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 const struct regexp_engine my_reg_engine;
89 #include "dquote_static.c"
90 #include "charclass_invlists.h"
91 #include "inline_invlist.c"
92 #include "unicode_constants.h"
94 #define HAS_NONLATIN1_FOLD_CLOSURE(i) _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(i)
95 #define IS_NON_FINAL_FOLD(c) _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c)
96 #define IS_IN_SOME_FOLD_L1(c) _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c)
103 # if defined(BUGGY_MSC6)
104 /* MSC 6.00A breaks on op/regexp.t test 85 unless we turn this off */
105 # pragma optimize("a",off)
106 /* But MSC 6.00A is happy with 'w', for aliases only across function calls*/
107 # pragma optimize("w",on )
108 # endif /* BUGGY_MSC6 */
112 #define STATIC static
116 typedef struct RExC_state_t {
117 U32 flags; /* RXf_* are we folding, multilining? */
118 U32 pm_flags; /* PMf_* stuff from the calling PMOP */
119 char *precomp; /* uncompiled string. */
120 REGEXP *rx_sv; /* The SV that is the regexp. */
121 regexp *rx; /* perl core regexp structure */
122 regexp_internal *rxi; /* internal data for regexp object pprivate field */
123 char *start; /* Start of input for compile */
124 char *end; /* End of input for compile */
125 char *parse; /* Input-scan pointer. */
126 I32 whilem_seen; /* number of WHILEM in this expr */
127 regnode *emit_start; /* Start of emitted-code area */
128 regnode *emit_bound; /* First regnode outside of the allocated space */
129 regnode *emit; /* Code-emit pointer; if = &emit_dummy,
130 implies compiling, so don't emit */
131 regnode emit_dummy; /* placeholder for emit to point to */
132 I32 naughty; /* How bad is this pattern? */
133 I32 sawback; /* Did we see \1, ...? */
135 I32 size; /* Code size. */
136 I32 npar; /* Capture buffer count, (OPEN). */
137 I32 cpar; /* Capture buffer count, (CLOSE). */
138 I32 nestroot; /* root parens we are in - used by accept */
141 regnode **open_parens; /* pointers to open parens */
142 regnode **close_parens; /* pointers to close parens */
143 regnode *opend; /* END node in program */
144 I32 utf8; /* whether the pattern is utf8 or not */
145 I32 orig_utf8; /* whether the pattern was originally in utf8 */
146 /* XXX use this for future optimisation of case
147 * where pattern must be upgraded to utf8. */
148 I32 uni_semantics; /* If a d charset modifier should use unicode
149 rules, even if the pattern is not in
151 HV *paren_names; /* Paren names */
153 regnode **recurse; /* Recurse regops */
154 I32 recurse_count; /* Number of recurse regops */
157 I32 override_recoding;
158 I32 in_multi_char_class;
159 struct reg_code_block *code_blocks; /* positions of literal (?{})
161 int num_code_blocks; /* size of code_blocks[] */
162 int code_index; /* next code_blocks[] slot */
164 char *starttry; /* -Dr: where regtry was called. */
165 #define RExC_starttry (pRExC_state->starttry)
167 SV *runtime_code_qr; /* qr with the runtime code blocks */
169 const char *lastparse;
171 AV *paren_name_list; /* idx -> name */
172 #define RExC_lastparse (pRExC_state->lastparse)
173 #define RExC_lastnum (pRExC_state->lastnum)
174 #define RExC_paren_name_list (pRExC_state->paren_name_list)
178 #define RExC_flags (pRExC_state->flags)
179 #define RExC_pm_flags (pRExC_state->pm_flags)
180 #define RExC_precomp (pRExC_state->precomp)
181 #define RExC_rx_sv (pRExC_state->rx_sv)
182 #define RExC_rx (pRExC_state->rx)
183 #define RExC_rxi (pRExC_state->rxi)
184 #define RExC_start (pRExC_state->start)
185 #define RExC_end (pRExC_state->end)
186 #define RExC_parse (pRExC_state->parse)
187 #define RExC_whilem_seen (pRExC_state->whilem_seen)
188 #ifdef RE_TRACK_PATTERN_OFFSETS
189 #define RExC_offsets (pRExC_state->rxi->u.offsets) /* I am not like the others */
191 #define RExC_emit (pRExC_state->emit)
192 #define RExC_emit_dummy (pRExC_state->emit_dummy)
193 #define RExC_emit_start (pRExC_state->emit_start)
194 #define RExC_emit_bound (pRExC_state->emit_bound)
195 #define RExC_naughty (pRExC_state->naughty)
196 #define RExC_sawback (pRExC_state->sawback)
197 #define RExC_seen (pRExC_state->seen)
198 #define RExC_size (pRExC_state->size)
199 #define RExC_npar (pRExC_state->npar)
200 #define RExC_nestroot (pRExC_state->nestroot)
201 #define RExC_extralen (pRExC_state->extralen)
202 #define RExC_seen_zerolen (pRExC_state->seen_zerolen)
203 #define RExC_utf8 (pRExC_state->utf8)
204 #define RExC_uni_semantics (pRExC_state->uni_semantics)
205 #define RExC_orig_utf8 (pRExC_state->orig_utf8)
206 #define RExC_open_parens (pRExC_state->open_parens)
207 #define RExC_close_parens (pRExC_state->close_parens)
208 #define RExC_opend (pRExC_state->opend)
209 #define RExC_paren_names (pRExC_state->paren_names)
210 #define RExC_recurse (pRExC_state->recurse)
211 #define RExC_recurse_count (pRExC_state->recurse_count)
212 #define RExC_in_lookbehind (pRExC_state->in_lookbehind)
213 #define RExC_contains_locale (pRExC_state->contains_locale)
214 #define RExC_override_recoding (pRExC_state->override_recoding)
215 #define RExC_in_multi_char_class (pRExC_state->in_multi_char_class)
218 #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
219 #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
220 ((*s) == '{' && regcurly(s, FALSE)))
223 #undef SPSTART /* dratted cpp namespace... */
226 * Flags to be passed up and down.
228 #define WORST 0 /* Worst case. */
229 #define HASWIDTH 0x01 /* Known to match non-null strings. */
231 /* Simple enough to be STAR/PLUS operand; in an EXACTish node must be a single
232 * character. (There needs to be a case: in the switch statement in regexec.c
233 * for any node marked SIMPLE.) Note that this is not the same thing as
236 #define SPSTART 0x04 /* Starts with * or + */
237 #define POSTPONED 0x08 /* (?1),(?&name), (??{...}) or similar */
238 #define TRYAGAIN 0x10 /* Weeded out a declaration. */
239 #define RESTART_UTF8 0x20 /* Restart, need to calcuate sizes as UTF-8 */
241 #define REG_NODE_NUM(x) ((x) ? (int)((x)-RExC_emit_start) : -1)
243 /* whether trie related optimizations are enabled */
244 #if PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION
245 #define TRIE_STUDY_OPT
246 #define FULL_TRIE_STUDY
252 #define PBYTE(u8str,paren) ((U8*)(u8str))[(paren) >> 3]
253 #define PBITVAL(paren) (1 << ((paren) & 7))
254 #define PAREN_TEST(u8str,paren) ( PBYTE(u8str,paren) & PBITVAL(paren))
255 #define PAREN_SET(u8str,paren) PBYTE(u8str,paren) |= PBITVAL(paren)
256 #define PAREN_UNSET(u8str,paren) PBYTE(u8str,paren) &= (~PBITVAL(paren))
258 #define REQUIRE_UTF8 STMT_START { \
260 *flagp = RESTART_UTF8; \
265 /* This converts the named class defined in regcomp.h to its equivalent class
266 * number defined in handy.h. */
267 #define namedclass_to_classnum(class) ((int) ((class) / 2))
268 #define classnum_to_namedclass(classnum) ((classnum) * 2)
270 /* About scan_data_t.
272 During optimisation we recurse through the regexp program performing
273 various inplace (keyhole style) optimisations. In addition study_chunk
274 and scan_commit populate this data structure with information about
275 what strings MUST appear in the pattern. We look for the longest
276 string that must appear at a fixed location, and we look for the
277 longest string that may appear at a floating location. So for instance
282 Both 'FOO' and 'A' are fixed strings. Both 'B' and 'BAR' are floating
283 strings (because they follow a .* construct). study_chunk will identify
284 both FOO and BAR as being the longest fixed and floating strings respectively.
286 The strings can be composites, for instance
290 will result in a composite fixed substring 'foo'.
292 For each string some basic information is maintained:
294 - offset or min_offset
295 This is the position the string must appear at, or not before.
296 It also implicitly (when combined with minlenp) tells us how many
297 characters must match before the string we are searching for.
298 Likewise when combined with minlenp and the length of the string it
299 tells us how many characters must appear after the string we have
303 Only used for floating strings. This is the rightmost point that
304 the string can appear at. If set to I32 max it indicates that the
305 string can occur infinitely far to the right.
308 A pointer to the minimum number of characters of the pattern that the
309 string was found inside. This is important as in the case of positive
310 lookahead or positive lookbehind we can have multiple patterns
315 The minimum length of the pattern overall is 3, the minimum length
316 of the lookahead part is 3, but the minimum length of the part that
317 will actually match is 1. So 'FOO's minimum length is 3, but the
318 minimum length for the F is 1. This is important as the minimum length
319 is used to determine offsets in front of and behind the string being
320 looked for. Since strings can be composites this is the length of the
321 pattern at the time it was committed with a scan_commit. Note that
322 the length is calculated by study_chunk, so that the minimum lengths
323 are not known until the full pattern has been compiled, thus the
324 pointer to the value.
328 In the case of lookbehind the string being searched for can be
329 offset past the start point of the final matching string.
330 If this value was just blithely removed from the min_offset it would
331 invalidate some of the calculations for how many chars must match
332 before or after (as they are derived from min_offset and minlen and
333 the length of the string being searched for).
334 When the final pattern is compiled and the data is moved from the
335 scan_data_t structure into the regexp structure the information
336 about lookbehind is factored in, with the information that would
337 have been lost precalculated in the end_shift field for the
340 The fields pos_min and pos_delta are used to store the minimum offset
341 and the delta to the maximum offset at the current point in the pattern.
345 typedef struct scan_data_t {
346 /*I32 len_min; unused */
347 /*I32 len_delta; unused */
351 I32 last_end; /* min value, <0 unless valid. */
354 SV **longest; /* Either &l_fixed, or &l_float. */
355 SV *longest_fixed; /* longest fixed string found in pattern */
356 I32 offset_fixed; /* offset where it starts */
357 I32 *minlen_fixed; /* pointer to the minlen relevant to the string */
358 I32 lookbehind_fixed; /* is the position of the string modfied by LB */
359 SV *longest_float; /* longest floating string found in pattern */
360 I32 offset_float_min; /* earliest point in string it can appear */
361 I32 offset_float_max; /* latest point in string it can appear */
362 I32 *minlen_float; /* pointer to the minlen relevant to the string */
363 I32 lookbehind_float; /* is the position of the string modified by LB */
367 struct regnode_charclass_class *start_class;
370 /* The below is perhaps overboard, but this allows us to save a test at the
371 * expense of a mask. This is because on both EBCDIC and ASCII machines, 'A'
372 * and 'a' differ by a single bit; the same with the upper and lower case of
373 * all other ASCII-range alphabetics. On ASCII platforms, they are 32 apart;
374 * on EBCDIC, they are 64. This uses an exclusive 'or' to find that bit and
375 * then inverts it to form a mask, with just a single 0, in the bit position
376 * where the upper- and lowercase differ. XXX There are about 40 other
377 * instances in the Perl core where this micro-optimization could be used.
378 * Should decide if maintenance cost is worse, before changing those
380 * Returns a boolean as to whether or not 'v' is either a lowercase or
381 * uppercase instance of 'c', where 'c' is in [A-Za-z]. If 'c' is a
382 * compile-time constant, the generated code is better than some optimizing
383 * compilers figure out, amounting to a mask and test. The results are
384 * meaningless if 'c' is not one of [A-Za-z] */
385 #define isARG2_lower_or_UPPER_ARG1(c, v) \
386 (((v) & ~('A' ^ 'a')) == ((c) & ~('A' ^ 'a')))
389 * Forward declarations for pregcomp()'s friends.
392 static const scan_data_t zero_scan_data =
393 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ,0};
395 #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
396 #define SF_BEFORE_SEOL 0x0001
397 #define SF_BEFORE_MEOL 0x0002
398 #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
399 #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
402 # define SF_FIX_SHIFT_EOL (0+2)
403 # define SF_FL_SHIFT_EOL (0+4)
405 # define SF_FIX_SHIFT_EOL (+2)
406 # define SF_FL_SHIFT_EOL (+4)
409 #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
410 #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
412 #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
413 #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
414 #define SF_IS_INF 0x0040
415 #define SF_HAS_PAR 0x0080
416 #define SF_IN_PAR 0x0100
417 #define SF_HAS_EVAL 0x0200
418 #define SCF_DO_SUBSTR 0x0400
419 #define SCF_DO_STCLASS_AND 0x0800
420 #define SCF_DO_STCLASS_OR 0x1000
421 #define SCF_DO_STCLASS (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
422 #define SCF_WHILEM_VISITED_POS 0x2000
424 #define SCF_TRIE_RESTUDY 0x4000 /* Do restudy? */
425 #define SCF_SEEN_ACCEPT 0x8000
426 #define SCF_TRIE_DOING_RESTUDY 0x10000
428 #define UTF cBOOL(RExC_utf8)
430 /* The enums for all these are ordered so things work out correctly */
431 #define LOC (get_regex_charset(RExC_flags) == REGEX_LOCALE_CHARSET)
432 #define DEPENDS_SEMANTICS (get_regex_charset(RExC_flags) == REGEX_DEPENDS_CHARSET)
433 #define UNI_SEMANTICS (get_regex_charset(RExC_flags) == REGEX_UNICODE_CHARSET)
434 #define AT_LEAST_UNI_SEMANTICS (get_regex_charset(RExC_flags) >= REGEX_UNICODE_CHARSET)
435 #define ASCII_RESTRICTED (get_regex_charset(RExC_flags) == REGEX_ASCII_RESTRICTED_CHARSET)
436 #define AT_LEAST_ASCII_RESTRICTED (get_regex_charset(RExC_flags) >= REGEX_ASCII_RESTRICTED_CHARSET)
437 #define ASCII_FOLD_RESTRICTED (get_regex_charset(RExC_flags) == REGEX_ASCII_MORE_RESTRICTED_CHARSET)
439 #define FOLD cBOOL(RExC_flags & RXf_PMf_FOLD)
441 #define OOB_NAMEDCLASS -1
443 /* There is no code point that is out-of-bounds, so this is problematic. But
444 * its only current use is to initialize a variable that is always set before
446 #define OOB_UNICODE 0xDEADBEEF
448 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
449 #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
452 /* length of regex to show in messages that don't mark a position within */
453 #define RegexLengthToShowInErrorMessages 127
456 * If MARKER[12] are adjusted, be sure to adjust the constants at the top
457 * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
458 * op/pragma/warn/regcomp.
460 #define MARKER1 "<-- HERE" /* marker as it appears in the description */
461 #define MARKER2 " <-- HERE " /* marker as it appears within the regex */
463 #define REPORT_LOCATION " in regex; marked by " MARKER1 " in m/%.*s" MARKER2 "%s/"
466 * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
467 * arg. Show regex, up to a maximum length. If it's too long, chop and add
470 #define _FAIL(code) STMT_START { \
471 const char *ellipses = ""; \
472 IV len = RExC_end - RExC_precomp; \
475 SAVEFREESV(RExC_rx_sv); \
476 if (len > RegexLengthToShowInErrorMessages) { \
477 /* chop 10 shorter than the max, to ensure meaning of "..." */ \
478 len = RegexLengthToShowInErrorMessages - 10; \
484 #define FAIL(msg) _FAIL( \
485 Perl_croak(aTHX_ "%s in regex m/%.*s%s/", \
486 msg, (int)len, RExC_precomp, ellipses))
488 #define FAIL2(msg,arg) _FAIL( \
489 Perl_croak(aTHX_ msg " in regex m/%.*s%s/", \
490 arg, (int)len, RExC_precomp, ellipses))
493 * Simple_vFAIL -- like FAIL, but marks the current location in the scan
495 #define Simple_vFAIL(m) STMT_START { \
496 const IV offset = RExC_parse - RExC_precomp; \
497 Perl_croak(aTHX_ "%s" REPORT_LOCATION, \
498 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
502 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
504 #define vFAIL(m) STMT_START { \
506 SAVEFREESV(RExC_rx_sv); \
511 * Like Simple_vFAIL(), but accepts two arguments.
513 #define Simple_vFAIL2(m,a1) STMT_START { \
514 const IV offset = RExC_parse - RExC_precomp; \
515 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, \
516 (int)offset, RExC_precomp, RExC_precomp + offset); \
520 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
522 #define vFAIL2(m,a1) STMT_START { \
524 SAVEFREESV(RExC_rx_sv); \
525 Simple_vFAIL2(m, a1); \
530 * Like Simple_vFAIL(), but accepts three arguments.
532 #define Simple_vFAIL3(m, a1, a2) STMT_START { \
533 const IV offset = RExC_parse - RExC_precomp; \
534 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, \
535 (int)offset, RExC_precomp, RExC_precomp + offset); \
539 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
541 #define vFAIL3(m,a1,a2) STMT_START { \
543 SAVEFREESV(RExC_rx_sv); \
544 Simple_vFAIL3(m, a1, a2); \
548 * Like Simple_vFAIL(), but accepts four arguments.
550 #define Simple_vFAIL4(m, a1, a2, a3) STMT_START { \
551 const IV offset = RExC_parse - RExC_precomp; \
552 S_re_croak2(aTHX_ m, REPORT_LOCATION, a1, a2, a3, \
553 (int)offset, RExC_precomp, RExC_precomp + offset); \
556 #define vFAIL4(m,a1,a2,a3) STMT_START { \
558 SAVEFREESV(RExC_rx_sv); \
559 Simple_vFAIL4(m, a1, a2, a3); \
562 /* m is not necessarily a "literal string", in this macro */
563 #define reg_warn_non_literal_string(loc, m) STMT_START { \
564 const IV offset = loc - RExC_precomp; \
565 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s" REPORT_LOCATION, \
566 m, (int)offset, RExC_precomp, RExC_precomp + offset); \
569 #define ckWARNreg(loc,m) STMT_START { \
570 const IV offset = loc - RExC_precomp; \
571 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
572 (int)offset, RExC_precomp, RExC_precomp + offset); \
575 #define vWARN_dep(loc, m) STMT_START { \
576 const IV offset = loc - RExC_precomp; \
577 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), m REPORT_LOCATION, \
578 (int)offset, RExC_precomp, RExC_precomp + offset); \
581 #define ckWARNdep(loc,m) STMT_START { \
582 const IV offset = loc - RExC_precomp; \
583 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
585 (int)offset, RExC_precomp, RExC_precomp + offset); \
588 #define ckWARNregdep(loc,m) STMT_START { \
589 const IV offset = loc - RExC_precomp; \
590 Perl_ck_warner_d(aTHX_ packWARN2(WARN_DEPRECATED, WARN_REGEXP), \
592 (int)offset, RExC_precomp, RExC_precomp + offset); \
595 #define ckWARN2reg_d(loc,m, a1) STMT_START { \
596 const IV offset = loc - RExC_precomp; \
597 Perl_ck_warner_d(aTHX_ packWARN(WARN_REGEXP), \
599 a1, (int)offset, RExC_precomp, RExC_precomp + offset); \
602 #define ckWARN2reg(loc, m, a1) STMT_START { \
603 const IV offset = loc - RExC_precomp; \
604 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
605 a1, (int)offset, RExC_precomp, RExC_precomp + offset); \
608 #define vWARN3(loc, m, a1, a2) STMT_START { \
609 const IV offset = loc - RExC_precomp; \
610 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
611 a1, a2, (int)offset, RExC_precomp, RExC_precomp + offset); \
614 #define ckWARN3reg(loc, m, a1, a2) STMT_START { \
615 const IV offset = loc - RExC_precomp; \
616 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
617 a1, a2, (int)offset, RExC_precomp, RExC_precomp + offset); \
620 #define vWARN4(loc, m, a1, a2, a3) STMT_START { \
621 const IV offset = loc - RExC_precomp; \
622 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
623 a1, a2, a3, (int)offset, RExC_precomp, RExC_precomp + offset); \
626 #define ckWARN4reg(loc, m, a1, a2, a3) STMT_START { \
627 const IV offset = loc - RExC_precomp; \
628 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
629 a1, a2, a3, (int)offset, RExC_precomp, RExC_precomp + offset); \
632 #define vWARN5(loc, m, a1, a2, a3, a4) STMT_START { \
633 const IV offset = loc - RExC_precomp; \
634 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
635 a1, a2, a3, a4, (int)offset, RExC_precomp, RExC_precomp + offset); \
639 /* Allow for side effects in s */
640 #define REGC(c,s) STMT_START { \
641 if (!SIZE_ONLY) *(s) = (c); else (void)(s); \
644 /* Macros for recording node offsets. 20001227 mjd@plover.com
645 * Nodes are numbered 1, 2, 3, 4. Node #n's position is recorded in
646 * element 2*n-1 of the array. Element #2n holds the byte length node #n.
647 * Element 0 holds the number n.
648 * Position is 1 indexed.
650 #ifndef RE_TRACK_PATTERN_OFFSETS
651 #define Set_Node_Offset_To_R(node,byte)
652 #define Set_Node_Offset(node,byte)
653 #define Set_Cur_Node_Offset
654 #define Set_Node_Length_To_R(node,len)
655 #define Set_Node_Length(node,len)
656 #define Set_Node_Cur_Length(node,start)
657 #define Node_Offset(n)
658 #define Node_Length(n)
659 #define Set_Node_Offset_Length(node,offset,len)
660 #define ProgLen(ri) ri->u.proglen
661 #define SetProgLen(ri,x) ri->u.proglen = x
663 #define ProgLen(ri) ri->u.offsets[0]
664 #define SetProgLen(ri,x) ri->u.offsets[0] = x
665 #define Set_Node_Offset_To_R(node,byte) STMT_START { \
667 MJD_OFFSET_DEBUG(("** (%d) offset of node %d is %d.\n", \
668 __LINE__, (int)(node), (int)(byte))); \
670 Perl_croak(aTHX_ "value of node is %d in Offset macro", (int)(node)); \
672 RExC_offsets[2*(node)-1] = (byte); \
677 #define Set_Node_Offset(node,byte) \
678 Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
679 #define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
681 #define Set_Node_Length_To_R(node,len) STMT_START { \
683 MJD_OFFSET_DEBUG(("** (%d) size of node %d is %d.\n", \
684 __LINE__, (int)(node), (int)(len))); \
686 Perl_croak(aTHX_ "value of node is %d in Length macro", (int)(node)); \
688 RExC_offsets[2*(node)] = (len); \
693 #define Set_Node_Length(node,len) \
694 Set_Node_Length_To_R((node)-RExC_emit_start, len)
695 #define Set_Node_Cur_Length(node, start) \
696 Set_Node_Length(node, RExC_parse - start)
698 /* Get offsets and lengths */
699 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
700 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
702 #define Set_Node_Offset_Length(node,offset,len) STMT_START { \
703 Set_Node_Offset_To_R((node)-RExC_emit_start, (offset)); \
704 Set_Node_Length_To_R((node)-RExC_emit_start, (len)); \
708 #if PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS
709 #define EXPERIMENTAL_INPLACESCAN
710 #endif /*PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS*/
712 #define DEBUG_STUDYDATA(str,data,depth) \
713 DEBUG_OPTIMISE_MORE_r(if(data){ \
714 PerlIO_printf(Perl_debug_log, \
715 "%*s" str "Pos:%"IVdf"/%"IVdf \
716 " Flags: 0x%"UVXf" Whilem_c: %"IVdf" Lcp: %"IVdf" %s", \
717 (int)(depth)*2, "", \
718 (IV)((data)->pos_min), \
719 (IV)((data)->pos_delta), \
720 (UV)((data)->flags), \
721 (IV)((data)->whilem_c), \
722 (IV)((data)->last_closep ? *((data)->last_closep) : -1), \
723 is_inf ? "INF " : "" \
725 if ((data)->last_found) \
726 PerlIO_printf(Perl_debug_log, \
727 "Last:'%s' %"IVdf":%"IVdf"/%"IVdf" %sFixed:'%s' @ %"IVdf \
728 " %sFloat: '%s' @ %"IVdf"/%"IVdf"", \
729 SvPVX_const((data)->last_found), \
730 (IV)((data)->last_end), \
731 (IV)((data)->last_start_min), \
732 (IV)((data)->last_start_max), \
733 ((data)->longest && \
734 (data)->longest==&((data)->longest_fixed)) ? "*" : "", \
735 SvPVX_const((data)->longest_fixed), \
736 (IV)((data)->offset_fixed), \
737 ((data)->longest && \
738 (data)->longest==&((data)->longest_float)) ? "*" : "", \
739 SvPVX_const((data)->longest_float), \
740 (IV)((data)->offset_float_min), \
741 (IV)((data)->offset_float_max) \
743 PerlIO_printf(Perl_debug_log,"\n"); \
746 /* Mark that we cannot extend a found fixed substring at this point.
747 Update the longest found anchored substring and the longest found
748 floating substrings if needed. */
751 S_scan_commit(pTHX_ const RExC_state_t *pRExC_state, scan_data_t *data, I32 *minlenp, int is_inf)
753 const STRLEN l = CHR_SVLEN(data->last_found);
754 const STRLEN old_l = CHR_SVLEN(*data->longest);
755 GET_RE_DEBUG_FLAGS_DECL;
757 PERL_ARGS_ASSERT_SCAN_COMMIT;
759 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
760 SvSetMagicSV(*data->longest, data->last_found);
761 if (*data->longest == data->longest_fixed) {
762 data->offset_fixed = l ? data->last_start_min : data->pos_min;
763 if (data->flags & SF_BEFORE_EOL)
765 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
767 data->flags &= ~SF_FIX_BEFORE_EOL;
768 data->minlen_fixed=minlenp;
769 data->lookbehind_fixed=0;
771 else { /* *data->longest == data->longest_float */
772 data->offset_float_min = l ? data->last_start_min : data->pos_min;
773 data->offset_float_max = (l
774 ? data->last_start_max
775 : (data->pos_delta == I32_MAX ? I32_MAX : data->pos_min + data->pos_delta));
776 if (is_inf || (U32)data->offset_float_max > (U32)I32_MAX)
777 data->offset_float_max = I32_MAX;
778 if (data->flags & SF_BEFORE_EOL)
780 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
782 data->flags &= ~SF_FL_BEFORE_EOL;
783 data->minlen_float=minlenp;
784 data->lookbehind_float=0;
787 SvCUR_set(data->last_found, 0);
789 SV * const sv = data->last_found;
790 if (SvUTF8(sv) && SvMAGICAL(sv)) {
791 MAGIC * const mg = mg_find(sv, PERL_MAGIC_utf8);
797 data->flags &= ~SF_BEFORE_EOL;
798 DEBUG_STUDYDATA("commit: ",data,0);
801 /* These macros set, clear and test whether the synthetic start class ('ssc',
802 * given by the parameter) matches an empty string (EOS). This uses the
803 * 'next_off' field in the node, to save a bit in the flags field. The ssc
804 * stands alone, so there is never a next_off, so this field is otherwise
805 * unused. The EOS information is used only for compilation, but theoretically
806 * it could be passed on to the execution code. This could be used to store
807 * more than one bit of information, but only this one is currently used. */
808 #define SET_SSC_EOS(node) STMT_START { (node)->next_off = TRUE; } STMT_END
809 #define CLEAR_SSC_EOS(node) STMT_START { (node)->next_off = FALSE; } STMT_END
810 #define TEST_SSC_EOS(node) cBOOL((node)->next_off)
812 /* Can match anything (initialization) */
814 S_cl_anything(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
816 PERL_ARGS_ASSERT_CL_ANYTHING;
818 ANYOF_BITMAP_SETALL(cl);
819 cl->flags = ANYOF_UNICODE_ALL;
822 /* If any portion of the regex is to operate under locale rules,
823 * initialization includes it. The reason this isn't done for all regexes
824 * is that the optimizer was written under the assumption that locale was
825 * all-or-nothing. Given the complexity and lack of documentation in the
826 * optimizer, and that there are inadequate test cases for locale, so many
827 * parts of it may not work properly, it is safest to avoid locale unless
829 if (RExC_contains_locale) {
830 ANYOF_CLASS_SETALL(cl); /* /l uses class */
831 cl->flags |= ANYOF_LOCALE|ANYOF_CLASS|ANYOF_LOC_FOLD;
834 ANYOF_CLASS_ZERO(cl); /* Only /l uses class now */
838 /* Can match anything (initialization) */
840 S_cl_is_anything(const struct regnode_charclass_class *cl)
844 PERL_ARGS_ASSERT_CL_IS_ANYTHING;
846 for (value = 0; value < ANYOF_MAX; value += 2)
847 if (ANYOF_CLASS_TEST(cl, value) && ANYOF_CLASS_TEST(cl, value + 1))
849 if (!(cl->flags & ANYOF_UNICODE_ALL))
851 if (!ANYOF_BITMAP_TESTALLSET((const void*)cl))
856 /* Can match anything (initialization) */
858 S_cl_init(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl)
860 PERL_ARGS_ASSERT_CL_INIT;
862 Zero(cl, 1, struct regnode_charclass_class);
864 cl_anything(pRExC_state, cl);
865 ARG_SET(cl, ANYOF_NONBITMAP_EMPTY);
868 /* These two functions currently do the exact same thing */
869 #define cl_init_zero cl_init
871 /* 'AND' a given class with another one. Can create false positives. 'cl'
872 * should not be inverted. 'and_with->flags & ANYOF_CLASS' should be 0 if
873 * 'and_with' is a regnode_charclass instead of a regnode_charclass_class. */
875 S_cl_and(struct regnode_charclass_class *cl,
876 const struct regnode_charclass_class *and_with)
878 PERL_ARGS_ASSERT_CL_AND;
880 assert(PL_regkind[and_with->type] == ANYOF);
882 /* I (khw) am not sure all these restrictions are necessary XXX */
883 if (!(ANYOF_CLASS_TEST_ANY_SET(and_with))
884 && !(ANYOF_CLASS_TEST_ANY_SET(cl))
885 && (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
886 && !(and_with->flags & ANYOF_LOC_FOLD)
887 && !(cl->flags & ANYOF_LOC_FOLD)) {
890 if (and_with->flags & ANYOF_INVERT)
891 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
892 cl->bitmap[i] &= ~and_with->bitmap[i];
894 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
895 cl->bitmap[i] &= and_with->bitmap[i];
896 } /* XXXX: logic is complicated otherwise, leave it along for a moment. */
898 if (and_with->flags & ANYOF_INVERT) {
900 /* Here, the and'ed node is inverted. Get the AND of the flags that
901 * aren't affected by the inversion. Those that are affected are
902 * handled individually below */
903 U8 affected_flags = cl->flags & ~INVERSION_UNAFFECTED_FLAGS;
904 cl->flags &= (and_with->flags & INVERSION_UNAFFECTED_FLAGS);
905 cl->flags |= affected_flags;
907 /* We currently don't know how to deal with things that aren't in the
908 * bitmap, but we know that the intersection is no greater than what
909 * is already in cl, so let there be false positives that get sorted
910 * out after the synthetic start class succeeds, and the node is
911 * matched for real. */
913 /* The inversion of these two flags indicate that the resulting
914 * intersection doesn't have them */
915 if (and_with->flags & ANYOF_UNICODE_ALL) {
916 cl->flags &= ~ANYOF_UNICODE_ALL;
918 if (and_with->flags & ANYOF_NON_UTF8_LATIN1_ALL) {
919 cl->flags &= ~ANYOF_NON_UTF8_LATIN1_ALL;
922 else { /* and'd node is not inverted */
923 U8 outside_bitmap_but_not_utf8; /* Temp variable */
925 if (! ANYOF_NONBITMAP(and_with)) {
927 /* Here 'and_with' doesn't match anything outside the bitmap
928 * (except possibly ANYOF_UNICODE_ALL), which means the
929 * intersection can't either, except for ANYOF_UNICODE_ALL, in
930 * which case we don't know what the intersection is, but it's no
931 * greater than what cl already has, so can just leave it alone,
932 * with possible false positives */
933 if (! (and_with->flags & ANYOF_UNICODE_ALL)) {
934 ARG_SET(cl, ANYOF_NONBITMAP_EMPTY);
935 cl->flags &= ~ANYOF_NONBITMAP_NON_UTF8;
938 else if (! ANYOF_NONBITMAP(cl)) {
940 /* Here, 'and_with' does match something outside the bitmap, and cl
941 * doesn't have a list of things to match outside the bitmap. If
942 * cl can match all code points above 255, the intersection will
943 * be those above-255 code points that 'and_with' matches. If cl
944 * can't match all Unicode code points, it means that it can't
945 * match anything outside the bitmap (since the 'if' that got us
946 * into this block tested for that), so we leave the bitmap empty.
