583 #if !defined(DEBUGGING) && !defined(NOJUMPTRIE)
4584 regnode * const noper_next = regnext( noper );
4585 U8 noper_next_type = (noper_next && noper_next < tail) ? OP(noper_next) : 0;
4586 U8 noper_next_trietype = noper_next_type ? TRIE_TYPE( noper_next_type ) :0;
4589 if ( noper_next_trietype ) {
4590 trietype = noper_next_trietype;
4591 } else if (noper_next_type) {
4592 /* a NOTHING regop is 1 regop wide.
4593 * We need at least two for a trie
4594 * so we can't merge this in */
4598 trietype = noper_trietype;
4601 if ( trietype == NOTHING )
4602 trietype = noper_trietype;
4607 } /* end handle mergable triable node */
4609 /* handle unmergable node -
4610 * noper may either be a triable node which can
4611 * not be tried together with the current trie,
4612 * or a non triable node */
4614 /* If last is set and trietype is not
4615 * NOTHING then we have found at least two
4616 * triable branch sequences in a row of a
4617 * similar trietype so we can turn them
4618 * into a trie. If/when we allow NOTHING to
4619 * start a trie sequence this condition
4620 * will be required, and it isn't expensive
4621 * so we leave it in for now. */
4622 if ( trietype && trietype != NOTHING )
4623 make_trie( pRExC_state,
4624 startbranch, first, cur, tail,
4625 count, trietype, depth+1 );
4626 last = NULL; /* note: we clear/update
4627 first, trietype etc below,
4628 so we dont do it here */
4632 && noper_next >= tail
4635 /* noper is triable, so we can start a new
4639 trietype = noper_trietype;
4641 /* if we already saw a first but the
4642 * current node is not triable then we have
4643 * to reset the first information. */
4648 } /* end handle unmergable node */
4649 } /* loop over branches */
4650 DEBUG_TRIE_COMPILE_r({
4651 regprop(RExC_rx, RExC_mysv, cur, NULL, pRExC_state);
4652 Perl_re_indentf( aTHX_ "- %s (%d) <SCAN FINISHED> ",
4653 depth+1, SvPV_nolen_const( RExC_mysv ),REG_NODE_NUM(cur));
4654 Perl_re_printf( aTHX_ "(First==%d, Last==%d, Cur==%d, tt==%s)\n",
4655 REG_NODE_NUM(first), REG_NODE_NUM(last), REG_NODE_NUM(cur),
4656 PL_reg_name[trietype]
4660 if ( last && trietype ) {
4661 if ( trietype != NOTHING ) {
4662 /* the last branch of the sequence was part of
4663 * a trie, so we have to construct it here
4664 * outside of the loop */
4665 made= make_trie( pRExC_state, startbranch,
4666 first, scan, tail, count,
4667 trietype, depth+1 );
4668 #ifdef TRIE_STUDY_OPT
4669 if ( ((made == MADE_EXACT_TRIE &&
4670 startbranch == first)
4671 || ( first_non_open == first )) &&
4673 flags |= SCF_TRIE_RESTUDY;
4674 if ( startbranch == first
4677 RExC_seen &=~REG_TOP_LEVEL_BRANCHES_SEEN;
4682 /* at this point we know whatever we have is a
4683 * NOTHING sequence/branch AND if 'startbranch'
4684 * is 'first' then we can turn the whole thing
4687 if ( startbranch == first ) {
4689 /* the entire thing is a NOTHING sequence,
4690 * something like this: (?:|) So we can
4691 * turn it into a plain NOTHING op. */
4692 DEBUG_TRIE_COMPILE_r({
4693 regprop(RExC_rx, RExC_mysv, cur, NULL, pRExC_state);
4694 Perl_re_indentf( aTHX_ "- %s (%d) <NOTHING BRANCH SEQUENCE>\n",
4696 SvPV_nolen_const( RExC_mysv ),REG_NODE_NUM(cur));
4699 OP(startbranch)= NOTHING;
4700 NEXT_OFF(startbranch)= tail - startbranch;
4701 for ( opt= startbranch + 1; opt < tail ; opt++ )
4705 } /* end if ( last) */
4706 } /* TRIE_MAXBUF is non zero */
4711 else if ( code == BRANCHJ ) { /* single branch is optimized. */
4712 scan = NEXTOPER(NEXTOPER(scan));
4713 } else /* single branch is optimized. */
4714 scan = NEXTOPER(scan);
4716 } else if (OP(scan) == SUSPEND || OP(scan) == GOSUB) {
4718 regnode *start = NULL;
4719 regnode *end = NULL;
4720 U32 my_recursed_depth= recursed_depth;
4722 if (OP(scan) != SUSPEND) { /* GOSUB */
4723 /* Do setup, note this code has side effects beyond
4724 * the rest of this block. Specifically setting
4725 * RExC_recurse[] must happen at least once during
4728 RExC_recurse[ARG2L(scan)] = scan;
4729 start = RExC_open_parens[paren];
4730 end = RExC_close_parens[paren];
4732 /* NOTE we MUST always execute the above code, even
4733 * if we do nothing with a GOSUB */
4735 ( flags & SCF_IN_DEFINE )
4738 (is_inf_internal || is_inf || (data && data->flags & SF_IS_INF))
4740 ( (flags & (SCF_DO_STCLASS | SCF_DO_SUBSTR)) == 0 )
4743 /* no need to do anything here if we are in a define. */
4744 /* or we are after some kind of infinite construct
4745 * so we can skip recursing into this item.
4746 * Since it is infinite we will not change the maxlen
4747 * or delta, and if we miss something that might raise
4748 * the minlen it will merely pessimise a little.
4750 * Iow /(?(DEFINE)(?<foo>foo|food))a+(?&foo)/
4751 * might result in a minlen of 1 and not of 4,
4752 * but this doesn't make us mismatch, just try a bit
4753 * harder than we should.
4755 scan= regnext(scan);
4762 !PAREN_TEST(RExC_study_chunk_recursed + ((recursed_depth-1) * RExC_study_chunk_recursed_bytes), paren)
4764 /* it is quite possible that there are more efficient ways
4765 * to do this. We maintain a bitmap per level of recursion
4766 * of which patterns we have entered so we can detect if a
4767 * pattern creates a possible infinite loop. When we
4768 * recurse down a level we copy the previous levels bitmap
4769 * down. When we are at recursion level 0 we zero the top
4770 * level bitmap. It would be nice to implement a different
4771 * more efficient way of doing this. In particular the top
4772 * level bitmap may be unnecessary.
4774 if (!recursed_depth) {
4775 Zero(RExC_study_chunk_recursed, RExC_study_chunk_recursed_bytes, U8);
4777 Copy(RExC_study_chunk_recursed + ((recursed_depth-1) * RExC_study_chunk_recursed_bytes),
4778 RExC_study_chunk_recursed + (recursed_depth * RExC_study_chunk_recursed_bytes),
4779 RExC_study_chunk_recursed_bytes, U8);
4781 /* we havent recursed into this paren yet, so recurse into it */
4782 DEBUG_STUDYDATA("gosub-set:", data,depth);
4783 PAREN_SET(RExC_study_chunk_recursed + (recursed_depth * RExC_study_chunk_recursed_bytes), paren);
4784 my_recursed_depth= recursed_depth + 1;
4786 DEBUG_STUDYDATA("gosub-inf:", data,depth);
4787 /* some form of infinite recursion, assume infinite length
4789 if (flags & SCF_DO_SUBSTR) {
4790 scan_commit(pRExC_state, data, minlenp, is_inf);
4791 data->longest = &(data->longest_float);
4793 is_inf = is_inf_internal = 1;
4794 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
4795 ssc_anything(data->start_class);
4796 flags &= ~SCF_DO_STCLASS;
4798 start= NULL; /* reset start so we dont recurse later on. */
4803 end = regnext(scan);
4806 scan_frame *newframe;
4808 if (!RExC_frame_last) {
4809 Newxz(newframe, 1, scan_frame);
4810 SAVEDESTRUCTOR_X(S_unwind_scan_frames, newframe);
4811 RExC_frame_head= newframe;
4813 } else if (!RExC_frame_last->next_frame) {
4814 Newxz(newframe,1,scan_frame);
4815 RExC_frame_last->next_frame= newframe;
4816 newframe->prev_frame= RExC_frame_last;
4819 newframe= RExC_frame_last->next_frame;
4821 RExC_frame_last= newframe;
4823 newframe->next_regnode = regnext(scan);
4824 newframe->last_regnode = last;
4825 newframe->stopparen = stopparen;
4826 newframe->prev_recursed_depth = recursed_depth;
4827 newframe->this_prev_frame= frame;
4829 DEBUG_STUDYDATA("frame-new:",data,depth);
4830 DEBUG_PEEP("fnew", scan, depth);
4837 recursed_depth= my_recursed_depth;
4842 else if (OP(scan) == EXACT || OP(scan) == EXACTL) {
4843 SSize_t l = STR_LEN(scan);
4846 const U8 * const s = (U8*)STRING(scan);
4847 uc = utf8_to_uvchr_buf(s, s + l, NULL);
4848 l = utf8_length(s, s + l);
4850 uc = *((U8*)STRING(scan));
4853 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
4854 /* The code below prefers earlier match for fixed
4855 offset, later match for variable offset. */
4856 if (data->last_end == -1) { /* Update the start info. */
4857 data->last_start_min = data->pos_min;
4858 data->last_start_max = is_inf
4859 ? SSize_t_MAX : data->pos_min + data->pos_delta;
4861 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
4863 SvUTF8_on(data->last_found);
4865 SV * const sv = data->last_found;
4866 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
4867 mg_find(sv, PERL_MAGIC_utf8) : NULL;
4868 if (mg && mg->mg_len >= 0)
4869 mg->mg_len += utf8_length((U8*)STRING(scan),
4870 (U8*)STRING(scan)+STR_LEN(scan));
4872 data->last_end = data->pos_min + l;
4873 data->pos_min += l; /* As in the first entry. */
4874 data->flags &= ~SF_BEFORE_EOL;
4877 /* ANDing the code point leaves at most it, and not in locale, and
4878 * can't match null string */
4879 if (flags & SCF_DO_STCLASS_AND) {
4880 ssc_cp_and(data->start_class, uc);
4881 ANYOF_FLAGS(data->start_class) &= ~SSC_MATCHES_EMPTY_STRING;
4882 ssc_clear_locale(data->start_class);
4884 else if (flags & SCF_DO_STCLASS_OR) {
4885 ssc_add_cp(data->start_class, uc);
4886 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
4888 /* See commit msg 749e076fceedeb708a624933726e7989f2302f6a */
4889 ANYOF_FLAGS(data->start_class) &= ~SSC_MATCHES_EMPTY_STRING;
4891 flags &= ~SCF_DO_STCLASS;
4893 else if (PL_regkind[OP(scan)] == EXACT) {
4894 /* But OP != EXACT!, so is EXACTFish */
4895 SSize_t l = STR_LEN(scan);
4896 const U8 * s = (U8*)STRING(scan);
4898 /* Search for fixed substrings supports EXACT only. */
4899 if (flags & SCF_DO_SUBSTR) {
4901 scan_commit(pRExC_state, data, minlenp, is_inf);
4904 l = utf8_length(s, s + l);
4906 if (unfolded_multi_char) {
4907 RExC_seen |= REG_UNFOLDED_MULTI_SEEN;
4909 min += l - min_subtract;
4911 delta += min_subtract;
4912 if (flags & SCF_DO_SUBSTR) {
4913 data->pos_min += l - min_subtract;
4914 if (data->pos_min < 0) {
4917 data->pos_delta += min_subtract;
4919 data->longest = &(data->longest_float);
4923 if (flags & SCF_DO_STCLASS) {
4924 SV* EXACTF_invlist = _make_exactf_invlist(pRExC_state, scan);
4926 assert(EXACTF_invlist);
4927 if (flags & SCF_DO_STCLASS_AND) {
4928 if (OP(scan) != EXACTFL)
4929 ssc_clear_locale(data->start_class);
4930 ANYOF_FLAGS(data->start_class) &= ~SSC_MATCHES_EMPTY_STRING;
4931 ANYOF_POSIXL_ZERO(data->start_class);
4932 ssc_intersection(data->start_class, EXACTF_invlist, FALSE);
4934 else { /* SCF_DO_STCLASS_OR */
4935 ssc_union(data->start_class, EXACTF_invlist, FALSE);
4936 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
4938 /* See commit msg 749e076fceedeb708a624933726e7989f2302f6a */
4939 ANYOF_FLAGS(data->start_class) &= ~SSC_MATCHES_EMPTY_STRING;
4941 flags &= ~SCF_DO_STCLASS;
4942 SvREFCNT_dec(EXACTF_invlist);
4945 else if (REGNODE_VARIES(OP(scan))) {
4946 SSize_t mincount, maxcount, minnext, deltanext, pos_before = 0;
4947 I32 fl = 0, f = flags;
4948 regnode * const oscan = scan;
4949 regnode_ssc this_class;
4950 regnode_ssc *oclass = NULL;
4951 I32 next_is_eval = 0;
4953 switch (PL_regkind[OP(scan)]) {
4954 case WHILEM: /* End of (?:...)* . */
4955 scan = NEXTOPER(scan);
4958 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
4959 next = NEXTOPER(scan);
4960 if (OP(next) == EXACT
4961 || OP(next) == EXACTL
4962 || (flags & SCF_DO_STCLASS))
4965 maxcount = REG_INFTY;
4966 next = regnext(scan);
4967 scan = NEXTOPER(scan);
4971 if (flags & SCF_DO_SUBSTR)
4976 if (flags & SCF_DO_STCLASS) {
4978 maxcount = REG_INFTY;
4979 next = regnext(scan);
4980 scan = NEXTOPER(scan);
4983 if (flags & SCF_DO_SUBSTR) {
4984 scan_commit(pRExC_state, data, minlenp, is_inf);
4985 /* Cannot extend fixed substrings */
4986 data->longest = &(data->longest_float);
4988 is_inf = is_inf_internal = 1;
4989 scan = regnext(scan);
4990 goto optimize_curly_tail;
4992 if (stopparen>0 && (OP(scan)==CURLYN || OP(scan)==CURLYM)
4993 && (scan->flags == stopparen))
4998 mincount = ARG1(scan);
4999 maxcount = ARG2(scan);
5001 next = regnext(scan);
5002 if (OP(scan) == CURLYX) {
5003 I32 lp = (data ? *(data->last_closep) : 0);
5004 scan->flags = ((lp <= (I32)U8_MAX) ? (U8)lp : U8_MAX);
5006 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
5007 next_is_eval = (OP(scan) == EVAL);
5009 if (flags & SCF_DO_SUBSTR) {
5011 scan_commit(pRExC_state, data, minlenp, is_inf);
5012 /* Cannot extend fixed substrings */
5013 pos_before = data->pos_min;
5017 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
5019 data->flags |= SF_IS_INF;
5021 if (flags & SCF_DO_STCLASS) {
5022 ssc_init(pRExC_state, &this_class);
5023 oclass = data->start_class;
5024 data->start_class = &this_class;
5025 f |= SCF_DO_STCLASS_AND;
5026 f &= ~SCF_DO_STCLASS_OR;
5028 /* Exclude from super-linear cache processing any {n,m}
5029 regops for which the combination of input pos and regex
5030 pos is not enough information to determine if a match
5033 For example, in the regex /foo(bar\s*){4,8}baz/ with the
5034 regex pos at the \s*, the prospects for a match depend not
5035 only on the input position but also on how many (bar\s*)
5036 repeats into the {4,8} we are. */
5037 if ((mincount > 1) || (maxcount > 1 && maxcount != REG_INFTY))
5038 f &= ~SCF_WHILEM_VISITED_POS;
5040 /* This will finish on WHILEM, setting scan, or on NULL: */
5041 minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext,
5042 last, data, stopparen, recursed_depth, NULL,
5044 ? (f & ~SCF_DO_SUBSTR)
5048 if (flags & SCF_DO_STCLASS)
5049 data->start_class = oclass;
5050 if (mincount == 0 || minnext == 0) {
5051 if (flags & SCF_DO_STCLASS_OR) {
5052 ssc_or(pRExC_state, data->start_class, (regnode_charclass *) &this_class);
5054 else if (flags & SCF_DO_STCLASS_AND) {
5055 /* Switch to OR mode: cache the old value of
5056 * data->start_class */
5058 StructCopy(data->start_class, and_withp, regnode_ssc);
5059 flags &= ~SCF_DO_STCLASS_AND;
5060 StructCopy(&this_class, data->start_class, regnode_ssc);
5061 flags |= SCF_DO_STCLASS_OR;
5062 ANYOF_FLAGS(data->start_class)
5063 |= SSC_MATCHES_EMPTY_STRING;
5065 } else { /* Non-zero len */
5066 if (flags & SCF_DO_STCLASS_OR) {
5067 ssc_or(pRExC_state, data->start_class, (regnode_charclass *) &this_class);
5068 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
5070 else if (flags & SCF_DO_STCLASS_AND)
5071 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) &this_class);
5072 flags &= ~SCF_DO_STCLASS;
5074 if (!scan) /* It was not CURLYX, but CURLY. */
5076 if (!(flags & SCF_TRIE_DOING_RESTUDY)
5077 /* ? quantifier ok, except for (?{ ... }) */
5078 && (next_is_eval || !(mincount == 0 && maxcount == 1))
5079 && (minnext == 0) && (deltanext == 0)
5080 && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
5081 && maxcount <= REG_INFTY/3) /* Complement check for big
5084 /* Fatal warnings may leak the regexp without this: */
5085 SAVEFREESV(RExC_rx_sv);
5086 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP),
5087 "Quantifier unexpected on zero-length expression "
5088 "in regex m/%" UTF8f "/",
5089 UTF8fARG(UTF, RExC_precomp_end - RExC_precomp,
5091 (void)ReREFCNT_inc(RExC_rx_sv);
5094 min += minnext * mincount;
5095 is_inf_internal |= deltanext == SSize_t_MAX
5096 || (maxcount == REG_INFTY && minnext + deltanext > 0);
5097 is_inf |= is_inf_internal;
5099 delta = SSize_t_MAX;
5101 delta += (minnext + deltanext) * maxcount
5102 - minnext * mincount;
5104 /* Try powerful optimization CURLYX => CURLYN. */
5105 if ( OP(oscan) == CURLYX && data
5106 && data->flags & SF_IN_PAR
5107 && !(data->flags & SF_HAS_EVAL)
5108 && !deltanext && minnext == 1 ) {
5109 /* Try to optimize to CURLYN. */
5110 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
5111 regnode * const nxt1 = nxt;
5118 if (!REGNODE_SIMPLE(OP(nxt))
5119 && !(PL_regkind[OP(nxt)] == EXACT
5120 && STR_LEN(nxt) == 1))
5126 if (OP(nxt) != CLOSE)
5128 if (RExC_open_parens) {
5129 RExC_open_parens[ARG(nxt1)]=oscan; /*open->CURLYM*/
5130 RExC_close_parens[ARG(nxt1)]=nxt+2; /*close->while*/
5132 /* Now we know that nxt2 is the only contents: */
5133 oscan->flags = (U8)ARG(nxt);
5135 OP(nxt1) = NOTHING; /* was OPEN. */
5138 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
5139 NEXT_OFF(nxt1+ 1) = 0; /* just for consistency. */
5140 NEXT_OFF(nxt2) = 0; /* just for consistency with CURLY. */
5141 OP(nxt) = OPTIMIZED; /* was CLOSE. */
5142 OP(nxt + 1) = OPTIMIZED; /* was count. */
5143 NEXT_OFF(nxt+ 1) = 0; /* just for consistency. */
5148 /* Try optimization CURLYX => CURLYM. */
5149 if ( OP(oscan) == CURLYX && data
5150 && !(data->flags & SF_HAS_PAR)
5151 && !(data->flags & SF_HAS_EVAL)
5152 && !deltanext /* atom is fixed width */
5153 && minnext != 0 /* CURLYM can't handle zero width */
5155 /* Nor characters whose fold at run-time may be
5156 * multi-character */
5157 && ! (RExC_seen & REG_UNFOLDED_MULTI_SEEN)
5159 /* XXXX How to optimize if data == 0? */
5160 /* Optimize to a simpler form. */
5161 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
5165 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
5166 && (OP(nxt2) != WHILEM))
5168 OP(nxt2) = SUCCEED; /* Whas WHILEM */
5169 /* Need to optimize away parenths. */
5170 if ((data->flags & SF_IN_PAR) && OP(nxt) == CLOSE) {
5171 /* Set the parenth number. */
5172 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
5174 oscan->flags = (U8)ARG(nxt);
5175 if (RExC_open_parens) {
5176 RExC_open_parens[ARG(nxt1)]=oscan; /*open->CURLYM*/
5177 RExC_close_parens[ARG(nxt1)]=nxt2+1; /*close->NOTHING*/
5179 OP(nxt1) = OPTIMIZED; /* was OPEN. */
5180 OP(nxt) = OPTIMIZED; /* was CLOSE. */
5183 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
5184 OP(nxt + 1) = OPTIMIZED; /* was count. */
5185 NEXT_OFF(nxt1 + 1) = 0; /* just for consistency. */
5186 NEXT_OFF(nxt + 1) = 0; /* just for consistency. */
5189 while ( nxt1 && (OP(nxt1) != WHILEM)) {
5190 regnode *nnxt = regnext(nxt1);
5192 if (reg_off_by_arg[OP(nxt1)])
5193 ARG_SET(nxt1, nxt2 - nxt1);
5194 else if (nxt2 - nxt1 < U16_MAX)
5195 NEXT_OFF(nxt1) = nxt2 - nxt1;
5197 OP(nxt) = NOTHING; /* Cannot beautify */
5202 /* Optimize again: */
5203 study_chunk(pRExC_state, &nxt1, minlenp, &deltanext, nxt,
5204 NULL, stopparen, recursed_depth, NULL, 0,depth+1);
5209 else if ((OP(oscan) == CURLYX)
5210 && (flags & SCF_WHILEM_VISITED_POS)
5211 /* See the comment on a similar expression above.
5212 However, this time it's not a subexpression
5213 we care about, but the expression itself. */
5214 && (maxcount == REG_INFTY)
5216 /* This stays as CURLYX, we can put the count/of pair. */
5217 /* Find WHILEM (as in regexec.c) */
5218 regnode *nxt = oscan + NEXT_OFF(oscan);
5220 if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
5222 nxt = PREVOPER(nxt);
5223 if (nxt->flags & 0xf) {
5224 /* we've already set whilem count on this node */
5225 } else if (++data->whilem_c < 16) {
5226 assert(data->whilem_c <= RExC_whilem_seen);
5227 nxt->flags = (U8)(data->whilem_c
5228 | (RExC_whilem_seen << 4)); /* On WHILEM */
5231 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
5233 if (flags & SCF_DO_SUBSTR) {
5234 SV *last_str = NULL;
5235 STRLEN last_chrs = 0;
5236 int counted = mincount != 0;
5238 if (data->last_end > 0 && mincount != 0) { /* Ends with a
5240 SSize_t b = pos_before >= data->last_start_min
5241 ? pos_before : data->last_start_min;
5243 const char * const s = SvPV_const(data->last_found, l);
5244 SSize_t old = b - data->last_start_min;
5247 old = utf8_hop((U8*)s, old) - (U8*)s;
5249 /* Get the added string: */
5250 last_str = newSVpvn_utf8(s + old, l, UTF);
5251 last_chrs = UTF ? utf8_length((U8*)(s + old),
5252 (U8*)(s + old + l)) : l;
5253 if (deltanext == 0 && pos_before == b) {
5254 /* What was added is a constant string */
5257 SvGROW(last_str, (mincount * l) + 1);
5258 repeatcpy(SvPVX(last_str) + l,
5259 SvPVX_const(last_str), l,
5261 SvCUR_set(last_str, SvCUR(last_str) * mincount);
5262 /* Add additional parts. */
5263 SvCUR_set(data->last_found,
5264 SvCUR(data->last_found) - l);
5265 sv_catsv(data->last_found, last_str);
5267 SV * sv = data->last_found;
5269 SvUTF8(sv) && SvMAGICAL(sv) ?
5270 mg_find(sv, PERL_MAGIC_utf8) : NULL;
5271 if (mg && mg->mg_len >= 0)
5272 mg->mg_len += last_chrs * (mincount-1);
5274 last_chrs *= mincount;
5275 data->last_end += l * (mincount - 1);
5278 /* start offset must point into the last copy */
5279 data->last_start_min += minnext * (mincount - 1);
5280 data->last_start_max =
5283 : data->last_start_max +
5284 (maxcount - 1) * (minnext + data->pos_delta);
5287 /* It is counted once already... */
5288 data->pos_min += minnext * (mincount - counted);
5290 Perl_re_printf( aTHX_ "counted=%" UVuf " deltanext=%" UVuf
5291 " SSize_t_MAX=%" UVuf " minnext=%" UVuf
5292 " maxcount=%" UVuf " mincount=%" UVuf "\n",
5293 (UV)counted, (UV)deltanext, (UV)SSize_t_MAX, (UV)minnext, (UV)maxcount,
5295 if (deltanext != SSize_t_MAX)
5296 Perl_re_printf( aTHX_ "LHS=%" UVuf " RHS=%" UVuf "\n",
5297 (UV)(-counted * deltanext + (minnext + deltanext) * maxcount
5298 - minnext * mincount), (UV)(SSize_t_MAX - data->pos_delta));
5300 if (deltanext == SSize_t_MAX
5301 || -counted * deltanext + (minnext + deltanext) * maxcount - minnext * mincount >= SSize_t_MAX - data->pos_delta)
5302 data->pos_delta = SSize_t_MAX;
5304 data->pos_delta += - counted * deltanext +
5305 (minnext + deltanext) * maxcount - minnext * mincount;
5306 if (mincount != maxcount) {
5307 /* Cannot extend fixed substrings found inside
5309 scan_commit(pRExC_state, data, minlenp, is_inf);
5310 if (mincount && last_str) {
5311 SV * const sv = data->last_found;
5312 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
5313 mg_find(sv, PERL_MAGIC_utf8) : NULL;
5317 sv_setsv(sv, last_str);
5318 data->last_end = data->pos_min;
5319 data->last_start_min = data->pos_min - last_chrs;
5320 data->last_start_max = is_inf
5322 : data->pos_min + data->pos_delta - last_chrs;
5324 data->longest = &(data->longest_float);
5326 SvREFCNT_dec(last_str);
5328 if (data && (fl & SF_HAS_EVAL))
5329 data->flags |= SF_HAS_EVAL;
5330 optimize_curly_tail:
5331 if (OP(oscan) != CURLYX) {
5332 while (PL_regkind[OP(next = regnext(oscan))] == NOTHING
5334 NEXT_OFF(oscan) += NEXT_OFF(next);
5340 Perl_croak(aTHX_ "panic: unexpected varying REx opcode %d",
5345 if (flags & SCF_DO_SUBSTR) {
5346 /* Cannot expect anything... */
5347 scan_commit(pRExC_state, data, minlenp, is_inf);
5348 data->longest = &(data->longest_float);
5350 is_inf = is_inf_internal = 1;
5351 if (flags & SCF_DO_STCLASS_OR) {
5352 if (OP(scan) == CLUMP) {
5353 /* Actually is any start char, but very few code points
5354 * aren't start characters */
5355 ssc_match_all_cp(data->start_class);
5358 ssc_anything(data->start_class);
5361 flags &= ~SCF_DO_STCLASS;
5365 else if (OP(scan) == LNBREAK) {
5366 if (flags & SCF_DO_STCLASS) {
5367 if (flags & SCF_DO_STCLASS_AND) {
5368 ssc_intersection(data->start_class,
5369 PL_XPosix_ptrs[_CC_VERTSPACE], FALSE);
5370 ssc_clear_locale(data->start_class);
5371 ANYOF_FLAGS(data->start_class)
5372 &= ~SSC_MATCHES_EMPTY_STRING;
5374 else if (flags & SCF_DO_STCLASS_OR) {
5375 ssc_union(data->start_class,
5376 PL_XPosix_ptrs[_CC_VERTSPACE],
5378 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
5380 /* See commit msg for
5381 * 749e076fceedeb708a624933726e7989f2302f6a */
5382 ANYOF_FLAGS(data->start_class)
5383 &= ~SSC_MATCHES_EMPTY_STRING;
5385 flags &= ~SCF_DO_STCLASS;
5388 if (delta != SSize_t_MAX)
5389 delta++; /* Because of the 2 char string cr-lf */
5390 if (flags & SCF_DO_SUBSTR) {
5391 /* Cannot expect anything... */
5392 scan_commit(pRExC_state, data, minlenp, is_inf);
5394 data->pos_delta += 1;
5395 data->longest = &(data->longest_float);
5398 else if (REGNODE_SIMPLE(OP(scan))) {
5400 if (flags & SCF_DO_SUBSTR) {
5401 scan_commit(pRExC_state, data, minlenp, is_inf);
5405 if (flags & SCF_DO_STCLASS) {
5407 SV* my_invlist = NULL;
5410 /* See commit msg 749e076fceedeb708a624933726e7989f2302f6a */
5411 ANYOF_FLAGS(data->start_class) &= ~SSC_MATCHES_EMPTY_STRING;
5413 /* Some of the logic below assumes that switching
5414 locale on will only add false positives. */
5419 Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d",
5423 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
5424 ssc_match_all_cp(data->start_class);
5429 SV* REG_ANY_invlist = _new_invlist(2);
5430 REG_ANY_invlist = add_cp_to_invlist(REG_ANY_invlist,
5432 if (flags & SCF_DO_STCLASS_OR) {
5433 ssc_union(data->start_class,
5435 TRUE /* TRUE => invert, hence all but \n
5439 else if (flags & SCF_DO_STCLASS_AND) {
5440 ssc_intersection(data->start_class,
5442 TRUE /* TRUE => invert */
5444 ssc_clear_locale(data->start_class);
5446 SvREFCNT_dec_NN(REG_ANY_invlist);
5453 if (flags & SCF_DO_STCLASS_AND)
5454 ssc_and(pRExC_state, data->start_class,
5455 (regnode_charclass *) scan);
5457 ssc_or(pRExC_state, data->start_class,
5458 (regnode_charclass *) scan);
5466 namedclass = classnum_to_namedclass(FLAGS(scan)) + invert;
5467 if (flags & SCF_DO_STCLASS_AND) {
5468 bool was_there = cBOOL(
5469 ANYOF_POSIXL_TEST(data->start_class,
5471 ANYOF_POSIXL_ZERO(data->start_class);
5472 if (was_there) { /* Do an AND */
5473 ANYOF_POSIXL_SET(data->start_class, namedclass);
5475 /* No individual code points can now match */
5476 data->start_class->invlist
5477 = sv_2mortal(_new_invlist(0));
5480 int complement = namedclass + ((invert) ? -1 : 1);
5482 assert(flags & SCF_DO_STCLASS_OR);
5484 /* If the complement of this class was already there,
5485 * the result is that they match all code points,
5486 * (\d + \D == everything). Remove the classes from
5487 * future consideration. Locale is not relevant in
5489 if (ANYOF_POSIXL_TEST(data->start_class, complement)) {
5490 ssc_match_all_cp(data->start_class);
5491 ANYOF_POSIXL_CLEAR(data->start_class, namedclass);
5492 ANYOF_POSIXL_CLEAR(data->start_class, complement);
5494 else { /* The usual case; just add this class to the
5496 ANYOF_POSIXL_SET(data->start_class, namedclass);
5501 case NPOSIXA: /* For these, we always know the exact set of
5506 if (FLAGS(scan) == _CC_ASCII) {
5507 my_invlist = invlist_clone(PL_XPosix_ptrs[_CC_ASCII]);
5510 _invlist_intersection(PL_XPosix_ptrs[FLAGS(scan)],
5511 PL_XPosix_ptrs[_CC_ASCII],
5522 my_invlist = invlist_clone(PL_XPosix_ptrs[FLAGS(scan)]);
5524 /* NPOSIXD matches all upper Latin1 code points unless the
5525 * target string being matched is UTF-8, which is
5526 * unknowable until match time. Since we are going to
5527 * invert, we want to get rid of all of them so that the
5528 * inversion will match all */
5529 if (OP(scan) == NPOSIXD) {
5530 _invlist_subtract(my_invlist, PL_UpperLatin1,
5536 if (flags & SCF_DO_STCLASS_AND) {
5537 ssc_intersection(data->start_class, my_invlist, invert);
5538 ssc_clear_locale(data->start_class);
5541 assert(flags & SCF_DO_STCLASS_OR);
5542 ssc_union(data->start_class, my_invlist, invert);
5544 SvREFCNT_dec(my_invlist);
5546 if (flags & SCF_DO_STCLASS_OR)
5547 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
5548 flags &= ~SCF_DO_STCLASS;
5551 else if (PL_regkind[OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
5552 data->flags |= (OP(scan) == MEOL
5555 scan_commit(pRExC_state, data, minlenp, is_inf);
5558 else if ( PL_regkind[OP(scan)] == BRANCHJ
5559 /* Lookbehind, or need to calculate parens/evals/stclass: */
5560 && (scan->flags || data || (flags & SCF_DO_STCLASS))
5561 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM))
5563 if ( !PERL_ENABLE_POSITIVE_ASSERTION_STUDY
5564 || OP(scan) == UNLESSM )
5566 /* Negative Lookahead/lookbehind
5567 In this case we can't do fixed string optimisation.
5570 SSize_t deltanext, minnext, fake = 0;
5575 StructCopy(&zero_scan_data, &data_fake, scan_data_t);
5577 data_fake.whilem_c = data->whilem_c;
5578 data_fake.last_closep = data->last_closep;
5581 data_fake.last_closep = &fake;
5582 data_fake.pos_delta = delta;
5583 if ( flags & SCF_DO_STCLASS && !scan->flags
5584 && OP(scan) == IFMATCH ) { /* Lookahead */
5585 ssc_init(pRExC_state, &intrnl);
5586 data_fake.start_class = &intrnl;
5587 f |= SCF_DO_STCLASS_AND;
5589 if (flags & SCF_WHILEM_VISITED_POS)
5590 f |= SCF_WHILEM_VISITED_POS;
5591 next = regnext(scan);
5592 nscan = NEXTOPER(NEXTOPER(scan));
5593 minnext = study_chunk(pRExC_state, &nscan, minlenp, &deltanext,
5594 last, &data_fake, stopparen,
5595 recursed_depth, NULL, f, depth+1);
5598 FAIL("Variable length lookbehind not implemented");
5600 else if (minnext > (I32)U8_MAX) {
5601 FAIL2("Lookbehind longer than %" UVuf " not implemented",
5604 scan->flags = (U8)minnext;
5607 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
5609 if (data_fake.flags & SF_HAS_EVAL)
5610 data->flags |= SF_HAS_EVAL;
5611 data->whilem_c = data_fake.whilem_c;
5613 if (f & SCF_DO_STCLASS_AND) {
5614 if (flags & SCF_DO_STCLASS_OR) {
5615 /* OR before, AND after: ideally we would recurse with
5616 * data_fake to get the AND applied by study of the
5617 * remainder of the pattern, and then derecurse;
5618 * *** HACK *** for now just treat as "no information".
5619 * See [perl #56690].
5621 ssc_init(pRExC_state, data->start_class);
5623 /* AND before and after: combine and continue. These
5624 * assertions are zero-length, so can match an EMPTY
5626 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) &intrnl);
5627 ANYOF_FLAGS(data->start_class)
5628 |= SSC_MATCHES_EMPTY_STRING;
5632 #if PERL_ENABLE_POSITIVE_ASSERTION_STUDY
5634 /* Positive Lookahead/lookbehind
5635 In this case we can do fixed string optimisation,
5636 but we must be careful about it. Note in the case of
5637 lookbehind the positions will be offset by the minimum
5638 length of the pattern, something we won't know about
5639 until after the recurse.
5641 SSize_t deltanext, fake = 0;
5645 /* We use SAVEFREEPV so that when the full compile
5646 is finished perl will clean up the allocated
5647 minlens when it's all done. This way we don't
5648 have to worry about freeing them when we know
5649 they wont be used, which would be a pain.
5652 Newx( minnextp, 1, SSize_t );
5653 SAVEFREEPV(minnextp);
5656 StructCopy(data, &data_fake, scan_data_t);
5657 if ((flags & SCF_DO_SUBSTR) && data->last_found) {
5660 scan_commit(pRExC_state, &data_fake, minlenp, is_inf);
5661 data_fake.last_found=newSVsv(data->last_found);
5665 data_fake.last_closep = &fake;
5666 data_fake.flags = 0;
5667 data_fake.pos_delta = delta;
5669 data_fake.flags |= SF_IS_INF;
5670 if ( flags & SCF_DO_STCLASS && !scan->flags
5671 && OP(scan) == IFMATCH ) { /* Lookahead */
5672 ssc_init(pRExC_state, &intrnl);
5673 data_fake.start_class = &intrnl;
5674 f |= SCF_DO_STCLASS_AND;
5676 if (flags & SCF_WHILEM_VISITED_POS)
5677 f |= SCF_WHILEM_VISITED_POS;
5678 next = regnext(scan);
5679 nscan = NEXTOPER(NEXTOPER(scan));
5681 *minnextp = study_chunk(pRExC_state, &nscan, minnextp,
5682 &deltanext, last, &data_fake,
5683 stopparen, recursed_depth, NULL,
5687 FAIL("Variable length lookbehind not implemented");
5689 else if (*minnextp > (I32)U8_MAX) {
5690 FAIL2("Lookbehind longer than %" UVuf " not implemented",
5693 scan->flags = (U8)*minnextp;
5698 if (f & SCF_DO_STCLASS_AND) {
5699 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) &intrnl);
5700 ANYOF_FLAGS(data->start_class) |= SSC_MATCHES_EMPTY_STRING;
5703 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
5705 if (data_fake.flags & SF_HAS_EVAL)
5706 data->flags |= SF_HAS_EVAL;
5707 data->whilem_c = data_fake.whilem_c;
5708 if ((flags & SCF_DO_SUBSTR) && data_fake.last_found) {
5709 if (RExC_rx->minlen<*minnextp)
5710 RExC_rx->minlen=*minnextp;
5711 scan_commit(pRExC_state, &data_fake, minnextp, is_inf);
5712 SvREFCNT_dec_NN(data_fake.last_found);
5714 if ( data_fake.minlen_fixed != minlenp )
5716 data->offset_fixed= data_fake.offset_fixed;
5717 data->minlen_fixed= data_fake.minlen_fixed;
5718 data->lookbehind_fixed+= scan->flags;
5720 if ( data_fake.minlen_float != minlenp )
5722 data->minlen_float= data_fake.minlen_float;
5723 data->offset_float_min=data_fake.offset_float_min;
5724 data->offset_float_max=data_fake.offset_float_max;
5725 data->lookbehind_float+= scan->flags;
5732 else if (OP(scan) == OPEN) {
5733 if (stopparen != (I32)ARG(scan))
5736 else if (OP(scan) == CLOSE) {
5737 if (stopparen == (I32)ARG(scan)) {
5740 if ((I32)ARG(scan) == is_par) {
5741 next = regnext(scan);
5743 if ( next && (OP(next) != WHILEM) && next < last)
5744 is_par = 0; /* Disable optimization */
5747 *(data->last_closep) = ARG(scan);
5749 else if (OP(scan) == EVAL) {
5751 data->flags |= SF_HAS_EVAL;
5753 else if ( PL_regkind[OP(scan)] == ENDLIKE ) {
5754 if (flags & SCF_DO_SUBSTR) {
5755 scan_commit(pRExC_state, data, minlenp, is_inf);
5756 flags &= ~SCF_DO_SUBSTR;
5758 if (data && OP(scan)==ACCEPT) {
5759 data->flags |= SCF_SEEN_ACCEPT;
5764 else if (OP(scan) == LOGICAL && scan->flags == 2) /* Embedded follows */
5766 if (flags & SCF_DO_SUBSTR) {
5767 scan_commit(pRExC_state, data, minlenp, is_inf);
5768 data->longest = &(data->longest_float);
5770 is_inf = is_inf_internal = 1;
5771 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
5772 ssc_anything(data->start_class);
5773 flags &= ~SCF_DO_STCLASS;
5775 else if (OP(scan) == GPOS) {
5776 if (!(RExC_rx->intflags & PREGf_GPOS_FLOAT) &&
5777 !(delta || is_inf || (data && data->pos_delta)))
5779 if (!(RExC_rx->intflags & PREGf_ANCH) && (flags & SCF_DO_SUBSTR))
5780 RExC_rx->intflags |= PREGf_ANCH_GPOS;
5781 if (RExC_rx->gofs < (STRLEN)min)
5782 RExC_rx->gofs = min;
5784 RExC_rx->intflags |= PREGf_GPOS_FLOAT;
5788 #ifdef TRIE_STUDY_OPT
5789 #ifdef FULL_TRIE_STUDY
5790 else if (PL_regkind[OP(scan)] == TRIE) {
5791 /* NOTE - There is similar code to this block above for handling
5792 BRANCH nodes on the initial study. If you change stuff here
5794 regnode *trie_node= scan;
5795 regnode *tail= regnext(scan);
5796 reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
5797 SSize_t max1 = 0, min1 = SSize_t_MAX;
5800 if (flags & SCF_DO_SUBSTR) { /* XXXX Add !SUSPEND? */
5801 /* Cannot merge strings after this. */
5802 scan_commit(pRExC_state, data, minlenp, is_inf);
5804 if (flags & SCF_DO_STCLASS)
5805 ssc_init_zero(pRExC_state, &accum);
5811 const regnode *nextbranch= NULL;
5814 for ( word=1 ; word <= trie->wordcount ; word++)
5816 SSize_t deltanext=0, minnext=0, f = 0, fake;
5817 regnode_ssc this_class;
5819 StructCopy(&zero_scan_data, &data_fake, scan_data_t);
5821 data_fake.whilem_c = data->whilem_c;
5822 data_fake.last_closep = data->last_closep;
5825 data_fake.last_closep = &fake;
5826 data_fake.pos_delta = delta;
5827 if (flags & SCF_DO_STCLASS) {
5828 ssc_init(pRExC_state, &this_class);
5829 data_fake.start_class = &this_class;
5830 f = SCF_DO_STCLASS_AND;
5832 if (flags & SCF_WHILEM_VISITED_POS)
5833 f |= SCF_WHILEM_VISITED_POS;
5835 if (trie->jump[word]) {
5837 nextbranch = trie_node + trie->jump[0];
5838 scan= trie_node + trie->jump[word];
5839 /* We go from the jump point to the branch that follows
5840 it. Note this means we need the vestigal unused
5841 branches even though they arent otherwise used. */
5842 minnext = study_chunk(pRExC_state, &scan, minlenp,
5843 &deltanext, (regnode *)nextbranch, &data_fake,
5844 stopparen, recursed_depth, NULL, f,depth+1);
5846 if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
5847 nextbranch= regnext((regnode*)nextbranch);
5849 if (min1 > (SSize_t)(minnext + trie->minlen))
5850 min1 = minnext + trie->minlen;
5851 if (deltanext == SSize_t_MAX) {
5852 is_inf = is_inf_internal = 1;
5854 } else if (max1 < (SSize_t)(minnext + deltanext + trie->maxlen))
5855 max1 = minnext + deltanext + trie->maxlen;
5857 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
5859 if (data_fake.flags & SCF_SEEN_ACCEPT) {
5860 if ( stopmin > min + min1)
5861 stopmin = min + min1;
5862 flags &= ~SCF_DO_SUBSTR;
5864 data->flags |= SCF_SEEN_ACCEPT;
5867 if (data_fake.flags & SF_HAS_EVAL)
5868 data->flags |= SF_HAS_EVAL;
5869 data->whilem_c = data_fake.whilem_c;
5871 if (flags & SCF_DO_STCLASS)
5872 ssc_or(pRExC_state, &accum, (regnode_charclass *) &this_class);
5875 if (flags & SCF_DO_SUBSTR) {
5876 data->pos_min += min1;
5877 data->pos_delta += max1 - min1;
5878 if (max1 != min1 || is_inf)
5879 data->longest = &(data->longest_float);
5882 if (delta != SSize_t_MAX)
5883 delta += max1 - min1;
5884 if (flags & SCF_DO_STCLASS_OR) {
5885 ssc_or(pRExC_state, data->start_class, (regnode_charclass *) &accum);
5887 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
5888 flags &= ~SCF_DO_STCLASS;
5891 else if (flags & SCF_DO_STCLASS_AND) {
5893 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) &accum);
5894 flags &= ~SCF_DO_STCLASS;
5897 /* Switch to OR mode: cache the old value of
5898 * data->start_class */
5900 StructCopy(data->start_class, and_withp, regnode_ssc);
5901 flags &= ~SCF_DO_STCLASS_AND;
5902 StructCopy(&accum, data->start_class, regnode_ssc);
5903 flags |= SCF_DO_STCLASS_OR;
5910 else if (PL_regkind[OP(scan)] == TRIE) {
5911 reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
5914 min += trie->minlen;
5915 delta += (trie->maxlen - trie->minlen);
5916 flags &= ~SCF_DO_STCLASS; /* xxx */
5917 if (flags & SCF_DO_SUBSTR) {
5918 /* Cannot expect anything... */
5919 scan_commit(pRExC_state, data, minlenp, is_inf);
5920 data->pos_min += trie->minlen;
5921 data->pos_delta += (trie->maxlen - trie->minlen);
5922 if (trie->maxlen != trie->minlen)
5923 data->longest = &(data->longest_float);
5925 if (trie->jump) /* no more substrings -- for now /grr*/
5926 flags &= ~SCF_DO_SUBSTR;
5928 #endif /* old or new */
5929 #endif /* TRIE_STUDY_OPT */
5931 /* Else: zero-length, ignore. */
5932 scan = regnext(scan);
5937 /* we need to unwind recursion. */
5940 DEBUG_STUDYDATA("frame-end:",data,depth);
5941 DEBUG_PEEP("fend", scan, depth);
5943 /* restore previous context */
5944 last = frame->last_regnode;
5945 scan = frame->next_regnode;
5946 stopparen = frame->stopparen;
5947 recursed_depth = frame->prev_recursed_depth;
5949 RExC_frame_last = frame->prev_frame;
5950 frame = frame->this_prev_frame;
5951 goto fake_study_recurse;
5955 DEBUG_STUDYDATA("pre-fin:",data,depth);
5958 *deltap = is_inf_internal ? SSize_t_MAX : delta;
5960 if (flags & SCF_DO_SUBSTR && is_inf)
5961 data->pos_delta = SSize_t_MAX - data->pos_min;
5962 if (is_par > (I32)U8_MAX)
5964 if (is_par && pars==1 && data) {
5965 data->flags |= SF_IN_PAR;
5966 data->flags &= ~SF_HAS_PAR;
5968 else if (pars && data) {
5969 data->flags |= SF_HAS_PAR;
5970 data->flags &= ~SF_IN_PAR;
5972 if (flags & SCF_DO_STCLASS_OR)
5973 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
5974 if (flags & SCF_TRIE_RESTUDY)
5975 data->flags |= SCF_TRIE_RESTUDY;
5977 DEBUG_STUDYDATA("post-fin:",data,depth);
5980 SSize_t final_minlen= min < stopmin ? min : stopmin;
5982 if (!(RExC_seen & REG_UNBOUNDED_QUANTIFIER_SEEN)) {
5983 if (final_minlen > SSize_t_MAX - delta)
5984 RExC_maxlen = SSize_t_MAX;
5985 else if (RExC_maxlen < final_minlen + delta)
5986 RExC_maxlen = final_minlen + delta;
5988 return final_minlen;
5990 NOT_REACHED; /* NOTREACHED */
5994 S_add_data(RExC_state_t* const pRExC_state, const char* const s, const U32 n)
5996 U32 count = RExC_rxi->data ? RExC_rxi->data->count : 0;
5998 PERL_ARGS_ASSERT_ADD_DATA;
6000 Renewc(RExC_rxi->data,
6001 sizeof(*RExC_rxi->data) + sizeof(void*) * (count + n - 1),
6002 char, struct reg_data);
6004 Renew(RExC_rxi->data->what, count + n, U8);
6006 Newx(RExC_rxi->data->what, n, U8);
6007 RExC_rxi->data->count = count + n;
6008 Copy(s, RExC_rxi->data->what + count, n, U8);
6012 /*XXX: todo make this not included in a non debugging perl, but appears to be
6013 * used anyway there, in 'use re' */
6014 #ifndef PERL_IN_XSUB_RE
6016 Perl_reginitcolors(pTHX)
6018 const char * const s = PerlEnv_getenv("PERL_RE_COLORS");
6020 char *t = savepv(s);
6024 t = strchr(t, '\t');
6030 PL_colors[i] = t = (char *)"";
6035 PL_colors[i++] = (char *)"";
6042 #ifdef TRIE_STUDY_OPT
6043 #define CHECK_RESTUDY_GOTO_butfirst(dOsomething) \
6046 (data.flags & SCF_TRIE_RESTUDY) \
6054 #define CHECK_RESTUDY_GOTO_butfirst
6058 * pregcomp - compile a regular expression into internal code
6060 * Decides which engine's compiler to call based on the hint currently in
6064 #ifndef PERL_IN_XSUB_RE
6066 /* return the currently in-scope regex engine (or the default if none) */
6068 regexp_engine const *
6069 Perl_current_re_engine(pTHX)
6071 if (IN_PERL_COMPILETIME) {
6072 HV * const table = GvHV(PL_hintgv);
6075 if (!table || !(PL_hints & HINT_LOCALIZE_HH))
6076 return &PL_core_reg_engine;
6077 ptr = hv_fetchs(table, "regcomp", FALSE);
6078 if ( !(ptr && SvIOK(*ptr) && SvIV(*ptr)))
6079 return &PL_core_reg_engine;
6080 return INT2PTR(regexp_engine*,SvIV(*ptr));
6084 if (!PL_curcop->cop_hints_hash)
6085 return &PL_core_reg_engine;
6086 ptr = cop_hints_fetch_pvs(PL_curcop, "regcomp", 0);
6087 if ( !(ptr && SvIOK(ptr) && SvIV(ptr)))
6088 return &PL_core_reg_engine;
6089 return INT2PTR(regexp_engine*,SvIV(ptr));
6095 Perl_pregcomp(pTHX_ SV * const pattern, const U32 flags)
6097 regexp_engine const *eng = current_re_engine();
6098 GET_RE_DEBUG_FLAGS_DECL;
6100 PERL_ARGS_ASSERT_PREGCOMP;
6102 /* Dispatch a request to compile a regexp to correct regexp engine. */
6104 Perl_re_printf( aTHX_ "Using engine %" UVxf "\n",
6107 return CALLREGCOMP_ENG(eng, pattern, flags);
6111 /* public(ish) entry point for the perl core's own regex compiling code.
6112 * It's actually a wrapper for Perl_re_op_compile that only takes an SV
6113 * pattern rather than a list of OPs, and uses the internal engine rather
6114 * than the current one */
6117 Perl_re_compile(pTHX_ SV * const pattern, U32 rx_flags)
6119 SV *pat = pattern; /* defeat constness! */
6120 PERL_ARGS_ASSERT_RE_COMPILE;
6121 return Perl_re_op_compile(aTHX_ &pat, 1, NULL,
6122 #ifdef PERL_IN_XSUB_RE
6125 &PL_core_reg_engine,
6127 NULL, NULL, rx_flags, 0);
6132 S_free_codeblocks(pTHX_ struct reg_code_blocks *cbs)
6136 if (--cbs->refcnt > 0)
6138 for (n = 0; n < cbs->count; n++) {
6139 REGEXP *rx = cbs->cb[n].src_regex;
6140 cbs->cb[n].src_regex = NULL;
6148 static struct reg_code_blocks *
6149 S_alloc_code_blocks(pTHX_ int ncode)
6151 struct reg_code_blocks *cbs;
6152 Newx(cbs, 1, struct reg_code_blocks);
6155 SAVEDESTRUCTOR_X(S_free_codeblocks, cbs);
6157 Newx(cbs->cb, ncode, struct reg_code_block);
6164 /* upgrade pattern pat_p of length plen_p to UTF8, and if there are code
6165 * blocks, recalculate the indices. Update pat_p and plen_p in-place to
6166 * point to the realloced string and length.
6168 * This is essentially a copy of Perl_bytes_to_utf8() with the code index
6172 S_pat_upgrade_to_utf8(pTHX_ RExC_state_t * const pRExC_state,
6173 char **pat_p, STRLEN *plen_p, int num_code_blocks)
6175 U8 *const src = (U8*)*pat_p;
6180 GET_RE_DEBUG_FLAGS_DECL;
6182 DEBUG_PARSE_r(Perl_re_printf( aTHX_
6183 "UTF8 mismatch! Converting to utf8 for resizing and compile\n"));
6185 Newx(dst, *plen_p * 2 + 1, U8);
6188 while (s < *plen_p) {
6189 append_utf8_from_native_byte(src[s], &d);
6191 if (n < num_code_blocks) {
6192 assert(pRExC_state->code_blocks);
6193 if (!do_end && pRExC_state->code_blocks->cb[n].start == s) {
6194 pRExC_state->code_blocks->cb[n].start = d - dst - 1;
6195 assert(*(d - 1) == '(');
6198 else if (do_end && pRExC_state->code_blocks->cb[n].end == s) {
6199 pRExC_state->code_blocks->cb[n].end = d - dst - 1;
6200 assert(*(d - 1) == ')');
6209 *pat_p = (char*) dst;
6211 RExC_orig_utf8 = RExC_utf8 = 1;
6216 /* S_concat_pat(): concatenate a list of args to the pattern string pat,
6217 * while recording any code block indices, and handling overloading,
6218 * nested qr// objects etc. If pat is null, it will allocate a new
6219 * string, or just return the first arg, if there's only one.
6221 * Returns the malloced/updated pat.
6222 * patternp and pat_count is the array of SVs to be concatted;
6223 * oplist is the optional list of ops that generated the SVs;
6224 * recompile_p is a pointer to a boolean that will be set if
6225 * the regex will need to be recompiled.
6226 * delim, if non-null is an SV that will be inserted between each element
6230 S_concat_pat(pTHX_ RExC_state_t * const pRExC_state,
6231 SV *pat, SV ** const patternp, int pat_count,
6232 OP *oplist, bool *recompile_p, SV *delim)
6236 bool use_delim = FALSE;
6237 bool alloced = FALSE;
6239 /* if we know we have at least two args, create an empty string,
6240 * then concatenate args to that. For no args, return an empty string */
6241 if (!pat && pat_count != 1) {
6247 for (svp = patternp; svp < patternp + pat_count; svp++) {
6250 STRLEN orig_patlen = 0;
6252 SV *msv = use_delim ? delim : *svp;
6253 if (!msv) msv = &PL_sv_undef;
6255 /* if we've got a delimiter, we go round the loop twice for each
6256 * svp slot (except the last), using the delimiter the second
6265 if (SvTYPE(msv) == SVt_PVAV) {
6266 /* we've encountered an interpolated array within
6267 * the pattern, e.g. /...@a..../. Expand the list of elements,
6268 * then recursively append elements.
6269 * The code in this block is based on S_pushav() */
6271 AV *const av = (AV*)msv;
6272 const SSize_t maxarg = AvFILL(av) + 1;
6276 assert(oplist->op_type == OP_PADAV
6277 || oplist->op_type == OP_RV2AV);
6278 oplist = OpSIBLING(oplist);
6281 if (SvRMAGICAL(av)) {
6284 Newx(array, maxarg, SV*);
6286 for (i=0; i < maxarg; i++) {
6287 SV ** const svp = av_fetch(av, i, FALSE);
6288 array[i] = svp ? *svp : &PL_sv_undef;
6292 array = AvARRAY(av);
6294 pat = S_concat_pat(aTHX_ pRExC_state, pat,
6295 array, maxarg, NULL, recompile_p,
6297 GvSV((gv_fetchpvs("\"", GV_ADDMULTI, SVt_PV))));
6303 /* we make the assumption here that each op in the list of
6304 * op_siblings maps to one SV pushed onto the stack,
6305 * except for code blocks, with have both an OP_NULL and
6307 * This allows us to match up the list of SVs against the
6308 * list of OPs to find the next code block.
6310 * Note that PUSHMARK PADSV PADSV ..
6312 * PADRANGE PADSV PADSV ..
6313 * so the alignment still works. */
6316 if (oplist->op_type == OP_NULL
6317 && (oplist->op_flags & OPf_SPECIAL))
6319 assert(n < pRExC_state->code_blocks->count);
6320 pRExC_state->code_blocks->cb[n].start = pat ? SvCUR(pat) : 0;
6321 pRExC_state->code_blocks->cb[n].block = oplist;
6322 pRExC_state->code_blocks->cb[n].src_regex = NULL;
6325 oplist = OpSIBLING(oplist); /* skip CONST */
6328 oplist = OpSIBLING(oplist);;
6331 /* apply magic and QR overloading to arg */
6334 if (SvROK(msv) && SvAMAGIC(msv)) {
6335 SV *sv = AMG_CALLunary(msv, regexp_amg);
6339 if (SvTYPE(sv) != SVt_REGEXP)
6340 Perl_croak(aTHX_ "Overloaded qr did not return a REGEXP");
6345 /* try concatenation overload ... */
6346 if (pat && (SvAMAGIC(pat) || SvAMAGIC(msv)) &&
6347 (sv = amagic_call(pat, msv, concat_amg, AMGf_assign)))
6350 /* overloading involved: all bets are off over literal
6351 * code. Pretend we haven't seen it */
6353 pRExC_state->code_blocks->count -= n;
6357 /* ... or failing that, try "" overload */
6358 while (SvAMAGIC(msv)
6359 && (sv = AMG_CALLunary(msv, string_amg))
6363 && SvRV(msv) == SvRV(sv))
6368 if (SvROK(msv) && SvTYPE(SvRV(msv)) == SVt_REGEXP)
6372 /* this is a partially unrolled
6373 * sv_catsv_nomg(pat, msv);
6374 * that allows us to adjust code block indices if
6377 char *dst = SvPV_force_nomg(pat, dlen);
6379 if (SvUTF8(msv) && !SvUTF8(pat)) {
6380 S_pat_upgrade_to_utf8(aTHX_ pRExC_state, &dst, &dlen, n);
6381 sv_setpvn(pat, dst, dlen);
6384 sv_catsv_nomg(pat, msv);
6388 /* We have only one SV to process, but we need to verify
6389 * it is properly null terminated or we will fail asserts
6390 * later. In theory we probably shouldn't get such SV's,
6391 * but if we do we should handle it gracefully. */
6392 if ( SvTYPE(msv) != SVt_PV || (SvLEN(msv) > SvCUR(msv) && *(SvEND(msv)) == 0) ) {
6393 /* not a string, or a string with a trailing null */
6396 /* a string with no trailing null, we need to copy it
6397 * so it we have a trailing null */
6403 pRExC_state->code_blocks->cb[n-1].end = SvCUR(pat)-1;
6406 /* extract any code blocks within any embedded qr//'s */
6407 if (rx && SvTYPE(rx) == SVt_REGEXP
6408 && RX_ENGINE((REGEXP*)rx)->op_comp)
6411 RXi_GET_DECL(ReANY((REGEXP *)rx), ri);
6412 if (ri->code_blocks && ri->code_blocks->count) {
6414 /* the presence of an embedded qr// with code means
6415 * we should always recompile: the text of the
6416 * qr// may not have changed, but it may be a
6417 * different closure than last time */
6419 if (pRExC_state->code_blocks) {
6420 int new_count = pRExC_state->code_blocks->count
6421 + ri->code_blocks->count;
6422 Renew(pRExC_state->code_blocks->cb,
6423 new_count, struct reg_code_block);
6424 pRExC_state->code_blocks->count = new_count;
6427 pRExC_state->code_blocks = S_alloc_code_blocks(aTHX_
6428 ri->code_blocks->count);
6430 for (i=0; i < ri->code_blocks->count; i++) {
6431 struct reg_code_block *src, *dst;
6432 STRLEN offset = orig_patlen
6433 + ReANY((REGEXP *)rx)->pre_prefix;
6434 assert(n < pRExC_state->code_blocks->count);
6435 src = &ri->code_blocks->cb[i];
6436 dst = &pRExC_state->code_blocks->cb[n];
6437 dst->start = src->start + offset;
6438 dst->end = src->end + offset;
6439 dst->block = src->block;
6440 dst->src_regex = (REGEXP*) SvREFCNT_inc( (SV*)
6449 /* avoid calling magic multiple times on a single element e.g. =~ $qr */
6458 /* see if there are any run-time code blocks in the pattern.
6459 * False positives are allowed */
6462 S_has_runtime_code(pTHX_ RExC_state_t * const pRExC_state,
6463 char *pat, STRLEN plen)
6468 PERL_UNUSED_CONTEXT;
6470 for (s = 0; s < plen; s++) {
6471 if ( pRExC_state->code_blocks
6472 && n < pRExC_state->code_blocks->count
6473 && s == pRExC_state->code_blocks->cb[n].start)
6475 s = pRExC_state->code_blocks->cb[n].end;
6479 /* TODO ideally should handle [..], (#..), /#.../x to reduce false
6481 if (pat[s] == '(' && s+2 <= plen && pat[s+1] == '?' &&
6483 || (s + 2 <= plen && pat[s+2] == '?' && pat[s+3] == '{'))
6490 /* Handle run-time code blocks. We will already have compiled any direct
6491 * or indirect literal code blocks. Now, take the pattern 'pat' and make a
6492 * copy of it, but with any literal code blocks blanked out and
6493 * appropriate chars escaped; then feed it into
6495 * eval "qr'modified_pattern'"
6499 * a\bc(?{"this was literal"})def'ghi\\jkl(?{"this is runtime"})mno
6503 * qr'a\\bc_______________________def\'ghi\\\\jkl(?{"this is runtime"})mno'
6505 * After eval_sv()-ing that, grab any new code blocks from the returned qr
6506 * and merge them with any code blocks of the original regexp.
6508 * If the pat is non-UTF8, while the evalled qr is UTF8, don't merge;
6509 * instead, just save the qr and return FALSE; this tells our caller that
6510 * the original pattern needs upgrading to utf8.
6514 S_compile_runtime_code(pTHX_ RExC_state_t * const pRExC_state,
6515 char *pat, STRLEN plen)
6519 GET_RE_DEBUG_FLAGS_DECL;
6521 if (pRExC_state->runtime_code_qr) {
6522 /* this is the second time we've been called; this should
6523 * only happen if the main pattern got upgraded to utf8
6524 * during compilation; re-use the qr we compiled first time
6525 * round (which should be utf8 too)
6527 qr = pRExC_state->runtime_code_qr;
6528 pRExC_state->runtime_code_qr = NULL;
6529 assert(RExC_utf8 && SvUTF8(qr));
6535 int newlen = plen + 7; /* allow for "qr''xx\0" extra chars */
6539 /* determine how many extra chars we need for ' and \ escaping */
6540 for (s = 0; s < plen; s++) {
6541 if (pat[s] == '\'' || pat[s] == '\\')
6545 Newx(newpat, newlen, char);
6547 *p++ = 'q'; *p++ = 'r'; *p++ = '\'';
6549 for (s = 0; s < plen; s++) {
6550 if ( pRExC_state->code_blocks
6551 && n < pRExC_state->code_blocks->count
6552 && s == pRExC_state->code_blocks->cb[n].start)
6554 /* blank out literal code block */
6555 assert(pat[s] == '(');
6556 while (s <= pRExC_state->code_blocks->cb[n].end) {
6564 if (pat[s] == '\'' || pat[s] == '\\')
6569 if (pRExC_state->pm_flags & RXf_PMf_EXTENDED) {
6571 if (pRExC_state->pm_flags & RXf_PMf_EXTENDED_MORE) {
6577 Perl_re_printf( aTHX_
6578 "%sre-parsing pattern for runtime code:%s %s\n",
6579 PL_colors[4],PL_colors[5],newpat);
6582 sv = newSVpvn_flags(newpat, p-newpat-1, RExC_utf8 ? SVf_UTF8 : 0);
6588 PUSHSTACKi(PERLSI_REQUIRE);
6589 /* G_RE_REPARSING causes the toker to collapse \\ into \ when
6590 * parsing qr''; normally only q'' does this. It also alters
6592 eval_sv(sv, G_SCALAR|G_RE_REPARSING);
6593 SvREFCNT_dec_NN(sv);
6598 SV * const errsv = ERRSV;
6599 if (SvTRUE_NN(errsv))
6600 /* use croak_sv ? */
6601 Perl_croak_nocontext("%" SVf, SVfARG(errsv));
6603 assert(SvROK(qr_ref));
6605 assert(SvTYPE(qr) == SVt_REGEXP && RX_ENGINE((REGEXP*)qr)->op_comp);
6606 /* the leaving below frees the tmp qr_ref.