948 if (cl->flags & ANYOF_UNICODE_ALL) {
949 ARG_SET(cl, ARG(and_with));
951 /* and_with's ARG may match things that don't require UTF8.
952 * And now cl's will too, in spite of this being an 'and'. See
953 * the comments below about the kludge */
954 cl->flags |= and_with->flags & ANYOF_NONBITMAP_NON_UTF8;
958 /* Here, both 'and_with' and cl match something outside the
959 * bitmap. Currently we do not do the intersection, so just match
960 * whatever cl had at the beginning. */
964 /* Take the intersection of the two sets of flags. However, the
965 * ANYOF_NONBITMAP_NON_UTF8 flag is treated as an 'or'. This is a
966 * kludge around the fact that this flag is not treated like the others
967 * which are initialized in cl_anything(). The way the optimizer works
968 * is that the synthetic start class (SSC) is initialized to match
969 * anything, and then the first time a real node is encountered, its
970 * values are AND'd with the SSC's with the result being the values of
971 * the real node. However, there are paths through the optimizer where
972 * the AND never gets called, so those initialized bits are set
973 * inappropriately, which is not usually a big deal, as they just cause
974 * false positives in the SSC, which will just mean a probably
975 * imperceptible slow down in execution. However this bit has a
976 * higher false positive consequence in that it can cause utf8.pm,
977 * utf8_heavy.pl ... to be loaded when not necessary, which is a much
978 * bigger slowdown and also causes significant extra memory to be used.
979 * In order to prevent this, the code now takes a different tack. The
980 * bit isn't set unless some part of the regular expression needs it,
981 * but once set it won't get cleared. This means that these extra
982 * modules won't get loaded unless there was some path through the
983 * pattern that would have required them anyway, and so any false
984 * positives that occur by not ANDing them out when they could be
985 * aren't as severe as they would be if we treated this bit like all
987 outside_bitmap_but_not_utf8 = (cl->flags | and_with->flags)
988 & ANYOF_NONBITMAP_NON_UTF8;
989 cl->flags &= and_with->flags;
990 cl->flags |= outside_bitmap_but_not_utf8;
994 /* 'OR' a given class with another one. Can create false positives. 'cl'
995 * should not be inverted. 'or_with->flags & ANYOF_CLASS' should be 0 if
996 * 'or_with' is a regnode_charclass instead of a regnode_charclass_class. */
998 S_cl_or(const RExC_state_t *pRExC_state, struct regnode_charclass_class *cl, const struct regnode_charclass_class *or_with)
1000 PERL_ARGS_ASSERT_CL_OR;
1002 if (or_with->flags & ANYOF_INVERT) {
1004 /* Here, the or'd node is to be inverted. This means we take the
1005 * complement of everything not in the bitmap, but currently we don't
1006 * know what that is, so give up and match anything */
1007 if (ANYOF_NONBITMAP(or_with)) {
1008 cl_anything(pRExC_state, cl);
1011 * (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
1012 * <= (B1 | !B2) | (CL1 | !CL2)
1013 * which is wasteful if CL2 is small, but we ignore CL2:
1014 * (B1 | CL1) | (!B2 & !CL2) <= (B1 | CL1) | !B2 = (B1 | !B2) | CL1
1015 * XXXX Can we handle case-fold? Unclear:
1016 * (OK1(i) | OK1(i')) | !(OK1(i) | OK1(i')) =
1017 * (OK1(i) | OK1(i')) | (!OK1(i) & !OK1(i'))
1019 else if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
1020 && !(or_with->flags & ANYOF_LOC_FOLD)
1021 && !(cl->flags & ANYOF_LOC_FOLD) ) {
1024 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
1025 cl->bitmap[i] |= ~or_with->bitmap[i];
1026 } /* XXXX: logic is complicated otherwise */
1028 cl_anything(pRExC_state, cl);
1031 /* And, we can just take the union of the flags that aren't affected
1032 * by the inversion */
1033 cl->flags |= or_with->flags & INVERSION_UNAFFECTED_FLAGS;
1035 /* For the remaining flags:
1036 ANYOF_UNICODE_ALL and inverted means to not match anything above
1037 255, which means that the union with cl should just be
1038 what cl has in it, so can ignore this flag
1039 ANYOF_NON_UTF8_LATIN1_ALL and inverted means if not utf8 and ord
1040 is 127-255 to match them, but then invert that, so the
1041 union with cl should just be what cl has in it, so can
1044 } else { /* 'or_with' is not inverted */
1045 /* (B1 | CL1) | (B2 | CL2) = (B1 | B2) | (CL1 | CL2)) */
1046 if ( (or_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
1047 && (!(or_with->flags & ANYOF_LOC_FOLD)
1048 || (cl->flags & ANYOF_LOC_FOLD)) ) {
1051 /* OR char bitmap and class bitmap separately */
1052 for (i = 0; i < ANYOF_BITMAP_SIZE; i++)
1053 cl->bitmap[i] |= or_with->bitmap[i];
1054 if (or_with->flags & ANYOF_CLASS) {
1055 ANYOF_CLASS_OR(or_with, cl);
1058 else { /* XXXX: logic is complicated, leave it along for a moment. */
1059 cl_anything(pRExC_state, cl);
1062 if (ANYOF_NONBITMAP(or_with)) {
1064 /* Use the added node's outside-the-bit-map match if there isn't a
1065 * conflict. If there is a conflict (both nodes match something
1066 * outside the bitmap, but what they match outside is not the same
1067 * pointer, and hence not easily compared until XXX we extend
1068 * inversion lists this far), give up and allow the start class to
1069 * match everything outside the bitmap. If that stuff is all above
1070 * 255, can just set UNICODE_ALL, otherwise caould be anything. */
1071 if (! ANYOF_NONBITMAP(cl)) {
1072 ARG_SET(cl, ARG(or_with));
1074 else if (ARG(cl) != ARG(or_with)) {
1076 if ((or_with->flags & ANYOF_NONBITMAP_NON_UTF8)) {
1077 cl_anything(pRExC_state, cl);
1080 cl->flags |= ANYOF_UNICODE_ALL;
1085 /* Take the union */
1086 cl->flags |= or_with->flags;
1090 #define TRIE_LIST_ITEM(state,idx) (trie->states[state].trans.list)[ idx ]
1091 #define TRIE_LIST_CUR(state) ( TRIE_LIST_ITEM( state, 0 ).forid )
1092 #define TRIE_LIST_LEN(state) ( TRIE_LIST_ITEM( state, 0 ).newstate )
1093 #define TRIE_LIST_USED(idx) ( trie->states[state].trans.list ? (TRIE_LIST_CUR( idx ) - 1) : 0 )
1098 dump_trie(trie,widecharmap,revcharmap)
1099 dump_trie_interim_list(trie,widecharmap,revcharmap,next_alloc)
1100 dump_trie_interim_table(trie,widecharmap,revcharmap,next_alloc)
1102 These routines dump out a trie in a somewhat readable format.
1103 The _interim_ variants are used for debugging the interim
1104 tables that are used to generate the final compressed
1105 representation which is what dump_trie expects.
1107 Part of the reason for their existence is to provide a form
1108 of documentation as to how the different representations function.
1113 Dumps the final compressed table form of the trie to Perl_debug_log.
1114 Used for debugging make_trie().
1118 S_dump_trie(pTHX_ const struct _reg_trie_data *trie, HV *widecharmap,
1119 AV *revcharmap, U32 depth)
1122 SV *sv=sv_newmortal();
1123 int colwidth= widecharmap ? 6 : 4;
1125 GET_RE_DEBUG_FLAGS_DECL;
1127 PERL_ARGS_ASSERT_DUMP_TRIE;
1129 PerlIO_printf( Perl_debug_log, "%*sChar : %-6s%-6s%-4s ",
1130 (int)depth * 2 + 2,"",
1131 "Match","Base","Ofs" );
1133 for( state = 0 ; state < trie->uniquecharcount ; state++ ) {
1134 SV ** const tmp = av_fetch( revcharmap, state, 0);
1136 PerlIO_printf( Perl_debug_log, "%*s",
1138 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
1139 PL_colors[0], PL_colors[1],
1140 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
1141 PERL_PV_ESCAPE_FIRSTCHAR
1146 PerlIO_printf( Perl_debug_log, "\n%*sState|-----------------------",
1147 (int)depth * 2 + 2,"");
1149 for( state = 0 ; state < trie->uniquecharcount ; state++ )
1150 PerlIO_printf( Perl_debug_log, "%.*s", colwidth, "--------");
1151 PerlIO_printf( Perl_debug_log, "\n");
1153 for( state = 1 ; state < trie->statecount ; state++ ) {
1154 const U32 base = trie->states[ state ].trans.base;
1156 PerlIO_printf( Perl_debug_log, "%*s#%4"UVXf"|", (int)depth * 2 + 2,"", (UV)state);
1158 if ( trie->states[ state ].wordnum ) {
1159 PerlIO_printf( Perl_debug_log, " W%4X", trie->states[ state ].wordnum );
1161 PerlIO_printf( Perl_debug_log, "%6s", "" );
1164 PerlIO_printf( Perl_debug_log, " @%4"UVXf" ", (UV)base );
1169 while( ( base + ofs < trie->uniquecharcount ) ||
1170 ( base + ofs - trie->uniquecharcount < trie->lasttrans
1171 && trie->trans[ base + ofs - trie->uniquecharcount ].check != state))
1174 PerlIO_printf( Perl_debug_log, "+%2"UVXf"[ ", (UV)ofs);
1176 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
1177 if ( ( base + ofs >= trie->uniquecharcount ) &&
1178 ( base + ofs - trie->uniquecharcount < trie->lasttrans ) &&
1179 trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
1181 PerlIO_printf( Perl_debug_log, "%*"UVXf,
1183 (UV)trie->trans[ base + ofs - trie->uniquecharcount ].next );
1185 PerlIO_printf( Perl_debug_log, "%*s",colwidth," ." );
1189 PerlIO_printf( Perl_debug_log, "]");
1192 PerlIO_printf( Perl_debug_log, "\n" );
1194 PerlIO_printf(Perl_debug_log, "%*sword_info N:(prev,len)=", (int)depth*2, "");
1195 for (word=1; word <= trie->wordcount; word++) {
1196 PerlIO_printf(Perl_debug_log, " %d:(%d,%d)",
1197 (int)word, (int)(trie->wordinfo[word].prev),
1198 (int)(trie->wordinfo[word].len));
1200 PerlIO_printf(Perl_debug_log, "\n" );
1203 Dumps a fully constructed but uncompressed trie in list form.
1204 List tries normally only are used for construction when the number of
1205 possible chars (trie->uniquecharcount) is very high.
1206 Used for debugging make_trie().
1209 S_dump_trie_interim_list(pTHX_ const struct _reg_trie_data *trie,
1210 HV *widecharmap, AV *revcharmap, U32 next_alloc,
1214 SV *sv=sv_newmortal();
1215 int colwidth= widecharmap ? 6 : 4;
1216 GET_RE_DEBUG_FLAGS_DECL;
1218 PERL_ARGS_ASSERT_DUMP_TRIE_INTERIM_LIST;
1220 /* print out the table precompression. */
1221 PerlIO_printf( Perl_debug_log, "%*sState :Word | Transition Data\n%*s%s",
1222 (int)depth * 2 + 2,"", (int)depth * 2 + 2,"",
1223 "------:-----+-----------------\n" );
1225 for( state=1 ; state < next_alloc ; state ++ ) {
1228 PerlIO_printf( Perl_debug_log, "%*s %4"UVXf" :",
1229 (int)depth * 2 + 2,"", (UV)state );
1230 if ( ! trie->states[ state ].wordnum ) {
1231 PerlIO_printf( Perl_debug_log, "%5s| ","");
1233 PerlIO_printf( Perl_debug_log, "W%4x| ",
1234 trie->states[ state ].wordnum
1237 for( charid = 1 ; charid <= TRIE_LIST_USED( state ) ; charid++ ) {
1238 SV ** const tmp = av_fetch( revcharmap, TRIE_LIST_ITEM(state,charid).forid, 0);
1240 PerlIO_printf( Perl_debug_log, "%*s:%3X=%4"UVXf" | ",
1242 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
1243 PL_colors[0], PL_colors[1],
1244 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
1245 PERL_PV_ESCAPE_FIRSTCHAR
1247 TRIE_LIST_ITEM(state,charid).forid,
1248 (UV)TRIE_LIST_ITEM(state,charid).newstate
1251 PerlIO_printf(Perl_debug_log, "\n%*s| ",
1252 (int)((depth * 2) + 14), "");
1255 PerlIO_printf( Perl_debug_log, "\n");
1260 Dumps a fully constructed but uncompressed trie in table form.
1261 This is the normal DFA style state transition table, with a few
1262 twists to facilitate compression later.
1263 Used for debugging make_trie().
1266 S_dump_trie_interim_table(pTHX_ const struct _reg_trie_data *trie,
1267 HV *widecharmap, AV *revcharmap, U32 next_alloc,
1272 SV *sv=sv_newmortal();
1273 int colwidth= widecharmap ? 6 : 4;
1274 GET_RE_DEBUG_FLAGS_DECL;
1276 PERL_ARGS_ASSERT_DUMP_TRIE_INTERIM_TABLE;
1279 print out the table precompression so that we can do a visual check
1280 that they are identical.
1283 PerlIO_printf( Perl_debug_log, "%*sChar : ",(int)depth * 2 + 2,"" );
1285 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
1286 SV ** const tmp = av_fetch( revcharmap, charid, 0);
1288 PerlIO_printf( Perl_debug_log, "%*s",
1290 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
1291 PL_colors[0], PL_colors[1],
1292 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
1293 PERL_PV_ESCAPE_FIRSTCHAR
1299 PerlIO_printf( Perl_debug_log, "\n%*sState+-",(int)depth * 2 + 2,"" );
1301 for( charid=0 ; charid < trie->uniquecharcount ; charid++ ) {
1302 PerlIO_printf( Perl_debug_log, "%.*s", colwidth,"--------");
1305 PerlIO_printf( Perl_debug_log, "\n" );
1307 for( state=1 ; state < next_alloc ; state += trie->uniquecharcount ) {
1309 PerlIO_printf( Perl_debug_log, "%*s%4"UVXf" : ",
1310 (int)depth * 2 + 2,"",
1311 (UV)TRIE_NODENUM( state ) );
1313 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
1314 UV v=(UV)SAFE_TRIE_NODENUM( trie->trans[ state + charid ].next );
1316 PerlIO_printf( Perl_debug_log, "%*"UVXf, colwidth, v );
1318 PerlIO_printf( Perl_debug_log, "%*s", colwidth, "." );
1320 if ( ! trie->states[ TRIE_NODENUM( state ) ].wordnum ) {
1321 PerlIO_printf( Perl_debug_log, " (%4"UVXf")\n", (UV)trie->trans[ state ].check );
1323 PerlIO_printf( Perl_debug_log, " (%4"UVXf") W%4X\n", (UV)trie->trans[ state ].check,
1324 trie->states[ TRIE_NODENUM( state ) ].wordnum );
1332 /* make_trie(startbranch,first,last,tail,word_count,flags,depth)
1333 startbranch: the first branch in the whole branch sequence
1334 first : start branch of sequence of branch-exact nodes.
1335 May be the same as startbranch
1336 last : Thing following the last branch.
1337 May be the same as tail.
1338 tail : item following the branch sequence
1339 count : words in the sequence
1340 flags : currently the OP() type we will be building one of /EXACT(|F|Fl)/
1341 depth : indent depth
1343 Inplace optimizes a sequence of 2 or more Branch-Exact nodes into a TRIE node.
1345 A trie is an N'ary tree where the branches are determined by digital
1346 decomposition of the key. IE, at the root node you look up the 1st character and
1347 follow that branch repeat until you find the end of the branches. Nodes can be
1348 marked as "accepting" meaning they represent a complete word. Eg:
1352 would convert into the following structure. Numbers represent states, letters
1353 following numbers represent valid transitions on the letter from that state, if
1354 the number is in square brackets it represents an accepting state, otherwise it
1355 will be in parenthesis.
1357 +-h->+-e->[3]-+-r->(8)-+-s->[9]
1361 (1) +-i->(6)-+-s->[7]
1363 +-s->(3)-+-h->(4)-+-e->[5]
1365 Accept Word Mapping: 3=>1 (he),5=>2 (she), 7=>3 (his), 9=>4 (hers)
1367 This shows that when matching against the string 'hers' we will begin at state 1
1368 read 'h' and move to state 2, read 'e' and move to state 3 which is accepting,
1369 then read 'r' and go to state 8 followed by 's' which takes us to state 9 which
1370 is also accepting. Thus we know that we can match both 'he' and 'hers' with a
1371 single traverse. We store a mapping from accepting to state to which word was
1372 matched, and then when we have multiple possibilities we try to complete the
1373 rest of the regex in the order in which they occured in the alternation.
1375 The only prior NFA like behaviour that would be changed by the TRIE support is
1376 the silent ignoring of duplicate alternations which are of the form:
1378 / (DUPE|DUPE) X? (?{ ... }) Y /x
1380 Thus EVAL blocks following a trie may be called a different number of times with
1381 and without the optimisation. With the optimisations dupes will be silently
1382 ignored. This inconsistent behaviour of EVAL type nodes is well established as
1383 the following demonstrates:
1385 'words'=~/(word|word|word)(?{ print $1 })[xyz]/
1387 which prints out 'word' three times, but
1389 'words'=~/(word|word|word)(?{ print $1 })S/
1391 which doesnt print it out at all. This is due to other optimisations kicking in.
1393 Example of what happens on a structural level:
1395 The regexp /(ac|ad|ab)+/ will produce the following debug output:
1397 1: CURLYM[1] {1,32767}(18)
1408 This would be optimizable with startbranch=5, first=5, last=16, tail=16
1409 and should turn into:
1411 1: CURLYM[1] {1,32767}(18)
1413 [Words:3 Chars Stored:6 Unique Chars:4 States:5 NCP:1]
1421 Cases where tail != last would be like /(?foo|bar)baz/:
1431 which would be optimizable with startbranch=1, first=1, last=7, tail=8
1432 and would end up looking like:
1435 [Words:2 Chars Stored:6 Unique Chars:5 States:7 NCP:1]
1442 d = uvuni_to_utf8_flags(d, uv, 0);
1444 is the recommended Unicode-aware way of saying
1449 #define TRIE_STORE_REVCHAR(val) \
1452 SV *zlopp = newSV(7); /* XXX: optimize me */ \
1453 unsigned char *flrbbbbb = (unsigned char *) SvPVX(zlopp); \
1454 unsigned const char *const kapow = uvuni_to_utf8(flrbbbbb, val); \
1455 SvCUR_set(zlopp, kapow - flrbbbbb); \
1458 av_push(revcharmap, zlopp); \
1460 char ooooff = (char)val; \
1461 av_push(revcharmap, newSVpvn(&ooooff, 1)); \
1465 #define TRIE_READ_CHAR STMT_START { \
1468 /* if it is UTF then it is either already folded, or does not need folding */ \
1469 uvc = utf8n_to_uvuni( (const U8*) uc, UTF8_MAXLEN, &len, uniflags); \
1471 else if (folder == PL_fold_latin1) { \
1472 /* if we use this folder we have to obey unicode rules on latin-1 data */ \
1473 if ( foldlen > 0 ) { \
1474 uvc = utf8n_to_uvuni( (const U8*) scan, UTF8_MAXLEN, &len, uniflags ); \
1480 uvc = _to_fold_latin1( (U8) *uc, foldbuf, &foldlen, FOLD_FLAGS_FULL); \
1481 skiplen = UNISKIP(uvc); \
1482 foldlen -= skiplen; \
1483 scan = foldbuf + skiplen; \
1486 /* raw data, will be folded later if needed */ \
1494 #define TRIE_LIST_PUSH(state,fid,ns) STMT_START { \
1495 if ( TRIE_LIST_CUR( state ) >=TRIE_LIST_LEN( state ) ) { \
1496 U32 ging = TRIE_LIST_LEN( state ) *= 2; \
1497 Renew( trie->states[ state ].trans.list, ging, reg_trie_trans_le ); \
1499 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).forid = fid; \
1500 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).newstate = ns; \
1501 TRIE_LIST_CUR( state )++; \
1504 #define TRIE_LIST_NEW(state) STMT_START { \
1505 Newxz( trie->states[ state ].trans.list, \
1506 4, reg_trie_trans_le ); \
1507 TRIE_LIST_CUR( state ) = 1; \
1508 TRIE_LIST_LEN( state ) = 4; \
1511 #define TRIE_HANDLE_WORD(state) STMT_START { \
1512 U16 dupe= trie->states[ state ].wordnum; \
1513 regnode * const noper_next = regnext( noper ); \
1516 /* store the word for dumping */ \
1518 if (OP(noper) != NOTHING) \
1519 tmp = newSVpvn_utf8(STRING(noper), STR_LEN(noper), UTF); \
1521 tmp = newSVpvn_utf8( "", 0, UTF ); \
1522 av_push( trie_words, tmp ); \
1526 trie->wordinfo[curword].prev = 0; \
1527 trie->wordinfo[curword].len = wordlen; \
1528 trie->wordinfo[curword].accept = state; \
1530 if ( noper_next < tail ) { \
1532 trie->jump = (U16 *) PerlMemShared_calloc( word_count + 1, sizeof(U16) ); \
1533 trie->jump[curword] = (U16)(noper_next - convert); \
1535 jumper = noper_next; \
1537 nextbranch= regnext(cur); \
1541 /* It's a dupe. Pre-insert into the wordinfo[].prev */\
1542 /* chain, so that when the bits of chain are later */\
1543 /* linked together, the dups appear in the chain */\
1544 trie->wordinfo[curword].prev = trie->wordinfo[dupe].prev; \
1545 trie->wordinfo[dupe].prev = curword; \
1547 /* we haven't inserted this word yet. */ \
1548 trie->states[ state ].wordnum = curword; \
1553 #define TRIE_TRANS_STATE(state,base,ucharcount,charid,special) \
1554 ( ( base + charid >= ucharcount \
1555 && base + charid < ubound \
1556 && state == trie->trans[ base - ucharcount + charid ].check \
1557 && trie->trans[ base - ucharcount + charid ].next ) \
1558 ? trie->trans[ base - ucharcount + charid ].next \
1559 : ( state==1 ? special : 0 ) \
1563 #define MADE_JUMP_TRIE 2
1564 #define MADE_EXACT_TRIE 4
1567 S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch, regnode *first, regnode *last, regnode *tail, U32 word_count, U32 flags, U32 depth)
1570 /* first pass, loop through and scan words */
1571 reg_trie_data *trie;
1572 HV *widecharmap = NULL;
1573 AV *revcharmap = newAV();
1575 const U32 uniflags = UTF8_ALLOW_DEFAULT;
1580 regnode *jumper = NULL;
1581 regnode *nextbranch = NULL;
1582 regnode *convert = NULL;
1583 U32 *prev_states; /* temp array mapping each state to previous one */
1584 /* we just use folder as a flag in utf8 */
1585 const U8 * folder = NULL;
1588 const U32 data_slot = add_data( pRExC_state, 4, "tuuu" );
1589 AV *trie_words = NULL;
1590 /* along with revcharmap, this only used during construction but both are
1591 * useful during debugging so we store them in the struct when debugging.
1594 const U32 data_slot = add_data( pRExC_state, 2, "tu" );
1595 STRLEN trie_charcount=0;
1597 SV *re_trie_maxbuff;
1598 GET_RE_DEBUG_FLAGS_DECL;
1600 PERL_ARGS_ASSERT_MAKE_TRIE;
1602 PERL_UNUSED_ARG(depth);
1609 case EXACTFU_TRICKYFOLD:
1610 case EXACTFU: folder = PL_fold_latin1; break;
1611 case EXACTF: folder = PL_fold; break;
1612 case EXACTFL: folder = PL_fold_locale; break;
1613 default: Perl_croak( aTHX_ "panic! In trie construction, unknown node type %u %s", (unsigned) flags, PL_reg_name[flags] );
1616 trie = (reg_trie_data *) PerlMemShared_calloc( 1, sizeof(reg_trie_data) );
1618 trie->startstate = 1;
1619 trie->wordcount = word_count;
1620 RExC_rxi->data->data[ data_slot ] = (void*)trie;
1621 trie->charmap = (U16 *) PerlMemShared_calloc( 256, sizeof(U16) );
1623 trie->bitmap = (char *) PerlMemShared_calloc( ANYOF_BITMAP_SIZE, 1 );
1624 trie->wordinfo = (reg_trie_wordinfo *) PerlMemShared_calloc(
1625 trie->wordcount+1, sizeof(reg_trie_wordinfo));
1628 trie_words = newAV();
1631 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
1632 if (!SvIOK(re_trie_maxbuff)) {
1633 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
1635 DEBUG_TRIE_COMPILE_r({
1636 PerlIO_printf( Perl_debug_log,
1637 "%*smake_trie start==%d, first==%d, last==%d, tail==%d depth=%d\n",
1638 (int)depth * 2 + 2, "",
1639 REG_NODE_NUM(startbranch),REG_NODE_NUM(first),
1640 REG_NODE_NUM(last), REG_NODE_NUM(tail),
1644 /* Find the node we are going to overwrite */
1645 if ( first == startbranch && OP( last ) != BRANCH ) {
1646 /* whole branch chain */
1649 /* branch sub-chain */
1650 convert = NEXTOPER( first );
1653 /* -- First loop and Setup --
1655 We first traverse the branches and scan each word to determine if it
1656 contains widechars, and how many unique chars there are, this is
1657 important as we have to build a table with at least as many columns as we
1660 We use an array of integers to represent the character codes 0..255
1661 (trie->charmap) and we use a an HV* to store Unicode characters. We use the
1662 native representation of the character value as the key and IV's for the
1665 *TODO* If we keep track of how many times each character is used we can
1666 remap the columns so that the table compression later on is more
1667 efficient in terms of memory by ensuring the most common value is in the
1668 middle and the least common are on the outside. IMO this would be better
1669 than a most to least common mapping as theres a decent chance the most
1670 common letter will share a node with the least common, meaning the node
1671 will not be compressible. With a middle is most common approach the worst
1672 case is when we have the least common nodes twice.
1676 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1677 regnode *noper = NEXTOPER( cur );
1678 const U8 *uc = (U8*)STRING( noper );
1679 const U8 *e = uc + STR_LEN( noper );
1681 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1683 const U8 *scan = (U8*)NULL;
1684 U32 wordlen = 0; /* required init */
1686 bool set_bit = trie->bitmap ? 1 : 0; /*store the first char in the bitmap?*/
1688 if (OP(noper) == NOTHING) {
1689 regnode *noper_next= regnext(noper);
1690 if (noper_next != tail && OP(noper_next) == flags) {
1692 uc= (U8*)STRING(noper);
1693 e= uc + STR_LEN(noper);
1694 trie->minlen= STR_LEN(noper);
1701 if ( set_bit ) { /* bitmap only alloced when !(UTF&&Folding) */
1702 TRIE_BITMAP_SET(trie,*uc); /* store the raw first byte
1703 regardless of encoding */
1704 if (OP( noper ) == EXACTFU_SS) {
1705 /* false positives are ok, so just set this */
1706 TRIE_BITMAP_SET(trie,0xDF);
1709 for ( ; uc < e ; uc += len ) {
1710 TRIE_CHARCOUNT(trie)++;
1715 U8 folded= folder[ (U8) uvc ];
1716 if ( !trie->charmap[ folded ] ) {
1717 trie->charmap[ folded ]=( ++trie->uniquecharcount );
1718 TRIE_STORE_REVCHAR( folded );
1721 if ( !trie->charmap[ uvc ] ) {
1722 trie->charmap[ uvc ]=( ++trie->uniquecharcount );
1723 TRIE_STORE_REVCHAR( uvc );
1726 /* store the codepoint in the bitmap, and its folded
1728 TRIE_BITMAP_SET(trie, uvc);
1730 /* store the folded codepoint */
1731 if ( folder ) TRIE_BITMAP_SET(trie, folder[(U8) uvc ]);
1734 /* store first byte of utf8 representation of
1735 variant codepoints */
1736 if (! UNI_IS_INVARIANT(uvc)) {
1737 TRIE_BITMAP_SET(trie, UTF8_TWO_BYTE_HI(uvc));
1740 set_bit = 0; /* We've done our bit :-) */
1745 widecharmap = newHV();
1747 svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 1 );
1750 Perl_croak( aTHX_ "error creating/fetching widecharmap entry for 0x%"UVXf, uvc );
1752 if ( !SvTRUE( *svpp ) ) {
1753 sv_setiv( *svpp, ++trie->uniquecharcount );
1754 TRIE_STORE_REVCHAR(uvc);
1758 if( cur == first ) {
1759 trie->minlen = chars;
1760 trie->maxlen = chars;
1761 } else if (chars < trie->minlen) {
1762 trie->minlen = chars;
1763 } else if (chars > trie->maxlen) {
1764 trie->maxlen = chars;
1766 if (OP( noper ) == EXACTFU_SS) {
1767 /* XXX: workaround - 'ss' could match "\x{DF}" so minlen could be 1 and not 2*/
1768 if (trie->minlen > 1)
1771 if (OP( noper ) == EXACTFU_TRICKYFOLD) {
1772 /* XXX: workround - things like "\x{1FBE}\x{0308}\x{0301}" can match "\x{0390}"
1773 * - We assume that any such sequence might match a 2 byte string */
1774 if (trie->minlen > 2 )
1778 } /* end first pass */
1779 DEBUG_TRIE_COMPILE_r(
1780 PerlIO_printf( Perl_debug_log, "%*sTRIE(%s): W:%d C:%d Uq:%d Min:%d Max:%d\n",
1781 (int)depth * 2 + 2,"",
1782 ( widecharmap ? "UTF8" : "NATIVE" ), (int)word_count,
1783 (int)TRIE_CHARCOUNT(trie), trie->uniquecharcount,
1784 (int)trie->minlen, (int)trie->maxlen )
1788 We now know what we are dealing with in terms of unique chars and
1789 string sizes so we can calculate how much memory a naive
1790 representation using a flat table will take. If it's over a reasonable
1791 limit (as specified by ${^RE_TRIE_MAXBUF}) we use a more memory
1792 conservative but potentially much slower representation using an array
1795 At the end we convert both representations into the same compressed
1796 form that will be used in regexec.c for matching with. The latter
1797 is a form that cannot be used to construct with but has memory
1798 properties similar to the list form and access properties similar
1799 to the table form making it both suitable for fast searches and
1800 small enough that its feasable to store for the duration of a program.
1802 See the comment in the code where the compressed table is produced
1803 inplace from the flat tabe representation for an explanation of how
1804 the compression works.
1809 Newx(prev_states, TRIE_CHARCOUNT(trie) + 2, U32);
1812 if ( (IV)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1) > SvIV(re_trie_maxbuff) ) {
1814 Second Pass -- Array Of Lists Representation
1816 Each state will be represented by a list of charid:state records
1817 (reg_trie_trans_le) the first such element holds the CUR and LEN
1818 points of the allocated array. (See defines above).
1820 We build the initial structure using the lists, and then convert
1821 it into the compressed table form which allows faster lookups
1822 (but cant be modified once converted).
1825 STRLEN transcount = 1;
1827 DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log,
1828 "%*sCompiling trie using list compiler\n",
1829 (int)depth * 2 + 2, ""));
1831 trie->states = (reg_trie_state *)
1832 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
1833 sizeof(reg_trie_state) );
1837 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
1839 regnode *noper = NEXTOPER( cur );
1840 U8 *uc = (U8*)STRING( noper );
1841 const U8 *e = uc + STR_LEN( noper );
1842 U32 state = 1; /* required init */
1843 U16 charid = 0; /* sanity init */
1844 U8 *scan = (U8*)NULL; /* sanity init */
1845 STRLEN foldlen = 0; /* required init */
1846 U32 wordlen = 0; /* required init */
1847 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
1850 if (OP(noper) == NOTHING) {
1851 regnode *noper_next= regnext(noper);
1852 if (noper_next != tail && OP(noper_next) == flags) {
1854 uc= (U8*)STRING(noper);
1855 e= uc + STR_LEN(noper);
1859 if (OP(noper) != NOTHING) {
1860 for ( ; uc < e ; uc += len ) {
1865 charid = trie->charmap[ uvc ];
1867 SV** const svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 0);
1871 charid=(U16)SvIV( *svpp );
1874 /* charid is now 0 if we dont know the char read, or nonzero if we do */
1881 if ( !trie->states[ state ].trans.list ) {
1882 TRIE_LIST_NEW( state );
1884 for ( check = 1; check <= TRIE_LIST_USED( state ); check++ ) {
1885 if ( TRIE_LIST_ITEM( state, check ).forid == charid ) {
1886 newstate = TRIE_LIST_ITEM( state, check ).newstate;
1891 newstate = next_alloc++;
1892 prev_states[newstate] = state;
1893 TRIE_LIST_PUSH( state, charid, newstate );
1898 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %"IVdf, uvc );
1902 TRIE_HANDLE_WORD(state);
1904 } /* end second pass */
1906 /* next alloc is the NEXT state to be allocated */
1907 trie->statecount = next_alloc;
1908 trie->states = (reg_trie_state *)
1909 PerlMemShared_realloc( trie->states,
1911 * sizeof(reg_trie_state) );
1913 /* and now dump it out before we compress it */
1914 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_list(trie, widecharmap,
1915 revcharmap, next_alloc,
1919 trie->trans = (reg_trie_trans *)
1920 PerlMemShared_calloc( transcount, sizeof(reg_trie_trans) );
1927 for( state=1 ; state < next_alloc ; state ++ ) {
1931 DEBUG_TRIE_COMPILE_MORE_r(
1932 PerlIO_printf( Perl_debug_log, "tp: %d zp: %d ",tp,zp)
1936 if (trie->states[state].trans.list) {
1937 U16 minid=TRIE_LIST_ITEM( state, 1).forid;
1941 for( idx = 2 ; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
1942 const U16 forid = TRIE_LIST_ITEM( state, idx).forid;
1943 if ( forid < minid ) {
1945 } else if ( forid > maxid ) {
1949 if ( transcount < tp + maxid - minid + 1) {
1951 trie->trans = (reg_trie_trans *)
1952 PerlMemShared_realloc( trie->trans,
1954 * sizeof(reg_trie_trans) );
1955 Zero( trie->trans + (transcount / 2), transcount / 2 , reg_trie_trans );
1957 base = trie->uniquecharcount + tp - minid;
1958 if ( maxid == minid ) {
1960 for ( ; zp < tp ; zp++ ) {
1961 if ( ! trie->trans[ zp ].next ) {
1962 base = trie->uniquecharcount + zp - minid;
1963 trie->trans[ zp ].next = TRIE_LIST_ITEM( state, 1).newstate;
1964 trie->trans[ zp ].check = state;
1970 trie->trans[ tp ].next = TRIE_LIST_ITEM( state, 1).newstate;
1971 trie->trans[ tp ].check = state;
1976 for ( idx=1; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
1977 const U32 tid = base - trie->uniquecharcount + TRIE_LIST_ITEM( state, idx ).forid;
1978 trie->trans[ tid ].next = TRIE_LIST_ITEM( state, idx ).newstate;
1979 trie->trans[ tid ].check = state;
1981 tp += ( maxid - minid + 1 );
1983 Safefree(trie->states[ state ].trans.list);
1986 DEBUG_TRIE_COMPILE_MORE_r(
1987 PerlIO_printf( Perl_debug_log, " base: %d\n",base);
1990 trie->states[ state ].trans.base=base;
1992 trie->lasttrans = tp + 1;
1996 Second Pass -- Flat Table Representation.
1998 we dont use the 0 slot of either trans[] or states[] so we add 1 to each.
1999 We know that we will need Charcount+1 trans at most to store the data
2000 (one row per char at worst case) So we preallocate both structures
2001 assuming worst case.
2003 We then construct the trie using only the .next slots of the entry
2006 We use the .check field of the first entry of the node temporarily to
2007 make compression both faster and easier by keeping track of how many non
2008 zero fields are in the node.
2010 Since trans are numbered from 1 any 0 pointer in the table is a FAIL
2013 There are two terms at use here: state as a TRIE_NODEIDX() which is a
2014 number representing the first entry of the node, and state as a
2015 TRIE_NODENUM() which is the trans number. state 1 is TRIE_NODEIDX(1) and
2016 TRIE_NODENUM(1), state 2 is TRIE_NODEIDX(2) and TRIE_NODENUM(3) if there
2017 are 2 entrys per node. eg:
2025 The table is internally in the right hand, idx form. However as we also
2026 have to deal with the states array which is indexed by nodenum we have to
2027 use TRIE_NODENUM() to convert.
2030 DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log,
2031 "%*sCompiling trie using table compiler\n",
2032 (int)depth * 2 + 2, ""));
2034 trie->trans = (reg_trie_trans *)
2035 PerlMemShared_calloc( ( TRIE_CHARCOUNT(trie) + 1 )
2036 * trie->uniquecharcount + 1,
2037 sizeof(reg_trie_trans) );
2038 trie->states = (reg_trie_state *)
2039 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
2040 sizeof(reg_trie_state) );
2041 next_alloc = trie->uniquecharcount + 1;
2044 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
2046 regnode *noper = NEXTOPER( cur );
2047 const U8 *uc = (U8*)STRING( noper );
2048 const U8 *e = uc + STR_LEN( noper );
2050 U32 state = 1; /* required init */
2052 U16 charid = 0; /* sanity init */
2053 U32 accept_state = 0; /* sanity init */
2054 U8 *scan = (U8*)NULL; /* sanity init */
2056 STRLEN foldlen = 0; /* required init */
2057 U32 wordlen = 0; /* required init */
2059 U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
2061 if (OP(noper) == NOTHING) {
2062 regnode *noper_next= regnext(noper);
2063 if (noper_next != tail && OP(noper_next) == flags) {
2065 uc= (U8*)STRING(noper);
2066 e= uc + STR_LEN(noper);
2070 if ( OP(noper) != NOTHING ) {
2071 for ( ; uc < e ; uc += len ) {
2076 charid = trie->charmap[ uvc ];
2078 SV* const * const svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 0);
2079 charid = svpp ? (U16)SvIV(*svpp) : 0;
2083 if ( !trie->trans[ state + charid ].next ) {
2084 trie->trans[ state + charid ].next = next_alloc;
2085 trie->trans[ state ].check++;
2086 prev_states[TRIE_NODENUM(next_alloc)]
2087 = TRIE_NODENUM(state);
2088 next_alloc += trie->uniquecharcount;
2090 state = trie->trans[ state + charid ].next;
2092 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %"IVdf, uvc );
2094 /* charid is now 0 if we dont know the char read, or nonzero if we do */
2097 accept_state = TRIE_NODENUM( state );
2098 TRIE_HANDLE_WORD(accept_state);
2100 } /* end second pass */
2102 /* and now dump it out before we compress it */
2103 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_table(trie, widecharmap,
2105 next_alloc, depth+1));
2109 * Inplace compress the table.*
2111 For sparse data sets the table constructed by the trie algorithm will
2112 be mostly 0/FAIL transitions or to put it another way mostly empty.
2113 (Note that leaf nodes will not contain any transitions.)
2115 This algorithm compresses the tables by eliminating most such
2116 transitions, at the cost of a modest bit of extra work during lookup:
2118 - Each states[] entry contains a .base field which indicates the
2119 index in the state[] array wheres its transition data is stored.
2121 - If .base is 0 there are no valid transitions from that node.
2123 - If .base is nonzero then charid is added to it to find an entry in
2126 -If trans[states[state].base+charid].check!=state then the
2127 transition is taken to be a 0/Fail transition. Thus if there are fail
2128 transitions at the front of the node then the .base offset will point
2129 somewhere inside the previous nodes data (or maybe even into a node
2130 even earlier), but the .check field determines if the transition is
2134 The following process inplace converts the table to the compressed
2135 table: We first do not compress the root node 1,and mark all its
2136 .check pointers as 1 and set its .base pointer as 1 as well. This
2137 allows us to do a DFA construction from the compressed table later,
2138 and ensures that any .base pointers we calculate later are greater
2141 - We set 'pos' to indicate the first entry of the second node.
2143 - We then iterate over the columns of the node, finding the first and
2144 last used entry at l and m. We then copy l..m into pos..(pos+m-l),
2145 and set the .check pointers accordingly, and advance pos
2146 appropriately and repreat for the next node. Note that when we copy
2147 the next pointers we have to convert them from the original
2148 NODEIDX form to NODENUM form as the former is not valid post
2151 - If a node has no transitions used we mark its base as 0 and do not
2152 advance the pos pointer.
2154 - If a node only has one transition we use a second pointer into the
2155 structure to fill in allocated fail transitions from other states.
2156 This pointer is independent of the main pointer and scans forward
2157 looking for null transitions that are allocated to a state. When it
2158 finds one it writes the single transition into the "hole". If the
2159 pointer doesnt find one the single transition is appended as normal.
2161 - Once compressed we can Renew/realloc the structures to release the
2164 See "Table-Compression Methods" in sec 3.9 of the Red Dragon,
2165 specifically Fig 3.47 and the associated pseudocode.
2169 const U32 laststate = TRIE_NODENUM( next_alloc );
2172 trie->statecount = laststate;
2174 for ( state = 1 ; state < laststate ; state++ ) {
2176 const U32 stateidx = TRIE_NODEIDX( state );
2177 const U32 o_used = trie->trans[ stateidx ].check;
2178 U32 used = trie->trans[ stateidx ].check;
2179 trie->trans[ stateidx ].check = 0;
2181 for ( charid = 0 ; used && charid < trie->uniquecharcount ; charid++ ) {
2182 if ( flag || trie->trans[ stateidx + charid ].next ) {
2183 if ( trie->trans[ stateidx + charid ].next ) {
2185 for ( ; zp < pos ; zp++ ) {
2186 if ( ! trie->trans[ zp ].next ) {
2190 trie->states[ state ].trans.base = zp + trie->uniquecharcount - charid ;
2191 trie->trans[ zp ].next = SAFE_TRIE_NODENUM( trie->trans[ stateidx + charid ].next );
2192 trie->trans[ zp ].check = state;
2193 if ( ++zp > pos ) pos = zp;
2200 trie->states[ state ].trans.base = pos + trie->uniquecharcount - charid ;
2202 trie->trans[ pos ].next = SAFE_TRIE_NODENUM( trie->trans[ stateidx + charid ].next );
2203 trie->trans[ pos ].check = state;
2208 trie->lasttrans = pos + 1;
2209 trie->states = (reg_trie_state *)
2210 PerlMemShared_realloc( trie->states, laststate
2211 * sizeof(reg_trie_state) );
2212 DEBUG_TRIE_COMPILE_MORE_r(
2213 PerlIO_printf( Perl_debug_log,
2214 "%*sAlloc: %d Orig: %"IVdf" elements, Final:%"IVdf". Savings of %%%5.2f\n",
2215 (int)depth * 2 + 2,"",
2216 (int)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1 ),
2219 ( ( next_alloc - pos ) * 100 ) / (double)next_alloc );
2222 } /* end table compress */
2224 DEBUG_TRIE_COMPILE_MORE_r(
2225 PerlIO_printf(Perl_debug_log, "%*sStatecount:%"UVxf" Lasttrans:%"UVxf"\n",
2226 (int)depth * 2 + 2, "",
2227 (UV)trie->statecount,
2228 (UV)trie->lasttrans)
2230 /* resize the trans array to remove unused space */
2231 trie->trans = (reg_trie_trans *)
2232 PerlMemShared_realloc( trie->trans, trie->lasttrans
2233 * sizeof(reg_trie_trans) );
2235 { /* Modify the program and insert the new TRIE node */
2236 U8 nodetype =(U8)(flags & 0xFF);
2240 regnode *optimize = NULL;
2241 #ifdef RE_TRACK_PATTERN_OFFSETS
2244 U32 mjd_nodelen = 0;
2245 #endif /* RE_TRACK_PATTERN_OFFSETS */
2246 #endif /* DEBUGGING */
2248 This means we convert either the first branch or the first Exact,
2249 depending on whether the thing following (in 'last') is a branch
2250 or not and whther first is the startbranch (ie is it a sub part of
2251 the alternation or is it the whole thing.)
2252 Assuming its a sub part we convert the EXACT otherwise we convert
2253 the whole branch sequence, including the first.
2255 /* Find the node we are going to overwrite */
2256 if ( first != startbranch || OP( last ) == BRANCH ) {
2257 /* branch sub-chain */
2258 NEXT_OFF( first ) = (U16)(last - first);
2259 #ifdef RE_TRACK_PATTERN_OFFSETS
2261 mjd_offset= Node_Offset((convert));
2262 mjd_nodelen= Node_Length((convert));
2265 /* whole branch chain */
2267 #ifdef RE_TRACK_PATTERN_OFFSETS
2270 const regnode *nop = NEXTOPER( convert );
2271 mjd_offset= Node_Offset((nop));
2272 mjd_nodelen= Node_Length((nop));
2276 PerlIO_printf(Perl_debug_log, "%*sMJD offset:%"UVuf" MJD length:%"UVuf"\n",
2277 (int)depth * 2 + 2, "",
2278 (UV)mjd_offset, (UV)mjd_nodelen)
2281 /* But first we check to see if there is a common prefix we can
2282 split out as an EXACT and put in front of the TRIE node. */
2283 trie->startstate= 1;
2284 if ( trie->bitmap && !widecharmap && !trie->jump ) {
2286 for ( state = 1 ; state < trie->statecount-1 ; state++ ) {
2290 const U32 base = trie->states[ state ].trans.base;
2292 if ( trie->states[state].wordnum )
2295 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
2296 if ( ( base + ofs >= trie->uniquecharcount ) &&
2297 ( base + ofs - trie->uniquecharcount < trie->lasttrans ) &&
2298 trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
2300 if ( ++count > 1 ) {
2301 SV **tmp = av_fetch( revcharmap, ofs, 0);
2302 const U8 *ch = (U8*)SvPV_nolen_const( *tmp );
2303 if ( state == 1 ) break;
2305 Zero(trie->bitmap, ANYOF_BITMAP_SIZE, char);
2307 PerlIO_printf(Perl_debug_log,
2308 "%*sNew Start State=%"UVuf" Class: [",
2309 (int)depth * 2 + 2, "",
2312 SV ** const tmp = av_fetch( revcharmap, idx, 0);
2313 const U8 * const ch = (U8*)SvPV_nolen_const( *tmp );
2315 TRIE_BITMAP_SET(trie,*ch);
2317 TRIE_BITMAP_SET(trie, folder[ *ch ]);
2319 PerlIO_printf(Perl_debug_log, "%s", (char*)ch)
2323 TRIE_BITMAP_SET(trie,*ch);
2325 TRIE_BITMAP_SET(trie,folder[ *ch ]);
2326 DEBUG_OPTIMISE_r(PerlIO_printf( Perl_debug_log,"%s", ch));
2332 SV **tmp = av_fetch( revcharmap, idx, 0);
2334 char *ch = SvPV( *tmp, len );
2336 SV *sv=sv_newmortal();
2337 PerlIO_printf( Perl_debug_log,
2338 "%*sPrefix State: %"UVuf" Idx:%"UVuf" Char='%s'\n",
2339 (int)depth * 2 + 2, "",
2341 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 6,
2342 PL_colors[0], PL_colors[1],
2343 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
2344 PERL_PV_ESCAPE_FIRSTCHAR
2349 OP( convert ) = nodetype;
2350 str=STRING(convert);
2353 STR_LEN(convert) += len;
2359 DEBUG_OPTIMISE_r(PerlIO_printf( Perl_debug_log,"]\n"));
2364 trie->prefixlen = (state-1);
2366 regnode *n = convert+NODE_SZ_STR(convert);
2367 NEXT_OFF(convert) = NODE_SZ_STR(convert);
2368 trie->startstate = state;
2369 trie->minlen -= (state - 1);
2370 trie->maxlen -= (state - 1);
2372 /* At least the UNICOS C compiler choked on this
2373 * being argument to DEBUG_r(), so let's just have
2376 #ifdef PERL_EXT_RE_BUILD
2382 regnode *fix = convert;
2383 U32 word = trie->wordcount;
2385 Set_Node_Offset_Length(convert, mjd_offset, state - 1);
2386 while( ++fix < n ) {
2387 Set_Node_Offset_Length(fix, 0, 0);
2390 SV ** const tmp = av_fetch( trie_words, word, 0 );
2392 if ( STR_LEN(convert) <= SvCUR(*tmp) )
2393 sv_chop(*tmp, SvPV_nolen(*tmp) + STR_LEN(convert));
2395 sv_chop(*tmp, SvPV_nolen(*tmp) + SvCUR(*tmp));
2403 NEXT_OFF(convert) = (U16)(tail - convert);
2404 DEBUG_r(optimize= n);
2410 if ( trie->maxlen ) {
2411 NEXT_OFF( convert ) = (U16)(tail - convert);
2412 ARG_SET( convert, data_slot );
2413 /* Store the offset to the first unabsorbed branch in
2414 jump[0], which is otherwise unused by the jump logic.
2415 We use this when dumping a trie and during optimisation. */
2417 trie->jump[0] = (U16)(nextbranch - convert);
2419 /* If the start state is not accepting (meaning there is no empty string/NOTHING)
2420 * and there is a bitmap
2421 * and the first "jump target" node we found leaves enough room
2422 * then convert the TRIE node into a TRIEC node, with the bitmap
2423 * embedded inline in the opcode - this is hypothetically faster.
2425 if ( !trie->states[trie->startstate].wordnum
2427 && ( (char *)jumper - (char *)convert) >= (int)sizeof(struct regnode_charclass) )
2429 OP( convert ) = TRIEC;
2430 Copy(trie->bitmap, ((struct regnode_charclass *)convert)->bitmap, ANYOF_BITMAP_SIZE, char);
2431 PerlMemShared_free(trie->bitmap);
2434 OP( convert ) = TRIE;
2436 /* store the type in the flags */
2437 convert->flags = nodetype;
2441 + regarglen[ OP( convert ) ];
2443 /* XXX We really should free up the resource in trie now,
2444 as we won't use them - (which resources?) dmq */
2446 /* needed for dumping*/
2447 DEBUG_r(if (optimize) {
2448 regnode *opt = convert;
2450 while ( ++opt < optimize) {
2451 Set_Node_Offset_Length(opt,0,0);
2454 Try to clean up some of the debris left after the
2457 while( optimize < jumper ) {
2458 mjd_nodelen += Node_Length((optimize));
2459 OP( optimize ) = OPTIMIZED;
2460 Set_Node_Offset_Length(optimize,0,0);
2463 Set_Node_Offset_Length(convert,mjd_offset,mjd_nodelen);
2465 } /* end node insert */
2466 REH_CALL_COMP_NODE_HOOK(pRExC_state->rx, convert);
2468 /* Finish populating the prev field of the wordinfo array. Walk back
2469 * from each accept state until we find another accept state, and if
2470 * so, point the first word's .prev field at the second word. If the
2471 * second already has a .prev field set, stop now. This will be the
2472 * case either if we've already processed that word's accept state,
2473 * or that state had multiple words, and the overspill words were
2474 * already linked up earlier.
2481 for (word=1; word <= trie->wordcount; word++) {
2483 if (trie->wordinfo[word].prev)
2485 state = trie->wordinfo[word].accept;
2487 state = prev_states[state];
2490 prev = trie->states[state].wordnum;
2494 trie->wordinfo[word].prev = prev;
2496 Safefree(prev_states);
2500 /* and now dump out the compressed format */
2501 DEBUG_TRIE_COMPILE_r(dump_trie(trie, widecharmap, revcharmap, depth+1));
2503 RExC_rxi->data->data[ data_slot + 1 ] = (void*)widecharmap;
2505 RExC_rxi->data->data[ data_slot + TRIE_WORDS_OFFSET ] = (void*)trie_words;
2506 RExC_rxi->data->data[ data_slot + 3 ] = (void*)revcharmap;
2508 SvREFCNT_dec_NN(revcharmap);
2512 : trie->startstate>1
2518 S_make_trie_failtable(pTHX_ RExC_state_t *pRExC_state, regnode *source, regnode *stclass, U32 depth)
2520 /* The Trie is constructed and compressed now so we can build a fail array if it's needed
2522 This is basically the Aho-Corasick algorithm. Its from exercise 3.31 and 3.32 in the
2523 "Red Dragon" -- Compilers, principles, techniques, and tools. Aho, Sethi, Ullman 1985/88
2526 We find the fail state for each state in the trie, this state is the longest proper
2527 suffix of the current state's 'word' that is also a proper prefix of another word in our
2528 trie. State 1 represents the word '' and is thus the default fail state. This allows
2529 the DFA not to have to restart after its tried and failed a word at a given point, it
2530 simply continues as though it had been matching the other word in the first place.
2532 'abcdgu'=~/abcdefg|cdgu/
2533 When we get to 'd' we are still matching the first word, we would encounter 'g' which would
2534 fail, which would bring us to the state representing 'd' in the second word where we would
2535 try 'g' and succeed, proceeding to match 'cdgu'.
2537 /* add a fail transition */
2538 const U32 trie_offset = ARG(source);
2539 reg_trie_data *trie=(reg_trie_data *)RExC_rxi->data->data[trie_offset];
2541 const U32 ucharcount = trie->uniquecharcount;
2542 const U32 numstates = trie->statecount;
2543 const U32 ubound = trie->lasttrans + ucharcount;
2547 U32 base = trie->states[ 1 ].trans.base;
2550 const U32 data_slot = add_data( pRExC_state, 1, "T" );
2551 GET_RE_DEBUG_FLAGS_DECL;
2553 PERL_ARGS_ASSERT_MAKE_TRIE_FAILTABLE;
2555 PERL_UNUSED_ARG(depth);
2559 ARG_SET( stclass, data_slot );
2560 aho = (reg_ac_data *) PerlMemShared_calloc( 1, sizeof(reg_ac_data) );
2561 RExC_rxi->data->data[ data_slot ] = (void*)aho;
2562 aho->trie=trie_offset;
2563 aho->states=(reg_trie_state *)PerlMemShared_malloc( numstates * sizeof(reg_trie_state) );
2564 Copy( trie->states, aho->states, numstates, reg_trie_state );
2565 Newxz( q, numstates, U32);
2566 aho->fail = (U32 *) PerlMemShared_calloc( numstates, sizeof(U32) );
2569 /* initialize fail[0..1] to be 1 so that we always have
2570 a valid final fail state */
2571 fail[ 0 ] = fail[ 1 ] = 1;
2573 for ( charid = 0; charid < ucharcount ; charid++ ) {
2574 const U32 newstate = TRIE_TRANS_STATE( 1, base, ucharcount, charid, 0 );
2576 q[ q_write ] = newstate;
2577 /* set to point at the root */
2578 fail[ q[ q_write++ ] ]=1;
2581 while ( q_read < q_write) {
2582 const U32 cur = q[ q_read++ % numstates ];
2583 base = trie->states[ cur ].trans.base;
2585 for ( charid = 0 ; charid < ucharcount ; charid++ ) {
2586 const U32 ch_state = TRIE_TRANS_STATE( cur, base, ucharcount, charid, 1 );
2588 U32 fail_state = cur;
2591 fail_state = fail[ fail_state ];
2592 fail_base = aho->states[ fail_state ].trans.base;
2593 } while ( !TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 ) );
2595 fail_state = TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 );
2596 fail[ ch_state ] = fail_state;
2597 if ( !aho->states[ ch_state ].wordnum && aho->states[ fail_state ].wordnum )
2599 aho->states[ ch_state ].wordnum = aho->states[ fail_state ].wordnum;
2601 q[ q_write++ % numstates] = ch_state;
2605 /* restore fail[0..1] to 0 so that we "fall out" of the AC loop
2606 when we fail in state 1, this allows us to use the
2607 charclass scan to find a valid start char. This is based on the principle
2608 that theres a good chance the string being searched contains lots of stuff
2609 that cant be a start char.
2611 fail[ 0 ] = fail[ 1 ] = 0;
2612 DEBUG_TRIE_COMPILE_r({
2613 PerlIO_printf(Perl_debug_log,
2614 "%*sStclass Failtable (%"UVuf" states): 0",
2615 (int)(depth * 2), "", (UV)numstates
2617 for( q_read=1; q_read<numstates; q_read++ ) {
2618 PerlIO_printf(Perl_debug_log, ", %"UVuf, (UV)fail[q_read]);
2620 PerlIO_printf(Perl_debug_log, "\n");
2623 /*RExC_seen |= REG_SEEN_TRIEDFA;*/
2628 * There are strange code-generation bugs caused on sparc64 by gcc-2.95.2.
2629 * These need to be revisited when a newer toolchain becomes available.
2631 #if defined(__sparc64__) && defined(__GNUC__)
2632 # if __GNUC__ < 2 || (__GNUC__ == 2 && __GNUC_MINOR__ < 96)
2633 # undef SPARC64_GCC_WORKAROUND
2634 # define SPARC64_GCC_WORKAROUND 1
2638 #define DEBUG_PEEP(str,scan,depth) \
2639 DEBUG_OPTIMISE_r({if (scan){ \
2640 SV * const mysv=sv_newmortal(); \
2641 regnode *Next = regnext(scan); \
2642 regprop(RExC_rx, mysv, scan); \
2643 PerlIO_printf(Perl_debug_log, "%*s" str ">%3d: %s (%d)\n", \
2644 (int)depth*2, "", REG_NODE_NUM(scan), SvPV_nolen_const(mysv),\
2645 Next ? (REG_NODE_NUM(Next)) : 0 ); \
2649 /* The below joins as many adjacent EXACTish nodes as possible into a single
2650 * one. The regop may be changed if the node(s) contain certain sequences that
2651 * require special handling. The joining is only done if:
2652 * 1) there is room in the current conglomerated node to entirely contain the
2654 * 2) they are the exact same node type
2656 * The adjacent nodes actually may be separated by NOTHING-kind nodes, and
2657 * these get optimized out
2659 * If a node is to match under /i (folded), the number of characters it matches
2660 * can be different than its character length if it contains a multi-character
2661 * fold. *min_subtract is set to the total delta of the input nodes.
2663 * And *has_exactf_sharp_s is set to indicate whether or not the node is EXACTF
2664 * and contains LATIN SMALL LETTER SHARP S
2666 * This is as good a place as any to discuss the design of handling these
2667 * multi-character fold sequences. It's been wrong in Perl for a very long
2668 * time. There are three code points in Unicode whose multi-character folds
2669 * were long ago discovered to mess things up. The previous designs for
2670 * dealing with these involved assigning a special node for them. This
2671 * approach doesn't work, as evidenced by this example:
2672 * "\xDFs" =~ /s\xDF/ui # Used to fail before these patches
2673 * Both these fold to "sss", but if the pattern is parsed to create a node that
2674 * would match just the \xDF, it won't be able to handle the case where a
2675 * successful match would have to cross the node's boundary. The new approach
2676 * that hopefully generally solves the problem generates an EXACTFU_SS node
2679 * It turns out that there are problems with all multi-character folds, and not
2680 * just these three. Now the code is general, for all such cases, but the
2681 * three still have some special handling. The approach taken is:
2682 * 1) This routine examines each EXACTFish node that could contain multi-
2683 * character fold sequences. It returns in *min_subtract how much to
2684 * subtract from the the actual length of the string to get a real minimum
2685 * match length; it is 0 if there are no multi-char folds. This delta is
2686 * used by the caller to adjust the min length of the match, and the delta
2687 * between min and max, so that the optimizer doesn't reject these
2688 * possibilities based on size constraints.
2689 * 2) Certain of these sequences require special handling by the trie code,
2690 * so, if found, this code changes the joined node type to special ops:
2691 * EXACTFU_TRICKYFOLD and EXACTFU_SS.
2692 * 3) For the sequence involving the Sharp s (\xDF), the node type EXACTFU_SS
2693 * is used for an EXACTFU node that contains at least one "ss" sequence in
2694 * it. For non-UTF-8 patterns and strings, this is the only case where
2695 * there is a possible fold length change. That means that a regular
2696 * EXACTFU node without UTF-8 involvement doesn't have to concern itself
2697 * with length changes, and so can be processed faster. regexec.c takes
2698 * advantage of this. Generally, an EXACTFish node that is in UTF-8 is
2699 * pre-folded by regcomp.c. This saves effort in regex matching.
2700 * However, the pre-folding isn't done for non-UTF8 patterns because the
2701 * fold of the MICRO SIGN requires UTF-8, and we don't want to slow things
2702 * down by forcing the pattern into UTF8 unless necessary. Also what
2703 * EXACTF and EXACTFL nodes fold to isn't known until runtime. The fold
2704 * possibilities for the non-UTF8 patterns are quite simple, except for
2705 * the sharp s. All the ones that don't involve a UTF-8 target string are
2706 * members of a fold-pair, and arrays are set up for all of them so that
2707 * the other member of the pair can be found quickly. Code elsewhere in
2708 * this file makes sure that in EXACTFU nodes, the sharp s gets folded to
2709 * 'ss', even if the pattern isn't UTF-8. This avoids the issues
2710 * described in the next item.