6607 * Give qr a life of its own */
6615 if (!RExC_utf8 && SvUTF8(qr)) {
6616 /* first time through; the pattern got upgraded; save the
6617 * qr for the next time through */
6618 assert(!pRExC_state->runtime_code_qr);
6619 pRExC_state->runtime_code_qr = qr;
6624 /* extract any code blocks within the returned qr// */
6627 /* merge the main (r1) and run-time (r2) code blocks into one */
6629 RXi_GET_DECL(ReANY((REGEXP *)qr), r2);
6630 struct reg_code_block *new_block, *dst;
6631 RExC_state_t * const r1 = pRExC_state; /* convenient alias */
6635 if (!r2->code_blocks || !r2->code_blocks->count) /* we guessed wrong */
6637 SvREFCNT_dec_NN(qr);
6641 if (!r1->code_blocks)
6642 r1->code_blocks = S_alloc_code_blocks(aTHX_ 0);
6644 r1c = r1->code_blocks->count;
6645 r2c = r2->code_blocks->count;
6647 Newx(new_block, r1c + r2c, struct reg_code_block);
6651 while (i1 < r1c || i2 < r2c) {
6652 struct reg_code_block *src;
6656 src = &r2->code_blocks->cb[i2++];
6660 src = &r1->code_blocks->cb[i1++];
6661 else if ( r1->code_blocks->cb[i1].start
6662 < r2->code_blocks->cb[i2].start)
6664 src = &r1->code_blocks->cb[i1++];
6665 assert(src->end < r2->code_blocks->cb[i2].start);
6668 assert( r1->code_blocks->cb[i1].start
6669 > r2->code_blocks->cb[i2].start);
6670 src = &r2->code_blocks->cb[i2++];
6672 assert(src->end < r1->code_blocks->cb[i1].start);
6675 assert(pat[src->start] == '(');
6676 assert(pat[src->end] == ')');
6677 dst->start = src->start;
6678 dst->end = src->end;
6679 dst->block = src->block;
6680 dst->src_regex = is_qr ? (REGEXP*) SvREFCNT_inc( (SV*) qr)
6684 r1->code_blocks->count += r2c;
6685 Safefree(r1->code_blocks->cb);
6686 r1->code_blocks->cb = new_block;
6689 SvREFCNT_dec_NN(qr);
6695 S_setup_longest(pTHX_ RExC_state_t *pRExC_state, SV* sv_longest,
6696 SV** rx_utf8, SV** rx_substr, SSize_t* rx_end_shift,
6697 SSize_t lookbehind, SSize_t offset, SSize_t *minlen,
6698 STRLEN longest_length, bool eol, bool meol)
6700 /* This is the common code for setting up the floating and fixed length
6701 * string data extracted from Perl_re_op_compile() below. Returns a boolean
6702 * as to whether succeeded or not */
6707 if (! (longest_length
6708 || (eol /* Can't have SEOL and MULTI */
6709 && (! meol || (RExC_flags & RXf_PMf_MULTILINE)))
6711 /* See comments for join_exact for why REG_UNFOLDED_MULTI_SEEN */
6712 || (RExC_seen & REG_UNFOLDED_MULTI_SEEN))
6717 /* copy the information about the longest from the reg_scan_data
6718 over to the program. */
6719 if (SvUTF8(sv_longest)) {
6720 *rx_utf8 = sv_longest;
6723 *rx_substr = sv_longest;
6726 /* end_shift is how many chars that must be matched that
6727 follow this item. We calculate it ahead of time as once the
6728 lookbehind offset is added in we lose the ability to correctly
6730 ml = minlen ? *(minlen) : (SSize_t)longest_length;
6731 *rx_end_shift = ml - offset
6733 /* XXX SvTAIL is always false here - did you mean FBMcf_TAIL
6735 + (SvTAIL(sv_longest) != 0)
6739 t = (eol/* Can't have SEOL and MULTI */
6740 && (! meol || (RExC_flags & RXf_PMf_MULTILINE)));
6741 fbm_compile(sv_longest, t ? FBMcf_TAIL : 0);
6747 * Perl_re_op_compile - the perl internal RE engine's function to compile a
6748 * regular expression into internal code.
6749 * The pattern may be passed either as:
6750 * a list of SVs (patternp plus pat_count)
6751 * a list of OPs (expr)
6752 * If both are passed, the SV list is used, but the OP list indicates
6753 * which SVs are actually pre-compiled code blocks
6755 * The SVs in the list have magic and qr overloading applied to them (and
6756 * the list may be modified in-place with replacement SVs in the latter
6759 * If the pattern hasn't changed from old_re, then old_re will be
6762 * eng is the current engine. If that engine has an op_comp method, then
6763 * handle directly (i.e. we assume that op_comp was us); otherwise, just
6764 * do the initial concatenation of arguments and pass on to the external
6767 * If is_bare_re is not null, set it to a boolean indicating whether the
6768 * arg list reduced (after overloading) to a single bare regex which has
6769 * been returned (i.e. /$qr/).
6771 * orig_rx_flags contains RXf_* flags. See perlreapi.pod for more details.
6773 * pm_flags contains the PMf_* flags, typically based on those from the
6774 * pm_flags field of the related PMOP. Currently we're only interested in
6775 * PMf_HAS_CV, PMf_IS_QR, PMf_USE_RE_EVAL.
6777 * We can't allocate space until we know how big the compiled form will be,
6778 * but we can't compile it (and thus know how big it is) until we've got a
6779 * place to put the code. So we cheat: we compile it twice, once with code
6780 * generation turned off and size counting turned on, and once "for real".
6781 * This also means that we don't allocate space until we are sure that the
6782 * thing really will compile successfully, and we never have to move the
6783 * code and thus invalidate pointers into it. (Note that it has to be in
6784 * one piece because free() must be able to free it all.) [NB: not true in perl]
6786 * Beware that the optimization-preparation code in here knows about some
6787 * of the structure of the compiled regexp. [I'll say.]
6791 Perl_re_op_compile(pTHX_ SV ** const patternp, int pat_count,
6792 OP *expr, const regexp_engine* eng, REGEXP *old_re,
6793 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags)
6797 regexp_internal *ri;
6805 SV** new_patternp = patternp;
6807 /* these are all flags - maybe they should be turned
6808 * into a single int with different bit masks */
6809 I32 sawlookahead = 0;
6814 regex_charset initial_charset = get_regex_charset(orig_rx_flags);
6816 bool runtime_code = 0;
6818 RExC_state_t RExC_state;
6819 RExC_state_t * const pRExC_state = &RExC_state;
6820 #ifdef TRIE_STUDY_OPT
6822 RExC_state_t copyRExC_state;
6824 GET_RE_DEBUG_FLAGS_DECL;
6826 PERL_ARGS_ASSERT_RE_OP_COMPILE;
6828 DEBUG_r(if (!PL_colorset) reginitcolors());
6830 /* Initialize these here instead of as-needed, as is quick and avoids
6831 * having to test them each time otherwise */
6832 if (! PL_AboveLatin1) {
6834 char * dump_len_string;
6837 PL_AboveLatin1 = _new_invlist_C_array(AboveLatin1_invlist);
6838 PL_Latin1 = _new_invlist_C_array(Latin1_invlist);
6839 PL_UpperLatin1 = _new_invlist_C_array(UpperLatin1_invlist);
6840 PL_utf8_foldable = _new_invlist_C_array(_Perl_Any_Folds_invlist);
6841 PL_HasMultiCharFold =
6842 _new_invlist_C_array(_Perl_Folds_To_Multi_Char_invlist);
6844 /* This is calculated here, because the Perl program that generates the
6845 * static global ones doesn't currently have access to
6846 * NUM_ANYOF_CODE_POINTS */
6847 PL_InBitmap = _new_invlist(2);
6848 PL_InBitmap = _add_range_to_invlist(PL_InBitmap, 0,
6849 NUM_ANYOF_CODE_POINTS - 1);
6851 dump_len_string = PerlEnv_getenv("PERL_DUMP_RE_MAX_LEN");
6852 if ( ! dump_len_string
6853 || ! grok_atoUV(dump_len_string, (UV *)&PL_dump_re_max_len, NULL))
6855 PL_dump_re_max_len = 0;
6860 pRExC_state->warn_text = NULL;
6861 pRExC_state->code_blocks = NULL;
6864 *is_bare_re = FALSE;
6866 if (expr && (expr->op_type == OP_LIST ||
6867 (expr->op_type == OP_NULL && expr->op_targ == OP_LIST))) {
6868 /* allocate code_blocks if needed */
6872 for (o = cLISTOPx(expr)->op_first; o; o = OpSIBLING(o))
6873 if (o->op_type == OP_NULL && (o->op_flags & OPf_SPECIAL))
6874 ncode++; /* count of DO blocks */
6877 pRExC_state->code_blocks = S_alloc_code_blocks(aTHX_ ncode);
6881 /* compile-time pattern with just OP_CONSTs and DO blocks */
6886 /* find how many CONSTs there are */
6889 if (expr->op_type == OP_CONST)
6892 for (o = cLISTOPx(expr)->op_first; o; o = OpSIBLING(o)) {
6893 if (o->op_type == OP_CONST)
6897 /* fake up an SV array */
6899 assert(!new_patternp);
6900 Newx(new_patternp, n, SV*);
6901 SAVEFREEPV(new_patternp);
6905 if (expr->op_type == OP_CONST)
6906 new_patternp[n] = cSVOPx_sv(expr);
6908 for (o = cLISTOPx(expr)->op_first; o; o = OpSIBLING(o)) {
6909 if (o->op_type == OP_CONST)
6910 new_patternp[n++] = cSVOPo_sv;
6915 DEBUG_PARSE_r(Perl_re_printf( aTHX_
6916 "Assembling pattern from %d elements%s\n", pat_count,
6917 orig_rx_flags & RXf_SPLIT ? " for split" : ""));
6919 /* set expr to the first arg op */
6921 if (pRExC_state->code_blocks && pRExC_state->code_blocks->count
6922 && expr->op_type != OP_CONST)
6924 expr = cLISTOPx(expr)->op_first;
6925 assert( expr->op_type == OP_PUSHMARK
6926 || (expr->op_type == OP_NULL && expr->op_targ == OP_PUSHMARK)
6927 || expr->op_type == OP_PADRANGE);
6928 expr = OpSIBLING(expr);
6931 pat = S_concat_pat(aTHX_ pRExC_state, NULL, new_patternp, pat_count,
6932 expr, &recompile, NULL);
6934 /* handle bare (possibly after overloading) regex: foo =~ $re */
6939 if (SvTYPE(re) == SVt_REGEXP) {
6943 DEBUG_PARSE_r(Perl_re_printf( aTHX_
6944 "Precompiled pattern%s\n",
6945 orig_rx_flags & RXf_SPLIT ? " for split" : ""));
6951 exp = SvPV_nomg(pat, plen);
6953 if (!eng->op_comp) {
6954 if ((SvUTF8(pat) && IN_BYTES)
6955 || SvGMAGICAL(pat) || SvAMAGIC(pat))
6957 /* make a temporary copy; either to convert to bytes,
6958 * or to avoid repeating get-magic / overloaded stringify */
6959 pat = newSVpvn_flags(exp, plen, SVs_TEMP |
6960 (IN_BYTES ? 0 : SvUTF8(pat)));
6962 return CALLREGCOMP_ENG(eng, pat, orig_rx_flags);
6965 /* ignore the utf8ness if the pattern is 0 length */
6966 RExC_utf8 = RExC_orig_utf8 = (plen == 0 || IN_BYTES) ? 0 : SvUTF8(pat);
6968 RExC_uni_semantics = 0;
6969 RExC_seen_unfolded_sharp_s = 0;
6970 RExC_contains_locale = 0;
6971 RExC_strict = cBOOL(pm_flags & RXf_PMf_STRICT);
6972 RExC_study_started = 0;
6973 pRExC_state->runtime_code_qr = NULL;
6974 RExC_frame_head= NULL;
6975 RExC_frame_last= NULL;
6976 RExC_frame_count= 0;
6979 RExC_mysv1= sv_newmortal();
6980 RExC_mysv2= sv_newmortal();
6983 SV *dsv= sv_newmortal();
6984 RE_PV_QUOTED_DECL(s, RExC_utf8, dsv, exp, plen, 60);
6985 Perl_re_printf( aTHX_ "%sCompiling REx%s %s\n",
6986 PL_colors[4],PL_colors[5],s);
6990 /* we jump here if we have to recompile, e.g., from upgrading the pattern
6993 if ((pm_flags & PMf_USE_RE_EVAL)
6994 /* this second condition covers the non-regex literal case,
6995 * i.e. $foo =~ '(?{})'. */
6996 || (IN_PERL_COMPILETIME && (PL_hints & HINT_RE_EVAL))
6998 runtime_code = S_has_runtime_code(aTHX_ pRExC_state, exp, plen);
7000 /* return old regex if pattern hasn't changed */
7001 /* XXX: note in the below we have to check the flags as well as the
7004 * Things get a touch tricky as we have to compare the utf8 flag
7005 * independently from the compile flags. */
7009 && !!RX_UTF8(old_re) == !!RExC_utf8
7010 && ( RX_COMPFLAGS(old_re) == ( orig_rx_flags & RXf_PMf_FLAGCOPYMASK ) )
7011 && RX_PRECOMP(old_re)
7012 && RX_PRELEN(old_re) == plen
7013 && memEQ(RX_PRECOMP(old_re), exp, plen)
7014 && !runtime_code /* with runtime code, always recompile */ )
7019 rx_flags = orig_rx_flags;
7021 if ( initial_charset == REGEX_DEPENDS_CHARSET
7022 && (RExC_utf8 ||RExC_uni_semantics))
7025 /* Set to use unicode semantics if the pattern is in utf8 and has the
7026 * 'depends' charset specified, as it means unicode when utf8 */
7027 set_regex_charset(&rx_flags, REGEX_UNICODE_CHARSET);
7031 RExC_precomp_adj = 0;
7032 RExC_flags = rx_flags;
7033 RExC_pm_flags = pm_flags;
7036 assert(TAINTING_get || !TAINT_get);
7038 Perl_croak(aTHX_ "Eval-group in insecure regular expression");
7040 if (!S_compile_runtime_code(aTHX_ pRExC_state, exp, plen)) {
7041 /* whoops, we have a non-utf8 pattern, whilst run-time code
7042 * got compiled as utf8. Try again with a utf8 pattern */
7043 S_pat_upgrade_to_utf8(aTHX_ pRExC_state, &exp, &plen,
7044 pRExC_state->code_blocks ? pRExC_state->code_blocks->count : 0);
7045 goto redo_first_pass;
7048 assert(!pRExC_state->runtime_code_qr);
7054 RExC_in_lookbehind = 0;
7055 RExC_seen_zerolen = *exp == '^' ? -1 : 0;
7058 RExC_recode_x_to_native = 0;
7060 RExC_in_multi_char_class = 0;
7062 /* First pass: determine size, legality. */
7064 RExC_start = RExC_adjusted_start = exp;
7065 RExC_end = exp + plen;
7066 RExC_precomp_end = RExC_end;
7071 RExC_emit = (regnode *) &RExC_emit_dummy;
7072 RExC_whilem_seen = 0;
7073 RExC_open_parens = NULL;
7074 RExC_close_parens = NULL;
7076 RExC_paren_names = NULL;
7078 RExC_paren_name_list = NULL;
7080 RExC_recurse = NULL;
7081 RExC_study_chunk_recursed = NULL;
7082 RExC_study_chunk_recursed_bytes= 0;
7083 RExC_recurse_count = 0;
7084 pRExC_state->code_index = 0;
7086 /* This NUL is guaranteed because the pattern comes from an SV*, and the sv
7087 * code makes sure the final byte is an uncounted NUL. But should this
7088 * ever not be the case, lots of things could read beyond the end of the
7089 * buffer: loops like
7090 * while(isFOO(*RExC_parse)) RExC_parse++;
7091 * strchr(RExC_parse, "foo");
7092 * etc. So it is worth noting. */
7093 assert(*RExC_end == '\0');
7096 Perl_re_printf( aTHX_ "Starting first pass (sizing)\n");
7098 RExC_lastparse=NULL;
7101 if (reg(pRExC_state, 0, &flags,1) == NULL) {
7102 /* It's possible to write a regexp in ascii that represents Unicode
7103 codepoints outside of the byte range, such as via \x{100}. If we
7104 detect such a sequence we have to convert the entire pattern to utf8
7105 and then recompile, as our sizing calculation will have been based
7106 on 1 byte == 1 character, but we will need to use utf8 to encode
7107 at least some part of the pattern, and therefore must convert the whole
7110 if (flags & RESTART_PASS1) {
7111 if (flags & NEED_UTF8) {
7112 S_pat_upgrade_to_utf8(aTHX_ pRExC_state, &exp, &plen,
7113 pRExC_state->code_blocks ? pRExC_state->code_blocks->count : 0);
7116 DEBUG_PARSE_r(Perl_re_printf( aTHX_
7117 "Need to redo pass 1\n"));
7120 goto redo_first_pass;
7122 Perl_croak(aTHX_ "panic: reg returned NULL to re_op_compile for sizing pass, flags=%#" UVxf, (UV) flags);
7126 Perl_re_printf( aTHX_
7127 "Required size %" IVdf " nodes\n"
7128 "Starting second pass (creation)\n",
7131 RExC_lastparse=NULL;
7134 /* The first pass could have found things that force Unicode semantics */
7135 if ((RExC_utf8 || RExC_uni_semantics)
7136 && get_regex_charset(rx_flags) == REGEX_DEPENDS_CHARSET)
7138 set_regex_charset(&rx_flags, REGEX_UNICODE_CHARSET);
7141 /* Small enough for pointer-storage convention?
7142 If extralen==0, this means that we will not need long jumps. */
7143 if (RExC_size >= 0x10000L && RExC_extralen)
7144 RExC_size += RExC_extralen;
7147 if (RExC_whilem_seen > 15)
7148 RExC_whilem_seen = 15;
7150 /* Allocate space and zero-initialize. Note, the two step process
7151 of zeroing when in debug mode, thus anything assigned has to
7152 happen after that */
7153 rx = (REGEXP*) newSV_type(SVt_REGEXP);
7155 Newxc(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode),
7156 char, regexp_internal);
7157 if ( r == NULL || ri == NULL )
7158 FAIL("Regexp out of space");
7160 /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
7161 Zero(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode),
7164 /* bulk initialize base fields with 0. */
7165 Zero(ri, sizeof(regexp_internal), char);
7168 /* non-zero initialization begins here */
7171 r->extflags = rx_flags;
7172 RXp_COMPFLAGS(r) = orig_rx_flags & RXf_PMf_FLAGCOPYMASK;
7174 if (pm_flags & PMf_IS_QR) {
7175 ri->code_blocks = pRExC_state->code_blocks;
7176 if (ri->code_blocks)
7177 ri->code_blocks->refcnt++;
7181 bool has_p = ((r->extflags & RXf_PMf_KEEPCOPY) == RXf_PMf_KEEPCOPY);
7182 bool has_charset = (get_regex_charset(r->extflags)
7183 != REGEX_DEPENDS_CHARSET);
7185 /* The caret is output if there are any defaults: if not all the STD
7186 * flags are set, or if no character set specifier is needed */
7188 (((r->extflags & RXf_PMf_STD_PMMOD) != RXf_PMf_STD_PMMOD)
7190 bool has_runon = ((RExC_seen & REG_RUN_ON_COMMENT_SEEN)
7191 == REG_RUN_ON_COMMENT_SEEN);
7192 U8 reganch = (U8)((r->extflags & RXf_PMf_STD_PMMOD)
7193 >> RXf_PMf_STD_PMMOD_SHIFT);
7194 const char *fptr = STD_PAT_MODS; /*"msixxn"*/
7197 /* We output all the necessary flags; we never output a minus, as all
7198 * those are defaults, so are
7199 * covered by the caret */
7200 const STRLEN wraplen = plen + has_p + has_runon
7201 + has_default /* If needs a caret */
7202 + PL_bitcount[reganch] /* 1 char for each set standard flag */
7204 /* If needs a character set specifier */
7205 + ((has_charset) ? MAX_CHARSET_NAME_LENGTH : 0)
7206 + (sizeof("(?:)") - 1);
7208 /* make sure PL_bitcount bounds not exceeded */
7209 assert(sizeof(STD_PAT_MODS) <= 8);
7211 Newx(p, wraplen + 1, char); /* +1 for the ending NUL */
7212 r->xpv_len_u.xpvlenu_pv = p;
7214 SvFLAGS(rx) |= SVf_UTF8;
7217 /* If a default, cover it using the caret */
7219 *p++= DEFAULT_PAT_MOD;
7223 const char* const name = get_regex_charset_name(r->extflags, &len);
7224 Copy(name, p, len, char);
7228 *p++ = KEEPCOPY_PAT_MOD; /*'p'*/
7231 while((ch = *fptr++)) {
7239 Copy(RExC_precomp, p, plen, char);
7240 assert ((RX_WRAPPED(rx) - p) < 16);
7241 r->pre_prefix = p - RX_WRAPPED(rx);
7247 SvCUR_set(rx, p - RX_WRAPPED(rx));
7251 r->nparens = RExC_npar - 1; /* set early to validate backrefs */
7253 /* Useful during FAIL. */
7254 #ifdef RE_TRACK_PATTERN_OFFSETS
7255 Newxz(ri->u.offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
7256 DEBUG_OFFSETS_r(Perl_re_printf( aTHX_
7257 "%s %" UVuf " bytes for offset annotations.\n",
7258 ri->u.offsets ? "Got" : "Couldn't get",
7259 (UV)((2*RExC_size+1) * sizeof(U32))));
7261 SetProgLen(ri,RExC_size);
7266 /* Second pass: emit code. */
7267 RExC_flags = rx_flags; /* don't let top level (?i) bleed */
7268 RExC_pm_flags = pm_flags;
7270 RExC_end = exp + plen;
7272 RExC_emit_start = ri->program;
7273 RExC_emit = ri->program;
7274 RExC_emit_bound = ri->program + RExC_size + 1;
7275 pRExC_state->code_index = 0;
7277 *((char*) RExC_emit++) = (char) REG_MAGIC;
7278 /* setup various meta data about recursion, this all requires
7279 * RExC_npar to be correctly set, and a bit later on we clear it */
7280 if (RExC_seen & REG_RECURSE_SEEN) {
7281 DEBUG_OPTIMISE_MORE_r(Perl_re_printf( aTHX_
7282 "%*s%*s Setting up open/close parens\n",
7283 22, "| |", (int)(0 * 2 + 1), ""));
7285 /* setup RExC_open_parens, which holds the address of each
7286 * OPEN tag, and to make things simpler for the 0 index
7287 * the start of the program - this is used later for offsets */
7288 Newxz(RExC_open_parens, RExC_npar,regnode *);
7289 SAVEFREEPV(RExC_open_parens);
7290 RExC_open_parens[0] = RExC_emit;
7292 /* setup RExC_close_parens, which holds the address of each
7293 * CLOSE tag, and to make things simpler for the 0 index
7294 * the end of the program - this is used later for offsets */
7295 Newxz(RExC_close_parens, RExC_npar,regnode *);
7296 SAVEFREEPV(RExC_close_parens);
7297 /* we dont know where end op starts yet, so we dont
7298 * need to set RExC_close_parens[0] like we do RExC_open_parens[0] above */
7300 /* Note, RExC_npar is 1 + the number of parens in a pattern.
7301 * So its 1 if there are no parens. */
7302 RExC_study_chunk_recursed_bytes= (RExC_npar >> 3) +
7303 ((RExC_npar & 0x07) != 0);
7304 Newx(RExC_study_chunk_recursed,
7305 RExC_study_chunk_recursed_bytes * RExC_npar, U8);
7306 SAVEFREEPV(RExC_study_chunk_recursed);
7309 if (reg(pRExC_state, 0, &flags,1) == NULL) {
7311 Perl_croak(aTHX_ "panic: reg returned NULL to re_op_compile for generation pass, flags=%#" UVxf, (UV) flags);
7314 Perl_re_printf( aTHX_ "Starting post parse optimization\n");
7317 /* XXXX To minimize changes to RE engine we always allocate
7318 3-units-long substrs field. */
7319 Newx(r->substrs, 1, struct reg_substr_data);
7320 if (RExC_recurse_count) {
7321 Newxz(RExC_recurse,RExC_recurse_count,regnode *);
7322 SAVEFREEPV(RExC_recurse);
7326 r->minlen = minlen = sawlookahead = sawplus = sawopen = sawminmod = 0;
7328 RExC_study_chunk_recursed_count= 0;
7330 Zero(r->substrs, 1, struct reg_substr_data);
7331 if (RExC_study_chunk_recursed) {
7332 Zero(RExC_study_chunk_recursed,
7333 RExC_study_chunk_recursed_bytes * RExC_npar, U8);
7337 #ifdef TRIE_STUDY_OPT
7339 StructCopy(&zero_scan_data, &data, scan_data_t);
7340 copyRExC_state = RExC_state;
7343 DEBUG_OPTIMISE_r(Perl_re_printf( aTHX_ "Restudying\n"));
7345 RExC_state = copyRExC_state;
7346 if (seen & REG_TOP_LEVEL_BRANCHES_SEEN)
7347 RExC_seen |= REG_TOP_LEVEL_BRANCHES_SEEN;
7349 RExC_seen &= ~REG_TOP_LEVEL_BRANCHES_SEEN;
7350 StructCopy(&zero_scan_data, &data, scan_data_t);
7353 StructCopy(&zero_scan_data, &data, scan_data_t);
7356 /* Dig out information for optimizations. */
7357 r->extflags = RExC_flags; /* was pm_op */
7358 /*dmq: removed as part of de-PMOP: pm->op_pmflags = RExC_flags; */
7361 SvUTF8_on(rx); /* Unicode in it? */
7362 ri->regstclass = NULL;
7363 if (RExC_naughty >= TOO_NAUGHTY) /* Probably an expensive pattern. */
7364 r->intflags |= PREGf_NAUGHTY;
7365 scan = ri->program + 1; /* First BRANCH. */
7367 /* testing for BRANCH here tells us whether there is "must appear"
7368 data in the pattern. If there is then we can use it for optimisations */
7369 if (!(RExC_seen & REG_TOP_LEVEL_BRANCHES_SEEN)) { /* Only one top-level choice.
7372 STRLEN longest_float_length, longest_fixed_length;
7373 regnode_ssc ch_class; /* pointed to by data */
7375 SSize_t last_close = 0; /* pointed to by data */
7376 regnode *first= scan;
7377 regnode *first_next= regnext(first);
7379 * Skip introductions and multiplicators >= 1
7380 * so that we can extract the 'meat' of the pattern that must
7381 * match in the large if() sequence following.
7382 * NOTE that EXACT is NOT covered here, as it is normally
7383 * picked up by the optimiser separately.
7385 * This is unfortunate as the optimiser isnt handling lookahead
7386 * properly currently.
7389 while ((OP(first) == OPEN && (sawopen = 1)) ||
7390 /* An OR of *one* alternative - should not happen now. */
7391 (OP(first) == BRANCH && OP(first_next) != BRANCH) ||
7392 /* for now we can't handle lookbehind IFMATCH*/
7393 (OP(first) == IFMATCH && !first->flags && (sawlookahead = 1)) ||
7394 (OP(first) == PLUS) ||
7395 (OP(first) == MINMOD) ||
7396 /* An {n,m} with n>0 */
7397 (PL_regkind[OP(first)] == CURLY && ARG1(first) > 0) ||
7398 (OP(first) == NOTHING && PL_regkind[OP(first_next)] != END ))
7401 * the only op that could be a regnode is PLUS, all the rest
7402 * will be regnode_1 or regnode_2.
7404 * (yves doesn't think this is true)
7406 if (OP(first) == PLUS)
7409 if (OP(first) == MINMOD)
7411 first += regarglen[OP(first)];
7413 first = NEXTOPER(first);
7414 first_next= regnext(first);
7417 /* Starting-point info. */
7419 DEBUG_PEEP("first:",first,0);
7420 /* Ignore EXACT as we deal with it later. */
7421 if (PL_regkind[OP(first)] == EXACT) {
7422 if (OP(first) == EXACT || OP(first) == EXACTL)
7423 NOOP; /* Empty, get anchored substr later. */
7425 ri->regstclass = first;
7428 else if (PL_regkind[OP(first)] == TRIE &&
7429 ((reg_trie_data *)ri->data->data[ ARG(first) ])->minlen>0)
7431 /* this can happen only on restudy */
7432 ri->regstclass = construct_ahocorasick_from_trie(pRExC_state, (regnode *)first, 0);
7435 else if (REGNODE_SIMPLE(OP(first)))
7436 ri->regstclass = first;
7437 else if (PL_regkind[OP(first)] == BOUND ||
7438 PL_regkind[OP(first)] == NBOUND)
7439 ri->regstclass = first;
7440 else if (PL_regkind[OP(first)] == BOL) {
7441 r->intflags |= (OP(first) == MBOL
7444 first = NEXTOPER(first);
7447 else if (OP(first) == GPOS) {
7448 r->intflags |= PREGf_ANCH_GPOS;
7449 first = NEXTOPER(first);
7452 else if ((!sawopen || !RExC_sawback) &&
7454 (OP(first) == STAR &&
7455 PL_regkind[OP(NEXTOPER(first))] == REG_ANY) &&
7456 !(r->intflags & PREGf_ANCH) && !pRExC_state->code_blocks)
7458 /* turn .* into ^.* with an implied $*=1 */
7460 (OP(NEXTOPER(first)) == REG_ANY)
7463 r->intflags |= (type | PREGf_IMPLICIT);
7464 first = NEXTOPER(first);
7467 if (sawplus && !sawminmod && !sawlookahead
7468 && (!sawopen || !RExC_sawback)
7469 && !pRExC_state->code_blocks) /* May examine pos and $& */
7470 /* x+ must match at the 1st pos of run of x's */
7471 r->intflags |= PREGf_SKIP;
7473 /* Scan is after the zeroth branch, first is atomic matcher. */
7474 #ifdef TRIE_STUDY_OPT
7477 Perl_re_printf( aTHX_ "first at %" IVdf "\n",
7478 (IV)(first - scan + 1))
7482 Perl_re_printf( aTHX_ "first at %" IVdf "\n",
7483 (IV)(first - scan + 1))
7489 * If there's something expensive in the r.e., find the
7490 * longest literal string that must appear and make it the
7491 * regmust. Resolve ties in favor of later strings, since
7492 * the regstart check works with the beginning of the r.e.
7493 * and avoiding duplication strengthens checking. Not a
7494 * strong reason, but sufficient in the absence of others.
7495 * [Now we resolve ties in favor of the earlier string if
7496 * it happens that c_offset_min has been invalidated, since the
7497 * earlier string may buy us something the later one won't.]
7500 data.longest_fixed = newSVpvs("");
7501 data.longest_float = newSVpvs("");
7502 data.last_found = newSVpvs("");
7503 data.longest = &(data.longest_fixed);
7504 ENTER_with_name("study_chunk");
7505 SAVEFREESV(data.longest_fixed);
7506 SAVEFREESV(data.longest_float);
7507 SAVEFREESV(data.last_found);
7509 if (!ri->regstclass) {
7510 ssc_init(pRExC_state, &ch_class);
7511 data.start_class = &ch_class;
7512 stclass_flag = SCF_DO_STCLASS_AND;
7513 } else /* XXXX Check for BOUND? */
7515 data.last_closep = &last_close;
7518 minlen = study_chunk(pRExC_state, &first, &minlen, &fake,
7519 scan + RExC_size, /* Up to end */
7521 SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag
7522 | (restudied ? SCF_TRIE_DOING_RESTUDY : 0),
7526 CHECK_RESTUDY_GOTO_butfirst(LEAVE_with_name("study_chunk"));
7529 if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
7530 && data.last_start_min == 0 && data.last_end > 0
7531 && !RExC_seen_zerolen
7532 && !(RExC_seen & REG_VERBARG_SEEN)
7533 && !(RExC_seen & REG_GPOS_SEEN)
7535 r->extflags |= RXf_CHECK_ALL;
7537 scan_commit(pRExC_state, &data,&minlen,0);
7539 longest_float_length = CHR_SVLEN(data.longest_float);
7541 if (! ((SvCUR(data.longest_fixed) /* ok to leave SvCUR */
7542 && data.offset_fixed == data.offset_float_min
7543 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float)))
7544 && S_setup_longest (aTHX_ pRExC_state,
7548 &(r->float_end_shift),
7549 data.lookbehind_float,
7550 data.offset_float_min,
7552 longest_float_length,
7553 cBOOL(data.flags & SF_FL_BEFORE_EOL),
7554 cBOOL(data.flags & SF_FL_BEFORE_MEOL)))
7556 r->float_min_offset = data.offset_float_min - data.lookbehind_float;
7557 r->float_max_offset = data.offset_float_max;
7558 if (data.offset_float_max < SSize_t_MAX) /* Don't offset infinity */
7559 r->float_max_offset -= data.lookbehind_float;
7560 SvREFCNT_inc_simple_void_NN(data.longest_float);
7563 r->float_substr = r->float_utf8 = NULL;
7564 longest_float_length = 0;
7567 longest_fixed_length = CHR_SVLEN(data.longest_fixed);
7569 if (S_setup_longest (aTHX_ pRExC_state,
7571 &(r->anchored_utf8),
7572 &(r->anchored_substr),
7573 &(r->anchored_end_shift),
7574 data.lookbehind_fixed,
7577 longest_fixed_length,
7578 cBOOL(data.flags & SF_FIX_BEFORE_EOL),
7579 cBOOL(data.flags & SF_FIX_BEFORE_MEOL)))
7581 r->anchored_offset = data.offset_fixed - data.lookbehind_fixed;
7582 SvREFCNT_inc_simple_void_NN(data.longest_fixed);
7585 r->anchored_substr = r->anchored_utf8 = NULL;
7586 longest_fixed_length = 0;
7588 LEAVE_with_name("study_chunk");
7591 && (OP(ri->regstclass) == REG_ANY || OP(ri->regstclass) == SANY))
7592 ri->regstclass = NULL;
7594 if ((!(r->anchored_substr || r->anchored_utf8) || r->anchored_offset)
7596 && ! (ANYOF_FLAGS(data.start_class) & SSC_MATCHES_EMPTY_STRING)
7597 && is_ssc_worth_it(pRExC_state, data.start_class))
7599 const U32 n = add_data(pRExC_state, STR_WITH_LEN("f"));
7601 ssc_finalize(pRExC_state, data.start_class);
7603 Newx(RExC_rxi->data->data[n], 1, regnode_ssc);
7604 StructCopy(data.start_class,
7605 (regnode_ssc*)RExC_rxi->data->data[n],
7607 ri->regstclass = (regnode*)RExC_rxi->data->data[n];
7608 r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
7609 DEBUG_COMPILE_r({ SV *sv = sv_newmortal();
7610 regprop(r, sv, (regnode*)data.start_class, NULL, pRExC_state);
7611 Perl_re_printf( aTHX_
7612 "synthetic stclass \"%s\".\n",
7613 SvPVX_const(sv));});
7614 data.start_class = NULL;
7617 /* A temporary algorithm prefers floated substr to fixed one to dig
7619 if (longest_fixed_length > longest_float_length) {
7620 r->substrs->check_ix = 0;
7621 r->check_end_shift = r->anchored_end_shift;
7622 r->check_substr = r->anchored_substr;
7623 r->check_utf8 = r->anchored_utf8;
7624 r->check_offset_min = r->check_offset_max = r->anchored_offset;
7625 if (r->intflags & (PREGf_ANCH_SBOL|PREGf_ANCH_GPOS))
7626 r->intflags |= PREGf_NOSCAN;
7629 r->substrs->check_ix = 1;
7630 r->check_end_shift = r->float_end_shift;
7631 r->check_substr = r->float_substr;
7632 r->check_utf8 = r->float_utf8;
7633 r->check_offset_min = r->float_min_offset;
7634 r->check_offset_max = r->float_max_offset;
7636 if ((r->check_substr || r->check_utf8) ) {
7637 r->extflags |= RXf_USE_INTUIT;
7638 if (SvTAIL(r->check_substr ? r->check_substr : r->check_utf8))
7639 r->extflags |= RXf_INTUIT_TAIL;
7641 r->substrs->data[0].max_offset = r->substrs->data[0].min_offset;
7643 /* XXX Unneeded? dmq (shouldn't as this is handled elsewhere)
7644 if ( (STRLEN)minlen < longest_float_length )
7645 minlen= longest_float_length;
7646 if ( (STRLEN)minlen < longest_fixed_length )
7647 minlen= longest_fixed_length;
7651 /* Several toplevels. Best we can is to set minlen. */
7653 regnode_ssc ch_class;
7654 SSize_t last_close = 0;
7656 DEBUG_PARSE_r(Perl_re_printf( aTHX_ "\nMulti Top Level\n"));
7658 scan = ri->program + 1;
7659 ssc_init(pRExC_state, &ch_class);
7660 data.start_class = &ch_class;
7661 data.last_closep = &last_close;
7664 minlen = study_chunk(pRExC_state,
7665 &scan, &minlen, &fake, scan + RExC_size, &data, -1, 0, NULL,
7666 SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS|(restudied
7667 ? SCF_TRIE_DOING_RESTUDY
7671 CHECK_RESTUDY_GOTO_butfirst(NOOP);
7673 r->check_substr = r->check_utf8 = r->anchored_substr = r->anchored_utf8
7674 = r->float_substr = r->float_utf8 = NULL;
7676 if (! (ANYOF_FLAGS(data.start_class) & SSC_MATCHES_EMPTY_STRING)
7677 && is_ssc_worth_it(pRExC_state, data.start_class))
7679 const U32 n = add_data(pRExC_state, STR_WITH_LEN("f"));
7681 ssc_finalize(pRExC_state, data.start_class);
7683 Newx(RExC_rxi->data->data[n], 1, regnode_ssc);
7684 StructCopy(data.start_class,
7685 (regnode_ssc*)RExC_rxi->data->data[n],
7687 ri->regstclass = (regnode*)RExC_rxi->data->data[n];
7688 r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
7689 DEBUG_COMPILE_r({ SV* sv = sv_newmortal();
7690 regprop(r, sv, (regnode*)data.start_class, NULL, pRExC_state);
7691 Perl_re_printf( aTHX_
7692 "synthetic stclass \"%s\".\n",
7693 SvPVX_const(sv));});
7694 data.start_class = NULL;
7698 if (RExC_seen & REG_UNBOUNDED_QUANTIFIER_SEEN) {
7699 r->extflags |= RXf_UNBOUNDED_QUANTIFIER_SEEN;
7700 r->maxlen = REG_INFTY;
7703 r->maxlen = RExC_maxlen;
7706 /* Guard against an embedded (?=) or (?<=) with a longer minlen than
7707 the "real" pattern. */
7709 Perl_re_printf( aTHX_ "minlen: %" IVdf " r->minlen:%" IVdf " maxlen:%" IVdf "\n",
7710 (IV)minlen, (IV)r->minlen, (IV)RExC_maxlen);
7712 r->minlenret = minlen;
7713 if (r->minlen < minlen)
7716 if (RExC_seen & REG_RECURSE_SEEN ) {
7717 r->intflags |= PREGf_RECURSE_SEEN;
7718 Newxz(r->recurse_locinput, r->nparens + 1, char *);
7720 if (RExC_seen & REG_GPOS_SEEN)
7721 r->intflags |= PREGf_GPOS_SEEN;
7722 if (RExC_seen & REG_LOOKBEHIND_SEEN)
7723 r->extflags |= RXf_NO_INPLACE_SUBST; /* inplace might break the
7725 if (pRExC_state->code_blocks)
7726 r->extflags |= RXf_EVAL_SEEN;
7727 if (RExC_seen & REG_VERBARG_SEEN)
7729 r->intflags |= PREGf_VERBARG_SEEN;
7730 r->extflags |= RXf_NO_INPLACE_SUBST; /* don't understand this! Yves */
7732 if (RExC_seen & REG_CUTGROUP_SEEN)
7733 r->intflags |= PREGf_CUTGROUP_SEEN;
7734 if (pm_flags & PMf_USE_RE_EVAL)
7735 r->intflags |= PREGf_USE_RE_EVAL;
7736 if (RExC_paren_names)
7737 RXp_PAREN_NAMES(r) = MUTABLE_HV(SvREFCNT_inc(RExC_paren_names));
7739 RXp_PAREN_NAMES(r) = NULL;
7741 /* If we have seen an anchor in our pattern then we set the extflag RXf_IS_ANCHORED
7742 * so it can be used in pp.c */
7743 if (r->intflags & PREGf_ANCH)
7744 r->extflags |= RXf_IS_ANCHORED;
7748 /* this is used to identify "special" patterns that might result
7749 * in Perl NOT calling the regex engine and instead doing the match "itself",
7750 * particularly special cases in split//. By having the regex compiler
7751 * do this pattern matching at a regop level (instead of by inspecting the pattern)
7752 * we avoid weird issues with equivalent patterns resulting in different behavior,
7753 * AND we allow non Perl engines to get the same optimizations by the setting the
7754 * flags appropriately - Yves */
7755 regnode *first = ri->program + 1;
7757 regnode *next = regnext(first);
7760 if (PL_regkind[fop] == NOTHING && nop == END)
7761 r->extflags |= RXf_NULL;
7762 else if ((fop == MBOL || (fop == SBOL && !first->flags)) && nop == END)
7763 /* when fop is SBOL first->flags will be true only when it was
7764 * produced by parsing /\A/, and not when parsing /^/. This is
7765 * very important for the split code as there we want to
7766 * treat /^/ as /^/m, but we do not want to treat /\A/ as /^/m.
7767 * See rt #122761 for more details. -- Yves */
7768 r->extflags |= RXf_START_ONLY;
7769 else if (fop == PLUS
7770 && PL_regkind[nop] == POSIXD && FLAGS(next) == _CC_SPACE
7772 r->extflags |= RXf_WHITE;
7773 else if ( r->extflags & RXf_SPLIT
7774 && (fop == EXACT || fop == EXACTL)
7775 && STR_LEN(first) == 1
7776 && *(STRING(first)) == ' '
7778 r->extflags |= (RXf_SKIPWHITE|RXf_WHITE);
7782 if (RExC_contains_locale) {
7783 RXp_EXTFLAGS(r) |= RXf_TAINTED;
7787 if (RExC_paren_names) {
7788 ri->name_list_idx = add_data( pRExC_state, STR_WITH_LEN("a"));
7789 ri->data->data[ri->name_list_idx]
7790 = (void*)SvREFCNT_inc(RExC_paren_name_list);
7793 ri->name_list_idx = 0;
7795 while ( RExC_recurse_count > 0 ) {
7796 const regnode *scan = RExC_recurse[ --RExC_recurse_count ];
7798 * This data structure is set up in study_chunk() and is used
7799 * to calculate the distance between a GOSUB regopcode and
7800 * the OPEN/CURLYM (CURLYM's are special and can act like OPEN's)
7803 * If for some reason someone writes code that optimises
7804 * away a GOSUB opcode then the assert should be changed to
7805 * an if(scan) to guard the ARG2L_SET() - Yves
7808 assert(scan && OP(scan) == GOSUB);
7809 ARG2L_SET( scan, RExC_open_parens[ARG(scan)] - scan );
7812 Newxz(r->offs, RExC_npar, regexp_paren_pair);
7813 /* assume we don't need to swap parens around before we match */
7815 Perl_re_printf( aTHX_ "study_chunk_recursed_count: %lu\n",
7816 (unsigned long)RExC_study_chunk_recursed_count);
7820 Perl_re_printf( aTHX_ "Final program:\n");
7823 #ifdef RE_TRACK_PATTERN_OFFSETS
7824 DEBUG_OFFSETS_r(if (ri->u.offsets) {
7825 const STRLEN len = ri->u.offsets[0];
7827 GET_RE_DEBUG_FLAGS_DECL;
7828 Perl_re_printf( aTHX_
7829 "Offsets: [%" UVuf "]\n\t", (UV)ri->u.offsets[0]);
7830 for (i = 1; i <= len; i++) {
7831 if (ri->u.offsets[i*2-1] || ri->u.offsets[i*2])
7832 Perl_re_printf( aTHX_ "%" UVuf ":%" UVuf "[%" UVuf "] ",
7833 (UV)i, (UV)ri->u.offsets[i*2-1], (UV)ri->u.offsets[i*2]);
7835 Perl_re_printf( aTHX_ "\n");
7840 /* under ithreads the ?pat? PMf_USED flag on the pmop is simulated
7841 * by setting the regexp SV to readonly-only instead. If the
7842 * pattern's been recompiled, the USEDness should remain. */
7843 if (old_re && SvREADONLY(old_re))
7851 Perl_reg_named_buff(pTHX_ REGEXP * const rx, SV * const key, SV * const value,
7854 PERL_ARGS_ASSERT_REG_NAMED_BUFF;
7856 PERL_UNUSED_ARG(value);
7858 if (flags & RXapif_FETCH) {
7859 return reg_named_buff_fetch(rx, key, flags);
7860 } else if (flags & (RXapif_STORE | RXapif_DELETE | RXapif_CLEAR)) {
7861 Perl_croak_no_modify();
7863 } else if (flags & RXapif_EXISTS) {
7864 return reg_named_buff_exists(rx, key, flags)
7867 } else if (flags & RXapif_REGNAMES) {
7868 return reg_named_buff_all(rx, flags);
7869 } else if (flags & (RXapif_SCALAR | RXapif_REGNAMES_COUNT)) {
7870 return reg_named_buff_scalar(rx, flags);
7872 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff", (int)flags);
7878 Perl_reg_named_buff_iter(pTHX_ REGEXP * const rx, const SV * const lastkey,
7881 PERL_ARGS_ASSERT_REG_NAMED_BUFF_ITER;
7882 PERL_UNUSED_ARG(lastkey);
7884 if (flags & RXapif_FIRSTKEY)
7885 return reg_named_buff_firstkey(rx, flags);
7886 else if (flags & RXapif_NEXTKEY)
7887 return reg_named_buff_nextkey(rx, flags);
7889 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_iter",
7896 Perl_reg_named_buff_fetch(pTHX_ REGEXP * const r, SV * const namesv,
7900 struct regexp *const rx = ReANY(r);
7902 PERL_ARGS_ASSERT_REG_NAMED_BUFF_FETCH;
7904 if (rx && RXp_PAREN_NAMES(rx)) {
7905 HE *he_str = hv_fetch_ent( RXp_PAREN_NAMES(rx), namesv, 0, 0 );
7908 SV* sv_dat=HeVAL(he_str);
7909 I32 *nums=(I32*)SvPVX(sv_dat);
7910 AV * const retarray = (flags & RXapif_ALL) ? newAV() : NULL;
7911 for ( i=0; i<SvIVX(sv_dat); i++ ) {
7912 if ((I32)(rx->nparens) >= nums[i]
7913 && rx->offs[nums[i]].start != -1
7914 && rx->offs[nums[i]].end != -1)
7917 CALLREG_NUMBUF_FETCH(r,nums[i],ret);
7922 ret = newSVsv(&PL_sv_undef);
7925 av_push(retarray, ret);
7928 return newRV_noinc(MUTABLE_SV(retarray));
7935 Perl_reg_named_buff_exists(pTHX_ REGEXP * const r, SV * const key,
7938 struct regexp *const rx = ReANY(r);
7940 PERL_ARGS_ASSERT_REG_NAMED_BUFF_EXISTS;
7942 if (rx && RXp_PAREN_NAMES(rx)) {
7943 if (flags & RXapif_ALL) {
7944 return hv_exists_ent(RXp_PAREN_NAMES(rx), key, 0);
7946 SV *sv = CALLREG_NAMED_BUFF_FETCH(r, key, flags);
7948 SvREFCNT_dec_NN(sv);
7960 Perl_reg_named_buff_firstkey(pTHX_ REGEXP * const r, const U32 flags)
7962 struct regexp *const rx = ReANY(r);
7964 PERL_ARGS_ASSERT_REG_NAMED_BUFF_FIRSTKEY;
7966 if ( rx && RXp_PAREN_NAMES(rx) ) {
7967 (void)hv_iterinit(RXp_PAREN_NAMES(rx));
7969 return CALLREG_NAMED_BUFF_NEXTKEY(r, NULL, flags & ~RXapif_FIRSTKEY);
7976 Perl_reg_named_buff_nextkey(pTHX_ REGEXP * const r, const U32 flags)
7978 struct regexp *const rx = ReANY(r);
7979 GET_RE_DEBUG_FLAGS_DECL;
7981 PERL_ARGS_ASSERT_REG_NAMED_BUFF_NEXTKEY;
7983 if (rx && RXp_PAREN_NAMES(rx)) {
7984 HV *hv = RXp_PAREN_NAMES(rx);
7986 while ( (temphe = hv_iternext_flags(hv,0)) ) {
7989 SV* sv_dat = HeVAL(temphe);
7990 I32 *nums = (I32*)SvPVX(sv_dat);
7991 for ( i = 0; i < SvIVX(sv_dat); i++ ) {
7992 if ((I32)(rx->lastparen) >= nums[i] &&
7993 rx->offs[nums[i]].start != -1 &&
7994 rx->offs[nums[i]].end != -1)
8000 if (parno || flags & RXapif_ALL) {
8001 return newSVhek(HeKEY_hek(temphe));
8009 Perl_reg_named_buff_scalar(pTHX_ REGEXP * const r, const U32 flags)
8014 struct regexp *const rx = ReANY(r);
8016 PERL_ARGS_ASSERT_REG_NAMED_BUFF_SCALAR;
8018 if (rx && RXp_PAREN_NAMES(rx)) {
8019 if (flags & (RXapif_ALL | RXapif_REGNAMES_COUNT)) {
8020 return newSViv(HvTOTALKEYS(RXp_PAREN_NAMES(rx)));
8021 } else if (flags & RXapif_ONE) {
8022 ret = CALLREG_NAMED_BUFF_ALL(r, (flags | RXapif_REGNAMES));
8023 av = MUTABLE_AV(SvRV(ret));
8024 length = av_tindex(av);
8025 SvREFCNT_dec_NN(ret);
8026 return newSViv(length + 1);
8028 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_scalar",
8033 return &PL_sv_undef;
8037 Perl_reg_named_buff_all(pTHX_ REGEXP * const r, const U32 flags)
8039 struct regexp *const rx = ReANY(r);
8042 PERL_ARGS_ASSERT_REG_NAMED_BUFF_ALL;
8044 if (rx && RXp_PAREN_NAMES(rx)) {
8045 HV *hv= RXp_PAREN_NAMES(rx);
8047 (void)hv_iterinit(hv);
8048 while ( (temphe = hv_iternext_flags(hv,0)) ) {
8051 SV* sv_dat = HeVAL(temphe);
8052 I32 *nums = (I32*)SvPVX(sv_dat);
8053 for ( i = 0; i < SvIVX(sv_dat); i++ ) {
8054 if ((I32)(rx->lastparen) >= nums[i] &&
8055 rx->offs[nums[i]].start != -1 &&
8056 rx->offs[nums[i]].end != -1)
8062 if (parno || flags & RXapif_ALL) {
8063 av_push(av, newSVhek(HeKEY_hek(temphe)));
8068 return newRV_noinc(MUTABLE_SV(av));
8072 Perl_reg_numbered_buff_fetch(pTHX_ REGEXP * const r, const I32 paren,
8075 struct regexp *const rx = ReANY(r);
8081 PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_FETCH;
8083 if ( n == RX_BUFF_IDX_CARET_PREMATCH
8084 || n == RX_BUFF_IDX_CARET_FULLMATCH
8085 || n == RX_BUFF_IDX_CARET_POSTMATCH
8088 bool keepcopy = cBOOL(rx->extflags & RXf_PMf_KEEPCOPY);
8090 /* on something like
8093 * the KEEPCOPY is set on the PMOP rather than the regex */
8094 if (PL_curpm && r == PM_GETRE(PL_curpm))
8095 keepcopy = cBOOL(PL_curpm->op_pmflags & PMf_KEEPCOPY);
8104 if (n == RX_BUFF_IDX_CARET_FULLMATCH)
8105 /* no need to distinguish between them any more */
8106 n = RX_BUFF_IDX_FULLMATCH;
8108 if ((n == RX_BUFF_IDX_PREMATCH || n == RX_BUFF_IDX_CARET_PREMATCH)
8109 && rx->offs[0].start != -1)
8111 /* $`, ${^PREMATCH} */
8112 i = rx->offs[0].start;
8116 if ((n == RX_BUFF_IDX_POSTMATCH || n == RX_BUFF_IDX_CARET_POSTMATCH)
8117 && rx->offs[0].end != -1)
8119 /* $', ${^POSTMATCH} */
8120 s = rx->subbeg - rx->suboffset + rx->offs[0].end;
8121 i = rx->sublen + rx->suboffset - rx->offs[0].end;
8124 if ( 0 <= n && n <= (I32)rx->nparens &&
8125 (s1 = rx->offs[n].start) != -1 &&
8126 (t1 = rx->offs[n].end) != -1)
8128 /* $&, ${^MATCH}, $1 ... */
8130 s = rx->subbeg + s1 - rx->suboffset;
8135 assert(s >= rx->subbeg);
8136 assert((STRLEN)rx->sublen >= (STRLEN)((s - rx->subbeg) + i) );
8138 #ifdef NO_TAINT_SUPPORT
8139 sv_setpvn(sv, s, i);
8141 const int oldtainted = TAINT_get;
8143 sv_setpvn(sv, s, i);
8144 TAINT_set(oldtainted);
8146 if (RXp_MATCH_UTF8(rx))
8151 if (RXp_MATCH_TAINTED(rx)) {
8152 if (SvTYPE(sv) >= SVt_PVMG) {
8153 MAGIC* const mg = SvMAGIC(sv);
8156 SvMAGIC_set(sv, mg->mg_moremagic);
8158 if ((mgt = SvMAGIC(sv))) {
8159 mg->mg_moremagic = mgt;
8160 SvMAGIC_set(sv, mg);
8177 Perl_reg_numbered_buff_store(pTHX_ REGEXP * const rx, const I32 paren,
8178 SV const * const value)
8180 PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_STORE;
8182 PERL_UNUSED_ARG(rx);
8183 PERL_UNUSED_ARG(paren);
8184 PERL_UNUSED_ARG(value);
8187 Perl_croak_no_modify();
8191 Perl_reg_numbered_buff_length(pTHX_ REGEXP * const r, const SV * const sv,
8194 struct regexp *const rx = ReANY(r);
8198 PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_LENGTH;
8200 if ( paren == RX_BUFF_IDX_CARET_PREMATCH
8201 || paren == RX_BUFF_IDX_CARET_FULLMATCH
8202 || paren == RX_BUFF_IDX_CARET_POSTMATCH
8205 bool keepcopy = cBOOL(rx->extflags & RXf_PMf_KEEPCOPY);
8207 /* on something like
8210 * the KEEPCOPY is set on the PMOP rather than the regex */
8211 if (PL_curpm && r == PM_GETRE(PL_curpm))
8212 keepcopy = cBOOL(PL_curpm->op_pmflags & PMf_KEEPCOPY);
8218 /* Some of this code was originally in C<Perl_magic_len> in F<mg.c> */
8220 case RX_BUFF_IDX_CARET_PREMATCH: /* ${^PREMATCH} */
8221 case RX_BUFF_IDX_PREMATCH: /* $` */
8222 if (rx->offs[0].start != -1) {
8223 i = rx->offs[0].start;
8232 case RX_BUFF_IDX_CARET_POSTMATCH: /* ${^POSTMATCH} */
8233 case RX_BUFF_IDX_POSTMATCH: /* $' */
8234 if (rx->offs[0].end != -1) {
8235 i = rx->sublen - rx->offs[0].end;
8237 s1 = rx->offs[0].end;
8244 default: /* $& / ${^MATCH}, $1, $2, ... */
8245 if (paren <= (I32)rx->nparens &&
8246 (s1 = rx->offs[paren].start) != -1 &&
8247 (t1 = rx->offs[paren].end) != -1)
8253 if (ckWARN(WARN_UNINITIALIZED))
8254 report_uninit((const SV *)sv);
8259 if (i > 0 && RXp_MATCH_UTF8(rx)) {
8260 const char * const s = rx->subbeg - rx->suboffset + s1;
8265 if (is_utf8_string_loclen((U8*)s, i, &ep, &el))
8272 Perl_reg_qr_package(pTHX_ REGEXP * const rx)
8274 PERL_ARGS_ASSERT_REG_QR_PACKAGE;
8275 PERL_UNUSED_ARG(rx);
8279 return newSVpvs("Regexp");
8282 /* Scans the name of a named buffer from the pattern.
8283 * If flags is REG_RSN_RETURN_NULL returns null.
8284 * If flags is REG_RSN_RETURN_NAME returns an SV* containing the name
8285 * If flags is REG_RSN_RETURN_DATA returns the data SV* corresponding
8286 * to the parsed name as looked up in the RExC_paren_names hash.
8287 * If there is an error throws a vFAIL().. type exception.
8290 #define REG_RSN_RETURN_NULL 0
8291 #define REG_RSN_RETURN_NAME 1
8292 #define REG_RSN_RETURN_DATA 2
8295 S_reg_scan_name(pTHX_ RExC_state_t *pRExC_state, U32 flags)
8297 char *name_start = RExC_parse;
8299 PERL_ARGS_ASSERT_REG_SCAN_NAME;
8301 assert (RExC_parse <= RExC_end);
8302 if (RExC_parse == RExC_end) NOOP;
8303 else if (isIDFIRST_lazy_if_safe(RExC_parse, RExC_end, UTF)) {
8304 /* Note that the code here assumes well-formed UTF-8. Skip IDFIRST by
8305 * using do...while */
8308 RExC_parse += UTF8SKIP(RExC_parse);
8309 } while ( RExC_parse < RExC_end
8310 && isWORDCHAR_utf8_safe((U8*)RExC_parse, (U8*) RExC_end));
8314 } while (RExC_parse < RExC_end && isWORDCHAR(*RExC_parse));
8316 RExC_parse++; /* so the <- from the vFAIL is after the offending
8318 vFAIL("Group name must start with a non-digit word character");
8322 = newSVpvn_flags(name_start, (int)(RExC_parse - name_start),
8323 SVs_TEMP | (UTF ? SVf_UTF8 : 0));
8324 if ( flags == REG_RSN_RETURN_NAME)
8326 else if (flags==REG_RSN_RETURN_DATA) {
8329 if ( ! sv_name ) /* should not happen*/
8330 Perl_croak(aTHX_ "panic: no svname in reg_scan_name");
8331 if (RExC_paren_names)
8332 he_str = hv_fetch_ent( RExC_paren_names, sv_name, 0, 0 );
8334 sv_dat = HeVAL(he_str);
8336 vFAIL("Reference to nonexistent named group");
8340 Perl_croak(aTHX_ "panic: bad flag %lx in reg_scan_name",
8341 (unsigned long) flags);
8343 NOT_REACHED; /* NOTREACHED */
8348 #define DEBUG_PARSE_MSG(funcname) DEBUG_PARSE_r({ \
8350 if (RExC_lastparse!=RExC_parse) { \
8351 Perl_re_printf( aTHX_ "%s", \
8352 Perl_pv_pretty(aTHX_ RExC_mysv1, RExC_parse, \
8353 RExC_end - RExC_parse, 16, \
8355 PERL_PV_ESCAPE_UNI_DETECT | \
8356 PERL_PV_PRETTY_ELLIPSES | \
8357 PERL_PV_PRETTY_LTGT | \
8358 PERL_PV_ESCAPE_RE | \
8359 PERL_PV_PRETTY_EXACTSIZE \
8363 Perl_re_printf( aTHX_ "%16s",""); \
8366 num = RExC_size + 1; \
8368 num=REG_NODE_NUM(RExC_emit); \
8369 if (RExC_lastnum!=num) \
8370 Perl_re_printf( aTHX_ "|%4d",num); \
8372 Perl_re_printf( aTHX_ "|%4s",""); \
8373 Perl_re_printf( aTHX_ "|%*s%-4s", \
8374 (int)((depth*2)), "", \
8378 RExC_lastparse=RExC_parse; \
8383 #define DEBUG_PARSE(funcname) DEBUG_PARSE_r({ \
8384 DEBUG_PARSE_MSG((funcname)); \
8385 Perl_re_printf( aTHX_ "%4s","\n"); \
8387 #define DEBUG_PARSE_FMT(funcname,fmt,args) DEBUG_PARSE_r({\
8388 DEBUG_PARSE_MSG((funcname)); \
8389 Perl_re_printf( aTHX_ fmt "\n",args); \
8392 /* This section of code defines the inversion list object and its methods. The
8393 * interfaces are highly subject to change, so as much as possible is static to
8394 * this file. An inversion list is here implemented as a malloc'd C UV array
8395 * as an SVt_INVLIST scalar.
8397 * An inversion list for Unicode is an array of code points, sorted by ordinal
8398 * number. Each element gives the code point that begins a range that extends
8399 * up-to but not including the code point given by the next element. The final
8400 * element gives the first code point of a range that extends to the platform's
8401 * infinity. The even-numbered elements (invlist[0], invlist[2], invlist[4],
8402 * ...) give ranges whose code points are all in the inversion list. We say
8403 * that those ranges are in the set. The odd-numbered elements give ranges
8404 * whose code points are not in the inversion list, and hence not in the set.
8405 * Thus, element [0] is the first code point in the list. Element [1]
8406 * is the first code point beyond that not in the list; and element [2] is the
8407 * first code point beyond that that is in the list. In other words, the first
8408 * range is invlist[0]..(invlist[1]-1), and all code points in that range are
8409 * in the inversion list. The second range is invlist[1]..(invlist[2]-1), and
8410 * all code points in that range are not in the inversion list. The third
8411 * range invlist[2]..(invlist[3]-1) gives code points that are in the inversion
8412 * list, and so forth. Thus every element whose index is divisible by two
8413 * gives the beginning of a range that is in the list, and every element whose
8414 * index is not divisible by two gives the beginning of a range not in the
8415 * list. If the final element's index is divisible by two, the inversion list
8416 * extends to the platform's infinity; otherwise the highest code point in the
8417 * inversion list is the contents of that element minus 1.
8419 * A range that contains just a single code point N will look like
8421 * invlist[i+1] == N+1
8423 * If N is UV_MAX (the highest representable code point on the machine), N+1 is
8424 * impossible to represent, so element [i+1] is omitted. The single element
8426 * invlist[0] == UV_MAX
8427 * contains just UV_MAX, but is interpreted as matching to infinity.
8429 * Taking the complement (inverting) an inversion list is quite simple, if the
8430 * first element is 0, remove it; otherwise add a 0 element at the beginning.
8431 * This implementation reserves an element at the beginning of each inversion
8432 * list to always contain 0; there is an additional flag in the header which
8433 * indicates if the list begins at the 0, or is offset to begin at the next
8434 * element. This means that the inversion list can be inverted without any
8435 * copying; just flip the flag.
8437 * More about inversion lists can be found in "Unicode Demystified"
8438 * Chapter 13 by Richard Gillam, published by Addison-Wesley.
8440 * The inversion list data structure is currently implemented as an SV pointing
8441 * to an array of UVs that the SV thinks are bytes. This allows us to have an
8442 * array of UV whose memory management is automatically handled by the existing
8443 * facilities for SV's.