2711 * 4) A problem remains for the sharp s in EXACTF and EXACTFA nodes when the
2712 * pattern isn't in UTF-8. (BTW, there cannot be an EXACTF node with a
2713 * UTF-8 pattern.) An assumption that the optimizer part of regexec.c
2714 * (probably unwittingly, in Perl_regexec_flags()) makes is that a
2715 * character in the pattern corresponds to at most a single character in
2716 * the target string. (And I do mean character, and not byte here, unlike
2717 * other parts of the documentation that have never been updated to
2718 * account for multibyte Unicode.) sharp s in EXACTF nodes can match the
2719 * two character string 'ss'; in EXACTFA nodes it can match
2720 * "\x{17F}\x{17F}". These violate the assumption, and they are the only
2721 * instances where it is violated. I'm reluctant to try to change the
2722 * assumption, as the code involved is impenetrable to me (khw), so
2723 * instead the code here punts. This routine examines (when the pattern
2724 * isn't UTF-8) EXACTF and EXACTFA nodes for the sharp s, and returns a
2725 * boolean indicating whether or not the node contains a sharp s. When it
2726 * is true, the caller sets a flag that later causes the optimizer in this
2727 * file to not set values for the floating and fixed string lengths, and
2728 * thus avoids the optimizer code in regexec.c that makes the invalid
2729 * assumption. Thus, there is no optimization based on string lengths for
2730 * non-UTF8-pattern EXACTF and EXACTFA nodes that contain the sharp s.
2731 * (The reason the assumption is wrong only in these two cases is that all
2732 * other non-UTF-8 folds are 1-1; and, for UTF-8 patterns, we pre-fold all
2733 * other folds to their expanded versions. We can't prefold sharp s to
2734 * 'ss' in EXACTF nodes because we don't know at compile time if it
2735 * actually matches 'ss' or not. It will match iff the target string is
2736 * in UTF-8, unlike the EXACTFU nodes, where it always matches; and
2737 * EXACTFA and EXACTFL where it never does. In an EXACTFA node in a UTF-8
2738 * pattern, sharp s is folded to "\x{17F}\x{17F}, avoiding the problem;
2739 * but in a non-UTF8 pattern, folding it to that above-Latin1 string would
2740 * require the pattern to be forced into UTF-8, the overhead of which we
2744 #define JOIN_EXACT(scan,min_subtract,has_exactf_sharp_s, flags) \
2745 if (PL_regkind[OP(scan)] == EXACT) \
2746 join_exact(pRExC_state,(scan),(min_subtract),has_exactf_sharp_s, (flags),NULL,depth+1)
2749 S_join_exact(pTHX_ RExC_state_t *pRExC_state, regnode *scan, UV *min_subtract, bool *has_exactf_sharp_s, U32 flags,regnode *val, U32 depth) {
2750 /* Merge several consecutive EXACTish nodes into one. */
2751 regnode *n = regnext(scan);
2753 regnode *next = scan + NODE_SZ_STR(scan);
2757 regnode *stop = scan;
2758 GET_RE_DEBUG_FLAGS_DECL;
2760 PERL_UNUSED_ARG(depth);
2763 PERL_ARGS_ASSERT_JOIN_EXACT;
2764 #ifndef EXPERIMENTAL_INPLACESCAN
2765 PERL_UNUSED_ARG(flags);
2766 PERL_UNUSED_ARG(val);
2768 DEBUG_PEEP("join",scan,depth);
2770 /* Look through the subsequent nodes in the chain. Skip NOTHING, merge
2771 * EXACT ones that are mergeable to the current one. */
2773 && (PL_regkind[OP(n)] == NOTHING
2774 || (stringok && OP(n) == OP(scan)))
2776 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX)
2779 if (OP(n) == TAIL || n > next)
2781 if (PL_regkind[OP(n)] == NOTHING) {
2782 DEBUG_PEEP("skip:",n,depth);
2783 NEXT_OFF(scan) += NEXT_OFF(n);
2784 next = n + NODE_STEP_REGNODE;
2791 else if (stringok) {
2792 const unsigned int oldl = STR_LEN(scan);
2793 regnode * const nnext = regnext(n);
2795 /* XXX I (khw) kind of doubt that this works on platforms where
2796 * U8_MAX is above 255 because of lots of other assumptions */
2797 /* Don't join if the sum can't fit into a single node */
2798 if (oldl + STR_LEN(n) > U8_MAX)
2801 DEBUG_PEEP("merg",n,depth);
2804 NEXT_OFF(scan) += NEXT_OFF(n);
2805 STR_LEN(scan) += STR_LEN(n);
2806 next = n + NODE_SZ_STR(n);
2807 /* Now we can overwrite *n : */
2808 Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
2816 #ifdef EXPERIMENTAL_INPLACESCAN
2817 if (flags && !NEXT_OFF(n)) {
2818 DEBUG_PEEP("atch", val, depth);
2819 if (reg_off_by_arg[OP(n)]) {
2820 ARG_SET(n, val - n);
2823 NEXT_OFF(n) = val - n;
2831 *has_exactf_sharp_s = FALSE;
2833 /* Here, all the adjacent mergeable EXACTish nodes have been merged. We
2834 * can now analyze for sequences of problematic code points. (Prior to
2835 * this final joining, sequences could have been split over boundaries, and
2836 * hence missed). The sequences only happen in folding, hence for any
2837 * non-EXACT EXACTish node */
2838 if (OP(scan) != EXACT) {
2839 const U8 * const s0 = (U8*) STRING(scan);
2841 const U8 * const s_end = s0 + STR_LEN(scan);
2843 /* One pass is made over the node's string looking for all the
2844 * possibilities. to avoid some tests in the loop, there are two main
2845 * cases, for UTF-8 patterns (which can't have EXACTF nodes) and
2849 /* Examine the string for a multi-character fold sequence. UTF-8
2850 * patterns have all characters pre-folded by the time this code is
2852 while (s < s_end - 1) /* Can stop 1 before the end, as minimum
2853 length sequence we are looking for is 2 */
2856 int len = is_MULTI_CHAR_FOLD_utf8_safe(s, s_end);
2857 if (! len) { /* Not a multi-char fold: get next char */
2862 /* Nodes with 'ss' require special handling, except for EXACTFL
2863 * and EXACTFA for which there is no multi-char fold to this */
2864 if (len == 2 && *s == 's' && *(s+1) == 's'
2865 && OP(scan) != EXACTFL && OP(scan) != EXACTFA)
2868 OP(scan) = EXACTFU_SS;
2871 else if (len == 6 /* len is the same in both ASCII and EBCDIC
2873 && (memEQ(s, GREEK_SMALL_LETTER_IOTA_UTF8
2874 COMBINING_DIAERESIS_UTF8
2875 COMBINING_ACUTE_ACCENT_UTF8,
2877 || memEQ(s, GREEK_SMALL_LETTER_UPSILON_UTF8
2878 COMBINING_DIAERESIS_UTF8
2879 COMBINING_ACUTE_ACCENT_UTF8,
2884 /* These two folds require special handling by trie's, so
2885 * change the node type to indicate this. If EXACTFA and
2886 * EXACTFL were ever to be handled by trie's, this would
2887 * have to be changed. If this node has already been
2888 * changed to EXACTFU_SS in this loop, leave it as is. (I
2889 * (khw) think it doesn't matter in regexec.c for UTF
2890 * patterns, but no need to change it */
2891 if (OP(scan) == EXACTFU) {
2892 OP(scan) = EXACTFU_TRICKYFOLD;
2896 else { /* Here is a generic multi-char fold. */
2897 const U8* multi_end = s + len;
2899 /* Count how many characters in it. In the case of /l and
2900 * /aa, no folds which contain ASCII code points are
2901 * allowed, so check for those, and skip if found. (In
2902 * EXACTFL, no folds are allowed to any Latin1 code point,
2903 * not just ASCII. But there aren't any of these
2904 * currently, nor ever likely, so don't take the time to
2905 * test for them. The code that generates the
2906 * is_MULTI_foo() macros croaks should one actually get put
2907 * into Unicode .) */
2908 if (OP(scan) != EXACTFL && OP(scan) != EXACTFA) {
2909 count = utf8_length(s, multi_end);
2913 while (s < multi_end) {
2916 goto next_iteration;
2926 /* The delta is how long the sequence is minus 1 (1 is how long
2927 * the character that folds to the sequence is) */
2928 *min_subtract += count - 1;
2932 else if (OP(scan) == EXACTFA) {
2934 /* Non-UTF-8 pattern, EXACTFA node. There can't be a multi-char
2935 * fold to the ASCII range (and there are no existing ones in the
2936 * upper latin1 range). But, as outlined in the comments preceding
2937 * this function, we need to flag any occurrences of the sharp s */
2939 if (*s == LATIN_SMALL_LETTER_SHARP_S) {
2940 *has_exactf_sharp_s = TRUE;
2947 else if (OP(scan) != EXACTFL) {
2949 /* Non-UTF-8 pattern, not EXACTFA nor EXACTFL node. Look for the
2950 * multi-char folds that are all Latin1. (This code knows that
2951 * there are no current multi-char folds possible with EXACTFL,
2952 * relying on fold_grind.t to catch any errors if the very unlikely
2953 * event happens that some get added in future Unicode versions.)
2954 * As explained in the comments preceding this function, we look
2955 * also for the sharp s in EXACTF nodes; it can be in the final
2956 * position. Otherwise we can stop looking 1 byte earlier because
2957 * have to find at least two characters for a multi-fold */
2958 const U8* upper = (OP(scan) == EXACTF) ? s_end : s_end -1;
2961 int len = is_MULTI_CHAR_FOLD_latin1_safe(s, s_end);
2962 if (! len) { /* Not a multi-char fold. */
2963 if (*s == LATIN_SMALL_LETTER_SHARP_S && OP(scan) == EXACTF)
2965 *has_exactf_sharp_s = TRUE;
2972 && isARG2_lower_or_UPPER_ARG1('s', *s)
2973 && isARG2_lower_or_UPPER_ARG1('s', *(s+1)))
2976 /* EXACTF nodes need to know that the minimum length
2977 * changed so that a sharp s in the string can match this
2978 * ss in the pattern, but they remain EXACTF nodes, as they
2979 * won't match this unless the target string is is UTF-8,
2980 * which we don't know until runtime */
2981 if (OP(scan) != EXACTF) {
2982 OP(scan) = EXACTFU_SS;
2986 *min_subtract += len - 1;
2993 /* Allow dumping but overwriting the collection of skipped
2994 * ops and/or strings with fake optimized ops */
2995 n = scan + NODE_SZ_STR(scan);
3003 DEBUG_OPTIMISE_r(if (merged){DEBUG_PEEP("finl",scan,depth)});
3007 /* REx optimizer. Converts nodes into quicker variants "in place".
3008 Finds fixed substrings. */
3010 /* Stops at toplevel WHILEM as well as at "last". At end *scanp is set
3011 to the position after last scanned or to NULL. */
3013 #define INIT_AND_WITHP \
3014 assert(!and_withp); \
3015 Newx(and_withp,1,struct regnode_charclass_class); \
3016 SAVEFREEPV(and_withp)
3018 /* this is a chain of data about sub patterns we are processing that
3019 need to be handled separately/specially in study_chunk. Its so
3020 we can simulate recursion without losing state. */
3022 typedef struct scan_frame {
3023 regnode *last; /* last node to process in this frame */
3024 regnode *next; /* next node to process when last is reached */
3025 struct scan_frame *prev; /*previous frame*/
3026 I32 stop; /* what stopparen do we use */
3030 #define SCAN_COMMIT(s, data, m) scan_commit(s, data, m, is_inf)
3033 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
3034 I32 *minlenp, I32 *deltap,
3039 struct regnode_charclass_class *and_withp,
3040 U32 flags, U32 depth)
3041 /* scanp: Start here (read-write). */
3042 /* deltap: Write maxlen-minlen here. */
3043 /* last: Stop before this one. */
3044 /* data: string data about the pattern */
3045 /* stopparen: treat close N as END */
3046 /* recursed: which subroutines have we recursed into */
3047 /* and_withp: Valid if flags & SCF_DO_STCLASS_OR */
3050 I32 min = 0; /* There must be at least this number of characters to match */
3052 regnode *scan = *scanp, *next;
3054 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
3055 int is_inf_internal = 0; /* The studied chunk is infinite */
3056 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
3057 scan_data_t data_fake;
3058 SV *re_trie_maxbuff = NULL;
3059 regnode *first_non_open = scan;
3060 I32 stopmin = I32_MAX;
3061 scan_frame *frame = NULL;
3062 GET_RE_DEBUG_FLAGS_DECL;
3064 PERL_ARGS_ASSERT_STUDY_CHUNK;
3067 StructCopy(&zero_scan_data, &data_fake, scan_data_t);
3071 while (first_non_open && OP(first_non_open) == OPEN)
3072 first_non_open=regnext(first_non_open);
3077 while ( scan && OP(scan) != END && scan < last ){
3078 UV min_subtract = 0; /* How mmany chars to subtract from the minimum
3079 node length to get a real minimum (because
3080 the folded version may be shorter) */
3081 bool has_exactf_sharp_s = FALSE;
3082 /* Peephole optimizer: */
3083 DEBUG_STUDYDATA("Peep:", data,depth);
3084 DEBUG_PEEP("Peep",scan,depth);
3086 /* Its not clear to khw or hv why this is done here, and not in the
3087 * clauses that deal with EXACT nodes. khw's guess is that it's
3088 * because of a previous design */
3089 JOIN_EXACT(scan,&min_subtract, &has_exactf_sharp_s, 0);
3091 /* Follow the next-chain of the current node and optimize
3092 away all the NOTHINGs from it. */
3093 if (OP(scan) != CURLYX) {
3094 const int max = (reg_off_by_arg[OP(scan)]
3096 /* I32 may be smaller than U16 on CRAYs! */
3097 : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
3098 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
3102 /* Skip NOTHING and LONGJMP. */
3103 while ((n = regnext(n))
3104 && ((PL_regkind[OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
3105 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
3106 && off + noff < max)
3108 if (reg_off_by_arg[OP(scan)])
3111 NEXT_OFF(scan) = off;
3116 /* The principal pseudo-switch. Cannot be a switch, since we
3117 look into several different things. */
3118 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
3119 || OP(scan) == IFTHEN) {
3120 next = regnext(scan);
3122 /* demq: the op(next)==code check is to see if we have "branch-branch" AFAICT */
3124 if (OP(next) == code || code == IFTHEN) {
3125 /* NOTE - There is similar code to this block below for handling
3126 TRIE nodes on a re-study. If you change stuff here check there
3128 I32 max1 = 0, min1 = I32_MAX, num = 0;
3129 struct regnode_charclass_class accum;
3130 regnode * const startbranch=scan;
3132 if (flags & SCF_DO_SUBSTR)
3133 SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot merge strings after this. */
3134 if (flags & SCF_DO_STCLASS)
3135 cl_init_zero(pRExC_state, &accum);
3137 while (OP(scan) == code) {
3138 I32 deltanext, minnext, f = 0, fake;
3139 struct regnode_charclass_class this_class;
3142 data_fake.flags = 0;
3144 data_fake.whilem_c = data->whilem_c;
3145 data_fake.last_closep = data->last_closep;
3148 data_fake.last_closep = &fake;
3150 data_fake.pos_delta = delta;
3151 next = regnext(scan);
3152 scan = NEXTOPER(scan);
3154 scan = NEXTOPER(scan);
3155 if (flags & SCF_DO_STCLASS) {
3156 cl_init(pRExC_state, &this_class);
3157 data_fake.start_class = &this_class;
3158 f = SCF_DO_STCLASS_AND;
3160 if (flags & SCF_WHILEM_VISITED_POS)
3161 f |= SCF_WHILEM_VISITED_POS;
3163 /* we suppose the run is continuous, last=next...*/
3164 minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext,
3166 stopparen, recursed, NULL, f,depth+1);
3169 if (deltanext == I32_MAX) {
3170 is_inf = is_inf_internal = 1;
3172 } else if (max1 < minnext + deltanext)
3173 max1 = minnext + deltanext;
3175 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
3177 if (data_fake.flags & SCF_SEEN_ACCEPT) {
3178 if ( stopmin > minnext)
3179 stopmin = min + min1;
3180 flags &= ~SCF_DO_SUBSTR;
3182 data->flags |= SCF_SEEN_ACCEPT;
3185 if (data_fake.flags & SF_HAS_EVAL)
3186 data->flags |= SF_HAS_EVAL;
3187 data->whilem_c = data_fake.whilem_c;
3189 if (flags & SCF_DO_STCLASS)
3190 cl_or(pRExC_state, &accum, &this_class);
3192 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
3194 if (flags & SCF_DO_SUBSTR) {
3195 data->pos_min += min1;
3196 if (data->pos_delta >= I32_MAX - (max1 - min1))
3197 data->pos_delta = I32_MAX;
3199 data->pos_delta += max1 - min1;
3200 if (max1 != min1 || is_inf)
3201 data->longest = &(data->longest_float);
3204 if (delta == I32_MAX || I32_MAX - delta - (max1 - min1) < 0)
3207 delta += max1 - min1;
3208 if (flags & SCF_DO_STCLASS_OR) {
3209 cl_or(pRExC_state, data->start_class, &accum);
3211 cl_and(data->start_class, and_withp);
3212 flags &= ~SCF_DO_STCLASS;
3215 else if (flags & SCF_DO_STCLASS_AND) {
3217 cl_and(data->start_class, &accum);
3218 flags &= ~SCF_DO_STCLASS;
3221 /* Switch to OR mode: cache the old value of
3222 * data->start_class */
3224 StructCopy(data->start_class, and_withp,
3225 struct regnode_charclass_class);
3226 flags &= ~SCF_DO_STCLASS_AND;
3227 StructCopy(&accum, data->start_class,
3228 struct regnode_charclass_class);
3229 flags |= SCF_DO_STCLASS_OR;
3230 SET_SSC_EOS(data->start_class);
3234 if (PERL_ENABLE_TRIE_OPTIMISATION && OP( startbranch ) == BRANCH ) {
3237 Assuming this was/is a branch we are dealing with: 'scan' now
3238 points at the item that follows the branch sequence, whatever
3239 it is. We now start at the beginning of the sequence and look
3246 which would be constructed from a pattern like /A|LIST|OF|WORDS/
3248 If we can find such a subsequence we need to turn the first
3249 element into a trie and then add the subsequent branch exact
3250 strings to the trie.
3254 1. patterns where the whole set of branches can be converted.
3256 2. patterns where only a subset can be converted.
3258 In case 1 we can replace the whole set with a single regop
3259 for the trie. In case 2 we need to keep the start and end
3262 'BRANCH EXACT; BRANCH EXACT; BRANCH X'
3263 becomes BRANCH TRIE; BRANCH X;
3265 There is an additional case, that being where there is a
3266 common prefix, which gets split out into an EXACT like node
3267 preceding the TRIE node.
3269 If x(1..n)==tail then we can do a simple trie, if not we make
3270 a "jump" trie, such that when we match the appropriate word
3271 we "jump" to the appropriate tail node. Essentially we turn
3272 a nested if into a case structure of sorts.
3277 if (!re_trie_maxbuff) {
3278 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
3279 if (!SvIOK(re_trie_maxbuff))
3280 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
3282 if ( SvIV(re_trie_maxbuff)>=0 ) {
3284 regnode *first = (regnode *)NULL;
3285 regnode *last = (regnode *)NULL;
3286 regnode *tail = scan;
3291 SV * const mysv = sv_newmortal(); /* for dumping */
3293 /* var tail is used because there may be a TAIL
3294 regop in the way. Ie, the exacts will point to the
3295 thing following the TAIL, but the last branch will
3296 point at the TAIL. So we advance tail. If we
3297 have nested (?:) we may have to move through several
3301 while ( OP( tail ) == TAIL ) {
3302 /* this is the TAIL generated by (?:) */
3303 tail = regnext( tail );
3307 DEBUG_TRIE_COMPILE_r({
3308 regprop(RExC_rx, mysv, tail );
3309 PerlIO_printf( Perl_debug_log, "%*s%s%s\n",
3310 (int)depth * 2 + 2, "",
3311 "Looking for TRIE'able sequences. Tail node is: ",
3312 SvPV_nolen_const( mysv )
3318 Step through the branches
3319 cur represents each branch,
3320 noper is the first thing to be matched as part of that branch
3321 noper_next is the regnext() of that node.
3323 We normally handle a case like this /FOO[xyz]|BAR[pqr]/
3324 via a "jump trie" but we also support building with NOJUMPTRIE,
3325 which restricts the trie logic to structures like /FOO|BAR/.
3327 If noper is a trieable nodetype then the branch is a possible optimization
3328 target. If we are building under NOJUMPTRIE then we require that noper_next
3329 is the same as scan (our current position in the regex program).
3331 Once we have two or more consecutive such branches we can create a
3332 trie of the EXACT's contents and stitch it in place into the program.
3334 If the sequence represents all of the branches in the alternation we
3335 replace the entire thing with a single TRIE node.
3337 Otherwise when it is a subsequence we need to stitch it in place and
3338 replace only the relevant branches. This means the first branch has
3339 to remain as it is used by the alternation logic, and its next pointer,
3340 and needs to be repointed at the item on the branch chain following
3341 the last branch we have optimized away.
3343 This could be either a BRANCH, in which case the subsequence is internal,
3344 or it could be the item following the branch sequence in which case the
3345 subsequence is at the end (which does not necessarily mean the first node
3346 is the start of the alternation).
3348 TRIE_TYPE(X) is a define which maps the optype to a trietype.
3351 ----------------+-----------
3355 EXACTFU_SS | EXACTFU
3356 EXACTFU_TRICKYFOLD | EXACTFU
3361 #define TRIE_TYPE(X) ( ( NOTHING == (X) ) ? NOTHING : \
3362 ( EXACT == (X) ) ? EXACT : \
3363 ( EXACTFU == (X) || EXACTFU_SS == (X) || EXACTFU_TRICKYFOLD == (X) ) ? EXACTFU : \
3366 /* dont use tail as the end marker for this traverse */
3367 for ( cur = startbranch ; cur != scan ; cur = regnext( cur ) ) {
3368 regnode * const noper = NEXTOPER( cur );
3369 U8 noper_type = OP( noper );
3370 U8 noper_trietype = TRIE_TYPE( noper_type );
3371 #if defined(DEBUGGING) || defined(NOJUMPTRIE)
3372 regnode * const noper_next = regnext( noper );
3373 U8 noper_next_type = (noper_next && noper_next != tail) ? OP(noper_next) : 0;
3374 U8 noper_next_trietype = (noper_next && noper_next != tail) ? TRIE_TYPE( noper_next_type ) :0;
3377 DEBUG_TRIE_COMPILE_r({
3378 regprop(RExC_rx, mysv, cur);
3379 PerlIO_printf( Perl_debug_log, "%*s- %s (%d)",
3380 (int)depth * 2 + 2,"", SvPV_nolen_const( mysv ), REG_NODE_NUM(cur) );
3382 regprop(RExC_rx, mysv, noper);
3383 PerlIO_printf( Perl_debug_log, " -> %s",
3384 SvPV_nolen_const(mysv));
3387 regprop(RExC_rx, mysv, noper_next );
3388 PerlIO_printf( Perl_debug_log,"\t=> %s\t",
3389 SvPV_nolen_const(mysv));
3391 PerlIO_printf( Perl_debug_log, "(First==%d,Last==%d,Cur==%d,tt==%s,nt==%s,nnt==%s)\n",
3392 REG_NODE_NUM(first), REG_NODE_NUM(last), REG_NODE_NUM(cur),
3393 PL_reg_name[trietype], PL_reg_name[noper_trietype], PL_reg_name[noper_next_trietype]
3397 /* Is noper a trieable nodetype that can be merged with the
3398 * current trie (if there is one)? */
3402 ( noper_trietype == NOTHING)
3403 || ( trietype == NOTHING )
3404 || ( trietype == noper_trietype )
3407 && noper_next == tail
3411 /* Handle mergable triable node
3412 * Either we are the first node in a new trieable sequence,
3413 * in which case we do some bookkeeping, otherwise we update
3414 * the end pointer. */
3417 if ( noper_trietype == NOTHING ) {
3418 #if !defined(DEBUGGING) && !defined(NOJUMPTRIE)
3419 regnode * const noper_next = regnext( noper );
3420 U8 noper_next_type = (noper_next && noper_next!=tail) ? OP(noper_next) : 0;
3421 U8 noper_next_trietype = noper_next_type ? TRIE_TYPE( noper_next_type ) :0;
3424 if ( noper_next_trietype ) {
3425 trietype = noper_next_trietype;
3426 } else if (noper_next_type) {
3427 /* a NOTHING regop is 1 regop wide. We need at least two
3428 * for a trie so we can't merge this in */
3432 trietype = noper_trietype;
3435 if ( trietype == NOTHING )
3436 trietype = noper_trietype;
3441 } /* end handle mergable triable node */
3443 /* handle unmergable node -
3444 * noper may either be a triable node which can not be tried
3445 * together with the current trie, or a non triable node */
3447 /* If last is set and trietype is not NOTHING then we have found
3448 * at least two triable branch sequences in a row of a similar
3449 * trietype so we can turn them into a trie. If/when we
3450 * allow NOTHING to start a trie sequence this condition will be
3451 * required, and it isn't expensive so we leave it in for now. */
3452 if ( trietype && trietype != NOTHING )
3453 make_trie( pRExC_state,
3454 startbranch, first, cur, tail, count,
3455 trietype, depth+1 );
3456 last = NULL; /* note: we clear/update first, trietype etc below, so we dont do it here */
3460 && noper_next == tail
3463 /* noper is triable, so we can start a new trie sequence */
3466 trietype = noper_trietype;
3468 /* if we already saw a first but the current node is not triable then we have
3469 * to reset the first information. */
3474 } /* end handle unmergable node */
3475 } /* loop over branches */
3476 DEBUG_TRIE_COMPILE_r({
3477 regprop(RExC_rx, mysv, cur);
3478 PerlIO_printf( Perl_debug_log,
3479 "%*s- %s (%d) <SCAN FINISHED>\n", (int)depth * 2 + 2,
3480 "", SvPV_nolen_const( mysv ),REG_NODE_NUM(cur));
3483 if ( last && trietype ) {
3484 if ( trietype != NOTHING ) {
3485 /* the last branch of the sequence was part of a trie,
3486 * so we have to construct it here outside of the loop
3488 made= make_trie( pRExC_state, startbranch, first, scan, tail, count, trietype, depth+1 );
3489 #ifdef TRIE_STUDY_OPT
3490 if ( ((made == MADE_EXACT_TRIE &&
3491 startbranch == first)
3492 || ( first_non_open == first )) &&
3494 flags |= SCF_TRIE_RESTUDY;
3495 if ( startbranch == first
3498 RExC_seen &=~REG_TOP_LEVEL_BRANCHES;
3503 /* at this point we know whatever we have is a NOTHING sequence/branch
3504 * AND if 'startbranch' is 'first' then we can turn the whole thing into a NOTHING
3506 if ( startbranch == first ) {
3508 /* the entire thing is a NOTHING sequence, something like this:
3509 * (?:|) So we can turn it into a plain NOTHING op. */
3510 DEBUG_TRIE_COMPILE_r({
3511 regprop(RExC_rx, mysv, cur);
3512 PerlIO_printf( Perl_debug_log,
3513 "%*s- %s (%d) <NOTHING BRANCH SEQUENCE>\n", (int)depth * 2 + 2,
3514 "", SvPV_nolen_const( mysv ),REG_NODE_NUM(cur));
3517 OP(startbranch)= NOTHING;
3518 NEXT_OFF(startbranch)= tail - startbranch;
3519 for ( opt= startbranch + 1; opt < tail ; opt++ )
3523 } /* end if ( last) */
3524 } /* TRIE_MAXBUF is non zero */
3529 else if ( code == BRANCHJ ) { /* single branch is optimized. */
3530 scan = NEXTOPER(NEXTOPER(scan));
3531 } else /* single branch is optimized. */
3532 scan = NEXTOPER(scan);
3534 } else if (OP(scan) == SUSPEND || OP(scan) == GOSUB || OP(scan) == GOSTART) {
3535 scan_frame *newframe = NULL;
3540 if (OP(scan) != SUSPEND) {
3541 /* set the pointer */
3542 if (OP(scan) == GOSUB) {
3544 RExC_recurse[ARG2L(scan)] = scan;
3545 start = RExC_open_parens[paren-1];
3546 end = RExC_close_parens[paren-1];
3549 start = RExC_rxi->program + 1;
3553 Newxz(recursed, (((RExC_npar)>>3) +1), U8);
3554 SAVEFREEPV(recursed);
3556 if (!PAREN_TEST(recursed,paren+1)) {
3557 PAREN_SET(recursed,paren+1);
3558 Newx(newframe,1,scan_frame);
3560 if (flags & SCF_DO_SUBSTR) {
3561 SCAN_COMMIT(pRExC_state,data,minlenp);
3562 data->longest = &(data->longest_float);
3564 is_inf = is_inf_internal = 1;
3565 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
3566 cl_anything(pRExC_state, data->start_class);
3567 flags &= ~SCF_DO_STCLASS;
3570 Newx(newframe,1,scan_frame);
3573 end = regnext(scan);
3578 SAVEFREEPV(newframe);
3579 newframe->next = regnext(scan);
3580 newframe->last = last;
3581 newframe->stop = stopparen;
3582 newframe->prev = frame;
3592 else if (OP(scan) == EXACT) {
3593 I32 l = STR_LEN(scan);
3596 const U8 * const s = (U8*)STRING(scan);
3597 uc = utf8_to_uvchr_buf(s, s + l, NULL);
3598 l = utf8_length(s, s + l);
3600 uc = *((U8*)STRING(scan));
3603 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
3604 /* The code below prefers earlier match for fixed
3605 offset, later match for variable offset. */
3606 if (data->last_end == -1) { /* Update the start info. */
3607 data->last_start_min = data->pos_min;
3608 data->last_start_max = is_inf
3609 ? I32_MAX : data->pos_min + data->pos_delta;
3611 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
3613 SvUTF8_on(data->last_found);
3615 SV * const sv = data->last_found;
3616 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
3617 mg_find(sv, PERL_MAGIC_utf8) : NULL;
3618 if (mg && mg->mg_len >= 0)
3619 mg->mg_len += utf8_length((U8*)STRING(scan),
3620 (U8*)STRING(scan)+STR_LEN(scan));
3622 data->last_end = data->pos_min + l;
3623 data->pos_min += l; /* As in the first entry. */
3624 data->flags &= ~SF_BEFORE_EOL;
3626 if (flags & SCF_DO_STCLASS_AND) {
3627 /* Check whether it is compatible with what we know already! */
3631 /* If compatible, we or it in below. It is compatible if is
3632 * in the bitmp and either 1) its bit or its fold is set, or 2)
3633 * it's for a locale. Even if there isn't unicode semantics
3634 * here, at runtime there may be because of matching against a
3635 * utf8 string, so accept a possible false positive for
3636 * latin1-range folds */
3638 (!(data->start_class->flags & ANYOF_LOCALE)
3639 && !ANYOF_BITMAP_TEST(data->start_class, uc)
3640 && (!(data->start_class->flags & ANYOF_LOC_FOLD)
3641 || !ANYOF_BITMAP_TEST(data->start_class, PL_fold_latin1[uc])))
3646 ANYOF_CLASS_ZERO(data->start_class);
3647 ANYOF_BITMAP_ZERO(data->start_class);
3649 ANYOF_BITMAP_SET(data->start_class, uc);
3650 else if (uc >= 0x100) {
3653 /* Some Unicode code points fold to the Latin1 range; as
3654 * XXX temporary code, instead of figuring out if this is
3655 * one, just assume it is and set all the start class bits
3656 * that could be some such above 255 code point's fold
3657 * which will generate fals positives. As the code
3658 * elsewhere that does compute the fold settles down, it
3659 * can be extracted out and re-used here */
3660 for (i = 0; i < 256; i++){
3661 if (HAS_NONLATIN1_FOLD_CLOSURE(i)) {
3662 ANYOF_BITMAP_SET(data->start_class, i);
3666 CLEAR_SSC_EOS(data->start_class);
3668 data->start_class->flags &= ~ANYOF_UNICODE_ALL;
3670 else if (flags & SCF_DO_STCLASS_OR) {
3671 /* false positive possible if the class is case-folded */
3673 ANYOF_BITMAP_SET(data->start_class, uc);
3675 data->start_class->flags |= ANYOF_UNICODE_ALL;
3676 CLEAR_SSC_EOS(data->start_class);
3677 cl_and(data->start_class, and_withp);
3679 flags &= ~SCF_DO_STCLASS;
3681 else if (PL_regkind[OP(scan)] == EXACT) { /* But OP != EXACT! */
3682 I32 l = STR_LEN(scan);
3683 UV uc = *((U8*)STRING(scan));
3685 /* Search for fixed substrings supports EXACT only. */
3686 if (flags & SCF_DO_SUBSTR) {
3688 SCAN_COMMIT(pRExC_state, data, minlenp);
3691 const U8 * const s = (U8 *)STRING(scan);
3692 uc = utf8_to_uvchr_buf(s, s + l, NULL);
3693 l = utf8_length(s, s + l);
3695 if (has_exactf_sharp_s) {
3696 RExC_seen |= REG_SEEN_EXACTF_SHARP_S;
3698 min += l - min_subtract;
3700 delta += min_subtract;
3701 if (flags & SCF_DO_SUBSTR) {
3702 data->pos_min += l - min_subtract;
3703 if (data->pos_min < 0) {
3706 data->pos_delta += min_subtract;
3708 data->longest = &(data->longest_float);
3711 if (flags & SCF_DO_STCLASS_AND) {
3712 /* Check whether it is compatible with what we know already! */
3715 (!(data->start_class->flags & ANYOF_LOCALE)
3716 && !ANYOF_BITMAP_TEST(data->start_class, uc)
3717 && !ANYOF_BITMAP_TEST(data->start_class, PL_fold_latin1[uc])))
3721 ANYOF_CLASS_ZERO(data->start_class);
3722 ANYOF_BITMAP_ZERO(data->start_class);
3724 ANYOF_BITMAP_SET(data->start_class, uc);
3725 CLEAR_SSC_EOS(data->start_class);
3726 if (OP(scan) == EXACTFL) {
3727 /* XXX This set is probably no longer necessary, and
3728 * probably wrong as LOCALE now is on in the initial
3730 data->start_class->flags |= ANYOF_LOCALE|ANYOF_LOC_FOLD;
3734 /* Also set the other member of the fold pair. In case
3735 * that unicode semantics is called for at runtime, use
3736 * the full latin1 fold. (Can't do this for locale,
3737 * because not known until runtime) */
3738 ANYOF_BITMAP_SET(data->start_class, PL_fold_latin1[uc]);
3740 /* All other (EXACTFL handled above) folds except under
3741 * /iaa that include s, S, and sharp_s also may include
3743 if (OP(scan) != EXACTFA) {
3744 if (uc == 's' || uc == 'S') {
3745 ANYOF_BITMAP_SET(data->start_class,
3746 LATIN_SMALL_LETTER_SHARP_S);
3748 else if (uc == LATIN_SMALL_LETTER_SHARP_S) {
3749 ANYOF_BITMAP_SET(data->start_class, 's');
3750 ANYOF_BITMAP_SET(data->start_class, 'S');
3755 else if (uc >= 0x100) {
3757 for (i = 0; i < 256; i++){
3758 if (_HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(i)) {
3759 ANYOF_BITMAP_SET(data->start_class, i);
3764 else if (flags & SCF_DO_STCLASS_OR) {
3765 if (data->start_class->flags & ANYOF_LOC_FOLD) {
3766 /* false positive possible if the class is case-folded.
3767 Assume that the locale settings are the same... */
3769 ANYOF_BITMAP_SET(data->start_class, uc);
3770 if (OP(scan) != EXACTFL) {
3772 /* And set the other member of the fold pair, but
3773 * can't do that in locale because not known until
3775 ANYOF_BITMAP_SET(data->start_class,
3776 PL_fold_latin1[uc]);
3778 /* All folds except under /iaa that include s, S,
3779 * and sharp_s also may include the others */
3780 if (OP(scan) != EXACTFA) {
3781 if (uc == 's' || uc == 'S') {
3782 ANYOF_BITMAP_SET(data->start_class,
3783 LATIN_SMALL_LETTER_SHARP_S);
3785 else if (uc == LATIN_SMALL_LETTER_SHARP_S) {
3786 ANYOF_BITMAP_SET(data->start_class, 's');
3787 ANYOF_BITMAP_SET(data->start_class, 'S');
3792 CLEAR_SSC_EOS(data->start_class);
3794 cl_and(data->start_class, and_withp);
3796 flags &= ~SCF_DO_STCLASS;
3798 else if (REGNODE_VARIES(OP(scan))) {
3799 I32 mincount, maxcount, minnext, deltanext, fl = 0;
3800 I32 f = flags, pos_before = 0;
3801 regnode * const oscan = scan;
3802 struct regnode_charclass_class this_class;
3803 struct regnode_charclass_class *oclass = NULL;
3804 I32 next_is_eval = 0;
3806 switch (PL_regkind[OP(scan)]) {
3807 case WHILEM: /* End of (?:...)* . */
3808 scan = NEXTOPER(scan);
3811 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
3812 next = NEXTOPER(scan);
3813 if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
3815 maxcount = REG_INFTY;
3816 next = regnext(scan);
3817 scan = NEXTOPER(scan);
3821 if (flags & SCF_DO_SUBSTR)
3826 if (flags & SCF_DO_STCLASS) {
3828 maxcount = REG_INFTY;
3829 next = regnext(scan);
3830 scan = NEXTOPER(scan);
3833 is_inf = is_inf_internal = 1;
3834 scan = regnext(scan);
3835 if (flags & SCF_DO_SUBSTR) {
3836 SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot extend fixed substrings */
3837 data->longest = &(data->longest_float);
3839 goto optimize_curly_tail;
3841 if (stopparen>0 && (OP(scan)==CURLYN || OP(scan)==CURLYM)
3842 && (scan->flags == stopparen))
3847 mincount = ARG1(scan);
3848 maxcount = ARG2(scan);
3850 next = regnext(scan);
3851 if (OP(scan) == CURLYX) {
3852 I32 lp = (data ? *(data->last_closep) : 0);
3853 scan->flags = ((lp <= (I32)U8_MAX) ? (U8)lp : U8_MAX);
3855 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
3856 next_is_eval = (OP(scan) == EVAL);
3858 if (flags & SCF_DO_SUBSTR) {
3859 if (mincount == 0) SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot extend fixed substrings */
3860 pos_before = data->pos_min;
3864 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
3866 data->flags |= SF_IS_INF;
3868 if (flags & SCF_DO_STCLASS) {
3869 cl_init(pRExC_state, &this_class);
3870 oclass = data->start_class;
3871 data->start_class = &this_class;
3872 f |= SCF_DO_STCLASS_AND;
3873 f &= ~SCF_DO_STCLASS_OR;
3875 /* Exclude from super-linear cache processing any {n,m}
3876 regops for which the combination of input pos and regex
3877 pos is not enough information to determine if a match
3880 For example, in the regex /foo(bar\s*){4,8}baz/ with the
3881 regex pos at the \s*, the prospects for a match depend not
3882 only on the input position but also on how many (bar\s*)
3883 repeats into the {4,8} we are. */
3884 if ((mincount > 1) || (maxcount > 1 && maxcount != REG_INFTY))
3885 f &= ~SCF_WHILEM_VISITED_POS;
3887 /* This will finish on WHILEM, setting scan, or on NULL: */
3888 minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext,
3889 last, data, stopparen, recursed, NULL,
3891 ? (f & ~SCF_DO_SUBSTR) : f),depth+1);
3893 if (flags & SCF_DO_STCLASS)
3894 data->start_class = oclass;
3895 if (mincount == 0 || minnext == 0) {
3896 if (flags & SCF_DO_STCLASS_OR) {
3897 cl_or(pRExC_state, data->start_class, &this_class);
3899 else if (flags & SCF_DO_STCLASS_AND) {
3900 /* Switch to OR mode: cache the old value of
3901 * data->start_class */
3903 StructCopy(data->start_class, and_withp,
3904 struct regnode_charclass_class);
3905 flags &= ~SCF_DO_STCLASS_AND;
3906 StructCopy(&this_class, data->start_class,
3907 struct regnode_charclass_class);
3908 flags |= SCF_DO_STCLASS_OR;
3909 SET_SSC_EOS(data->start_class);
3911 } else { /* Non-zero len */
3912 if (flags & SCF_DO_STCLASS_OR) {
3913 cl_or(pRExC_state, data->start_class, &this_class);
3914 cl_and(data->start_class, and_withp);
3916 else if (flags & SCF_DO_STCLASS_AND)
3917 cl_and(data->start_class, &this_class);
3918 flags &= ~SCF_DO_STCLASS;
3920 if (!scan) /* It was not CURLYX, but CURLY. */
3922 if (!(flags & SCF_TRIE_DOING_RESTUDY)
3923 /* ? quantifier ok, except for (?{ ... }) */
3924 && (next_is_eval || !(mincount == 0 && maxcount == 1))
3925 && (minnext == 0) && (deltanext == 0)
3926 && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
3927 && maxcount <= REG_INFTY/3) /* Complement check for big count */
3929 /* Fatal warnings may leak the regexp without this: */
3930 SAVEFREESV(RExC_rx_sv);
3931 ckWARNreg(RExC_parse,
3932 "Quantifier unexpected on zero-length expression");
3933 (void)ReREFCNT_inc(RExC_rx_sv);
3936 min += minnext * mincount;
3937 is_inf_internal |= deltanext == I32_MAX
3938 || (maxcount == REG_INFTY && minnext + deltanext > 0);
3939 is_inf |= is_inf_internal;
3943 delta += (minnext + deltanext) * maxcount - minnext * mincount;
3945 /* Try powerful optimization CURLYX => CURLYN. */
3946 if ( OP(oscan) == CURLYX && data
3947 && data->flags & SF_IN_PAR
3948 && !(data->flags & SF_HAS_EVAL)
3949 && !deltanext && minnext == 1 ) {
3950 /* Try to optimize to CURLYN. */
3951 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
3952 regnode * const nxt1 = nxt;
3959 if (!REGNODE_SIMPLE(OP(nxt))
3960 && !(PL_regkind[OP(nxt)] == EXACT
3961 && STR_LEN(nxt) == 1))
3967 if (OP(nxt) != CLOSE)
3969 if (RExC_open_parens) {
3970 RExC_open_parens[ARG(nxt1)-1]=oscan; /*open->CURLYM*/
3971 RExC_close_parens[ARG(nxt1)-1]=nxt+2; /*close->while*/
3973 /* Now we know that nxt2 is the only contents: */
3974 oscan->flags = (U8)ARG(nxt);
3976 OP(nxt1) = NOTHING; /* was OPEN. */
3979 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
3980 NEXT_OFF(nxt1+ 1) = 0; /* just for consistency. */
3981 NEXT_OFF(nxt2) = 0; /* just for consistency with CURLY. */
3982 OP(nxt) = OPTIMIZED; /* was CLOSE. */
3983 OP(nxt + 1) = OPTIMIZED; /* was count. */
3984 NEXT_OFF(nxt+ 1) = 0; /* just for consistency. */
3989 /* Try optimization CURLYX => CURLYM. */
3990 if ( OP(oscan) == CURLYX && data
3991 && !(data->flags & SF_HAS_PAR)
3992 && !(data->flags & SF_HAS_EVAL)
3993 && !deltanext /* atom is fixed width */
3994 && minnext != 0 /* CURLYM can't handle zero width */
3995 && ! (RExC_seen & REG_SEEN_EXACTF_SHARP_S) /* Nor \xDF */
3997 /* XXXX How to optimize if data == 0? */
3998 /* Optimize to a simpler form. */
3999 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
4003 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
4004 && (OP(nxt2) != WHILEM))
4006 OP(nxt2) = SUCCEED; /* Whas WHILEM */
4007 /* Need to optimize away parenths. */
4008 if ((data->flags & SF_IN_PAR) && OP(nxt) == CLOSE) {
4009 /* Set the parenth number. */
4010 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
4012 oscan->flags = (U8)ARG(nxt);
4013 if (RExC_open_parens) {
4014 RExC_open_parens[ARG(nxt1)-1]=oscan; /*open->CURLYM*/
4015 RExC_close_parens[ARG(nxt1)-1]=nxt2+1; /*close->NOTHING*/
4017 OP(nxt1) = OPTIMIZED; /* was OPEN. */
4018 OP(nxt) = OPTIMIZED; /* was CLOSE. */
4021 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
4022 OP(nxt + 1) = OPTIMIZED; /* was count. */
4023 NEXT_OFF(nxt1 + 1) = 0; /* just for consistency. */
4024 NEXT_OFF(nxt + 1) = 0; /* just for consistency. */
4027 while ( nxt1 && (OP(nxt1) != WHILEM)) {
4028 regnode *nnxt = regnext(nxt1);
4030 if (reg_off_by_arg[OP(nxt1)])
4031 ARG_SET(nxt1, nxt2 - nxt1);
4032 else if (nxt2 - nxt1 < U16_MAX)
4033 NEXT_OFF(nxt1) = nxt2 - nxt1;
4035 OP(nxt) = NOTHING; /* Cannot beautify */
4040 /* Optimize again: */
4041 study_chunk(pRExC_state, &nxt1, minlenp, &deltanext, nxt,
4042 NULL, stopparen, recursed, NULL, 0,depth+1);
4047 else if ((OP(oscan) == CURLYX)
4048 && (flags & SCF_WHILEM_VISITED_POS)
4049 /* See the comment on a similar expression above.
4050 However, this time it's not a subexpression
4051 we care about, but the expression itself. */
4052 && (maxcount == REG_INFTY)
4053 && data && ++data->whilem_c < 16) {
4054 /* This stays as CURLYX, we can put the count/of pair. */
4055 /* Find WHILEM (as in regexec.c) */
4056 regnode *nxt = oscan + NEXT_OFF(oscan);
4058 if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
4060 PREVOPER(nxt)->flags = (U8)(data->whilem_c
4061 | (RExC_whilem_seen << 4)); /* On WHILEM */
4063 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
4065 if (flags & SCF_DO_SUBSTR) {
4066 SV *last_str = NULL;
4067 int counted = mincount != 0;
4069 if (data->last_end > 0 && mincount != 0) { /* Ends with a string. */
4070 #if defined(SPARC64_GCC_WORKAROUND)
4073 const char *s = NULL;
4076 if (pos_before >= data->last_start_min)
4079 b = data->last_start_min;
4082 s = SvPV_const(data->last_found, l);
4083 old = b - data->last_start_min;
4086 I32 b = pos_before >= data->last_start_min
4087 ? pos_before : data->last_start_min;
4089 const char * const s = SvPV_const(data->last_found, l);
4090 I32 old = b - data->last_start_min;
4094 old = utf8_hop((U8*)s, old) - (U8*)s;
4096 /* Get the added string: */
4097 last_str = newSVpvn_utf8(s + old, l, UTF);
4098 if (deltanext == 0 && pos_before == b) {
4099 /* What was added is a constant string */
4101 SvGROW(last_str, (mincount * l) + 1);
4102 repeatcpy(SvPVX(last_str) + l,
4103 SvPVX_const(last_str), l, mincount - 1);
4104 SvCUR_set(last_str, SvCUR(last_str) * mincount);
4105 /* Add additional parts. */
4106 SvCUR_set(data->last_found,
4107 SvCUR(data->last_found) - l);
4108 sv_catsv(data->last_found, last_str);
4110 SV * sv = data->last_found;
4112 SvUTF8(sv) && SvMAGICAL(sv) ?
4113 mg_find(sv, PERL_MAGIC_utf8) : NULL;
4114 if (mg && mg->mg_len >= 0)
4115 mg->mg_len += CHR_SVLEN(last_str) - l;
4117 data->last_end += l * (mincount - 1);
4120 /* start offset must point into the last copy */
4121 data->last_start_min += minnext * (mincount - 1);
4122 data->last_start_max += is_inf ? I32_MAX
4123 : (maxcount - 1) * (minnext + data->pos_delta);
4126 /* It is counted once already... */
4127 data->pos_min += minnext * (mincount - counted);
4129 PerlIO_printf(Perl_debug_log, "counted=%d deltanext=%d I32_MAX=%d minnext=%d maxcount=%d mincount=%d\n",
4130 counted, deltanext, I32_MAX, minnext, maxcount, mincount);
4131 if (deltanext != I32_MAX)
4132 PerlIO_printf(Perl_debug_log, "LHS=%d RHS=%d\n", -counted * deltanext + (minnext + deltanext) * maxcount - minnext * mincount, I32_MAX - data->pos_delta);
4134 if (deltanext == I32_MAX || -counted * deltanext + (minnext + deltanext) * maxcount - minnext * mincount >= I32_MAX - data->pos_delta)
4135 data->pos_delta = I32_MAX;
4137 data->pos_delta += - counted * deltanext +
4138 (minnext + deltanext) * maxcount - minnext * mincount;
4139 if (mincount != maxcount) {
4140 /* Cannot extend fixed substrings found inside
4142 SCAN_COMMIT(pRExC_state,data,minlenp);
4143 if (mincount && last_str) {
4144 SV * const sv = data->last_found;
4145 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
4146 mg_find(sv, PERL_MAGIC_utf8) : NULL;
4150 sv_setsv(sv, last_str);
4151 data->last_end = data->pos_min;
4152 data->last_start_min =
4153 data->pos_min - CHR_SVLEN(last_str);
4154 data->last_start_max = is_inf
4156 : data->pos_min + data->pos_delta
4157 - CHR_SVLEN(last_str);
4159 data->longest = &(data->longest_float);
4161 SvREFCNT_dec(last_str);
4163 if (data && (fl & SF_HAS_EVAL))
4164 data->flags |= SF_HAS_EVAL;
4165 optimize_curly_tail:
4166 if (OP(oscan) != CURLYX) {
4167 while (PL_regkind[OP(next = regnext(oscan))] == NOTHING
4169 NEXT_OFF(oscan) += NEXT_OFF(next);
4172 default: /* REF, and CLUMP only? */
4173 if (flags & SCF_DO_SUBSTR) {
4174 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
4175 data->longest = &(data->longest_float);
4177 is_inf = is_inf_internal = 1;
4178 if (flags & SCF_DO_STCLASS_OR)
4179 cl_anything(pRExC_state, data->start_class);
4180 flags &= ~SCF_DO_STCLASS;
4184 else if (OP(scan) == LNBREAK) {
4185 if (flags & SCF_DO_STCLASS) {
4187 CLEAR_SSC_EOS(data->start_class); /* No match on empty */
4188 if (flags & SCF_DO_STCLASS_AND) {
4189 for (value = 0; value < 256; value++)
4190 if (!is_VERTWS_cp(value))
4191 ANYOF_BITMAP_CLEAR(data->start_class, value);
4194 for (value = 0; value < 256; value++)
4195 if (is_VERTWS_cp(value))
4196 ANYOF_BITMAP_SET(data->start_class, value);
4198 if (flags & SCF_DO_STCLASS_OR)
4199 cl_and(data->start_class, and_withp);
4200 flags &= ~SCF_DO_STCLASS;
4203 delta++; /* Because of the 2 char string cr-lf */
4204 if (flags & SCF_DO_SUBSTR) {
4205 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
4207 data->pos_delta += 1;
4208 data->longest = &(data->longest_float);
4211 else if (REGNODE_SIMPLE(OP(scan))) {
4214 if (flags & SCF_DO_SUBSTR) {
4215 SCAN_COMMIT(pRExC_state,data,minlenp);
4219 if (flags & SCF_DO_STCLASS) {
4221 CLEAR_SSC_EOS(data->start_class); /* No match on empty */
4223 /* Some of the logic below assumes that switching
4224 locale on will only add false positives. */
4225 switch (PL_regkind[OP(scan)]) {
4231 Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d", OP(scan));
4234 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
4235 cl_anything(pRExC_state, data->start_class);
4238 if (OP(scan) == SANY)
4240 if (flags & SCF_DO_STCLASS_OR) { /* Everything but \n */
4241 value = (ANYOF_BITMAP_TEST(data->start_class,'\n')
4242 || ANYOF_CLASS_TEST_ANY_SET(data->start_class));
4243 cl_anything(pRExC_state, data->start_class);
4245 if (flags & SCF_DO_STCLASS_AND || !value)
4246 ANYOF_BITMAP_CLEAR(data->start_class,'\n');
4249 if (flags & SCF_DO_STCLASS_AND)
4250 cl_and(data->start_class,
4251 (struct regnode_charclass_class*)scan);
4253 cl_or(pRExC_state, data->start_class,
4254 (struct regnode_charclass_class*)scan);
4262 classnum = FLAGS(scan);
4263 if (flags & SCF_DO_STCLASS_AND) {
4264 if (!(data->start_class->flags & ANYOF_LOCALE)) {
4265 ANYOF_CLASS_CLEAR(data->start_class, classnum_to_namedclass(classnum) + 1);
4266 for (value = 0; value < loop_max; value++) {
4267 if (! _generic_isCC(UNI_TO_NATIVE(value), classnum)) {
4268 ANYOF_BITMAP_CLEAR(data->start_class, UNI_TO_NATIVE(value));
4274 if (data->start_class->flags & ANYOF_LOCALE) {
4275 ANYOF_CLASS_SET(data->start_class, classnum_to_namedclass(classnum));
4279 /* Even if under locale, set the bits for non-locale
4280 * in case it isn't a true locale-node. This will
4281 * create false positives if it truly is locale */
4282 for (value = 0; value < loop_max; value++) {
4283 if (_generic_isCC(UNI_TO_NATIVE(value), classnum)) {
4284 ANYOF_BITMAP_SET(data->start_class, UNI_TO_NATIVE(value));
4296 classnum = FLAGS(scan);
4297 if (flags & SCF_DO_STCLASS_AND) {
4298 if (!(data->start_class->flags & ANYOF_LOCALE)) {
4299 ANYOF_CLASS_CLEAR(data->start_class, classnum_to_namedclass(classnum));
4300 for (value = 0; value < loop_max; value++) {
4301 if (_generic_isCC(UNI_TO_NATIVE(value), classnum)) {
4302 ANYOF_BITMAP_CLEAR(data->start_class, UNI_TO_NATIVE(value));
4308 if (data->start_class->flags & ANYOF_LOCALE) {
4309 ANYOF_CLASS_SET(data->start_class, classnum_to_namedclass(classnum) + 1);
4313 /* Even if under locale, set the bits for non-locale in
4314 * case it isn't a true locale-node. This will create
4315 * false positives if it truly is locale */
4316 for (value = 0; value < loop_max; value++) {
4317 if (! _generic_isCC(UNI_TO_NATIVE(value), classnum)) {
4318 ANYOF_BITMAP_SET(data->start_class, UNI_TO_NATIVE(value));
4321 if (PL_regkind[OP(scan)] == NPOSIXD) {
4322 data->start_class->flags |= ANYOF_NON_UTF8_LATIN1_ALL;
4328 if (flags & SCF_DO_STCLASS_OR)
4329 cl_and(data->start_class, and_withp);
4330 flags &= ~SCF_DO_STCLASS;
4333 else if (PL_regkind[OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
4334 data->flags |= (OP(scan) == MEOL
4337 SCAN_COMMIT(pRExC_state, data, minlenp);
4340 else if ( PL_regkind[OP(scan)] == BRANCHJ
4341 /* Lookbehind, or need to calculate parens/evals/stclass: */
4342 && (scan->flags || data || (flags & SCF_DO_STCLASS))
4343 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
4344 if ( OP(scan) == UNLESSM &&
4346 OP(NEXTOPER(NEXTOPER(scan))) == NOTHING &&
4347 OP(regnext(NEXTOPER(NEXTOPER(scan)))) == SUCCEED
4350 regnode *upto= regnext(scan);
4352 SV * const mysv_val=sv_newmortal();
4353 DEBUG_STUDYDATA("OPFAIL",data,depth);
4355 /*DEBUG_PARSE_MSG("opfail");*/
4356 regprop(RExC_rx, mysv_val, upto);
4357 PerlIO_printf(Perl_debug_log, "~ replace with OPFAIL pointed at %s (%"IVdf") offset %"IVdf"\n",
4358 SvPV_nolen_const(mysv_val),
4359 (IV)REG_NODE_NUM(upto),
4364 NEXT_OFF(scan) = upto - scan;
4365 for (opt= scan + 1; opt < upto ; opt++)
4366 OP(opt) = OPTIMIZED;
4370 if ( !PERL_ENABLE_POSITIVE_ASSERTION_STUDY
4371 || OP(scan) == UNLESSM )
4373 /* Negative Lookahead/lookbehind
4374 In this case we can't do fixed string optimisation.
4377 I32 deltanext, minnext, fake = 0;
4379 struct regnode_charclass_class intrnl;
4382 data_fake.flags = 0;
4384 data_fake.whilem_c = data->whilem_c;
4385 data_fake.last_closep = data->last_closep;
4388 data_fake.last_closep = &fake;
4389 data_fake.pos_delta = delta;
4390 if ( flags & SCF_DO_STCLASS && !scan->flags
4391 && OP(scan) == IFMATCH ) { /* Lookahead */
4392 cl_init(pRExC_state, &intrnl);
4393 data_fake.start_class = &intrnl;
4394 f |= SCF_DO_STCLASS_AND;
4396 if (flags & SCF_WHILEM_VISITED_POS)
4397 f |= SCF_WHILEM_VISITED_POS;
4398 next = regnext(scan);
4399 nscan = NEXTOPER(NEXTOPER(scan));
4400 minnext = study_chunk(pRExC_state, &nscan, minlenp, &deltanext,
4401 last, &data_fake, stopparen, recursed, NULL, f, depth+1);
4404 FAIL("Variable length lookbehind not implemented");
4406 else if (minnext > (I32)U8_MAX) {
4407 FAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
4409 scan->flags = (U8)minnext;
4412 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
4414 if (data_fake.flags & SF_HAS_EVAL)
4415 data->flags |= SF_HAS_EVAL;
4416 data->whilem_c = data_fake.whilem_c;
4418 if (f & SCF_DO_STCLASS_AND) {
4419 if (flags & SCF_DO_STCLASS_OR) {
4420 /* OR before, AND after: ideally we would recurse with
4421 * data_fake to get the AND applied by study of the
4422 * remainder of the pattern, and then derecurse;
4423 * *** HACK *** for now just treat as "no information".
4424 * See [perl #56690].
4426 cl_init(pRExC_state, data->start_class);
4428 /* AND before and after: combine and continue */
4429 const int was = TEST_SSC_EOS(data->start_class);
4431 cl_and(data->start_class, &intrnl);
4433 SET_SSC_EOS(data->start_class);
4437 #if PERL_ENABLE_POSITIVE_ASSERTION_STUDY
4439 /* Positive Lookahead/lookbehind
4440 In this case we can do fixed string optimisation,
4441 but we must be careful about it. Note in the case of
4442 lookbehind the positions will be offset by the minimum
4443 length of the pattern, something we won't know about
4444 until after the recurse.
4446 I32 deltanext, fake = 0;
4448 struct regnode_charclass_class intrnl;
4450 /* We use SAVEFREEPV so that when the full compile
4451 is finished perl will clean up the allocated
4452 minlens when it's all done. This way we don't
4453 have to worry about freeing them when we know
4454 they wont be used, which would be a pain.
4457 Newx( minnextp, 1, I32 );
4458 SAVEFREEPV(minnextp);
4461 StructCopy(data, &data_fake, scan_data_t);
4462 if ((flags & SCF_DO_SUBSTR) && data->last_found) {
4465 SCAN_COMMIT(pRExC_state, &data_fake,minlenp);
4466 data_fake.last_found=newSVsv(data->last_found);
4470 data_fake.last_closep = &fake;
4471 data_fake.flags = 0;
4472 data_fake.pos_delta = delta;
4474 data_fake.flags |= SF_IS_INF;
4475 if ( flags & SCF_DO_STCLASS && !scan->flags
4476 && OP(scan) == IFMATCH ) { /* Lookahead */
4477 cl_init(pRExC_state, &intrnl);
4478 data_fake.start_class = &intrnl;
4479 f |= SCF_DO_STCLASS_AND;
4481 if (flags & SCF_WHILEM_VISITED_POS)
4482 f |= SCF_WHILEM_VISITED_POS;
4483 next = regnext(scan);
4484 nscan = NEXTOPER(NEXTOPER(scan));
4486 *minnextp = study_chunk(pRExC_state, &nscan, minnextp, &deltanext,
4487 last, &data_fake, stopparen, recursed, NULL, f,depth+1);
4490 FAIL("Variable length lookbehind not implemented");
4492 else if (*minnextp > (I32)U8_MAX) {
4493 FAIL2("Lookbehind longer than %"UVuf" not implemented", (UV)U8_MAX);
4495 scan->flags = (U8)*minnextp;
4500 if (f & SCF_DO_STCLASS_AND) {
4501 const int was = TEST_SSC_EOS(data.start_class);
4503 cl_and(data->start_class, &intrnl);
4505 SET_SSC_EOS(data->start_class);
4508 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
4510 if (data_fake.flags & SF_HAS_EVAL)
4511 data->flags |= SF_HAS_EVAL;
4512 data->whilem_c = data_fake.whilem_c;
4513 if ((flags & SCF_DO_SUBSTR) && data_fake.last_found) {
4514 if (RExC_rx->minlen<*minnextp)
4515 RExC_rx->minlen=*minnextp;
4516 SCAN_COMMIT(pRExC_state, &data_fake, minnextp);
4517 SvREFCNT_dec_NN(data_fake.last_found);
4519 if ( data_fake.minlen_fixed != minlenp )
4521 data->offset_fixed= data_fake.offset_fixed;
4522 data->minlen_fixed= data_fake.minlen_fixed;
4523 data->lookbehind_fixed+= scan->flags;
4525 if ( data_fake.minlen_float != minlenp )
4527 data->minlen_float= data_fake.minlen_float;
4528 data->offset_float_min=data_fake.offset_float_min;
4529 data->offset_float_max=data_fake.offset_float_max;
4530 data->lookbehind_float+= scan->flags;
4537 else if (OP(scan) == OPEN) {
4538 if (stopparen != (I32)ARG(scan))
4541 else if (OP(scan) == CLOSE) {
4542 if (stopparen == (I32)ARG(scan)) {
4545 if ((I32)ARG(scan) == is_par) {
4546 next = regnext(scan);
4548 if ( next && (OP(next) != WHILEM) && next < last)
4549 is_par = 0; /* Disable optimization */
4552 *(data->last_closep) = ARG(scan);
4554 else if (OP(scan) == EVAL) {
4556 data->flags |= SF_HAS_EVAL;
4558 else if ( PL_regkind[OP(scan)] == ENDLIKE ) {
4559 if (flags & SCF_DO_SUBSTR) {
4560 SCAN_COMMIT(pRExC_state,data,minlenp);
4561 flags &= ~SCF_DO_SUBSTR;
4563 if (data && OP(scan)==ACCEPT) {
4564 data->flags |= SCF_SEEN_ACCEPT;
4569 else if (OP(scan) == LOGICAL && scan->flags == 2) /* Embedded follows */
4571 if (flags & SCF_DO_SUBSTR) {
4572 SCAN_COMMIT(pRExC_state,data,minlenp);
4573 data->longest = &(data->longest_float);
4575 is_inf = is_inf_internal = 1;
4576 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
4577 cl_anything(pRExC_state, data->start_class);
4578 flags &= ~SCF_DO_STCLASS;
4580 else if (OP(scan) == GPOS) {
4581 if (!(RExC_rx->extflags & RXf_GPOS_FLOAT) &&
4582 !(delta || is_inf || (data && data->pos_delta)))
4584 if (!(RExC_rx->extflags & RXf_ANCH) && (flags & SCF_DO_SUBSTR))
4585 RExC_rx->extflags |= RXf_ANCH_GPOS;
4586 if (RExC_rx->gofs < (U32)min)
4587 RExC_rx->gofs = min;
4589 RExC_rx->extflags |= RXf_GPOS_FLOAT;
4593 #ifdef TRIE_STUDY_OPT
4594 #ifdef FULL_TRIE_STUDY
4595 else if (PL_regkind[OP(scan)] == TRIE) {
4596 /* NOTE - There is similar code to this block above for handling
4597 BRANCH nodes on the initial study. If you change stuff here
4599 regnode *trie_node= scan;
4600 regnode *tail= regnext(scan);
4601 reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
4602 I32 max1 = 0, min1 = I32_MAX;
4603 struct regnode_charclass_class accum;
4605 if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
4606 SCAN_COMMIT(pRExC_state, data,minlenp); /* Cannot merge strings after this. */
4607 if (flags & SCF_DO_STCLASS)
4608 cl_init_zero(pRExC_state, &accum);
4614 const regnode *nextbranch= NULL;
4617 for ( word=1 ; word <= trie->wordcount ; word++)
4619 I32 deltanext=0, minnext=0, f = 0, fake;
4620 struct regnode_charclass_class this_class;
4622 data_fake.flags = 0;
4624 data_fake.whilem_c = data->whilem_c;
4625 data_fake.last_closep = data->last_closep;
4628 data_fake.last_closep = &fake;
4629 data_fake.pos_delta = delta;
4630 if (flags & SCF_DO_STCLASS) {
4631 cl_init(pRExC_state, &this_class);
4632 data_fake.start_class = &this_class;
4633 f = SCF_DO_STCLASS_AND;
4635 if (flags & SCF_WHILEM_VISITED_POS)
4636 f |= SCF_WHILEM_VISITED_POS;
4638 if (trie->jump[word]) {
4640 nextbranch = trie_node + trie->jump[0];
4641 scan= trie_node + trie->jump[word];
4642 /* We go from the jump point to the branch that follows
4643 it. Note this means we need the vestigal unused branches
4644 even though they arent otherwise used.