8445 * Some of the methods should always be private to the implementation, and some
8446 * should eventually be made public */
8448 /* The header definitions are in F<invlist_inline.h> */
8450 #ifndef PERL_IN_XSUB_RE
8452 PERL_STATIC_INLINE UV*
8453 S__invlist_array_init(SV* const invlist, const bool will_have_0)
8455 /* Returns a pointer to the first element in the inversion list's array.
8456 * This is called upon initialization of an inversion list. Where the
8457 * array begins depends on whether the list has the code point U+0000 in it
8458 * or not. The other parameter tells it whether the code that follows this
8459 * call is about to put a 0 in the inversion list or not. The first
8460 * element is either the element reserved for 0, if TRUE, or the element
8461 * after it, if FALSE */
8463 bool* offset = get_invlist_offset_addr(invlist);
8464 UV* zero_addr = (UV *) SvPVX(invlist);
8466 PERL_ARGS_ASSERT__INVLIST_ARRAY_INIT;
8469 assert(! _invlist_len(invlist));
8473 /* 1^1 = 0; 1^0 = 1 */
8474 *offset = 1 ^ will_have_0;
8475 return zero_addr + *offset;
8480 PERL_STATIC_INLINE void
8481 S_invlist_set_len(pTHX_ SV* const invlist, const UV len, const bool offset)
8483 /* Sets the current number of elements stored in the inversion list.
8484 * Updates SvCUR correspondingly */
8485 PERL_UNUSED_CONTEXT;
8486 PERL_ARGS_ASSERT_INVLIST_SET_LEN;
8488 assert(SvTYPE(invlist) == SVt_INVLIST);
8493 : TO_INTERNAL_SIZE(len + offset));
8494 assert(SvLEN(invlist) == 0 || SvCUR(invlist) <= SvLEN(invlist));
8497 #ifndef PERL_IN_XSUB_RE
8500 S_invlist_replace_list_destroys_src(pTHX_ SV * dest, SV * src)
8502 /* Replaces the inversion list in 'dest' with the one from 'src'. It
8503 * steals the list from 'src', so 'src' is made to have a NULL list. This
8504 * is similar to what SvSetMagicSV() would do, if it were implemented on
8505 * inversion lists, though this routine avoids a copy */
8507 const UV src_len = _invlist_len(src);
8508 const bool src_offset = *get_invlist_offset_addr(src);
8509 const STRLEN src_byte_len = SvLEN(src);
8510 char * array = SvPVX(src);
8512 const int oldtainted = TAINT_get;
8514 PERL_ARGS_ASSERT_INVLIST_REPLACE_LIST_DESTROYS_SRC;
8516 assert(SvTYPE(src) == SVt_INVLIST);
8517 assert(SvTYPE(dest) == SVt_INVLIST);
8518 assert(! invlist_is_iterating(src));
8519 assert(SvCUR(src) == 0 || SvCUR(src) < SvLEN(src));
8521 /* Make sure it ends in the right place with a NUL, as our inversion list
8522 * manipulations aren't careful to keep this true, but sv_usepvn_flags()
8524 array[src_byte_len - 1] = '\0';
8526 TAINT_NOT; /* Otherwise it breaks */
8527 sv_usepvn_flags(dest,
8531 /* This flag is documented to cause a copy to be avoided */
8532 SV_HAS_TRAILING_NUL);
8533 TAINT_set(oldtainted);
8538 /* Finish up copying over the other fields in an inversion list */
8539 *get_invlist_offset_addr(dest) = src_offset;
8540 invlist_set_len(dest, src_len, src_offset);
8541 *get_invlist_previous_index_addr(dest) = 0;
8542 invlist_iterfinish(dest);
8545 PERL_STATIC_INLINE IV*
8546 S_get_invlist_previous_index_addr(SV* invlist)
8548 /* Return the address of the IV that is reserved to hold the cached index
8550 PERL_ARGS_ASSERT_GET_INVLIST_PREVIOUS_INDEX_ADDR;
8552 assert(SvTYPE(invlist) == SVt_INVLIST);
8554 return &(((XINVLIST*) SvANY(invlist))->prev_index);
8557 PERL_STATIC_INLINE IV
8558 S_invlist_previous_index(SV* const invlist)
8560 /* Returns cached index of previous search */
8562 PERL_ARGS_ASSERT_INVLIST_PREVIOUS_INDEX;
8564 return *get_invlist_previous_index_addr(invlist);
8567 PERL_STATIC_INLINE void
8568 S_invlist_set_previous_index(SV* const invlist, const IV index)
8570 /* Caches <index> for later retrieval */
8572 PERL_ARGS_ASSERT_INVLIST_SET_PREVIOUS_INDEX;
8574 assert(index == 0 || index < (int) _invlist_len(invlist));
8576 *get_invlist_previous_index_addr(invlist) = index;
8579 PERL_STATIC_INLINE void
8580 S_invlist_trim(SV* invlist)
8582 /* Free the not currently-being-used space in an inversion list */
8584 /* But don't free up the space needed for the 0 UV that is always at the
8585 * beginning of the list, nor the trailing NUL */
8586 const UV min_size = TO_INTERNAL_SIZE(1) + 1;
8588 PERL_ARGS_ASSERT_INVLIST_TRIM;
8590 assert(SvTYPE(invlist) == SVt_INVLIST);
8592 SvPV_renew(invlist, MAX(min_size, SvCUR(invlist) + 1));
8595 PERL_STATIC_INLINE void
8596 S_invlist_clear(pTHX_ SV* invlist) /* Empty the inversion list */
8598 PERL_ARGS_ASSERT_INVLIST_CLEAR;
8600 assert(SvTYPE(invlist) == SVt_INVLIST);
8602 invlist_set_len(invlist, 0, 0);
8603 invlist_trim(invlist);
8606 #endif /* ifndef PERL_IN_XSUB_RE */
8608 PERL_STATIC_INLINE bool
8609 S_invlist_is_iterating(SV* const invlist)
8611 PERL_ARGS_ASSERT_INVLIST_IS_ITERATING;
8613 return *(get_invlist_iter_addr(invlist)) < (STRLEN) UV_MAX;
8616 #ifndef PERL_IN_XSUB_RE
8618 PERL_STATIC_INLINE UV
8619 S_invlist_max(SV* const invlist)
8621 /* Returns the maximum number of elements storable in the inversion list's
8622 * array, without having to realloc() */
8624 PERL_ARGS_ASSERT_INVLIST_MAX;
8626 assert(SvTYPE(invlist) == SVt_INVLIST);
8628 /* Assumes worst case, in which the 0 element is not counted in the
8629 * inversion list, so subtracts 1 for that */
8630 return SvLEN(invlist) == 0 /* This happens under _new_invlist_C_array */
8631 ? FROM_INTERNAL_SIZE(SvCUR(invlist)) - 1
8632 : FROM_INTERNAL_SIZE(SvLEN(invlist)) - 1;
8635 Perl__new_invlist(pTHX_ IV initial_size)
8638 /* Return a pointer to a newly constructed inversion list, with enough
8639 * space to store 'initial_size' elements. If that number is negative, a
8640 * system default is used instead */
8644 if (initial_size < 0) {
8648 /* Allocate the initial space */
8649 new_list = newSV_type(SVt_INVLIST);
8651 /* First 1 is in case the zero element isn't in the list; second 1 is for
8653 SvGROW(new_list, TO_INTERNAL_SIZE(initial_size + 1) + 1);
8654 invlist_set_len(new_list, 0, 0);
8656 /* Force iterinit() to be used to get iteration to work */
8657 *get_invlist_iter_addr(new_list) = (STRLEN) UV_MAX;
8659 *get_invlist_previous_index_addr(new_list) = 0;
8665 Perl__new_invlist_C_array(pTHX_ const UV* const list)
8667 /* Return a pointer to a newly constructed inversion list, initialized to
8668 * point to <list>, which has to be in the exact correct inversion list
8669 * form, including internal fields. Thus this is a dangerous routine that
8670 * should not be used in the wrong hands. The passed in 'list' contains
8671 * several header fields at the beginning that are not part of the
8672 * inversion list body proper */
8674 const STRLEN length = (STRLEN) list[0];
8675 const UV version_id = list[1];
8676 const bool offset = cBOOL(list[2]);
8677 #define HEADER_LENGTH 3
8678 /* If any of the above changes in any way, you must change HEADER_LENGTH
8679 * (if appropriate) and regenerate INVLIST_VERSION_ID by running
8680 * perl -E 'say int(rand 2**31-1)'
8682 #define INVLIST_VERSION_ID 148565664 /* This is a combination of a version and
8683 data structure type, so that one being
8684 passed in can be validated to be an
8685 inversion list of the correct vintage.
8688 SV* invlist = newSV_type(SVt_INVLIST);
8690 PERL_ARGS_ASSERT__NEW_INVLIST_C_ARRAY;
8692 if (version_id != INVLIST_VERSION_ID) {
8693 Perl_croak(aTHX_ "panic: Incorrect version for previously generated inversion list");
8696 /* The generated array passed in includes header elements that aren't part
8697 * of the list proper, so start it just after them */
8698 SvPV_set(invlist, (char *) (list + HEADER_LENGTH));
8700 SvLEN_set(invlist, 0); /* Means we own the contents, and the system
8701 shouldn't touch it */
8703 *(get_invlist_offset_addr(invlist)) = offset;
8705 /* The 'length' passed to us is the physical number of elements in the
8706 * inversion list. But if there is an offset the logical number is one
8708 invlist_set_len(invlist, length - offset, offset);
8710 invlist_set_previous_index(invlist, 0);
8712 /* Initialize the iteration pointer. */
8713 invlist_iterfinish(invlist);
8715 SvREADONLY_on(invlist);
8721 S_invlist_extend(pTHX_ SV* const invlist, const UV new_max)
8723 /* Grow the maximum size of an inversion list */
8725 PERL_ARGS_ASSERT_INVLIST_EXTEND;
8727 assert(SvTYPE(invlist) == SVt_INVLIST);
8729 /* Add one to account for the zero element at the beginning which may not
8730 * be counted by the calling parameters */
8731 SvGROW((SV *)invlist, TO_INTERNAL_SIZE(new_max + 1));
8735 S__append_range_to_invlist(pTHX_ SV* const invlist,
8736 const UV start, const UV end)
8738 /* Subject to change or removal. Append the range from 'start' to 'end' at
8739 * the end of the inversion list. The range must be above any existing
8743 UV max = invlist_max(invlist);
8744 UV len = _invlist_len(invlist);
8747 PERL_ARGS_ASSERT__APPEND_RANGE_TO_INVLIST;
8749 if (len == 0) { /* Empty lists must be initialized */
8750 offset = start != 0;
8751 array = _invlist_array_init(invlist, ! offset);
8754 /* Here, the existing list is non-empty. The current max entry in the
8755 * list is generally the first value not in the set, except when the
8756 * set extends to the end of permissible values, in which case it is
8757 * the first entry in that final set, and so this call is an attempt to
8758 * append out-of-order */
8760 UV final_element = len - 1;
8761 array = invlist_array(invlist);
8762 if ( array[final_element] > start
8763 || ELEMENT_RANGE_MATCHES_INVLIST(final_element))
8765 Perl_croak(aTHX_ "panic: attempting to append to an inversion list, but wasn't at the end of the list, final=%" UVuf ", start=%" UVuf ", match=%c",
8766 array[final_element], start,
8767 ELEMENT_RANGE_MATCHES_INVLIST(final_element) ? 't' : 'f');
8770 /* Here, it is a legal append. If the new range begins 1 above the end
8771 * of the range below it, it is extending the range below it, so the
8772 * new first value not in the set is one greater than the newly
8773 * extended range. */
8774 offset = *get_invlist_offset_addr(invlist);
8775 if (array[final_element] == start) {
8776 if (end != UV_MAX) {
8777 array[final_element] = end + 1;
8780 /* But if the end is the maximum representable on the machine,
8781 * assume that infinity was actually what was meant. Just let
8782 * the range that this would extend to have no end */
8783 invlist_set_len(invlist, len - 1, offset);
8789 /* Here the new range doesn't extend any existing set. Add it */
8791 len += 2; /* Includes an element each for the start and end of range */
8793 /* If wll overflow the existing space, extend, which may cause the array to
8796 invlist_extend(invlist, len);
8798 /* Have to set len here to avoid assert failure in invlist_array() */
8799 invlist_set_len(invlist, len, offset);
8801 array = invlist_array(invlist);
8804 invlist_set_len(invlist, len, offset);
8807 /* The next item on the list starts the range, the one after that is
8808 * one past the new range. */
8809 array[len - 2] = start;
8810 if (end != UV_MAX) {
8811 array[len - 1] = end + 1;
8814 /* But if the end is the maximum representable on the machine, just let
8815 * the range have no end */
8816 invlist_set_len(invlist, len - 1, offset);
8821 Perl__invlist_search(SV* const invlist, const UV cp)
8823 /* Searches the inversion list for the entry that contains the input code
8824 * point <cp>. If <cp> is not in the list, -1 is returned. Otherwise, the
8825 * return value is the index into the list's array of the range that
8826 * contains <cp>, that is, 'i' such that
8827 * array[i] <= cp < array[i+1]
8832 IV high = _invlist_len(invlist);
8833 const IV highest_element = high - 1;
8836 PERL_ARGS_ASSERT__INVLIST_SEARCH;
8838 /* If list is empty, return failure. */
8843 /* (We can't get the array unless we know the list is non-empty) */
8844 array = invlist_array(invlist);
8846 mid = invlist_previous_index(invlist);
8848 if (mid > highest_element) {
8849 mid = highest_element;
8852 /* <mid> contains the cache of the result of the previous call to this
8853 * function (0 the first time). See if this call is for the same result,
8854 * or if it is for mid-1. This is under the theory that calls to this
8855 * function will often be for related code points that are near each other.
8856 * And benchmarks show that caching gives better results. We also test
8857 * here if the code point is within the bounds of the list. These tests
8858 * replace others that would have had to be made anyway to make sure that
8859 * the array bounds were not exceeded, and these give us extra information
8860 * at the same time */
8861 if (cp >= array[mid]) {
8862 if (cp >= array[highest_element]) {
8863 return highest_element;
8866 /* Here, array[mid] <= cp < array[highest_element]. This means that
8867 * the final element is not the answer, so can exclude it; it also
8868 * means that <mid> is not the final element, so can refer to 'mid + 1'
8870 if (cp < array[mid + 1]) {
8876 else { /* cp < aray[mid] */
8877 if (cp < array[0]) { /* Fail if outside the array */
8881 if (cp >= array[mid - 1]) {
8886 /* Binary search. What we are looking for is <i> such that
8887 * array[i] <= cp < array[i+1]
8888 * The loop below converges on the i+1. Note that there may not be an
8889 * (i+1)th element in the array, and things work nonetheless */
8890 while (low < high) {
8891 mid = (low + high) / 2;
8892 assert(mid <= highest_element);
8893 if (array[mid] <= cp) { /* cp >= array[mid] */
8896 /* We could do this extra test to exit the loop early.
8897 if (cp < array[low]) {
8902 else { /* cp < array[mid] */
8909 invlist_set_previous_index(invlist, high);
8914 Perl__invlist_populate_swatch(SV* const invlist,
8915 const UV start, const UV end, U8* swatch)
8917 /* populates a swatch of a swash the same way swatch_get() does in utf8.c,
8918 * but is used when the swash has an inversion list. This makes this much
8919 * faster, as it uses a binary search instead of a linear one. This is
8920 * intimately tied to that function, and perhaps should be in utf8.c,
8921 * except it is intimately tied to inversion lists as well. It assumes
8922 * that <swatch> is all 0's on input */
8925 const IV len = _invlist_len(invlist);
8929 PERL_ARGS_ASSERT__INVLIST_POPULATE_SWATCH;
8931 if (len == 0) { /* Empty inversion list */
8935 array = invlist_array(invlist);
8937 /* Find which element it is */
8938 i = _invlist_search(invlist, start);
8940 /* We populate from <start> to <end> */
8941 while (current < end) {
8944 /* The inversion list gives the results for every possible code point
8945 * after the first one in the list. Only those ranges whose index is
8946 * even are ones that the inversion list matches. For the odd ones,
8947 * and if the initial code point is not in the list, we have to skip
8948 * forward to the next element */
8949 if (i == -1 || ! ELEMENT_RANGE_MATCHES_INVLIST(i)) {
8951 if (i >= len) { /* Finished if beyond the end of the array */
8955 if (current >= end) { /* Finished if beyond the end of what we
8957 if (LIKELY(end < UV_MAX)) {
8961 /* We get here when the upper bound is the maximum
8962 * representable on the machine, and we are looking for just
8963 * that code point. Have to special case it */
8965 goto join_end_of_list;
8968 assert(current >= start);
8970 /* The current range ends one below the next one, except don't go past
8973 upper = (i < len && array[i] < end) ? array[i] : end;
8975 /* Here we are in a range that matches. Populate a bit in the 3-bit U8
8976 * for each code point in it */
8977 for (; current < upper; current++) {
8978 const STRLEN offset = (STRLEN)(current - start);
8979 swatch[offset >> 3] |= 1 << (offset & 7);
8984 /* Quit if at the end of the list */
8987 /* But first, have to deal with the highest possible code point on
8988 * the platform. The previous code assumes that <end> is one
8989 * beyond where we want to populate, but that is impossible at the
8990 * platform's infinity, so have to handle it specially */
8991 if (UNLIKELY(end == UV_MAX && ELEMENT_RANGE_MATCHES_INVLIST(len-1)))
8993 const STRLEN offset = (STRLEN)(end - start);
8994 swatch[offset >> 3] |= 1 << (offset & 7);
8999 /* Advance to the next range, which will be for code points not in the
9008 Perl__invlist_union_maybe_complement_2nd(pTHX_ SV* const a, SV* const b,
9009 const bool complement_b, SV** output)
9011 /* Take the union of two inversion lists and point '*output' to it. On
9012 * input, '*output' MUST POINT TO NULL OR TO AN SV* INVERSION LIST (possibly
9013 * even 'a' or 'b'). If to an inversion list, the contents of the original
9014 * list will be replaced by the union. The first list, 'a', may be
9015 * NULL, in which case a copy of the second list is placed in '*output'.
9016 * If 'complement_b' is TRUE, the union is taken of the complement
9017 * (inversion) of 'b' instead of b itself.
9019 * The basis for this comes from "Unicode Demystified" Chapter 13 by
9020 * Richard Gillam, published by Addison-Wesley, and explained at some
9021 * length there. The preface says to incorporate its examples into your
9022 * code at your own risk.
9024 * The algorithm is like a merge sort. */
9026 const UV* array_a; /* a's array */
9028 UV len_a; /* length of a's array */
9031 SV* u; /* the resulting union */
9035 UV i_a = 0; /* current index into a's array */
9039 /* running count, as explained in the algorithm source book; items are
9040 * stopped accumulating and are output when the count changes to/from 0.
9041 * The count is incremented when we start a range that's in an input's set,
9042 * and decremented when we start a range that's not in a set. So this
9043 * variable can be 0, 1, or 2. When it is 0 neither input is in their set,
9044 * and hence nothing goes into the union; 1, just one of the inputs is in
9045 * its set (and its current range gets added to the union); and 2 when both
9046 * inputs are in their sets. */
9049 PERL_ARGS_ASSERT__INVLIST_UNION_MAYBE_COMPLEMENT_2ND;
9051 assert(*output == NULL || SvTYPE(*output) == SVt_INVLIST);
9053 len_b = _invlist_len(b);
9056 /* Here, 'b' is empty, hence it's complement is all possible code
9057 * points. So if the union includes the complement of 'b', it includes
9058 * everything, and we need not even look at 'a'. It's easiest to
9059 * create a new inversion list that matches everything. */
9061 SV* everything = _add_range_to_invlist(NULL, 0, UV_MAX);
9063 if (*output == NULL) { /* If the output didn't exist, just point it
9065 *output = everything;
9067 else { /* Otherwise, replace its contents with the new list */
9068 invlist_replace_list_destroys_src(*output, everything);
9069 SvREFCNT_dec_NN(everything);
9075 /* Here, we don't want the complement of 'b', and since 'b' is empty,
9076 * the union will come entirely from 'a'. If 'a' is NULL or empty, the
9077 * output will be empty */
9079 if (a == NULL || _invlist_len(a) == 0) {
9080 if (*output == NULL) {
9081 *output = _new_invlist(0);
9084 invlist_clear(*output);
9089 /* Here, 'a' is not empty, but 'b' is, so 'a' entirely determines the
9090 * union. We can just return a copy of 'a' if '*output' doesn't point
9091 * to an existing list */
9092 if (*output == NULL) {
9093 *output = invlist_clone(a);
9097 /* If the output is to overwrite 'a', we have a no-op, as it's
9103 /* Here, '*output' is to be overwritten by 'a' */
9104 u = invlist_clone(a);
9105 invlist_replace_list_destroys_src(*output, u);
9111 /* Here 'b' is not empty. See about 'a' */
9113 if (a == NULL || ((len_a = _invlist_len(a)) == 0)) {
9115 /* Here, 'a' is empty (and b is not). That means the union will come
9116 * entirely from 'b'. If '*output' is NULL, we can directly return a
9117 * clone of 'b'. Otherwise, we replace the contents of '*output' with
9120 SV ** dest = (*output == NULL) ? output : &u;
9121 *dest = invlist_clone(b);
9123 _invlist_invert(*dest);
9127 invlist_replace_list_destroys_src(*output, u);
9134 /* Here both lists exist and are non-empty */
9135 array_a = invlist_array(a);
9136 array_b = invlist_array(b);
9138 /* If are to take the union of 'a' with the complement of b, set it
9139 * up so are looking at b's complement. */
9142 /* To complement, we invert: if the first element is 0, remove it. To
9143 * do this, we just pretend the array starts one later */
9144 if (array_b[0] == 0) {
9150 /* But if the first element is not zero, we pretend the list starts
9151 * at the 0 that is always stored immediately before the array. */
9157 /* Size the union for the worst case: that the sets are completely
9159 u = _new_invlist(len_a + len_b);
9161 /* Will contain U+0000 if either component does */
9162 array_u = _invlist_array_init(u, ( len_a > 0 && array_a[0] == 0)
9163 || (len_b > 0 && array_b[0] == 0));
9165 /* Go through each input list item by item, stopping when have exhausted
9167 while (i_a < len_a && i_b < len_b) {
9168 UV cp; /* The element to potentially add to the union's array */
9169 bool cp_in_set; /* is it in the the input list's set or not */
9171 /* We need to take one or the other of the two inputs for the union.
9172 * Since we are merging two sorted lists, we take the smaller of the
9173 * next items. In case of a tie, we take first the one that is in its
9174 * set. If we first took the one not in its set, it would decrement
9175 * the count, possibly to 0 which would cause it to be output as ending
9176 * the range, and the next time through we would take the same number,
9177 * and output it again as beginning the next range. By doing it the
9178 * opposite way, there is no possibility that the count will be
9179 * momentarily decremented to 0, and thus the two adjoining ranges will
9180 * be seamlessly merged. (In a tie and both are in the set or both not
9181 * in the set, it doesn't matter which we take first.) */
9182 if ( array_a[i_a] < array_b[i_b]
9183 || ( array_a[i_a] == array_b[i_b]
9184 && ELEMENT_RANGE_MATCHES_INVLIST(i_a)))
9186 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_a);
9187 cp = array_a[i_a++];
9190 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_b);
9191 cp = array_b[i_b++];
9194 /* Here, have chosen which of the two inputs to look at. Only output
9195 * if the running count changes to/from 0, which marks the
9196 * beginning/end of a range that's in the set */
9199 array_u[i_u++] = cp;
9206 array_u[i_u++] = cp;
9212 /* The loop above increments the index into exactly one of the input lists
9213 * each iteration, and ends when either index gets to its list end. That
9214 * means the other index is lower than its end, and so something is
9215 * remaining in that one. We decrement 'count', as explained below, if
9216 * that list is in its set. (i_a and i_b each currently index the element
9217 * beyond the one we care about.) */
9218 if ( (i_a != len_a && PREV_RANGE_MATCHES_INVLIST(i_a))
9219 || (i_b != len_b && PREV_RANGE_MATCHES_INVLIST(i_b)))
9224 /* Above we decremented 'count' if the list that had unexamined elements in
9225 * it was in its set. This has made it so that 'count' being non-zero
9226 * means there isn't anything left to output; and 'count' equal to 0 means
9227 * that what is left to output is precisely that which is left in the
9228 * non-exhausted input list.
9230 * To see why, note first that the exhausted input obviously has nothing
9231 * left to add to the union. If it was in its set at its end, that means
9232 * the set extends from here to the platform's infinity, and hence so does
9233 * the union and the non-exhausted set is irrelevant. The exhausted set
9234 * also contributed 1 to 'count'. If 'count' was 2, it got decremented to
9235 * 1, but if it was 1, the non-exhausted set wasn't in its set, and so
9236 * 'count' remains at 1. This is consistent with the decremented 'count'
9237 * != 0 meaning there's nothing left to add to the union.
9239 * But if the exhausted input wasn't in its set, it contributed 0 to
9240 * 'count', and the rest of the union will be whatever the other input is.
9241 * If 'count' was 0, neither list was in its set, and 'count' remains 0;
9242 * otherwise it gets decremented to 0. This is consistent with 'count'
9243 * == 0 meaning the remainder of the union is whatever is left in the
9244 * non-exhausted list. */
9249 IV copy_count = len_a - i_a;
9250 if (copy_count > 0) { /* The non-exhausted input is 'a' */
9251 Copy(array_a + i_a, array_u + i_u, copy_count, UV);
9253 else { /* The non-exhausted input is b */
9254 copy_count = len_b - i_b;
9255 Copy(array_b + i_b, array_u + i_u, copy_count, UV);
9257 len_u = i_u + copy_count;
9260 /* Set the result to the final length, which can change the pointer to
9261 * array_u, so re-find it. (Note that it is unlikely that this will
9262 * change, as we are shrinking the space, not enlarging it) */
9263 if (len_u != _invlist_len(u)) {
9264 invlist_set_len(u, len_u, *get_invlist_offset_addr(u));
9266 array_u = invlist_array(u);
9269 if (*output == NULL) { /* Simply return the new inversion list */
9273 /* Otherwise, overwrite the inversion list that was in '*output'. We
9274 * could instead free '*output', and then set it to 'u', but experience
9275 * has shown [perl #127392] that if the input is a mortal, we can get a
9276 * huge build-up of these during regex compilation before they get
9278 invlist_replace_list_destroys_src(*output, u);
9286 Perl__invlist_intersection_maybe_complement_2nd(pTHX_ SV* const a, SV* const b,
9287 const bool complement_b, SV** i)
9289 /* Take the intersection of two inversion lists and point '*i' to it. On
9290 * input, '*i' MUST POINT TO NULL OR TO AN SV* INVERSION LIST (possibly
9291 * even 'a' or 'b'). If to an inversion list, the contents of the original
9292 * list will be replaced by the intersection. The first list, 'a', may be
9293 * NULL, in which case '*i' will be an empty list. If 'complement_b' is
9294 * TRUE, the result will be the intersection of 'a' and the complement (or
9295 * inversion) of 'b' instead of 'b' directly.
9297 * The basis for this comes from "Unicode Demystified" Chapter 13 by
9298 * Richard Gillam, published by Addison-Wesley, and explained at some
9299 * length there. The preface says to incorporate its examples into your
9300 * code at your own risk. In fact, it had bugs
9302 * The algorithm is like a merge sort, and is essentially the same as the
9306 const UV* array_a; /* a's array */
9308 UV len_a; /* length of a's array */
9311 SV* r; /* the resulting intersection */
9315 UV i_a = 0; /* current index into a's array */
9319 /* running count of how many of the two inputs are postitioned at ranges
9320 * that are in their sets. As explained in the algorithm source book,
9321 * items are stopped accumulating and are output when the count changes
9322 * to/from 2. The count is incremented when we start a range that's in an
9323 * input's set, and decremented when we start a range that's not in a set.
9324 * Only when it is 2 are we in the intersection. */
9327 PERL_ARGS_ASSERT__INVLIST_INTERSECTION_MAYBE_COMPLEMENT_2ND;
9329 assert(*i == NULL || SvTYPE(*i) == SVt_INVLIST);
9331 /* Special case if either one is empty */
9332 len_a = (a == NULL) ? 0 : _invlist_len(a);
9333 if ((len_a == 0) || ((len_b = _invlist_len(b)) == 0)) {
9334 if (len_a != 0 && complement_b) {
9336 /* Here, 'a' is not empty, therefore from the enclosing 'if', 'b'
9337 * must be empty. Here, also we are using 'b's complement, which
9338 * hence must be every possible code point. Thus the intersection
9341 if (*i == a) { /* No-op */
9346 *i = invlist_clone(a);
9350 r = invlist_clone(a);
9351 invlist_replace_list_destroys_src(*i, r);
9356 /* Here, 'a' or 'b' is empty and not using the complement of 'b'. The
9357 * intersection must be empty */
9359 *i = _new_invlist(0);
9367 /* Here both lists exist and are non-empty */
9368 array_a = invlist_array(a);
9369 array_b = invlist_array(b);
9371 /* If are to take the intersection of 'a' with the complement of b, set it
9372 * up so are looking at b's complement. */
9375 /* To complement, we invert: if the first element is 0, remove it. To
9376 * do this, we just pretend the array starts one later */
9377 if (array_b[0] == 0) {
9383 /* But if the first element is not zero, we pretend the list starts
9384 * at the 0 that is always stored immediately before the array. */
9390 /* Size the intersection for the worst case: that the intersection ends up
9391 * fragmenting everything to be completely disjoint */
9392 r= _new_invlist(len_a + len_b);
9394 /* Will contain U+0000 iff both components do */
9395 array_r = _invlist_array_init(r, len_a > 0 && array_a[0] == 0
9396 && len_b > 0 && array_b[0] == 0);
9398 /* Go through each list item by item, stopping when have exhausted one of
9400 while (i_a < len_a && i_b < len_b) {
9401 UV cp; /* The element to potentially add to the intersection's
9403 bool cp_in_set; /* Is it in the input list's set or not */
9405 /* We need to take one or the other of the two inputs for the
9406 * intersection. Since we are merging two sorted lists, we take the
9407 * smaller of the next items. In case of a tie, we take first the one
9408 * that is not in its set (a difference from the union algorithm). If
9409 * we first took the one in its set, it would increment the count,
9410 * possibly to 2 which would cause it to be output as starting a range
9411 * in the intersection, and the next time through we would take that
9412 * same number, and output it again as ending the set. By doing the
9413 * opposite of this, there is no possibility that the count will be
9414 * momentarily incremented to 2. (In a tie and both are in the set or
9415 * both not in the set, it doesn't matter which we take first.) */
9416 if ( array_a[i_a] < array_b[i_b]
9417 || ( array_a[i_a] == array_b[i_b]
9418 && ! ELEMENT_RANGE_MATCHES_INVLIST(i_a)))
9420 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_a);
9421 cp = array_a[i_a++];
9424 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_b);
9428 /* Here, have chosen which of the two inputs to look at. Only output
9429 * if the running count changes to/from 2, which marks the
9430 * beginning/end of a range that's in the intersection */
9434 array_r[i_r++] = cp;
9439 array_r[i_r++] = cp;
9446 /* The loop above increments the index into exactly one of the input lists
9447 * each iteration, and ends when either index gets to its list end. That
9448 * means the other index is lower than its end, and so something is
9449 * remaining in that one. We increment 'count', as explained below, if the
9450 * exhausted list was in its set. (i_a and i_b each currently index the
9451 * element beyond the one we care about.) */
9452 if ( (i_a == len_a && PREV_RANGE_MATCHES_INVLIST(i_a))
9453 || (i_b == len_b && PREV_RANGE_MATCHES_INVLIST(i_b)))
9458 /* Above we incremented 'count' if the exhausted list was in its set. This
9459 * has made it so that 'count' being below 2 means there is nothing left to
9460 * output; otheriwse what's left to add to the intersection is precisely
9461 * that which is left in the non-exhausted input list.
9463 * To see why, note first that the exhausted input obviously has nothing
9464 * left to affect the intersection. If it was in its set at its end, that
9465 * means the set extends from here to the platform's infinity, and hence
9466 * anything in the non-exhausted's list will be in the intersection, and
9467 * anything not in it won't be. Hence, the rest of the intersection is
9468 * precisely what's in the non-exhausted list The exhausted set also
9469 * contributed 1 to 'count', meaning 'count' was at least 1. Incrementing
9470 * it means 'count' is now at least 2. This is consistent with the
9471 * incremented 'count' being >= 2 means to add the non-exhausted list to
9474 * But if the exhausted input wasn't in its set, it contributed 0 to
9475 * 'count', and the intersection can't include anything further; the
9476 * non-exhausted set is irrelevant. 'count' was at most 1, and doesn't get
9477 * incremented. This is consistent with 'count' being < 2 meaning nothing
9478 * further to add to the intersection. */
9479 if (count < 2) { /* Nothing left to put in the intersection. */
9482 else { /* copy the non-exhausted list, unchanged. */
9483 IV copy_count = len_a - i_a;
9484 if (copy_count > 0) { /* a is the one with stuff left */
9485 Copy(array_a + i_a, array_r + i_r, copy_count, UV);
9487 else { /* b is the one with stuff left */
9488 copy_count = len_b - i_b;
9489 Copy(array_b + i_b, array_r + i_r, copy_count, UV);
9491 len_r = i_r + copy_count;
9494 /* Set the result to the final length, which can change the pointer to
9495 * array_r, so re-find it. (Note that it is unlikely that this will
9496 * change, as we are shrinking the space, not enlarging it) */
9497 if (len_r != _invlist_len(r)) {
9498 invlist_set_len(r, len_r, *get_invlist_offset_addr(r));
9500 array_r = invlist_array(r);
9503 if (*i == NULL) { /* Simply return the calculated intersection */
9506 else { /* Otherwise, replace the existing inversion list in '*i'. We could
9507 instead free '*i', and then set it to 'r', but experience has
9508 shown [perl #127392] that if the input is a mortal, we can get a
9509 huge build-up of these during regex compilation before they get
9512 invlist_replace_list_destroys_src(*i, r);
9524 Perl__add_range_to_invlist(pTHX_ SV* invlist, UV start, UV end)
9526 /* Add the range from 'start' to 'end' inclusive to the inversion list's
9527 * set. A pointer to the inversion list is returned. This may actually be
9528 * a new list, in which case the passed in one has been destroyed. The
9529 * passed-in inversion list can be NULL, in which case a new one is created
9530 * with just the one range in it. The new list is not necessarily
9531 * NUL-terminated. Space is not freed if the inversion list shrinks as a
9532 * result of this function. The gain would not be large, and in many
9533 * cases, this is called multiple times on a single inversion list, so
9534 * anything freed may almost immediately be needed again.
9536 * This used to mostly call the 'union' routine, but that is much more
9537 * heavyweight than really needed for a single range addition */
9539 UV* array; /* The array implementing the inversion list */
9540 UV len; /* How many elements in 'array' */
9541 SSize_t i_s; /* index into the invlist array where 'start'
9543 SSize_t i_e = 0; /* And the index where 'end' should go */
9544 UV cur_highest; /* The highest code point in the inversion list
9545 upon entry to this function */
9547 /* This range becomes the whole inversion list if none already existed */
9548 if (invlist == NULL) {
9549 invlist = _new_invlist(2);
9550 _append_range_to_invlist(invlist, start, end);
9554 /* Likewise, if the inversion list is currently empty */
9555 len = _invlist_len(invlist);
9557 _append_range_to_invlist(invlist, start, end);
9561 /* Starting here, we have to know the internals of the list */
9562 array = invlist_array(invlist);
9564 /* If the new range ends higher than the current highest ... */
9565 cur_highest = invlist_highest(invlist);
9566 if (end > cur_highest) {
9568 /* If the whole range is higher, we can just append it */
9569 if (start > cur_highest) {
9570 _append_range_to_invlist(invlist, start, end);
9574 /* Otherwise, add the portion that is higher ... */
9575 _append_range_to_invlist(invlist, cur_highest + 1, end);
9577 /* ... and continue on below to handle the rest. As a result of the
9578 * above append, we know that the index of the end of the range is the
9579 * final even numbered one of the array. Recall that the final element
9580 * always starts a range that extends to infinity. If that range is in
9581 * the set (meaning the set goes from here to infinity), it will be an
9582 * even index, but if it isn't in the set, it's odd, and the final
9583 * range in the set is one less, which is even. */
9584 if (end == UV_MAX) {
9592 /* We have dealt with appending, now see about prepending. If the new
9593 * range starts lower than the current lowest ... */
9594 if (start < array[0]) {
9596 /* Adding something which has 0 in it is somewhat tricky, and uncommon.
9597 * Let the union code handle it, rather than having to know the
9598 * trickiness in two code places. */
9599 if (UNLIKELY(start == 0)) {
9602 range_invlist = _new_invlist(2);
9603 _append_range_to_invlist(range_invlist, start, end);
9605 _invlist_union(invlist, range_invlist, &invlist);
9607 SvREFCNT_dec_NN(range_invlist);
9612 /* If the whole new range comes before the first entry, and doesn't
9613 * extend it, we have to insert it as an additional range */
9614 if (end < array[0] - 1) {
9616 goto splice_in_new_range;
9619 /* Here the new range adjoins the existing first range, extending it
9623 /* And continue on below to handle the rest. We know that the index of
9624 * the beginning of the range is the first one of the array */
9627 else { /* Not prepending any part of the new range to the existing list.
9628 * Find where in the list it should go. This finds i_s, such that:
9629 * invlist[i_s] <= start < array[i_s+1]
9631 i_s = _invlist_search(invlist, start);
9634 /* At this point, any extending before the beginning of the inversion list
9635 * and/or after the end has been done. This has made it so that, in the
9636 * code below, each endpoint of the new range is either in a range that is
9637 * in the set, or is in a gap between two ranges that are. This means we
9638 * don't have to worry about exceeding the array bounds.
9640 * Find where in the list the new range ends (but we can skip this if we
9641 * have already determined what it is, or if it will be the same as i_s,
9642 * which we already have computed) */
9644 i_e = (start == end)
9646 : _invlist_search(invlist, end);
9649 /* Here generally invlist[i_e] <= end < array[i_e+1]. But if invlist[i_e]
9650 * is a range that goes to infinity there is no element at invlist[i_e+1],
9651 * so only the first relation holds. */
9653 if ( ! ELEMENT_RANGE_MATCHES_INVLIST(i_s)) {
9655 /* Here, the ranges on either side of the beginning of the new range
9656 * are in the set, and this range starts in the gap between them.
9658 * The new range extends the range above it downwards if the new range
9659 * ends at or above that range's start */
9660 const bool extends_the_range_above = ( end == UV_MAX
9661 || end + 1 >= array[i_s+1]);
9663 /* The new range extends the range below it upwards if it begins just
9664 * after where that range ends */
9665 if (start == array[i_s]) {
9667 /* If the new range fills the entire gap between the other ranges,
9668 * they will get merged together. Other ranges may also get
9669 * merged, depending on how many of them the new range spans. In
9670 * the general case, we do the merge later, just once, after we
9671 * figure out how many to merge. But in the case where the new
9672 * range exactly spans just this one gap (possibly extending into
9673 * the one above), we do the merge here, and an early exit. This
9674 * is done here to avoid having to special case later. */
9675 if (i_e - i_s <= 1) {
9677 /* If i_e - i_s == 1, it means that the new range terminates
9678 * within the range above, and hence 'extends_the_range_above'
9679 * must be true. (If the range above it extends to infinity,
9680 * 'i_s+2' will be above the array's limit, but 'len-i_s-2'
9681 * will be 0, so no harm done.) */
9682 if (extends_the_range_above) {
9683 Move(array + i_s + 2, array + i_s, len - i_s - 2, UV);
9684 invlist_set_len(invlist,
9686 *(get_invlist_offset_addr(invlist)));
9690 /* Here, i_e must == i_s. We keep them in sync, as they apply
9691 * to the same range, and below we are about to decrement i_s
9696 /* Here, the new range is adjacent to the one below. (It may also
9697 * span beyond the range above, but that will get resolved later.)
9698 * Extend the range below to include this one. */
9699 array[i_s] = (end == UV_MAX) ? UV_MAX : end + 1;
9703 else if (extends_the_range_above) {
9705 /* Here the new range only extends the range above it, but not the
9706 * one below. It merges with the one above. Again, we keep i_e
9707 * and i_s in sync if they point to the same range */
9716 /* Here, we've dealt with the new range start extending any adjoining
9719 * If the new range extends to infinity, it is now the final one,
9720 * regardless of what was there before */
9721 if (UNLIKELY(end == UV_MAX)) {
9722 invlist_set_len(invlist, i_s + 1, *(get_invlist_offset_addr(invlist)));
9726 /* If i_e started as == i_s, it has also been dealt with,
9727 * and been updated to the new i_s, which will fail the following if */
9728 if (! ELEMENT_RANGE_MATCHES_INVLIST(i_e)) {
9730 /* Here, the ranges on either side of the end of the new range are in
9731 * the set, and this range ends in the gap between them.
9733 * If this range is adjacent to (hence extends) the range above it, it
9734 * becomes part of that range; likewise if it extends the range below,
9735 * it becomes part of that range */
9736 if (end + 1 == array[i_e+1]) {
9740 else if (start <= array[i_e]) {
9741 array[i_e] = end + 1;
9748 /* If the range fits entirely in an existing range (as possibly already
9749 * extended above), it doesn't add anything new */
9750 if (ELEMENT_RANGE_MATCHES_INVLIST(i_s)) {
9754 /* Here, no part of the range is in the list. Must add it. It will
9755 * occupy 2 more slots */
9756 splice_in_new_range:
9758 invlist_extend(invlist, len + 2);
9759 array = invlist_array(invlist);
9760 /* Move the rest of the array down two slots. Don't include any
9762 Move(array + i_e + 1, array + i_e + 3, len - i_e - 1, UV);
9764 /* Do the actual splice */
9765 array[i_e+1] = start;
9766 array[i_e+2] = end + 1;
9767 invlist_set_len(invlist, len + 2, *(get_invlist_offset_addr(invlist)));
9771 /* Here the new range crossed the boundaries of a pre-existing range. The
9772 * code above has adjusted things so that both ends are in ranges that are
9773 * in the set. This means everything in between must also be in the set.
9774 * Just squash things together */
9775 Move(array + i_e + 1, array + i_s + 1, len - i_e - 1, UV);
9776 invlist_set_len(invlist,
9778 *(get_invlist_offset_addr(invlist)));
9784 Perl__setup_canned_invlist(pTHX_ const STRLEN size, const UV element0,
9785 UV** other_elements_ptr)
9787 /* Create and return an inversion list whose contents are to be populated
9788 * by the caller. The caller gives the number of elements (in 'size') and
9789 * the very first element ('element0'). This function will set
9790 * '*other_elements_ptr' to an array of UVs, where the remaining elements
9793 * Obviously there is some trust involved that the caller will properly
9794 * fill in the other elements of the array.
9796 * (The first element needs to be passed in, as the underlying code does
9797 * things differently depending on whether it is zero or non-zero) */
9799 SV* invlist = _new_invlist(size);
9802 PERL_ARGS_ASSERT__SETUP_CANNED_INVLIST;
9804 invlist = add_cp_to_invlist(invlist, element0);
9805 offset = *get_invlist_offset_addr(invlist);
9807 invlist_set_len(invlist, size, offset);
9808 *other_elements_ptr = invlist_array(invlist) + 1;
9814 PERL_STATIC_INLINE SV*
9815 S_add_cp_to_invlist(pTHX_ SV* invlist, const UV cp) {
9816 return _add_range_to_invlist(invlist, cp, cp);
9819 #ifndef PERL_IN_XSUB_RE
9821 Perl__invlist_invert(pTHX_ SV* const invlist)
9823 /* Complement the input inversion list. This adds a 0 if the list didn't
9824 * have a zero; removes it otherwise. As described above, the data
9825 * structure is set up so that this is very efficient */
9827 PERL_ARGS_ASSERT__INVLIST_INVERT;
9829 assert(! invlist_is_iterating(invlist));
9831 /* The inverse of matching nothing is matching everything */
9832 if (_invlist_len(invlist) == 0) {
9833 _append_range_to_invlist(invlist, 0, UV_MAX);
9837 *get_invlist_offset_addr(invlist) = ! *get_invlist_offset_addr(invlist);
9842 PERL_STATIC_INLINE SV*
9843 S_invlist_clone(pTHX_ SV* const invlist)
9846 /* Return a new inversion list that is a copy of the input one, which is
9847 * unchanged. The new list will not be mortal even if the old one was. */
9849 /* Need to allocate extra space to accommodate Perl's addition of a
9850 * trailing NUL to SvPV's, since it thinks they are always strings */
9851 SV* new_invlist = _new_invlist(_invlist_len(invlist) + 1);
9852 STRLEN physical_length = SvCUR(invlist);
9853 bool offset = *(get_invlist_offset_addr(invlist));
9855 PERL_ARGS_ASSERT_INVLIST_CLONE;
9857 *(get_invlist_offset_addr(new_invlist)) = offset;
9858 invlist_set_len(new_invlist, _invlist_len(invlist), offset);
9859 Copy(SvPVX(invlist), SvPVX(new_invlist), physical_length, char);
9864 PERL_STATIC_INLINE STRLEN*
9865 S_get_invlist_iter_addr(SV* invlist)
9867 /* Return the address of the UV that contains the current iteration
9870 PERL_ARGS_ASSERT_GET_INVLIST_ITER_ADDR;
9872 assert(SvTYPE(invlist) == SVt_INVLIST);
9874 return &(((XINVLIST*) SvANY(invlist))->iterator);
9877 PERL_STATIC_INLINE void
9878 S_invlist_iterinit(SV* invlist) /* Initialize iterator for invlist */
9880 PERL_ARGS_ASSERT_INVLIST_ITERINIT;
9882 *get_invlist_iter_addr(invlist) = 0;
9885 PERL_STATIC_INLINE void
9886 S_invlist_iterfinish(SV* invlist)
9888 /* Terminate iterator for invlist. This is to catch development errors.
9889 * Any iteration that is interrupted before completed should call this
9890 * function. Functions that add code points anywhere else but to the end
9891 * of an inversion list assert that they are not in the middle of an
9892 * iteration. If they were, the addition would make the iteration
9893 * problematical: if the iteration hadn't reached the place where things
9894 * were being added, it would be ok */
9896 PERL_ARGS_ASSERT_INVLIST_ITERFINISH;
9898 *get_invlist_iter_addr(invlist) = (STRLEN) UV_MAX;
9902 S_invlist_iternext(SV* invlist, UV* start, UV* end)
9904 /* An C<invlist_iterinit> call on <invlist> must be used to set this up.
9905 * This call sets in <*start> and <*end>, the next range in <invlist>.
9906 * Returns <TRUE> if successful and the next call will return the next
9907 * range; <FALSE> if was already at the end of the list. If the latter,
9908 * <*start> and <*end> are unchanged, and the next call to this function
9909 * will start over at the beginning of the list */
9911 STRLEN* pos = get_invlist_iter_addr(invlist);
9912 UV len = _invlist_len(invlist);
9915 PERL_ARGS_ASSERT_INVLIST_ITERNEXT;
9918 *pos = (STRLEN) UV_MAX; /* Force iterinit() to be required next time */
9922 array = invlist_array(invlist);
9924 *start = array[(*pos)++];
9930 *end = array[(*pos)++] - 1;
9936 PERL_STATIC_INLINE UV
9937 S_invlist_highest(SV* const invlist)
9939 /* Returns the highest code point that matches an inversion list. This API
9940 * has an ambiguity, as it returns 0 under either the highest is actually
9941 * 0, or if the list is empty. If this distinction matters to you, check
9942 * for emptiness before calling this function */
9944 UV len = _invlist_len(invlist);
9947 PERL_ARGS_ASSERT_INVLIST_HIGHEST;
9953 array = invlist_array(invlist);
9955 /* The last element in the array in the inversion list always starts a
9956 * range that goes to infinity. That range may be for code points that are
9957 * matched in the inversion list, or it may be for ones that aren't
9958 * matched. In the latter case, the highest code point in the set is one
9959 * less than the beginning of this range; otherwise it is the final element
9960 * of this range: infinity */
9961 return (ELEMENT_RANGE_MATCHES_INVLIST(len - 1))
9963 : array[len - 1] - 1;
9967 S_invlist_contents(pTHX_ SV* const invlist, const bool traditional_style)
9969 /* Get the contents of an inversion list into a string SV so that they can
9970 * be printed out. If 'traditional_style' is TRUE, it uses the format
9971 * traditionally done for debug tracing; otherwise it uses a format
9972 * suitable for just copying to the output, with blanks between ranges and
9973 * a dash between range components */
9977 const char intra_range_delimiter = (traditional_style ? '\t' : '-');
9978 const char inter_range_delimiter = (traditional_style ? '\n' : ' ');
9980 if (traditional_style) {
9981 output = newSVpvs("\n");
9984 output = newSVpvs("");
9987 PERL_ARGS_ASSERT_INVLIST_CONTENTS;
9989 assert(! invlist_is_iterating(invlist));
9991 invlist_iterinit(invlist);
9992 while (invlist_iternext(invlist, &start, &end)) {
9993 if (end == UV_MAX) {
9994 Perl_sv_catpvf(aTHX_ output, "%04" UVXf "%cINFINITY%c",
9995 start, intra_range_delimiter,
9996 inter_range_delimiter);
9998 else if (end != start) {
9999 Perl_sv_catpvf(aTHX_ output, "%04" UVXf "%c%04" UVXf "%c",
10001 intra_range_delimiter,
10002 end, inter_range_delimiter);
10005 Perl_sv_catpvf(aTHX_ output, "%04" UVXf "%c",
10006 start, inter_range_delimiter);
10010 if (SvCUR(output) && ! traditional_style) {/* Get rid of trailing blank */
10011 SvCUR_set(output, SvCUR(output) - 1);
10017 #ifndef PERL_IN_XSUB_RE
10019 Perl__invlist_dump(pTHX_ PerlIO *file, I32 level,
10020 const char * const indent, SV* const invlist)
10022 /* Designed to be called only by do_sv_dump(). Dumps out the ranges of the
10023 * inversion list 'invlist' to 'file' at 'level' Each line is prefixed by
10024 * the string 'indent'. The output looks like this:
10025 [0] 0x000A .. 0x000D
10027 [4] 0x2028 .. 0x2029
10028 [6] 0x3104 .. INFINITY
10029 * This means that the first range of code points matched by the list are
10030 * 0xA through 0xD; the second range contains only the single code point
10031 * 0x85, etc. An inversion list is an array of UVs. Two array elements
10032 * are used to define each range (except if the final range extends to
10033 * infinity, only a single element is needed). The array index of the
10034 * first element for the corresponding range is given in brackets. */
10039 PERL_ARGS_ASSERT__INVLIST_DUMP;
10041 if (invlist_is_iterating(invlist)) {
10042 Perl_dump_indent(aTHX_ level, file,
10043 "%sCan't dump inversion list because is in middle of iterating\n",
10048 invlist_iterinit(invlist);
10049 while (invlist_iternext(invlist, &start, &end)) {
10050 if (end == UV_MAX) {
10051 Perl_dump_indent(aTHX_ level, file,
10052 "%s[%" UVuf "] 0x%04" UVXf " .. INFINITY\n",
10053 indent, (UV)count, start);
10055 else if (end != start) {
10056 Perl_dump_indent(aTHX_ level, file,
10057 "%s[%" UVuf "] 0x%04" UVXf " .. 0x%04" UVXf "\n",
10058 indent, (UV)count, start, end);
10061 Perl_dump_indent(aTHX_ level, file, "%s[%" UVuf "] 0x%04" UVXf "\n",
10062 indent, (UV)count, start);
10069 Perl__load_PL_utf8_foldclosures (pTHX)
10071 assert(! PL_utf8_foldclosures);
10073 /* If the folds haven't been read in, call a fold function
10075 if (! PL_utf8_tofold) {
10076 U8 dummy[UTF8_MAXBYTES_CASE+1];
10077 const U8 hyphen[] = HYPHEN_UTF8;
10079 /* This string is just a short named one above \xff */
10080 toFOLD_utf8_safe(hyphen, hyphen + sizeof(hyphen) - 1, dummy, NULL);
10081 assert(PL_utf8_tofold); /* Verify that worked */
10083 PL_utf8_foldclosures = _swash_inversion_hash(PL_utf8_tofold);
10087 #if defined(PERL_ARGS_ASSERT__INVLISTEQ) && !defined(PERL_IN_XSUB_RE)
10089 Perl__invlistEQ(pTHX_ SV* const a, SV* const b, const bool complement_b)
10091 /* Return a boolean as to if the two passed in inversion lists are
10092 * identical. The final argument, if TRUE, says to take the complement of
10093 * the second inversion list before doing the comparison */
10095 const UV* array_a = invlist_array(a);
10096 const UV* array_b = invlist_array(b);
10097 UV len_a = _invlist_len(a);
10098 UV len_b = _invlist_len(b);
10100 PERL_ARGS_ASSERT__INVLISTEQ;
10102 /* If are to compare 'a' with the complement of b, set it
10103 * up so are looking at b's complement. */
10104 if (complement_b) {
10106 /* The complement of nothing is everything, so <a> would have to have
10107 * just one element, starting at zero (ending at infinity) */
10109 return (len_a == 1 && array_a[0] == 0);
10111 else if (array_b[0] == 0) {
10113 /* Otherwise, to complement, we invert. Here, the first element is
10114 * 0, just remove it. To do this, we just pretend the array starts
10122 /* But if the first element is not zero, we pretend the list starts
10123 * at the 0 that is always stored immediately before the array. */
10129 return len_a == len_b
10130 && memEQ(array_a, array_b, len_a * sizeof(array_a[0]));
10136 * As best we can, determine the characters that can match the start of
10137 * the given EXACTF-ish node.
10139 * Returns the invlist as a new SV*; it is the caller's responsibility to
10140 * call SvREFCNT_dec() when done with it.
10143 S__make_exactf_invlist(pTHX_ RExC_state_t *pRExC_state, regnode *node)
10145 const U8 * s = (U8*)STRING(node);
10146 SSize_t bytelen = STR_LEN(node);
10148 /* Start out big enough for 2 separate code points */
10149 SV* invlist = _new_invlist(4);
10151 PERL_ARGS_ASSERT__MAKE_EXACTF_INVLIST;
10156 /* We punt and assume can match anything if the node begins
10157 * with a multi-character fold. Things are complicated. For
10158 * example, /ffi/i could match any of:
10159 * "\N{LATIN SMALL LIGATURE FFI}"
10160 * "\N{LATIN SMALL LIGATURE FF}I"
10161 * "F\N{LATIN SMALL LIGATURE FI}"
10162 * plus several other things; and making sure we have all the
10163 * possibilities is hard. */
10164 if (is_MULTI_CHAR_FOLD_latin1_safe(s, s + bytelen)) {
10165 invlist = _add_range_to_invlist(invlist, 0, UV_MAX);
10168 /* Any Latin1 range character can potentially match any
10169 * other depending on the locale */
10170 if (OP(node) == EXACTFL) {
10171 _invlist_union(invlist, PL_Latin1, &invlist);
10174 /* But otherwise, it matches at least itself. We can
10175 * quickly tell if it has a distinct fold, and if so,
10176 * it matches that as well */
10177 invlist = add_cp_to_invlist(invlist, uc);
10178 if (IS_IN_SOME_FOLD_L1(uc))
10179 invlist = add_cp_to_invlist(invlist, PL_fold_latin1[uc]);
10182 /* Some characters match above-Latin1 ones under /i. This
10183 * is true of EXACTFL ones when the locale is UTF-8 */
10184 if (HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(uc)
10185 && (! isASCII(uc) || (OP(node) != EXACTFA
10186 && OP(node) != EXACTFA_NO_TRIE)))
10188 add_above_Latin1_folds(pRExC_state, (U8) uc, &invlist);
10192 else { /* Pattern is UTF-8 */
10193 U8 folded[UTF8_MAX_FOLD_CHAR_EXPAND * UTF8_MAXBYTES_CASE + 1] = { '\0' };
10194 STRLEN foldlen = UTF8SKIP(s);
10195 const U8* e = s + bytelen;
10198 uc = utf8_to_uvchr_buf(s, s + bytelen, NULL);
10200 /* The only code points that aren't folded in a UTF EXACTFish
10201 * node are are the problematic ones in EXACTFL nodes */
10202 if (OP(node) == EXACTFL && is_PROBLEMATIC_LOCALE_FOLDEDS_START_cp(uc)) {
10203 /* We need to check for the possibility that this EXACTFL
10204 * node begins with a multi-char fold. Therefore we fold
10205 * the first few characters of it so that we can make that
10210 for (i = 0; i < UTF8_MAX_FOLD_CHAR_EXPAND && s < e; i++) {
10212 *(d++) = (U8) toFOLD(*s);
10217 toFOLD_utf8_safe(s, e, d, &len);
10223 /* And set up so the code below that looks in this folded
10224 * buffer instead of the node's string */
10226 foldlen = UTF8SKIP(folded);
10230 /* When we reach here 's' points to the fold of the first
10231 * character(s) of the node; and 'e' points to far enough along
10232 * the folded string to be just past any possible multi-char
10233 * fold. 'foldlen' is the length in bytes of the first
10236 * Unlike the non-UTF-8 case, the macro for determining if a
10237 * string is a multi-char fold requires all the characters to
10238 * already be folded. This is because of all the complications
10239 * if not. Note that they are folded anyway, except in EXACTFL
10240 * nodes. Like the non-UTF case above, we punt if the node
10241 * begins with a multi-char fold */
10243 if (is_MULTI_CHAR_FOLD_utf8_safe(s, e)) {
10244 invlist = _add_range_to_invlist(invlist, 0, UV_MAX);
10246 else { /* Single char fold */
10248 /* It matches all the things that fold to it, which are
10249 * found in PL_utf8_foldclosures (including itself) */
10250 invlist = add_cp_to_invlist(invlist, uc);
10251 if (! PL_utf8_foldclosures)
10252 _load_PL_utf8_foldclosures();
10253 if ((listp = hv_fetch(PL_utf8_foldclosures,
10254 (char *) s, foldlen, FALSE)))
10256 AV* list = (AV*) *listp;
10258 for (k = 0; k <= av_tindex_skip_len_mg(list); k++) {
10259 SV** c_p = av_fetch(list, k, FALSE);
10265 /* /aa doesn't allow folds between ASCII and non- */
10266 if ((OP(node) == EXACTFA || OP(node) == EXACTFA_NO_TRIE)
10267 && isASCII(c) != isASCII(uc))
10272 invlist = add_cp_to_invlist(invlist, c);
10281 #undef HEADER_LENGTH
10282 #undef TO_INTERNAL_SIZE
10283 #undef FROM_INTERNAL_SIZE
10284 #undef INVLIST_VERSION_ID
10286 /* End of inversion list object */
10289 S_parse_lparen_question_flags(pTHX_ RExC_state_t *pRExC_state)
10291 /* This parses the flags that are in either the '(?foo)' or '(?foo:bar)'
10292 * constructs, and updates RExC_flags with them. On input, RExC_parse
10293 * should point to the first flag; it is updated on output to point to the
10294 * final ')' or ':'. There needs to be at least one flag, or this will
10297 /* for (?g), (?gc), and (?o) warnings; warning
10298 about (?c) will warn about (?g) -- japhy */
10300 #define WASTED_O 0x01
10301 #define WASTED_G 0x02
10302 #define WASTED_C 0x04
10303 #define WASTED_GC (WASTED_G|WASTED_C)
10304 I32 wastedflags = 0x00;
10305 U32 posflags = 0, negflags = 0;
10306 U32 *flagsp = &posflags;
10307 char has_charset_modifier = '\0';
10309 bool has_use_defaults = FALSE;
10310 const char* const seqstart = RExC_parse - 1; /* Point to the '?' */
10311 int x_mod_count = 0;
10313 PERL_ARGS_ASSERT_PARSE_LPAREN_QUESTION_FLAGS;
10315 /* '^' as an initial flag sets certain defaults */
10316 if (UCHARAT(RExC_parse) == '^') {
10318 has_use_defaults = TRUE;
10319 STD_PMMOD_FLAGS_CLEAR(&RExC_flags);
10320 set_regex_charset(&RExC_flags, (RExC_utf8 || RExC_uni_semantics)
10321 ? REGEX_UNICODE_CHARSET
10322 : REGEX_DEPENDS_CHARSET);
10325 cs = get_regex_charset(RExC_flags);
10326 if (cs == REGEX_DEPENDS_CHARSET
10327 && (RExC_utf8 || RExC_uni_semantics))
10329 cs = REGEX_UNICODE_CHARSET;
10332 while (RExC_parse < RExC_end) {
10333 /* && strchr("iogcmsx", *RExC_parse) */
10334 /* (?g), (?gc) and (?o) are useless here
10335 and must be globally applied -- japhy */
10336 switch (*RExC_parse) {
10338 /* Code for the imsxn flags */
10339 CASE_STD_PMMOD_FLAGS_PARSE_SET(flagsp, x_mod_count);
10341 case LOCALE_PAT_MOD:
10342 if (has_charset_modifier) {
10343 goto excess_modifier;
10345 else if (flagsp == &negflags) {
10348 cs = REGEX_LOCALE_CHARSET;
10349 has_charset_modifier = LOCALE_PAT_MOD;
10351 case UNICODE_PAT_MOD:
10352 if (has_charset_modifier) {
10353 goto excess_modifier;
10355 else if (flagsp == &negflags) {
10358 cs = REGEX_UNICODE_CHARSET;
10359 has_charset_modifier = UNICODE_PAT_MOD;
10361 case ASCII_RESTRICT_PAT_MOD:
10362 if (flagsp == &negflags) {
10365 if (has_charset_modifier) {
10366 if (cs != REGEX_ASCII_RESTRICTED_CHARSET) {
10367 goto excess_modifier;
10369 /* Doubled modifier implies more restricted */
10370 cs = REGEX_ASCII_MORE_RESTRICTED_CHARSET;
10373 cs = REGEX_ASCII_RESTRICTED_CHARSET;
10375 has_charset_modifier = ASCII_RESTRICT_PAT_MOD;
10377 case DEPENDS_PAT_MOD:
10378 if (has_use_defaults) {
10379 goto fail_modifiers;
10381 else if (flagsp == &negflags) {
10384 else if (has_charset_modifier) {
10385 goto excess_modifier;
10388 /* The dual charset means unicode semantics if the
10389 * pattern (or target, not known until runtime) are
10390 * utf8, or something in the pattern indicates unicode
10392 cs = (RExC_utf8 || RExC_uni_semantics)
10393 ? REGEX_UNICODE_CHARSET
10394 : REGEX_DEPENDS_CHARSET;
10395 has_charset_modifier = DEPENDS_PAT_MOD;
10399 if (has_charset_modifier == ASCII_RESTRICT_PAT_MOD) {
10400 vFAIL2("Regexp modifier \"%c\" may appear a maximum of twice", ASCII_RESTRICT_PAT_MOD);
10402 else if (has_charset_modifier == *(RExC_parse - 1)) {
10403 vFAIL2("Regexp modifier \"%c\" may not appear twice",
10404 *(RExC_parse - 1));
10407 vFAIL3("Regexp modifiers \"%c\" and \"%c\" are mutually exclusive", has_charset_modifier, *(RExC_parse - 1));
10409 NOT_REACHED; /*NOTREACHED*/
10412 vFAIL2("Regexp modifier \"%c\" may not appear after the \"-\"",
10413 *(RExC_parse - 1));
10414 NOT_REACHED; /*NOTREACHED*/
10415 case ONCE_PAT_MOD: /* 'o' */
10416 case GLOBAL_PAT_MOD: /* 'g' */
10417 if (PASS2 && ckWARN(WARN_REGEXP)) {
10418 const I32 wflagbit = *RExC_parse == 'o'
10421 if (! (wastedflags & wflagbit) ) {
10422 wastedflags |= wflagbit;
10423 /* diag_listed_as: Useless (?-%s) - don't use /%s modifier in regex; marked by <-- HERE in m/%s/ */
10426 "Useless (%s%c) - %suse /%c modifier",
10427 flagsp == &negflags ? "?-" : "?",
10429 flagsp == &negflags ? "don't " : "",
10436 case CONTINUE_PAT_MOD: /* 'c' */
10437 if (PASS2 && ckWARN(WARN_REGEXP)) {
10438 if (! (wastedflags & WASTED_C) ) {
10439 wastedflags |= WASTED_GC;
10440 /* diag_listed_as: Useless (?-%s) - don't use /%s modifier in regex; marked by <-- HERE in m/%s/ */
10443 "Useless (%sc) - %suse /gc modifier",
10444 flagsp == &negflags ? "?-" : "?",
10445 flagsp == &negflags ? "don't " : ""
10450 case KEEPCOPY_PAT_MOD: /* 'p' */
10451 if (flagsp == &negflags) {
10453 ckWARNreg(RExC_parse + 1,"Useless use of (?-p)");
10455 *flagsp |= RXf_PMf_KEEPCOPY;
10459 /* A flag is a default iff it is following a minus, so
10460 * if there is a minus, it means will be trying to
10461 * re-specify a default which is an error */
10462 if (has_use_defaults || flagsp == &negflags) {
10463 goto fail_modifiers;
10465 flagsp = &negflags;
10466 wastedflags = 0; /* reset so (?g-c) warns twice */
10472 if ((posflags & (RXf_PMf_EXTENDED|RXf_PMf_EXTENDED_MORE)) == RXf_PMf_EXTENDED) {
10473 negflags |= RXf_PMf_EXTENDED_MORE;
10475 RExC_flags |= posflags;
10477 if (negflags & RXf_PMf_EXTENDED) {
10478 negflags |= RXf_PMf_EXTENDED_MORE;
10480 RExC_flags &= ~negflags;
10481 set_regex_charset(&RExC_flags, cs);
10486 RExC_parse += SKIP_IF_CHAR(RExC_parse);
10487 /* diag_listed_as: Sequence (?%s...) not recognized in regex; marked by <-- HERE in m/%s/ */
10488 vFAIL2utf8f("Sequence (%" UTF8f "...) not recognized",
10489 UTF8fARG(UTF, RExC_parse-seqstart, seqstart));
10490 NOT_REACHED; /*NOTREACHED*/
10493 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
10496 vFAIL("Sequence (?... not terminated");
10500 - reg - regular expression, i.e. main body or parenthesized thing
10502 * Caller must absorb opening parenthesis.