4646 minnext = study_chunk(pRExC_state, &scan, minlenp,
4647 &deltanext, (regnode *)nextbranch, &data_fake,
4648 stopparen, recursed, NULL, f,depth+1);
4650 if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
4651 nextbranch= regnext((regnode*)nextbranch);
4653 if (min1 > (I32)(minnext + trie->minlen))
4654 min1 = minnext + trie->minlen;
4655 if (deltanext == I32_MAX) {
4656 is_inf = is_inf_internal = 1;
4658 } else if (max1 < (I32)(minnext + deltanext + trie->maxlen))
4659 max1 = minnext + deltanext + trie->maxlen;
4661 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
4663 if (data_fake.flags & SCF_SEEN_ACCEPT) {
4664 if ( stopmin > min + min1)
4665 stopmin = min + min1;
4666 flags &= ~SCF_DO_SUBSTR;
4668 data->flags |= SCF_SEEN_ACCEPT;
4671 if (data_fake.flags & SF_HAS_EVAL)
4672 data->flags |= SF_HAS_EVAL;
4673 data->whilem_c = data_fake.whilem_c;
4675 if (flags & SCF_DO_STCLASS)
4676 cl_or(pRExC_state, &accum, &this_class);
4679 if (flags & SCF_DO_SUBSTR) {
4680 data->pos_min += min1;
4681 data->pos_delta += max1 - min1;
4682 if (max1 != min1 || is_inf)
4683 data->longest = &(data->longest_float);
4686 delta += max1 - min1;
4687 if (flags & SCF_DO_STCLASS_OR) {
4688 cl_or(pRExC_state, data->start_class, &accum);
4690 cl_and(data->start_class, and_withp);
4691 flags &= ~SCF_DO_STCLASS;
4694 else if (flags & SCF_DO_STCLASS_AND) {
4696 cl_and(data->start_class, &accum);
4697 flags &= ~SCF_DO_STCLASS;
4700 /* Switch to OR mode: cache the old value of
4701 * data->start_class */
4703 StructCopy(data->start_class, and_withp,
4704 struct regnode_charclass_class);
4705 flags &= ~SCF_DO_STCLASS_AND;
4706 StructCopy(&accum, data->start_class,
4707 struct regnode_charclass_class);
4708 flags |= SCF_DO_STCLASS_OR;
4709 SET_SSC_EOS(data->start_class);
4716 else if (PL_regkind[OP(scan)] == TRIE) {
4717 reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
4720 min += trie->minlen;
4721 delta += (trie->maxlen - trie->minlen);
4722 flags &= ~SCF_DO_STCLASS; /* xxx */
4723 if (flags & SCF_DO_SUBSTR) {
4724 SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect anything... */
4725 data->pos_min += trie->minlen;
4726 data->pos_delta += (trie->maxlen - trie->minlen);
4727 if (trie->maxlen != trie->minlen)
4728 data->longest = &(data->longest_float);
4730 if (trie->jump) /* no more substrings -- for now /grr*/
4731 flags &= ~SCF_DO_SUBSTR;
4733 #endif /* old or new */
4734 #endif /* TRIE_STUDY_OPT */
4736 /* Else: zero-length, ignore. */
4737 scan = regnext(scan);
4742 stopparen = frame->stop;
4743 frame = frame->prev;
4744 goto fake_study_recurse;
4749 DEBUG_STUDYDATA("pre-fin:",data,depth);
4752 *deltap = is_inf_internal ? I32_MAX : delta;
4753 if (flags & SCF_DO_SUBSTR && is_inf)
4754 data->pos_delta = I32_MAX - data->pos_min;
4755 if (is_par > (I32)U8_MAX)
4757 if (is_par && pars==1 && data) {
4758 data->flags |= SF_IN_PAR;
4759 data->flags &= ~SF_HAS_PAR;
4761 else if (pars && data) {
4762 data->flags |= SF_HAS_PAR;
4763 data->flags &= ~SF_IN_PAR;
4765 if (flags & SCF_DO_STCLASS_OR)
4766 cl_and(data->start_class, and_withp);
4767 if (flags & SCF_TRIE_RESTUDY)
4768 data->flags |= SCF_TRIE_RESTUDY;
4770 DEBUG_STUDYDATA("post-fin:",data,depth);
4772 return min < stopmin ? min : stopmin;
4776 S_add_data(RExC_state_t *pRExC_state, U32 n, const char *s)
4778 U32 count = RExC_rxi->data ? RExC_rxi->data->count : 0;
4780 PERL_ARGS_ASSERT_ADD_DATA;
4782 Renewc(RExC_rxi->data,
4783 sizeof(*RExC_rxi->data) + sizeof(void*) * (count + n - 1),
4784 char, struct reg_data);
4786 Renew(RExC_rxi->data->what, count + n, U8);
4788 Newx(RExC_rxi->data->what, n, U8);
4789 RExC_rxi->data->count = count + n;
4790 Copy(s, RExC_rxi->data->what + count, n, U8);
4794 /*XXX: todo make this not included in a non debugging perl */
4795 #ifndef PERL_IN_XSUB_RE
4797 Perl_reginitcolors(pTHX)
4800 const char * const s = PerlEnv_getenv("PERL_RE_COLORS");
4802 char *t = savepv(s);
4806 t = strchr(t, '\t');
4812 PL_colors[i] = t = (char *)"";
4817 PL_colors[i++] = (char *)"";
4824 #ifdef TRIE_STUDY_OPT
4825 #define CHECK_RESTUDY_GOTO_butfirst(dOsomething) \
4828 (data.flags & SCF_TRIE_RESTUDY) \
4836 #define CHECK_RESTUDY_GOTO_butfirst
4840 * pregcomp - compile a regular expression into internal code
4842 * Decides which engine's compiler to call based on the hint currently in
4846 #ifndef PERL_IN_XSUB_RE
4848 /* return the currently in-scope regex engine (or the default if none) */
4850 regexp_engine const *
4851 Perl_current_re_engine(pTHX)
4855 if (IN_PERL_COMPILETIME) {
4856 HV * const table = GvHV(PL_hintgv);
4860 return &reh_regexp_engine;
4861 ptr = hv_fetchs(table, "regcomp", FALSE);
4862 if ( !(ptr && SvIOK(*ptr) && SvIV(*ptr)))
4863 return &reh_regexp_engine;
4864 return INT2PTR(regexp_engine*,SvIV(*ptr));
4868 if (!PL_curcop->cop_hints_hash)
4869 return &reh_regexp_engine;
4870 ptr = cop_hints_fetch_pvs(PL_curcop, "regcomp", 0);
4871 if ( !(ptr && SvIOK(ptr) && SvIV(ptr)))
4872 return &reh_regexp_engine;
4873 return INT2PTR(regexp_engine*,SvIV(ptr));
4879 Perl_pregcomp(pTHX_ SV * const pattern, const U32 flags)
4882 regexp_engine const *eng = current_re_engine();
4883 GET_RE_DEBUG_FLAGS_DECL;
4885 PERL_ARGS_ASSERT_PREGCOMP;
4887 /* Dispatch a request to compile a regexp to correct regexp engine. */
4889 PerlIO_printf(Perl_debug_log, "Using engine %"UVxf"\n",
4892 return CALLREGCOMP_ENG(eng, pattern, flags);
4896 /* public(ish) entry point for the perl core's own regex compiling code.
4897 * It's actually a wrapper for Perl_re_op_compile that only takes an SV
4898 * pattern rather than a list of OPs, and uses the internal engine rather
4899 * than the current one */
4902 Perl_re_compile(pTHX_ SV * const pattern, U32 rx_flags)
4904 SV *pat = pattern; /* defeat constness! */
4905 PERL_ARGS_ASSERT_RE_COMPILE;
4906 return Perl_re_op_compile(aTHX_ &pat, 1, NULL,
4907 #ifdef PERL_IN_XSUB_RE
4912 NULL, NULL, rx_flags, 0);
4916 /* upgrade pattern pat_p of length plen_p to UTF8, and if there are code
4917 * blocks, recalculate the indices. Update pat_p and plen_p in-place to
4918 * point to the realloced string and length.
4920 * This is essentially a copy of Perl_bytes_to_utf8() with the code index
4924 S_pat_upgrade_to_utf8(pTHX_ RExC_state_t * const pRExC_state,
4925 char **pat_p, STRLEN *plen_p, int num_code_blocks)
4927 U8 *const src = (U8*)*pat_p;
4930 STRLEN s = 0, d = 0;
4932 GET_RE_DEBUG_FLAGS_DECL;
4934 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log,
4935 "UTF8 mismatch! Converting to utf8 for resizing and compile\n"));
4937 Newx(dst, *plen_p * 2 + 1, U8);
4939 while (s < *plen_p) {
4940 const UV uv = NATIVE_TO_ASCII(src[s]);
4941 if (UNI_IS_INVARIANT(uv))
4942 dst[d] = (U8)UTF_TO_NATIVE(uv);
4944 dst[d++] = (U8)UTF8_EIGHT_BIT_HI(uv);
4945 dst[d] = (U8)UTF8_EIGHT_BIT_LO(uv);
4947 if (n < num_code_blocks) {
4948 if (!do_end && pRExC_state->code_blocks[n].start == s) {
4949 pRExC_state->code_blocks[n].start = d;
4950 assert(dst[d] == '(');
4953 else if (do_end && pRExC_state->code_blocks[n].end == s) {
4954 pRExC_state->code_blocks[n].end = d;
4955 assert(dst[d] == ')');
4965 *pat_p = (char*) dst;
4967 RExC_orig_utf8 = RExC_utf8 = 1;
4972 /* S_concat_pat(): concatenate a list of args to the pattern string pat,
4973 * while recording any code block indices, and handling overloading,
4974 * nested qr// objects etc. If pat is null, it will allocate a new
4975 * string, or just return the first arg, if there's only one.
4977 * Returns the malloced/updated pat.
4978 * patternp and pat_count is the array of SVs to be concatted;
4979 * oplist is the optional list of ops that generated the SVs;
4980 * recompile_p is a pointer to a boolean that will be set if
4981 * the regex will need to be recompiled.
4982 * delim, if non-null is an SV that will be inserted between each element
4986 S_concat_pat(pTHX_ RExC_state_t * const pRExC_state,
4987 SV *pat, SV ** const patternp, int pat_count,
4988 OP *oplist, bool *recompile_p, SV *delim)
4992 bool use_delim = FALSE;
4993 bool alloced = FALSE;
4995 /* if we know we have at least two args, create an empty string,
4996 * then concatenate args to that. For no args, return an empty string */
4997 if (!pat && pat_count != 1) {
4998 pat = newSVpvn("", 0);
5003 for (svp = patternp; svp < patternp + pat_count; svp++) {
5006 STRLEN orig_patlen = 0;
5008 SV *msv = use_delim ? delim : *svp;
5010 /* if we've got a delimiter, we go round the loop twice for each
5011 * svp slot (except the last), using the delimiter the second
5020 if (SvTYPE(msv) == SVt_PVAV) {
5021 /* we've encountered an interpolated array within
5022 * the pattern, e.g. /...@a..../. Expand the list of elements,
5023 * then recursively append elements.
5024 * The code in this block is based on S_pushav() */
5026 AV *const av = (AV*)msv;
5027 const I32 maxarg = AvFILL(av) + 1;
5031 assert(oplist->op_type == OP_PADAV
5032 || oplist->op_type == OP_RV2AV);
5033 oplist = oplist->op_sibling;;
5036 if (SvRMAGICAL(av)) {
5039 Newx(array, maxarg, SV*);
5041 for (i=0; i < (U32)maxarg; i++) {
5042 SV ** const svp = av_fetch(av, i, FALSE);
5043 array[i] = svp ? *svp : &PL_sv_undef;
5047 array = AvARRAY(av);
5049 pat = S_concat_pat(aTHX_ pRExC_state, pat,
5050 array, maxarg, NULL, recompile_p,
5052 GvSV((gv_fetchpvs("\"", GV_ADDMULTI, SVt_PV))));
5058 /* we make the assumption here that each op in the list of
5059 * op_siblings maps to one SV pushed onto the stack,
5060 * except for code blocks, with have both an OP_NULL and
5062 * This allows us to match up the list of SVs against the
5063 * list of OPs to find the next code block.
5065 * Note that PUSHMARK PADSV PADSV ..
5067 * PADRANGE PADSV PADSV ..
5068 * so the alignment still works. */
5071 if (oplist->op_type == OP_NULL
5072 && (oplist->op_flags & OPf_SPECIAL))
5074 assert(n < pRExC_state->num_code_blocks);
5075 pRExC_state->code_blocks[n].start = pat ? SvCUR(pat) : 0;
5076 pRExC_state->code_blocks[n].block = oplist;
5077 pRExC_state->code_blocks[n].src_regex = NULL;
5080 oplist = oplist->op_sibling; /* skip CONST */
5083 oplist = oplist->op_sibling;;
5086 /* apply magic and QR overloading to arg */
5089 if (SvROK(msv) && SvAMAGIC(msv)) {
5090 SV *sv = AMG_CALLunary(msv, regexp_amg);
5094 if (SvTYPE(sv) != SVt_REGEXP)
5095 Perl_croak(aTHX_ "Overloaded qr did not return a REGEXP");
5100 /* try concatenation overload ... */
5101 if (pat && (SvAMAGIC(pat) || SvAMAGIC(msv)) &&
5102 (sv = amagic_call(pat, msv, concat_amg, AMGf_assign)))
5105 /* overloading involved: all bets are off over literal
5106 * code. Pretend we haven't seen it */
5107 pRExC_state->num_code_blocks -= n;
5111 /* ... or failing that, try "" overload */
5112 while (SvAMAGIC(msv)
5113 && (sv = AMG_CALLunary(msv, string_amg))
5117 && SvRV(msv) == SvRV(sv))
5122 if (SvROK(msv) && SvTYPE(SvRV(msv)) == SVt_REGEXP)
5126 /* this is a partially unrolled
5127 * sv_catsv_nomg(pat, msv);
5128 * that allows us to adjust code block indices if
5131 char *dst = SvPV_force_nomg(pat, dlen);
5133 if (SvUTF8(msv) && !SvUTF8(pat)) {
5134 S_pat_upgrade_to_utf8(aTHX_ pRExC_state, &dst, &dlen, n);
5135 sv_setpvn(pat, dst, dlen);
5138 sv_catsv_nomg(pat, msv);
5145 pRExC_state->code_blocks[n-1].end = SvCUR(pat)-1;
5148 /* extract any code blocks within any embedded qr//'s */
5149 if (rx && SvTYPE(rx) == SVt_REGEXP
5150 && RX_ENGINE((REGEXP*)rx)->op_comp)
5153 RXi_GET_DECL(ReANY((REGEXP *)rx), ri);
5154 if (ri->num_code_blocks) {
5156 /* the presence of an embedded qr// with code means
5157 * we should always recompile: the text of the
5158 * qr// may not have changed, but it may be a
5159 * different closure than last time */
5161 Renew(pRExC_state->code_blocks,
5162 pRExC_state->num_code_blocks + ri->num_code_blocks,
5163 struct reg_code_block);
5164 pRExC_state->num_code_blocks += ri->num_code_blocks;
5166 for (i=0; i < ri->num_code_blocks; i++) {
5167 struct reg_code_block *src, *dst;
5168 STRLEN offset = orig_patlen
5169 + ReANY((REGEXP *)rx)->pre_prefix;
5170 assert(n < pRExC_state->num_code_blocks);
5171 src = &ri->code_blocks[i];
5172 dst = &pRExC_state->code_blocks[n];
5173 dst->start = src->start + offset;
5174 dst->end = src->end + offset;
5175 dst->block = src->block;
5176 dst->src_regex = (REGEXP*) SvREFCNT_inc( (SV*)
5185 /* avoid calling magic multiple times on a single element e.g. =~ $qr */
5194 /* see if there are any run-time code blocks in the pattern.
5195 * False positives are allowed */
5198 S_has_runtime_code(pTHX_ RExC_state_t * const pRExC_state,
5199 char *pat, STRLEN plen)
5204 for (s = 0; s < plen; s++) {
5205 if (n < pRExC_state->num_code_blocks
5206 && s == pRExC_state->code_blocks[n].start)
5208 s = pRExC_state->code_blocks[n].end;
5212 /* TODO ideally should handle [..], (#..), /#.../x to reduce false
5214 if (pat[s] == '(' && s+2 <= plen && pat[s+1] == '?' &&
5216 || (s + 2 <= plen && pat[s+2] == '?' && pat[s+3] == '{'))
5223 /* Handle run-time code blocks. We will already have compiled any direct
5224 * or indirect literal code blocks. Now, take the pattern 'pat' and make a
5225 * copy of it, but with any literal code blocks blanked out and
5226 * appropriate chars escaped; then feed it into
5228 * eval "qr'modified_pattern'"
5232 * a\bc(?{"this was literal"})def'ghi\\jkl(?{"this is runtime"})mno
5236 * qr'a\\bc_______________________def\'ghi\\\\jkl(?{"this is runtime"})mno'
5238 * After eval_sv()-ing that, grab any new code blocks from the returned qr
5239 * and merge them with any code blocks of the original regexp.
5241 * If the pat is non-UTF8, while the evalled qr is UTF8, don't merge;
5242 * instead, just save the qr and return FALSE; this tells our caller that
5243 * the original pattern needs upgrading to utf8.
5247 S_compile_runtime_code(pTHX_ RExC_state_t * const pRExC_state,
5248 char *pat, STRLEN plen)
5252 GET_RE_DEBUG_FLAGS_DECL;
5254 if (pRExC_state->runtime_code_qr) {
5255 /* this is the second time we've been called; this should
5256 * only happen if the main pattern got upgraded to utf8
5257 * during compilation; re-use the qr we compiled first time
5258 * round (which should be utf8 too)
5260 qr = pRExC_state->runtime_code_qr;
5261 pRExC_state->runtime_code_qr = NULL;
5262 assert(RExC_utf8 && SvUTF8(qr));
5268 int newlen = plen + 6; /* allow for "qr''x\0" extra chars */
5272 /* determine how many extra chars we need for ' and \ escaping */
5273 for (s = 0; s < plen; s++) {
5274 if (pat[s] == '\'' || pat[s] == '\\')
5278 Newx(newpat, newlen, char);
5280 *p++ = 'q'; *p++ = 'r'; *p++ = '\'';
5282 for (s = 0; s < plen; s++) {
5283 if (n < pRExC_state->num_code_blocks
5284 && s == pRExC_state->code_blocks[n].start)
5286 /* blank out literal code block */
5287 assert(pat[s] == '(');
5288 while (s <= pRExC_state->code_blocks[n].end) {
5296 if (pat[s] == '\'' || pat[s] == '\\')
5301 if (pRExC_state->pm_flags & RXf_PMf_EXTENDED)
5305 PerlIO_printf(Perl_debug_log,
5306 "%sre-parsing pattern for runtime code:%s %s\n",
5307 PL_colors[4],PL_colors[5],newpat);
5310 sv = newSVpvn_flags(newpat, p-newpat-1, RExC_utf8 ? SVf_UTF8 : 0);
5316 PUSHSTACKi(PERLSI_REQUIRE);
5317 /* G_RE_REPARSING causes the toker to collapse \\ into \ when
5318 * parsing qr''; normally only q'' does this. It also alters
5320 eval_sv(sv, G_SCALAR|G_RE_REPARSING);
5321 SvREFCNT_dec_NN(sv);
5326 SV * const errsv = ERRSV;
5327 if (SvTRUE_NN(errsv))
5329 Safefree(pRExC_state->code_blocks);
5330 /* use croak_sv ? */
5331 Perl_croak_nocontext("%s", SvPV_nolen_const(errsv));
5334 assert(SvROK(qr_ref));
5336 assert(SvTYPE(qr) == SVt_REGEXP && RX_ENGINE((REGEXP*)qr)->op_comp);
5337 /* the leaving below frees the tmp qr_ref.
5338 * Give qr a life of its own */
5346 if (!RExC_utf8 && SvUTF8(qr)) {
5347 /* first time through; the pattern got upgraded; save the
5348 * qr for the next time through */
5349 assert(!pRExC_state->runtime_code_qr);
5350 pRExC_state->runtime_code_qr = qr;
5355 /* extract any code blocks within the returned qr// */
5358 /* merge the main (r1) and run-time (r2) code blocks into one */
5360 RXi_GET_DECL(ReANY((REGEXP *)qr), r2);
5361 struct reg_code_block *new_block, *dst;
5362 RExC_state_t * const r1 = pRExC_state; /* convenient alias */
5365 if (!r2->num_code_blocks) /* we guessed wrong */
5367 SvREFCNT_dec_NN(qr);
5372 r1->num_code_blocks + r2->num_code_blocks,
5373 struct reg_code_block);
5376 while ( i1 < r1->num_code_blocks
5377 || i2 < r2->num_code_blocks)
5379 struct reg_code_block *src;
5382 if (i1 == r1->num_code_blocks) {
5383 src = &r2->code_blocks[i2++];
5386 else if (i2 == r2->num_code_blocks)
5387 src = &r1->code_blocks[i1++];
5388 else if ( r1->code_blocks[i1].start
5389 < r2->code_blocks[i2].start)
5391 src = &r1->code_blocks[i1++];
5392 assert(src->end < r2->code_blocks[i2].start);
5395 assert( r1->code_blocks[i1].start
5396 > r2->code_blocks[i2].start);
5397 src = &r2->code_blocks[i2++];
5399 assert(src->end < r1->code_blocks[i1].start);
5402 assert(pat[src->start] == '(');
5403 assert(pat[src->end] == ')');
5404 dst->start = src->start;
5405 dst->end = src->end;
5406 dst->block = src->block;
5407 dst->src_regex = is_qr ? (REGEXP*) SvREFCNT_inc( (SV*) qr)
5411 r1->num_code_blocks += r2->num_code_blocks;
5412 Safefree(r1->code_blocks);
5413 r1->code_blocks = new_block;
5416 SvREFCNT_dec_NN(qr);
5422 S_setup_longest(pTHX_ RExC_state_t *pRExC_state, SV* sv_longest, SV** rx_utf8, SV** rx_substr, I32* rx_end_shift, I32 lookbehind, I32 offset, I32 *minlen, STRLEN longest_length, bool eol, bool meol)
5424 /* This is the common code for setting up the floating and fixed length
5425 * string data extracted from Perl_re_op_compile() below. Returns a boolean
5426 * as to whether succeeded or not */
5430 if (! (longest_length
5431 || (eol /* Can't have SEOL and MULTI */
5432 && (! meol || (RExC_flags & RXf_PMf_MULTILINE)))
5434 /* See comments for join_exact for why REG_SEEN_EXACTF_SHARP_S */
5435 || (RExC_seen & REG_SEEN_EXACTF_SHARP_S))
5440 /* copy the information about the longest from the reg_scan_data
5441 over to the program. */
5442 if (SvUTF8(sv_longest)) {
5443 *rx_utf8 = sv_longest;
5446 *rx_substr = sv_longest;
5449 /* end_shift is how many chars that must be matched that
5450 follow this item. We calculate it ahead of time as once the
5451 lookbehind offset is added in we lose the ability to correctly
5453 ml = minlen ? *(minlen) : (I32)longest_length;
5454 *rx_end_shift = ml - offset
5455 - longest_length + (SvTAIL(sv_longest) != 0)
5458 t = (eol/* Can't have SEOL and MULTI */
5459 && (! meol || (RExC_flags & RXf_PMf_MULTILINE)));
5460 fbm_compile(sv_longest, t ? FBMcf_TAIL : 0);
5466 * Perl_re_op_compile - the perl internal RE engine's function to compile a
5467 * regular expression into internal code.
5468 * The pattern may be passed either as:
5469 * a list of SVs (patternp plus pat_count)
5470 * a list of OPs (expr)
5471 * If both are passed, the SV list is used, but the OP list indicates
5472 * which SVs are actually pre-compiled code blocks
5474 * The SVs in the list have magic and qr overloading applied to them (and
5475 * the list may be modified in-place with replacement SVs in the latter
5478 * If the pattern hasn't changed from old_re, then old_re will be
5481 * eng is the current engine. If that engine has an op_comp method, then
5482 * handle directly (i.e. we assume that op_comp was us); otherwise, just
5483 * do the initial concatenation of arguments and pass on to the external
5486 * If is_bare_re is not null, set it to a boolean indicating whether the
5487 * arg list reduced (after overloading) to a single bare regex which has
5488 * been returned (i.e. /$qr/).
5490 * orig_rx_flags contains RXf_* flags. See perlreapi.pod for more details.
5492 * pm_flags contains the PMf_* flags, typically based on those from the
5493 * pm_flags field of the related PMOP. Currently we're only interested in
5494 * PMf_HAS_CV, PMf_IS_QR, PMf_USE_RE_EVAL.
5496 * We can't allocate space until we know how big the compiled form will be,
5497 * but we can't compile it (and thus know how big it is) until we've got a
5498 * place to put the code. So we cheat: we compile it twice, once with code
5499 * generation turned off and size counting turned on, and once "for real".
5500 * This also means that we don't allocate space until we are sure that the
5501 * thing really will compile successfully, and we never have to move the
5502 * code and thus invalidate pointers into it. (Note that it has to be in
5503 * one piece because free() must be able to free it all.) [NB: not true in perl]
5505 * Beware that the optimization-preparation code in here knows about some
5506 * of the structure of the compiled regexp. [I'll say.]