10504 * Combining parenthesis handling with the base level of regular expression
10505 * is a trifle forced, but the need to tie the tails of the branches to what
10506 * follows makes it hard to avoid.
10508 #define REGTAIL(x,y,z) regtail((x),(y),(z),depth+1)
10510 #define REGTAIL_STUDY(x,y,z) regtail_study((x),(y),(z),depth+1)
10512 #define REGTAIL_STUDY(x,y,z) regtail((x),(y),(z),depth+1)
10515 PERL_STATIC_INLINE regnode *
10516 S_handle_named_backref(pTHX_ RExC_state_t *pRExC_state,
10518 char * parse_start,
10523 char* name_start = RExC_parse;
10525 SV *sv_dat = reg_scan_name(pRExC_state, SIZE_ONLY
10526 ? REG_RSN_RETURN_NULL
10527 : REG_RSN_RETURN_DATA);
10528 GET_RE_DEBUG_FLAGS_DECL;
10530 PERL_ARGS_ASSERT_HANDLE_NAMED_BACKREF;
10532 if (RExC_parse == name_start || *RExC_parse != ch) {
10533 /* diag_listed_as: Sequence \%s... not terminated in regex; marked by <-- HERE in m/%s/ */
10534 vFAIL2("Sequence %.3s... not terminated",parse_start);
10538 num = add_data( pRExC_state, STR_WITH_LEN("S"));
10539 RExC_rxi->data->data[num]=(void*)sv_dat;
10540 SvREFCNT_inc_simple_void(sv_dat);
10543 ret = reganode(pRExC_state,
10546 : (ASCII_FOLD_RESTRICTED)
10548 : (AT_LEAST_UNI_SEMANTICS)
10554 *flagp |= HASWIDTH;
10556 Set_Node_Offset(ret, parse_start+1);
10557 Set_Node_Cur_Length(ret, parse_start);
10559 nextchar(pRExC_state);
10563 /* Returns NULL, setting *flagp to TRYAGAIN at the end of (?) that only sets
10564 flags. Returns NULL, setting *flagp to RESTART_PASS1 if the sizing scan
10565 needs to be restarted, or'd with NEED_UTF8 if the pattern needs to be
10566 upgraded to UTF-8. Otherwise would only return NULL if regbranch() returns
10567 NULL, which cannot happen. */
10569 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
10570 /* paren: Parenthesized? 0=top; 1,2=inside '(': changed to letter.
10571 * 2 is like 1, but indicates that nextchar() has been called to advance
10572 * RExC_parse beyond the '('. Things like '(?' are indivisible tokens, and
10573 * this flag alerts us to the need to check for that */
10575 regnode *ret; /* Will be the head of the group. */
10578 regnode *ender = NULL;
10581 U32 oregflags = RExC_flags;
10582 bool have_branch = 0;
10584 I32 freeze_paren = 0;
10585 I32 after_freeze = 0;
10586 I32 num; /* numeric backreferences */
10588 char * parse_start = RExC_parse; /* MJD */
10589 char * const oregcomp_parse = RExC_parse;
10591 GET_RE_DEBUG_FLAGS_DECL;
10593 PERL_ARGS_ASSERT_REG;
10594 DEBUG_PARSE("reg ");
10596 *flagp = 0; /* Tentatively. */
10598 /* Having this true makes it feasible to have a lot fewer tests for the
10599 * parse pointer being in scope. For example, we can write
10600 * while(isFOO(*RExC_parse)) RExC_parse++;
10602 * while(RExC_parse < RExC_end && isFOO(*RExC_parse)) RExC_parse++;
10604 assert(*RExC_end == '\0');
10606 /* Make an OPEN node, if parenthesized. */
10609 /* Under /x, space and comments can be gobbled up between the '(' and
10610 * here (if paren ==2). The forms '(*VERB' and '(?...' disallow such
10611 * intervening space, as the sequence is a token, and a token should be
10613 bool has_intervening_patws = paren == 2 && *(RExC_parse - 1) != '(';
10615 if (RExC_parse >= RExC_end) {
10616 vFAIL("Unmatched (");
10619 if ( *RExC_parse == '*') { /* (*VERB:ARG) */
10620 char *start_verb = RExC_parse + 1;
10622 char *start_arg = NULL;
10623 unsigned char op = 0;
10624 int arg_required = 0;
10625 int internal_argval = -1; /* if >-1 we are not allowed an argument*/
10627 if (has_intervening_patws) {
10628 RExC_parse++; /* past the '*' */
10629 vFAIL("In '(*VERB...)', the '(' and '*' must be adjacent");
10631 while (RExC_parse < RExC_end && *RExC_parse != ')' ) {
10632 if ( *RExC_parse == ':' ) {
10633 start_arg = RExC_parse + 1;
10636 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
10638 verb_len = RExC_parse - start_verb;
10640 if (RExC_parse >= RExC_end) {
10641 goto unterminated_verb_pattern;
10643 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
10644 while ( RExC_parse < RExC_end && *RExC_parse != ')' )
10645 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
10646 if ( RExC_parse >= RExC_end || *RExC_parse != ')' )
10647 unterminated_verb_pattern:
10648 vFAIL("Unterminated verb pattern argument");
10649 if ( RExC_parse == start_arg )
10652 if ( RExC_parse >= RExC_end || *RExC_parse != ')' )
10653 vFAIL("Unterminated verb pattern");
10656 /* Here, we know that RExC_parse < RExC_end */
10658 switch ( *start_verb ) {
10659 case 'A': /* (*ACCEPT) */
10660 if ( memEQs(start_verb,verb_len,"ACCEPT") ) {
10662 internal_argval = RExC_nestroot;
10665 case 'C': /* (*COMMIT) */
10666 if ( memEQs(start_verb,verb_len,"COMMIT") )
10669 case 'F': /* (*FAIL) */
10670 if ( verb_len==1 || memEQs(start_verb,verb_len,"FAIL") ) {
10674 case ':': /* (*:NAME) */
10675 case 'M': /* (*MARK:NAME) */
10676 if ( verb_len==0 || memEQs(start_verb,verb_len,"MARK") ) {
10681 case 'P': /* (*PRUNE) */
10682 if ( memEQs(start_verb,verb_len,"PRUNE") )
10685 case 'S': /* (*SKIP) */
10686 if ( memEQs(start_verb,verb_len,"SKIP") )
10689 case 'T': /* (*THEN) */
10690 /* [19:06] <TimToady> :: is then */
10691 if ( memEQs(start_verb,verb_len,"THEN") ) {
10693 RExC_seen |= REG_CUTGROUP_SEEN;
10698 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
10700 "Unknown verb pattern '%" UTF8f "'",
10701 UTF8fARG(UTF, verb_len, start_verb));
10703 if ( arg_required && !start_arg ) {
10704 vFAIL3("Verb pattern '%.*s' has a mandatory argument",
10705 verb_len, start_verb);
10707 if (internal_argval == -1) {
10708 ret = reganode(pRExC_state, op, 0);
10710 ret = reg2Lanode(pRExC_state, op, 0, internal_argval);
10712 RExC_seen |= REG_VERBARG_SEEN;
10713 if ( ! SIZE_ONLY ) {
10715 SV *sv = newSVpvn( start_arg,
10716 RExC_parse - start_arg);
10717 ARG(ret) = add_data( pRExC_state,
10718 STR_WITH_LEN("S"));
10719 RExC_rxi->data->data[ARG(ret)]=(void*)sv;
10724 if ( internal_argval != -1 )
10725 ARG2L_SET(ret, internal_argval);
10727 nextchar(pRExC_state);
10730 else if (*RExC_parse == '?') { /* (?...) */
10731 bool is_logical = 0;
10732 const char * const seqstart = RExC_parse;
10733 const char * endptr;
10734 if (has_intervening_patws) {
10736 vFAIL("In '(?...)', the '(' and '?' must be adjacent");
10739 RExC_parse++; /* past the '?' */
10740 paren = *RExC_parse; /* might be a trailing NUL, if not
10742 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
10743 if (RExC_parse > RExC_end) {
10746 ret = NULL; /* For look-ahead/behind. */
10749 case 'P': /* (?P...) variants for those used to PCRE/Python */
10750 paren = *RExC_parse;
10751 if ( paren == '<') { /* (?P<...>) named capture */
10753 if (RExC_parse >= RExC_end) {
10754 vFAIL("Sequence (?P<... not terminated");
10756 goto named_capture;
10758 else if (paren == '>') { /* (?P>name) named recursion */
10760 if (RExC_parse >= RExC_end) {
10761 vFAIL("Sequence (?P>... not terminated");
10763 goto named_recursion;
10765 else if (paren == '=') { /* (?P=...) named backref */
10767 return handle_named_backref(pRExC_state, flagp,
10770 RExC_parse += SKIP_IF_CHAR(RExC_parse);
10771 /* diag_listed_as: Sequence (?%s...) not recognized in regex; marked by <-- HERE in m/%s/ */
10772 vFAIL3("Sequence (%.*s...) not recognized",
10773 RExC_parse-seqstart, seqstart);
10774 NOT_REACHED; /*NOTREACHED*/
10775 case '<': /* (?<...) */
10776 if (*RExC_parse == '!')
10778 else if (*RExC_parse != '=')
10785 case '\'': /* (?'...') */
10786 name_start = RExC_parse;
10787 svname = reg_scan_name(pRExC_state,
10788 SIZE_ONLY /* reverse test from the others */
10789 ? REG_RSN_RETURN_NAME
10790 : REG_RSN_RETURN_NULL);
10791 if ( RExC_parse == name_start
10792 || RExC_parse >= RExC_end
10793 || *RExC_parse != paren)
10795 vFAIL2("Sequence (?%c... not terminated",
10796 paren=='>' ? '<' : paren);
10801 if (!svname) /* shouldn't happen */
10803 "panic: reg_scan_name returned NULL");
10804 if (!RExC_paren_names) {
10805 RExC_paren_names= newHV();
10806 sv_2mortal(MUTABLE_SV(RExC_paren_names));
10808 RExC_paren_name_list= newAV();
10809 sv_2mortal(MUTABLE_SV(RExC_paren_name_list));
10812 he_str = hv_fetch_ent( RExC_paren_names, svname, 1, 0 );
10814 sv_dat = HeVAL(he_str);
10816 /* croak baby croak */
10818 "panic: paren_name hash element allocation failed");
10819 } else if ( SvPOK(sv_dat) ) {
10820 /* (?|...) can mean we have dupes so scan to check
10821 its already been stored. Maybe a flag indicating
10822 we are inside such a construct would be useful,
10823 but the arrays are likely to be quite small, so
10824 for now we punt -- dmq */
10825 IV count = SvIV(sv_dat);
10826 I32 *pv = (I32*)SvPVX(sv_dat);
10828 for ( i = 0 ; i < count ; i++ ) {
10829 if ( pv[i] == RExC_npar ) {
10835 pv = (I32*)SvGROW(sv_dat,
10836 SvCUR(sv_dat) + sizeof(I32)+1);
10837 SvCUR_set(sv_dat, SvCUR(sv_dat) + sizeof(I32));
10838 pv[count] = RExC_npar;
10839 SvIV_set(sv_dat, SvIVX(sv_dat) + 1);
10842 (void)SvUPGRADE(sv_dat,SVt_PVNV);
10843 sv_setpvn(sv_dat, (char *)&(RExC_npar),
10846 SvIV_set(sv_dat, 1);
10849 /* Yes this does cause a memory leak in debugging Perls
10851 if (!av_store(RExC_paren_name_list,
10852 RExC_npar, SvREFCNT_inc(svname)))
10853 SvREFCNT_dec_NN(svname);
10856 /*sv_dump(sv_dat);*/
10858 nextchar(pRExC_state);
10860 goto capturing_parens;
10862 RExC_seen |= REG_LOOKBEHIND_SEEN;
10863 RExC_in_lookbehind++;
10865 if (RExC_parse >= RExC_end) {
10866 vFAIL("Sequence (?... not terminated");
10870 case '=': /* (?=...) */
10871 RExC_seen_zerolen++;
10873 case '!': /* (?!...) */
10874 RExC_seen_zerolen++;
10875 /* check if we're really just a "FAIL" assertion */
10876 skip_to_be_ignored_text(pRExC_state, &RExC_parse,
10877 FALSE /* Don't force to /x */ );
10878 if (*RExC_parse == ')') {
10879 ret=reganode(pRExC_state, OPFAIL, 0);
10880 nextchar(pRExC_state);
10884 case '|': /* (?|...) */
10885 /* branch reset, behave like a (?:...) except that
10886 buffers in alternations share the same numbers */
10888 after_freeze = freeze_paren = RExC_npar;
10890 case ':': /* (?:...) */
10891 case '>': /* (?>...) */
10893 case '$': /* (?$...) */
10894 case '@': /* (?@...) */
10895 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
10897 case '0' : /* (?0) */
10898 case 'R' : /* (?R) */
10899 if (RExC_parse == RExC_end || *RExC_parse != ')')
10900 FAIL("Sequence (?R) not terminated");
10902 RExC_seen |= REG_RECURSE_SEEN;
10903 *flagp |= POSTPONED;
10904 goto gen_recurse_regop;
10906 /* named and numeric backreferences */
10907 case '&': /* (?&NAME) */
10908 parse_start = RExC_parse - 1;
10911 SV *sv_dat = reg_scan_name(pRExC_state,
10912 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
10913 num = sv_dat ? *((I32 *)SvPVX(sv_dat)) : 0;
10915 if (RExC_parse >= RExC_end || *RExC_parse != ')')
10916 vFAIL("Sequence (?&... not terminated");
10917 goto gen_recurse_regop;
10920 if (!(RExC_parse[0] >= '1' && RExC_parse[0] <= '9')) {
10922 vFAIL("Illegal pattern");
10924 goto parse_recursion;
10926 case '-': /* (?-1) */
10927 if (!(RExC_parse[0] >= '1' && RExC_parse[0] <= '9')) {
10928 RExC_parse--; /* rewind to let it be handled later */
10932 case '1': case '2': case '3': case '4': /* (?1) */
10933 case '5': case '6': case '7': case '8': case '9':
10934 RExC_parse = (char *) seqstart + 1; /* Point to the digit */
10937 bool is_neg = FALSE;
10939 parse_start = RExC_parse - 1; /* MJD */
10940 if (*RExC_parse == '-') {
10944 if (grok_atoUV(RExC_parse, &unum, &endptr)
10948 RExC_parse = (char*)endptr;
10952 /* Some limit for num? */
10956 if (*RExC_parse!=')')
10957 vFAIL("Expecting close bracket");
10960 if ( paren == '-' ) {
10962 Diagram of capture buffer numbering.
10963 Top line is the normal capture buffer numbers
10964 Bottom line is the negative indexing as from
10968 /(a(x)y)(a(b(c(?-2)d)e)f)(g(h))/
10972 num = RExC_npar + num;
10975 vFAIL("Reference to nonexistent group");
10977 } else if ( paren == '+' ) {
10978 num = RExC_npar + num - 1;
10980 /* We keep track how many GOSUB items we have produced.
10981 To start off the ARG2L() of the GOSUB holds its "id",
10982 which is used later in conjunction with RExC_recurse
10983 to calculate the offset we need to jump for the GOSUB,
10984 which it will store in the final representation.
10985 We have to defer the actual calculation until much later
10986 as the regop may move.
10989 ret = reg2Lanode(pRExC_state, GOSUB, num, RExC_recurse_count);
10991 if (num > (I32)RExC_rx->nparens) {
10993 vFAIL("Reference to nonexistent group");
10995 RExC_recurse_count++;
10996 DEBUG_OPTIMISE_MORE_r(Perl_re_printf( aTHX_
10997 "%*s%*s Recurse #%" UVuf " to %" IVdf "\n",
10998 22, "| |", (int)(depth * 2 + 1), "",
10999 (UV)ARG(ret), (IV)ARG2L(ret)));
11001 RExC_seen |= REG_RECURSE_SEEN;
11003 Set_Node_Length(ret, 1 + regarglen[OP(ret)]); /* MJD */
11004 Set_Node_Offset(ret, parse_start); /* MJD */
11006 *flagp |= POSTPONED;
11007 assert(*RExC_parse == ')');
11008 nextchar(pRExC_state);
11013 case '?': /* (??...) */
11015 if (*RExC_parse != '{') {
11016 RExC_parse += SKIP_IF_CHAR(RExC_parse);
11017 /* diag_listed_as: Sequence (?%s...) not recognized in regex; marked by <-- HERE in m/%s/ */
11019 "Sequence (%" UTF8f "...) not recognized",
11020 UTF8fARG(UTF, RExC_parse-seqstart, seqstart));
11021 NOT_REACHED; /*NOTREACHED*/
11023 *flagp |= POSTPONED;
11027 case '{': /* (?{...}) */
11030 struct reg_code_block *cb;
11032 RExC_seen_zerolen++;
11034 if ( !pRExC_state->code_blocks
11035 || pRExC_state->code_index
11036 >= pRExC_state->code_blocks->count
11037 || pRExC_state->code_blocks->cb[pRExC_state->code_index].start
11038 != (STRLEN)((RExC_parse -3 - (is_logical ? 1 : 0))
11041 if (RExC_pm_flags & PMf_USE_RE_EVAL)
11042 FAIL("panic: Sequence (?{...}): no code block found\n");
11043 FAIL("Eval-group not allowed at runtime, use re 'eval'");
11045 /* this is a pre-compiled code block (?{...}) */
11046 cb = &pRExC_state->code_blocks->cb[pRExC_state->code_index];
11047 RExC_parse = RExC_start + cb->end;
11050 if (cb->src_regex) {
11051 n = add_data(pRExC_state, STR_WITH_LEN("rl"));
11052 RExC_rxi->data->data[n] =
11053 (void*)SvREFCNT_inc((SV*)cb->src_regex);
11054 RExC_rxi->data->data[n+1] = (void*)o;
11057 n = add_data(pRExC_state,
11058 (RExC_pm_flags & PMf_HAS_CV) ? "L" : "l", 1);
11059 RExC_rxi->data->data[n] = (void*)o;
11062 pRExC_state->code_index++;
11063 nextchar(pRExC_state);
11067 ret = reg_node(pRExC_state, LOGICAL);
11069 eval = reg2Lanode(pRExC_state, EVAL,
11072 /* for later propagation into (??{})
11074 RExC_flags & RXf_PMf_COMPILETIME
11079 REGTAIL(pRExC_state, ret, eval);
11080 /* deal with the length of this later - MJD */
11083 ret = reg2Lanode(pRExC_state, EVAL, n, 0);
11084 Set_Node_Length(ret, RExC_parse - parse_start + 1);
11085 Set_Node_Offset(ret, parse_start);
11088 case '(': /* (?(?{...})...) and (?(?=...)...) */
11091 const int DEFINE_len = sizeof("DEFINE") - 1;
11092 if (RExC_parse[0] == '?') { /* (?(?...)) */
11093 if ( RExC_parse < RExC_end - 1
11094 && ( RExC_parse[1] == '='
11095 || RExC_parse[1] == '!'
11096 || RExC_parse[1] == '<'
11097 || RExC_parse[1] == '{')
11098 ) { /* Lookahead or eval. */
11102 ret = reg_node(pRExC_state, LOGICAL);
11106 tail = reg(pRExC_state, 1, &flag, depth+1);
11107 if (flag & (RESTART_PASS1|NEED_UTF8)) {
11108 *flagp = flag & (RESTART_PASS1|NEED_UTF8);
11111 REGTAIL(pRExC_state, ret, tail);
11114 /* Fall through to ‘Unknown switch condition’ at the
11115 end of the if/else chain. */
11117 else if ( RExC_parse[0] == '<' /* (?(<NAME>)...) */
11118 || RExC_parse[0] == '\'' ) /* (?('NAME')...) */
11120 char ch = RExC_parse[0] == '<' ? '>' : '\'';
11121 char *name_start= RExC_parse++;
11123 SV *sv_dat=reg_scan_name(pRExC_state,
11124 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
11125 if ( RExC_parse == name_start
11126 || RExC_parse >= RExC_end
11127 || *RExC_parse != ch)
11129 vFAIL2("Sequence (?(%c... not terminated",
11130 (ch == '>' ? '<' : ch));
11134 num = add_data( pRExC_state, STR_WITH_LEN("S"));
11135 RExC_rxi->data->data[num]=(void*)sv_dat;
11136 SvREFCNT_inc_simple_void(sv_dat);
11138 ret = reganode(pRExC_state,NGROUPP,num);
11139 goto insert_if_check_paren;
11141 else if (RExC_end - RExC_parse >= DEFINE_len
11142 && strnEQ(RExC_parse, "DEFINE", DEFINE_len))
11144 ret = reganode(pRExC_state,DEFINEP,0);
11145 RExC_parse += DEFINE_len;
11147 goto insert_if_check_paren;
11149 else if (RExC_parse[0] == 'R') {
11151 /* parno == 0 => /(?(R)YES|NO)/ "in any form of recursion OR eval"
11152 * parno == 1 => /(?(R0)YES|NO)/ "in GOSUB (?0) / (?R)"
11153 * parno == 2 => /(?(R1)YES|NO)/ "in GOSUB (?1) (parno-1)"
11156 if (RExC_parse[0] == '0') {
11160 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
11162 if (grok_atoUV(RExC_parse, &uv, &endptr)
11165 parno = (I32)uv + 1;
11166 RExC_parse = (char*)endptr;
11168 /* else "Switch condition not recognized" below */
11169 } else if (RExC_parse[0] == '&') {
11172 sv_dat = reg_scan_name(pRExC_state,
11174 ? REG_RSN_RETURN_NULL
11175 : REG_RSN_RETURN_DATA);
11177 /* we should only have a false sv_dat when
11178 * SIZE_ONLY is true, and we always have false
11179 * sv_dat when SIZE_ONLY is true.
11180 * reg_scan_name() will VFAIL() if the name is
11181 * unknown when SIZE_ONLY is false, and otherwise
11182 * will return something, and when SIZE_ONLY is
11183 * true, reg_scan_name() just parses the string,
11184 * and doesnt return anything. (in theory) */
11185 assert(SIZE_ONLY ? !sv_dat : !!sv_dat);
11188 parno = 1 + *((I32 *)SvPVX(sv_dat));
11190 ret = reganode(pRExC_state,INSUBP,parno);
11191 goto insert_if_check_paren;
11193 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
11197 if (grok_atoUV(RExC_parse, &uv, &endptr)
11201 RExC_parse = (char*)endptr;
11204 vFAIL("panic: grok_atoUV returned FALSE");
11206 ret = reganode(pRExC_state, GROUPP, parno);
11208 insert_if_check_paren:
11209 if (UCHARAT(RExC_parse) != ')') {
11210 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
11211 vFAIL("Switch condition not recognized");
11213 nextchar(pRExC_state);
11215 REGTAIL(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
11216 br = regbranch(pRExC_state, &flags, 1,depth+1);
11218 if (flags & (RESTART_PASS1|NEED_UTF8)) {
11219 *flagp = flags & (RESTART_PASS1|NEED_UTF8);
11222 FAIL2("panic: regbranch returned NULL, flags=%#" UVxf,
11225 REGTAIL(pRExC_state, br, reganode(pRExC_state,
11227 c = UCHARAT(RExC_parse);
11228 nextchar(pRExC_state);
11229 if (flags&HASWIDTH)
11230 *flagp |= HASWIDTH;
11233 vFAIL("(?(DEFINE)....) does not allow branches");
11235 /* Fake one for optimizer. */
11236 lastbr = reganode(pRExC_state, IFTHEN, 0);
11238 if (!regbranch(pRExC_state, &flags, 1,depth+1)) {
11239 if (flags & (RESTART_PASS1|NEED_UTF8)) {
11240 *flagp = flags & (RESTART_PASS1|NEED_UTF8);
11243 FAIL2("panic: regbranch returned NULL, flags=%#" UVxf,
11246 REGTAIL(pRExC_state, ret, lastbr);
11247 if (flags&HASWIDTH)
11248 *flagp |= HASWIDTH;
11249 c = UCHARAT(RExC_parse);
11250 nextchar(pRExC_state);
11255 if (RExC_parse >= RExC_end)
11256 vFAIL("Switch (?(condition)... not terminated");
11258 vFAIL("Switch (?(condition)... contains too many branches");
11260 ender = reg_node(pRExC_state, TAIL);
11261 REGTAIL(pRExC_state, br, ender);
11263 REGTAIL(pRExC_state, lastbr, ender);
11264 REGTAIL(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
11267 REGTAIL(pRExC_state, ret, ender);
11268 RExC_size++; /* XXX WHY do we need this?!!
11269 For large programs it seems to be required
11270 but I can't figure out why. -- dmq*/
11273 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
11274 vFAIL("Unknown switch condition (?(...))");
11276 case '[': /* (?[ ... ]) */
11277 return handle_regex_sets(pRExC_state, NULL, flagp, depth,
11279 case 0: /* A NUL */
11280 RExC_parse--; /* for vFAIL to print correctly */
11281 vFAIL("Sequence (? incomplete");
11283 default: /* e.g., (?i) */
11284 RExC_parse = (char *) seqstart + 1;
11286 parse_lparen_question_flags(pRExC_state);
11287 if (UCHARAT(RExC_parse) != ':') {
11288 if (RExC_parse < RExC_end)
11289 nextchar(pRExC_state);
11294 nextchar(pRExC_state);
11299 else if (!(RExC_flags & RXf_PMf_NOCAPTURE)) { /* (...) */
11304 ret = reganode(pRExC_state, OPEN, parno);
11306 if (!RExC_nestroot)
11307 RExC_nestroot = parno;
11308 if (RExC_open_parens && !RExC_open_parens[parno])
11310 DEBUG_OPTIMISE_MORE_r(Perl_re_printf( aTHX_
11311 "%*s%*s Setting open paren #%" IVdf " to %d\n",
11312 22, "| |", (int)(depth * 2 + 1), "",
11313 (IV)parno, REG_NODE_NUM(ret)));
11314 RExC_open_parens[parno]= ret;
11317 Set_Node_Length(ret, 1); /* MJD */
11318 Set_Node_Offset(ret, RExC_parse); /* MJD */
11321 /* with RXf_PMf_NOCAPTURE treat (...) as (?:...) */
11330 /* Pick up the branches, linking them together. */
11331 parse_start = RExC_parse; /* MJD */
11332 br = regbranch(pRExC_state, &flags, 1,depth+1);
11334 /* branch_len = (paren != 0); */
11337 if (flags & (RESTART_PASS1|NEED_UTF8)) {
11338 *flagp = flags & (RESTART_PASS1|NEED_UTF8);
11341 FAIL2("panic: regbranch returned NULL, flags=%#" UVxf, (UV) flags);
11343 if (*RExC_parse == '|') {
11344 if (!SIZE_ONLY && RExC_extralen) {
11345 reginsert(pRExC_state, BRANCHJ, br, depth+1);
11348 reginsert(pRExC_state, BRANCH, br, depth+1);
11349 Set_Node_Length(br, paren != 0);
11350 Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
11354 RExC_extralen += 1; /* For BRANCHJ-BRANCH. */
11356 else if (paren == ':') {
11357 *flagp |= flags&SIMPLE;
11359 if (is_open) { /* Starts with OPEN. */
11360 REGTAIL(pRExC_state, ret, br); /* OPEN -> first. */
11362 else if (paren != '?') /* Not Conditional */
11364 *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
11366 while (*RExC_parse == '|') {
11367 if (!SIZE_ONLY && RExC_extralen) {
11368 ender = reganode(pRExC_state, LONGJMP,0);
11370 /* Append to the previous. */
11371 REGTAIL(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
11374 RExC_extralen += 2; /* Account for LONGJMP. */
11375 nextchar(pRExC_state);
11376 if (freeze_paren) {
11377 if (RExC_npar > after_freeze)
11378 after_freeze = RExC_npar;
11379 RExC_npar = freeze_paren;
11381 br = regbranch(pRExC_state, &flags, 0, depth+1);
11384 if (flags & (RESTART_PASS1|NEED_UTF8)) {
11385 *flagp = flags & (RESTART_PASS1|NEED_UTF8);
11388 FAIL2("panic: regbranch returned NULL, flags=%#" UVxf, (UV) flags);
11390 REGTAIL(pRExC_state, lastbr, br); /* BRANCH -> BRANCH. */
11392 *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
11395 if (have_branch || paren != ':') {
11396 /* Make a closing node, and hook it on the end. */
11399 ender = reg_node(pRExC_state, TAIL);
11402 ender = reganode(pRExC_state, CLOSE, parno);
11403 if ( RExC_close_parens ) {
11404 DEBUG_OPTIMISE_MORE_r(Perl_re_printf( aTHX_
11405 "%*s%*s Setting close paren #%" IVdf " to %d\n",
11406 22, "| |", (int)(depth * 2 + 1), "", (IV)parno, REG_NODE_NUM(ender)));
11407 RExC_close_parens[parno]= ender;
11408 if (RExC_nestroot == parno)
11411 Set_Node_Offset(ender,RExC_parse+1); /* MJD */
11412 Set_Node_Length(ender,1); /* MJD */
11418 *flagp &= ~HASWIDTH;
11421 ender = reg_node(pRExC_state, SUCCEED);
11424 ender = reg_node(pRExC_state, END);
11426 assert(!RExC_end_op); /* there can only be one! */
11427 RExC_end_op = ender;
11428 if (RExC_close_parens) {
11429 DEBUG_OPTIMISE_MORE_r(Perl_re_printf( aTHX_
11430 "%*s%*s Setting close paren #0 (END) to %d\n",
11431 22, "| |", (int)(depth * 2 + 1), "", REG_NODE_NUM(ender)));
11433 RExC_close_parens[0]= ender;
11438 DEBUG_PARSE_r(if (!SIZE_ONLY) {
11439 DEBUG_PARSE_MSG("lsbr");
11440 regprop(RExC_rx, RExC_mysv1, lastbr, NULL, pRExC_state);
11441 regprop(RExC_rx, RExC_mysv2, ender, NULL, pRExC_state);
11442 Perl_re_printf( aTHX_ "~ tying lastbr %s (%" IVdf ") to ender %s (%" IVdf ") offset %" IVdf "\n",
11443 SvPV_nolen_const(RExC_mysv1),
11444 (IV)REG_NODE_NUM(lastbr),
11445 SvPV_nolen_const(RExC_mysv2),
11446 (IV)REG_NODE_NUM(ender),
11447 (IV)(ender - lastbr)
11450 REGTAIL(pRExC_state, lastbr, ender);
11452 if (have_branch && !SIZE_ONLY) {
11453 char is_nothing= 1;
11455 RExC_seen |= REG_TOP_LEVEL_BRANCHES_SEEN;
11457 /* Hook the tails of the branches to the closing node. */
11458 for (br = ret; br; br = regnext(br)) {
11459 const U8 op = PL_regkind[OP(br)];
11460 if (op == BRANCH) {
11461 REGTAIL_STUDY(pRExC_state, NEXTOPER(br), ender);
11462 if ( OP(NEXTOPER(br)) != NOTHING
11463 || regnext(NEXTOPER(br)) != ender)
11466 else if (op == BRANCHJ) {
11467 REGTAIL_STUDY(pRExC_state, NEXTOPER(NEXTOPER(br)), ender);
11468 /* for now we always disable this optimisation * /
11469 if ( OP(NEXTOPER(NEXTOPER(br))) != NOTHING
11470 || regnext(NEXTOPER(NEXTOPER(br))) != ender)
11476 br= PL_regkind[OP(ret)] != BRANCH ? regnext(ret) : ret;
11477 DEBUG_PARSE_r(if (!SIZE_ONLY) {
11478 DEBUG_PARSE_MSG("NADA");
11479 regprop(RExC_rx, RExC_mysv1, ret, NULL, pRExC_state);
11480 regprop(RExC_rx, RExC_mysv2, ender, NULL, pRExC_state);
11481 Perl_re_printf( aTHX_ "~ converting ret %s (%" IVdf ") to ender %s (%" IVdf ") offset %" IVdf "\n",
11482 SvPV_nolen_const(RExC_mysv1),
11483 (IV)REG_NODE_NUM(ret),
11484 SvPV_nolen_const(RExC_mysv2),
11485 (IV)REG_NODE_NUM(ender),
11490 if (OP(ender) == TAIL) {
11495 for ( opt= br + 1; opt < ender ; opt++ )
11496 OP(opt)= OPTIMIZED;
11497 NEXT_OFF(br)= ender - br;
11505 static const char parens[] = "=!<,>";
11507 if (paren && (p = strchr(parens, paren))) {
11508 U8 node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
11509 int flag = (p - parens) > 1;
11512 node = SUSPEND, flag = 0;
11513 reginsert(pRExC_state, node,ret, depth+1);
11514 Set_Node_Cur_Length(ret, parse_start);
11515 Set_Node_Offset(ret, parse_start + 1);
11517 REGTAIL_STUDY(pRExC_state, ret, reg_node(pRExC_state, TAIL));
11521 /* Check for proper termination. */
11523 /* restore original flags, but keep (?p) and, if we've changed from /d
11524 * rules to /u, keep the /u */
11525 RExC_flags = oregflags | (RExC_flags & RXf_PMf_KEEPCOPY);
11526 if (DEPENDS_SEMANTICS && RExC_uni_semantics) {
11527 set_regex_charset(&RExC_flags, REGEX_UNICODE_CHARSET);
11529 if (RExC_parse >= RExC_end || UCHARAT(RExC_parse) != ')') {
11530 RExC_parse = oregcomp_parse;
11531 vFAIL("Unmatched (");
11533 nextchar(pRExC_state);
11535 else if (!paren && RExC_parse < RExC_end) {
11536 if (*RExC_parse == ')') {
11538 vFAIL("Unmatched )");
11541 FAIL("Junk on end of regexp"); /* "Can't happen". */
11542 NOT_REACHED; /* NOTREACHED */
11545 if (RExC_in_lookbehind) {
11546 RExC_in_lookbehind--;
11548 if (after_freeze > RExC_npar)
11549 RExC_npar = after_freeze;
11554 - regbranch - one alternative of an | operator
11556 * Implements the concatenation operator.
11558 * Returns NULL, setting *flagp to RESTART_PASS1 if the sizing scan needs to be
11559 * restarted, or'd with NEED_UTF8 if the pattern needs to be upgraded to UTF-8
11562 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first, U32 depth)
11565 regnode *chain = NULL;
11567 I32 flags = 0, c = 0;
11568 GET_RE_DEBUG_FLAGS_DECL;
11570 PERL_ARGS_ASSERT_REGBRANCH;
11572 DEBUG_PARSE("brnc");
11577 if (!SIZE_ONLY && RExC_extralen)
11578 ret = reganode(pRExC_state, BRANCHJ,0);
11580 ret = reg_node(pRExC_state, BRANCH);
11581 Set_Node_Length(ret, 1);
11585 if (!first && SIZE_ONLY)
11586 RExC_extralen += 1; /* BRANCHJ */
11588 *flagp = WORST; /* Tentatively. */
11590 skip_to_be_ignored_text(pRExC_state, &RExC_parse,
11591 FALSE /* Don't force to /x */ );
11592 while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
11593 flags &= ~TRYAGAIN;
11594 latest = regpiece(pRExC_state, &flags,depth+1);
11595 if (latest == NULL) {
11596 if (flags & TRYAGAIN)
11598 if (flags & (RESTART_PASS1|NEED_UTF8)) {
11599 *flagp = flags & (RESTART_PASS1|NEED_UTF8);
11602 FAIL2("panic: regpiece returned NULL, flags=%#" UVxf, (UV) flags);
11604 else if (ret == NULL)
11606 *flagp |= flags&(HASWIDTH|POSTPONED);
11607 if (chain == NULL) /* First piece. */
11608 *flagp |= flags&SPSTART;
11610 /* FIXME adding one for every branch after the first is probably
11611 * excessive now we have TRIE support. (hv) */
11613 REGTAIL(pRExC_state, chain, latest);
11618 if (chain == NULL) { /* Loop ran zero times. */
11619 chain = reg_node(pRExC_state, NOTHING);
11624 *flagp |= flags&SIMPLE;
11631 - regpiece - something followed by possible quantifier * + ? {n,m}
11633 * Note that the branching code sequences used for ? and the general cases
11634 * of * and + are somewhat optimized: they use the same NOTHING node as
11635 * both the endmarker for their branch list and the body of the last branch.
11636 * It might seem that this node could be dispensed with entirely, but the
11637 * endmarker role is not redundant.
11639 * Returns NULL, setting *flagp to TRYAGAIN if regatom() returns NULL with
11641 * Returns NULL, setting *flagp to RESTART_PASS1 if the sizing scan needs to be
11642 * restarted, or'd with NEED_UTF8 if the pattern needs to be upgraded to UTF-8
11645 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
11651 const char * const origparse = RExC_parse;
11653 I32 max = REG_INFTY;
11654 #ifdef RE_TRACK_PATTERN_OFFSETS
11657 const char *maxpos = NULL;
11660 /* Save the original in case we change the emitted regop to a FAIL. */
11661 regnode * const orig_emit = RExC_emit;
11663 GET_RE_DEBUG_FLAGS_DECL;
11665 PERL_ARGS_ASSERT_REGPIECE;
11667 DEBUG_PARSE("piec");
11669 ret = regatom(pRExC_state, &flags,depth+1);
11671 if (flags & (TRYAGAIN|RESTART_PASS1|NEED_UTF8))
11672 *flagp |= flags & (TRYAGAIN|RESTART_PASS1|NEED_UTF8);
11674 FAIL2("panic: regatom returned NULL, flags=%#" UVxf, (UV) flags);
11680 if (op == '{' && regcurly(RExC_parse)) {
11682 #ifdef RE_TRACK_PATTERN_OFFSETS
11683 parse_start = RExC_parse; /* MJD */
11685 next = RExC_parse + 1;
11686 while (isDIGIT(*next) || *next == ',') {
11687 if (*next == ',') {
11695 if (*next == '}') { /* got one */
11696 const char* endptr;
11700 if (isDIGIT(*RExC_parse)) {
11701 if (!grok_atoUV(RExC_parse, &uv, &endptr))
11702 vFAIL("Invalid quantifier in {,}");
11703 if (uv >= REG_INFTY)
11704 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
11709 if (*maxpos == ',')
11712 maxpos = RExC_parse;
11713 if (isDIGIT(*maxpos)) {
11714 if (!grok_atoUV(maxpos, &uv, &endptr))
11715 vFAIL("Invalid quantifier in {,}");
11716 if (uv >= REG_INFTY)
11717 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
11720 max = REG_INFTY; /* meaning "infinity" */
11723 nextchar(pRExC_state);
11724 if (max < min) { /* If can't match, warn and optimize to fail
11726 reginsert(pRExC_state, OPFAIL, orig_emit, depth+1);
11728 ckWARNreg(RExC_parse, "Quantifier {n,m} with n > m can't match");
11729 NEXT_OFF(orig_emit)= regarglen[OPFAIL] + NODE_STEP_REGNODE;
11733 else if (min == max && *RExC_parse == '?')
11736 ckWARN2reg(RExC_parse + 1,
11737 "Useless use of greediness modifier '%c'",
11743 if ((flags&SIMPLE)) {
11744 if (min == 0 && max == REG_INFTY) {
11745 reginsert(pRExC_state, STAR, ret, depth+1);
11748 RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
11751 if (min == 1 && max == REG_INFTY) {
11752 reginsert(pRExC_state, PLUS, ret, depth+1);
11755 RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
11758 MARK_NAUGHTY_EXP(2, 2);
11759 reginsert(pRExC_state, CURLY, ret, depth+1);
11760 Set_Node_Offset(ret, parse_start+1); /* MJD */
11761 Set_Node_Cur_Length(ret, parse_start);
11764 regnode * const w = reg_node(pRExC_state, WHILEM);
11767 REGTAIL(pRExC_state, ret, w);
11768 if (!SIZE_ONLY && RExC_extralen) {
11769 reginsert(pRExC_state, LONGJMP,ret, depth+1);
11770 reginsert(pRExC_state, NOTHING,ret, depth+1);
11771 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
11773 reginsert(pRExC_state, CURLYX,ret, depth+1);
11775 Set_Node_Offset(ret, parse_start+1);
11776 Set_Node_Length(ret,
11777 op == '{' ? (RExC_parse - parse_start) : 1);
11779 if (!SIZE_ONLY && RExC_extralen)
11780 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
11781 REGTAIL(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
11783 RExC_whilem_seen++, RExC_extralen += 3;
11784 MARK_NAUGHTY_EXP(1, 4); /* compound interest */
11791 *flagp |= HASWIDTH;
11793 ARG1_SET(ret, (U16)min);
11794 ARG2_SET(ret, (U16)max);
11796 if (max == REG_INFTY)
11797 RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
11803 if (!ISMULT1(op)) {
11808 #if 0 /* Now runtime fix should be reliable. */
11810 /* if this is reinstated, don't forget to put this back into perldiag:
11812 =item Regexp *+ operand could be empty at {#} in regex m/%s/
11814 (F) The part of the regexp subject to either the * or + quantifier
11815 could match an empty string. The {#} shows in the regular
11816 expression about where the problem was discovered.
11820 if (!(flags&HASWIDTH) && op != '?')
11821 vFAIL("Regexp *+ operand could be empty");
11824 #ifdef RE_TRACK_PATTERN_OFFSETS
11825 parse_start = RExC_parse;
11827 nextchar(pRExC_state);
11829 *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
11835 else if (op == '+') {
11839 else if (op == '?') {
11844 if (!SIZE_ONLY && !(flags&(HASWIDTH|POSTPONED)) && max > REG_INFTY/3) {
11845 SAVEFREESV(RExC_rx_sv); /* in case of fatal warnings */
11846 ckWARN2reg(RExC_parse,
11847 "%" UTF8f " matches null string many times",
11848 UTF8fARG(UTF, (RExC_parse >= origparse
11849 ? RExC_parse - origparse
11852 (void)ReREFCNT_inc(RExC_rx_sv);
11855 if (*RExC_parse == '?') {
11856 nextchar(pRExC_state);
11857 reginsert(pRExC_state, MINMOD, ret, depth+1);
11858 REGTAIL(pRExC_state, ret, ret + NODE_STEP_REGNODE);
11860 else if (*RExC_parse == '+') {
11862 nextchar(pRExC_state);
11863 ender = reg_node(pRExC_state, SUCCEED);
11864 REGTAIL(pRExC_state, ret, ender);
11865 reginsert(pRExC_state, SUSPEND, ret, depth+1);
11867 ender = reg_node(pRExC_state, TAIL);
11868 REGTAIL(pRExC_state, ret, ender);
11871 if (ISMULT2(RExC_parse)) {
11873 vFAIL("Nested quantifiers");
11880 S_grok_bslash_N(pTHX_ RExC_state_t *pRExC_state,
11889 /* This routine teases apart the various meanings of \N and returns
11890 * accordingly. The input parameters constrain which meaning(s) is/are valid
11891 * in the current context.
11893 * Exactly one of <node_p> and <code_point_p> must be non-NULL.
11895 * If <code_point_p> is not NULL, the context is expecting the result to be a
11896 * single code point. If this \N instance turns out to a single code point,
11897 * the function returns TRUE and sets *code_point_p to that code point.
11899 * If <node_p> is not NULL, the context is expecting the result to be one of
11900 * the things representable by a regnode. If this \N instance turns out to be
11901 * one such, the function generates the regnode, returns TRUE and sets *node_p
11902 * to point to that regnode.
11904 * If this instance of \N isn't legal in any context, this function will
11905 * generate a fatal error and not return.
11907 * On input, RExC_parse should point to the first char following the \N at the
11908 * time of the call. On successful return, RExC_parse will have been updated
11909 * to point to just after the sequence identified by this routine. Also
11910 * *flagp has been updated as needed.
11912 * When there is some problem with the current context and this \N instance,
11913 * the function returns FALSE, without advancing RExC_parse, nor setting
11914 * *node_p, nor *code_point_p, nor *flagp.
11916 * If <cp_count> is not NULL, the caller wants to know the length (in code
11917 * points) that this \N sequence matches. This is set even if the function
11918 * returns FALSE, as detailed below.
11920 * There are 5 possibilities here, as detailed in the next 5 paragraphs.
11922 * Probably the most common case is for the \N to specify a single code point.
11923 * *cp_count will be set to 1, and *code_point_p will be set to that code
11926 * Another possibility is for the input to be an empty \N{}, which for
11927 * backwards compatibility we accept. *cp_count will be set to 0. *node_p
11928 * will be set to a generated NOTHING node.
11930 * Still another possibility is for the \N to mean [^\n]. *cp_count will be
11931 * set to 0. *node_p will be set to a generated REG_ANY node.
11933 * The fourth possibility is that \N resolves to a sequence of more than one
11934 * code points. *cp_count will be set to the number of code points in the
11935 * sequence. *node_p * will be set to a generated node returned by this
11936 * function calling S_reg().
11938 * The final possibility is that it is premature to be calling this function;
11939 * that pass1 needs to be restarted. This can happen when this changes from
11940 * /d to /u rules, or when the pattern needs to be upgraded to UTF-8. The
11941 * latter occurs only when the fourth possibility would otherwise be in
11942 * effect, and is because one of those code points requires the pattern to be
11943 * recompiled as UTF-8. The function returns FALSE, and sets the
11944 * RESTART_PASS1 and NEED_UTF8 flags in *flagp, as appropriate. When this
11945 * happens, the caller needs to desist from continuing parsing, and return
11946 * this information to its caller. This is not set for when there is only one
11947 * code point, as this can be called as part of an ANYOF node, and they can
11948 * store above-Latin1 code points without the pattern having to be in UTF-8.
11950 * For non-single-quoted regexes, the tokenizer has resolved character and
11951 * sequence names inside \N{...} into their Unicode values, normalizing the
11952 * result into what we should see here: '\N{U+c1.c2...}', where c1... are the
11953 * hex-represented code points in the sequence. This is done there because
11954 * the names can vary based on what charnames pragma is in scope at the time,
11955 * so we need a way to take a snapshot of what they resolve to at the time of
11956 * the original parse. [perl #56444].
11958 * That parsing is skipped for single-quoted regexes, so we may here get
11959 * '\N{NAME}'. This is a fatal error. These names have to be resolved by the
11960 * parser. But if the single-quoted regex is something like '\N{U+41}', that
11961 * is legal and handled here. The code point is Unicode, and has to be
11962 * translated into the native character set for non-ASCII platforms.
11965 char * endbrace; /* points to '}' following the name */
11966 char *endchar; /* Points to '.' or '}' ending cur char in the input
11968 char* p = RExC_parse; /* Temporary */
11970 GET_RE_DEBUG_FLAGS_DECL;
11972 PERL_ARGS_ASSERT_GROK_BSLASH_N;
11974 GET_RE_DEBUG_FLAGS;
11976 assert(cBOOL(node_p) ^ cBOOL(code_point_p)); /* Exactly one should be set */
11977 assert(! (node_p && cp_count)); /* At most 1 should be set */
11979 if (cp_count) { /* Initialize return for the most common case */
11983 /* The [^\n] meaning of \N ignores spaces and comments under the /x
11984 * modifier. The other meanings do not, so use a temporary until we find
11985 * out which we are being called with */
11986 skip_to_be_ignored_text(pRExC_state, &p,
11987 FALSE /* Don't force to /x */ );
11989 /* Disambiguate between \N meaning a named character versus \N meaning
11990 * [^\n]. The latter is assumed when the {...} following the \N is a legal
11991 * quantifier, or there is no '{' at all */
11992 if (*p != '{' || regcurly(p)) {
12002 *node_p = reg_node(pRExC_state, REG_ANY);
12003 *flagp |= HASWIDTH|SIMPLE;
12005 Set_Node_Length(*node_p, 1); /* MJD */
12009 /* Here, we have decided it should be a named character or sequence */
12011 /* The test above made sure that the next real character is a '{', but
12012 * under the /x modifier, it could be separated by space (or a comment and
12013 * \n) and this is not allowed (for consistency with \x{...} and the
12014 * tokenizer handling of \N{NAME}). */
12015 if (*RExC_parse != '{') {
12016 vFAIL("Missing braces on \\N{}");
12019 RExC_parse++; /* Skip past the '{' */
12021 if (! (endbrace = strchr(RExC_parse, '}'))) { /* no trailing brace */
12022 vFAIL2("Missing right brace on \\%c{}", 'N');
12024 else if(!(endbrace == RExC_parse /* nothing between the {} */
12025 || (endbrace - RExC_parse >= 2 /* U+ (bad hex is checked... */
12026 && strnEQ(RExC_parse, "U+", 2)))) /* ... below for a better
12029 RExC_parse = endbrace; /* position msg's '<--HERE' */
12030 vFAIL("\\N{NAME} must be resolved by the lexer");
12033 REQUIRE_UNI_RULES(flagp, FALSE); /* Unicode named chars imply Unicode
12036 if (endbrace == RExC_parse) { /* empty: \N{} */
12038 RExC_parse++; /* Position after the "}" */
12039 vFAIL("Zero length \\N{}");
12044 nextchar(pRExC_state);
12049 *node_p = reg_node(pRExC_state,NOTHING);
12053 RExC_parse += 2; /* Skip past the 'U+' */
12055 /* Because toke.c has generated a special construct for us guaranteed not
12056 * to have NULs, we can use a str function */
12057 endchar = RExC_parse + strcspn(RExC_parse, ".}");
12059 /* Code points are separated by dots. If none, there is only one code
12060 * point, and is terminated by the brace */
12062 if (endchar >= endbrace) {
12063 STRLEN length_of_hex;
12064 I32 grok_hex_flags;
12066 /* Here, exactly one code point. If that isn't what is wanted, fail */
12067 if (! code_point_p) {
12072 /* Convert code point from hex */
12073 length_of_hex = (STRLEN)(endchar - RExC_parse);
12074 grok_hex_flags = PERL_SCAN_ALLOW_UNDERSCORES
12075 | PERL_SCAN_DISALLOW_PREFIX
12077 /* No errors in the first pass (See [perl
12078 * #122671].) We let the code below find the
12079 * errors when there are multiple chars. */
12081 ? PERL_SCAN_SILENT_ILLDIGIT
12084 /* This routine is the one place where both single- and double-quotish
12085 * \N{U+xxxx} are evaluated. The value is a Unicode code point which
12086 * must be converted to native. */
12087 *code_point_p = UNI_TO_NATIVE(grok_hex(RExC_parse,
12092 /* The tokenizer should have guaranteed validity, but it's possible to
12093 * bypass it by using single quoting, so check. Don't do the check
12094 * here when there are multiple chars; we do it below anyway. */
12095 if (length_of_hex == 0
12096 || length_of_hex != (STRLEN)(endchar - RExC_parse) )
12098 RExC_parse += length_of_hex; /* Includes all the valid */
12099 RExC_parse += (RExC_orig_utf8) /* point to after 1st invalid */
12100 ? UTF8SKIP(RExC_parse)
12102 /* Guard against malformed utf8 */
12103 if (RExC_parse >= endchar) {
12104 RExC_parse = endchar;
12106 vFAIL("Invalid hexadecimal number in \\N{U+...}");
12109 RExC_parse = endbrace + 1;
12112 else { /* Is a multiple character sequence */
12113 SV * substitute_parse;
12115 char *orig_end = RExC_end;
12116 char *save_start = RExC_start;
12119 /* Count the code points, if desired, in the sequence */
12122 while (RExC_parse < endbrace) {
12123 /* Point to the beginning of the next character in the sequence. */
12124 RExC_parse = endchar + 1;
12125 endchar = RExC_parse + strcspn(RExC_parse, ".}");
12130 /* Fail if caller doesn't want to handle a multi-code-point sequence.
12131 * But don't backup up the pointer if the caller want to know how many
12132 * code points there are (they can then handle things) */
12140 /* What is done here is to convert this to a sub-pattern of the form
12141 * \x{char1}\x{char2}... and then call reg recursively to parse it
12142 * (enclosing in "(?: ... )" ). That way, it retains its atomicness,
12143 * while not having to worry about special handling that some code
12144 * points may have. */
12146 substitute_parse = newSVpvs("?:");
12148 while (RExC_parse < endbrace) {
12150 /* Convert to notation the rest of the code understands */
12151 sv_catpv(substitute_parse, "\\x{");
12152 sv_catpvn(substitute_parse, RExC_parse, endchar - RExC_parse);
12153 sv_catpv(substitute_parse, "}");
12155 /* Point to the beginning of the next character in the sequence. */
12156 RExC_parse = endchar + 1;
12157 endchar = RExC_parse + strcspn(RExC_parse, ".}");
12160 sv_catpv(substitute_parse, ")");
12162 RExC_parse = RExC_start = RExC_adjusted_start = SvPV(substitute_parse,
12165 /* Don't allow empty number */
12166 if (len < (STRLEN) 8) {
12167 RExC_parse = endbrace;
12168 vFAIL("Invalid hexadecimal number in \\N{U+...}");
12170 RExC_end = RExC_parse + len;
12172 /* The values are Unicode, and therefore not subject to recoding, but
12173 * have to be converted to native on a non-Unicode (meaning non-ASCII)
12176 RExC_recode_x_to_native = 1;
12180 if (!(*node_p = reg(pRExC_state, 1, &flags, depth+1))) {
12181 if (flags & (RESTART_PASS1|NEED_UTF8)) {
12182 *flagp = flags & (RESTART_PASS1|NEED_UTF8);
12185 FAIL2("panic: reg returned NULL to grok_bslash_N, flags=%#" UVxf,
12188 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
12191 /* Restore the saved values */
12192 RExC_start = RExC_adjusted_start = save_start;
12193 RExC_parse = endbrace;
12194 RExC_end = orig_end;
12196 RExC_recode_x_to_native = 0;
12199 SvREFCNT_dec_NN(substitute_parse);
12200 nextchar(pRExC_state);
12207 PERL_STATIC_INLINE U8
12208 S_compute_EXACTish(RExC_state_t *pRExC_state)
12212 PERL_ARGS_ASSERT_COMPUTE_EXACTISH;
12220 op = get_regex_charset(RExC_flags);
12221 if (op >= REGEX_ASCII_RESTRICTED_CHARSET) {
12222 op--; /* /a is same as /u, and map /aa's offset to what /a's would have
12223 been, so there is no hole */
12226 return op + EXACTF;
12229 PERL_STATIC_INLINE void
12230 S_alloc_maybe_populate_EXACT(pTHX_ RExC_state_t *pRExC_state,
12231 regnode *node, I32* flagp, STRLEN len, UV code_point,
12234 /* This knows the details about sizing an EXACTish node, setting flags for
12235 * it (by setting <*flagp>, and potentially populating it with a single
12238 * If <len> (the length in bytes) is non-zero, this function assumes that
12239 * the node has already been populated, and just does the sizing. In this
12240 * case <code_point> should be the final code point that has already been
12241 * placed into the node. This value will be ignored except that under some
12242 * circumstances <*flagp> is set based on it.
12244 * If <len> is zero, the function assumes that the node is to contain only
12245 * the single character given by <code_point> and calculates what <len>
12246 * should be. In pass 1, it sizes the node appropriately. In pass 2, it
12247 * additionally will populate the node's STRING with <code_point> or its
12250 * In both cases <*flagp> is appropriately set
12252 * It knows that under FOLD, the Latin Sharp S and UTF characters above
12253 * 255, must be folded (the former only when the rules indicate it can
12256 * When it does the populating, it looks at the flag 'downgradable'. If
12257 * true with a node that folds, it checks if the single code point
12258 * participates in a fold, and if not downgrades the node to an EXACT.
12259 * This helps the optimizer */
12261 bool len_passed_in = cBOOL(len != 0);
12262 U8 character[UTF8_MAXBYTES_CASE+1];
12264 PERL_ARGS_ASSERT_ALLOC_MAYBE_POPULATE_EXACT;
12266 /* Don't bother to check for downgrading in PASS1, as it doesn't make any
12267 * sizing difference, and is extra work that is thrown away */
12268 if (downgradable && ! PASS2) {
12269 downgradable = FALSE;
12272 if (! len_passed_in) {
12274 if (UVCHR_IS_INVARIANT(code_point)) {
12275 if (LOC || ! FOLD) { /* /l defers folding until runtime */
12276 *character = (U8) code_point;
12278 else { /* Here is /i and not /l. (toFOLD() is defined on just
12279 ASCII, which isn't the same thing as INVARIANT on
12280 EBCDIC, but it works there, as the extra invariants
12281 fold to themselves) */
12282 *character = toFOLD((U8) code_point);
12284 /* We can downgrade to an EXACT node if this character
12285 * isn't a folding one. Note that this assumes that
12286 * nothing above Latin1 folds to some other invariant than
12287 * one of these alphabetics; otherwise we would also have
12289 * && (! HAS_NONLATIN1_FOLD_CLOSURE(code_point)
12290 * || ASCII_FOLD_RESTRICTED))
12292 if (downgradable && PL_fold[code_point] == code_point) {
12298 else if (FOLD && (! LOC
12299 || ! is_PROBLEMATIC_LOCALE_FOLD_cp(code_point)))
12300 { /* Folding, and ok to do so now */
12301 UV folded = _to_uni_fold_flags(
12305 FOLD_FLAGS_FULL | ((ASCII_FOLD_RESTRICTED)
12306 ? FOLD_FLAGS_NOMIX_ASCII
12309 && folded == code_point /* This quickly rules out many
12310 cases, avoiding the
12311 _invlist_contains_cp() overhead
12313 && ! _invlist_contains_cp(PL_utf8_foldable, code_point))
12320 else if (code_point <= MAX_UTF8_TWO_BYTE) {
12322 /* Not folding this cp, and can output it directly */
12323 *character = UTF8_TWO_BYTE_HI(code_point);
12324 *(character + 1) = UTF8_TWO_BYTE_LO(code_point);
12328 uvchr_to_utf8( character, code_point);
12329 len = UTF8SKIP(character);
12331 } /* Else pattern isn't UTF8. */
12333 *character = (U8) code_point;
12335 } /* Else is folded non-UTF8 */
12336 #if UNICODE_MAJOR_VERSION > 3 /* no multifolds in early Unicode */ \
12337 || (UNICODE_MAJOR_VERSION == 3 && ( UNICODE_DOT_VERSION > 0) \
12338 || UNICODE_DOT_DOT_VERSION > 0)
12339 else if (LIKELY(code_point != LATIN_SMALL_LETTER_SHARP_S)) {
12343 /* We don't fold any non-UTF8 except possibly the Sharp s (see
12344 * comments at join_exact()); */
12345 *character = (U8) code_point;
12348 /* Can turn into an EXACT node if we know the fold at compile time,
12349 * and it folds to itself and doesn't particpate in other folds */
12352 && PL_fold_latin1[code_point] == code_point
12353 && (! HAS_NONLATIN1_FOLD_CLOSURE(code_point)
12354 || (isASCII(code_point) && ASCII_FOLD_RESTRICTED)))
12358 } /* else is Sharp s. May need to fold it */
12359 else if (AT_LEAST_UNI_SEMANTICS && ! ASCII_FOLD_RESTRICTED) {
12361 *(character + 1) = 's';
12365 *character = LATIN_SMALL_LETTER_SHARP_S;
12371 RExC_size += STR_SZ(len);
12374 RExC_emit += STR_SZ(len);
12375 STR_LEN(node) = len;
12376 if (! len_passed_in) {
12377 Copy((char *) character, STRING(node), len, char);
12381 *flagp |= HASWIDTH;
12383 /* A single character node is SIMPLE, except for the special-cased SHARP S
12385 if ((len == 1 || (UTF && len == UVCHR_SKIP(code_point)))
12386 #if UNICODE_MAJOR_VERSION > 3 /* no multifolds in early Unicode */ \
12387 || (UNICODE_MAJOR_VERSION == 3 && ( UNICODE_DOT_VERSION > 0) \
12388 || UNICODE_DOT_DOT_VERSION > 0)
12389 && ( code_point != LATIN_SMALL_LETTER_SHARP_S
12390 || ! FOLD || ! DEPENDS_SEMANTICS)
12396 /* The OP may not be well defined in PASS1 */
12397 if (PASS2 && OP(node) == EXACTFL) {
12398 RExC_contains_locale = 1;
12403 /* Parse backref decimal value, unless it's too big to sensibly be a backref,
12404 * in which case return I32_MAX (rather than possibly 32-bit wrapping) */
12407 S_backref_value(char *p)
12409 const char* endptr;
12411 if (grok_atoUV(p, &val, &endptr) && val <= I32_MAX)
12418 - regatom - the lowest level
12420 Try to identify anything special at the start of the current parse position.
12421 If there is, then handle it as required. This may involve generating a
12422 single regop, such as for an assertion; or it may involve recursing, such as
12423 to handle a () structure.
12425 If the string doesn't start with something special then we gobble up
12426 as much literal text as we can. If we encounter a quantifier, we have to
12427 back off the final literal character, as that quantifier applies to just it
12428 and not to the whole string of literals.
12430 Once we have been able to handle whatever type of thing started the
12431 sequence, we return.
12433 Note: we have to be careful with escapes, as they can be both literal
12434 and special, and in the case of \10 and friends, context determines which.
12436 A summary of the code structure is:
12438 switch (first_byte) {
12439 cases for each special:
12440 handle this special;
12443 switch (2nd byte) {
12444 cases for each unambiguous special:
12445 handle this special;
12447 cases for each ambigous special/literal:
12449 if (special) handle here
12451 default: // unambiguously literal:
12454 default: // is a literal char
12457 create EXACTish node for literal;
12458 while (more input and node isn't full) {
12459 switch (input_byte) {
12460 cases for each special;
12461 make sure parse pointer is set so that the next call to
12462 regatom will see this special first
12463 goto loopdone; // EXACTish node terminated by prev. char
12465 append char to EXACTISH node;
12467 get next input byte;
12471 return the generated node;
12473 Specifically there are two separate switches for handling
12474 escape sequences, with the one for handling literal escapes requiring
12475 a dummy entry for all of the special escapes that are actually handled
12478 Returns NULL, setting *flagp to TRYAGAIN if reg() returns NULL with
12480 Returns NULL, setting *flagp to RESTART_PASS1 if the sizing scan needs to be
12481 restarted, or'd with NEED_UTF8 if the pattern needs to be upgraded to UTF-8
12482 Otherwise does not return NULL.
12486 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
12488 regnode *ret = NULL;
12495 GET_RE_DEBUG_FLAGS_DECL;
12497 *flagp = WORST; /* Tentatively. */
12499 DEBUG_PARSE("atom");
12501 PERL_ARGS_ASSERT_REGATOM;
12504 parse_start = RExC_parse;
12505 assert(RExC_parse < RExC_end);
12506 switch ((U8)*RExC_parse) {
12508 RExC_seen_zerolen++;
12509 nextchar(pRExC_state);
12510 if (RExC_flags & RXf_PMf_MULTILINE)
12511 ret = reg_node(pRExC_state, MBOL);
12513 ret = reg_node(pRExC_state, SBOL);
12514 Set_Node_Length(ret, 1); /* MJD */
12517 nextchar(pRExC_state);
12519 RExC_seen_zerolen++;
12520 if (RExC_flags & RXf_PMf_MULTILINE)
12521 ret = reg_node(pRExC_state, MEOL);
12523 ret = reg_node(pRExC_state, SEOL);
12524 Set_Node_Length(ret, 1); /* MJD */
12527 nextchar(pRExC_state);
12528 if (RExC_flags & RXf_PMf_SINGLELINE)
12529 ret = reg_node(pRExC_state, SANY);
12531 ret = reg_node(pRExC_state, REG_ANY);
12532 *flagp |= HASWIDTH|SIMPLE;
12534 Set_Node_Length(ret, 1); /* MJD */
12538 char * const oregcomp_parse = ++RExC_parse;
12539 ret = regclass(pRExC_state, flagp,depth+1,
12540 FALSE, /* means parse the whole char class */
12541 TRUE, /* allow multi-char folds */
12542 FALSE, /* don't silence non-portable warnings. */
12543 (bool) RExC_strict,
12544 TRUE, /* Allow an optimized regnode result */
12548 if (*flagp & (RESTART_PASS1|NEED_UTF8))
12550 FAIL2("panic: regclass returned NULL to regatom, flags=%#" UVxf,
12553 if (*RExC_parse != ']') {
12554 RExC_parse = oregcomp_parse;
12555 vFAIL("Unmatched [");
12557 nextchar(pRExC_state);
12558 Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
12562 nextchar(pRExC_state);
12563 ret = reg(pRExC_state, 2, &flags,depth+1);
12565 if (flags & TRYAGAIN) {
12566 if (RExC_parse >= RExC_end) {
12567 /* Make parent create an empty node if needed. */
12568 *flagp |= TRYAGAIN;
12573 if (flags & (RESTART_PASS1|NEED_UTF8)) {
12574 *flagp = flags & (RESTART_PASS1|NEED_UTF8);
12577 FAIL2("panic: reg returned NULL to regatom, flags=%#" UVxf,
12580 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
12584 if (flags & TRYAGAIN) {
12585 *flagp |= TRYAGAIN;
12588 vFAIL("Internal urp");
12589 /* Supposed to be caught earlier. */
12595 vFAIL("Quantifier follows nothing");
12600 This switch handles escape sequences that resolve to some kind
12601 of special regop and not to literal text. Escape sequnces that
12602 resolve to literal text are handled below in the switch marked
12605 Every entry in this switch *must* have a corresponding entry
12606 in the literal escape switch. However, the opposite is not
12607 required, as the default for this switch is to jump to the
12608 literal text handling code.
12611 switch ((U8)*RExC_parse) {
12612 /* Special Escapes */
12614 RExC_seen_zerolen++;
12615 ret = reg_node(pRExC_state, SBOL);
12616 /* SBOL is shared with /^/ so we set the flags so we can tell
12617 * /\A/ from /^/ in split. We check ret because first pass we
12618 * have no regop struct to set the flags on. */
12622 goto finish_meta_pat;
12624 ret = reg_node(pRExC_state, GPOS);
12625 RExC_seen |= REG_GPOS_SEEN;
12627 goto finish_meta_pat;
12629 RExC_seen_zerolen++;
12630 ret = reg_node(pRExC_state, KEEPS);
12632 /* XXX:dmq : disabling in-place substitution seems to
12633 * be necessary here to avoid cases of memory corruption, as
12634 * with: C<$_="x" x 80; s/x\K/y/> -- rgs
12636 RExC_seen |= REG_LOOKBEHIND_SEEN;
12637 goto finish_meta_pat;
12639 ret = reg_node(pRExC_state, SEOL);
12641 RExC_seen_zerolen++; /* Do not optimize RE away */
12642 goto finish_meta_pat;
12644 ret = reg_node(pRExC_state, EOS);
12646 RExC_seen_zerolen++; /* Do not optimize RE away */
12647 goto finish_meta_pat;
12649 vFAIL("\\C no longer supported");
12651 ret = reg_node(pRExC_state, CLUMP);
12652 *flagp |= HASWIDTH;
12653 goto finish_meta_pat;
12659 arg = ANYOF_WORDCHAR;
12667 regex_charset charset = get_regex_charset(RExC_flags);
12669 RExC_seen_zerolen++;
12670 RExC_seen |= REG_LOOKBEHIND_SEEN;
12671 op = BOUND + charset;
12673 if (op == BOUNDL) {
12674 RExC_contains_locale = 1;
12677 ret = reg_node(pRExC_state, op);
12679 if (RExC_parse >= RExC_end || *(RExC_parse + 1) != '{') {
12680 FLAGS(ret) = TRADITIONAL_BOUND;
12681 if (PASS2 && op > BOUNDA) { /* /aa is same as /a */
12687 char name = *RExC_parse;
12690 endbrace = strchr(RExC_parse, '}');
12693 vFAIL2("Missing right brace on \\%c{}", name);
12695 /* XXX Need to decide whether to take spaces or not. Should be
12696 * consistent with \p{}, but that currently is SPACE, which
12697 * means vertical too, which seems wrong
12698 * while (isBLANK(*RExC_parse)) {
12701 if (endbrace == RExC_parse) {
12702 RExC_parse++; /* After the '}' */
12703 vFAIL2("Empty \\%c{}", name);
12705 length = endbrace - RExC_parse;
12706 /*while (isBLANK(*(RExC_parse + length - 1))) {
12709 switch (*RExC_parse) {
12712 && (length != 3 || strnNE(RExC_parse + 1, "cb", 2)))
12714 goto bad_bound_type;
12716 FLAGS(ret) = GCB_BOUND;
12719 if (length != 2 || *(RExC_parse + 1) != 'b') {
12720 goto bad_bound_type;
12722 FLAGS(ret) = LB_BOUND;
12725 if (length != 2 || *(RExC_parse + 1) != 'b') {
12726 goto bad_bound_type;
12728 FLAGS(ret) = SB_BOUND;
12731 if (length != 2 || *(RExC_parse + 1) != 'b') {
12732 goto bad_bound_type;
12734 FLAGS(ret) = WB_BOUND;
12738 RExC_parse = endbrace;
12740 "'%" UTF8f "' is an unknown bound type",
12741 UTF8fARG(UTF, length, endbrace - length));
12742 NOT_REACHED; /*NOTREACHED*/
12744 RExC_parse = endbrace;
12745 REQUIRE_UNI_RULES(flagp, NULL);
12747 if (PASS2 && op >= BOUNDA) { /* /aa is same as /a */
12751 /* Don't have to worry about UTF-8, in this message because
12752 * to get here the contents of the \b must be ASCII */
12753 ckWARN4reg(RExC_parse + 1, /* Include the '}' in msg */
12754 "Using /u for '%.*s' instead of /%s",
12756 endbrace - length + 1,
12757 (charset == REGEX_ASCII_RESTRICTED_CHARSET)
12758 ? ASCII_RESTRICT_PAT_MODS
12759 : ASCII_MORE_RESTRICT_PAT_MODS);
12763 if (PASS2 && invert) {
12764 OP(ret) += NBOUND - BOUND;
12766 goto finish_meta_pat;
12774 if (! DEPENDS_SEMANTICS) {
12778 /* \d doesn't have any matches in the upper Latin1 range, hence /d
12779 * is equivalent to /u. Changing to /u saves some branches at
12782 goto join_posix_op_known;
12785 ret = reg_node(pRExC_state, LNBREAK);
12786 *flagp |= HASWIDTH|SIMPLE;
12787 goto finish_meta_pat;
12795 goto join_posix_op_known;
12801 arg = ANYOF_VERTWS;
12803 goto join_posix_op_known;
12813 op = POSIXD + get_regex_charset(RExC_flags);
12814 if (op > POSIXA) { /* /aa is same as /a */
12817 else if (op == POSIXL) {
12818 RExC_contains_locale = 1;
12821 join_posix_op_known:
12824 op += NPOSIXD - POSIXD;
12827 ret = reg_node(pRExC_state, op);
12829 FLAGS(ret) = namedclass_to_classnum(arg);
12832 *flagp |= HASWIDTH|SIMPLE;
12836 nextchar(pRExC_state);
12837 Set_Node_Length(ret, 2); /* MJD */
12843 ret = regclass(pRExC_state, flagp,depth+1,
12844 TRUE, /* means just parse this element */
12845 FALSE, /* don't allow multi-char folds */
12846 FALSE, /* don't silence non-portable warnings. It
12847 would be a bug if these returned
12849 (bool) RExC_strict,
12850 TRUE, /* Allow an optimized regnode result */
12853 if (*flagp & RESTART_PASS1)
12855 /* regclass() can only return RESTART_PASS1 and NEED_UTF8 if
12856 * multi-char folds are allowed. */
12858 FAIL2("panic: regclass returned NULL to regatom, flags=%#" UVxf,
12863 Set_Node_Offset(ret, parse_start);
12864 Set_Node_Cur_Length(ret, parse_start - 2);
12865 nextchar(pRExC_state);
12868 /* Handle \N, \N{} and \N{NAMED SEQUENCE} (the latter meaning the
12869 * \N{...} evaluates to a sequence of more than one code points).
12870 * The function call below returns a regnode, which is our result.
12871 * The parameters cause it to fail if the \N{} evaluates to a
12872 * single code point; we handle those like any other literal. The
12873 * reason that the multicharacter case is handled here and not as
12874 * part of the EXACtish code is because of quantifiers. In
12875 * /\N{BLAH}+/, the '+' applies to the whole thing, and doing it
12876 * this way makes that Just Happen. dmq.
12877 * join_exact() will join this up with adjacent EXACTish nodes
12878 * later on, if appropriate. */
12880 if (grok_bslash_N(pRExC_state,
12881 &ret, /* Want a regnode returned */
12882 NULL, /* Fail if evaluates to a single code
12884 NULL, /* Don't need a count of how many code
12893 if (*flagp & RESTART_PASS1)
12896 /* Here, evaluates to a single code point. Go get that */
12897 RExC_parse = parse_start;
12900 case 'k': /* Handle \k<NAME> and \k'NAME' */
12904 if ( RExC_parse >= RExC_end - 1
12905 || (( ch = RExC_parse[1]) != '<'
12910 /* diag_listed_as: Sequence \%s... not terminated in regex; marked by <-- HERE in m/%s/ */
12911 vFAIL2("Sequence %.2s... not terminated",parse_start);
12914 ret = handle_named_backref(pRExC_state,
12926 case '1': case '2': case '3': case '4':
12927 case '5': case '6': case '7': case '8': case '9':
12932 if (*RExC_parse == 'g') {
12936 if (*RExC_parse == '{') {
12940 if (*RExC_parse == '-') {
12944 if (hasbrace && !isDIGIT(*RExC_parse)) {
12945 if (isrel) RExC_parse--;
12947 goto parse_named_seq;
12950 if (RExC_parse >= RExC_end) {
12951 goto unterminated_g;
12953 num = S_backref_value(RExC_parse);
12955 vFAIL("Reference to invalid group 0");
12956 else if (num == I32_MAX) {
12957 if (isDIGIT(*RExC_parse))
12958 vFAIL("Reference to nonexistent group");
12961 vFAIL("Unterminated \\g... pattern");
12965 num = RExC_npar - num;
12967 vFAIL("Reference to nonexistent or unclosed group");
12971 num = S_backref_value(RExC_parse);
12972 /* bare \NNN might be backref or octal - if it is larger
12973 * than or equal RExC_npar then it is assumed to be an
12974 * octal escape. Note RExC_npar is +1 from the actual
12975 * number of parens. */
12976 /* Note we do NOT check if num == I32_MAX here, as that is
12977 * handled by the RExC_npar check */
12980 /* any numeric escape < 10 is always a backref */
12982 /* any numeric escape < RExC_npar is a backref */
12983 && num >= RExC_npar
12984 /* cannot be an octal escape if it starts with 8 */
12985 && *RExC_parse != '8'
12986 /* cannot be an octal escape it it starts with 9 */
12987 && *RExC_parse != '9'
12990 /* Probably not a backref, instead likely to be an
12991 * octal character escape, e.g. \35 or \777.
12992 * The above logic should make it obvious why using
12993 * octal escapes in patterns is problematic. - Yves */
12994 RExC_parse = parse_start;
12999 /* At this point RExC_parse points at a numeric escape like
13000 * \12 or \88 or something similar, which we should NOT treat
13001 * as an octal escape. It may or may not be a valid backref
13002 * escape. For instance \88888888 is unlikely to be a valid
13004 while (isDIGIT(*RExC_parse))
13007 if (*RExC_parse != '}')
13008 vFAIL("Unterminated \\g{...} pattern");
13012 if (num > (I32)RExC_rx->nparens)
13013 vFAIL("Reference to nonexistent group");
13016 ret = reganode(pRExC_state,
13019 : (ASCII_FOLD_RESTRICTED)
13021 : (AT_LEAST_UNI_SEMANTICS)
13027 *flagp |= HASWIDTH;
13029 /* override incorrect value set in reganode MJD */
13030 Set_Node_Offset(ret, parse_start);
13031 Set_Node_Cur_Length(ret, parse_start-1);
13032 skip_to_be_ignored_text(pRExC_state, &RExC_parse,
13033 FALSE /* Don't force to /x */ );
13037 if (RExC_parse >= RExC_end)
13038 FAIL("Trailing \\");
13041 /* Do not generate "unrecognized" warnings here, we fall
13042 back into the quick-grab loop below */
13043 RExC_parse = parse_start;
13045 } /* end of switch on a \foo sequence */
13050 /* '#' comments should have been spaced over before this function was
13052 assert((RExC_flags & RXf_PMf_EXTENDED) == 0);
13054 if (RExC_flags & RXf_PMf_EXTENDED) {
13055 RExC_parse = reg_skipcomment( pRExC_state, RExC_parse );
13056 if (RExC_parse < RExC_end)
13066 /* Here, we have determined that the next thing is probably a
13067 * literal character. RExC_parse points to the first byte of its
13068 * definition. (It still may be an escape sequence that evaluates
13069 * to a single character) */
13075 #define MAX_NODE_STRING_SIZE 127
13076 char foldbuf[MAX_NODE_STRING_SIZE+UTF8_MAXBYTES_CASE];
13078 U8 upper_parse = MAX_NODE_STRING_SIZE;
13079 U8 node_type = compute_EXACTish(pRExC_state);
13080 bool next_is_quantifier;
13081 char * oldp = NULL;
13083 /* We can convert EXACTF nodes to EXACTFU if they contain only
13084 * characters that match identically regardless of the target
13085 * string's UTF8ness. The reason to do this is that EXACTF is not
13086 * trie-able, EXACTFU is.
13088 * Similarly, we can convert EXACTFL nodes to EXACTFLU8 if they
13089 * contain only above-Latin1 characters (hence must be in UTF8),
13090 * which don't participate in folds with Latin1-range characters,
13091 * as the latter's folds aren't known until runtime. (We don't
13092 * need to figure this out until pass 2) */
13093 bool maybe_exactfu = PASS2
13094 && (node_type == EXACTF || node_type == EXACTFL);
13096 /* If a folding node contains only code points that don't
13097 * participate in folds, it can be changed into an EXACT node,
13098 * which allows the optimizer more things to look for */
13101 ret = reg_node(pRExC_state, node_type);
13103 /* In pass1, folded, we use a temporary buffer instead of the
13104 * actual node, as the node doesn't exist yet */
13105 s = (SIZE_ONLY && FOLD) ? foldbuf : STRING(ret);
13111 /* We look for the EXACTFish to EXACT node optimizaton only if
13112 * folding. (And we don't need to figure this out until pass 2).
13113 * XXX It might actually make sense to split the node into portions
13114 * that are exact and ones that aren't, so that we could later use
13115 * the exact ones to find the longest fixed and floating strings.
13116 * One would want to join them back into a larger node. One could
13117 * use a pseudo regnode like 'EXACT_ORIG_FOLD' */
13118 maybe_exact = FOLD && PASS2;
13120 /* XXX The node can hold up to 255 bytes, yet this only goes to
13121 * 127. I (khw) do not know why. Keeping it somewhat less than
13122 * 255 allows us to not have to worry about overflow due to
13123 * converting to utf8 and fold expansion, but that value is
13124 * 255-UTF8_MAXBYTES_CASE. join_exact() may join adjacent nodes
13125 * split up by this limit into a single one using the real max of
13126 * 255. Even at 127, this breaks under rare circumstances. If
13127 * folding, we do not want to split a node at a character that is a
13128 * non-final in a multi-char fold, as an input string could just
13129 * happen to want to match across the node boundary. The join
13130 * would solve that problem if the join actually happens. But a
13131 * series of more than two nodes in a row each of 127 would cause
13132 * the first join to succeed to get to 254, but then there wouldn't
13133 * be room for the next one, which could at be one of those split
13134 * multi-char folds. I don't know of any fool-proof solution. One
13135 * could back off to end with only a code point that isn't such a
13136 * non-final, but it is possible for there not to be any in the
13139 assert( ! UTF /* Is at the beginning of a character */
13140 || UTF8_IS_INVARIANT(UCHARAT(RExC_parse))
13141 || UTF8_IS_START(UCHARAT(RExC_parse)));
13143 /* Here, we have a literal character. Find the maximal string of
13144 * them in the input that we can fit into a single EXACTish node.
13145 * We quit at the first non-literal or when the node gets full */
13146 for (p = RExC_parse;
13147 len < upper_parse && p < RExC_end;
13152 /* White space has already been ignored */
13153 assert( (RExC_flags & RXf_PMf_EXTENDED) == 0
13154 || ! is_PATWS_safe((p), RExC_end, UTF));
13166 /* Literal Escapes Switch
13168 This switch is meant to handle escape sequences that
13169 resolve to a literal character.
13171 Every escape sequence that represents something
13172 else, like an assertion or a char class, is handled
13173 in the switch marked 'Special Escapes' above in this
13174 routine, but also has an entry here as anything that
13175 isn't explicitly mentioned here will be treated as
13176 an unescaped equivalent literal.
13179 switch ((U8)*++p) {
13180 /* These are all the special escapes. */
13181 case 'A': /* Start assertion */
13182 case 'b': case 'B': /* Word-boundary assertion*/
13183 case 'C': /* Single char !DANGEROUS! */
13184 case 'd': case 'D': /* digit class */
13185 case 'g': case 'G': /* generic-backref, pos assertion */
13186 case 'h': case 'H': /* HORIZWS */
13187 case 'k': case 'K': /* named backref, keep marker */
13188 case 'p': case 'P': /* Unicode property */
13189 case 'R': /* LNBREAK */
13190 case 's': case 'S': /* space class */
13191 case 'v': case 'V': /* VERTWS */
13192 case 'w': case 'W': /* word class */
13193 case 'X': /* eXtended Unicode "combining
13194 character sequence" */
13195 case 'z': case 'Z': /* End of line/string assertion */
13199 /* Anything after here is an escape that resolves to a
13200 literal. (Except digits, which may or may not)
13206 case 'N': /* Handle a single-code point named character. */
13207 RExC_parse = p + 1;
13208 if (! grok_bslash_N(pRExC_state,
13209 NULL, /* Fail if evaluates to
13210 anything other than a
13211 single code point */
13212 &ender, /* The returned single code
13214 NULL, /* Don't need a count of
13215 how many code points */
13220 if (*flagp & NEED_UTF8)
13221 FAIL("panic: grok_bslash_N set NEED_UTF8");
13222 if (*flagp & RESTART_PASS1)
13225 /* Here, it wasn't a single code point. Go close
13226 * up this EXACTish node. The switch() prior to
13227 * this switch handles the other cases */
13228 RExC_parse = p = oldp;
13232 if (ender > 0xff) {
13233 REQUIRE_UTF8(flagp);
13249 ender = ESC_NATIVE;
13259 const char* error_msg;
13261 bool valid = grok_bslash_o(&p,
13264 PASS2, /* out warnings */
13265 (bool) RExC_strict,
13266 TRUE, /* Output warnings
13271 RExC_parse = p; /* going to die anyway; point
13272 to exact spot of failure */
13276 if (ender > 0xff) {
13277 REQUIRE_UTF8(flagp);
13283 UV result = UV_MAX; /* initialize to erroneous
13285 const char* error_msg;
13287 bool valid = grok_bslash_x(&p,
13290 PASS2, /* out warnings */
13291 (bool) RExC_strict,
13292 TRUE, /* Silence warnings
13297 RExC_parse = p; /* going to die anyway; point
13298 to exact spot of failure */
13303 if (ender < 0x100) {
13305 if (RExC_recode_x_to_native) {
13306 ender = LATIN1_TO_NATIVE(ender);
13311 REQUIRE_UTF8(flagp);
13317 ender = grok_bslash_c(*p++, PASS2);
13319 case '8': case '9': /* must be a backreference */
13321 /* we have an escape like \8 which cannot be an octal escape
13322 * so we exit the loop, and let the outer loop handle this
13323 * escape which may or may not be a legitimate backref. */
13325 case '1': case '2': case '3':case '4':
13326 case '5': case '6': case '7':
13327 /* When we parse backslash escapes there is ambiguity
13328 * between backreferences and octal escapes. Any escape
13329 * from \1 - \9 is a backreference, any multi-digit
13330 * escape which does not start with 0 and which when
13331 * evaluated as decimal could refer to an already
13332 * parsed capture buffer is a back reference. Anything
13335 * Note this implies that \118 could be interpreted as
13336 * 118 OR as "\11" . "8" depending on whether there
13337 * were 118 capture buffers defined already in the
13340 /* NOTE, RExC_npar is 1 more than the actual number of
13341 * parens we have seen so far, hence the < RExC_npar below. */
13343 if ( !isDIGIT(p[1]) || S_backref_value(p) < RExC_npar)
13344 { /* Not to be treated as an octal constant, go
13352 I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
13354 ender = grok_oct(p, &numlen, &flags, NULL);
13355 if (ender > 0xff) {
13356 REQUIRE_UTF8(flagp);
13359 if (PASS2 /* like \08, \178 */
13361 && isDIGIT(*p) && ckWARN(WARN_REGEXP))
13363 reg_warn_non_literal_string(
13365 form_short_octal_warning(p, numlen));
13371 FAIL("Trailing \\");
13374 if (!SIZE_ONLY&& isALPHANUMERIC(*p)) {
13375 /* Include any left brace following the alpha to emphasize
13376 * that it could be part of an escape at some point
13378 int len = (isALPHA(*p) && *(p + 1) == '{') ? 2 : 1;
13379 ckWARN3reg(p + len, "Unrecognized escape \\%.*s passed through", len, p);
13381 goto normal_default;
13382 } /* End of switch on '\' */
13385 /* Currently we don't care if the lbrace is at the start
13386 * of a construct. This catches it in the middle of a
13387 * literal string, or when it's the first thing after
13388 * something like "\b" */
13389 if (len || (p > RExC_start && isALPHA_A(*(p -1)))) {
13391 /* GNU Autoconf is depended on by a lot of code, and
13392 * can't seem to release a new version that avoids the
13393 * deprecation now made fatal. (A commit to do so has
13394 * been in its repository since early 2013; only one
13395 * pattern is affected.) As a work-around, don't
13396 * fatalize this if the pattern being compiled is the
13397 * precise one that trips up Autoconf. See [perl
13398 * #130497] for more details. */
13399 if (memNEs(RExC_start, RExC_end - RExC_start,
13402 RExC_parse = p + 1;
13403 vFAIL("Unescaped left brace in regex is "
13407 ckWARNregdep(p + 1,
13408 "Unescaped left brace in regex is "
13409 "deprecated here (and will be fatal "
13410 "in Perl 5.30), passed through");
13413 goto normal_default;
13416 if (PASS2 && p > RExC_parse && RExC_strict) {
13417 ckWARN2reg(p + 1, "Unescaped literal '%c'", *p);
13420 default: /* A literal character */
13422 if (! UTF8_IS_INVARIANT(*p) && UTF) {
13424 ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
13425 &numlen, UTF8_ALLOW_DEFAULT);
13431 } /* End of switch on the literal */
13433 /* Here, have looked at the literal character and <ender>
13434 * contains its ordinal, <p> points to the character after it.
13435 * We need to check if the next non-ignored thing is a
13436 * quantifier. Move <p> to after anything that should be
13437 * ignored, which, as a side effect, positions <p> for the next
13438 * loop iteration */
13439 skip_to_be_ignored_text(pRExC_state, &p,
13440 FALSE /* Don't force to /x */ );
13442 /* If the next thing is a quantifier, it applies to this
13443 * character only, which means that this character has to be in
13444 * its own node and can't just be appended to the string in an
13445 * existing node, so if there are already other characters in
13446 * the node, close the node with just them, and set up to do
13447 * this character again next time through, when it will be the
13448 * only thing in its new node */
13450 if ((next_is_quantifier = ( LIKELY(p < RExC_end)
13451 && UNLIKELY(ISMULT2(p))))
13458 /* Ready to add 'ender' to the node */
13460 if (! FOLD) { /* The simple case, just append the literal */
13462 /* In the sizing pass, we need only the size of the
13463 * character we are appending, hence we can delay getting
13464 * its representation until PASS2. */
13467 const STRLEN unilen = UVCHR_SKIP(ender);
13470 /* We have to subtract 1 just below (and again in
13471 * the corresponding PASS2 code) because the loop
13472 * increments <len> each time, as all but this path
13473 * (and one other) through it add a single byte to
13474 * the EXACTish node. But these paths would change
13475 * len to be the correct final value, so cancel out
13476 * the increment that follows */
13482 } else { /* PASS2 */
13485 U8 * new_s = uvchr_to_utf8((U8*)s, ender);
13486 len += (char *) new_s - s - 1;
13487 s = (char *) new_s;
13490 *(s++) = (char) ender;
13494 else if (LOC && is_PROBLEMATIC_LOCALE_FOLD_cp(ender)) {
13496 /* Here are folding under /l, and the code point is
13497 * problematic. First, we know we can't simplify things */
13498 maybe_exact = FALSE;
13499 maybe_exactfu = FALSE;
13501 /* A problematic code point in this context means that its
13502 * fold isn't known until runtime, so we can't fold it now.
13503 * (The non-problematic code points are the above-Latin1
13504 * ones that fold to also all above-Latin1. Their folds
13505 * don't vary no matter what the locale is.) But here we
13506 * have characters whose fold depends on the locale.
13507 * Unlike the non-folding case above, we have to keep track
13508 * of these in the sizing pass, so that we can make sure we
13509 * don't split too-long nodes in the middle of a potential
13510 * multi-char fold. And unlike the regular fold case
13511 * handled in the else clauses below, we don't actually
13512 * fold and don't have special cases to consider. What we
13513 * do for both passes is the PASS2 code for non-folding */
13514 goto not_fold_common;
13516 else /* A regular FOLD code point */
13518 #if UNICODE_MAJOR_VERSION > 3 /* no multifolds in early Unicode */ \
13519 || (UNICODE_MAJOR_VERSION == 3 && ( UNICODE_DOT_VERSION > 0) \
13520 || UNICODE_DOT_DOT_VERSION > 0)
13521 /* See comments for join_exact() as to why we fold
13522 * this non-UTF at compile time */
13523 || ( node_type == EXACTFU
13524 && ender == LATIN_SMALL_LETTER_SHARP_S)
13527 /* Here, are folding and are not UTF-8 encoded; therefore
13528 * the character must be in the range 0-255, and is not /l
13529 * (Not /l because we already handled these under /l in
13530 * is_PROBLEMATIC_LOCALE_FOLD_cp) */
13531 if (IS_IN_SOME_FOLD_L1(ender)) {
13532 maybe_exact = FALSE;
13534 /* See if the character's fold differs between /d and
13535 * /u. This includes the multi-char fold SHARP S to
13537 if (UNLIKELY(ender == LATIN_SMALL_LETTER_SHARP_S)) {
13538 RExC_seen_unfolded_sharp_s = 1;
13539 maybe_exactfu = FALSE;
13541 else if (maybe_exactfu
13542 && (PL_fold[ender] != PL_fold_latin1[ender]
13543 #if UNICODE_MAJOR_VERSION > 3 /* no multifolds in early Unicode */ \
13544 || (UNICODE_MAJOR_VERSION == 3 && ( UNICODE_DOT_VERSION > 0) \
13545 || UNICODE_DOT_DOT_VERSION > 0)
13547 && isALPHA_FOLD_EQ(ender, 's')
13548 && isALPHA_FOLD_EQ(*(s-1), 's'))
13551 maybe_exactfu = FALSE;
13555 /* Even when folding, we store just the input character, as
13556 * we have an array that finds its fold quickly */
13557 *(s++) = (char) ender;
13559 else { /* FOLD, and UTF (or sharp s) */
13560 /* Unlike the non-fold case, we do actually have to
13561 * calculate the results here in pass 1. This is for two
13562 * reasons, the folded length may be longer than the
13563 * unfolded, and we have to calculate how many EXACTish
13564 * nodes it will take; and we may run out of room in a node
13565 * in the middle of a potential multi-char fold, and have
13566 * to back off accordingly. */
13569 if (isASCII_uni(ender)) {
13570 folded = toFOLD(ender);
13571 *(s)++ = (U8) folded;
13576 folded = _to_uni_fold_flags(
13580 FOLD_FLAGS_FULL | ((ASCII_FOLD_RESTRICTED)
13581 ? FOLD_FLAGS_NOMIX_ASCII
13585 /* The loop increments <len> each time, as all but this
13586 * path (and one other) through it add a single byte to
13587 * the EXACTish node. But this one has changed len to
13588 * be the correct final value, so subtract one to
13589 * cancel out the increment that follows */
13590 len += foldlen - 1;
13592 /* If this node only contains non-folding code points so
13593 * far, see if this new one is also non-folding */
13595 if (folded != ender) {
13596 maybe_exact = FALSE;
13599 /* Here the fold is the original; we have to check
13600 * further to see if anything folds to it */
13601 if (_invlist_contains_cp(PL_utf8_foldable,
13604 maybe_exact = FALSE;
13611 if (next_is_quantifier) {
13613 /* Here, the next input is a quantifier, and to get here,
13614 * the current character is the only one in the node.
13615 * Also, here <len> doesn't include the final byte for this
13621 } /* End of loop through literal characters */
13623 /* Here we have either exhausted the input or ran out of room in
13624 * the node. (If we encountered a character that can't be in the
13625 * node, transfer is made directly to <loopdone>, and so we
13626 * wouldn't have fallen off the end of the loop.) In the latter
13627 * case, we artificially have to split the node into two, because
13628 * we just don't have enough space to hold everything. This
13629 * creates a problem if the final character participates in a
13630 * multi-character fold in the non-final position, as a match that
13631 * should have occurred won't, due to the way nodes are matched,
13632 * and our artificial boundary. So back off until we find a non-
13633 * problematic character -- one that isn't at the beginning or
13634 * middle of such a fold. (Either it doesn't participate in any
13635 * folds, or appears only in the final position of all the folds it
13636 * does participate in.) A better solution with far fewer false
13637 * positives, and that would fill the nodes more completely, would
13638 * be to actually have available all the multi-character folds to
13639 * test against, and to back-off only far enough to be sure that
13640 * this node isn't ending with a partial one. <upper_parse> is set
13641 * further below (if we need to reparse the node) to include just
13642 * up through that final non-problematic character that this code
13643 * identifies, so when it is set to less than the full node, we can
13644 * skip the rest of this */
13645 if (FOLD && p < RExC_end && upper_parse == MAX_NODE_STRING_SIZE) {
13647 const STRLEN full_len = len;
13649 assert(len >= MAX_NODE_STRING_SIZE);
13651 /* Here, <s> points to the final byte of the final character.
13652 * Look backwards through the string until find a non-
13653 * problematic character */
13657 /* This has no multi-char folds to non-UTF characters */
13658 if (ASCII_FOLD_RESTRICTED) {
13662 while (--s >= s0 && IS_NON_FINAL_FOLD(*s)) { }
13666 if (! PL_NonL1NonFinalFold) {
13667 PL_NonL1NonFinalFold = _new_invlist_C_array(
13668 NonL1_Perl_Non_Final_Folds_invlist);
13671 /* Point to the first byte of the final character */
13672 s = (char *) utf8_hop((U8 *) s, -1);
13674 while (s >= s0) { /* Search backwards until find
13675 non-problematic char */
13676 if (UTF8_IS_INVARIANT(*s)) {
13678 /* There are no ascii characters that participate
13679 * in multi-char folds under /aa. In EBCDIC, the
13680 * non-ascii invariants are all control characters,
13681 * so don't ever participate in any folds. */
13682 if (ASCII_FOLD_RESTRICTED
13683 || ! IS_NON_FINAL_FOLD(*s))
13688 else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
13689 if (! IS_NON_FINAL_FOLD(EIGHT_BIT_UTF8_TO_NATIVE(
13695 else if (! _invlist_contains_cp(
13696 PL_NonL1NonFinalFold,
13697 valid_utf8_to_uvchr((U8 *) s, NULL)))
13702 /* Here, the current character is problematic in that
13703 * it does occur in the non-final position of some
13704 * fold, so try the character before it, but have to
13705 * special case the very first byte in the string, so
13706 * we don't read outside the string */
13707 s = (s == s0) ? s -1 : (char *) utf8_hop((U8 *) s, -1);
13708 } /* End of loop backwards through the string */
13710 /* If there were only problematic characters in the string,
13711 * <s> will point to before s0, in which case the length
13712 * should be 0, otherwise include the length of the
13713 * non-problematic character just found */
13714 len = (s < s0) ? 0 : s - s0 + UTF8SKIP(s);
13717 /* Here, have found the final character, if any, that is
13718 * non-problematic as far as ending the node without splitting
13719 * it across a potential multi-char fold. <len> contains the
13720 * number of bytes in the node up-to and including that
13721 * character, or is 0 if there is no such character, meaning
13722 * the whole node contains only problematic characters. In
13723 * this case, give up and just take the node as-is. We can't
13728 /* If the node ends in an 's' we make sure it stays EXACTF,
13729 * as if it turns into an EXACTFU, it could later get
13730 * joined with another 's' that would then wrongly match
13732 if (maybe_exactfu && isALPHA_FOLD_EQ(ender, 's'))
13734 maybe_exactfu = FALSE;
13738 /* Here, the node does contain some characters that aren't
13739 * problematic. If one such is the final character in the
13740 * node, we are done */
13741 if (len == full_len) {
13744 else if (len + ((UTF) ? UTF8SKIP(s) : 1) == full_len) {
13746 /* If the final character is problematic, but the
13747 * penultimate is not, back-off that last character to
13748 * later start a new node with it */
13753 /* Here, the final non-problematic character is earlier
13754 * in the input than the penultimate character. What we do
13755 * is reparse from the beginning, going up only as far as
13756 * this final ok one, thus guaranteeing that the node ends
13757 * in an acceptable character. The reason we reparse is
13758 * that we know how far in the character is, but we don't
13759 * know how to correlate its position with the input parse.
13760 * An alternate implementation would be to build that
13761 * correlation as we go along during the original parse,
13762 * but that would entail extra work for every node, whereas
13763 * this code gets executed only when the string is too
13764 * large for the node, and the final two characters are
13765 * problematic, an infrequent occurrence. Yet another
13766 * possible strategy would be to save the tail of the
13767 * string, and the next time regatom is called, initialize
13768 * with that. The problem with this is that unless you
13769 * back off one more character, you won't be guaranteed
13770 * regatom will get called again, unless regbranch,
13771 * regpiece ... are also changed. If you do back off that
13772 * extra character, so that there is input guaranteed to
13773 * force calling regatom, you can't handle the case where
13774 * just the first character in the node is acceptable. I
13775 * (khw) decided to try this method which doesn't have that
13776 * pitfall; if performance issues are found, we can do a
13777 * combination of the current approach plus that one */
13783 } /* End of verifying node ends with an appropriate char */
13785 loopdone: /* Jumped to when encounters something that shouldn't be
13788 /* I (khw) don't know if you can get here with zero length, but the
13789 * old code handled this situation by creating a zero-length EXACT
13790 * node. Might as well be NOTHING instead */
13796 /* If 'maybe_exact' is still set here, means there are no
13797 * code points in the node that participate in folds;
13798 * similarly for 'maybe_exactfu' and code points that match
13799 * differently depending on UTF8ness of the target string
13800 * (for /u), or depending on locale for /l */
13806 else if (maybe_exactfu) {
13812 alloc_maybe_populate_EXACT(pRExC_state, ret, flagp, len, ender,
13813 FALSE /* Don't look to see if could
13814 be turned into an EXACT
13815 node, as we have already
13820 RExC_parse = p - 1;
13821 Set_Node_Cur_Length(ret, parse_start);
13824 /* len is STRLEN which is unsigned, need to copy to signed */
13827 vFAIL("Internal disaster");
13830 } /* End of label 'defchar:' */
13832 } /* End of giant switch on input character */
13834 /* Position parse to next real character */
13835 skip_to_be_ignored_text(pRExC_state, &RExC_parse,
13836 FALSE /* Don't force to /x */ );
13837 if (PASS2 && *RExC_parse == '{' && OP(ret) != SBOL && ! regcurly(RExC_parse)) {
13838 ckWARNregdep(RExC_parse + 1, "Unescaped left brace in regex is deprecated here (and will be fatal in Perl 5.30), passed through");
13846 S_populate_ANYOF_from_invlist(pTHX_ regnode *node, SV** invlist_ptr)
13848 /* Uses the inversion list '*invlist_ptr' to populate the ANYOF 'node'. It
13849 * sets up the bitmap and any flags, removing those code points from the
13850 * inversion list, setting it to NULL should it become completely empty */
13852 PERL_ARGS_ASSERT_POPULATE_ANYOF_FROM_INVLIST;
13853 assert(PL_regkind[OP(node)] == ANYOF);
13855 ANYOF_BITMAP_ZERO(node);
13856 if (*invlist_ptr) {
13858 /* This gets set if we actually need to modify things */
13859 bool change_invlist = FALSE;
13863 /* Start looking through *invlist_ptr */
13864 invlist_iterinit(*invlist_ptr);
13865 while (invlist_iternext(*invlist_ptr, &start, &end)) {
13869 if (end == UV_MAX && start <= NUM_ANYOF_CODE_POINTS) {
13870 ANYOF_FLAGS(node) |= ANYOF_MATCHES_ALL_ABOVE_BITMAP;
13873 /* Quit if are above what we should change */
13874 if (start >= NUM_ANYOF_CODE_POINTS) {
13878 change_invlist = TRUE;
13880 /* Set all the bits in the range, up to the max that we are doing */
13881 high = (end < NUM_ANYOF_CODE_POINTS - 1)
13883 : NUM_ANYOF_CODE_POINTS - 1;
13884 for (i = start; i <= (int) high; i++) {
13885 if (! ANYOF_BITMAP_TEST(node, i)) {
13886 ANYOF_BITMAP_SET(node, i);
13890 invlist_iterfinish(*invlist_ptr);
13892 /* Done with loop; remove any code points that are in the bitmap from
13893 * *invlist_ptr; similarly for code points above the bitmap if we have
13894 * a flag to match all of them anyways */
13895 if (change_invlist) {
13896 _invlist_subtract(*invlist_ptr, PL_InBitmap, invlist_ptr);
13898 if (ANYOF_FLAGS(node) & ANYOF_MATCHES_ALL_ABOVE_BITMAP) {
13899 _invlist_intersection(*invlist_ptr, PL_InBitmap, invlist_ptr);
13902 /* If have completely emptied it, remove it completely */
13903 if (_invlist_len(*invlist_ptr) == 0) {
13904 SvREFCNT_dec_NN(*invlist_ptr);
13905 *invlist_ptr = NULL;
13910 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
13911 Character classes ([:foo:]) can also be negated ([:^foo:]).
13912 Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
13913 Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
13914 but trigger failures because they are currently unimplemented. */
13916 #define POSIXCC_DONE(c) ((c) == ':')
13917 #define POSIXCC_NOTYET(c) ((c) == '=' || (c) == '.')
13918 #define POSIXCC(c) (POSIXCC_DONE(c) || POSIXCC_NOTYET(c))
13919 #define MAYBE_POSIXCC(c) (POSIXCC(c) || (c) == '^' || (c) == ';')
13921 #define WARNING_PREFIX "Assuming NOT a POSIX class since "
13922 #define NO_BLANKS_POSIX_WARNING "no blanks are allowed in one"
13923 #define SEMI_COLON_POSIX_WARNING "a semi-colon was found instead of a colon"
13925 #define NOT_MEANT_TO_BE_A_POSIX_CLASS (OOB_NAMEDCLASS - 1)
13927 /* 'posix_warnings' and 'warn_text' are names of variables in the following
13929 #define ADD_POSIX_WARNING(p, text) STMT_START { \
13930 if (posix_warnings) { \
13931 if (! RExC_warn_text ) RExC_warn_text = (AV *) sv_2mortal((SV *) newAV()); \
13932 av_push(RExC_warn_text, Perl_newSVpvf(aTHX_ \
13936 REPORT_LOCATION_ARGS(p))); \
13941 S_handle_possible_posix(pTHX_ RExC_state_t *pRExC_state,
13943 const char * const s, /* Where the putative posix class begins.
13944 Normally, this is one past the '['. This
13945 parameter exists so it can be somewhere
13946 besides RExC_parse. */
13947 char ** updated_parse_ptr, /* Where to set the updated parse pointer, or
13949 AV ** posix_warnings, /* Where to place any generated warnings, or
13951 const bool check_only /* Don't die if error */
13954 /* This parses what the caller thinks may be one of the three POSIX
13956 * 1) a character class, like [:blank:]
13957 * 2) a collating symbol, like [. .]
13958 * 3) an equivalence class, like [= =]
13959 * In the latter two cases, it croaks if it finds a syntactically legal
13960 * one, as these are not handled by Perl.
13962 * The main purpose is to look for a POSIX character class. It returns:
13963 * a) the class number
13964 * if it is a completely syntactically and semantically legal class.
13965 * 'updated_parse_ptr', if not NULL, is set to point to just after the
13966 * closing ']' of the class
13967 * b) OOB_NAMEDCLASS
13968 * if it appears that one of the three POSIX constructs was meant, but
13969 * its specification was somehow defective. 'updated_parse_ptr', if
13970 * not NULL, is set to point to the character just after the end
13971 * character of the class. See below for handling of warnings.
13972 * c) NOT_MEANT_TO_BE_A_POSIX_CLASS
13973 * if it doesn't appear that a POSIX construct was intended.
13974 * 'updated_parse_ptr' is not changed. No warnings nor errors are
13977 * In b) there may be errors or warnings generated. If 'check_only' is
13978 * TRUE, then any errors are discarded. Warnings are returned to the
13979 * caller via an AV* created into '*posix_warnings' if it is not NULL. If
13980 * instead it is NULL, warnings are suppressed. This is done in all
13981 * passes. The reason for this is that the rest of the parsing is heavily
13982 * dependent on whether this routine found a valid posix class or not. If
13983 * it did, the closing ']' is absorbed as part of the class. If no class,
13984 * or an invalid one is found, any ']' will be considered the terminator of
13985 * the outer bracketed character class, leading to very different results.
13986 * In particular, a '(?[ ])' construct will likely have a syntax error if
13987 * the class is parsed other than intended, and this will happen in pass1,
13988 * before the warnings would normally be output. This mechanism allows the
13989 * caller to output those warnings in pass1 just before dieing, giving a
13990 * much better clue as to what is wrong.
13992 * The reason for this function, and its complexity is that a bracketed
13993 * character class can contain just about anything. But it's easy to
13994 * mistype the very specific posix class syntax but yielding a valid
13995 * regular bracketed class, so it silently gets compiled into something
13996 * quite unintended.
13998 * The solution adopted here maintains backward compatibility except that
13999 * it adds a warning if it looks like a posix class was intended but
14000 * improperly specified. The warning is not raised unless what is input
14001 * very closely resembles one of the 14 legal posix classes. To do this,
14002 * it uses fuzzy parsing. It calculates how many single-character edits it
14003 * would take to transform what was input into a legal posix class. Only
14004 * if that number is quite small does it think that the intention was a
14005 * posix class. Obviously these are heuristics, and there will be cases
14006 * where it errs on one side or another, and they can be tweaked as
14007 * experience informs.
14009 * The syntax for a legal posix class is:
14011 * qr/(?xa: \[ : \^? [:lower:]{4,6} : \] )/
14013 * What this routine considers syntactically to be an intended posix class
14014 * is this (the comments indicate some restrictions that the pattern
14017 * qr/(?x: \[? # The left bracket, possibly
14019 * \h* # possibly followed by blanks
14020 * (?: \^ \h* )? # possibly a misplaced caret
14021 * [:;]? # The opening class character,
14022 * # possibly omitted. A typo
14023 * # semi-colon can also be used.
14025 * \^? # possibly a correctly placed
14026 * # caret, but not if there was also
14027 * # a misplaced one
14029 * .{3,15} # The class name. If there are
14030 * # deviations from the legal syntax,
14031 * # its edit distance must be close
14032 * # to a real class name in order
14033 * # for it to be considered to be
14034 * # an intended posix class.
14036 * [:punct:]? # The closing class character,
14037 * # possibly omitted. If not a colon
14038 * # nor semi colon, the class name
14039 * # must be even closer to a valid
14042 * \]? # The right bracket, possibly
14046 * In the above, \h must be ASCII-only.
14048 * These are heuristics, and can be tweaked as field experience dictates.
14049 * There will be cases when someone didn't intend to specify a posix class
14050 * that this warns as being so. The goal is to minimize these, while
14051 * maximizing the catching of things intended to be a posix class that
14052 * aren't parsed as such.
14056 const char * const e = RExC_end;
14057 unsigned complement = 0; /* If to complement the class */
14058 bool found_problem = FALSE; /* Assume OK until proven otherwise */
14059 bool has_opening_bracket = FALSE;
14060 bool has_opening_colon = FALSE;
14061 int class_number = OOB_NAMEDCLASS; /* Out-of-bounds until find
14063 const char * possible_end = NULL; /* used for a 2nd parse pass */
14064 const char* name_start; /* ptr to class name first char */
14066 /* If the number of single-character typos the input name is away from a
14067 * legal name is no more than this number, it is considered to have meant
14068 * the legal name */
14069 int max_distance = 2;
14071 /* to store the name. The size determines the maximum length before we
14072 * decide that no posix class was intended. Should be at least
14073 * sizeof("alphanumeric") */
14076 PERL_ARGS_ASSERT_HANDLE_POSSIBLE_POSIX;
14078 if (posix_warnings && RExC_warn_text)
14079 av_clear(RExC_warn_text);
14082 return NOT_MEANT_TO_BE_A_POSIX_CLASS;
14085 if (*(p - 1) != '[') {
14086 ADD_POSIX_WARNING(p, "it doesn't start with a '['");
14087 found_problem = TRUE;
14090 has_opening_bracket = TRUE;
14093 /* They could be confused and think you can put spaces between the
14096 found_problem = TRUE;
14100 } while (p < e && isBLANK(*p));
14102 ADD_POSIX_WARNING(p, NO_BLANKS_POSIX_WARNING);
14105 /* For [. .] and [= =]. These are quite different internally from [: :],
14106 * so they are handled separately. */
14107 if (POSIXCC_NOTYET(*p) && p < e - 3) /* 1 for the close, and 1 for the ']'
14108 and 1 for at least one char in it
14111 const char open_char = *p;
14112 const char * temp_ptr = p + 1;
14114 /* These two constructs are not handled by perl, and if we find a
14115 * syntactically valid one, we croak. khw, who wrote this code, finds
14116 * this explanation of them very unclear:
14117 * http://pubs.opengroup.org/onlinepubs/009696899/basedefs/xbd_chap09.html
14118 * And searching the rest of the internet wasn't very helpful either.
14119 * It looks like just about any byte can be in these constructs,
14120 * depending on the locale. But unless the pattern is being compiled
14121 * under /l, which is very rare, Perl runs under the C or POSIX locale.
14122 * In that case, it looks like [= =] isn't allowed at all, and that
14123 * [. .] could be any single code point, but for longer strings the
14124 * constituent characters would have to be the ASCII alphabetics plus
14125 * the minus-hyphen. Any sensible locale definition would limit itself
14126 * to these. And any portable one definitely should. Trying to parse
14127 * the general case is a nightmare (see [perl #127604]). So, this code
14128 * looks only for interiors of these constructs that match:
14130 * Using \w relaxes the apparent rules a little, without adding much
14131 * danger of mistaking something else for one of these constructs.
14133 * [. .] in some implementations described on the internet is usable to
14134 * escape a character that otherwise is special in bracketed character
14135 * classes. For example [.].] means a literal right bracket instead of
14136 * the ending of the class
14138 * [= =] can legitimately contain a [. .] construct, but we don't
14139 * handle this case, as that [. .] construct will later get parsed
14140 * itself and croak then. And [= =] is checked for even when not under
14141 * /l, as Perl has long done so.
14143 * The code below relies on there being a trailing NUL, so it doesn't
14144 * have to keep checking if the parse ptr < e.
14146 if (temp_ptr[1] == open_char) {
14149 else while ( temp_ptr < e
14150 && (isWORDCHAR(*temp_ptr) || *temp_ptr == '-'))
14155 if (*temp_ptr == open_char) {
14157 if (*temp_ptr == ']') {
14159 if (! found_problem && ! check_only) {
14160 RExC_parse = (char *) temp_ptr;
14161 vFAIL3("POSIX syntax [%c %c] is reserved for future "
14162 "extensions", open_char, open_char);
14165 /* Here, the syntax wasn't completely valid, or else the call
14166 * is to check-only */
14167 if (updated_parse_ptr) {
14168 *updated_parse_ptr = (char *) temp_ptr;
14171 return OOB_NAMEDCLASS;
14175 /* If we find something that started out to look like one of these
14176 * constructs, but isn't, we continue below so that it can be checked
14177 * for being a class name with a typo of '.' or '=' instead of a colon.
14181 /* Here, we think there is a possibility that a [: :] class was meant, and
14182 * we have the first real character. It could be they think the '^' comes
14185 found_problem = TRUE;
14186 ADD_POSIX_WARNING(p + 1, "the '^' must come after the colon");
14191 found_problem = TRUE;
14195 } while (p < e && isBLANK(*p));
14197 ADD_POSIX_WARNING(p, NO_BLANKS_POSIX_WARNING);
14201 /* But the first character should be a colon, which they could have easily
14202 * mistyped on a qwerty keyboard as a semi-colon (and which may be hard to
14203 * distinguish from a colon, so treat that as a colon). */
14206 has_opening_colon = TRUE;
14208 else if (*p == ';') {
14209 found_problem = TRUE;
14211 ADD_POSIX_WARNING(p, SEMI_COLON_POSIX_WARNING);
14212 has_opening_colon = TRUE;
14215 found_problem = TRUE;
14216 ADD_POSIX_WARNING(p, "there must be a starting ':'");
14218 /* Consider an initial punctuation (not one of the recognized ones) to
14219 * be a left terminator */
14220 if (*p != '^' && *p != ']' && isPUNCT(*p)) {
14225 /* They may think that you can put spaces between the components */
14227 found_problem = TRUE;
14231 } while (p < e && isBLANK(*p));
14233 ADD_POSIX_WARNING(p, NO_BLANKS_POSIX_WARNING);
14238 /* We consider something like [^:^alnum:]] to not have been intended to
14239 * be a posix class, but XXX maybe we should */
14241 return NOT_MEANT_TO_BE_A_POSIX_CLASS;
14248 /* Again, they may think that you can put spaces between the components */
14250 found_problem = TRUE;
14254 } while (p < e && isBLANK(*p));
14256 ADD_POSIX_WARNING(p, NO_BLANKS_POSIX_WARNING);
14261 /* XXX This ']' may be a typo, and something else was meant. But
14262 * treating it as such creates enough complications, that that
14263 * possibility isn't currently considered here. So we assume that the
14264 * ']' is what is intended, and if we've already found an initial '[',
14265 * this leaves this construct looking like [:] or [:^], which almost
14266 * certainly weren't intended to be posix classes */
14267 if (has_opening_bracket) {
14268 return NOT_MEANT_TO_BE_A_POSIX_CLASS;
14271 /* But this function can be called when we parse the colon for
14272 * something like qr/[alpha:]]/, so we back up to look for the
14277 found_problem = TRUE;
14278 ADD_POSIX_WARNING(p, SEMI_COLON_POSIX_WARNING);
14280 else if (*p != ':') {
14282 /* XXX We are currently very restrictive here, so this code doesn't
14283 * consider the possibility that, say, /[alpha.]]/ was intended to
14284 * be a posix class. */
14285 return NOT_MEANT_TO_BE_A_POSIX_CLASS;
14288 /* Here we have something like 'foo:]'. There was no initial colon,
14289 * and we back up over 'foo. XXX Unlike the going forward case, we
14290 * don't handle typos of non-word chars in the middle */
14291 has_opening_colon = FALSE;
14294 while (p > RExC_start && isWORDCHAR(*p)) {
14299 /* Here, we have positioned ourselves to where we think the first
14300 * character in the potential class is */
14303 /* Now the interior really starts. There are certain key characters that
14304 * can end the interior, or these could just be typos. To catch both
14305 * cases, we may have to do two passes. In the first pass, we keep on
14306 * going unless we come to a sequence that matches
14307 * qr/ [[:punct:]] [[:blank:]]* \] /xa
14308 * This means it takes a sequence to end the pass, so two typos in a row if
14309 * that wasn't what was intended. If the class is perfectly formed, just
14310 * this one pass is needed. We also stop if there are too many characters
14311 * being accumulated, but this number is deliberately set higher than any
14312 * real class. It is set high enough so that someone who thinks that
14313 * 'alphanumeric' is a correct name would get warned that it wasn't.
14314 * While doing the pass, we keep track of where the key characters were in
14315 * it. If we don't find an end to the class, and one of the key characters
14316 * was found, we redo the pass, but stop when we get to that character.
14317 * Thus the key character was considered a typo in the first pass, but a
14318 * terminator in the second. If two key characters are found, we stop at
14319 * the second one in the first pass. Again this can miss two typos, but
14320 * catches a single one
14322 * In the first pass, 'possible_end' starts as NULL, and then gets set to
14323 * point to the first key character. For the second pass, it starts as -1.
14329 bool has_blank = FALSE;
14330 bool has_upper = FALSE;
14331 bool has_terminating_colon = FALSE;
14332 bool has_terminating_bracket = FALSE;
14333 bool has_semi_colon = FALSE;
14334 unsigned int name_len = 0;
14335 int punct_count = 0;
14339 /* Squeeze out blanks when looking up the class name below */
14340 if (isBLANK(*p) ) {
14342 found_problem = TRUE;
14347 /* The name will end with a punctuation */
14349 const char * peek = p + 1;
14351 /* Treat any non-']' punctuation followed by a ']' (possibly
14352 * with intervening blanks) as trying to terminate the class.
14353 * ']]' is very likely to mean a class was intended (but
14354 * missing the colon), but the warning message that gets
14355 * generated shows the error position better if we exit the
14356 * loop at the bottom (eventually), so skip it here. */
14358 if (peek < e && isBLANK(*peek)) {
14360 found_problem = TRUE;
14363 } while (peek < e && isBLANK(*peek));
14366 if (peek < e && *peek == ']') {
14367 has_terminating_bracket = TRUE;
14369 has_terminating_colon = TRUE;
14371 else if (*p == ';') {
14372 has_semi_colon = TRUE;
14373 has_terminating_colon = TRUE;
14376 found_problem = TRUE;
14383 /* Here we have punctuation we thought didn't end the class.
14384 * Keep track of the position of the key characters that are
14385 * more likely to have been class-enders */
14386 if (*p == ']' || *p == '[' || *p == ':' || *p == ';') {
14388 /* Allow just one such possible class-ender not actually
14389 * ending the class. */
14390 if (possible_end) {
14396 /* If we have too many punctuation characters, no use in
14398 if (++punct_count > max_distance) {
14402 /* Treat the punctuation as a typo. */
14403 input_text[name_len++] = *p;
14406 else if (isUPPER(*p)) { /* Use lowercase for lookup */
14407 input_text[name_len++] = toLOWER(*p);
14409 found_problem = TRUE;
14411 } else if (! UTF || UTF8_IS_INVARIANT(*p)) {
14412 input_text[name_len++] = *p;
14416 input_text[name_len++] = utf8_to_uvchr_buf((U8 *) p, e, NULL);
14420 /* The declaration of 'input_text' is how long we allow a potential
14421 * class name to be, before saying they didn't mean a class name at
14423 if (name_len >= C_ARRAY_LENGTH(input_text)) {
14428 /* We get to here when the possible class name hasn't been properly
14429 * terminated before:
14430 * 1) we ran off the end of the pattern; or
14431 * 2) found two characters, each of which might have been intended to
14432 * be the name's terminator
14433 * 3) found so many punctuation characters in the purported name,
14434 * that the edit distance to a valid one is exceeded
14435 * 4) we decided it was more characters than anyone could have
14436 * intended to be one. */
14438 found_problem = TRUE;
14440 /* In the final two cases, we know that looking up what we've
14441 * accumulated won't lead to a match, even a fuzzy one. */
14442 if ( name_len >= C_ARRAY_LENGTH(input_text)
14443 || punct_count > max_distance)
14445 /* If there was an intermediate key character that could have been
14446 * an intended end, redo the parse, but stop there */
14447 if (possible_end && possible_end != (char *) -1) {
14448 possible_end = (char *) -1; /* Special signal value to say
14449 we've done a first pass */
14454 /* Otherwise, it can't have meant to have been a class */
14455 return NOT_MEANT_TO_BE_A_POSIX_CLASS;
14458 /* If we ran off the end, and the final character was a punctuation
14459 * one, back up one, to look at that final one just below. Later, we
14460 * will restore the parse pointer if appropriate */
14461 if (name_len && p == e && isPUNCT(*(p-1))) {
14466 if (p < e && isPUNCT(*p)) {
14468 has_terminating_bracket = TRUE;
14470 /* If this is a 2nd ']', and the first one is just below this
14471 * one, consider that to be the real terminator. This gives a
14472 * uniform and better positioning for the warning message */
14474 && possible_end != (char *) -1
14475 && *possible_end == ']'
14476 && name_len && input_text[name_len - 1] == ']')
14481 /* And this is actually equivalent to having done the 2nd
14482 * pass now, so set it to not try again */
14483 possible_end = (char *) -1;
14488 has_terminating_colon = TRUE;
14490 else if (*p == ';') {
14491 has_semi_colon = TRUE;
14492 has_terminating_colon = TRUE;
14500 /* Here, we have a class name to look up. We can short circuit the
14501 * stuff below for short names that can't possibly be meant to be a
14502 * class name. (We can do this on the first pass, as any second pass
14503 * will yield an even shorter name) */
14504 if (name_len < 3) {
14505 return NOT_MEANT_TO_BE_A_POSIX_CLASS;
14508 /* Find which class it is. Initially switch on the length of the name.