5510 Perl_re_op_compile(pTHX_ SV ** const patternp, int pat_count,
5511 OP *expr, const regexp_engine* eng, REGEXP *old_re,
5512 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags)
5517 regexp_internal *ri;
5525 SV *code_blocksv = NULL;
5526 SV** new_patternp = patternp;
5528 /* these are all flags - maybe they should be turned
5529 * into a single int with different bit masks */
5530 I32 sawlookahead = 0;
5535 regex_charset initial_charset = get_regex_charset(orig_rx_flags);
5537 bool runtime_code = 0;
5539 RExC_state_t RExC_state;
5540 RExC_state_t * const pRExC_state = &RExC_state;
5541 #ifdef TRIE_STUDY_OPT
5543 RExC_state_t copyRExC_state;
5545 GET_RE_DEBUG_FLAGS_DECL;
5547 PERL_ARGS_ASSERT_RE_OP_COMPILE;
5549 DEBUG_r(if (!PL_colorset) reginitcolors());
5551 #ifndef PERL_IN_XSUB_RE
5552 /* Initialize these here instead of as-needed, as is quick and avoids
5553 * having to test them each time otherwise */
5554 if (! PL_AboveLatin1) {
5555 PL_AboveLatin1 = _new_invlist_C_array(AboveLatin1_invlist);
5556 PL_ASCII = _new_invlist_C_array(ASCII_invlist);
5557 PL_Latin1 = _new_invlist_C_array(Latin1_invlist);
5559 PL_L1Posix_ptrs[_CC_ALPHANUMERIC]
5560 = _new_invlist_C_array(L1PosixAlnum_invlist);
5561 PL_Posix_ptrs[_CC_ALPHANUMERIC]
5562 = _new_invlist_C_array(PosixAlnum_invlist);
5564 PL_L1Posix_ptrs[_CC_ALPHA]
5565 = _new_invlist_C_array(L1PosixAlpha_invlist);
5566 PL_Posix_ptrs[_CC_ALPHA] = _new_invlist_C_array(PosixAlpha_invlist);
5568 PL_Posix_ptrs[_CC_BLANK] = _new_invlist_C_array(PosixBlank_invlist);
5569 PL_XPosix_ptrs[_CC_BLANK] = _new_invlist_C_array(XPosixBlank_invlist);
5571 /* Cased is the same as Alpha in the ASCII range */
5572 PL_L1Posix_ptrs[_CC_CASED] = _new_invlist_C_array(L1Cased_invlist);
5573 PL_Posix_ptrs[_CC_CASED] = _new_invlist_C_array(PosixAlpha_invlist);
5575 PL_Posix_ptrs[_CC_CNTRL] = _new_invlist_C_array(PosixCntrl_invlist);
5576 PL_XPosix_ptrs[_CC_CNTRL] = _new_invlist_C_array(XPosixCntrl_invlist);
5578 PL_Posix_ptrs[_CC_DIGIT] = _new_invlist_C_array(PosixDigit_invlist);
5579 PL_L1Posix_ptrs[_CC_DIGIT] = _new_invlist_C_array(PosixDigit_invlist);
5581 PL_L1Posix_ptrs[_CC_GRAPH] = _new_invlist_C_array(L1PosixGraph_invlist);
5582 PL_Posix_ptrs[_CC_GRAPH] = _new_invlist_C_array(PosixGraph_invlist);
5584 PL_L1Posix_ptrs[_CC_LOWER] = _new_invlist_C_array(L1PosixLower_invlist);
5585 PL_Posix_ptrs[_CC_LOWER] = _new_invlist_C_array(PosixLower_invlist);
5587 PL_L1Posix_ptrs[_CC_PRINT] = _new_invlist_C_array(L1PosixPrint_invlist);
5588 PL_Posix_ptrs[_CC_PRINT] = _new_invlist_C_array(PosixPrint_invlist);
5590 PL_L1Posix_ptrs[_CC_PUNCT] = _new_invlist_C_array(L1PosixPunct_invlist);
5591 PL_Posix_ptrs[_CC_PUNCT] = _new_invlist_C_array(PosixPunct_invlist);
5593 PL_Posix_ptrs[_CC_SPACE] = _new_invlist_C_array(PerlSpace_invlist);
5594 PL_XPosix_ptrs[_CC_SPACE] = _new_invlist_C_array(XPerlSpace_invlist);
5595 PL_Posix_ptrs[_CC_PSXSPC] = _new_invlist_C_array(PosixSpace_invlist);
5596 PL_XPosix_ptrs[_CC_PSXSPC] = _new_invlist_C_array(XPosixSpace_invlist);
5598 PL_L1Posix_ptrs[_CC_UPPER] = _new_invlist_C_array(L1PosixUpper_invlist);
5599 PL_Posix_ptrs[_CC_UPPER] = _new_invlist_C_array(PosixUpper_invlist);
5601 PL_XPosix_ptrs[_CC_VERTSPACE] = _new_invlist_C_array(VertSpace_invlist);
5603 PL_Posix_ptrs[_CC_WORDCHAR] = _new_invlist_C_array(PosixWord_invlist);
5604 PL_L1Posix_ptrs[_CC_WORDCHAR]
5605 = _new_invlist_C_array(L1PosixWord_invlist);
5607 PL_Posix_ptrs[_CC_XDIGIT] = _new_invlist_C_array(PosixXDigit_invlist);
5608 PL_XPosix_ptrs[_CC_XDIGIT] = _new_invlist_C_array(XPosixXDigit_invlist);
5610 PL_HasMultiCharFold = _new_invlist_C_array(_Perl_Multi_Char_Folds_invlist);
5614 pRExC_state->code_blocks = NULL;
5615 pRExC_state->num_code_blocks = 0;
5618 *is_bare_re = FALSE;
5620 if (expr && (expr->op_type == OP_LIST ||
5621 (expr->op_type == OP_NULL && expr->op_targ == OP_LIST))) {
5622 /* allocate code_blocks if needed */
5626 for (o = cLISTOPx(expr)->op_first; o; o = o->op_sibling)
5627 if (o->op_type == OP_NULL && (o->op_flags & OPf_SPECIAL))
5628 ncode++; /* count of DO blocks */
5630 pRExC_state->num_code_blocks = ncode;
5631 Newx(pRExC_state->code_blocks, ncode, struct reg_code_block);
5636 /* compile-time pattern with just OP_CONSTs and DO blocks */
5641 /* find how many CONSTs there are */
5644 if (expr->op_type == OP_CONST)
5647 for (o = cLISTOPx(expr)->op_first; o; o = o->op_sibling) {
5648 if (o->op_type == OP_CONST)
5652 /* fake up an SV array */
5654 assert(!new_patternp);
5655 Newx(new_patternp, n, SV*);
5656 SAVEFREEPV(new_patternp);
5660 if (expr->op_type == OP_CONST)
5661 new_patternp[n] = cSVOPx_sv(expr);
5663 for (o = cLISTOPx(expr)->op_first; o; o = o->op_sibling) {
5664 if (o->op_type == OP_CONST)
5665 new_patternp[n++] = cSVOPo_sv;
5670 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log,
5671 "Assembling pattern from %d elements%s\n", pat_count,
5672 orig_rx_flags & RXf_SPLIT ? " for split" : ""));
5674 /* set expr to the first arg op */
5676 if (pRExC_state->num_code_blocks
5677 && expr->op_type != OP_CONST)
5679 expr = cLISTOPx(expr)->op_first;
5680 assert( expr->op_type == OP_PUSHMARK
5681 || (expr->op_type == OP_NULL && expr->op_targ == OP_PUSHMARK)
5682 || expr->op_type == OP_PADRANGE);
5683 expr = expr->op_sibling;
5686 pat = S_concat_pat(aTHX_ pRExC_state, NULL, new_patternp, pat_count,
5687 expr, &recompile, NULL);
5689 /* handle bare (possibly after overloading) regex: foo =~ $re */
5694 if (SvTYPE(re) == SVt_REGEXP) {
5698 Safefree(pRExC_state->code_blocks);
5699 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log,
5700 "Precompiled pattern%s\n",
5701 orig_rx_flags & RXf_SPLIT ? " for split" : ""));
5707 exp = SvPV_nomg(pat, plen);
5709 if (!eng->op_comp) {
5710 if ((SvUTF8(pat) && IN_BYTES)
5711 || SvGMAGICAL(pat) || SvAMAGIC(pat))
5713 /* make a temporary copy; either to convert to bytes,
5714 * or to avoid repeating get-magic / overloaded stringify */
5715 pat = newSVpvn_flags(exp, plen, SVs_TEMP |
5716 (IN_BYTES ? 0 : SvUTF8(pat)));
5718 Safefree(pRExC_state->code_blocks);
5719 return CALLREGCOMP_ENG(eng, pat, orig_rx_flags);
5722 /* ignore the utf8ness if the pattern is 0 length */
5723 RExC_utf8 = RExC_orig_utf8 = (plen == 0 || IN_BYTES) ? 0 : SvUTF8(pat);
5724 RExC_uni_semantics = 0;
5725 RExC_contains_locale = 0;
5726 pRExC_state->runtime_code_qr = NULL;
5729 SV *dsv= sv_newmortal();
5730 RE_PV_QUOTED_DECL(s, RExC_utf8, dsv, exp, plen, 60);
5731 PerlIO_printf(Perl_debug_log, "%sCompiling REx%s %s\n",
5732 PL_colors[4],PL_colors[5],s);
5736 /* we jump here if we upgrade the pattern to utf8 and have to
5739 if ((pm_flags & PMf_USE_RE_EVAL)
5740 /* this second condition covers the non-regex literal case,
5741 * i.e. $foo =~ '(?{})'. */
5742 || (IN_PERL_COMPILETIME && (PL_hints & HINT_RE_EVAL))
5744 runtime_code = S_has_runtime_code(aTHX_ pRExC_state, exp, plen);
5746 /* return old regex if pattern hasn't changed */
5747 /* XXX: note in the below we have to check the flags as well as the pattern.
5749 * Things get a touch tricky as we have to compare the utf8 flag independently
5750 * from the compile flags.
5755 && !!RX_UTF8(old_re) == !!RExC_utf8
5756 && ( RX_COMPFLAGS(old_re) == ( orig_rx_flags & RXf_PMf_FLAGCOPYMASK ) )
5757 && RX_PRECOMP(old_re)
5758 && RX_PRELEN(old_re) == plen
5759 && memEQ(RX_PRECOMP(old_re), exp, plen)
5760 && !runtime_code /* with runtime code, always recompile */ )
5762 Safefree(pRExC_state->code_blocks);
5766 rx_flags = orig_rx_flags;
5768 if (initial_charset == REGEX_LOCALE_CHARSET) {
5769 RExC_contains_locale = 1;
5771 else if (RExC_utf8 && initial_charset == REGEX_DEPENDS_CHARSET) {
5773 /* Set to use unicode semantics if the pattern is in utf8 and has the
5774 * 'depends' charset specified, as it means unicode when utf8 */
5775 set_regex_charset(&rx_flags, REGEX_UNICODE_CHARSET);
5779 RExC_flags = rx_flags;
5780 RExC_pm_flags = pm_flags;
5783 if (TAINTING_get && TAINT_get)
5784 Perl_croak(aTHX_ "Eval-group in insecure regular expression");
5786 if (!S_compile_runtime_code(aTHX_ pRExC_state, exp, plen)) {
5787 /* whoops, we have a non-utf8 pattern, whilst run-time code
5788 * got compiled as utf8. Try again with a utf8 pattern */
5789 S_pat_upgrade_to_utf8(aTHX_ pRExC_state, &exp, &plen,
5790 pRExC_state->num_code_blocks);
5791 goto redo_first_pass;
5794 assert(!pRExC_state->runtime_code_qr);
5799 RExC_in_lookbehind = 0;
5800 RExC_seen_zerolen = *exp == '^' ? -1 : 0;
5802 RExC_override_recoding = 0;
5803 RExC_in_multi_char_class = 0;
5805 /* First pass: determine size, legality. */
5808 RExC_end = exp + plen;
5813 RExC_emit = &RExC_emit_dummy;
5814 RExC_whilem_seen = 0;
5815 RExC_open_parens = NULL;
5816 RExC_close_parens = NULL;
5818 RExC_paren_names = NULL;
5820 RExC_paren_name_list = NULL;
5822 RExC_recurse = NULL;
5823 RExC_recurse_count = 0;
5824 pRExC_state->code_index = 0;
5826 #if 0 /* REGC() is (currently) a NOP at the first pass.
5827 * Clever compilers notice this and complain. --jhi */
5828 REGC((U8)REG_MAGIC, (char*)RExC_emit);
5831 PerlIO_printf(Perl_debug_log, "Starting first pass (sizing)\n");
5833 RExC_lastparse=NULL;
5835 /* reg may croak on us, not giving us a chance to free
5836 pRExC_state->code_blocks. We cannot SAVEFREEPV it now, as we may
5837 need it to survive as long as the regexp (qr/(?{})/).
5838 We must check that code_blocksv is not already set, because we may
5839 have jumped back to restart the sizing pass. */
5840 if (pRExC_state->code_blocks && !code_blocksv) {
5841 code_blocksv = newSV_type(SVt_PV);
5842 SAVEFREESV(code_blocksv);
5843 SvPV_set(code_blocksv, (char *)pRExC_state->code_blocks);
5844 SvLEN_set(code_blocksv, 1); /*sufficient to make sv_clear free it*/
5846 if (reg(pRExC_state, 0, &flags,1) == NULL) {
5847 /* It's possible to write a regexp in ascii that represents Unicode
5848 codepoints outside of the byte range, such as via \x{100}. If we
5849 detect such a sequence we have to convert the entire pattern to utf8
5850 and then recompile, as our sizing calculation will have been based
5851 on 1 byte == 1 character, but we will need to use utf8 to encode
5852 at least some part of the pattern, and therefore must convert the whole
5855 if (flags & RESTART_UTF8) {
5856 S_pat_upgrade_to_utf8(aTHX_ pRExC_state, &exp, &plen,
5857 pRExC_state->num_code_blocks);
5858 goto redo_first_pass;
5860 Perl_croak(aTHX_ "panic: reg returned NULL to re_op_compile for sizing pass, flags=%#"UVxf"", (UV) flags);
5863 SvLEN_set(code_blocksv,0); /* no you can't have it, sv_clear */
5866 PerlIO_printf(Perl_debug_log,
5867 "Required size %"IVdf" nodes\n"
5868 "Starting second pass (creation)\n",
5871 RExC_lastparse=NULL;
5874 /* The first pass could have found things that force Unicode semantics */
5875 if ((RExC_utf8 || RExC_uni_semantics)
5876 && get_regex_charset(rx_flags) == REGEX_DEPENDS_CHARSET)
5878 set_regex_charset(&rx_flags, REGEX_UNICODE_CHARSET);
5881 /* Small enough for pointer-storage convention?
5882 If extralen==0, this means that we will not need long jumps. */
5883 if (RExC_size >= 0x10000L && RExC_extralen)
5884 RExC_size += RExC_extralen;
5887 if (RExC_whilem_seen > 15)
5888 RExC_whilem_seen = 15;
5890 /* Allocate space and zero-initialize. Note, the two step process
5891 of zeroing when in debug mode, thus anything assigned has to
5892 happen after that */
5893 rx = (REGEXP*) newSV_type(SVt_REGEXP);
5895 Newxc(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode),
5896 char, regexp_internal);
5897 if ( r == NULL || ri == NULL )
5898 FAIL("Regexp out of space");
5900 /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
5901 Zero(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode), char);
5903 /* bulk initialize base fields with 0. */
5904 Zero(ri, sizeof(regexp_internal), char);
5907 /* non-zero initialization begins here */
5910 r->extflags = rx_flags;
5911 RXp_COMPFLAGS(r) = orig_rx_flags & RXf_PMf_FLAGCOPYMASK;
5913 if (pm_flags & PMf_IS_QR) {
5914 ri->code_blocks = pRExC_state->code_blocks;
5915 ri->num_code_blocks = pRExC_state->num_code_blocks;
5920 for (n = 0; n < pRExC_state->num_code_blocks; n++)
5921 if (pRExC_state->code_blocks[n].src_regex)
5922 SAVEFREESV(pRExC_state->code_blocks[n].src_regex);
5923 SAVEFREEPV(pRExC_state->code_blocks);
5927 bool has_p = ((r->extflags & RXf_PMf_KEEPCOPY) == RXf_PMf_KEEPCOPY);
5928 bool has_charset = (get_regex_charset(r->extflags) != REGEX_DEPENDS_CHARSET);
5930 /* The caret is output if there are any defaults: if not all the STD
5931 * flags are set, or if no character set specifier is needed */
5933 (((r->extflags & RXf_PMf_STD_PMMOD) != RXf_PMf_STD_PMMOD)
5935 bool has_runon = ((RExC_seen & REG_SEEN_RUN_ON_COMMENT)==REG_SEEN_RUN_ON_COMMENT);
5936 U16 reganch = (U16)((r->extflags & RXf_PMf_STD_PMMOD)
5937 >> RXf_PMf_STD_PMMOD_SHIFT);
5938 const char *fptr = STD_PAT_MODS; /*"msix"*/
5940 /* Allocate for the worst case, which is all the std flags are turned
5941 * on. If more precision is desired, we could do a population count of
5942 * the flags set. This could be done with a small lookup table, or by
5943 * shifting, masking and adding, or even, when available, assembly
5944 * language for a machine-language population count.
5945 * We never output a minus, as all those are defaults, so are
5946 * covered by the caret */
5947 const STRLEN wraplen = plen + has_p + has_runon
5948 + has_default /* If needs a caret */
5950 /* If needs a character set specifier */
5951 + ((has_charset) ? MAX_CHARSET_NAME_LENGTH : 0)
5952 + (sizeof(STD_PAT_MODS) - 1)
5953 + (sizeof("(?:)") - 1);
5955 Newx(p, wraplen + 1, char); /* +1 for the ending NUL */
5956 r->xpv_len_u.xpvlenu_pv = p;
5958 SvFLAGS(rx) |= SVf_UTF8;
5961 /* If a default, cover it using the caret */
5963 *p++= DEFAULT_PAT_MOD;
5967 const char* const name = get_regex_charset_name(r->extflags, &len);
5968 Copy(name, p, len, char);
5972 *p++ = KEEPCOPY_PAT_MOD; /*'p'*/
5975 while((ch = *fptr++)) {
5983 Copy(RExC_precomp, p, plen, char);
5984 assert ((RX_WRAPPED(rx) - p) < 16);
5985 r->pre_prefix = p - RX_WRAPPED(rx);
5991 SvCUR_set(rx, p - RX_WRAPPED(rx));
5995 r->nparens = RExC_npar - 1; /* set early to validate backrefs */
5997 if (RExC_seen & REG_SEEN_RECURSE) {
5998 Newxz(RExC_open_parens, RExC_npar,regnode *);
5999 SAVEFREEPV(RExC_open_parens);
6000 Newxz(RExC_close_parens,RExC_npar,regnode *);
6001 SAVEFREEPV(RExC_close_parens);
6004 /* Useful during FAIL. */
6005 #ifdef RE_TRACK_PATTERN_OFFSETS
6006 Newxz(ri->u.offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
6007 DEBUG_OFFSETS_r(PerlIO_printf(Perl_debug_log,
6008 "%s %"UVuf" bytes for offset annotations.\n",
6009 ri->u.offsets ? "Got" : "Couldn't get",
6010 (UV)((2*RExC_size+1) * sizeof(U32))));
6012 SetProgLen(ri,RExC_size);
6016 REH_CALL_COMP_BEGIN_HOOK(pRExC_state->rx);
6018 /* Second pass: emit code. */
6019 RExC_flags = rx_flags; /* don't let top level (?i) bleed */
6020 RExC_pm_flags = pm_flags;
6022 RExC_end = exp + plen;
6025 RExC_emit_start = ri->program;
6026 RExC_emit = ri->program;
6027 RExC_emit_bound = ri->program + RExC_size + 1;
6028 pRExC_state->code_index = 0;
6030 REGC((U8)REG_MAGIC, (char*) RExC_emit++);
6031 if (reg(pRExC_state, 0, &flags,1) == NULL) {
6033 Perl_croak(aTHX_ "panic: reg returned NULL to re_op_compile for generation pass, flags=%#"UVxf"", (UV) flags);
6035 /* XXXX To minimize changes to RE engine we always allocate
6036 3-units-long substrs field. */
6037 Newx(r->substrs, 1, struct reg_substr_data);
6038 if (RExC_recurse_count) {
6039 Newxz(RExC_recurse,RExC_recurse_count,regnode *);
6040 SAVEFREEPV(RExC_recurse);
6044 r->minlen = minlen = sawlookahead = sawplus = sawopen = sawminmod = 0;
6045 Zero(r->substrs, 1, struct reg_substr_data);
6047 #ifdef TRIE_STUDY_OPT
6049 StructCopy(&zero_scan_data, &data, scan_data_t);
6050 copyRExC_state = RExC_state;
6053 DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log,"Restudying\n"));
6055 RExC_state = copyRExC_state;
6056 if (seen & REG_TOP_LEVEL_BRANCHES)
6057 RExC_seen |= REG_TOP_LEVEL_BRANCHES;
6059 RExC_seen &= ~REG_TOP_LEVEL_BRANCHES;
6060 StructCopy(&zero_scan_data, &data, scan_data_t);
6063 StructCopy(&zero_scan_data, &data, scan_data_t);
6066 /* Dig out information for optimizations. */
6067 r->extflags = RExC_flags; /* was pm_op */
6068 /*dmq: removed as part of de-PMOP: pm->op_pmflags = RExC_flags; */
6071 SvUTF8_on(rx); /* Unicode in it? */
6072 ri->regstclass = NULL;
6073 if (RExC_naughty >= 10) /* Probably an expensive pattern. */
6074 r->intflags |= PREGf_NAUGHTY;
6075 scan = ri->program + 1; /* First BRANCH. */
6077 /* testing for BRANCH here tells us whether there is "must appear"
6078 data in the pattern. If there is then we can use it for optimisations */
6079 if (!(RExC_seen & REG_TOP_LEVEL_BRANCHES)) { /* Only one top-level choice. */
6081 STRLEN longest_float_length, longest_fixed_length;
6082 struct regnode_charclass_class ch_class; /* pointed to by data */
6084 I32 last_close = 0; /* pointed to by data */
6085 regnode *first= scan;
6086 regnode *first_next= regnext(first);
6088 * Skip introductions and multiplicators >= 1
6089 * so that we can extract the 'meat' of the pattern that must
6090 * match in the large if() sequence following.
6091 * NOTE that EXACT is NOT covered here, as it is normally
6092 * picked up by the optimiser separately.
6094 * This is unfortunate as the optimiser isnt handling lookahead
6095 * properly currently.
6098 while ((OP(first) == OPEN && (sawopen = 1)) ||
6099 /* An OR of *one* alternative - should not happen now. */
6100 (OP(first) == BRANCH && OP(first_next) != BRANCH) ||
6101 /* for now we can't handle lookbehind IFMATCH*/
6102 (OP(first) == IFMATCH && !first->flags && (sawlookahead = 1)) ||
6103 (OP(first) == PLUS) ||
6104 (OP(first) == MINMOD) ||
6105 /* An {n,m} with n>0 */
6106 (PL_regkind[OP(first)] == CURLY && ARG1(first) > 0) ||
6107 (OP(first) == NOTHING && PL_regkind[OP(first_next)] != END ))
6110 * the only op that could be a regnode is PLUS, all the rest
6111 * will be regnode_1 or regnode_2.
6113 * (yves doesn't think this is true)
6115 if (OP(first) == PLUS)
6118 if (OP(first) == MINMOD)
6120 first += regarglen[OP(first)];
6122 first = NEXTOPER(first);
6123 first_next= regnext(first);
6126 /* Starting-point info. */
6128 DEBUG_PEEP("first:",first,0);
6129 /* Ignore EXACT as we deal with it later. */
6130 if (PL_regkind[OP(first)] == EXACT) {
6131 if (OP(first) == EXACT)
6132 NOOP; /* Empty, get anchored substr later. */
6134 ri->regstclass = first;
6137 else if (PL_regkind[OP(first)] == TRIE &&
6138 ((reg_trie_data *)ri->data->data[ ARG(first) ])->minlen>0)
6141 /* this can happen only on restudy */
6142 if ( OP(first) == TRIE ) {
6143 struct regnode_1 *trieop = (struct regnode_1 *)
6144 PerlMemShared_calloc(1, sizeof(struct regnode_1));
6145 StructCopy(first,trieop,struct regnode_1);
6146 trie_op=(regnode *)trieop;
6148 struct regnode_charclass *trieop = (struct regnode_charclass *)
6149 PerlMemShared_calloc(1, sizeof(struct regnode_charclass));
6150 StructCopy(first,trieop,struct regnode_charclass);
6151 trie_op=(regnode *)trieop;
6154 make_trie_failtable(pRExC_state, (regnode *)first, trie_op, 0);
6155 ri->regstclass = trie_op;
6158 else if (REGNODE_SIMPLE(OP(first)))
6159 ri->regstclass = first;
6160 else if (PL_regkind[OP(first)] == BOUND ||
6161 PL_regkind[OP(first)] == NBOUND)
6162 ri->regstclass = first;
6163 else if (PL_regkind[OP(first)] == BOL) {
6164 r->extflags |= (OP(first) == MBOL
6166 : (OP(first) == SBOL
6169 first = NEXTOPER(first);
6172 else if (OP(first) == GPOS) {
6173 r->extflags |= RXf_ANCH_GPOS;
6174 first = NEXTOPER(first);
6177 else if ((!sawopen || !RExC_sawback) &&
6178 (OP(first) == STAR &&
6179 PL_regkind[OP(NEXTOPER(first))] == REG_ANY) &&
6180 !(r->extflags & RXf_ANCH) && !pRExC_state->num_code_blocks)
6182 /* turn .* into ^.* with an implied $*=1 */
6184 (OP(NEXTOPER(first)) == REG_ANY)
6187 r->extflags |= type;
6188 r->intflags |= PREGf_IMPLICIT;
6189 first = NEXTOPER(first);
6192 if (sawplus && !sawminmod && !sawlookahead && (!sawopen || !RExC_sawback)
6193 && !pRExC_state->num_code_blocks) /* May examine pos and $& */
6194 /* x+ must match at the 1st pos of run of x's */
6195 r->intflags |= PREGf_SKIP;
6197 /* Scan is after the zeroth branch, first is atomic matcher. */
6198 #ifdef TRIE_STUDY_OPT
6201 PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
6202 (IV)(first - scan + 1))
6206 PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
6207 (IV)(first - scan + 1))
6213 * If there's something expensive in the r.e., find the
6214 * longest literal string that must appear and make it the
6215 * regmust. Resolve ties in favor of later strings, since
6216 * the regstart check works with the beginning of the r.e.
6217 * and avoiding duplication strengthens checking. Not a
6218 * strong reason, but sufficient in the absence of others.
6219 * [Now we resolve ties in favor of the earlier string if
6220 * it happens that c_offset_min has been invalidated, since the
6221 * earlier string may buy us something the later one won't.]