14510 switch (name_len) {
14512 if (memEQ(name_start, "word", 4)) {
14513 /* this is not POSIX, this is the Perl \w */
14514 class_number = ANYOF_WORDCHAR;
14518 /* Names all of length 5: alnum alpha ascii blank cntrl digit
14519 * graph lower print punct space upper
14520 * Offset 4 gives the best switch position. */
14521 switch (name_start[4]) {
14523 if (memEQ(name_start, "alph", 4)) /* alpha */
14524 class_number = ANYOF_ALPHA;
14527 if (memEQ(name_start, "spac", 4)) /* space */
14528 class_number = ANYOF_SPACE;
14531 if (memEQ(name_start, "grap", 4)) /* graph */
14532 class_number = ANYOF_GRAPH;
14535 if (memEQ(name_start, "asci", 4)) /* ascii */
14536 class_number = ANYOF_ASCII;
14539 if (memEQ(name_start, "blan", 4)) /* blank */
14540 class_number = ANYOF_BLANK;
14543 if (memEQ(name_start, "cntr", 4)) /* cntrl */
14544 class_number = ANYOF_CNTRL;
14547 if (memEQ(name_start, "alnu", 4)) /* alnum */
14548 class_number = ANYOF_ALPHANUMERIC;
14551 if (memEQ(name_start, "lowe", 4)) /* lower */
14552 class_number = (FOLD) ? ANYOF_CASED : ANYOF_LOWER;
14553 else if (memEQ(name_start, "uppe", 4)) /* upper */
14554 class_number = (FOLD) ? ANYOF_CASED : ANYOF_UPPER;
14557 if (memEQ(name_start, "digi", 4)) /* digit */
14558 class_number = ANYOF_DIGIT;
14559 else if (memEQ(name_start, "prin", 4)) /* print */
14560 class_number = ANYOF_PRINT;
14561 else if (memEQ(name_start, "punc", 4)) /* punct */
14562 class_number = ANYOF_PUNCT;
14567 if (memEQ(name_start, "xdigit", 6))
14568 class_number = ANYOF_XDIGIT;
14572 /* If the name exactly matches a posix class name the class number will
14573 * here be set to it, and the input almost certainly was meant to be a
14574 * posix class, so we can skip further checking. If instead the syntax
14575 * is exactly correct, but the name isn't one of the legal ones, we
14576 * will return that as an error below. But if neither of these apply,
14577 * it could be that no posix class was intended at all, or that one
14578 * was, but there was a typo. We tease these apart by doing fuzzy
14579 * matching on the name */
14580 if (class_number == OOB_NAMEDCLASS && found_problem) {
14581 const UV posix_names[][6] = {
14582 { 'a', 'l', 'n', 'u', 'm' },
14583 { 'a', 'l', 'p', 'h', 'a' },
14584 { 'a', 's', 'c', 'i', 'i' },
14585 { 'b', 'l', 'a', 'n', 'k' },
14586 { 'c', 'n', 't', 'r', 'l' },
14587 { 'd', 'i', 'g', 'i', 't' },
14588 { 'g', 'r', 'a', 'p', 'h' },
14589 { 'l', 'o', 'w', 'e', 'r' },
14590 { 'p', 'r', 'i', 'n', 't' },
14591 { 'p', 'u', 'n', 'c', 't' },
14592 { 's', 'p', 'a', 'c', 'e' },
14593 { 'u', 'p', 'p', 'e', 'r' },
14594 { 'w', 'o', 'r', 'd' },
14595 { 'x', 'd', 'i', 'g', 'i', 't' }
14597 /* The names of the above all have added NULs to make them the same
14598 * size, so we need to also have the real lengths */
14599 const UV posix_name_lengths[] = {
14600 sizeof("alnum") - 1,
14601 sizeof("alpha") - 1,
14602 sizeof("ascii") - 1,
14603 sizeof("blank") - 1,
14604 sizeof("cntrl") - 1,
14605 sizeof("digit") - 1,
14606 sizeof("graph") - 1,
14607 sizeof("lower") - 1,
14608 sizeof("print") - 1,
14609 sizeof("punct") - 1,
14610 sizeof("space") - 1,
14611 sizeof("upper") - 1,
14612 sizeof("word") - 1,
14613 sizeof("xdigit")- 1
14616 int temp_max = max_distance; /* Use a temporary, so if we
14617 reparse, we haven't changed the
14620 /* Use a smaller max edit distance if we are missing one of the
14622 if ( has_opening_bracket + has_opening_colon < 2
14623 || has_terminating_bracket + has_terminating_colon < 2)
14628 /* See if the input name is close to a legal one */
14629 for (i = 0; i < C_ARRAY_LENGTH(posix_names); i++) {
14631 /* Short circuit call if the lengths are too far apart to be
14633 if (abs( (int) (name_len - posix_name_lengths[i]))
14639 if (edit_distance(input_text,
14642 posix_name_lengths[i],
14646 { /* If it is close, it probably was intended to be a class */
14647 goto probably_meant_to_be;
14651 /* Here the input name is not close enough to a valid class name
14652 * for us to consider it to be intended to be a posix class. If
14653 * we haven't already done so, and the parse found a character that
14654 * could have been terminators for the name, but which we absorbed
14655 * as typos during the first pass, repeat the parse, signalling it
14656 * to stop at that character */
14657 if (possible_end && possible_end != (char *) -1) {
14658 possible_end = (char *) -1;
14663 /* Here neither pass found a close-enough class name */
14664 return NOT_MEANT_TO_BE_A_POSIX_CLASS;
14667 probably_meant_to_be:
14669 /* Here we think that a posix specification was intended. Update any
14671 if (updated_parse_ptr) {
14672 *updated_parse_ptr = (char *) p;
14675 /* If a posix class name was intended but incorrectly specified, we
14676 * output or return the warnings */
14677 if (found_problem) {
14679 /* We set flags for these issues in the parse loop above instead of
14680 * adding them to the list of warnings, because we can parse it
14681 * twice, and we only want one warning instance */
14683 ADD_POSIX_WARNING(p, "the name must be all lowercase letters");
14686 ADD_POSIX_WARNING(p, NO_BLANKS_POSIX_WARNING);
14688 if (has_semi_colon) {
14689 ADD_POSIX_WARNING(p, SEMI_COLON_POSIX_WARNING);
14691 else if (! has_terminating_colon) {
14692 ADD_POSIX_WARNING(p, "there is no terminating ':'");
14694 if (! has_terminating_bracket) {
14695 ADD_POSIX_WARNING(p, "there is no terminating ']'");
14698 if (posix_warnings && RExC_warn_text && av_top_index(RExC_warn_text) > -1) {
14699 *posix_warnings = RExC_warn_text;
14702 else if (class_number != OOB_NAMEDCLASS) {
14703 /* If it is a known class, return the class. The class number
14704 * #defines are structured so each complement is +1 to the normal
14706 return class_number + complement;
14708 else if (! check_only) {
14710 /* Here, it is an unrecognized class. This is an error (unless the
14711 * call is to check only, which we've already handled above) */
14712 const char * const complement_string = (complement)
14715 RExC_parse = (char *) p;
14716 vFAIL3utf8f("POSIX class [:%s%" UTF8f ":] unknown",
14718 UTF8fARG(UTF, RExC_parse - name_start - 2, name_start));
14722 return OOB_NAMEDCLASS;
14724 #undef ADD_POSIX_WARNING
14726 STATIC unsigned int
14727 S_regex_set_precedence(const U8 my_operator) {
14729 /* Returns the precedence in the (?[...]) construct of the input operator,
14730 * specified by its character representation. The precedence follows
14731 * general Perl rules, but it extends this so that ')' and ']' have (low)
14732 * precedence even though they aren't really operators */
14734 switch (my_operator) {
14750 NOT_REACHED; /* NOTREACHED */
14751 return 0; /* Silence compiler warning */
14755 S_handle_regex_sets(pTHX_ RExC_state_t *pRExC_state, SV** return_invlist,
14756 I32 *flagp, U32 depth,
14757 char * const oregcomp_parse)
14759 /* Handle the (?[...]) construct to do set operations */
14761 U8 curchar; /* Current character being parsed */
14762 UV start, end; /* End points of code point ranges */
14763 SV* final = NULL; /* The end result inversion list */
14764 SV* result_string; /* 'final' stringified */
14765 AV* stack; /* stack of operators and operands not yet
14767 AV* fence_stack = NULL; /* A stack containing the positions in
14768 'stack' of where the undealt-with left
14769 parens would be if they were actually
14771 /* The 'VOL' (expanding to 'volatile') is a workaround for an optimiser bug
14772 * in Solaris Studio 12.3. See RT #127455 */
14773 VOL IV fence = 0; /* Position of where most recent undealt-
14774 with left paren in stack is; -1 if none.
14776 STRLEN len; /* Temporary */
14777 regnode* node; /* Temporary, and final regnode returned by
14779 const bool save_fold = FOLD; /* Temporary */
14780 char *save_end, *save_parse; /* Temporaries */
14781 const bool in_locale = LOC; /* we turn off /l during processing */
14782 AV* posix_warnings = NULL;
14784 GET_RE_DEBUG_FLAGS_DECL;
14786 PERL_ARGS_ASSERT_HANDLE_REGEX_SETS;
14789 set_regex_charset(&RExC_flags, REGEX_UNICODE_CHARSET);
14792 REQUIRE_UNI_RULES(flagp, NULL); /* The use of this operator implies /u.
14793 This is required so that the compile
14794 time values are valid in all runtime
14797 /* This will return only an ANYOF regnode, or (unlikely) something smaller
14798 * (such as EXACT). Thus we can skip most everything if just sizing. We
14799 * call regclass to handle '[]' so as to not have to reinvent its parsing
14800 * rules here (throwing away the size it computes each time). And, we exit
14801 * upon an unescaped ']' that isn't one ending a regclass. To do both
14802 * these things, we need to realize that something preceded by a backslash
14803 * is escaped, so we have to keep track of backslashes */
14805 UV depth = 0; /* how many nested (?[...]) constructs */
14807 while (RExC_parse < RExC_end) {
14808 SV* current = NULL;
14810 skip_to_be_ignored_text(pRExC_state, &RExC_parse,
14811 TRUE /* Force /x */ );
14813 switch (*RExC_parse) {
14815 if (RExC_parse[1] == '[') depth++, RExC_parse++;
14820 /* Skip past this, so the next character gets skipped, after
14823 if (*RExC_parse == 'c') {
14824 /* Skip the \cX notation for control characters */
14825 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
14831 /* See if this is a [:posix:] class. */
14832 bool is_posix_class = (OOB_NAMEDCLASS
14833 < handle_possible_posix(pRExC_state,
14837 TRUE /* checking only */));
14838 /* If it is a posix class, leave the parse pointer at the
14839 * '[' to fool regclass() into thinking it is part of a
14840 * '[[:posix:]]'. */
14841 if (! is_posix_class) {
14845 /* regclass() can only return RESTART_PASS1 and NEED_UTF8
14846 * if multi-char folds are allowed. */
14847 if (!regclass(pRExC_state, flagp,depth+1,
14848 is_posix_class, /* parse the whole char
14849 class only if not a
14851 FALSE, /* don't allow multi-char folds */
14852 TRUE, /* silence non-portable warnings. */
14854 FALSE, /* Require return to be an ANYOF */
14858 FAIL2("panic: regclass returned NULL to handle_sets, "
14859 "flags=%#" UVxf, (UV) *flagp);
14861 /* function call leaves parse pointing to the ']', except
14862 * if we faked it */
14863 if (is_posix_class) {
14867 SvREFCNT_dec(current); /* In case it returned something */
14872 if (depth--) break;
14874 if (*RExC_parse == ')') {
14875 node = reganode(pRExC_state, ANYOF, 0);
14876 RExC_size += ANYOF_SKIP;
14877 nextchar(pRExC_state);
14878 Set_Node_Length(node,
14879 RExC_parse - oregcomp_parse + 1); /* MJD */
14881 set_regex_charset(&RExC_flags, REGEX_LOCALE_CHARSET);
14889 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
14893 /* We output the messages even if warnings are off, because we'll fail
14894 * the very next thing, and these give a likely diagnosis for that */
14895 if (posix_warnings && av_tindex_skip_len_mg(posix_warnings) >= 0) {
14896 output_or_return_posix_warnings(pRExC_state, posix_warnings, NULL);
14899 FAIL("Syntax error in (?[...])");
14902 /* Pass 2 only after this. */
14903 Perl_ck_warner_d(aTHX_
14904 packWARN(WARN_EXPERIMENTAL__REGEX_SETS),
14905 "The regex_sets feature is experimental" REPORT_LOCATION,
14906 REPORT_LOCATION_ARGS(RExC_parse));
14908 /* Everything in this construct is a metacharacter. Operands begin with
14909 * either a '\' (for an escape sequence), or a '[' for a bracketed
14910 * character class. Any other character should be an operator, or
14911 * parenthesis for grouping. Both types of operands are handled by calling
14912 * regclass() to parse them. It is called with a parameter to indicate to
14913 * return the computed inversion list. The parsing here is implemented via
14914 * a stack. Each entry on the stack is a single character representing one
14915 * of the operators; or else a pointer to an operand inversion list. */
14917 #define IS_OPERATOR(a) SvIOK(a)
14918 #define IS_OPERAND(a) (! IS_OPERATOR(a))
14920 /* The stack is kept in Łukasiewicz order. (That's pronounced similar
14921 * to luke-a-shave-itch (or -itz), but people who didn't want to bother
14922 * with pronouncing it called it Reverse Polish instead, but now that YOU
14923 * know how to pronounce it you can use the correct term, thus giving due
14924 * credit to the person who invented it, and impressing your geek friends.
14925 * Wikipedia says that the pronounciation of "Ł" has been changing so that
14926 * it is now more like an English initial W (as in wonk) than an L.)
14928 * This means that, for example, 'a | b & c' is stored on the stack as
14936 * where the numbers in brackets give the stack [array] element number.
14937 * In this implementation, parentheses are not stored on the stack.
14938 * Instead a '(' creates a "fence" so that the part of the stack below the
14939 * fence is invisible except to the corresponding ')' (this allows us to
14940 * replace testing for parens, by using instead subtraction of the fence
14941 * position). As new operands are processed they are pushed onto the stack
14942 * (except as noted in the next paragraph). New operators of higher
14943 * precedence than the current final one are inserted on the stack before
14944 * the lhs operand (so that when the rhs is pushed next, everything will be
14945 * in the correct positions shown above. When an operator of equal or
14946 * lower precedence is encountered in parsing, all the stacked operations
14947 * of equal or higher precedence are evaluated, leaving the result as the
14948 * top entry on the stack. This makes higher precedence operations
14949 * evaluate before lower precedence ones, and causes operations of equal
14950 * precedence to left associate.
14952 * The only unary operator '!' is immediately pushed onto the stack when
14953 * encountered. When an operand is encountered, if the top of the stack is
14954 * a '!", the complement is immediately performed, and the '!' popped. The
14955 * resulting value is treated as a new operand, and the logic in the
14956 * previous paragraph is executed. Thus in the expression
14958 * the stack looks like
14964 * as 'b' gets parsed, the latter gets evaluated to '!b', and the stack
14971 * A ')' is treated as an operator with lower precedence than all the
14972 * aforementioned ones, which causes all operations on the stack above the
14973 * corresponding '(' to be evaluated down to a single resultant operand.
14974 * Then the fence for the '(' is removed, and the operand goes through the
14975 * algorithm above, without the fence.
14977 * A separate stack is kept of the fence positions, so that the position of
14978 * the latest so-far unbalanced '(' is at the top of it.
14980 * The ']' ending the construct is treated as the lowest operator of all,
14981 * so that everything gets evaluated down to a single operand, which is the
14984 sv_2mortal((SV *)(stack = newAV()));
14985 sv_2mortal((SV *)(fence_stack = newAV()));
14987 while (RExC_parse < RExC_end) {
14988 I32 top_index; /* Index of top-most element in 'stack' */
14989 SV** top_ptr; /* Pointer to top 'stack' element */
14990 SV* current = NULL; /* To contain the current inversion list
14992 SV* only_to_avoid_leaks;
14994 skip_to_be_ignored_text(pRExC_state, &RExC_parse,
14995 TRUE /* Force /x */ );
14996 if (RExC_parse >= RExC_end) {
14997 Perl_croak(aTHX_ "panic: Read past end of '(?[ ])'");
15000 curchar = UCHARAT(RExC_parse);
15004 #ifdef ENABLE_REGEX_SETS_DEBUGGING
15005 /* Enable with -Accflags=-DENABLE_REGEX_SETS_DEBUGGING */
15006 DEBUG_U(dump_regex_sets_structures(pRExC_state,
15007 stack, fence, fence_stack));
15010 top_index = av_tindex_skip_len_mg(stack);
15013 SV** stacked_ptr; /* Ptr to something already on 'stack' */
15014 char stacked_operator; /* The topmost operator on the 'stack'. */
15015 SV* lhs; /* Operand to the left of the operator */
15016 SV* rhs; /* Operand to the right of the operator */
15017 SV* fence_ptr; /* Pointer to top element of the fence
15022 if ( RExC_parse < RExC_end - 1
15023 && (UCHARAT(RExC_parse + 1) == '?'))
15025 /* If is a '(?', could be an embedded '(?flags:(?[...])'.
15026 * This happens when we have some thing like
15028 * my $thai_or_lao = qr/(?[ \p{Thai} + \p{Lao} ])/;
15030 * qr/(?[ \p{Digit} & $thai_or_lao ])/;
15032 * Here we would be handling the interpolated
15033 * '$thai_or_lao'. We handle this by a recursive call to
15034 * ourselves which returns the inversion list the
15035 * interpolated expression evaluates to. We use the flags
15036 * from the interpolated pattern. */
15037 U32 save_flags = RExC_flags;
15038 const char * save_parse;
15040 RExC_parse += 2; /* Skip past the '(?' */
15041 save_parse = RExC_parse;
15043 /* Parse any flags for the '(?' */
15044 parse_lparen_question_flags(pRExC_state);
15046 if (RExC_parse == save_parse /* Makes sure there was at
15047 least one flag (or else
15048 this embedding wasn't
15050 || RExC_parse >= RExC_end - 4
15051 || UCHARAT(RExC_parse) != ':'
15052 || UCHARAT(++RExC_parse) != '('
15053 || UCHARAT(++RExC_parse) != '?'
15054 || UCHARAT(++RExC_parse) != '[')
15057 /* In combination with the above, this moves the
15058 * pointer to the point just after the first erroneous
15059 * character (or if there are no flags, to where they
15060 * should have been) */
15061 if (RExC_parse >= RExC_end - 4) {
15062 RExC_parse = RExC_end;
15064 else if (RExC_parse != save_parse) {
15065 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
15067 vFAIL("Expecting '(?flags:(?[...'");
15070 /* Recurse, with the meat of the embedded expression */
15072 (void) handle_regex_sets(pRExC_state, ¤t, flagp,
15073 depth+1, oregcomp_parse);
15075 /* Here, 'current' contains the embedded expression's
15076 * inversion list, and RExC_parse points to the trailing
15077 * ']'; the next character should be the ')' */
15079 assert(UCHARAT(RExC_parse) == ')');
15081 /* Then the ')' matching the original '(' handled by this
15082 * case: statement */
15084 assert(UCHARAT(RExC_parse) == ')');
15087 RExC_flags = save_flags;
15088 goto handle_operand;
15091 /* A regular '('. Look behind for illegal syntax */
15092 if (top_index - fence >= 0) {
15093 /* If the top entry on the stack is an operator, it had
15094 * better be a '!', otherwise the entry below the top
15095 * operand should be an operator */
15096 if ( ! (top_ptr = av_fetch(stack, top_index, FALSE))
15097 || (IS_OPERATOR(*top_ptr) && SvUV(*top_ptr) != '!')
15098 || ( IS_OPERAND(*top_ptr)
15099 && ( top_index - fence < 1
15100 || ! (stacked_ptr = av_fetch(stack,
15103 || ! IS_OPERATOR(*stacked_ptr))))
15106 vFAIL("Unexpected '(' with no preceding operator");
15110 /* Stack the position of this undealt-with left paren */
15111 av_push(fence_stack, newSViv(fence));
15112 fence = top_index + 1;
15116 /* regclass() can only return RESTART_PASS1 and NEED_UTF8 if
15117 * multi-char folds are allowed. */
15118 if (!regclass(pRExC_state, flagp,depth+1,
15119 TRUE, /* means parse just the next thing */
15120 FALSE, /* don't allow multi-char folds */
15121 FALSE, /* don't silence non-portable warnings. */
15123 FALSE, /* Require return to be an ANYOF */
15127 FAIL2("panic: regclass returned NULL to handle_sets, "
15128 "flags=%#" UVxf, (UV) *flagp);
15131 /* regclass() will return with parsing just the \ sequence,
15132 * leaving the parse pointer at the next thing to parse */
15134 goto handle_operand;
15136 case '[': /* Is a bracketed character class */
15138 /* See if this is a [:posix:] class. */
15139 bool is_posix_class = (OOB_NAMEDCLASS
15140 < handle_possible_posix(pRExC_state,
15144 TRUE /* checking only */));
15145 /* If it is a posix class, leave the parse pointer at the '['
15146 * to fool regclass() into thinking it is part of a
15147 * '[[:posix:]]'. */
15148 if (! is_posix_class) {
15152 /* regclass() can only return RESTART_PASS1 and NEED_UTF8 if
15153 * multi-char folds are allowed. */
15154 if (!regclass(pRExC_state, flagp,depth+1,
15155 is_posix_class, /* parse the whole char
15156 class only if not a
15158 FALSE, /* don't allow multi-char folds */
15159 TRUE, /* silence non-portable warnings. */
15161 FALSE, /* Require return to be an ANYOF */
15166 FAIL2("panic: regclass returned NULL to handle_sets, "
15167 "flags=%#" UVxf, (UV) *flagp);
15170 /* function call leaves parse pointing to the ']', except if we
15172 if (is_posix_class) {
15176 goto handle_operand;
15180 if (top_index >= 1) {
15181 goto join_operators;
15184 /* Only a single operand on the stack: are done */
15188 if (av_tindex_skip_len_mg(fence_stack) < 0) {
15190 vFAIL("Unexpected ')'");
15193 /* If nothing after the fence, is missing an operand */
15194 if (top_index - fence < 0) {
15198 /* If at least two things on the stack, treat this as an
15200 if (top_index - fence >= 1) {
15201 goto join_operators;
15204 /* Here only a single thing on the fenced stack, and there is a
15205 * fence. Get rid of it */
15206 fence_ptr = av_pop(fence_stack);
15208 fence = SvIV(fence_ptr) - 1;
15209 SvREFCNT_dec_NN(fence_ptr);
15216 /* Having gotten rid of the fence, we pop the operand at the
15217 * stack top and process it as a newly encountered operand */
15218 current = av_pop(stack);
15219 if (IS_OPERAND(current)) {
15220 goto handle_operand;
15232 /* These binary operators should have a left operand already
15234 if ( top_index - fence < 0
15235 || top_index - fence == 1
15236 || ( ! (top_ptr = av_fetch(stack, top_index, FALSE)))
15237 || ! IS_OPERAND(*top_ptr))
15239 goto unexpected_binary;
15242 /* If only the one operand is on the part of the stack visible
15243 * to us, we just place this operator in the proper position */
15244 if (top_index - fence < 2) {
15246 /* Place the operator before the operand */
15248 SV* lhs = av_pop(stack);
15249 av_push(stack, newSVuv(curchar));
15250 av_push(stack, lhs);
15254 /* But if there is something else on the stack, we need to
15255 * process it before this new operator if and only if the
15256 * stacked operation has equal or higher precedence than the
15261 /* The operator on the stack is supposed to be below both its
15263 if ( ! (stacked_ptr = av_fetch(stack, top_index - 2, FALSE))
15264 || IS_OPERAND(*stacked_ptr))
15266 /* But if not, it's legal and indicates we are completely
15267 * done if and only if we're currently processing a ']',
15268 * which should be the final thing in the expression */
15269 if (curchar == ']') {
15275 vFAIL2("Unexpected binary operator '%c' with no "
15276 "preceding operand", curchar);
15278 stacked_operator = (char) SvUV(*stacked_ptr);
15280 if (regex_set_precedence(curchar)
15281 > regex_set_precedence(stacked_operator))
15283 /* Here, the new operator has higher precedence than the
15284 * stacked one. This means we need to add the new one to
15285 * the stack to await its rhs operand (and maybe more
15286 * stuff). We put it before the lhs operand, leaving
15287 * untouched the stacked operator and everything below it
15289 lhs = av_pop(stack);
15290 assert(IS_OPERAND(lhs));
15292 av_push(stack, newSVuv(curchar));
15293 av_push(stack, lhs);
15297 /* Here, the new operator has equal or lower precedence than
15298 * what's already there. This means the operation already
15299 * there should be performed now, before the new one. */
15301 rhs = av_pop(stack);
15302 if (! IS_OPERAND(rhs)) {
15304 /* This can happen when a ! is not followed by an operand,
15305 * like in /(?[\t &!])/ */
15309 lhs = av_pop(stack);
15311 if (! IS_OPERAND(lhs)) {
15313 /* This can happen when there is an empty (), like in
15314 * /(?[[0]+()+])/ */
15318 switch (stacked_operator) {
15320 _invlist_intersection(lhs, rhs, &rhs);
15325 _invlist_union(lhs, rhs, &rhs);
15329 _invlist_subtract(lhs, rhs, &rhs);
15332 case '^': /* The union minus the intersection */
15337 _invlist_union(lhs, rhs, &u);
15338 _invlist_intersection(lhs, rhs, &i);
15339 _invlist_subtract(u, i, &rhs);
15340 SvREFCNT_dec_NN(i);
15341 SvREFCNT_dec_NN(u);
15347 /* Here, the higher precedence operation has been done, and the
15348 * result is in 'rhs'. We overwrite the stacked operator with
15349 * the result. Then we redo this code to either push the new
15350 * operator onto the stack or perform any higher precedence
15351 * stacked operation */
15352 only_to_avoid_leaks = av_pop(stack);
15353 SvREFCNT_dec(only_to_avoid_leaks);
15354 av_push(stack, rhs);
15357 case '!': /* Highest priority, right associative */
15359 /* If what's already at the top of the stack is another '!",
15360 * they just cancel each other out */
15361 if ( (top_ptr = av_fetch(stack, top_index, FALSE))
15362 && (IS_OPERATOR(*top_ptr) && SvUV(*top_ptr) == '!'))
15364 only_to_avoid_leaks = av_pop(stack);
15365 SvREFCNT_dec(only_to_avoid_leaks);
15367 else { /* Otherwise, since it's right associative, just push
15369 av_push(stack, newSVuv(curchar));
15374 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
15375 vFAIL("Unexpected character");
15379 /* Here 'current' is the operand. If something is already on the
15380 * stack, we have to check if it is a !. But first, the code above
15381 * may have altered the stack in the time since we earlier set
15384 top_index = av_tindex_skip_len_mg(stack);
15385 if (top_index - fence >= 0) {
15386 /* If the top entry on the stack is an operator, it had better
15387 * be a '!', otherwise the entry below the top operand should
15388 * be an operator */
15389 top_ptr = av_fetch(stack, top_index, FALSE);
15391 if (IS_OPERATOR(*top_ptr)) {
15393 /* The only permissible operator at the top of the stack is
15394 * '!', which is applied immediately to this operand. */
15395 curchar = (char) SvUV(*top_ptr);
15396 if (curchar != '!') {
15397 SvREFCNT_dec(current);
15398 vFAIL2("Unexpected binary operator '%c' with no "
15399 "preceding operand", curchar);
15402 _invlist_invert(current);
15404 only_to_avoid_leaks = av_pop(stack);
15405 SvREFCNT_dec(only_to_avoid_leaks);
15407 /* And we redo with the inverted operand. This allows
15408 * handling multiple ! in a row */
15409 goto handle_operand;
15411 /* Single operand is ok only for the non-binary ')'
15413 else if ((top_index - fence == 0 && curchar != ')')
15414 || (top_index - fence > 0
15415 && (! (stacked_ptr = av_fetch(stack,
15418 || IS_OPERAND(*stacked_ptr))))
15420 SvREFCNT_dec(current);
15421 vFAIL("Operand with no preceding operator");
15425 /* Here there was nothing on the stack or the top element was
15426 * another operand. Just add this new one */
15427 av_push(stack, current);
15429 } /* End of switch on next parse token */
15431 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
15432 } /* End of loop parsing through the construct */
15435 if (av_tindex_skip_len_mg(fence_stack) >= 0) {
15436 vFAIL("Unmatched (");
15439 if (av_tindex_skip_len_mg(stack) < 0 /* Was empty */
15440 || ((final = av_pop(stack)) == NULL)
15441 || ! IS_OPERAND(final)
15442 || SvTYPE(final) != SVt_INVLIST
15443 || av_tindex_skip_len_mg(stack) >= 0) /* More left on stack */
15446 SvREFCNT_dec(final);
15447 vFAIL("Incomplete expression within '(?[ ])'");
15450 /* Here, 'final' is the resultant inversion list from evaluating the
15451 * expression. Return it if so requested */
15452 if (return_invlist) {
15453 *return_invlist = final;
15457 /* Otherwise generate a resultant node, based on 'final'. regclass() is
15458 * expecting a string of ranges and individual code points */
15459 invlist_iterinit(final);
15460 result_string = newSVpvs("");
15461 while (invlist_iternext(final, &start, &end)) {
15462 if (start == end) {
15463 Perl_sv_catpvf(aTHX_ result_string, "\\x{%" UVXf "}", start);
15466 Perl_sv_catpvf(aTHX_ result_string, "\\x{%" UVXf "}-\\x{%" UVXf "}",
15471 /* About to generate an ANYOF (or similar) node from the inversion list we
15472 * have calculated */
15473 save_parse = RExC_parse;
15474 RExC_parse = SvPV(result_string, len);
15475 save_end = RExC_end;
15476 RExC_end = RExC_parse + len;
15478 /* We turn off folding around the call, as the class we have constructed
15479 * already has all folding taken into consideration, and we don't want
15480 * regclass() to add to that */
15481 RExC_flags &= ~RXf_PMf_FOLD;
15482 /* regclass() can only return RESTART_PASS1 and NEED_UTF8 if multi-char
15483 * folds are allowed. */
15484 node = regclass(pRExC_state, flagp,depth+1,
15485 FALSE, /* means parse the whole char class */
15486 FALSE, /* don't allow multi-char folds */
15487 TRUE, /* silence non-portable warnings. The above may very
15488 well have generated non-portable code points, but
15489 they're valid on this machine */
15490 FALSE, /* similarly, no need for strict */
15491 FALSE, /* Require return to be an ANYOF */
15496 FAIL2("panic: regclass returned NULL to handle_sets, flags=%#" UVxf,
15499 /* Fix up the node type if we are in locale. (We have pretended we are
15500 * under /u for the purposes of regclass(), as this construct will only
15501 * work under UTF-8 locales. But now we change the opcode to be ANYOFL (so
15502 * as to cause any warnings about bad locales to be output in regexec.c),
15503 * and add the flag that indicates to check if not in a UTF-8 locale. The
15504 * reason we above forbid optimization into something other than an ANYOF
15505 * node is simply to minimize the number of code changes in regexec.c.
15506 * Otherwise we would have to create new EXACTish node types and deal with
15507 * them. This decision could be revisited should this construct become
15510 * (One might think we could look at the resulting ANYOF node and suppress
15511 * the flag if everything is above 255, as those would be UTF-8 only,
15512 * but this isn't true, as the components that led to that result could
15513 * have been locale-affected, and just happen to cancel each other out
15514 * under UTF-8 locales.) */
15516 set_regex_charset(&RExC_flags, REGEX_LOCALE_CHARSET);
15518 assert(OP(node) == ANYOF);
15522 |= ANYOFL_SHARED_UTF8_LOCALE_fold_HAS_MATCHES_nonfold_REQD;
15526 RExC_flags |= RXf_PMf_FOLD;
15529 RExC_parse = save_parse + 1;
15530 RExC_end = save_end;
15531 SvREFCNT_dec_NN(final);
15532 SvREFCNT_dec_NN(result_string);
15534 nextchar(pRExC_state);
15535 Set_Node_Length(node, RExC_parse - oregcomp_parse + 1); /* MJD */
15539 #ifdef ENABLE_REGEX_SETS_DEBUGGING
15542 S_dump_regex_sets_structures(pTHX_ RExC_state_t *pRExC_state,
15543 AV * stack, const IV fence, AV * fence_stack)
15544 { /* Dumps the stacks in handle_regex_sets() */
15546 const SSize_t stack_top = av_tindex_skip_len_mg(stack);
15547 const SSize_t fence_stack_top = av_tindex_skip_len_mg(fence_stack);
15550 PERL_ARGS_ASSERT_DUMP_REGEX_SETS_STRUCTURES;
15552 PerlIO_printf(Perl_debug_log, "\nParse position is:%s\n", RExC_parse);
15554 if (stack_top < 0) {
15555 PerlIO_printf(Perl_debug_log, "Nothing on stack\n");
15558 PerlIO_printf(Perl_debug_log, "Stack: (fence=%d)\n", (int) fence);
15559 for (i = stack_top; i >= 0; i--) {
15560 SV ** element_ptr = av_fetch(stack, i, FALSE);
15561 if (! element_ptr) {
15564 if (IS_OPERATOR(*element_ptr)) {
15565 PerlIO_printf(Perl_debug_log, "[%d]: %c\n",
15566 (int) i, (int) SvIV(*element_ptr));
15569 PerlIO_printf(Perl_debug_log, "[%d] ", (int) i);
15570 sv_dump(*element_ptr);
15575 if (fence_stack_top < 0) {
15576 PerlIO_printf(Perl_debug_log, "Nothing on fence_stack\n");
15579 PerlIO_printf(Perl_debug_log, "Fence_stack: \n");
15580 for (i = fence_stack_top; i >= 0; i--) {
15581 SV ** element_ptr = av_fetch(fence_stack, i, FALSE);
15582 if (! element_ptr) {
15585 PerlIO_printf(Perl_debug_log, "[%d]: %d\n",
15586 (int) i, (int) SvIV(*element_ptr));
15597 S_add_above_Latin1_folds(pTHX_ RExC_state_t *pRExC_state, const U8 cp, SV** invlist)
15599 /* This hard-codes the Latin1/above-Latin1 folding rules, so that an
15600 * innocent-looking character class, like /[ks]/i won't have to go out to
15601 * disk to find the possible matches.
15603 * This should be called only for a Latin1-range code points, cp, which is
15604 * known to be involved in a simple fold with other code points above
15605 * Latin1. It would give false results if /aa has been specified.
15606 * Multi-char folds are outside the scope of this, and must be handled
15609 * XXX It would be better to generate these via regen, in case a new
15610 * version of the Unicode standard adds new mappings, though that is not
15611 * really likely, and may be caught by the default: case of the switch
15614 PERL_ARGS_ASSERT_ADD_ABOVE_LATIN1_FOLDS;
15616 assert(HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(cp));
15622 add_cp_to_invlist(*invlist, KELVIN_SIGN);
15626 *invlist = add_cp_to_invlist(*invlist, LATIN_SMALL_LETTER_LONG_S);
15629 *invlist = add_cp_to_invlist(*invlist, GREEK_CAPITAL_LETTER_MU);
15630 *invlist = add_cp_to_invlist(*invlist, GREEK_SMALL_LETTER_MU);
15632 case LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE:
15633 case LATIN_SMALL_LETTER_A_WITH_RING_ABOVE:
15634 *invlist = add_cp_to_invlist(*invlist, ANGSTROM_SIGN);
15636 case LATIN_SMALL_LETTER_Y_WITH_DIAERESIS:
15637 *invlist = add_cp_to_invlist(*invlist,
15638 LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS);
15641 #ifdef LATIN_CAPITAL_LETTER_SHARP_S /* not defined in early Unicode releases */
15643 case LATIN_SMALL_LETTER_SHARP_S:
15644 *invlist = add_cp_to_invlist(*invlist, LATIN_CAPITAL_LETTER_SHARP_S);
15649 #if UNICODE_MAJOR_VERSION < 3 \
15650 || (UNICODE_MAJOR_VERSION == 3 && UNICODE_DOT_VERSION == 0)
15652 /* In 3.0 and earlier, U+0130 folded simply to 'i'; and in 3.0.1 so did
15657 add_cp_to_invlist(*invlist, LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE);
15658 # if UNICODE_DOT_DOT_VERSION == 1
15659 *invlist = add_cp_to_invlist(*invlist, LATIN_SMALL_LETTER_DOTLESS_I);
15665 /* Use deprecated warning to increase the chances of this being
15668 ckWARN2reg_d(RExC_parse, "Perl folding rules are not up-to-date for 0x%02X; please use the perlbug utility to report;", cp);
15675 S_output_or_return_posix_warnings(pTHX_ RExC_state_t *pRExC_state, AV* posix_warnings, AV** return_posix_warnings)
15677 /* If the final parameter is NULL, output the elements of the array given
15678 * by '*posix_warnings' as REGEXP warnings. Otherwise, the elements are
15679 * pushed onto it, (creating if necessary) */
15682 const bool first_is_fatal = ! return_posix_warnings
15683 && ckDEAD(packWARN(WARN_REGEXP));
15685 PERL_ARGS_ASSERT_OUTPUT_OR_RETURN_POSIX_WARNINGS;
15687 while ((msg = av_shift(posix_warnings)) != &PL_sv_undef) {
15688 if (return_posix_warnings) {
15689 if (! *return_posix_warnings) { /* mortalize to not leak if
15690 warnings are fatal */
15691 *return_posix_warnings = (AV *) sv_2mortal((SV *) newAV());
15693 av_push(*return_posix_warnings, msg);
15696 if (first_is_fatal) { /* Avoid leaking this */
15697 av_undef(posix_warnings); /* This isn't necessary if the
15698 array is mortal, but is a
15700 (void) sv_2mortal(msg);
15702 SAVEFREESV(RExC_rx_sv);
15705 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s", SvPVX(msg));
15706 SvREFCNT_dec_NN(msg);
15712 S_add_multi_match(pTHX_ AV* multi_char_matches, SV* multi_string, const STRLEN cp_count)
15714 /* This adds the string scalar <multi_string> to the array
15715 * <multi_char_matches>. <multi_string> is known to have exactly
15716 * <cp_count> code points in it. This is used when constructing a
15717 * bracketed character class and we find something that needs to match more
15718 * than a single character.
15720 * <multi_char_matches> is actually an array of arrays. Each top-level
15721 * element is an array that contains all the strings known so far that are
15722 * the same length. And that length (in number of code points) is the same
15723 * as the index of the top-level array. Hence, the [2] element is an
15724 * array, each element thereof is a string containing TWO code points;
15725 * while element [3] is for strings of THREE characters, and so on. Since
15726 * this is for multi-char strings there can never be a [0] nor [1] element.
15728 * When we rewrite the character class below, we will do so such that the
15729 * longest strings are written first, so that it prefers the longest
15730 * matching strings first. This is done even if it turns out that any
15731 * quantifier is non-greedy, out of this programmer's (khw) laziness. Tom
15732 * Christiansen has agreed that this is ok. This makes the test for the
15733 * ligature 'ffi' come before the test for 'ff', for example */
15736 AV** this_array_ptr;
15738 PERL_ARGS_ASSERT_ADD_MULTI_MATCH;
15740 if (! multi_char_matches) {
15741 multi_char_matches = newAV();
15744 if (av_exists(multi_char_matches, cp_count)) {
15745 this_array_ptr = (AV**) av_fetch(multi_char_matches, cp_count, FALSE);
15746 this_array = *this_array_ptr;
15749 this_array = newAV();
15750 av_store(multi_char_matches, cp_count,
15753 av_push(this_array, multi_string);
15755 return multi_char_matches;
15758 /* The names of properties whose definitions are not known at compile time are
15759 * stored in this SV, after a constant heading. So if the length has been
15760 * changed since initialization, then there is a run-time definition. */
15761 #define HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION \
15762 (SvCUR(listsv) != initial_listsv_len)
15764 /* There is a restricted set of white space characters that are legal when
15765 * ignoring white space in a bracketed character class. This generates the
15766 * code to skip them.
15768 * There is a line below that uses the same white space criteria but is outside
15769 * this macro. Both here and there must use the same definition */
15770 #define SKIP_BRACKETED_WHITE_SPACE(do_skip, p) \
15773 while (isBLANK_A(UCHARAT(p))) \
15781 S_regclass(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth,
15782 const bool stop_at_1, /* Just parse the next thing, don't
15783 look for a full character class */
15784 bool allow_multi_folds,
15785 const bool silence_non_portable, /* Don't output warnings
15789 bool optimizable, /* ? Allow a non-ANYOF return
15791 SV** ret_invlist, /* Return an inversion list, not a node */
15792 AV** return_posix_warnings
15795 /* parse a bracketed class specification. Most of these will produce an
15796 * ANYOF node; but something like [a] will produce an EXACT node; [aA], an
15797 * EXACTFish node; [[:ascii:]], a POSIXA node; etc. It is more complex
15798 * under /i with multi-character folds: it will be rewritten following the
15799 * paradigm of this example, where the <multi-fold>s are characters which
15800 * fold to multiple character sequences:
15801 * /[abc\x{multi-fold1}def\x{multi-fold2}ghi]/i
15802 * gets effectively rewritten as:
15803 * /(?:\x{multi-fold1}|\x{multi-fold2}|[abcdefghi]/i
15804 * reg() gets called (recursively) on the rewritten version, and this
15805 * function will return what it constructs. (Actually the <multi-fold>s
15806 * aren't physically removed from the [abcdefghi], it's just that they are
15807 * ignored in the recursion by means of a flag:
15808 * <RExC_in_multi_char_class>.)
15810 * ANYOF nodes contain a bit map for the first NUM_ANYOF_CODE_POINTS
15811 * characters, with the corresponding bit set if that character is in the
15812 * list. For characters above this, a range list or swash is used. There
15813 * are extra bits for \w, etc. in locale ANYOFs, as what these match is not
15814 * determinable at compile time
15816 * Returns NULL, setting *flagp to RESTART_PASS1 if the sizing scan needs
15817 * to be restarted, or'd with NEED_UTF8 if the pattern needs to be upgraded
15818 * to UTF-8. This can only happen if ret_invlist is non-NULL.
15821 UV prevvalue = OOB_UNICODE, save_prevvalue = OOB_UNICODE;
15823 UV value = OOB_UNICODE, save_value = OOB_UNICODE;
15826 int namedclass = OOB_NAMEDCLASS;
15827 char *rangebegin = NULL;
15828 bool need_class = 0;
15830 STRLEN initial_listsv_len = 0; /* Kind of a kludge to see if it is more
15831 than just initialized. */
15832 SV* properties = NULL; /* Code points that match \p{} \P{} */
15833 SV* posixes = NULL; /* Code points that match classes like [:word:],
15834 extended beyond the Latin1 range. These have to
15835 be kept separate from other code points for much
15836 of this function because their handling is
15837 different under /i, and for most classes under
15839 SV* nposixes = NULL; /* Similarly for [:^word:]. These are kept
15840 separate for a while from the non-complemented
15841 versions because of complications with /d
15843 SV* simple_posixes = NULL; /* But under some conditions, the classes can be
15844 treated more simply than the general case,
15845 leading to less compilation and execution
15847 UV element_count = 0; /* Number of distinct elements in the class.
15848 Optimizations may be possible if this is tiny */
15849 AV * multi_char_matches = NULL; /* Code points that fold to more than one
15850 character; used under /i */
15852 char * stop_ptr = RExC_end; /* where to stop parsing */
15854 /* ignore unescaped whitespace? */
15855 const bool skip_white = cBOOL( ret_invlist
15856 || (RExC_flags & RXf_PMf_EXTENDED_MORE));
15858 /* Unicode properties are stored in a swash; this holds the current one
15859 * being parsed. If this swash is the only above-latin1 component of the
15860 * character class, an optimization is to pass it directly on to the
15861 * execution engine. Otherwise, it is set to NULL to indicate that there
15862 * are other things in the class that have to be dealt with at execution
15864 SV* swash = NULL; /* Code points that match \p{} \P{} */
15866 /* Set if a component of this character class is user-defined; just passed
15867 * on to the engine */
15868 bool has_user_defined_property = FALSE;
15870 /* inversion list of code points this node matches only when the target
15871 * string is in UTF-8. These are all non-ASCII, < 256. (Because is under
15873 SV* has_upper_latin1_only_utf8_matches = NULL;
15875 /* Inversion list of code points this node matches regardless of things
15876 * like locale, folding, utf8ness of the target string */
15877 SV* cp_list = NULL;
15879 /* Like cp_list, but code points on this list need to be checked for things
15880 * that fold to/from them under /i */
15881 SV* cp_foldable_list = NULL;
15883 /* Like cp_list, but code points on this list are valid only when the
15884 * runtime locale is UTF-8 */
15885 SV* only_utf8_locale_list = NULL;
15887 /* In a range, if one of the endpoints is non-character-set portable,
15888 * meaning that it hard-codes a code point that may mean a different
15889 * charactger in ASCII vs. EBCDIC, as opposed to, say, a literal 'A' or a
15890 * mnemonic '\t' which each mean the same character no matter which
15891 * character set the platform is on. */
15892 unsigned int non_portable_endpoint = 0;
15894 /* Is the range unicode? which means on a platform that isn't 1-1 native
15895 * to Unicode (i.e. non-ASCII), each code point in it should be considered
15896 * to be a Unicode value. */
15897 bool unicode_range = FALSE;
15898 bool invert = FALSE; /* Is this class to be complemented */
15900 bool warn_super = ALWAYS_WARN_SUPER;
15902 regnode * const orig_emit = RExC_emit; /* Save the original RExC_emit in
15903 case we need to change the emitted regop to an EXACT. */
15904 const char * orig_parse = RExC_parse;
15905 const SSize_t orig_size = RExC_size;
15906 bool posixl_matches_all = FALSE; /* Does /l class have both e.g. \W,\w ? */
15908 /* This variable is used to mark where the end in the input is of something
15909 * that looks like a POSIX construct but isn't. During the parse, when
15910 * something looks like it could be such a construct is encountered, it is
15911 * checked for being one, but not if we've already checked this area of the
15912 * input. Only after this position is reached do we check again */
15913 char *not_posix_region_end = RExC_parse - 1;
15915 AV* posix_warnings = NULL;
15916 const bool do_posix_warnings = return_posix_warnings
15917 || (PASS2 && ckWARN(WARN_REGEXP));
15919 GET_RE_DEBUG_FLAGS_DECL;
15921 PERL_ARGS_ASSERT_REGCLASS;
15923 PERL_UNUSED_ARG(depth);
15926 DEBUG_PARSE("clas");
15928 #if UNICODE_MAJOR_VERSION < 3 /* no multifolds in early Unicode */ \
15929 || (UNICODE_MAJOR_VERSION == 3 && UNICODE_DOT_VERSION == 0 \
15930 && UNICODE_DOT_DOT_VERSION == 0)
15931 allow_multi_folds = FALSE;
15934 /* Assume we are going to generate an ANYOF node. */
15935 ret = reganode(pRExC_state,
15942 RExC_size += ANYOF_SKIP;
15943 listsv = &PL_sv_undef; /* For code scanners: listsv always non-NULL. */
15946 ANYOF_FLAGS(ret) = 0;
15948 RExC_emit += ANYOF_SKIP;
15949 listsv = newSVpvs_flags("# comment\n", SVs_TEMP);
15950 initial_listsv_len = SvCUR(listsv);
15951 SvTEMP_off(listsv); /* Grr, TEMPs and mortals are conflated. */
15954 SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse);
15956 assert(RExC_parse <= RExC_end);
15958 if (UCHARAT(RExC_parse) == '^') { /* Complement the class */
15961 allow_multi_folds = FALSE;
15963 SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse);
15966 /* Check that they didn't say [:posix:] instead of [[:posix:]] */
15967 if (! ret_invlist && MAYBE_POSIXCC(UCHARAT(RExC_parse))) {
15968 int maybe_class = handle_possible_posix(pRExC_state,
15970 ¬_posix_region_end,
15972 TRUE /* checking only */);
15973 if (PASS2 && maybe_class >= OOB_NAMEDCLASS && do_posix_warnings) {
15974 SAVEFREESV(RExC_rx_sv);
15975 ckWARN4reg(not_posix_region_end,
15976 "POSIX syntax [%c %c] belongs inside character classes%s",
15977 *RExC_parse, *RExC_parse,
15978 (maybe_class == OOB_NAMEDCLASS)
15979 ? ((POSIXCC_NOTYET(*RExC_parse))
15980 ? " (but this one isn't implemented)"
15981 : " (but this one isn't fully valid)")
15984 (void)ReREFCNT_inc(RExC_rx_sv);
15988 /* If the caller wants us to just parse a single element, accomplish this
15989 * by faking the loop ending condition */
15990 if (stop_at_1 && RExC_end > RExC_parse) {
15991 stop_ptr = RExC_parse + 1;
15994 /* allow 1st char to be ']' (allowing it to be '-' is dealt with later) */
15995 if (UCHARAT(RExC_parse) == ']')
15996 goto charclassloop;
16000 if ( posix_warnings
16001 && av_tindex_skip_len_mg(posix_warnings) >= 0
16002 && RExC_parse > not_posix_region_end)
16004 /* Warnings about posix class issues are considered tentative until
16005 * we are far enough along in the parse that we can no longer
16006 * change our mind, at which point we either output them or add
16007 * them, if it has so specified, to what gets returned to the
16008 * caller. This is done each time through the loop so that a later
16009 * class won't zap them before they have been dealt with. */
16010 output_or_return_posix_warnings(pRExC_state, posix_warnings,
16011 return_posix_warnings);
16014 if (RExC_parse >= stop_ptr) {
16018 SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse);
16020 if (UCHARAT(RExC_parse) == ']') {
16026 namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
16027 save_value = value;
16028 save_prevvalue = prevvalue;
16031 rangebegin = RExC_parse;
16033 non_portable_endpoint = 0;
16035 if (UTF && ! UTF8_IS_INVARIANT(* RExC_parse)) {
16036 value = utf8n_to_uvchr((U8*)RExC_parse,
16037 RExC_end - RExC_parse,
16038 &numlen, UTF8_ALLOW_DEFAULT);
16039 RExC_parse += numlen;
16042 value = UCHARAT(RExC_parse++);
16044 if (value == '[') {
16045 char * posix_class_end;
16046 namedclass = handle_possible_posix(pRExC_state,
16049 do_posix_warnings ? &posix_warnings : NULL,
16050 FALSE /* die if error */);
16051 if (namedclass > OOB_NAMEDCLASS) {
16053 /* If there was an earlier attempt to parse this particular
16054 * posix class, and it failed, it was a false alarm, as this
16055 * successful one proves */
16056 if ( posix_warnings
16057 && av_tindex_skip_len_mg(posix_warnings) >= 0
16058 && not_posix_region_end >= RExC_parse
16059 && not_posix_region_end <= posix_class_end)
16061 av_undef(posix_warnings);
16064 RExC_parse = posix_class_end;
16066 else if (namedclass == OOB_NAMEDCLASS) {
16067 not_posix_region_end = posix_class_end;
16070 namedclass = OOB_NAMEDCLASS;
16073 else if ( RExC_parse - 1 > not_posix_region_end
16074 && MAYBE_POSIXCC(value))
16076 (void) handle_possible_posix(
16078 RExC_parse - 1, /* -1 because parse has already been
16080 ¬_posix_region_end,
16081 do_posix_warnings ? &posix_warnings : NULL,
16082 TRUE /* checking only */);
16084 else if (value == '\\') {
16085 /* Is a backslash; get the code point of the char after it */
16087 if (RExC_parse >= RExC_end) {
16088 vFAIL("Unmatched [");
16091 if (UTF && ! UTF8_IS_INVARIANT(UCHARAT(RExC_parse))) {
16092 value = utf8n_to_uvchr((U8*)RExC_parse,
16093 RExC_end - RExC_parse,
16094 &numlen, UTF8_ALLOW_DEFAULT);
16095 RExC_parse += numlen;
16098 value = UCHARAT(RExC_parse++);
16100 /* Some compilers cannot handle switching on 64-bit integer
16101 * values, therefore value cannot be an UV. Yes, this will
16102 * be a problem later if we want switch on Unicode.
16103 * A similar issue a little bit later when switching on
16104 * namedclass. --jhi */
16106 /* If the \ is escaping white space when white space is being
16107 * skipped, it means that that white space is wanted literally, and
16108 * is already in 'value'. Otherwise, need to translate the escape
16109 * into what it signifies. */
16110 if (! skip_white || ! isBLANK_A(value)) switch ((I32)value) {
16112 case 'w': namedclass = ANYOF_WORDCHAR; break;
16113 case 'W': namedclass = ANYOF_NWORDCHAR; break;
16114 case 's': namedclass = ANYOF_SPACE; break;
16115 case 'S': namedclass = ANYOF_NSPACE; break;
16116 case 'd': namedclass = ANYOF_DIGIT; break;
16117 case 'D': namedclass = ANYOF_NDIGIT; break;
16118 case 'v': namedclass = ANYOF_VERTWS; break;
16119 case 'V': namedclass = ANYOF_NVERTWS; break;
16120 case 'h': namedclass = ANYOF_HORIZWS; break;
16121 case 'H': namedclass = ANYOF_NHORIZWS; break;
16122 case 'N': /* Handle \N{NAME} in class */
16124 const char * const backslash_N_beg = RExC_parse - 2;
16127 if (! grok_bslash_N(pRExC_state,
16128 NULL, /* No regnode */
16129 &value, /* Yes single value */
16130 &cp_count, /* Multiple code pt count */
16136 if (*flagp & NEED_UTF8)
16137 FAIL("panic: grok_bslash_N set NEED_UTF8");
16138 if (*flagp & RESTART_PASS1)
16141 if (cp_count < 0) {
16142 vFAIL("\\N in a character class must be a named character: \\N{...}");
16144 else if (cp_count == 0) {
16146 ckWARNreg(RExC_parse,
16147 "Ignoring zero length \\N{} in character class");
16150 else { /* cp_count > 1 */
16151 if (! RExC_in_multi_char_class) {
16152 if (invert || range || *RExC_parse == '-') {
16155 vFAIL("\\N{} in inverted character class or as a range end-point is restricted to one character");
16158 ckWARNreg(RExC_parse, "Using just the first character returned by \\N{} in character class");
16160 break; /* <value> contains the first code
16161 point. Drop out of the switch to
16165 SV * multi_char_N = newSVpvn(backslash_N_beg,
16166 RExC_parse - backslash_N_beg);
16168 = add_multi_match(multi_char_matches,
16173 } /* End of cp_count != 1 */
16175 /* This element should not be processed further in this
16178 value = save_value;
16179 prevvalue = save_prevvalue;
16180 continue; /* Back to top of loop to get next char */
16183 /* Here, is a single code point, and <value> contains it */
16184 unicode_range = TRUE; /* \N{} are Unicode */
16192 /* We will handle any undefined properties ourselves */
16193 U8 swash_init_flags = _CORE_SWASH_INIT_RETURN_IF_UNDEF
16194 /* And we actually would prefer to get
16195 * the straight inversion list of the
16196 * swash, since we will be accessing it
16197 * anyway, to save a little time */
16198 |_CORE_SWASH_INIT_ACCEPT_INVLIST;
16200 if (RExC_parse >= RExC_end)
16201 vFAIL2("Empty \\%c", (U8)value);
16202 if (*RExC_parse == '{') {
16203 const U8 c = (U8)value;
16204 e = strchr(RExC_parse, '}');
16207 vFAIL2("Missing right brace on \\%c{}", c);
16211 while (isSPACE(*RExC_parse)) {
16215 if (UCHARAT(RExC_parse) == '^') {
16217 /* toggle. (The rhs xor gets the single bit that
16218 * differs between P and p; the other xor inverts just
16220 value ^= 'P' ^ 'p';
16223 while (isSPACE(*RExC_parse)) {
16228 if (e == RExC_parse)
16229 vFAIL2("Empty \\%c{}", c);
16231 n = e - RExC_parse;
16232 while (isSPACE(*(RExC_parse + n - 1)))
16234 } /* The \p isn't immediately followed by a '{' */
16235 else if (! isALPHA(*RExC_parse)) {
16236 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
16237 vFAIL2("Character following \\%c must be '{' or a "
16238 "single-character Unicode property name",
16248 char* base_name; /* name after any packages are stripped */
16249 char* lookup_name = NULL;
16250 const char * const colon_colon = "::";
16252 /* Try to get the definition of the property into
16253 * <invlist>. If /i is in effect, the effective property
16254 * will have its name be <__NAME_i>. The design is
16255 * discussed in commit
16256 * 2f833f5208e26b208886e51e09e2c072b5eabb46 */
16257 name = savepv(Perl_form(aTHX_ "%.*s", (int)n, RExC_parse));
16260 lookup_name = savepv(Perl_form(aTHX_ "__%s_i", name));
16262 /* The function call just below that uses this can fail
16263 * to return, leaking memory if we don't do this */
16264 SAVEFREEPV(lookup_name);
16267 /* Look up the property name, and get its swash and
16268 * inversion list, if the property is found */
16269 SvREFCNT_dec(swash); /* Free any left-overs */
16270 swash = _core_swash_init("utf8",
16277 NULL, /* No inversion list */
16280 if (! swash || ! (invlist = _get_swash_invlist(swash))) {
16281 HV* curpkg = (IN_PERL_COMPILETIME)
16283 : CopSTASH(PL_curcop);
16287 if (swash) { /* Got a swash but no inversion list.
16288 Something is likely wrong that will
16289 be sorted-out later */
16290 SvREFCNT_dec_NN(swash);
16294 /* Here didn't find it. It could be a an error (like a
16295 * typo) in specifying a Unicode property, or it could
16296 * be a user-defined property that will be available at
16297 * run-time. The names of these must begin with 'In'
16298 * or 'Is' (after any packages are stripped off). So
16299 * if not one of those, or if we accept only
16300 * compile-time properties, is an error; otherwise add
16301 * it to the list for run-time look up. */
16302 if ((base_name = rninstr(name, name + n,
16303 colon_colon, colon_colon + 2)))
16304 { /* Has ::. We know this must be a user-defined
16307 final_n -= base_name - name;
16316 || base_name[0] != 'I'
16317 || (base_name[1] != 's' && base_name[1] != 'n')
16320 const char * const msg
16322 ? "Illegal user-defined property name"
16323 : "Can't find Unicode property definition";
16324 RExC_parse = e + 1;
16326 /* diag_listed_as: Can't find Unicode property definition "%s" */
16327 vFAIL3utf8f("%s \"%" UTF8f "\"",
16328 msg, UTF8fARG(UTF, n, name));
16331 /* If the property name doesn't already have a package
16332 * name, add the current one to it so that it can be
16333 * referred to outside it. [perl #121777] */
16334 if (! has_pkg && curpkg) {
16335 char* pkgname = HvNAME(curpkg);
16336 if (strNE(pkgname, "main")) {
16337 char* full_name = Perl_form(aTHX_
16341 n = strlen(full_name);
16342 name = savepvn(full_name, n);
16346 Perl_sv_catpvf(aTHX_ listsv, "%cutf8::%s%" UTF8f "%s\n",
16347 (value == 'p' ? '+' : '!'),
16348 (FOLD) ? "__" : "",
16349 UTF8fARG(UTF, n, name),
16350 (FOLD) ? "_i" : "");
16351 has_user_defined_property = TRUE;
16352 optimizable = FALSE; /* Will have to leave this an
16355 /* We don't know yet what this matches, so have to flag
16357 ANYOF_FLAGS(ret) |= ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP;
16361 /* Here, did get the swash and its inversion list. If
16362 * the swash is from a user-defined property, then this
16363 * whole character class should be regarded as such */
16364 if (swash_init_flags
16365 & _CORE_SWASH_INIT_USER_DEFINED_PROPERTY)
16367 has_user_defined_property = TRUE;
16370 /* We warn on matching an above-Unicode code point
16371 * if the match would return true, except don't
16372 * warn for \p{All}, which has exactly one element
16374 (_invlist_contains_cp(invlist, 0x110000)
16375 && (! (_invlist_len(invlist) == 1
16376 && *invlist_array(invlist) == 0)))
16382 /* Invert if asking for the complement */
16383 if (value == 'P') {
16384 _invlist_union_complement_2nd(properties,
16388 /* The swash can't be used as-is, because we've
16389 * inverted things; delay removing it to here after
16390 * have copied its invlist above */
16391 SvREFCNT_dec_NN(swash);
16395 _invlist_union(properties, invlist, &properties);
16399 RExC_parse = e + 1;
16400 namedclass = ANYOF_UNIPROP; /* no official name, but it's
16403 /* \p means they want Unicode semantics */
16404 REQUIRE_UNI_RULES(flagp, NULL);
16407 case 'n': value = '\n'; break;
16408 case 'r': value = '\r'; break;
16409 case 't': value = '\t'; break;
16410 case 'f': value = '\f'; break;
16411 case 'b': value = '\b'; break;
16412 case 'e': value = ESC_NATIVE; break;
16413 case 'a': value = '\a'; break;
16415 RExC_parse--; /* function expects to be pointed at the 'o' */
16417 const char* error_msg;
16418 bool valid = grok_bslash_o(&RExC_parse,
16421 PASS2, /* warnings only in
16424 silence_non_portable,
16430 non_portable_endpoint++;
16433 RExC_parse--; /* function expects to be pointed at the 'x' */
16435 const char* error_msg;
16436 bool valid = grok_bslash_x(&RExC_parse,
16439 PASS2, /* Output warnings */
16441 silence_non_portable,
16447 non_portable_endpoint++;
16450 value = grok_bslash_c(*RExC_parse++, PASS2);
16451 non_portable_endpoint++;
16453 case '0': case '1': case '2': case '3': case '4':
16454 case '5': case '6': case '7':
16456 /* Take 1-3 octal digits */
16457 I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
16458 numlen = (strict) ? 4 : 3;
16459 value = grok_oct(--RExC_parse, &numlen, &flags, NULL);
16460 RExC_parse += numlen;
16463 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
16464 vFAIL("Need exactly 3 octal digits");
16466 else if (! SIZE_ONLY /* like \08, \178 */
16468 && RExC_parse < RExC_end
16469 && isDIGIT(*RExC_parse)
16470 && ckWARN(WARN_REGEXP))
16472 SAVEFREESV(RExC_rx_sv);
16473 reg_warn_non_literal_string(
16475 form_short_octal_warning(RExC_parse, numlen));
16476 (void)ReREFCNT_inc(RExC_rx_sv);
16479 non_portable_endpoint++;
16483 /* Allow \_ to not give an error */
16484 if (!SIZE_ONLY && isWORDCHAR(value) && value != '_') {
16486 vFAIL2("Unrecognized escape \\%c in character class",
16490 SAVEFREESV(RExC_rx_sv);
16491 ckWARN2reg(RExC_parse,
16492 "Unrecognized escape \\%c in character class passed through",
16494 (void)ReREFCNT_inc(RExC_rx_sv);
16498 } /* End of switch on char following backslash */
16499 } /* end of handling backslash escape sequences */
16501 /* Here, we have the current token in 'value' */
16503 if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
16506 /* a bad range like a-\d, a-[:digit:]. The '-' is taken as a
16507 * literal, as is the character that began the false range, i.e.
16508 * the 'a' in the examples */
16511 const int w = (RExC_parse >= rangebegin)
16512 ? RExC_parse - rangebegin
16516 "False [] range \"%" UTF8f "\"",
16517 UTF8fARG(UTF, w, rangebegin));
16520 SAVEFREESV(RExC_rx_sv); /* in case of fatal warnings */
16521 ckWARN2reg(RExC_parse,
16522 "False [] range \"%" UTF8f "\"",
16523 UTF8fARG(UTF, w, rangebegin));
16524 (void)ReREFCNT_inc(RExC_rx_sv);
16525 cp_list = add_cp_to_invlist(cp_list, '-');
16526 cp_foldable_list = add_cp_to_invlist(cp_foldable_list,
16531 range = 0; /* this was not a true range */
16532 element_count += 2; /* So counts for three values */
16535 classnum = namedclass_to_classnum(namedclass);
16537 if (LOC && namedclass < ANYOF_POSIXL_MAX
16538 #ifndef HAS_ISASCII
16539 && classnum != _CC_ASCII
16542 /* What the Posix classes (like \w, [:space:]) match in locale
16543 * isn't knowable under locale until actual match time. Room
16544 * must be reserved (one time per outer bracketed class) to
16545 * store such classes. The space will contain a bit for each
16546 * named class that is to be matched against. This isn't
16547 * needed for \p{} and pseudo-classes, as they are not affected
16548 * by locale, and hence are dealt with separately */
16549 if (! need_class) {
16552 RExC_size += ANYOF_POSIXL_SKIP - ANYOF_SKIP;
16555 RExC_emit += ANYOF_POSIXL_SKIP - ANYOF_SKIP;
16557 ANYOF_FLAGS(ret) |= ANYOF_MATCHES_POSIXL;
16558 ANYOF_POSIXL_ZERO(ret);
16560 /* We can't change this into some other type of node
16561 * (unless this is the only element, in which case there
16562 * are nodes that mean exactly this) as has runtime
16564 optimizable = FALSE;
16567 /* Coverity thinks it is possible for this to be negative; both
16568 * jhi and khw think it's not, but be safer */
16569 assert(! (ANYOF_FLAGS(ret) & ANYOF_MATCHES_POSIXL)
16570 || (namedclass + ((namedclass % 2) ? -1 : 1)) >= 0);
16572 /* See if it already matches the complement of this POSIX
16574 if ((ANYOF_FLAGS(ret) & ANYOF_MATCHES_POSIXL)
16575 && ANYOF_POSIXL_TEST(ret, namedclass + ((namedclass % 2)
16579 posixl_matches_all = TRUE;
16580 break; /* No need to continue. Since it matches both
16581 e.g., \w and \W, it matches everything, and the
16582 bracketed class can be optimized into qr/./s */
16585 /* Add this class to those that should be checked at runtime */
16586 ANYOF_POSIXL_SET(ret, namedclass);
16588 /* The above-Latin1 characters are not subject to locale rules.
16589 * Just add them, in the second pass, to the
16590 * unconditionally-matched list */
16592 SV* scratch_list = NULL;
16594 /* Get the list of the above-Latin1 code points this
16596 _invlist_intersection_maybe_complement_2nd(PL_AboveLatin1,
16597 PL_XPosix_ptrs[classnum],
16599 /* Odd numbers are complements, like
16600 * NDIGIT, NASCII, ... */
16601 namedclass % 2 != 0,
16603 /* Checking if 'cp_list' is NULL first saves an extra
16604 * clone. Its reference count will be decremented at the
16605 * next union, etc, or if this is the only instance, at the
16606 * end of the routine */
16608 cp_list = scratch_list;
16611 _invlist_union(cp_list, scratch_list, &cp_list);
16612 SvREFCNT_dec_NN(scratch_list);
16614 continue; /* Go get next character */
16617 else if (! SIZE_ONLY) {
16619 /* Here, not in pass1 (in that pass we skip calculating the
16620 * contents of this class), and is not /l, or is a POSIX class
16621 * for which /l doesn't matter (or is a Unicode property, which
16622 * is skipped here). */
16623 if (namedclass >= ANYOF_POSIXL_MAX) { /* If a special class */
16624 if (namedclass != ANYOF_UNIPROP) { /* UNIPROP = \p and \P */
16626 /* Here, should be \h, \H, \v, or \V. None of /d, /i
16627 * nor /l make a difference in what these match,
16628 * therefore we just add what they match to cp_list. */
16629 if (classnum != _CC_VERTSPACE) {
16630 assert( namedclass == ANYOF_HORIZWS
16631 || namedclass == ANYOF_NHORIZWS);
16633 /* It turns out that \h is just a synonym for
16635 classnum = _CC_BLANK;
16638 _invlist_union_maybe_complement_2nd(
16640 PL_XPosix_ptrs[classnum],
16641 namedclass % 2 != 0, /* Complement if odd
16642 (NHORIZWS, NVERTWS)
16647 else if ( UNI_SEMANTICS
16648 || classnum == _CC_ASCII
16649 || (DEPENDS_SEMANTICS && ( classnum == _CC_DIGIT
16650 || classnum == _CC_XDIGIT)))
16652 /* We usually have to worry about /d and /a affecting what
16653 * POSIX classes match, with special code needed for /d
16654 * because we won't know until runtime what all matches.