6224 data.longest_fixed = newSVpvs("");
6225 data.longest_float = newSVpvs("");
6226 data.last_found = newSVpvs("");
6227 data.longest = &(data.longest_fixed);
6228 ENTER_with_name("study_chunk");
6229 SAVEFREESV(data.longest_fixed);
6230 SAVEFREESV(data.longest_float);
6231 SAVEFREESV(data.last_found);
6233 if (!ri->regstclass) {
6234 cl_init(pRExC_state, &ch_class);
6235 data.start_class = &ch_class;
6236 stclass_flag = SCF_DO_STCLASS_AND;
6237 } else /* XXXX Check for BOUND? */
6239 data.last_closep = &last_close;
6241 minlen = study_chunk(pRExC_state, &first, &minlen, &fake, scan + RExC_size, /* Up to end */
6242 &data, -1, NULL, NULL,
6243 SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag
6244 | (restudied ? SCF_TRIE_DOING_RESTUDY : 0),
6248 CHECK_RESTUDY_GOTO_butfirst(LEAVE_with_name("study_chunk"));
6251 if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
6252 && data.last_start_min == 0 && data.last_end > 0
6253 && !RExC_seen_zerolen
6254 && !(RExC_seen & REG_SEEN_VERBARG)
6255 && !((RExC_seen & REG_SEEN_GPOS) || (r->extflags & RXf_ANCH_GPOS)))
6256 r->extflags |= RXf_CHECK_ALL;
6257 scan_commit(pRExC_state, &data,&minlen,0);
6259 longest_float_length = CHR_SVLEN(data.longest_float);
6261 if (! ((SvCUR(data.longest_fixed) /* ok to leave SvCUR */
6262 && data.offset_fixed == data.offset_float_min
6263 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float)))
6264 && S_setup_longest (aTHX_ pRExC_state,
6268 &(r->float_end_shift),
6269 data.lookbehind_float,
6270 data.offset_float_min,
6272 longest_float_length,
6273 cBOOL(data.flags & SF_FL_BEFORE_EOL),
6274 cBOOL(data.flags & SF_FL_BEFORE_MEOL)))
6276 r->float_min_offset = data.offset_float_min - data.lookbehind_float;
6277 r->float_max_offset = data.offset_float_max;
6278 if (data.offset_float_max < I32_MAX) /* Don't offset infinity */
6279 r->float_max_offset -= data.lookbehind_float;
6280 SvREFCNT_inc_simple_void_NN(data.longest_float);
6283 r->float_substr = r->float_utf8 = NULL;
6284 longest_float_length = 0;
6287 longest_fixed_length = CHR_SVLEN(data.longest_fixed);
6289 if (S_setup_longest (aTHX_ pRExC_state,
6291 &(r->anchored_utf8),
6292 &(r->anchored_substr),
6293 &(r->anchored_end_shift),
6294 data.lookbehind_fixed,
6297 longest_fixed_length,
6298 cBOOL(data.flags & SF_FIX_BEFORE_EOL),
6299 cBOOL(data.flags & SF_FIX_BEFORE_MEOL)))
6301 r->anchored_offset = data.offset_fixed - data.lookbehind_fixed;
6302 SvREFCNT_inc_simple_void_NN(data.longest_fixed);
6305 r->anchored_substr = r->anchored_utf8 = NULL;
6306 longest_fixed_length = 0;
6308 LEAVE_with_name("study_chunk");
6311 && (OP(ri->regstclass) == REG_ANY || OP(ri->regstclass) == SANY))
6312 ri->regstclass = NULL;
6314 if ((!(r->anchored_substr || r->anchored_utf8) || r->anchored_offset)
6316 && ! TEST_SSC_EOS(data.start_class)
6317 && !cl_is_anything(data.start_class))
6319 const U32 n = add_data(pRExC_state, 1, "f");
6320 OP(data.start_class) = ANYOF_SYNTHETIC;
6322 Newx(RExC_rxi->data->data[n], 1,
6323 struct regnode_charclass_class);
6324 StructCopy(data.start_class,
6325 (struct regnode_charclass_class*)RExC_rxi->data->data[n],
6326 struct regnode_charclass_class);
6327 ri->regstclass = (regnode*)RExC_rxi->data->data[n];
6328 r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
6329 DEBUG_COMPILE_r({ SV *sv = sv_newmortal();
6330 regprop(r, sv, (regnode*)data.start_class);
6331 PerlIO_printf(Perl_debug_log,
6332 "synthetic stclass \"%s\".\n",
6333 SvPVX_const(sv));});
6336 /* A temporary algorithm prefers floated substr to fixed one to dig more info. */
6337 if (longest_fixed_length > longest_float_length) {
6338 r->check_end_shift = r->anchored_end_shift;
6339 r->check_substr = r->anchored_substr;
6340 r->check_utf8 = r->anchored_utf8;
6341 r->check_offset_min = r->check_offset_max = r->anchored_offset;
6342 if (r->extflags & RXf_ANCH_SINGLE)
6343 r->extflags |= RXf_NOSCAN;
6346 r->check_end_shift = r->float_end_shift;
6347 r->check_substr = r->float_substr;
6348 r->check_utf8 = r->float_utf8;
6349 r->check_offset_min = r->float_min_offset;
6350 r->check_offset_max = r->float_max_offset;
6352 if ((r->check_substr || r->check_utf8) ) {
6353 r->extflags |= RXf_USE_INTUIT;
6354 if (SvTAIL(r->check_substr ? r->check_substr : r->check_utf8))
6355 r->extflags |= RXf_INTUIT_TAIL;
6357 /* XXX Unneeded? dmq (shouldn't as this is handled elsewhere)
6358 if ( (STRLEN)minlen < longest_float_length )
6359 minlen= longest_float_length;
6360 if ( (STRLEN)minlen < longest_fixed_length )
6361 minlen= longest_fixed_length;
6365 /* Several toplevels. Best we can is to set minlen. */
6367 struct regnode_charclass_class ch_class;
6370 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log, "\nMulti Top Level\n"));
6372 scan = ri->program + 1;
6373 cl_init(pRExC_state, &ch_class);
6374 data.start_class = &ch_class;
6375 data.last_closep = &last_close;
6378 minlen = study_chunk(pRExC_state, &scan, &minlen, &fake, scan + RExC_size,
6379 &data, -1, NULL, NULL,
6380 SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS
6381 |(restudied ? SCF_TRIE_DOING_RESTUDY : 0),
6384 CHECK_RESTUDY_GOTO_butfirst(NOOP);
6386 r->check_substr = r->check_utf8 = r->anchored_substr = r->anchored_utf8
6387 = r->float_substr = r->float_utf8 = NULL;
6389 if (! TEST_SSC_EOS(data.start_class)
6390 && !cl_is_anything(data.start_class))
6392 const U32 n = add_data(pRExC_state, 1, "f");
6393 OP(data.start_class) = ANYOF_SYNTHETIC;
6395 Newx(RExC_rxi->data->data[n], 1,
6396 struct regnode_charclass_class);
6397 StructCopy(data.start_class,
6398 (struct regnode_charclass_class*)RExC_rxi->data->data[n],
6399 struct regnode_charclass_class);
6400 ri->regstclass = (regnode*)RExC_rxi->data->data[n];
6401 r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
6402 DEBUG_COMPILE_r({ SV* sv = sv_newmortal();
6403 regprop(r, sv, (regnode*)data.start_class);
6404 PerlIO_printf(Perl_debug_log,
6405 "synthetic stclass \"%s\".\n",
6406 SvPVX_const(sv));});
6410 /* Guard against an embedded (?=) or (?<=) with a longer minlen than
6411 the "real" pattern. */
6413 PerlIO_printf(Perl_debug_log,"minlen: %"IVdf" r->minlen:%"IVdf"\n",
6414 (IV)minlen, (IV)r->minlen);
6416 r->minlenret = minlen;
6417 if (r->minlen < minlen)
6420 if (RExC_seen & REG_SEEN_GPOS)
6421 r->extflags |= RXf_GPOS_SEEN;
6422 if (RExC_seen & REG_SEEN_LOOKBEHIND)
6423 r->extflags |= RXf_NO_INPLACE_SUBST; /* inplace might break the lookbehind */
6424 if (pRExC_state->num_code_blocks)
6425 r->extflags |= RXf_EVAL_SEEN;
6426 if (RExC_seen & REG_SEEN_CANY)
6427 r->extflags |= RXf_CANY_SEEN;
6428 if (RExC_seen & REG_SEEN_VERBARG)
6430 r->intflags |= PREGf_VERBARG_SEEN;
6431 r->extflags |= RXf_NO_INPLACE_SUBST; /* don't understand this! Yves */
6433 if (RExC_seen & REG_SEEN_CUTGROUP)
6434 r->intflags |= PREGf_CUTGROUP_SEEN;
6435 if (pm_flags & PMf_USE_RE_EVAL)
6436 r->intflags |= PREGf_USE_RE_EVAL;
6437 if (RExC_paren_names)
6438 RXp_PAREN_NAMES(r) = MUTABLE_HV(SvREFCNT_inc(RExC_paren_names));
6440 RXp_PAREN_NAMES(r) = NULL;
6443 regnode *first = ri->program + 1;
6445 regnode *next = NEXTOPER(first);
6448 if (PL_regkind[fop] == NOTHING && nop == END)
6449 r->extflags |= RXf_NULL;
6450 else if (PL_regkind[fop] == BOL && nop == END)
6451 r->extflags |= RXf_START_ONLY;
6452 else if (fop == PLUS && PL_regkind[nop] == POSIXD && FLAGS(next) == _CC_SPACE && OP(regnext(first)) == END)
6453 r->extflags |= RXf_WHITE;
6454 else if ( r->extflags & RXf_SPLIT && fop == EXACT && STR_LEN(first) == 1 && *(STRING(first)) == ' ' && OP(regnext(first)) == END )
6455 r->extflags |= (RXf_SKIPWHITE|RXf_WHITE);
6459 if (RExC_paren_names) {
6460 ri->name_list_idx = add_data( pRExC_state, 1, "a" );
6461 ri->data->data[ri->name_list_idx] = (void*)SvREFCNT_inc(RExC_paren_name_list);
6464 ri->name_list_idx = 0;
6466 if (RExC_recurse_count) {
6467 for ( ; RExC_recurse_count ; RExC_recurse_count-- ) {
6468 const regnode *scan = RExC_recurse[RExC_recurse_count-1];
6469 ARG2L_SET( scan, RExC_open_parens[ARG(scan)-1] - scan );
6472 Newxz(r->offs, RExC_npar, regexp_paren_pair);
6473 /* assume we don't need to swap parens around before we match */
6476 PerlIO_printf(Perl_debug_log,"Final program:\n");
6479 #ifdef RE_TRACK_PATTERN_OFFSETS
6480 DEBUG_OFFSETS_r(if (ri->u.offsets) {
6481 const U32 len = ri->u.offsets[0];
6483 GET_RE_DEBUG_FLAGS_DECL;
6484 PerlIO_printf(Perl_debug_log, "Offsets: [%"UVuf"]\n\t", (UV)ri->u.offsets[0]);
6485 for (i = 1; i <= len; i++) {
6486 if (ri->u.offsets[i*2-1] || ri->u.offsets[i*2])
6487 PerlIO_printf(Perl_debug_log, "%"UVuf":%"UVuf"[%"UVuf"] ",
6488 (UV)i, (UV)ri->u.offsets[i*2-1], (UV)ri->u.offsets[i*2]);
6490 PerlIO_printf(Perl_debug_log, "\n");
6495 /* under ithreads the ?pat? PMf_USED flag on the pmop is simulated
6496 * by setting the regexp SV to readonly-only instead. If the
6497 * pattern's been recompiled, the USEDness should remain. */
6498 if (old_re && SvREADONLY(old_re))
6506 Perl_reg_named_buff(pTHX_ REGEXP * const rx, SV * const key, SV * const value,
6509 PERL_ARGS_ASSERT_REG_NAMED_BUFF;
6511 PERL_UNUSED_ARG(value);
6513 if (flags & RXapif_FETCH) {
6514 return reg_named_buff_fetch(rx, key, flags);
6515 } else if (flags & (RXapif_STORE | RXapif_DELETE | RXapif_CLEAR)) {
6516 Perl_croak_no_modify();
6518 } else if (flags & RXapif_EXISTS) {
6519 return reg_named_buff_exists(rx, key, flags)
6522 } else if (flags & RXapif_REGNAMES) {
6523 return reg_named_buff_all(rx, flags);
6524 } else if (flags & (RXapif_SCALAR | RXapif_REGNAMES_COUNT)) {
6525 return reg_named_buff_scalar(rx, flags);
6527 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff", (int)flags);
6533 Perl_reg_named_buff_iter(pTHX_ REGEXP * const rx, const SV * const lastkey,
6536 PERL_ARGS_ASSERT_REG_NAMED_BUFF_ITER;
6537 PERL_UNUSED_ARG(lastkey);
6539 if (flags & RXapif_FIRSTKEY)
6540 return reg_named_buff_firstkey(rx, flags);
6541 else if (flags & RXapif_NEXTKEY)
6542 return reg_named_buff_nextkey(rx, flags);
6544 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_iter", (int)flags);
6550 Perl_reg_named_buff_fetch(pTHX_ REGEXP * const r, SV * const namesv,
6553 AV *retarray = NULL;
6555 struct regexp *const rx = ReANY(r);
6557 PERL_ARGS_ASSERT_REG_NAMED_BUFF_FETCH;
6559 if (flags & RXapif_ALL)
6562 if (rx && RXp_PAREN_NAMES(rx)) {
6563 HE *he_str = hv_fetch_ent( RXp_PAREN_NAMES(rx), namesv, 0, 0 );
6566 SV* sv_dat=HeVAL(he_str);
6567 I32 *nums=(I32*)SvPVX(sv_dat);
6568 for ( i=0; i<SvIVX(sv_dat); i++ ) {
6569 if ((I32)(rx->nparens) >= nums[i]
6570 && rx->offs[nums[i]].start != -1
6571 && rx->offs[nums[i]].end != -1)
6574 CALLREG_NUMBUF_FETCH(r,nums[i],ret);
6579 ret = newSVsv(&PL_sv_undef);
6582 av_push(retarray, ret);
6585 return newRV_noinc(MUTABLE_SV(retarray));
6592 Perl_reg_named_buff_exists(pTHX_ REGEXP * const r, SV * const key,
6595 struct regexp *const rx = ReANY(r);
6597 PERL_ARGS_ASSERT_REG_NAMED_BUFF_EXISTS;
6599 if (rx && RXp_PAREN_NAMES(rx)) {
6600 if (flags & RXapif_ALL) {
6601 return hv_exists_ent(RXp_PAREN_NAMES(rx), key, 0);
6603 SV *sv = CALLREG_NAMED_BUFF_FETCH(r, key, flags);
6605 SvREFCNT_dec_NN(sv);
6617 Perl_reg_named_buff_firstkey(pTHX_ REGEXP * const r, const U32 flags)
6619 struct regexp *const rx = ReANY(r);
6621 PERL_ARGS_ASSERT_REG_NAMED_BUFF_FIRSTKEY;
6623 if ( rx && RXp_PAREN_NAMES(rx) ) {
6624 (void)hv_iterinit(RXp_PAREN_NAMES(rx));
6626 return CALLREG_NAMED_BUFF_NEXTKEY(r, NULL, flags & ~RXapif_FIRSTKEY);
6633 Perl_reg_named_buff_nextkey(pTHX_ REGEXP * const r, const U32 flags)
6635 struct regexp *const rx = ReANY(r);
6636 GET_RE_DEBUG_FLAGS_DECL;
6638 PERL_ARGS_ASSERT_REG_NAMED_BUFF_NEXTKEY;
6640 if (rx && RXp_PAREN_NAMES(rx)) {
6641 HV *hv = RXp_PAREN_NAMES(rx);
6643 while ( (temphe = hv_iternext_flags(hv,0)) ) {
6646 SV* sv_dat = HeVAL(temphe);
6647 I32 *nums = (I32*)SvPVX(sv_dat);
6648 for ( i = 0; i < SvIVX(sv_dat); i++ ) {
6649 if ((I32)(rx->lastparen) >= nums[i] &&
6650 rx->offs[nums[i]].start != -1 &&
6651 rx->offs[nums[i]].end != -1)
6657 if (parno || flags & RXapif_ALL) {
6658 return newSVhek(HeKEY_hek(temphe));
6666 Perl_reg_named_buff_scalar(pTHX_ REGEXP * const r, const U32 flags)
6671 struct regexp *const rx = ReANY(r);
6673 PERL_ARGS_ASSERT_REG_NAMED_BUFF_SCALAR;
6675 if (rx && RXp_PAREN_NAMES(rx)) {
6676 if (flags & (RXapif_ALL | RXapif_REGNAMES_COUNT)) {
6677 return newSViv(HvTOTALKEYS(RXp_PAREN_NAMES(rx)));
6678 } else if (flags & RXapif_ONE) {
6679 ret = CALLREG_NAMED_BUFF_ALL(r, (flags | RXapif_REGNAMES));
6680 av = MUTABLE_AV(SvRV(ret));
6681 length = av_len(av);
6682 SvREFCNT_dec_NN(ret);
6683 return newSViv(length + 1);
6685 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_scalar", (int)flags);
6689 return &PL_sv_undef;
6693 Perl_reg_named_buff_all(pTHX_ REGEXP * const r, const U32 flags)
6695 struct regexp *const rx = ReANY(r);
6698 PERL_ARGS_ASSERT_REG_NAMED_BUFF_ALL;
6700 if (rx && RXp_PAREN_NAMES(rx)) {
6701 HV *hv= RXp_PAREN_NAMES(rx);
6703 (void)hv_iterinit(hv);
6704 while ( (temphe = hv_iternext_flags(hv,0)) ) {
6707 SV* sv_dat = HeVAL(temphe);
6708 I32 *nums = (I32*)SvPVX(sv_dat);
6709 for ( i = 0; i < SvIVX(sv_dat); i++ ) {
6710 if ((I32)(rx->lastparen) >= nums[i] &&
6711 rx->offs[nums[i]].start != -1 &&
6712 rx->offs[nums[i]].end != -1)
6718 if (parno || flags & RXapif_ALL) {
6719 av_push(av, newSVhek(HeKEY_hek(temphe)));
6724 return newRV_noinc(MUTABLE_SV(av));
6728 Perl_reg_numbered_buff_fetch(pTHX_ REGEXP * const r, const I32 paren,
6731 struct regexp *const rx = ReANY(r);
6737 PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_FETCH;
6739 if ( n == RX_BUFF_IDX_CARET_PREMATCH
6740 || n == RX_BUFF_IDX_CARET_FULLMATCH
6741 || n == RX_BUFF_IDX_CARET_POSTMATCH
6744 bool keepcopy = cBOOL(rx->extflags & RXf_PMf_KEEPCOPY);
6746 /* on something like
6749 * the KEEPCOPY is set on the PMOP rather than the regex */
6750 if (PL_curpm && r == PM_GETRE(PL_curpm))
6751 keepcopy = cBOOL(PL_curpm->op_pmflags & PMf_KEEPCOPY);
6760 if (n == RX_BUFF_IDX_CARET_FULLMATCH)
6761 /* no need to distinguish between them any more */
6762 n = RX_BUFF_IDX_FULLMATCH;
6764 if ((n == RX_BUFF_IDX_PREMATCH || n == RX_BUFF_IDX_CARET_PREMATCH)
6765 && rx->offs[0].start != -1)
6767 /* $`, ${^PREMATCH} */
6768 i = rx->offs[0].start;
6772 if ((n == RX_BUFF_IDX_POSTMATCH || n == RX_BUFF_IDX_CARET_POSTMATCH)
6773 && rx->offs[0].end != -1)
6775 /* $', ${^POSTMATCH} */
6776 s = rx->subbeg - rx->suboffset + rx->offs[0].end;
6777 i = rx->sublen + rx->suboffset - rx->offs[0].end;
6780 if ( 0 <= n && n <= (I32)rx->nparens &&
6781 (s1 = rx->offs[n].start) != -1 &&
6782 (t1 = rx->offs[n].end) != -1)
6784 /* $&, ${^MATCH}, $1 ... */
6786 s = rx->subbeg + s1 - rx->suboffset;
6791 assert(s >= rx->subbeg);
6792 assert(rx->sublen >= (s - rx->subbeg) + i );
6794 #if NO_TAINT_SUPPORT
6795 sv_setpvn(sv, s, i);
6797 const int oldtainted = TAINT_get;
6799 sv_setpvn(sv, s, i);
6800 TAINT_set(oldtainted);
6802 if ( (rx->extflags & RXf_CANY_SEEN)
6803 ? (RXp_MATCH_UTF8(rx)
6804 && (!i || is_utf8_string((U8*)s, i)))
6805 : (RXp_MATCH_UTF8(rx)) )
6812 if (RXp_MATCH_TAINTED(rx)) {
6813 if (SvTYPE(sv) >= SVt_PVMG) {
6814 MAGIC* const mg = SvMAGIC(sv);
6817 SvMAGIC_set(sv, mg->mg_moremagic);
6819 if ((mgt = SvMAGIC(sv))) {
6820 mg->mg_moremagic = mgt;
6821 SvMAGIC_set(sv, mg);
6832 sv_setsv(sv,&PL_sv_undef);
6838 Perl_reg_numbered_buff_store(pTHX_ REGEXP * const rx, const I32 paren,
6839 SV const * const value)
6841 PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_STORE;
6843 PERL_UNUSED_ARG(rx);
6844 PERL_UNUSED_ARG(paren);
6845 PERL_UNUSED_ARG(value);
6848 Perl_croak_no_modify();
6852 Perl_reg_numbered_buff_length(pTHX_ REGEXP * const r, const SV * const sv,
6855 struct regexp *const rx = ReANY(r);
6859 PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_LENGTH;
6861 if ( paren == RX_BUFF_IDX_CARET_PREMATCH
6862 || paren == RX_BUFF_IDX_CARET_FULLMATCH
6863 || paren == RX_BUFF_IDX_CARET_POSTMATCH
6866 bool keepcopy = cBOOL(rx->extflags & RXf_PMf_KEEPCOPY);
6868 /* on something like
6871 * the KEEPCOPY is set on the PMOP rather than the regex */
6872 if (PL_curpm && r == PM_GETRE(PL_curpm))
6873 keepcopy = cBOOL(PL_curpm->op_pmflags & PMf_KEEPCOPY);
6879 /* Some of this code was originally in C<Perl_magic_len> in F<mg.c> */
6881 case RX_BUFF_IDX_CARET_PREMATCH: /* ${^PREMATCH} */
6882 case RX_BUFF_IDX_PREMATCH: /* $` */
6883 if (rx->offs[0].start != -1) {
6884 i = rx->offs[0].start;
6893 case RX_BUFF_IDX_CARET_POSTMATCH: /* ${^POSTMATCH} */
6894 case RX_BUFF_IDX_POSTMATCH: /* $' */
6895 if (rx->offs[0].end != -1) {
6896 i = rx->sublen - rx->offs[0].end;
6898 s1 = rx->offs[0].end;
6905 default: /* $& / ${^MATCH}, $1, $2, ... */
6906 if (paren <= (I32)rx->nparens &&
6907 (s1 = rx->offs[paren].start) != -1 &&
6908 (t1 = rx->offs[paren].end) != -1)
6914 if (ckWARN(WARN_UNINITIALIZED))
6915 report_uninit((const SV *)sv);
6920 if (i > 0 && RXp_MATCH_UTF8(rx)) {
6921 const char * const s = rx->subbeg - rx->suboffset + s1;
6926 if (is_utf8_string_loclen((U8*)s, i, &ep, &el))
6933 Perl_reg_qr_package(pTHX_ REGEXP * const rx)
6935 PERL_ARGS_ASSERT_REG_QR_PACKAGE;
6936 PERL_UNUSED_ARG(rx);
6940 return newSVpvs("Regexp");
6943 /* Scans the name of a named buffer from the pattern.
6944 * If flags is REG_RSN_RETURN_NULL returns null.
6945 * If flags is REG_RSN_RETURN_NAME returns an SV* containing the name
6946 * If flags is REG_RSN_RETURN_DATA returns the data SV* corresponding
6947 * to the parsed name as looked up in the RExC_paren_names hash.
6948 * If there is an error throws a vFAIL().. type exception.
6951 #define REG_RSN_RETURN_NULL 0
6952 #define REG_RSN_RETURN_NAME 1
6953 #define REG_RSN_RETURN_DATA 2
6956 S_reg_scan_name(pTHX_ RExC_state_t *pRExC_state, U32 flags)
6958 char *name_start = RExC_parse;
6960 PERL_ARGS_ASSERT_REG_SCAN_NAME;
6962 if (isIDFIRST_lazy_if(RExC_parse, UTF)) {
6963 /* skip IDFIRST by using do...while */
6966 RExC_parse += UTF8SKIP(RExC_parse);
6967 } while (isWORDCHAR_utf8((U8*)RExC_parse));
6971 } while (isWORDCHAR(*RExC_parse));
6973 RExC_parse++; /* so the <- from the vFAIL is after the offending character */
6974 vFAIL("Group name must start with a non-digit word character");
6978 = newSVpvn_flags(name_start, (int)(RExC_parse - name_start),
6979 SVs_TEMP | (UTF ? SVf_UTF8 : 0));
6980 if ( flags == REG_RSN_RETURN_NAME)
6982 else if (flags==REG_RSN_RETURN_DATA) {
6985 if ( ! sv_name ) /* should not happen*/
6986 Perl_croak(aTHX_ "panic: no svname in reg_scan_name");
6987 if (RExC_paren_names)
6988 he_str = hv_fetch_ent( RExC_paren_names, sv_name, 0, 0 );
6990 sv_dat = HeVAL(he_str);
6992 vFAIL("Reference to nonexistent named group");
6996 Perl_croak(aTHX_ "panic: bad flag %lx in reg_scan_name",
6997 (unsigned long) flags);
6999 assert(0); /* NOT REACHED */
7004 #define DEBUG_PARSE_MSG(funcname) DEBUG_PARSE_r({ \
7005 int rem=(int)(RExC_end - RExC_parse); \
7014 if (RExC_lastparse!=RExC_parse) \
7015 PerlIO_printf(Perl_debug_log," >%.*s%-*s", \
7018 iscut ? "..." : "<" \
7021 PerlIO_printf(Perl_debug_log,"%16s",""); \
7024 num = RExC_size + 1; \
7026 num=REG_NODE_NUM(RExC_emit); \
7027 if (RExC_lastnum!=num) \
7028 PerlIO_printf(Perl_debug_log,"|%4d",num); \
7030 PerlIO_printf(Perl_debug_log,"|%4s",""); \
7031 PerlIO_printf(Perl_debug_log,"|%*s%-4s", \
7032 (int)((depth*2)), "", \
7036 RExC_lastparse=RExC_parse; \
7041 #define DEBUG_PARSE(funcname) DEBUG_PARSE_r({ \
7042 DEBUG_PARSE_MSG((funcname)); \
7043 PerlIO_printf(Perl_debug_log,"%4s","\n"); \
7045 #define DEBUG_PARSE_FMT(funcname,fmt,args) DEBUG_PARSE_r({ \
7046 DEBUG_PARSE_MSG((funcname)); \
7047 PerlIO_printf(Perl_debug_log,fmt "\n",args); \
7050 /* This section of code defines the inversion list object and its methods. The
7051 * interfaces are highly subject to change, so as much as possible is static to
7052 * this file. An inversion list is here implemented as a malloc'd C UV array
7053 * as an SVt_INVLIST scalar.
7055 * An inversion list for Unicode is an array of code points, sorted by ordinal
7056 * number. The zeroth element is the first code point in the list. The 1th
7057 * element is the first element beyond that not in the list. In other words,
7058 * the first range is
7059 * invlist[0]..(invlist[1]-1)
7060 * The other ranges follow. Thus every element whose index is divisible by two
7061 * marks the beginning of a range that is in the list, and every element not
7062 * divisible by two marks the beginning of a range not in the list. A single
7063 * element inversion list that contains the single code point N generally
7064 * consists of two elements
7067 * (The exception is when N is the highest representable value on the
7068 * machine, in which case the list containing just it would be a single
7069 * element, itself. By extension, if the last range in the list extends to
7070 * infinity, then the first element of that range will be in the inversion list
7071 * at a position that is divisible by two, and is the final element in the
7073 * Taking the complement (inverting) an inversion list is quite simple, if the
7074 * first element is 0, remove it; otherwise add a 0 element at the beginning.
7075 * This implementation reserves an element at the beginning of each inversion
7076 * list to always contain 0; there is an additional flag in the header which
7077 * indicates if the list begins at the 0, or is offset to begin at the next
7080 * More about inversion lists can be found in "Unicode Demystified"
7081 * Chapter 13 by Richard Gillam, published by Addison-Wesley.
7082 * More will be coming when functionality is added later.
7084 * The inversion list data structure is currently implemented as an SV pointing
7085 * to an array of UVs that the SV thinks are bytes. This allows us to have an
7086 * array of UV whose memory management is automatically handled by the existing
7087 * facilities for SV's.
7089 * Some of the methods should always be private to the implementation, and some
7090 * should eventually be made public */
7092 /* The header definitions are in F<inline_invlist.c> */
7094 PERL_STATIC_INLINE UV*
7095 S__invlist_array_init(pTHX_ SV* const invlist, const bool will_have_0)
7097 /* Returns a pointer to the first element in the inversion list's array.
7098 * This is called upon initialization of an inversion list. Where the
7099 * array begins depends on whether the list has the code point U+0000 in it
7100 * or not. The other parameter tells it whether the code that follows this
7101 * call is about to put a 0 in the inversion list or not. The first
7102 * element is either the element reserved for 0, if TRUE, or the element
7103 * after it, if FALSE */
7105 bool* offset = get_invlist_offset_addr(invlist);
7106 UV* zero_addr = (UV *) SvPVX(invlist);
7108 PERL_ARGS_ASSERT__INVLIST_ARRAY_INIT;
7111 assert(! _invlist_len(invlist));
7115 /* 1^1 = 0; 1^0 = 1 */
7116 *offset = 1 ^ will_have_0;
7117 return zero_addr + *offset;
7120 PERL_STATIC_INLINE UV*
7121 S_invlist_array(pTHX_ SV* const invlist)
7123 /* Returns the pointer to the inversion list's array. Every time the
7124 * length changes, this needs to be called in case malloc or realloc moved
7127 PERL_ARGS_ASSERT_INVLIST_ARRAY;
7129 /* Must not be empty. If these fail, you probably didn't check for <len>
7130 * being non-zero before trying to get the array */
7131 assert(_invlist_len(invlist));
7133 /* The very first element always contains zero, The array begins either
7134 * there, or if the inversion list is offset, at the element after it.
7135 * The offset header field determines which; it contains 0 or 1 to indicate
7136 * how much additionally to add */
7137 assert(0 == *(SvPVX(invlist)));
7138 return ((UV *) SvPVX(invlist) + *get_invlist_offset_addr(invlist));
7141 PERL_STATIC_INLINE void
7142 S_invlist_set_len(pTHX_ SV* const invlist, const UV len, const bool offset)
7144 /* Sets the current number of elements stored in the inversion list.
7145 * Updates SvCUR correspondingly */
7147 PERL_ARGS_ASSERT_INVLIST_SET_LEN;
7149 assert(SvTYPE(invlist) == SVt_INVLIST);
7154 : TO_INTERNAL_SIZE(len + offset));
7155 assert(SvLEN(invlist) == 0 || SvCUR(invlist) <= SvLEN(invlist));
7158 PERL_STATIC_INLINE IV*
7159 S_get_invlist_previous_index_addr(pTHX_ SV* invlist)
7161 /* Return the address of the IV that is reserved to hold the cached index
7164 PERL_ARGS_ASSERT_GET_INVLIST_PREVIOUS_INDEX_ADDR;
7166 assert(SvTYPE(invlist) == SVt_INVLIST);
7168 return &(((XINVLIST*) SvANY(invlist))->prev_index);
7171 PERL_STATIC_INLINE IV
7172 S_invlist_previous_index(pTHX_ SV* const invlist)
7174 /* Returns cached index of previous search */
7176 PERL_ARGS_ASSERT_INVLIST_PREVIOUS_INDEX;
7178 return *get_invlist_previous_index_addr(invlist);
7181 PERL_STATIC_INLINE void
7182 S_invlist_set_previous_index(pTHX_ SV* const invlist, const IV index)
7184 /* Caches <index> for later retrieval */
7186 PERL_ARGS_ASSERT_INVLIST_SET_PREVIOUS_INDEX;
7188 assert(index == 0 || index < (int) _invlist_len(invlist));
7190 *get_invlist_previous_index_addr(invlist) = index;
7193 PERL_STATIC_INLINE UV
7194 S_invlist_max(pTHX_ SV* const invlist)
7196 /* Returns the maximum number of elements storable in the inversion list's
7197 * array, without having to realloc() */
7199 PERL_ARGS_ASSERT_INVLIST_MAX;
7201 assert(SvTYPE(invlist) == SVt_INVLIST);
7203 /* Assumes worst case, in which the 0 element is not counted in the
7204 * inversion list, so subtracts 1 for that */
7205 return SvLEN(invlist) == 0 /* This happens under _new_invlist_C_array */
7206 ? FROM_INTERNAL_SIZE(SvCUR(invlist)) - 1
7207 : FROM_INTERNAL_SIZE(SvLEN(invlist)) - 1;
7210 #ifndef PERL_IN_XSUB_RE
7212 Perl__new_invlist(pTHX_ IV initial_size)
7215 /* Return a pointer to a newly constructed inversion list, with enough
7216 * space to store 'initial_size' elements. If that number is negative, a
7217 * system default is used instead */
7221 if (initial_size < 0) {
7225 /* Allocate the initial space */
7226 new_list = newSV_type(SVt_INVLIST);
7228 /* First 1 is in case the zero element isn't in the list; second 1 is for
7230 SvGROW(new_list, TO_INTERNAL_SIZE(initial_size + 1) + 1);
7231 invlist_set_len(new_list, 0, 0);
7233 /* Force iterinit() to be used to get iteration to work */
7234 *get_invlist_iter_addr(new_list) = (STRLEN) UV_MAX;
7236 *get_invlist_previous_index_addr(new_list) = 0;
7243 S__new_invlist_C_array(pTHX_ const UV* const list)
7245 /* Return a pointer to a newly constructed inversion list, initialized to
7246 * point to <list>, which has to be in the exact correct inversion list
7247 * form, including internal fields. Thus this is a dangerous routine that
7248 * should not be used in the wrong hands. The passed in 'list' contains
7249 * several header fields at the beginning that are not part of the
7250 * inversion list body proper */
7252 const STRLEN length = (STRLEN) list[0];
7253 const UV version_id = list[1];
7254 const bool offset = cBOOL(list[2]);
7255 #define HEADER_LENGTH 3
7256 /* If any of the above changes in any way, you must change HEADER_LENGTH
7257 * (if appropriate) and regenerate INVLIST_VERSION_ID by running
7258 * perl -E 'say int(rand 2**31-1)'
7260 #define INVLIST_VERSION_ID 148565664 /* This is a combination of a version and
7261 data structure type, so that one being
7262 passed in can be validated to be an
7263 inversion list of the correct vintage.
7266 SV* invlist = newSV_type(SVt_INVLIST);
7268 PERL_ARGS_ASSERT__NEW_INVLIST_C_ARRAY;
7270 if (version_id != INVLIST_VERSION_ID) {
7271 Perl_croak(aTHX_ "panic: Incorrect version for previously generated inversion list");
7274 /* The generated array passed in includes header elements that aren't part
7275 * of the list proper, so start it just after them */
7276 SvPV_set(invlist, (char *) (list + HEADER_LENGTH));
7278 SvLEN_set(invlist, 0); /* Means we own the contents, and the system
7279 shouldn't touch it */
7281 *(get_invlist_offset_addr(invlist)) = offset;
7283 /* The 'length' passed to us is the physical number of elements in the
7284 * inversion list. But if there is an offset the logical number is one
7286 invlist_set_len(invlist, length - offset, offset);
7288 invlist_set_previous_index(invlist, 0);
7290 /* Initialize the iteration pointer. */
7291 invlist_iterfinish(invlist);
7297 S_invlist_extend(pTHX_ SV* const invlist, const UV new_max)
7299 /* Grow the maximum size of an inversion list */
7301 PERL_ARGS_ASSERT_INVLIST_EXTEND;
7303 assert(SvTYPE(invlist) == SVt_INVLIST);
7305 /* Add one to account for the zero element at the beginning which may not
7306 * be counted by the calling parameters */
7307 SvGROW((SV *)invlist, TO_INTERNAL_SIZE(new_max + 1));
7310 PERL_STATIC_INLINE void
7311 S_invlist_trim(pTHX_ SV* const invlist)
7313 PERL_ARGS_ASSERT_INVLIST_TRIM;
7315 assert(SvTYPE(invlist) == SVt_INVLIST);
7317 /* Change the length of the inversion list to how many entries it currently
7319 SvPV_shrink_to_cur((SV *) invlist);
7322 #define _invlist_union_complement_2nd(a, b, output) _invlist_union_maybe_complement_2nd(a, b, TRUE, output)
7325 S__append_range_to_invlist(pTHX_ SV* const invlist, const UV start, const UV end)
7327 /* Subject to change or removal. Append the range from 'start' to 'end' at
7328 * the end of the inversion list. The range must be above any existing
7332 UV max = invlist_max(invlist);
7333 UV len = _invlist_len(invlist);
7336 PERL_ARGS_ASSERT__APPEND_RANGE_TO_INVLIST;
7338 if (len == 0) { /* Empty lists must be initialized */
7339 offset = start != 0;
7340 array = _invlist_array_init(invlist, ! offset);
7343 /* Here, the existing list is non-empty. The current max entry in the
7344 * list is generally the first value not in the set, except when the
7345 * set extends to the end of permissible values, in which case it is
7346 * the first entry in that final set, and so this call is an attempt to
7347 * append out-of-order */
7349 UV final_element = len - 1;
7350 array = invlist_array(invlist);
7351 if (array[final_element] > start
7352 || ELEMENT_RANGE_MATCHES_INVLIST(final_element))
7354 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",
7355 array[final_element], start,
7356 ELEMENT_RANGE_MATCHES_INVLIST(final_element) ? 't' : 'f');
7359 /* Here, it is a legal append. If the new range begins with the first
7360 * value not in the set, it is extending the set, so the new first
7361 * value not in the set is one greater than the newly extended range.
7363 offset = *get_invlist_offset_addr(invlist);
7364 if (array[final_element] == start) {
7365 if (end != UV_MAX) {
7366 array[final_element] = end + 1;
7369 /* But if the end is the maximum representable on the machine,
7370 * just let the range that this would extend to have no end */
7371 invlist_set_len(invlist, len - 1, offset);
7377 /* Here the new range doesn't extend any existing set. Add it */
7379 len += 2; /* Includes an element each for the start and end of range */
7381 /* If wll overflow the existing space, extend, which may cause the array to
7384 invlist_extend(invlist, len);
7386 /* Have to set len here to avoid assert failure in invlist_array() */
7387 invlist_set_len(invlist, len, offset);
7389 array = invlist_array(invlist);
7392 invlist_set_len(invlist, len, offset);
7395 /* The next item on the list starts the range, the one after that is
7396 * one past the new range. */
7397 array[len - 2] = start;
7398 if (end != UV_MAX) {
7399 array[len - 1] = end + 1;
7402 /* But if the end is the maximum representable on the machine, just let
7403 * the range have no end */
7404 invlist_set_len(invlist, len - 1, offset);
7408 #ifndef PERL_IN_XSUB_RE
7411 Perl__invlist_search(pTHX_ SV* const invlist, const UV cp)
7413 /* Searches the inversion list for the entry that contains the input code
7414 * point <cp>. If <cp> is not in the list, -1 is returned. Otherwise, the
7415 * return value is the index into the list's array of the range that
7420 IV high = _invlist_