16655 * But there is no extra work needed under /u, and
16656 * [:ascii:] is unaffected by /a and /d; and :digit: and
16657 * :xdigit: don't have runtime differences under /d. So we
16658 * can special case these, and avoid some extra work below,
16659 * and at runtime. */
16660 _invlist_union_maybe_complement_2nd(
16662 PL_XPosix_ptrs[classnum],
16663 namedclass % 2 != 0,
16666 else { /* Garden variety class. If is NUPPER, NALPHA, ...
16667 complement and use nposixes */
16668 SV** posixes_ptr = namedclass % 2 == 0
16671 _invlist_union_maybe_complement_2nd(
16673 PL_XPosix_ptrs[classnum],
16674 namedclass % 2 != 0,
16678 } /* end of namedclass \blah */
16680 SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse);
16682 /* If 'range' is set, 'value' is the ending of a range--check its
16683 * validity. (If value isn't a single code point in the case of a
16684 * range, we should have figured that out above in the code that
16685 * catches false ranges). Later, we will handle each individual code
16686 * point in the range. If 'range' isn't set, this could be the
16687 * beginning of a range, so check for that by looking ahead to see if
16688 * the next real character to be processed is the range indicator--the
16693 /* For unicode ranges, we have to test that the Unicode as opposed
16694 * to the native values are not decreasing. (Above 255, there is
16695 * no difference between native and Unicode) */
16696 if (unicode_range && prevvalue < 255 && value < 255) {
16697 if (NATIVE_TO_LATIN1(prevvalue) > NATIVE_TO_LATIN1(value)) {
16698 goto backwards_range;
16703 if (prevvalue > value) /* b-a */ {
16708 w = RExC_parse - rangebegin;
16710 "Invalid [] range \"%" UTF8f "\"",
16711 UTF8fARG(UTF, w, rangebegin));
16712 NOT_REACHED; /* NOTREACHED */
16716 prevvalue = value; /* save the beginning of the potential range */
16717 if (! stop_at_1 /* Can't be a range if parsing just one thing */
16718 && *RExC_parse == '-')
16720 char* next_char_ptr = RExC_parse + 1;
16722 /* Get the next real char after the '-' */
16723 SKIP_BRACKETED_WHITE_SPACE(skip_white, next_char_ptr);
16725 /* If the '-' is at the end of the class (just before the ']',
16726 * it is a literal minus; otherwise it is a range */
16727 if (next_char_ptr < RExC_end && *next_char_ptr != ']') {
16728 RExC_parse = next_char_ptr;
16730 /* a bad range like \w-, [:word:]- ? */
16731 if (namedclass > OOB_NAMEDCLASS) {
16732 if (strict || (PASS2 && ckWARN(WARN_REGEXP))) {
16733 const int w = RExC_parse >= rangebegin
16734 ? RExC_parse - rangebegin
16737 vFAIL4("False [] range \"%*.*s\"",
16742 "False [] range \"%*.*s\"",
16747 cp_list = add_cp_to_invlist(cp_list, '-');
16751 range = 1; /* yeah, it's a range! */
16752 continue; /* but do it the next time */
16757 if (namedclass > OOB_NAMEDCLASS) {
16761 /* Here, we have a single value this time through the loop, and
16762 * <prevvalue> is the beginning of the range, if any; or <value> if
16765 /* non-Latin1 code point implies unicode semantics. Must be set in
16766 * pass1 so is there for the whole of pass 2 */
16768 REQUIRE_UNI_RULES(flagp, NULL);
16771 /* Ready to process either the single value, or the completed range.
16772 * For single-valued non-inverted ranges, we consider the possibility
16773 * of multi-char folds. (We made a conscious decision to not do this
16774 * for the other cases because it can often lead to non-intuitive
16775 * results. For example, you have the peculiar case that:
16776 * "s s" =~ /^[^\xDF]+$/i => Y
16777 * "ss" =~ /^[^\xDF]+$/i => N
16779 * See [perl #89750] */
16780 if (FOLD && allow_multi_folds && value == prevvalue) {
16781 if (value == LATIN_SMALL_LETTER_SHARP_S
16782 || (value > 255 && _invlist_contains_cp(PL_HasMultiCharFold,
16785 /* Here <value> is indeed a multi-char fold. Get what it is */
16787 U8 foldbuf[UTF8_MAXBYTES_CASE];
16790 UV folded = _to_uni_fold_flags(
16794 FOLD_FLAGS_FULL | (ASCII_FOLD_RESTRICTED
16795 ? FOLD_FLAGS_NOMIX_ASCII
16799 /* Here, <folded> should be the first character of the
16800 * multi-char fold of <value>, with <foldbuf> containing the
16801 * whole thing. But, if this fold is not allowed (because of
16802 * the flags), <fold> will be the same as <value>, and should
16803 * be processed like any other character, so skip the special
16805 if (folded != value) {
16807 /* Skip if we are recursed, currently parsing the class
16808 * again. Otherwise add this character to the list of
16809 * multi-char folds. */
16810 if (! RExC_in_multi_char_class) {
16811 STRLEN cp_count = utf8_length(foldbuf,
16812 foldbuf + foldlen);
16813 SV* multi_fold = sv_2mortal(newSVpvs(""));
16815 Perl_sv_catpvf(aTHX_ multi_fold, "\\x{%" UVXf "}", value);
16818 = add_multi_match(multi_char_matches,
16824 /* This element should not be processed further in this
16827 value = save_value;
16828 prevvalue = save_prevvalue;
16834 if (strict && PASS2 && ckWARN(WARN_REGEXP)) {
16837 /* If the range starts above 255, everything is portable and
16838 * likely to be so for any forseeable character set, so don't
16840 if (unicode_range && non_portable_endpoint && prevvalue < 256) {
16841 vWARN(RExC_parse, "Both or neither range ends should be Unicode");
16843 else if (prevvalue != value) {
16845 /* Under strict, ranges that stop and/or end in an ASCII
16846 * printable should have each end point be a portable value
16847 * for it (preferably like 'A', but we don't warn if it is
16848 * a (portable) Unicode name or code point), and the range
16849 * must be be all digits or all letters of the same case.
16850 * Otherwise, the range is non-portable and unclear as to
16851 * what it contains */
16852 if ( (isPRINT_A(prevvalue) || isPRINT_A(value))
16853 && ( non_portable_endpoint
16854 || ! ( (isDIGIT_A(prevvalue) && isDIGIT_A(value))
16855 || (isLOWER_A(prevvalue) && isLOWER_A(value))
16856 || (isUPPER_A(prevvalue) && isUPPER_A(value))
16858 vWARN(RExC_parse, "Ranges of ASCII printables should"
16859 " be some subset of \"0-9\","
16860 " \"A-Z\", or \"a-z\"");
16862 else if (prevvalue >= 0x660) { /* ARABIC_INDIC_DIGIT_ZERO */
16863 SSize_t index_start;
16864 SSize_t index_final;
16866 /* But the nature of Unicode and languages mean we
16867 * can't do the same checks for above-ASCII ranges,
16868 * except in the case of digit ones. These should
16869 * contain only digits from the same group of 10. The
16870 * ASCII case is handled just above. 0x660 is the
16871 * first digit character beyond ASCII. Hence here, the
16872 * range could be a range of digits. First some
16873 * unlikely special cases. Grandfather in that a range
16874 * ending in 19DA (NEW TAI LUE THAM DIGIT ONE) is bad
16875 * if its starting value is one of the 10 digits prior
16876 * to it. This is because it is an alternate way of
16877 * writing 19D1, and some people may expect it to be in
16878 * that group. But it is bad, because it won't give
16879 * the expected results. In Unicode 5.2 it was
16880 * considered to be in that group (of 11, hence), but
16881 * this was fixed in the next version */
16883 if (UNLIKELY(value == 0x19DA && prevvalue >= 0x19D0)) {
16884 goto warn_bad_digit_range;
16886 else if (UNLIKELY( prevvalue >= 0x1D7CE
16887 && value <= 0x1D7FF))
16889 /* This is the only other case currently in Unicode
16890 * where the algorithm below fails. The code
16891 * points just above are the end points of a single
16892 * range containing only decimal digits. It is 5
16893 * different series of 0-9. All other ranges of
16894 * digits currently in Unicode are just a single
16895 * series. (And mktables will notify us if a later
16896 * Unicode version breaks this.)
16898 * If the range being checked is at most 9 long,
16899 * and the digit values represented are in
16900 * numerical order, they are from the same series.
16902 if ( value - prevvalue > 9
16903 || ((( value - 0x1D7CE) % 10)
16904 <= (prevvalue - 0x1D7CE) % 10))
16906 goto warn_bad_digit_range;
16911 /* For all other ranges of digits in Unicode, the
16912 * algorithm is just to check if both end points
16913 * are in the same series, which is the same range.
16915 index_start = _invlist_search(
16916 PL_XPosix_ptrs[_CC_DIGIT],
16919 /* Warn if the range starts and ends with a digit,
16920 * and they are not in the same group of 10. */
16921 if ( index_start >= 0
16922 && ELEMENT_RANGE_MATCHES_INVLIST(index_start)
16924 _invlist_search(PL_XPosix_ptrs[_CC_DIGIT],
16925 value)) != index_start
16926 && index_final >= 0
16927 && ELEMENT_RANGE_MATCHES_INVLIST(index_final))
16929 warn_bad_digit_range:
16930 vWARN(RExC_parse, "Ranges of digits should be"
16931 " from the same group of"
16938 if ((! range || prevvalue == value) && non_portable_endpoint) {
16939 if (isPRINT_A(value)) {
16942 if (isBACKSLASHED_PUNCT(value)) {
16943 literal[d++] = '\\';
16945 literal[d++] = (char) value;
16946 literal[d++] = '\0';
16949 "\"%.*s\" is more clearly written simply as \"%s\"",
16950 (int) (RExC_parse - rangebegin),
16955 else if isMNEMONIC_CNTRL(value) {
16957 "\"%.*s\" is more clearly written simply as \"%s\"",
16958 (int) (RExC_parse - rangebegin),
16960 cntrl_to_mnemonic((U8) value)
16966 /* Deal with this element of the class */
16970 cp_foldable_list = _add_range_to_invlist(cp_foldable_list,
16973 /* On non-ASCII platforms, for ranges that span all of 0..255, and
16974 * ones that don't require special handling, we can just add the
16975 * range like we do for ASCII platforms */
16976 if ((UNLIKELY(prevvalue == 0) && value >= 255)
16977 || ! (prevvalue < 256
16979 || (! non_portable_endpoint
16980 && ((isLOWER_A(prevvalue) && isLOWER_A(value))
16981 || (isUPPER_A(prevvalue)
16982 && isUPPER_A(value)))))))
16984 cp_foldable_list = _add_range_to_invlist(cp_foldable_list,
16988 /* Here, requires special handling. This can be because it is
16989 * a range whose code points are considered to be Unicode, and
16990 * so must be individually translated into native, or because
16991 * its a subrange of 'A-Z' or 'a-z' which each aren't
16992 * contiguous in EBCDIC, but we have defined them to include
16993 * only the "expected" upper or lower case ASCII alphabetics.
16994 * Subranges above 255 are the same in native and Unicode, so
16995 * can be added as a range */
16996 U8 start = NATIVE_TO_LATIN1(prevvalue);
16998 U8 end = (value < 256) ? NATIVE_TO_LATIN1(value) : 255;
16999 for (j = start; j <= end; j++) {
17000 cp_foldable_list = add_cp_to_invlist(cp_foldable_list, LATIN1_TO_NATIVE(j));
17003 cp_foldable_list = _add_range_to_invlist(cp_foldable_list,
17010 range = 0; /* this range (if it was one) is done now */
17011 } /* End of loop through all the text within the brackets */
17014 if ( posix_warnings && av_tindex_skip_len_mg(posix_warnings) >= 0) {
17015 output_or_return_posix_warnings(pRExC_state, posix_warnings,
17016 return_posix_warnings);
17019 /* If anything in the class expands to more than one character, we have to
17020 * deal with them by building up a substitute parse string, and recursively
17021 * calling reg() on it, instead of proceeding */
17022 if (multi_char_matches) {
17023 SV * substitute_parse = newSVpvn_flags("?:", 2, SVs_TEMP);
17026 char *save_end = RExC_end;
17027 char *save_parse = RExC_parse;
17028 char *save_start = RExC_start;
17029 STRLEN prefix_end = 0; /* We copy the character class after a
17030 prefix supplied here. This is the size
17031 + 1 of that prefix */
17032 bool first_time = TRUE; /* First multi-char occurrence doesn't get
17037 assert(RExC_precomp_adj == 0); /* Only one level of recursion allowed */
17039 #if 0 /* Have decided not to deal with multi-char folds in inverted classes,
17040 because too confusing */
17042 sv_catpv(substitute_parse, "(?:");
17046 /* Look at the longest folds first */
17047 for (cp_count = av_tindex_skip_len_mg(multi_char_matches);
17052 if (av_exists(multi_char_matches, cp_count)) {
17053 AV** this_array_ptr;
17056 this_array_ptr = (AV**) av_fetch(multi_char_matches,
17058 while ((this_sequence = av_pop(*this_array_ptr)) !=
17061 if (! first_time) {
17062 sv_catpv(substitute_parse, "|");
17064 first_time = FALSE;
17066 sv_catpv(substitute_parse, SvPVX(this_sequence));
17071 /* If the character class contains anything else besides these
17072 * multi-character folds, have to include it in recursive parsing */
17073 if (element_count) {
17074 sv_catpv(substitute_parse, "|[");
17075 prefix_end = SvCUR(substitute_parse);
17076 sv_catpvn(substitute_parse, orig_parse, RExC_parse - orig_parse);
17078 /* Put in a closing ']' only if not going off the end, as otherwise
17079 * we are adding something that really isn't there */
17080 if (RExC_parse < RExC_end) {
17081 sv_catpv(substitute_parse, "]");
17085 sv_catpv(substitute_parse, ")");
17088 /* This is a way to get the parse to skip forward a whole named
17089 * sequence instead of matching the 2nd character when it fails the
17091 sv_catpv(substitute_parse, "(*THEN)(*SKIP)(*FAIL)|.)");
17095 /* Set up the data structure so that any errors will be properly
17096 * reported. See the comments at the definition of
17097 * REPORT_LOCATION_ARGS for details */
17098 RExC_precomp_adj = orig_parse - RExC_precomp;
17099 RExC_start = RExC_parse = SvPV(substitute_parse, len);
17100 RExC_adjusted_start = RExC_start + prefix_end;
17101 RExC_end = RExC_parse + len;
17102 RExC_in_multi_char_class = 1;
17103 RExC_emit = (regnode *)orig_emit;
17105 ret = reg(pRExC_state, 1, ®_flags, depth+1);
17107 *flagp |= reg_flags&(HASWIDTH|SIMPLE|SPSTART|POSTPONED|RESTART_PASS1|NEED_UTF8);
17109 /* And restore so can parse the rest of the pattern */
17110 RExC_parse = save_parse;
17111 RExC_start = RExC_adjusted_start = save_start;
17112 RExC_precomp_adj = 0;
17113 RExC_end = save_end;
17114 RExC_in_multi_char_class = 0;
17115 SvREFCNT_dec_NN(multi_char_matches);
17119 /* Here, we've gone through the entire class and dealt with multi-char
17120 * folds. We are now in a position that we can do some checks to see if we
17121 * can optimize this ANYOF node into a simpler one, even in Pass 1.
17122 * Currently we only do two checks:
17123 * 1) is in the unlikely event that the user has specified both, eg. \w and
17124 * \W under /l, then the class matches everything. (This optimization
17125 * is done only to make the optimizer code run later work.)
17126 * 2) if the character class contains only a single element (including a
17127 * single range), we see if there is an equivalent node for it.
17128 * Other checks are possible */
17130 && ! ret_invlist /* Can't optimize if returning the constructed
17132 && (UNLIKELY(posixl_matches_all) || element_count == 1))
17137 if (UNLIKELY(posixl_matches_all)) {
17140 else if (namedclass > OOB_NAMEDCLASS) { /* this is a single named
17141 class, like \w or [:digit:]
17144 /* All named classes are mapped into POSIXish nodes, with its FLAG
17145 * argument giving which class it is */
17146 switch ((I32)namedclass) {
17147 case ANYOF_UNIPROP:
17150 /* These don't depend on the charset modifiers. They always
17151 * match under /u rules */
17152 case ANYOF_NHORIZWS:
17153 case ANYOF_HORIZWS:
17154 namedclass = ANYOF_BLANK + namedclass - ANYOF_HORIZWS;
17157 case ANYOF_NVERTWS:
17162 /* The actual POSIXish node for all the rest depends on the
17163 * charset modifier. The ones in the first set depend only on
17164 * ASCII or, if available on this platform, also locale */
17168 op = (LOC) ? POSIXL : POSIXA;
17174 /* The following don't have any matches in the upper Latin1
17175 * range, hence /d is equivalent to /u for them. Making it /u
17176 * saves some branches at runtime */
17180 case ANYOF_NXDIGIT:
17181 if (! DEPENDS_SEMANTICS) {
17182 goto treat_as_default;
17188 /* The following change to CASED under /i */
17194 namedclass = ANYOF_CASED + (namedclass % 2);
17198 /* The rest have more possibilities depending on the charset.
17199 * We take advantage of the enum ordering of the charset
17200 * modifiers to get the exact node type, */
17203 op = POSIXD + get_regex_charset(RExC_flags);
17204 if (op > POSIXA) { /* /aa is same as /a */
17209 /* The odd numbered ones are the complements of the
17210 * next-lower even number one */
17211 if (namedclass % 2 == 1) {
17215 arg = namedclass_to_classnum(namedclass);
17219 else if (value == prevvalue) {
17221 /* Here, the class consists of just a single code point */
17224 if (! LOC && value == '\n') {
17225 op = REG_ANY; /* Optimize [^\n] */
17226 *flagp |= HASWIDTH|SIMPLE;
17230 else if (value < 256 || UTF) {
17232 /* Optimize a single value into an EXACTish node, but not if it
17233 * would require converting the pattern to UTF-8. */
17234 op = compute_EXACTish(pRExC_state);
17236 } /* Otherwise is a range */
17237 else if (! LOC) { /* locale could vary these */
17238 if (prevvalue == '0') {
17239 if (value == '9') {
17244 else if (! FOLD || ASCII_FOLD_RESTRICTED) {
17245 /* We can optimize A-Z or a-z, but not if they could match
17246 * something like the KELVIN SIGN under /i. */
17247 if (prevvalue == 'A') {
17250 && ! non_portable_endpoint
17253 arg = (FOLD) ? _CC_ALPHA : _CC_UPPER;
17257 else if (prevvalue == 'a') {
17260 && ! non_portable_endpoint
17263 arg = (FOLD) ? _CC_ALPHA : _CC_LOWER;
17270 /* Here, we have changed <op> away from its initial value iff we found
17271 * an optimization */
17274 /* Throw away this ANYOF regnode, and emit the calculated one,
17275 * which should correspond to the beginning, not current, state of
17277 const char * cur_parse = RExC_parse;
17278 RExC_parse = (char *)orig_parse;
17282 /* To get locale nodes to not use the full ANYOF size would
17283 * require moving the code above that writes the portions
17284 * of it that aren't in other nodes to after this point.
17285 * e.g. ANYOF_POSIXL_SET */
17286 RExC_size = orig_size;
17290 RExC_emit = (regnode *)orig_emit;
17291 if (PL_regkind[op] == POSIXD) {
17292 if (op == POSIXL) {
17293 RExC_contains_locale = 1;
17296 op += NPOSIXD - POSIXD;
17301 ret = reg_node(pRExC_state, op);
17303 if (PL_regkind[op] == POSIXD || PL_regkind[op] == NPOSIXD) {
17307 *flagp |= HASWIDTH|SIMPLE;
17309 else if (PL_regkind[op] == EXACT) {
17310 alloc_maybe_populate_EXACT(pRExC_state, ret, flagp, 0, value,
17311 TRUE /* downgradable to EXACT */
17315 RExC_parse = (char *) cur_parse;
17317 SvREFCNT_dec(posixes);
17318 SvREFCNT_dec(nposixes);
17319 SvREFCNT_dec(simple_posixes);
17320 SvREFCNT_dec(cp_list);
17321 SvREFCNT_dec(cp_foldable_list);
17328 /****** !SIZE_ONLY (Pass 2) AFTER HERE *********/
17330 /* If folding, we calculate all characters that could fold to or from the
17331 * ones already on the list */
17332 if (cp_foldable_list) {
17334 UV start, end; /* End points of code point ranges */
17336 SV* fold_intersection = NULL;
17339 /* Our calculated list will be for Unicode rules. For locale
17340 * matching, we have to keep a separate list that is consulted at
17341 * runtime only when the locale indicates Unicode rules. For
17342 * non-locale, we just use the general list */
17344 use_list = &only_utf8_locale_list;
17347 use_list = &cp_list;
17350 /* Only the characters in this class that participate in folds need
17351 * be checked. Get the intersection of this class and all the
17352 * possible characters that are foldable. This can quickly narrow
17353 * down a large class */
17354 _invlist_intersection(PL_utf8_foldable, cp_foldable_list,
17355 &fold_intersection);
17357 /* The folds for all the Latin1 characters are hard-coded into this
17358 * program, but we have to go out to disk to get the others. */
17359 if (invlist_highest(cp_foldable_list) >= 256) {
17361 /* This is a hash that for a particular fold gives all
17362 * characters that are involved in it */
17363 if (! PL_utf8_foldclosures) {
17364 _load_PL_utf8_foldclosures();
17368 /* Now look at the foldable characters in this class individually */
17369 invlist_iterinit(fold_intersection);
17370 while (invlist_iternext(fold_intersection, &start, &end)) {
17373 /* Look at every character in the range */
17374 for (j = start; j <= end; j++) {
17375 U8 foldbuf[UTF8_MAXBYTES_CASE+1];
17381 if (IS_IN_SOME_FOLD_L1(j)) {
17383 /* ASCII is always matched; non-ASCII is matched
17384 * only under Unicode rules (which could happen
17385 * under /l if the locale is a UTF-8 one */
17386 if (isASCII(j) || ! DEPENDS_SEMANTICS) {
17387 *use_list = add_cp_to_invlist(*use_list,
17388 PL_fold_latin1[j]);
17391 has_upper_latin1_only_utf8_matches
17392 = add_cp_to_invlist(
17393 has_upper_latin1_only_utf8_matches,
17394 PL_fold_latin1[j]);
17398 if (HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(j)
17399 && (! isASCII(j) || ! ASCII_FOLD_RESTRICTED))
17401 add_above_Latin1_folds(pRExC_state,
17408 /* Here is an above Latin1 character. We don't have the
17409 * rules hard-coded for it. First, get its fold. This is
17410 * the simple fold, as the multi-character folds have been
17411 * handled earlier and separated out */
17412 _to_uni_fold_flags(j, foldbuf, &foldlen,
17413 (ASCII_FOLD_RESTRICTED)
17414 ? FOLD_FLAGS_NOMIX_ASCII
17417 /* Single character fold of above Latin1. Add everything in
17418 * its fold closure to the list that this node should match.
17419 * The fold closures data structure is a hash with the keys
17420 * being the UTF-8 of every character that is folded to, like
17421 * 'k', and the values each an array of all code points that
17422 * fold to its key. e.g. [ 'k', 'K', KELVIN_SIGN ].
17423 * Multi-character folds are not included */
17424 if ((listp = hv_fetch(PL_utf8_foldclosures,
17425 (char *) foldbuf, foldlen, FALSE)))
17427 AV* list = (AV*) *listp;
17429 for (k = 0; k <= av_tindex_skip_len_mg(list); k++) {
17430 SV** c_p = av_fetch(list, k, FALSE);
17436 /* /aa doesn't allow folds between ASCII and non- */
17437 if ((ASCII_FOLD_RESTRICTED
17438 && (isASCII(c) != isASCII(j))))
17443 /* Folds under /l which cross the 255/256 boundary
17444 * are added to a separate list. (These are valid
17445 * only when the locale is UTF-8.) */
17446 if (c < 256 && LOC) {
17447 *use_list = add_cp_to_invlist(*use_list, c);
17451 if (isASCII(c) || c > 255 || AT_LEAST_UNI_SEMANTICS)
17453 cp_list = add_cp_to_invlist(cp_list, c);
17456 /* Similarly folds involving non-ascii Latin1
17457 * characters under /d are added to their list */
17458 has_upper_latin1_only_utf8_matches
17459 = add_cp_to_invlist(
17460 has_upper_latin1_only_utf8_matches,
17467 SvREFCNT_dec_NN(fold_intersection);
17470 /* Now that we have finished adding all the folds, there is no reason
17471 * to keep the foldable list separate */
17472 _invlist_union(cp_list, cp_foldable_list, &cp_list);
17473 SvREFCNT_dec_NN(cp_foldable_list);
17476 /* And combine the result (if any) with any inversion lists from posix
17477 * classes. The lists are kept separate up to now because we don't want to
17478 * fold the classes (folding of those is automatically handled by the swash
17479 * fetching code) */
17480 if (simple_posixes) { /* These are the classes known to be unaffected by
17483 _invlist_union(cp_list, simple_posixes, &cp_list);
17484 SvREFCNT_dec_NN(simple_posixes);
17487 cp_list = simple_posixes;
17490 if (posixes || nposixes) {
17492 /* We have to adjust /a and /aa */
17493 if (AT_LEAST_ASCII_RESTRICTED) {
17495 /* Under /a and /aa, nothing above ASCII matches these */
17497 _invlist_intersection(posixes,
17498 PL_XPosix_ptrs[_CC_ASCII],
17502 /* Under /a and /aa, everything above ASCII matches these
17505 _invlist_union_complement_2nd(nposixes,
17506 PL_XPosix_ptrs[_CC_ASCII],
17511 if (! DEPENDS_SEMANTICS) {
17513 /* For everything but /d, we can just add the current 'posixes' and
17514 * 'nposixes' to the main list */
17517 _invlist_union(cp_list, posixes, &cp_list);
17518 SvREFCNT_dec_NN(posixes);
17526 _invlist_union(cp_list, nposixes, &cp_list);
17527 SvREFCNT_dec_NN(nposixes);
17530 cp_list = nposixes;
17535 /* Under /d, things like \w match upper Latin1 characters only if
17536 * the target string is in UTF-8. But things like \W match all the
17537 * upper Latin1 characters if the target string is not in UTF-8.
17539 * Handle the case where there something like \W separately */
17541 SV* only_non_utf8_list = invlist_clone(PL_UpperLatin1);
17543 /* A complemented posix class matches all upper Latin1
17544 * characters if not in UTF-8. And it matches just certain
17545 * ones when in UTF-8. That means those certain ones are
17546 * matched regardless, so can just be added to the
17547 * unconditional list */
17549 _invlist_union(cp_list, nposixes, &cp_list);
17550 SvREFCNT_dec_NN(nposixes);
17554 cp_list = nposixes;
17557 /* Likewise for 'posixes' */
17558 _invlist_union(posixes, cp_list, &cp_list);
17560 /* Likewise for anything else in the range that matched only
17562 if (has_upper_latin1_only_utf8_matches) {
17563 _invlist_union(cp_list,
17564 has_upper_latin1_only_utf8_matches,
17566 SvREFCNT_dec_NN(has_upper_latin1_only_utf8_matches);
17567 has_upper_latin1_only_utf8_matches = NULL;
17570 /* If we don't match all the upper Latin1 characters regardless
17571 * of UTF-8ness, we have to set a flag to match the rest when
17573 _invlist_subtract(only_non_utf8_list, cp_list,
17574 &only_non_utf8_list);
17575 if (_invlist_len(only_non_utf8_list) != 0) {
17576 ANYOF_FLAGS(ret) |= ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER;
17580 /* Here there were no complemented posix classes. That means
17581 * the upper Latin1 characters in 'posixes' match only when the
17582 * target string is in UTF-8. So we have to add them to the
17583 * list of those types of code points, while adding the
17584 * remainder to the unconditional list.
17586 * First calculate what they are */
17587 SV* nonascii_but_latin1_properties = NULL;
17588 _invlist_intersection(posixes, PL_UpperLatin1,
17589 &nonascii_but_latin1_properties);
17591 /* And add them to the final list of such characters. */
17592 _invlist_union(has_upper_latin1_only_utf8_matches,
17593 nonascii_but_latin1_properties,
17594 &has_upper_latin1_only_utf8_matches);
17596 /* Remove them from what now becomes the unconditional list */
17597 _invlist_subtract(posixes, nonascii_but_latin1_properties,
17600 /* And add those unconditional ones to the final list */
17602 _invlist_union(cp_list, posixes, &cp_list);
17603 SvREFCNT_dec_NN(posixes);
17610 SvREFCNT_dec(nonascii_but_latin1_properties);
17612 /* Get rid of any characters that we now know are matched
17613 * unconditionally from the conditional list, which may make
17614 * that list empty */
17615 _invlist_subtract(has_upper_latin1_only_utf8_matches,
17617 &has_upper_latin1_only_utf8_matches);
17618 if (_invlist_len(has_upper_latin1_only_utf8_matches) == 0) {
17619 SvREFCNT_dec_NN(has_upper_latin1_only_utf8_matches);
17620 has_upper_latin1_only_utf8_matches = NULL;
17626 /* And combine the result (if any) with any inversion list from properties.
17627 * The lists are kept separate up to now so that we can distinguish the two
17628 * in regards to matching above-Unicode. A run-time warning is generated
17629 * if a Unicode property is matched against a non-Unicode code point. But,
17630 * we allow user-defined properties to match anything, without any warning,
17631 * and we also suppress the warning if there is a portion of the character
17632 * class that isn't a Unicode property, and which matches above Unicode, \W
17633 * or [\x{110000}] for example.
17634 * (Note that in this case, unlike the Posix one above, there is no
17635 * <has_upper_latin1_only_utf8_matches>, because having a Unicode property
17636 * forces Unicode semantics */
17640 /* If it matters to the final outcome, see if a non-property
17641 * component of the class matches above Unicode. If so, the
17642 * warning gets suppressed. This is true even if just a single
17643 * such code point is specified, as, though not strictly correct if
17644 * another such code point is matched against, the fact that they
17645 * are using above-Unicode code points indicates they should know
17646 * the issues involved */
17648 warn_super = ! (invert
17649 ^ (invlist_highest(cp_list) > PERL_UNICODE_MAX));
17652 _invlist_union(properties, cp_list, &cp_list);
17653 SvREFCNT_dec_NN(properties);
17656 cp_list = properties;
17661 |= ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER;
17663 /* Because an ANYOF node is the only one that warns, this node
17664 * can't be optimized into something else */
17665 optimizable = FALSE;
17669 /* Here, we have calculated what code points should be in the character
17672 * Now we can see about various optimizations. Fold calculation (which we
17673 * did above) needs to take place before inversion. Otherwise /[^k]/i
17674 * would invert to include K, which under /i would match k, which it
17675 * shouldn't. Therefore we can't invert folded locale now, as it won't be
17676 * folded until runtime */
17678 /* If we didn't do folding, it's because some information isn't available
17679 * until runtime; set the run-time fold flag for these. (We don't have to
17680 * worry about properties folding, as that is taken care of by the swash
17681 * fetching). We know to set the flag if we have a non-NULL list for UTF-8
17682 * locales, or the class matches at least one 0-255 range code point */
17685 /* Some things on the list might be unconditionally included because of
17686 * other components. Remove them, and clean up the list if it goes to
17688 if (only_utf8_locale_list && cp_list) {
17689 _invlist_subtract(only_utf8_locale_list, cp_list,
17690 &only_utf8_locale_list);
17692 if (_invlist_len(only_utf8_locale_list) == 0) {
17693 SvREFCNT_dec_NN(only_utf8_locale_list);
17694 only_utf8_locale_list = NULL;
17697 if (only_utf8_locale_list) {
17700 |ANYOFL_SHARED_UTF8_LOCALE_fold_HAS_MATCHES_nonfold_REQD;
17702 else if (cp_list) { /* Look to see if a 0-255 code point is in list */
17704 invlist_iterinit(cp_list);
17705 if (invlist_iternext(cp_list, &start, &end) && start < 256) {
17706 ANYOF_FLAGS(ret) |= ANYOFL_FOLD;
17708 invlist_iterfinish(cp_list);
17711 else if ( DEPENDS_SEMANTICS
17712 && ( has_upper_latin1_only_utf8_matches
17713 || (ANYOF_FLAGS(ret) & ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER)))
17716 optimizable = FALSE;
17720 /* Optimize inverted simple patterns (e.g. [^a-z]) when everything is known
17721 * at compile time. Besides not inverting folded locale now, we can't
17722 * invert if there are things such as \w, which aren't known until runtime
17726 && OP(ret) != ANYOFD
17727 && ! (ANYOF_FLAGS(ret) & (ANYOF_LOCALE_FLAGS))
17728 && ! HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION)
17730 _invlist_invert(cp_list);
17732 /* Any swash can't be used as-is, because we've inverted things */
17734 SvREFCNT_dec_NN(swash);
17738 /* Clear the invert flag since have just done it here */
17745 *ret_invlist = cp_list;
17746 SvREFCNT_dec(swash);
17748 /* Discard the generated node */
17750 RExC_size = orig_size;
17753 RExC_emit = orig_emit;
17758 /* Some character classes are equivalent to other nodes. Such nodes take
17759 * up less room and generally fewer operations to execute than ANYOF nodes.
17760 * Above, we checked for and optimized into some such equivalents for
17761 * certain common classes that are easy to test. Getting to this point in
17762 * the code means that the class didn't get optimized there. Since this
17763 * code is only executed in Pass 2, it is too late to save space--it has
17764 * been allocated in Pass 1, and currently isn't given back. But turning
17765 * things into an EXACTish node can allow the optimizer to join it to any
17766 * adjacent such nodes. And if the class is equivalent to things like /./,
17767 * expensive run-time swashes can be avoided. Now that we have more
17768 * complete information, we can find things necessarily missed by the
17769 * earlier code. Another possible "optimization" that isn't done is that
17770 * something like [Ee] could be changed into an EXACTFU. khw tried this
17771 * and found that the ANYOF is faster, including for code points not in the
17772 * bitmap. This still might make sense to do, provided it got joined with
17773 * an adjacent node(s) to create a longer EXACTFU one. This could be
17774 * accomplished by creating a pseudo ANYOF_EXACTFU node type that the join
17775 * routine would know is joinable. If that didn't happen, the node type
17776 * could then be made a straight ANYOF */
17778 if (optimizable && cp_list && ! invert) {
17780 U8 op = END; /* The optimzation node-type */
17781 int posix_class = -1; /* Illegal value */
17782 const char * cur_parse= RExC_parse;
17784 invlist_iterinit(cp_list);
17785 if (! invlist_iternext(cp_list, &start, &end)) {
17787 /* Here, the list is empty. This happens, for example, when a
17788 * Unicode property that doesn't match anything is the only element
17789 * in the character class (perluniprops.pod notes such properties).
17792 *flagp |= HASWIDTH|SIMPLE;
17794 else if (start == end) { /* The range is a single code point */
17795 if (! invlist_iternext(cp_list, &start, &end)
17797 /* Don't do this optimization if it would require changing
17798 * the pattern to UTF-8 */
17799 && (start < 256 || UTF))
17801 /* Here, the list contains a single code point. Can optimize
17802 * into an EXACTish node */
17813 /* A locale node under folding with one code point can be
17814 * an EXACTFL, as its fold won't be calculated until
17820 /* Here, we are generally folding, but there is only one
17821 * code point to match. If we have to, we use an EXACT
17822 * node, but it would be better for joining with adjacent
17823 * nodes in the optimization pass if we used the same
17824 * EXACTFish node that any such are likely to be. We can
17825 * do this iff the code point doesn't participate in any
17826 * folds. For example, an EXACTF of a colon is the same as
17827 * an EXACT one, since nothing folds to or from a colon. */
17829 if (IS_IN_SOME_FOLD_L1(value)) {
17834 if (_invlist_contains_cp(PL_utf8_foldable, value)) {
17839 /* If we haven't found the node type, above, it means we
17840 * can use the prevailing one */
17842 op = compute_EXACTish(pRExC_state);
17846 } /* End of first range contains just a single code point */
17847 else if (start == 0) {
17848 if (end == UV_MAX) {
17850 *flagp |= HASWIDTH|SIMPLE;
17853 else if (end == '\n' - 1
17854 && invlist_iternext(cp_list, &start, &end)
17855 && start == '\n' + 1 && end == UV_MAX)
17858 *flagp |= HASWIDTH|SIMPLE;
17862 invlist_iterfinish(cp_list);
17865 const UV cp_list_len = _invlist_len(cp_list);
17866 const UV* cp_list_array = invlist_array(cp_list);
17868 /* Here, didn't find an optimization. See if this matches any of
17869 * the POSIX classes. These run slightly faster for above-Unicode
17870 * code points, so don't bother with POSIXA ones nor the 2 that
17871 * have no above-Unicode matches. We can avoid these checks unless
17872 * the ANYOF matches at least as high as the lowest POSIX one
17873 * (which was manually found to be \v. The actual code point may
17874 * increase in later Unicode releases, if a higher code point is
17875 * assigned to be \v, but this code will never break. It would
17876 * just mean we could execute the checks for posix optimizations
17877 * unnecessarily) */
17879 if (cp_list_array[cp_list_len-1] > 0x2029) {
17880 for (posix_class = 0;
17881 posix_class <= _HIGHEST_REGCOMP_DOT_H_SYNC;
17885 if (posix_class == _CC_ASCII || posix_class == _CC_CNTRL) {
17888 for (try_inverted = 0; try_inverted < 2; try_inverted++) {
17890 /* Check if matches normal or inverted */
17891 if (_invlistEQ(cp_list,
17892 PL_XPosix_ptrs[posix_class],
17895 op = (try_inverted)
17898 *flagp |= HASWIDTH|SIMPLE;
17908 RExC_parse = (char *)orig_parse;
17909 RExC_emit = (regnode *)orig_emit;
17911 if (regarglen[op]) {
17912 ret = reganode(pRExC_state, op, 0);
17914 ret = reg_node(pRExC_state, op);
17917 RExC_parse = (char *)cur_parse;
17919 if (PL_regkind[op] == EXACT) {
17920 alloc_maybe_populate_EXACT(pRExC_state, ret, flagp, 0, value,
17921 TRUE /* downgradable to EXACT */
17924 else if (PL_regkind[op] == POSIXD || PL_regkind[op] == NPOSIXD) {
17925 FLAGS(ret) = posix_class;
17928 SvREFCNT_dec_NN(cp_list);
17933 /* Here, <cp_list> contains all the code points we can determine at
17934 * compile time that match under all conditions. Go through it, and
17935 * for things that belong in the bitmap, put them there, and delete from
17936 * <cp_list>. While we are at it, see if everything above 255 is in the
17937 * list, and if so, set a flag to speed up execution */
17939 populate_ANYOF_from_invlist(ret, &cp_list);
17942 ANYOF_FLAGS(ret) |= ANYOF_INVERT;
17945 /* Here, the bitmap has been populated with all the Latin1 code points that
17946 * always match. Can now add to the overall list those that match only
17947 * when the target string is UTF-8 (<has_upper_latin1_only_utf8_matches>).
17949 if (has_upper_latin1_only_utf8_matches) {
17951 _invlist_union(cp_list,
17952 has_upper_latin1_only_utf8_matches,
17954 SvREFCNT_dec_NN(has_upper_latin1_only_utf8_matches);
17957 cp_list = has_upper_latin1_only_utf8_matches;
17959 ANYOF_FLAGS(ret) |= ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP;
17962 /* If there is a swash and more than one element, we can't use the swash in
17963 * the optimization below. */
17964 if (swash && element_count > 1) {
17965 SvREFCNT_dec_NN(swash);
17969 /* Note that the optimization of using 'swash' if it is the only thing in
17970 * the class doesn't have us change swash at all, so it can include things
17971 * that are also in the bitmap; otherwise we have purposely deleted that
17972 * duplicate information */
17973 set_ANYOF_arg(pRExC_state, ret, cp_list,
17974 (HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION)
17976 only_utf8_locale_list,
17977 swash, has_user_defined_property);
17979 *flagp |= HASWIDTH|SIMPLE;
17981 if (ANYOF_FLAGS(ret) & ANYOF_LOCALE_FLAGS) {
17982 RExC_contains_locale = 1;
17988 #undef HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION
17991 S_set_ANYOF_arg(pTHX_ RExC_state_t* const pRExC_state,
17992 regnode* const node,
17994 SV* const runtime_defns,
17995 SV* const only_utf8_locale_list,
17997 const bool has_user_defined_property)
17999 /* Sets the arg field of an ANYOF-type node 'node', using information about
18000 * the node passed-in. If there is nothing outside the node's bitmap, the
18001 * arg is set to ANYOF_ONLY_HAS_BITMAP. Otherwise, it sets the argument to
18002 * the count returned by add_data(), having allocated and stored an array,
18003 * av, that that count references, as follows:
18004 * av[0] stores the character class description in its textual form.
18005 * This is used later (regexec.c:Perl_regclass_swash()) to
18006 * initialize the appropriate swash, and is also useful for dumping
18007 * the regnode. This is set to &PL_sv_undef if the textual
18008 * description is not needed at run-time (as happens if the other
18009 * elements completely define the class)
18010 * av[1] if &PL_sv_undef, is a placeholder to later contain the swash
18011 * computed from av[0]. But if no further computation need be done,
18012 * the swash is stored here now (and av[0] is &PL_sv_undef).
18013 * av[2] stores the inversion list of code points that match only if the
18014 * current locale is UTF-8
18015 * av[3] stores the cp_list inversion list for use in addition or instead
18016 * of av[0]; used only if cp_list exists and av[1] is &PL_sv_undef.
18017 * (Otherwise everything needed is already in av[0] and av[1])
18018 * av[4] is set if any component of the class is from a user-defined
18019 * property; used only if av[3] exists */
18023 PERL_ARGS_ASSERT_SET_ANYOF_ARG;
18025 if (! cp_list && ! runtime_defns && ! only_utf8_locale_list) {
18026 assert(! (ANYOF_FLAGS(node)
18027 & ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP));
18028 ARG_SET(node, ANYOF_ONLY_HAS_BITMAP);
18031 AV * const av = newAV();
18034 av_store(av, 0, (runtime_defns)
18035 ? SvREFCNT_inc(runtime_defns) : &PL_sv_undef);
18038 av_store(av, 1, swash);
18039 SvREFCNT_dec_NN(cp_list);
18042 av_store(av, 1, &PL_sv_undef);
18044 av_store(av, 3, cp_list);
18045 av_store(av, 4, newSVuv(has_user_defined_property));
18049 if (only_utf8_locale_list) {
18050 av_store(av, 2, only_utf8_locale_list);
18053 av_store(av, 2, &PL_sv_undef);
18056 rv = newRV_noinc(MUTABLE_SV(av));
18057 n = add_data(pRExC_state, STR_WITH_LEN("s"));
18058 RExC_rxi->data->data[n] = (void*)rv;
18063 #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION)
18065 Perl__get_regclass_nonbitmap_data(pTHX_ const regexp *prog,
18066 const regnode* node,
18069 SV** only_utf8_locale_ptr,
18070 SV** output_invlist)
18073 /* For internal core use only.
18074 * Returns the swash for the input 'node' in the regex 'prog'.
18075 * If <doinit> is 'true', will attempt to create the swash if not already
18077 * If <listsvp> is non-null, will return the printable contents of the
18078 * swash. This can be used to get debugging information even before the
18079 * swash exists, by calling this function with 'doinit' set to false, in
18080 * which case the components that will be used to eventually create the
18081 * swash are returned (in a printable form).
18082 * If <only_utf8_locale_ptr> is not NULL, it is where this routine is to
18083 * store an inversion list of code points that should match only if the
18084 * execution-time locale is a UTF-8 one.
18085 * If <output_invlist> is not NULL, it is where this routine is to store an
18086 * inversion list of the code points that would be instead returned in
18087 * <listsvp> if this were NULL. Thus, what gets output in <listsvp>
18088 * when this parameter is used, is just the non-code point data that
18089 * will go into creating the swash. This currently should be just
18090 * user-defined properties whose definitions were not known at compile
18091 * time. Using this parameter allows for easier manipulation of the
18092 * swash's data by the caller. It is illegal to call this function with
18093 * this parameter set, but not <listsvp>
18095 * Tied intimately to how S_set_ANYOF_arg sets up the data structure. Note
18096 * that, in spite of this function's name, the swash it returns may include
18097 * the bitmap data as well */
18100 SV *si = NULL; /* Input swash initialization string */
18101 SV* invlist = NULL;
18103 RXi_GET_DECL(prog,progi);
18104 const struct reg_data * const data = prog ? progi->data : NULL;
18106 PERL_ARGS_ASSERT__GET_REGCLASS_NONBITMAP_DATA;
18107 assert(! output_invlist || listsvp);
18109 if (data && data->count) {
18110 const U32 n = ARG(node);
18112 if (data->what[n] == 's') {
18113 SV * const rv = MUTABLE_SV(data->data[n]);
18114 AV * const av = MUTABLE_AV(SvRV(rv));
18115 SV **const ary = AvARRAY(av);
18116 U8 swash_init_flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
18118 si = *ary; /* ary[0] = the string to initialize the swash with */
18120 if (av_tindex_skip_len_mg(av) >= 2) {
18121 if (only_utf8_locale_ptr
18123 && ary[2] != &PL_sv_undef)
18125 *only_utf8_locale_ptr = ary[2];
18128 assert(only_utf8_locale_ptr);
18129 *only_utf8_locale_ptr = NULL;
18132 /* Elements 3 and 4 are either both present or both absent. [3]
18133 * is any inversion list generated at compile time; [4]
18134 * indicates if that inversion list has any user-defined
18135 * properties in it. */
18136 if (av_tindex_skip_len_mg(av) >= 3) {
18138 if (SvUV(ary[4])) {
18139 swash_init_flags |= _CORE_SWASH_INIT_USER_DEFINED_PROPERTY;
18147 /* Element [1] is reserved for the set-up swash. If already there,
18148 * return it; if not, create it and store it there */
18149 if (ary[1] && SvROK(ary[1])) {
18152 else if (doinit && ((si && si != &PL_sv_undef)
18153 || (invlist && invlist != &PL_sv_undef))) {
18155 sw = _core_swash_init("utf8", /* the utf8 package */
18159 0, /* not from tr/// */
18161 &swash_init_flags);
18162 (void)av_store(av, 1, sw);
18167 /* If requested, return a printable version of what this swash matches */
18169 SV* matches_string = NULL;
18171 /* The swash should be used, if possible, to get the data, as it
18172 * contains the resolved data. But this function can be called at
18173 * compile-time, before everything gets resolved, in which case we
18174 * return the currently best available information, which is the string
18175 * that will eventually be used to do that resolving, 'si' */
18176 if ((! sw || (invlist = _get_swash_invlist(sw)) == NULL)
18177 && (si && si != &PL_sv_undef))
18179 /* Here, we only have 'si' (and possibly some passed-in data in
18180 * 'invlist', which is handled below) If the caller only wants
18181 * 'si', use that. */
18182 if (! output_invlist) {
18183 matches_string = newSVsv(si);
18186 /* But if the caller wants an inversion list of the node, we
18187 * need to parse 'si' and place as much as possible in the
18188 * desired output inversion list, making 'matches_string' only
18189 * contain the currently unresolvable things */
18190 const char *si_string = SvPVX(si);
18191 STRLEN remaining = SvCUR(si);
18195 /* Ignore everything before the first new-line */
18196 while (*si_string != '\n' && remaining > 0) {
18200 assert(remaining > 0);
18205 while (remaining > 0) {
18207 /* The data consists of just strings defining user-defined
18208 * property names, but in prior incarnations, and perhaps
18209 * somehow from pluggable regex engines, it could still
18210 * hold hex code point definitions. Each component of a
18211 * range would be separated by a tab, and each range by a
18212 * new-line. If these are found, instead add them to the
18213 * inversion list */
18214 I32 grok_flags = PERL_SCAN_SILENT_ILLDIGIT
18215 |PERL_SCAN_SILENT_NON_PORTABLE;
18216 STRLEN len = remaining;
18217 UV cp = grok_hex(si_string, &len, &grok_flags, NULL);
18219 /* If the hex decode routine found something, it should go
18220 * up to the next \n */
18221 if ( *(si_string + len) == '\n') {
18222 if (count) { /* 2nd code point on line */
18223 *output_invlist = _add_range_to_invlist(*output_invlist, prev_cp, cp);
18226 *output_invlist = add_cp_to_invlist(*output_invlist, cp);
18229 goto prepare_for_next_iteration;
18232 /* If the hex decode was instead for the lower range limit,
18233 * save it, and go parse the upper range limit */
18234 if (*(si_string + len) == '\t') {
18235 assert(count == 0);
18239 prepare_for_next_iteration:
18240 si_string += len + 1;
18241 remaining -= len + 1;
18245 /* Here, didn't find a legal hex number. Just add it from
18246 * here to the next \n */
18249 while (*(si_string + len) != '\n' && remaining > 0) {
18253 if (*(si_string + len) == '\n') {
18257 if (matches_string) {
18258 sv_catpvn(matches_string, si_string, len - 1);
18261 matches_string = newSVpvn(si_string, len - 1);
18264 sv_catpvs(matches_string, " ");
18265 } /* end of loop through the text */
18267 assert(matches_string);
18268 if (SvCUR(matches_string)) { /* Get rid of trailing blank */
18269 SvCUR_set(matches_string, SvCUR(matches_string) - 1);
18271 } /* end of has an 'si' but no swash */
18274 /* If we have a swash in place, its equivalent inversion list was above
18275 * placed into 'invlist'. If not, this variable may contain a stored
18276 * inversion list which is information beyond what is in 'si' */
18279 /* Again, if the caller doesn't want the output inversion list, put
18280 * everything in 'matches-string' */
18281 if (! output_invlist) {
18282 if ( ! matches_string) {
18283 matches_string = newSVpvs("\n");
18285 sv_catsv(matches_string, invlist_contents(invlist,
18286 TRUE /* traditional style */
18289 else if (! *output_invlist) {
18290 *output_invlist = invlist_clone(invlist);
18293 _invlist_union(*output_invlist, invlist, output_invlist);
18297 *listsvp = matches_string;
18302 #endif /* !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION) */
18304 /* reg_skipcomment()
18306 Absorbs an /x style # comment from the input stream,
18307 returning a pointer to the first character beyond the comment, or if the
18308 comment terminates the pattern without anything following it, this returns
18309 one past the final character of the pattern (in other words, RExC_end) and
18310 sets the REG_RUN_ON_COMMENT_SEEN flag.
18312 Note it's the callers responsibility to ensure that we are
18313 actually in /x mode
18317 PERL_STATIC_INLINE char*
18318 S_reg_skipcomment(RExC_state_t *pRExC_state, char* p)
18320 PERL_ARGS_ASSERT_REG_SKIPCOMMENT;
18324 while (p < RExC_end) {
18325 if (*(++p) == '\n') {
18330 /* we ran off the end of the pattern without ending the comment, so we have
18331 * to add an \n when wrapping */
18332 RExC_seen |= REG_RUN_ON_COMMENT_SEEN;
18337 S_skip_to_be_ignored_text(pTHX_ RExC_state_t *pRExC_state,
18339 const bool force_to_xmod
18342 /* If the text at the current parse position '*p' is a '(?#...)' comment,
18343 * or if we are under /x or 'force_to_xmod' is TRUE, and the text at '*p'
18344 * is /x whitespace, advance '*p' so that on exit it points to the first
18345 * byte past all such white space and comments */
18347 const bool use_xmod = force_to_xmod || (RExC_flags & RXf_PMf_EXTENDED);
18349 PERL_ARGS_ASSERT_SKIP_TO_BE_IGNORED_TEXT;
18351 assert( ! UTF || UTF8_IS_INVARIANT(**p) || UTF8_IS_START(**p));
18354 if (RExC_end - (*p) >= 3
18356 && *(*p + 1) == '?'
18357 && *(*p + 2) == '#')
18359 while (*(*p) != ')') {
18360 if ((*p) == RExC_end)
18361 FAIL("Sequence (?#... not terminated");
18369 const char * save_p = *p;
18370 while ((*p) < RExC_end) {
18372 if ((len = is_PATWS_safe((*p), RExC_end, UTF))) {
18375 else if (*(*p) == '#') {
18376 (*p) = reg_skipcomment(pRExC_state, (*p));
18382 if (*p != save_p) {
18395 Advances the parse position by one byte, unless that byte is the beginning
18396 of a '(?#...)' style comment, or is /x whitespace and /x is in effect. In
18397 those two cases, the parse position is advanced beyond all such comments and
18400 This is the UTF, (?#...), and /x friendly way of saying RExC_parse++.
18404 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
18406 PERL_ARGS_ASSERT_NEXTCHAR;
18408 if (RExC_parse < RExC_end) {
18410 || UTF8_IS_INVARIANT(*RExC_parse)
18411 || UTF8_IS_START(*RExC_parse));
18413 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
18415 skip_to_be_ignored_text(pRExC_state, &RExC_parse,
18416 FALSE /* Don't force /x */ );
18421 S_regnode_guts(pTHX_ RExC_state_t *pRExC_state, const U8 op, const STRLEN extra_size, const char* const name)
18423 /* Allocate a regnode for 'op' and returns it, with 'extra_size' extra
18424 * space. In pass1, it aligns and increments RExC_size; in pass2,
18427 regnode * const ret = RExC_emit;
18428 GET_RE_DEBUG_FLAGS_DECL;
18430 PERL_ARGS_ASSERT_REGNODE_GUTS;
18432 assert(extra_size >= regarglen[op]);
18435 SIZE_ALIGN(RExC_size);
18436 RExC_size += 1 + extra_size;
18439 if (RExC_emit >= RExC_emit_bound)
18440 Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d, %p>=%p",
18441 op, (void*)RExC_emit, (void*)RExC_emit_bound);
18443 NODE_ALIGN_FILL(ret);
18444 #ifndef RE_TRACK_PATTERN_OFFSETS
18445 PERL_UNUSED_ARG(name);
18447 if (RExC_offsets) { /* MJD */
18449 ("%s:%d: (op %s) %s %" UVuf " (len %" UVuf ") (max %" UVuf ").\n",
18452 (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0]
18453 ? "Overwriting end of array!\n" : "OK",
18454 (UV)(RExC_emit - RExC_emit_start),
18455 (UV)(RExC_parse - RExC_start),
18456 (UV)RExC_offsets[0]));
18457 Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
18464 - reg_node - emit a node
18466 STATIC regnode * /* Location. */
18467 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
18469 regnode * const ret = regnode_guts(pRExC_state, op, regarglen[op], "reg_node");
18471 PERL_ARGS_ASSERT_REG_NODE;
18473 assert(regarglen[op] == 0);
18476 regnode *ptr = ret;
18477 FILL_ADVANCE_NODE(ptr, op);
18484 - reganode - emit a node with an argument
18486 STATIC regnode * /* Location. */
18487 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
18489 regnode * const ret = regnode_guts(pRExC_state, op, regarglen[op], "reganode");
18491 PERL_ARGS_ASSERT_REGANODE;
18493 assert(regarglen[op] == 1);
18496 regnode *ptr = ret;
18497 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
18504 S_reg2Lanode(pTHX_ RExC_state_t *pRExC_state, const U8 op, const U32 arg1, const I32 arg2)
18506 /* emit a node with U32 and I32 arguments */
18508 regnode * const ret = regnode_guts(pRExC_state, op, regarglen[op], "reg2Lanode");
18510 PERL_ARGS_ASSERT_REG2LANODE;
18512 assert(regarglen[op] == 2);
18515 regnode *ptr = ret;
18516 FILL_ADVANCE_NODE_2L_ARG(ptr, op, arg1, arg2);
18523 - reginsert - insert an operator in front of already-emitted operand
18525 * Means relocating the operand.
18527 * IMPORTANT NOTE - it is the *callers* responsibility to correctly
18528 * set up NEXT_OFF() of the inserted node if needed. Something like this:
18530 * reginsert(pRExC, OPFAIL, orig_emit, depth+1);
18532 * NEXT_OFF(orig_emit) = regarglen[OPFAIL] + NODE_STEP_REGNODE;
18536 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *operand, U32 depth)
18541 const int offset = regarglen[(U8)op];
18542 const int size = NODE_STEP_REGNODE + offset;
18543 GET_RE_DEBUG_FLAGS_DECL;
18545 PERL_ARGS_ASSERT_REGINSERT;
18546 PERL_UNUSED_CONTEXT;
18547 PERL_UNUSED_ARG(depth);
18548 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
18549 DEBUG_PARSE_FMT("inst"," - %s",PL_reg_name[op]);
18554 assert(!RExC_study_started); /* I believe we should never use reginsert once we have started
18555 studying. If this is wrong then we need to adjust RExC_recurse
18556 below like we do with RExC_open_parens/RExC_close_parens. */
18560 if (RExC_open_parens) {
18562 /*DEBUG_PARSE_FMT("inst"," - %" IVdf, (IV)RExC_npar);*/
18563 /* remember that RExC_npar is rex->nparens + 1,
18564 * iow it is 1 more than the number of parens seen in
18565 * the pattern so far. */
18566 for ( paren=0 ; paren < RExC_npar ; paren++ ) {
18567 /* note, RExC_open_parens[0] is the start of the
18568 * regex, it can't move. RExC_close_parens[0] is the end
18569 * of the regex, it *can* move. */
18570 if ( paren && RExC_open_parens[paren] >= operand ) {
18571 /*DEBUG_PARSE_FMT("open"," - %d",size);*/
18572 RExC_open_parens[paren] += size;
18574 /*DEBUG_PARSE_FMT("open"," - %s","ok");*/
18576 if ( RExC_close_parens[paren] >= operand ) {
18577 /*DEBUG_PARSE_FMT("close"," - %d",size);*/
18578 RExC_close_parens[paren] += size;
18580 /*DEBUG_PARSE_FMT("close"," - %s","ok");*/
18585 RExC_end_op += size;
18587 while (src > operand) {
18588 StructCopy(--src, --dst, regnode);
18589 #ifdef RE_TRACK_PATTERN_OFFSETS
18590 if (RExC_offsets) { /* MJD 20010112 */
18592 ("%s(%d): (op %s) %s copy %" UVuf " -> %" UVuf " (max %" UVuf ").\n",
18596 (UV)(dst - RExC_emit_start) > RExC_offsets[0]
18597 ? "Overwriting end of array!\n" : "OK",
18598 (UV)(src - RExC_emit_start),
18599 (UV)(dst - RExC_emit_start),
18600 (UV)RExC_offsets[0]));
18601 Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
18602 Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
18608 place = operand; /* Op node, where operand used to be. */
18609 #ifdef RE_TRACK_PATTERN_OFFSETS
18610 if (RExC_offsets) { /* MJD */
18612 ("%s(%d): (op %s) %s %" UVuf " <- %" UVuf " (max %" UVuf ").\n",
18616 (UV)(place - RExC_emit_start) > RExC_offsets[0]
18617 ? "Overwriting end of array!\n" : "OK",
18618 (UV)(place - RExC_emit_start),
18619 (UV)(RExC_parse - RExC_start),
18620 (UV)RExC_offsets[0]));
18621 Set_Node_Offset(place, RExC_parse);
18622 Set_Node_Length(place, 1);
18625 src = NEXTOPER(place);
18626 FILL_ADVANCE_NODE(place, op);
18627 Zero(src, offset, regnode);
18631 - regtail - set the next-pointer at the end of a node chain of p to val.
18632 - SEE ALSO: regtail_study
18635 S_regtail(pTHX_ RExC_state_t * pRExC_state,
18636 const regnode * const p,
18637 const regnode * const val,
18641 GET_RE_DEBUG_FLAGS_DECL;
18643 PERL_ARGS_ASSERT_REGTAIL;
18645 PERL_UNUSED_ARG(depth);
18651 /* Find last node. */
18652 scan = (regnode *) p;
18654 regnode * const temp = regnext(scan);
18656 DEBUG_PARSE_MSG((scan==p ? "tail" : ""));
18657 regprop(RExC_rx, RExC_mysv, scan, NULL, pRExC_state);
18658 Perl_re_printf( aTHX_ "~ %s (%d) %s %s\n",
18659 SvPV_nolen_const(RExC_mysv), REG_NODE_NUM(scan),
18660 (temp == NULL ? "->" : ""),
18661 (temp == NULL ? PL_reg_name[OP(val)] : "")
18669 if (reg_off_by_arg[OP(scan)]) {
18670 ARG_SET(scan, val - scan);
18673 NEXT_OFF(scan) = val - scan;
18679 - regtail_study - set the next-pointer at the end of a node chain of p to val.
18680 - Look for optimizable sequences at the same time.
18681 - currently only looks for EXACT chains.
18683 This is experimental code. The idea is to use this routine to perform
18684 in place optimizations on branches and groups as they are constructed,
18685 with the long term intention of removing optimization from study_chunk so
18686 that it is purely analytical.
18688 Currently only used when in DEBUG mode. The macro REGTAIL_STUDY() is used
18689 to control which is which.
18692 /* TODO: All four parms should be const */
18695 S_regtail_study(pTHX_ RExC_state_t *pRExC_state, regnode *p,
18696 const regnode *val,U32 depth)
18700 #ifdef EXPERIMENTAL_INPLACESCAN
18703 GET_RE_DEBUG_FLAGS_DECL;
18705 PERL_ARGS_ASSERT_REGTAIL_STUDY;
18711 /* Find last node. */
18715 regnode * const temp = regnext(scan);
18716 #ifdef EXPERIMENTAL_INPLACESCAN
18717 if (PL_regkind[OP(scan)] == EXACT) {
18718 bool unfolded_multi_char; /* Unexamined in this routine */
18719 if (join_exact(pRExC_state, scan, &min,
18720 &unfolded_multi_char, 1, val, depth+1))
18725 switch (OP(scan)) {
18729 case EXACTFA_NO_TRIE:
18735 if( exact == PSEUDO )
18737 else if ( exact != OP(scan) )
18746 DEBUG_PARSE_MSG((scan==p ? "tsdy" : ""));
18747 regprop(RExC_rx, RExC_mysv, scan, NULL, pRExC_state);
18748 Perl_re_printf( aTHX_ "~ %s (%d) -> %s\n",
18749 SvPV_nolen_const(RExC_mysv),
18750 REG_NODE_NUM(scan),
18751 PL_reg_name[exact]);
18758 DEBUG_PARSE_MSG("");
18759 regprop(RExC_rx, RExC_mysv, val, NULL, pRExC_state);
18760 Perl_re_printf( aTHX_
18761 "~ attach to %s (%" IVdf ") offset to %" IVdf "\n",
18762 SvPV_nolen_const(RExC_mysv),
18763 (IV)REG_NODE_NUM(val),
18767 if (reg_off_by_arg[OP(scan)]) {
18768 ARG_SET(scan, val - scan);
18771 NEXT_OFF(scan) = val - scan;
18779 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
18784 S_regdump_intflags(pTHX_ const char *lead, const U32 flags)
18789 ASSUME(REG_INTFLAGS_NAME_SIZE <= sizeof(flags)*8);
18791 for (bit=0; bit<REG_INTFLAGS_NAME_SIZE; bit++) {
18792 if (flags & (1<<bit)) {
18793 if (!set++ && lead)
18794 Perl_re_printf( aTHX_ "%s",lead);
18795 Perl_re_printf( aTHX_ "%s ",PL_reg_intflags_name[bit]);
18800 Perl_re_printf( aTHX_ "\n");
18802 Perl_re_printf( aTHX_ "%s[none-set]\n",lead);
18807 S_regdump_extflags(pTHX_ const char *lead, const U32 flags)
18813 ASSUME(REG_EXTFLAGS_NAME_SIZE <= sizeof(flags)*8);
18815 for (bit=0; bit<REG_EXTFLAGS_NAME_SIZE; bit++) {
18816 if (flags & (1<<bit)) {
18817 if ((1<<bit) & RXf_PMf_CHARSET) { /* Output separately, below */
18820 if (!set++ && lead)
18821 Perl_re_printf( aTHX_ "%s",lead);
18822 Perl_re_printf( aTHX_ "%s ",PL_reg_extflags_name[bit]);
18825 if ((cs = get_regex_charset(flags)) != REGEX_DEPENDS_CHARSET) {
18826 if (!set++ && lead) {
18827 Perl_re_printf( aTHX_ "%s",lead);
18830 case REGEX_UNICODE_CHARSET:
18831 Perl_re_printf( aTHX_ "UNICODE");
18833 case REGEX_LOCALE_CHARSET:
18834 Perl_re_printf( aTHX_ "LOCALE");
18836 case REGEX_ASCII_RESTRICTED_CHARSET:
18837 Perl_re_printf( aTHX_ "ASCII-RESTRICTED");
18839 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
18840 Perl_re_printf( aTHX_ "ASCII-MORE_RESTRICTED");
18843 Perl_re_printf( aTHX_ "UNKNOWN CHARACTER SET");
18849 Perl_re_printf( aTHX_ "\n");
18851 Perl_re_printf( aTHX_ "%s[none-set]\n",lead);
18857 Perl_regdump(pTHX_ const regexp *r)
18860 SV * const sv = sv_newmortal();
18861 SV *dsv= sv_newmortal();
18862 RXi_GET_DECL(r,ri);
18863 GET_RE_DEBUG_FLAGS_DECL;
18865 PERL_ARGS_ASSERT_REGDUMP;
18867 (void)dumpuntil(r, ri->program, ri->program + 1, NULL, NULL, sv, 0, 0);
18869 /* Header fields of interest. */
18870 if (r->anchored_substr) {
18871 RE_PV_QUOTED_DECL(s, 0, dsv, SvPVX_const(r->anchored_substr),
18872 RE_SV_DUMPLEN(r->anchored_substr), 30);
18873 Perl_re_printf( aTHX_
18874 "anchored %s%s at %" IVdf " ",
18875 s, RE_SV_TAIL(r->anchored_substr),
18876 (IV)r->anchored_offset);
18877 } else if (r->anchored_utf8) {
18878 RE_PV_QUOTED_DECL(s, 1, dsv, SvPVX_const(r->anchored_utf8),
18879 RE_SV_DUMPLEN(r->anchored_utf8), 30);
18880 Perl_re_printf( aTHX_
18881 "anchored utf8 %s%s at %" IVdf " ",
18882 s, RE_SV_TAIL(r->anchored_utf8),
18883 (IV)r->anchored_offset);
18885 if (r->float_substr) {
18886 RE_PV_QUOTED_DECL(s, 0, dsv, SvPVX_const(r->float_substr),
18887 RE_SV_DUMPLEN(r->float_substr), 30);
18888 Perl_re_printf( aTHX_
18889 "floating %s%s at %" IVdf "..%" UVuf " ",
18890 s, RE_SV_TAIL(r->float_substr),
18891 (IV)r->float_min_offset, (UV)r->float_max_offset);
18892 } else if (r->float_utf8) {
18893 RE_PV_QUOTED_DECL(s, 1, dsv, SvPVX_const(r->float_utf8),
18894 RE_SV_DUMPLEN(r->float_utf8), 30);
18895 Perl_re_printf( aTHX_
18896 "floating utf8 %s%s at %" IVdf "..%" UVuf " ",
18897 s, RE_SV_TAIL(r->float_utf8),
18898 (IV)r->float_min_offset, (UV)r->float_max_offset);
18900 if (r->check_substr || r->check_utf8)
18901 Perl_re_printf( aTHX_
18903 (r->check_substr == r->float_substr
18904 && r->check_utf8 == r->float_utf8
18905 ? "(checking floating" : "(checking anchored"));
18906 if (r->intflags & PREGf_NOSCAN)
18907 Perl_re_printf( aTHX_ " noscan");
18908 if (r->extflags & RXf_CHECK_ALL)
18909 Perl_re_printf( aTHX_ " isall");
18910 if (r->check_substr || r->check_utf8)
18911 Perl_re_printf( aTHX_ ") ");
18913 if (ri->regstclass) {
18914 regprop(r, sv, ri->regstclass, NULL, NULL);
18915 Perl_re_printf( aTHX_ "stclass %s ", SvPVX_const(sv));
18917 if (r->intflags & PREGf_ANCH) {
18918 Perl_re_printf( aTHX_ "anchored");
18919 if (r->intflags & PREGf_ANCH_MBOL)
18920 Perl_re_printf( aTHX_ "(MBOL)");
18921 if (r->intflags & PREGf_ANCH_SBOL)
18922 Perl_re_printf( aTHX_ "(SBOL)");
18923 if (r->intflags & PREGf_ANCH_GPOS)
18924 Perl_re_printf( aTHX_ "(GPOS)");
18925 Perl_re_printf( aTHX_ " ");
18927 if (r->intflags & PREGf_GPOS_SEEN)
18928 Perl_re_printf( aTHX_ "GPOS:%" UVuf " ", (UV)r->gofs);
18929 if (r->intflags & PREGf_SKIP)
18930 Perl_re_printf( aTHX_ "plus ");
18931 if (r->intflags & PREGf_IMPLICIT)
18932 Perl_re_printf( aTHX_ "implicit ");
18933 Perl_re_printf( aTHX_ "minlen %" IVdf " ", (IV)r->minlen);
18934 if (r->extflags & RXf_EVAL_SEEN)
18935 Perl_re_printf( aTHX_ "with eval ");
18936 Perl_re_printf( aTHX_ "\n");
18938 regdump_extflags("r->extflags: ",r->extflags);
18939 regdump_intflags("r->intflags: ",r->intflags);
18942 PERL_ARGS_ASSERT_REGDUMP;
18943 PERL_UNUSED_CONTEXT;
18944 PERL_UNUSED_ARG(r);
18945 #endif /* DEBUGGING */
18948 /* Should be synchronized with ANYOF_ #defines in regcomp.h */
18951 # if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 \
18952 || _CC_LOWER != 3 || _CC_UPPER != 4 || _CC_PUNCT != 5 \
18953 || _CC_PRINT != 6 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 \
18954 || _CC_CASED != 9 || _CC_SPACE != 10 || _CC_BLANK != 11 \
18955 || _CC_XDIGIT != 12 || _CC_CNTRL != 13 || _CC_ASCII != 14 \
18956 || _CC_VERTSPACE != 15
18957 # error Need to adjust order of anyofs[]
18959 static const char * const anyofs[] = {
18996 - regprop - printable representation of opcode, with run time support
19000 Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o, const regmatch_info *reginfo, const RExC_state_t *pRExC_state)
19004 RXi_GET_DECL(prog,progi);
19005 GET_RE_DEBUG_FLAGS_DECL;
19007 PERL_ARGS_ASSERT_REGPROP;
19011 if (OP(o) > REGNODE_MAX) /* regnode.type is unsigned */
19012 /* It would be nice to FAIL() here, but this may be called from
19013 regexec.c, and it would be hard to supply pRExC_state. */
19014 Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d",
19015 (int)OP(o), (int)REGNODE_MAX);
19016 sv_catpv(sv, PL_reg_name[OP(o)]); /* Take off const! */
19018 k = PL_regkind[OP(o)];
19021 sv_catpvs(sv, " ");
19022 /* Using is_utf8_string() (via PERL_PV_UNI_DETECT)
19023 * is a crude hack but it may be the best for now since
19024 * we have no flag "this EXACTish node was UTF-8"
19026 pv_pretty(sv, STRING(o), STR_LEN(o), 60, PL_colors[0], PL_colors[1],
19027 PERL_PV_ESCAPE_UNI_DETECT |
19028 PERL_PV_ESCAPE_NONASCII |
19029 PERL_PV_PRETTY_ELLIPSES |
19030 PERL_PV_PRETTY_LTGT |
19031 PERL_PV_PRETTY_NOCLEAR
19033 } else if (k == TRIE) {
19034 /* print the details of the trie in dumpuntil instead, as
19035 * progi->data isn't available here */
19036 const char op = OP(o);
19037 const U32 n = ARG(o);
19038 const reg_ac_data * const ac = IS_TRIE_AC(op) ?
19039 (reg_ac_data *)progi->data->data[n] :
19041 const reg_trie_data * const trie
19042 = (reg_trie_data*)progi->data->data[!IS_TRIE_AC(op) ? n : ac->trie];
19044 Perl_sv_catpvf(aTHX_ sv, "-%s",PL_reg_name[o->flags]);
19045 DEBUG_TRIE_COMPILE_r({
19047 sv_catpvs(sv, "(JUMP)");
19048 Perl_sv_catpvf(aTHX_ sv,
19049 "<S:%" UVuf "/%" IVdf " W:%" UVuf " L:%" UVuf "/%" UVuf " C:%" UVuf "/%" UVuf ">",
19050 (UV)trie->startstate,
19051 (IV)trie->statecount-1, /* -1 because of the unused 0 element */
19052 (UV)trie->wordcount,
19055 (UV)TRIE_CHARCOUNT(trie),
19056 (UV)trie->uniquecharcount
19059 if ( IS_ANYOF_TRIE(op) || trie->bitmap ) {
19060 sv_catpvs(sv, "[");
19061 (void) put_charclass_bitmap_innards(sv,
19062 ((IS_ANYOF_TRIE(op))
19064 : TRIE_BITMAP(trie)),
19070 sv_catpvs(sv, "]");
19072 } else if (k == CURLY) {
19073 U32 lo = ARG1(o), hi = ARG2(o);
19074 if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
19075 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
19076 Perl_sv_catpvf(aTHX_ sv, "{%u,", (unsigned) lo);
19077 if (hi == REG_INFTY)
19078 sv_catpvs(sv, "INFTY");
19080 Perl_sv_catpvf(aTHX_ sv, "%u", (unsigned) hi);
19081 sv_catpvs(sv, "}");
19083 else if (k == WHILEM && o->flags) /* Ordinal/of */
19084 Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
19085 else if (k == REF || k == OPEN || k == CLOSE
19086 || k == GROUPP || OP(o)==ACCEPT)
19088 AV *name_list= NULL;
19089 U32 parno= OP(o) == ACCEPT ? (U32)ARG2L(o) : ARG(o);
19090 Perl_sv_catpvf(aTHX_ sv, "%" UVuf, (UV)parno); /* Parenth number */
19091 if ( RXp_PAREN_NAMES(prog) ) {
19092 name_list= MUTABLE_AV(progi->data->data[progi->name_list_idx]);
19093 } else if ( pRExC_state ) {
19094 name_list= RExC_paren_name_list;
19097 if ( k != REF || (OP(o) < NREF)) {
19098 SV **name= av_fetch(name_list, parno, 0 );
19100 Perl_sv_catpvf(aTHX_ sv, " '%" SVf "'", SVfARG(*name));
19103 SV *sv_dat= MUTABLE_SV(progi->data->data[ parno ]);
19104 I32 *nums=(I32*)SvPVX(sv_dat);
19105 SV **name= av_fetch(name_list, nums[0], 0 );
19108 for ( n=0; n<SvIVX(sv_dat); n++ ) {
19109 Perl_sv_catpvf(aTHX_ sv, "%s%" IVdf,
19110 (n ? "," : ""), (IV)nums[n]);
19112 Perl_sv_catpvf(aTHX_ sv, " '%" SVf "'", SVfARG(*name));
19116 if ( k == REF && reginfo) {
19117 U32 n = ARG(o); /* which paren pair */
19118 I32 ln = prog->offs[n].start;
19119 if (prog->lastparen < n || ln == -1)
19120 Perl_sv_catpvf(aTHX_ sv, ": FAIL");
19121 else if (ln == prog->offs[n].end)
19122 Perl_sv_catpvf(aTHX_ sv, ": ACCEPT - EMPTY STRING");
19124 const char *s = reginfo->strbeg + ln;
19125 Perl_sv_catpvf(aTHX_ sv, ": ");
19126 Perl_pv_pretty( aTHX_ sv, s, prog->offs[n].end - prog->offs[n].start, 32, 0, 0,
19127 PERL_PV_ESCAPE_UNI_DETECT|PERL_PV_PRETTY_NOCLEAR|PERL_PV_PRETTY_ELLIPSES|PERL_PV_PRETTY_QUOTE );
19130 } else if (k == GOSUB) {
19131 AV *name_list= NULL;
19132 if ( RXp_PAREN_NAMES(prog) ) {
19133 name_list= MUTABLE_AV(progi->data->data[progi->name_list_idx]);
19134 } else if ( pRExC_state ) {
19135 name_list= RExC_paren_name_list;
19138 /* Paren and offset */
19139 Perl_sv_catpvf(aTHX_ sv, "%d[%+d:%d]", (int)ARG(o),(int)ARG2L(o),
19140 (int)((o + (int)ARG2L(o)) - progi->program) );
19142 SV **name= av_fetch(name_list, ARG(o), 0 );
19144 Perl_sv_catpvf(aTHX_ sv, " '%" SVf "'", SVfARG(*name));
19147 else if (k == LOGICAL)
19148 /* 2: embedded, otherwise 1 */
19149 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags);
19150 else if (k == ANYOF) {
19151 const U8 flags = ANYOF_FLAGS(o);
19152 bool do_sep = FALSE; /* Do we need to separate various components of
19154 /* Set if there is still an unresolved user-defined property */
19155 SV *unresolved = NULL;
19157 /* Things that are ignored except when the runtime locale is UTF-8 */
19158 SV *only_utf8_locale_invlist = NULL;
19160 /* Code points that don't fit in the bitmap */
19161 SV *nonbitmap_invlist = NULL;
19163 /* And things that aren't in the bitmap, but are small enough to be */
19164 SV* bitmap_range_not_in_bitmap = NULL;
19166 const bool inverted = flags & ANYOF_INVERT;
19168 if (OP(o) == ANYOFL) {
19169 if (ANYOFL_UTF8_LOCALE_REQD(flags)) {
19170 sv_catpvs(sv, "{utf8-locale-reqd}");
19172 if (flags & ANYOFL_FOLD) {
19173 sv_catpvs(sv, "{i}");
19177 /* If there is stuff outside the bitmap, get it */
19178 if (ARG(o) != ANYOF_ONLY_HAS_BITMAP) {
19179 (void) _get_regclass_nonbitmap_data(prog, o, FALSE,
19181 &only_utf8_locale_invlist,
19182 &nonbitmap_invlist);
19183 /* The non-bitmap data may contain stuff that could fit in the
19184 * bitmap. This could come from a user-defined property being
19185 * finally resolved when this call was done; or much more likely
19186 * because there are matches that require UTF-8 to be valid, and so
19187 * aren't in the bitmap. This is teased apart later */
19188 _invlist_intersection(nonbitmap_invlist,
19190 &bitmap_range_not_in_bitmap);
19191 /* Leave just the things that don't fit into the bitmap */
19192 _invlist_subtract(nonbitmap_invlist,
19194 &nonbitmap_invlist);
19197 /* Obey this flag to add all above-the-bitmap code points */
19198 if (flags & ANYOF_MATCHES_ALL_ABOVE_BITMAP) {
19199 nonbitmap_invlist = _add_range_to_invlist(nonbitmap_invlist,
19200 NUM_ANYOF_CODE_POINTS,
19204 /* Ready to start outputting. First, the initial left bracket */
19205 Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
19207 /* Then all the things that could fit in the bitmap */
19208 do_sep = put_charclass_bitmap_innards(sv,
19210 bitmap_range_not_in_bitmap,
19211 only_utf8_locale_invlist,
19214 /* Can't try inverting for a
19215 * better display if there are
19216 * things that haven't been
19218 unresolved != NULL);
19219 SvREFCNT_dec(bitmap_range_not_in_bitmap);
19221 /* If there are user-defined properties which haven't been defined yet,
19222 * output them. If the result is not to be inverted, it is clearest to
19223 * output them in a separate [] from the bitmap range stuff. If the
19224 * result is to be complemented, we have to show everything in one [],
19225 * as the inversion applies to the whole thing. Use {braces} to
19226 * separate them from anything in the bitmap and anything above the
19230 if (! do_sep) { /* If didn't output anything in the bitmap */
19231 sv_catpvs(sv, "^");
19233 sv_catpvs(sv, "{");
19236 Perl_sv_catpvf(aTHX_ sv,"%s][%s",PL_colors[1],PL_colors[0]);
19238 sv_catsv(sv, unresolved);
19240 sv_catpvs(sv, "}");
19242 do_sep = ! inverted;
19245 /* And, finally, add the above-the-bitmap stuff */
19246 if (nonbitmap_invlist && _invlist_len(nonbitmap_invlist)) {
19249 /* See if truncation size is overridden */
19250 const STRLEN dump_len = (PL_dump_re_max_len)
19251 ? PL_dump_re_max_len
19254 /* This is output in a separate [] */
19256 Perl_sv_catpvf(aTHX_ sv,"%s][%s",PL_colors[1],PL_colors[0]);
19259 /* And, for easy of understanding, it is shown in the
19260 * uncomplemented form if possible. The one exception being if
19261 * there are unresolved items, where the inversion has to be
19262 * delayed until runtime */
19263 if (inverted && ! unresolved) {
19264 _invlist_invert(nonbitmap_invlist);
19265 _invlist_subtract(nonbitmap_invlist, PL_InBitmap, &nonbitmap_invlist);
19268 contents = invlist_contents(nonbitmap_invlist,
19269 FALSE /* output suitable for catsv */
19272 /* If the output is shorter than the permissible maximum, just do it. */
19273 if (SvCUR(contents) <= dump_len) {
19274 sv_catsv(sv, contents);
19277 const char * contents_string = SvPVX(contents);
19278 STRLEN i = dump_len;
19280 /* Otherwise, start at the permissible max and work back to the
19281 * first break possibility */
19282 while (i > 0 && contents_string[i] != ' ') {
19285 if (i == 0) { /* Fail-safe. Use the max if we couldn't
19286 find a legal break */
19290 sv_catpvn(sv, contents_string, i);
19291 sv_catpvs(sv, "...");
19294 SvREFCNT_dec_NN(contents);
19295 SvREFCNT_dec_NN(nonbitmap_invlist);
19298 /* And finally the matching, closing ']' */
19299 Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
19301 SvREFCNT_dec(unresolved);
19303 else if (k == POSIXD || k == NPOSIXD) {
19304 U8 index = FLAGS(o) * 2;
19305 if (index < C_ARRAY_LENGTH(anyofs)) {
19306 if (*anyofs[index] != '[') {
19309 sv_catpv(sv, anyofs[index]);
19310 if (*anyofs[index] != '[') {
19315 Perl_sv_catpvf(aTHX_ sv, "[illegal type=%d])", index);
19318 else if (k == BOUND || k == NBOUND) {
19319 /* Must be synced with order of 'bound_type' in regcomp.h */
19320 const char * const bounds[] = {
19321 "", /* Traditional */
19327 assert(FLAGS(o) < C_ARRAY_LENGTH(bounds));
19328 sv_catpv(sv, bounds[FLAGS(o)]);
19330 else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
19331 Perl_sv_catpvf(aTHX_ sv, "[%d]", -(o->flags));
19332 else if (OP(o) == SBOL)
19333 Perl_sv_catpvf(aTHX_ sv, " /%s/", o->flags ? "\\A" : "^");
19335 /* add on the verb argument if there is one */
19336 if ( ( k == VERB || OP(o) == ACCEPT || OP(o) == OPFAIL ) && o->flags) {
19337 Perl_sv_catpvf(aTHX_ sv, ":%" SVf,
19338 SVfARG((MUTABLE_SV(progi->data->data[ ARG( o ) ]))));
19341 PERL_UNUSED_CONTEXT;
19342 PERL_UNUSED_ARG(sv);
19343 PERL_UNUSED_ARG(o);
19344 PERL_UNUSED_ARG(prog);
19345 PERL_UNUSED_ARG(reginfo);
19346 PERL_UNUSED_ARG(pRExC_state);
19347 #endif /* DEBUGGING */
19353 Perl_re_intuit_string(pTHX_ REGEXP * const r)
19354 { /* Assume that RE_INTUIT is set */
19355 struct regexp *const prog = ReANY(r);
19356 GET_RE_DEBUG_FLAGS_DECL;
19358 PERL_ARGS_ASSERT_RE_INTUIT_STRING;
19359 PERL_UNUSED_CONTEXT;
19363 const char * const s = SvPV_nolen_const(RX_UTF8(r)
19364 ? prog->check_utf8 : prog->check_substr);
19366 if (!PL_colorset) reginitcolors();
19367 Perl_re_printf( aTHX_
19368 "%sUsing REx %ssubstr:%s \"%s%.60s%s%s\"\n",
19370 RX_UTF8(r) ? "utf8 " : "",
19371 PL_colors[5],PL_colors[0],
19374 (strlen(s) > 60 ? "..." : ""));
19377 /* use UTF8 check substring if regexp pattern itself is in UTF8 */
19378 return RX_UTF8(r) ? prog->check_utf8 : prog->check_substr;
19384 handles refcounting and freeing the perl core regexp structure. When
19385 it is necessary to actually free the structure the first thing it
19386 does is call the 'free' method of the regexp_engine associated to
19387 the regexp, allowing the handling of the void *pprivate; member
19388 first. (This routine is not overridable by extensions, which is why
19389 the extensions free is called first.)
19391 See regdupe and regdupe_internal if you change anything here.
19393 #ifndef PERL_IN_XSUB_RE
19395 Perl_pregfree(pTHX_ REGEXP *r)
19401 Perl_pregfree2(pTHX_ REGEXP *rx)
19403 struct regexp *const r = ReANY(rx);
19404 GET_RE_DEBUG_FLAGS_DECL;
19406 PERL_ARGS_ASSERT_PREGFREE2;
19408 if (r->mother_re) {
19409 ReREFCNT_dec(r->mother_re);
19411 CALLREGFREE_PVT(rx); /* free the private data */
19412 SvREFCNT_dec(RXp_PAREN_NAMES(r));
19413 Safefree(r->xpv_len_u.xpvlenu_pv);
19416 SvREFCNT_dec(r->anchored_substr);
19417 SvREFCNT_dec(r->anchored_utf8);
19418 SvREFCNT_dec(r->float_substr);
19419 SvREFCNT_dec(r->float_utf8);
19420 Safefree(r->substrs);
19422 RX_MATCH_COPY_FREE(rx);
19423 #ifdef PERL_ANY_COW
19424 SvREFCNT_dec(r->saved_copy);
19427 SvREFCNT_dec(r->qr_anoncv);
19428 if (r->recurse_locinput)
19429 Safefree(r->recurse_locinput);
19430 rx->sv_u.svu_rx = 0;
19435 This is a hacky workaround to the structural issue of match results
19436 being stored in the regexp structure which is in turn stored in
19437 PL_curpm/PL_reg_curpm. The problem is that due to qr// the pattern
19438 could be PL_curpm in multiple contexts, and could require multiple
19439 result sets being associated with the pattern simultaneously, such
19440 as when doing a recursive match with (??{$qr})
19442 The solution is to make a lightweight copy of the regexp structure
19443 when a qr// is returned from the code executed by (??{$qr}) this
19444 lightweight copy doesn't actually own any of its data except for
19445 the starp/end and the actual regexp structure itself.
19451 Perl_reg_temp_copy (pTHX_ REGEXP *ret_x, REGEXP *rx)
19453 struct regexp *ret;
19454 struct regexp *const r = ReANY(rx);
19455 const bool islv = ret_x && SvTYPE(ret_x) == SVt_PVLV;
19457 PERL_ARGS_ASSERT_REG_TEMP_COPY;
19460 ret_x = (REGEXP*) newSV_type(SVt_REGEXP);
19462 SvOK_off((SV *)ret_x);
19464 /* For PVLVs, SvANY points to the xpvlv body while sv_u points
19465 to the regexp. (For SVt_REGEXPs, sv_upgrade has already
19466 made both spots point to the same regexp body.) */
19467 REGEXP *temp = (REGEXP *)newSV_type(SVt_REGEXP);
19468 assert(!SvPVX(ret_x));
19469 ret_x->sv_u.svu_rx = temp->sv_any;
19470 temp->sv_any = NULL;
19471 SvFLAGS(temp) = (SvFLAGS(temp) & ~SVTYPEMASK) | SVt_NULL;
19472 SvREFCNT_dec_NN(temp);
19473 /* SvCUR still resides in the xpvlv struct, so the regexp copy-
19474 ing below will not set it. */
19475 SvCUR_set(ret_x, SvCUR(rx));
19478 /* This ensures that SvTHINKFIRST(sv) is true, and hence that
19479 sv_force_normal(sv) is called. */
19481 ret = ReANY(ret_x);
19483 SvFLAGS(ret_x) |= SvUTF8(rx);
19484 /* We share the same string buffer as the original regexp, on which we
19485 hold a reference count, incremented when mother_re is set below.
19486 The string pointer is copied here, being part of the regexp struct.
19488 memcpy(&(ret->xpv_cur), &(r->xpv_cur),
19489 sizeof(regexp) - STRUCT_OFFSET(regexp, xpv_cur));
19491 const I32 npar = r->nparens+1;
19492 Newx(ret->offs, npar, regexp_paren_pair);
19493 Copy(r->offs, ret->offs, npar, regexp_paren_pair);
19496 Newx(ret->substrs, 1, struct reg_substr_data);
19497 StructCopy(r->substrs, ret->substrs, struct reg_substr_data);
19499 SvREFCNT_inc_void(ret->anchored_substr);
19500 SvREFCNT_inc_void(ret->anchored_utf8);
19501 SvREFCNT_inc_void(ret->float_substr);
19502 SvREFCNT_inc_void(ret->float_utf8);
19504 /* check_substr and check_utf8, if non-NULL, point to either their
19505 anchored or float namesakes, and don't hold a second reference. */
19507 RX_MATCH_COPIED_off(ret_x);
19508 #ifdef PERL_ANY_COW
19509 ret->saved_copy = NULL;
19511 ret->mother_re = ReREFCNT_inc(r->mother_re ? r->mother_re : rx);
19512 SvREFCNT_inc_void(ret->qr_anoncv);
19513 if (r->recurse_locinput)
19514 Newxz(ret->recurse_locinput,r->nparens + 1,char *);
19520 /* regfree_internal()
19522 Free the private data in a regexp. This is overloadable by
19523 extensions. Perl takes care of the regexp structure in pregfree(),
19524 this covers the *pprivate pointer which technically perl doesn't
19525 know about, however of course we have to handle the
19526 regexp_internal structure when no extension is in use.
19528 Note this is called before freeing anything in the regexp
19533 Perl_regfree_internal(pTHX_ REGEXP * const rx)
19535 struct regexp *const r = ReANY(rx);
19536 RXi_GET_DECL(r,ri);
19537 GET_RE_DEBUG_FLAGS_DECL;
19539 PERL_ARGS_ASSERT_REGFREE_INTERNAL;
19545 SV *dsv= sv_newmortal();
19546 RE_PV_QUOTED_DECL(s, RX_UTF8(rx),
19547 dsv, RX_PRECOMP(rx), RX_PRELEN(rx), 60);
19548 Perl_re_printf( aTHX_ "%sFreeing REx:%s %s\n",
19549 PL_colors[4],PL_colors[5],s);
19552 #ifdef RE_TRACK_PATTERN_OFFSETS
19554 Safefree(ri->u.offsets); /* 20010421 MJD */
19556 if (ri->code_blocks)
19557 S_free_codeblocks(aTHX_ ri->code_blocks);
19560 int n = ri->data->count;
19563 /* If you add a ->what type here, update the comment in regcomp.h */
19564 switch (ri->data->what[n]) {
19570 SvREFCNT_dec(MUTABLE_SV(ri->data->data[n]));
19573 Safefree(ri->data->data[n]);
19579 { /* Aho Corasick add-on structure for a trie node.
19580 Used in stclass optimization only */
19582 reg_ac_data *aho=(reg_ac_data*)ri->data->data[n];
19583 #ifdef USE_ITHREADS
19587 refcount = --aho->refcount;
19590 PerlMemShared_free(aho->states);
19591 PerlMemShared_free(aho->fail);
19592 /* do this last!!!! */
19593 PerlMemShared_free(ri->data->data[n]);
19594 /* we should only ever get called once, so
19595 * assert as much, and also guard the free
19596 * which /might/ happen twice. At the least
19597 * it will make code anlyzers happy and it
19598 * doesn't cost much. - Yves */
19599 assert(ri->regstclass);
19600 if (ri->regstclass) {
19601 PerlMemShared_free(ri->regstclass);
19602 ri->regstclass = 0;
19609 /* trie structure. */
19611 reg_trie_data *trie=(reg_trie_data*)ri->data->data[n];
19612 #ifdef USE_ITHREADS
19616 refcount = --trie->refcount;
19619 PerlMemShared_free(trie->charmap);
19620 PerlMemShared_free(trie->states);
19621 PerlMemShared_free(trie->trans);
19623 PerlMemShared_free(trie->bitmap);
19625 PerlMemShared_free(trie->jump);
19626 PerlMemShared_free(trie->wordinfo);
19627 /* do this last!!!! */
19628 PerlMemShared_free(ri->data->data[n]);
19633 Perl_croak(aTHX_ "panic: regfree data code '%c'",
19634 ri->data->what[n]);
19637 Safefree(ri->data->what);
19638 Safefree(ri->data);
19644 #define av_dup_inc(s,t) MUTABLE_AV(sv_dup_inc((const SV *)s,t))
19645 #define hv_dup_inc(s,t) MUTABLE_HV(sv_dup_inc((const SV *)s,t))
19646 #define SAVEPVN(p,n) ((p) ? savepvn(p,n) : NULL)
19649 re_dup_guts - duplicate a regexp.
19651 This routine is expected to clone a given regexp structure. It is only
19652 compiled under USE_ITHREADS.
19654 After all of the core data stored in struct regexp is duplicated
19655 the regexp_engine.dupe method is used to copy any private data
19656 stored in the *pprivate pointer. This allows extensions to handle
19657 any duplication it needs to do.
19659 See pregfree() and regfree_internal() if you change anything here.
19661 #if defined(USE_ITHREADS)
19662 #ifndef PERL_IN_XSUB_RE
19664 Perl_re_dup_guts(pTHX_ const REGEXP *sstr, REGEXP *dstr, CLONE_PARAMS *param)
19668 const struct regexp *r = ReANY(sstr);
19669 struct regexp *ret = ReANY(dstr);
19671 PERL_ARGS_ASSERT_RE_DUP_GUTS;
19673 npar = r->nparens+1;
19674 Newx(ret->offs, npar, regexp_paren_pair);
19675 Copy(r->offs, ret->offs, npar, regexp_paren_pair);
19677 if (ret->substrs) {
19678 /* Do it this way to avoid reading from *r after the StructCopy().
19679 That way, if any of the sv_dup_inc()s dislodge *r from the L1
19680 cache, it doesn't matter. */
19681 const bool anchored = r->check_substr
19682 ? r->check_substr == r->anchored_substr
19683 : r->check_utf8 == r->anchored_utf8;
19684 Newx(ret->substrs, 1, struct reg_substr_data);
19685 StructCopy(r->substrs, ret->substrs, struct reg_substr_data);
19687 ret->anchored_substr = sv_dup_inc(ret->anchored_substr, param);
19688 ret->anchored_utf8 = sv_dup_inc(ret->anchored_utf8, param);
19689 ret->float_substr = sv_dup_inc(ret->float_substr, param);
19690 ret->float_utf8 = sv_dup_inc(ret->float_utf8, param);
19692 /* check_substr and check_utf8, if non-NULL, point to either their
19693 anchored or float namesakes, and don't hold a second reference. */
19695 if (ret->check_substr) {
19697 assert(r->check_utf8 == r->anchored_utf8);
19698 ret->check_substr = ret->anchored_substr;
19699 ret->check_utf8 = ret->anchored_utf8;
19701 assert(r->check_substr == r->float_substr);
19702 assert(r->check_utf8 == r->float_utf8);
19703 ret->check_substr = ret->float_substr;
19704 ret->check_utf8 = ret->float_utf8;
19706 } else if (ret->check_utf8) {
19708 ret->check_utf8 = ret->anchored_utf8;
19710 ret->check_utf8 = ret->float_utf8;
19715 RXp_PAREN_NAMES(ret) = hv_dup_inc(RXp_PAREN_NAMES(ret), param);
19716 ret->qr_anoncv = MUTABLE_CV(sv_dup_inc((const SV *)ret->qr_anoncv, param));
19717 if (r->recurse_locinput)
19718 Newxz(ret->recurse_locinput,r->nparens + 1,char *);
19721 RXi_SET(ret,CALLREGDUPE_PVT(dstr,param));
19723 if (RX_MATCH_COPIED(dstr))
19724 ret->subbeg = SAVEPVN(ret->subbeg, ret->sublen);
19726 ret->subbeg = NULL;
19727 #ifdef PERL_ANY_COW
19728 ret->saved_copy = NULL;
19731 /* Whether mother_re be set or no, we need to copy the string. We
19732 cannot refrain from copying it when the storage points directly to
19733 our mother regexp, because that's
19734 1: a buffer in a different thread
19735 2: something we no longer hold a reference on
19736 so we need to copy it locally. */
19737 RX_WRAPPED(dstr) = SAVEPVN(RX_WRAPPED(sstr), SvCUR(sstr)+1);
19738 ret->mother_re = NULL;
19740 #endif /* PERL_IN_XSUB_RE */
19745 This is the internal complement to regdupe() which is used to copy
19746 the structure pointed to by the *pprivate pointer in the regexp.
19747 This is the core version of the extension overridable cloning hook.
19748 The regexp structure being duplicated will be copied by perl prior
19749 to this and will be provided as the regexp *r argument, however
19750 with the /old/ structures pprivate pointer value. Thus this routine
19751 may override any copying normally done by perl.
19753 It returns a pointer to the new regexp_internal structure.
19757 Perl_regdupe_internal(pTHX_ REGEXP * const rx, CLONE_PARAMS *param)
19760 struct regexp *const r = ReANY(rx);
19761 regexp_internal *reti;
19763 RXi_GET_DECL(r,ri);
19765 PERL_ARGS_ASSERT_REGDUPE_INTERNAL;
19769 Newxc(reti, sizeof(regexp_internal) + len*sizeof(regnode),
19770 char, regexp_internal);
19771 Copy(ri->program, reti->program, len+1, regnode);
19774 if (ri->code_blocks) {
19776 Newx(reti->code_blocks, 1, struct reg_code_blocks);
19777 Newx(reti->code_blocks->cb, ri->code_blocks->count,
19778 struct reg_code_block);
19779 Copy(ri->code_blocks->cb, reti->code_blocks->cb,
19780 ri->code_blocks->count, struct reg_code_block);
19781 for (n = 0; n < ri->code_blocks->count; n++)
19782 reti->code_blocks->cb[n].src_regex = (REGEXP*)
19783 sv_dup_inc((SV*)(ri->code_blocks->cb[n].src_regex), param);
19784 reti->code_blocks->count = ri->code_blocks->count;
19785 reti->code_blocks->refcnt = 1;
19788 reti->code_blocks = NULL;
19790 reti->regstclass = NULL;
19793 struct reg_data *d;
19794 const int count = ri->data->count;
19797 Newxc(d, sizeof(struct reg_data) + count*sizeof(void *),
19798 char, struct reg_data);
19799 Newx(d->what, count, U8);
19802 for (i = 0; i < count; i++) {
19803 d->what[i] = ri->data->what[i];
19804 switch (d->what[i]) {
19805 /* see also regcomp.h and regfree_internal() */
19806 case 'a': /* actually an AV, but the dup function is identical. */
19810 case 'u': /* actually an HV, but the dup function is identical. */
19811 d->data[i] = sv_dup_inc((const SV *)ri->data->data[i], param);
19814 /* This is cheating. */
19815 Newx(d->data[i], 1, regnode_ssc);
19816 StructCopy(ri->data->data[i], d->data[i], regnode_ssc);
19817 reti->regstclass = (regnode*)d->data[i];
19820 /* Trie stclasses are readonly and can thus be shared
19821 * without duplication. We free the stclass in pregfree
19822 * when the corresponding reg_ac_data struct is freed.
19824 reti->regstclass= ri->regstclass;
19828 ((reg_trie_data*)ri->data->data[i])->refcount++;
19833 d->data[i] = ri->data->data[i];
19836 Perl_croak(aTHX_ "panic: re_dup_guts unknown data code '%c'",
19837 ri->data->what[i]);
19846 reti->name_list_idx = ri->name_list_idx;
19848 #ifdef RE_TRACK_PATTERN_OFFSETS
19849 if (ri->u.offsets) {
19850 Newx(reti->u.offsets, 2*len+1, U32);
19851 Copy(ri->u.offsets, reti->u.offsets, 2*len+1, U32);
19854 SetProgLen(reti,len);
19857 return (void*)reti;
19860 #endif /* USE_ITHREADS */
19862 #ifndef PERL_IN_XSUB_RE
19865 - regnext - dig the "next" pointer out of a node
19868 Perl_regnext(pTHX_ regnode *p)
19875 if (OP(p) > REGNODE_MAX) { /* regnode.type is unsigned */
19876 Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d",
19877 (int)OP(p), (int)REGNODE_MAX);
19880 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
19889 S_re_croak2(pTHX_ bool utf8, const char* pat1,const char* pat2,...)
19892 STRLEN l1 = strlen(pat1);
19893 STRLEN l2 = strlen(pat2);
19896 const char *message;
19898 PERL_ARGS_ASSERT_RE_CROAK2;
19904 Copy(pat1, buf, l1 , char);
19905 Copy(pat2, buf + l1, l2 , char);
19906 buf[l1 + l2] = '\n';
19907 buf[l1 + l2 + 1] = '\0';
19908 va_start(args, pat2);
19909 msv = vmess(buf, &args);
19911 message = SvPV_const(msv,l1);
19914 Copy(message, buf, l1 , char);
19915 /* l1-1 to avoid \n */
19916 Perl_croak(aTHX_ "%" UTF8f, UTF8fARG(utf8, l1-1, buf));
19919 /* XXX Here's a total kludge. But we need to re-enter for swash routines. */
19921 #ifndef PERL_IN_XSUB_RE
19923 Perl_save_re_context(pTHX)
19928 /* Save $1..$n (#18107: UTF-8 s/(\w+)/uc($1)/e); AMS 20021106. */
19931 const REGEXP * const rx = PM_GETRE(PL_curpm);
19933 nparens = RX_NPARENS(rx);
19936 /* RT #124109. This is a complete hack; in the SWASHNEW case we know
19937 * that PL_curpm will be null, but that utf8.pm and the modules it
19938 * loads will only use $1..$3.
19939 * The t/porting/re_context.t test file checks this assumption.
19944 for (i = 1; i <= nparens; i++) {
19945 char digits[TYPE_CHARS(long)];
19946 const STRLEN len = my_snprintf(digits, sizeof(digits),
19948 GV *const *const gvp
19949 = (GV**)hv_fetch(PL_defstash, digits, len, 0);
19952 GV * const gv = *gvp;
19953 if (SvTYPE(gv) == SVt_PVGV && GvSV(gv))
19963 S_put_code_point(pTHX_ SV *sv, UV c)
19965 PERL_ARGS_ASSERT_PUT_CODE_POINT;
19968 Perl_sv_catpvf(aTHX_ sv, "\\x{%04" UVXf "}", c);
19970 else if (isPRINT(c)) {
19971 const char string = (char) c;
19973 /* We use {phrase} as metanotation in the class, so also escape literal
19975 if (isBACKSLASHED_PUNCT(c) || c == '{' || c == '}')
19976 sv_catpvs(sv, "\\");
19977 sv_catpvn(sv, &string, 1);
19979 else if (isMNEMONIC_CNTRL(c)) {
19980 Perl_sv_catpvf(aTHX_ sv, "%s", cntrl_to_mnemonic((U8) c));
19983 Perl_sv_catpvf(aTHX_ sv, "\\x%02X", (U8) c);
19987 #define MAX_PRINT_A MAX_PRINT_A_FOR_USE_ONLY_BY_REGCOMP_DOT_C
19990 S_put_range(pTHX_ SV *sv, UV start, const UV end, const bool allow_literals)
19992 /* Appends to 'sv' a displayable version of the range of code points from
19993 * 'start' to 'end'. Mnemonics (like '\r') are used for the few controls
19994 * that have them, when they occur at the beginning or end of the range.
19995 * It uses hex to output the remaining code points, unless 'allow_literals'
19996 * is true, in which case the printable ASCII ones are output as-is (though
19997 * some of these will be escaped by put_code_point()).
19999 * NOTE: This is designed only for printing ranges of code points that fit
20000 * inside an ANYOF bitmap. Higher code points are simply suppressed
20003 const unsigned int min_range_count = 3;
20005 assert(start <= end);
20007 PERL_ARGS_ASSERT_PUT_RANGE;
20009 while (start <= end) {
20011 const char * format;
20013 if (end - start < min_range_count) {
20015 /* Output chars individually when they occur in short ranges */
20016 for (; start <= end; start++) {
20017 put_code_point(sv, start);
20022 /* If permitted by the input options, and there is a possibility that
20023 * this range contains a printable literal, look to see if there is
20025 if (allow_literals && start <= MAX_PRINT_A) {
20027 /* If the character at the beginning of the range isn't an ASCII
20028 * printable, effectively split the range into two parts:
20029 * 1) the portion before the first such printable,
20031 * and output them separately. */
20032 if (! isPRINT_A(start)) {
20033 UV temp_end = start + 1;
20035 /* There is no point looking beyond the final possible
20036 * printable, in MAX_PRINT_A */
20037 UV max = MIN(end, MAX_PRINT_A);
20039 while (temp_end <= max && ! isPRINT_A(temp_end)) {
20043 /* Here, temp_end points to one beyond the first printable if
20044 * found, or to one beyond 'max' if not. If none found, make
20045 * sure that we use the entire range */
20046 if (temp_end > MAX_PRINT_A) {
20047 temp_end = end + 1;
20050 /* Output the first part of the split range: the part that
20051 * doesn't have printables, with the parameter set to not look
20052 * for literals (otherwise we would infinitely recurse) */
20053 put_range(sv, start, temp_end - 1, FALSE);
20055 /* The 2nd part of the range (if any) starts here. */
20058 /* We do a continue, instead of dropping down, because even if
20059 * the 2nd part is non-empty, it could be so short that we want
20060 * to output it as individual characters, as tested for at the
20061 * top of this loop. */
20065 /* Here, 'start' is a printable ASCII. If it is an alphanumeric,
20066 * output a sub-range of just the digits or letters, then process
20067 * the remaining portion as usual. */
20068 if (isALPHANUMERIC_A(start)) {
20069 UV mask = (isDIGIT_A(start))
20074 UV temp_end = start + 1;
20076 /* Find the end of the sub-range that includes just the
20077 * characters in the same class as the first character in it */
20078 while (temp_end <= end && _generic_isCC_A(temp_end, mask)) {
20083 /* For short ranges, don't duplicate the code above to output
20084 * them; just call recursively */
20085 if (temp_end - start < min_range_count) {
20086 put_range(sv, start, temp_end, FALSE);
20088 else { /* Output as a range */
20089 put_code_point(sv, start);
20090 sv_catpvs(sv, "-");
20091 put_code_point(sv, temp_end);
20093 start = temp_end + 1;
20097 /* We output any other printables as individual characters */
20098 if (isPUNCT_A(start) || isSPACE_A(start)) {
20099 while (start <= end && (isPUNCT_A(start)
20100 || isSPACE_A(start)))
20102 put_code_point(sv, start);
20107 } /* End of looking for literals */
20109 /* Here is not to output as a literal. Some control characters have
20110 * mnemonic names. Split off any of those at the beginning and end of
20111 * the range to print mnemonically. It isn't possible for many of
20112 * these to be in a row, so this won't overwhelm with output */
20114 && (isMNEMONIC_CNTRL(start) || isMNEMONIC_CNTRL(end)))
20116 while (isMNEMONIC_CNTRL(start) && start <= end) {
20117 put_code_point(sv, start);
20121 /* If this didn't take care of the whole range ... */
20122 if (start <= end) {
20124 /* Look backwards from the end to find the final non-mnemonic
20127 while (isMNEMONIC_CNTRL(temp_end)) {
20131 /* And separately output the interior range that doesn't start
20132 * or end with mnemonics */
20133 put_range(sv, start, temp_end, FALSE);
20135 /* Then output the mnemonic trailing controls */
20136 start = temp_end + 1;
20137 while (start <= end) {
20138 put_code_point(sv, start);
20145 /* As a final resort, output the range or subrange as hex. */
20147 this_end = (end < NUM_ANYOF_CODE_POINTS)
20149 : NUM_ANYOF_CODE_POINTS - 1;
20150 #if NUM_ANYOF_CODE_POINTS > 256
20151 format = (this_end < 256)
20152 ? "\\x%02" UVXf "-\\x%02" UVXf
20153 : "\\x{%04" UVXf "}-\\x{%04" UVXf "}";
20155 format = "\\x%02" UVXf "-\\x%02" UVXf;
20157 GCC_DIAG_IGNORE(-Wformat-nonliteral);
20158 Perl_sv_catpvf(aTHX_ sv, format, start, this_end);
20165 S_put_charclass_bitmap_innards_invlist(pTHX_ SV *sv, SV* invlist)
20167 /* Concatenate onto the PV in 'sv' a displayable form of the inversion list
20171 bool allow_literals = TRUE;
20173 PERL_ARGS_ASSERT_PUT_CHARCLASS_BITMAP_INNARDS_INVLIST;
20175 /* Generally, it is more readable if printable characters are output as
20176 * literals, but if a range (nearly) spans all of them, it's best to output
20177 * it as a single range. This code will use a single range if all but 2
20178 * ASCII printables are in it */
20179 invlist_iterinit(invlist);
20180 while (invlist_iternext(invlist, &start, &end)) {
20182 /* If the range starts beyond the final printable, it doesn't have any
20184 if (start > MAX_PRINT_A) {
20188 /* In both ASCII and EBCDIC, a SPACE is the lowest printable. To span
20189 * all but two, the range must start and end no later than 2 from
20191 if (start < ' ' + 2 && end > MAX_PRINT_A - 2) {
20192 if (end > MAX_PRINT_A) {
20198 if (end - start >= MAX_PRINT_A - ' ' - 2) {
20199 allow_literals = FALSE;
20204 invlist_iterfinish(invlist);
20206 /* Here we have figured things out. Output each range */
20207 invlist_iterinit(invlist);
20208 while (invlist_iternext(invlist, &start, &end)) {
20209 if (start >= NUM_ANYOF_CODE_POINTS) {
20212 put_range(sv, start, end, allow_literals);
20214 invlist_iterfinish(invlist);
20220 S_put_charclass_bitmap_innards_common(pTHX_
20221 SV* invlist, /* The bitmap */
20222 SV* posixes, /* Under /l, things like [:word:], \S */
20223 SV* only_utf8, /* Under /d, matches iff the target is UTF-8 */
20224 SV* not_utf8, /* /d, matches iff the target isn't UTF-8 */
20225 SV* only_utf8_locale, /* Under /l, matches if the locale is UTF-8 */
20226 const bool invert /* Is the result to be inverted? */
20229 /* Create and return an SV containing a displayable version of the bitmap
20230 * and associated information determined by the input parameters. If the
20231 * output would have been only the inversion indicator '^', NULL is instead
20236 PERL_ARGS_ASSERT_PUT_CHARCLASS_BITMAP_INNARDS_COMMON;
20239 output = newSVpvs("^");
20242 output = newSVpvs("");
20245 /* First, the code points in the bitmap that are unconditionally there */
20246 put_charclass_bitmap_innards_invlist(output, invlist);
20248 /* Traditionally, these have been placed after the main code points */
20250 sv_catsv(output, posixes);
20253 if (only_utf8 && _invlist_len(only_utf8)) {
20254 Perl_sv_catpvf(aTHX_ output, "%s{utf8}%s", PL_colors[1], PL_colors[0]);
20255 put_charclass_bitmap_innards_invlist(output, only_utf8);
20258 if (not_utf8 && _invlist_len(not_utf8)) {
20259 Perl_sv_catpvf(aTHX_ output, "%s{not utf8}%s", PL_colors[1], PL_colors[0]);
20260 put_charclass_bitmap_innards_invlist(output, not_utf8);
20263 if (only_utf8_locale && _invlist_len(only_utf8_locale)) {
20264 Perl_sv_catpvf(aTHX_ output, "%s{utf8 locale}%s", PL_colors[1], PL_colors[0]);
20265 put_charclass_bitmap_innards_invlist(output, only_utf8_locale);
20267 /* This is the only list in this routine that can legally contain code
20268 * points outside the bitmap range. The call just above to
20269 * 'put_charclass_bitmap_innards_invlist' will simply suppress them, so
20270 * output them here. There's about a half-dozen possible, and none in
20271 * contiguous ranges longer than 2 */
20272 if (invlist_highest(only_utf8_locale) >= NUM_ANYOF_CODE_POINTS) {
20274 SV* above_bitmap = NULL;
20276 _invlist_subtract(only_utf8_locale, PL_InBitmap, &above_bitmap);
20278 invlist_iterinit(above_bitmap);
20279 while (invlist_iternext(above_bitmap, &start, &end)) {
20282 for (i = start; i <= end; i++) {
20283 put_code_point(output, i);
20286 invlist_iterfinish(above_bitmap);
20287 SvREFCNT_dec_NN(above_bitmap);
20291 if (invert && SvCUR(output) == 1) {
20299 S_put_charclass_bitmap_innards(pTHX_ SV *sv,
20301 SV *nonbitmap_invlist,
20302 SV *only_utf8_locale_invlist,
20303 const regnode * const node,
20304 const bool force_as_is_display)
20306 /* Appends to 'sv' a displayable version of the innards of the bracketed
20307 * character class defined by the other arguments:
20308 * 'bitmap' points to the bitmap.
20309 * 'nonbitmap_invlist' is an inversion list of the code points that are in
20310 * the bitmap range, but for some reason aren't in the bitmap; NULL if
20311 * none. The reasons for this could be that they require some
20312 * condition such as the target string being or not being in UTF-8
20313 * (under /d), or because they came from a user-defined property that
20314 * was not resolved at the time of the regex compilation (under /u)
20315 * 'only_utf8_locale_invlist' is an inversion list of the code points that
20316 * are valid only if the runtime locale is a UTF-8 one; NULL if none
20317 * 'node' is the regex pattern node. It is needed only when the above two
20318 * parameters are not null, and is passed so that this routine can
20319 * tease apart the various reasons for them.
20320 * 'force_as_is_display' is TRUE if this routine should definitely NOT try
20321 * to invert things to see if that leads to a cleaner display. If
20322 * FALSE, this routine is free to use its judgment about doing this.
20324 * It returns TRUE if there was actually something output. (It may be that
20325 * the bitmap, etc is empty.)
20327 * When called for outputting the bitmap of a non-ANYOF node, just pass the
20328 * bitmap, with the succeeding parameters set to NULL, and the final one to
20332 /* In general, it tries to display the 'cleanest' representation of the
20333 * innards, choosing whether to display them inverted or not, regardless of
20334 * whether the class itself is to be inverted. However, there are some
20335 * cases where it can't try inverting, as what actually matches isn't known
20336 * until runtime, and hence the inversion isn't either. */
20337 bool inverting_allowed = ! force_as_is_display;
20340 STRLEN orig_sv_cur = SvCUR(sv);
20342 SV* invlist; /* Inversion list we accumulate of code points that
20343 are unconditionally matched */
20344 SV* only_utf8 = NULL; /* Under /d, list of matches iff the target is
20346 SV* not_utf8 = NULL; /* /d, list of matches iff the target isn't UTF-8
20348 SV* posixes = NULL; /* Under /l, string of things like [:word:], \D */
20349 SV* only_utf8_locale = NULL; /* Under /l, list of matches if the locale
20352 SV* as_is_display; /* The output string when we take the inputs
20354 SV* inverted_display; /* The output string when we invert the inputs */
20356 U8 flags = (node) ? ANYOF_FLAGS(node) : 0;
20358 bool invert = cBOOL(flags & ANYOF_INVERT); /* Is the input to be inverted
20360 /* We are biased in favor of displaying things without them being inverted,
20361 * as that is generally easier to understand */
20362 const int bias = 5;
20364 PERL_ARGS_ASSERT_PUT_CHARCLASS_BITMAP_INNARDS;
20366 /* Start off with whatever code points are passed in. (We clone, so we
20367 * don't change the caller's list) */
20368 if (nonbitmap_invlist) {
20369 assert(invlist_highest(nonbitmap_invlist) < NUM_ANYOF_CODE_POINTS);
20370 invlist = invlist_clone(nonbitmap_invlist);
20372 else { /* Worst case size is every other code point is matched */
20373 invlist = _new_invlist(NUM_ANYOF_CODE_POINTS / 2);
20377 if (OP(node) == ANYOFD) {
20379 /* This flag indicates that the code points below 0x100 in the
20380 * nonbitmap list are precisely the ones that match only when the
20381 * target is UTF-8 (they should all be non-ASCII). */
20382 if (flags & ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP)
20384 _invlist_intersection(invlist, PL_UpperLatin1, &only_utf8);
20385 _invlist_subtract(invlist, only_utf8, &invlist);
20388 /* And this flag for matching all non-ASCII 0xFF and below */
20389 if (flags & ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER)
20391 not_utf8 = invlist_clone(PL_UpperLatin1);
20394 else if (OP(node) == ANYOFL) {
20396 /* If either of these flags are set, what matches isn't
20397 * determinable except during execution, so don't know enough here
20399 if (flags & (ANYOFL_FOLD|ANYOF_MATCHES_POSIXL)) {
20400 inverting_allowed = FALSE;
20403 /* What the posix classes match also varies at runtime, so these
20404 * will be output symbolically. */
20405 if (ANYOF_POSIXL_TEST_ANY_SET(node)) {
20408 posixes = newSVpvs("");
20409 for (i = 0; i < ANYOF_POSIXL_MAX; i++) {
20410 if (ANYOF_POSIXL_TEST(node,i)) {
20411 sv_catpv(posixes, anyofs[i]);
20418 /* Accumulate the bit map into the unconditional match list */
20419 for (i = 0; i < NUM_ANYOF_CODE_POINTS; i++) {
20420 if (BITMAP_TEST(bitmap, i)) {
20422 for (; i < NUM_ANYOF_CODE_POINTS && BITMAP_TEST(bitmap, i); i++) {
20425 invlist = _add_range_to_invlist(invlist, start, i-1);
20429 /* Make sure that the conditional match lists don't have anything in them
20430 * that match unconditionally; otherwise the output is quite confusing.
20431 * This could happen if the code that populates these misses some
20434 _invlist_subtract(only_utf8, invlist, &only_utf8);
20437 _invlist_subtract(not_utf8, invlist, ¬_utf8);
20440 if (only_utf8_locale_invlist) {
20442 /* Since this list is passed in, we have to make a copy before
20444 only_utf8_locale = invlist_clone(only_utf8_locale_invlist);
20446 _invlist_subtract(only_utf8_locale, invlist, &only_utf8_locale);
20448 /* And, it can get really weird for us to try outputting an inverted
20449 * form of this list when it has things above the bitmap, so don't even
20451 if (invlist_highest(only_utf8_locale) >= NUM_ANYOF_CODE_POINTS) {
20452 inverting_allowed = FALSE;
20456 /* Calculate what the output would be if we take the input as-is */
20457 as_is_display = put_charclass_bitmap_innards_common(invlist,
20464 /* If have to take the output as-is, just do that */
20465 if (! inverting_allowed) {
20466 if (as_is_display) {
20467 sv_catsv(sv, as_is_display);
20468 SvREFCNT_dec_NN(as_is_display);
20471 else { /* But otherwise, create the output again on the inverted input, and
20472 use whichever version is shorter */
20474 int inverted_bias, as_is_bias;
20476 /* We will apply our bias to whichever of the the results doesn't have
20486 inverted_bias = bias;
20489 /* Now invert each of the lists that contribute to the output,
20490 * excluding from the result things outside the possible range */
20492 /* For the unconditional inversion list, we have to add in all the
20493 * conditional code points, so that when inverted, they will be gone
20495 _invlist_union(only_utf8, invlist, &invlist);
20496 _invlist_union(not_utf8, invlist, &invlist);
20497 _invlist_union(only_utf8_locale, invlist, &invlist);
20498 _invlist_invert(invlist);
20499 _invlist_intersection(invlist, PL_InBitmap, &invlist);
20502 _invlist_invert(only_utf8);
20503 _invlist_intersection(only_utf8, PL_UpperLatin1, &only_utf8);
20505 else if (not_utf8) {
20507 /* If a code point matches iff the target string is not in UTF-8,
20508 * then complementing the result has it not match iff not in UTF-8,
20509 * which is the same thing as matching iff it is UTF-8. */
20510 only_utf8 = not_utf8;
20514 if (only_utf8_locale) {
20515 _invlist_invert(only_utf8_locale);
20516 _invlist_intersection(only_utf8_locale,
20518 &only_utf8_locale);
20521 inverted_display = put_charclass_bitmap_innards_common(
20526 only_utf8_locale, invert);
20528 /* Use the shortest representation, taking into account our bias
20529 * against showing it inverted */
20530 if ( inverted_display
20531 && ( ! as_is_display
20532 || ( SvCUR(inverted_display) + inverted_bias
20533 < SvCUR(as_is_display) + as_is_bias)))
20535 sv_catsv(sv, inverted_display);
20537 else if (as_is_display) {
20538 sv_catsv(sv, as_is_display);
20541 SvREFCNT_dec(as_is_display);
20542 SvREFCNT_dec(inverted_display);
20545 SvREFCNT_dec_NN(invlist);
20546 SvREFCNT_dec(only_utf8);
20547 SvREFCNT_dec(not_utf8);
20548 SvREFCNT_dec(posixes);
20549 SvREFCNT_dec(only_utf8_locale);
20551 return SvCUR(sv) > orig_sv_cur;
20554 #define CLEAR_OPTSTART \
20555 if (optstart) STMT_START { \
20556 DEBUG_OPTIMISE_r(Perl_re_printf( aTHX_ \
20557 " (%" IVdf " nodes)\n", (IV)(node - optstart))); \
20561 #define DUMPUNTIL(b,e) \
20563 node=dumpuntil(r,start,(b),(e),last,sv,indent+1,depth+1);
20565 STATIC const regnode *
20566 S_dumpuntil(pTHX_ const regexp *r, const regnode *start, const regnode *node,
20567 const regnode *last, const regnode *plast,
20568 SV* sv, I32 indent, U32 depth)
20570 U8 op = PSEUDO; /* Arbitrary non-END op. */
20571 const regnode *next;
20572 const regnode *optstart= NULL;
20574 RXi_GET_DECL(r,ri);
20575 GET_RE_DEBUG_FLAGS_DECL;
20577 PERL_ARGS_ASSERT_DUMPUNTIL;
20579 #ifdef DEBUG_DUMPUNTIL
20580 Perl_re_printf( aTHX_ "--- %d : %d - %d - %d\n",indent,node-start,
20581 last ? last-start : 0,plast ? plast-start : 0);
20584 if (plast && plast < last)
20587 while (PL_regkind[op] != END && (!last || node < last)) {
20589 /* While that wasn't END last time... */
20592 if (op == CLOSE || op == WHILEM)
20594 next = regnext((regnode *)node);
20597 if (OP(node) == OPTIMIZED) {
20598 if (!optstart && RE_DEBUG_FLAG(RE_DEBUG_COMPILE_OPTIMISE))
20605 regprop(r, sv, node, NULL, NULL);
20606 Perl_re_printf( aTHX_ "%4" IVdf ":%*s%s", (IV)(node - start),
20607 (int)(2*indent + 1), "", SvPVX_const(sv));
20609 if (OP(node) != OPTIMIZED) {
20610 if (next == NULL) /* Next ptr. */
20611 Perl_re_printf( aTHX_ " (0)");
20612 else if (PL_regkind[(U8)op] == BRANCH
20613 && PL_regkind[OP(next)] != BRANCH )
20614 Perl_re_printf( aTHX_ " (FAIL)");
20616 Perl_re_printf( aTHX_ " (%" IVdf ")", (IV)(next - start));
20617 Perl_re_printf( aTHX_ "\n");
20621 if (PL_regkind[(U8)op] == BRANCHJ) {
20624 const regnode *nnode = (OP(next) == LONGJMP
20625 ? regnext((regnode *)next)
20627 if (last && nnode > last)
20629 DUMPUNTIL(NEXTOPER(NEXTOPER(node)), nnode);
20632 else if (PL_regkind[(U8)op] == BRANCH) {
20634 DUMPUNTIL(NEXTOPER(node), next);
20636 else if ( PL_regkind[(U8)op] == TRIE ) {
20637 const regnode *this_trie = node;
20638 const char op = OP(node);
20639 const U32 n = ARG(node);
20640 const reg_ac_data * const ac = op>=AHOCORASICK ?
20641 (reg_ac_data *)ri->data->data[n] :
20643 const reg_trie_data * const trie =
20644 (reg_trie_data*)ri->data->data[op<AHOCORASICK ? n : ac->trie];
20646 AV *const trie_words
20647 = MUTABLE_AV(ri->data->data[n + TRIE_WORDS_OFFSET]);
20649 const regnode *nextbranch= NULL;
20652 for (word_idx= 0; word_idx < (I32)trie->wordcount; word_idx++) {
20653 SV ** const elem_ptr = av_fetch(trie_words,word_idx,0);
20655 Perl_re_indentf( aTHX_ "%s ",
20658 ? pv_pretty(sv, SvPV_nolen_const(*elem_ptr),
20659 SvCUR(*elem_ptr), 60,
20660 PL_colors[0], PL_colors[1],
20662 ? PERL_PV_ESCAPE_UNI
20664 | PERL_PV_PRETTY_ELLIPSES
20665 | PERL_PV_PRETTY_LTGT
20670 U16 dist= trie->jump[word_idx+1];
20671 Perl_re_printf( aTHX_ "(%" UVuf ")\n",
20672 (UV)((dist ? this_trie + dist : next) - start));
20675 nextbranch= this_trie + trie->jump[0];
20676 DUMPUNTIL(this_trie + dist, nextbranch);
20678 if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
20679 nextbranch= regnext((regnode *)nextbranch);
20681 Perl_re_printf( aTHX_ "\n");
20684 if (last && next > last)
20689 else if ( op == CURLY ) { /* "next" might be very big: optimizer */
20690 DUMPUNTIL(NEXTOPER(node) + EXTRA_STEP_2ARGS,
20691 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1);
20693 else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
20695 DUMPUNTIL(NEXTOPER(node) + EXTRA_STEP_2ARGS, next);
20697 else if ( op == PLUS || op == STAR) {
20698 DUMPUNTIL(NEXTOPER(node), NEXTOPER(node) + 1);
20700 else if (PL_regkind[(U8)op] == ANYOF) {
20701 /* arglen 1 + class block */
20702 node += 1 + ((ANYOF_FLAGS(node) & ANYOF_MATCHES_POSIXL)
20703 ? ANYOF_POSIXL_SKIP
20705 node = NEXTOPER(node);
20707 else if (PL_regkind[(U8)op] == EXACT) {
20708 /* Literal string, where present. */
20709 node += NODE_SZ_STR(node) - 1;
20710 node = NEXTOPER(node);
20713 node = NEXTOPER(node);
20714 node += regarglen[(U8)op];
20716 if (op == CURLYX || op == OPEN)
20720 #ifdef DEBUG_DUMPUNTIL
20721 Perl_re_printf( aTHX_ "--- %d\n", (int)indent);
20726 #endif /* DEBUGGING */
20729 * ex: set ts=8 sts=4 sw=4 et: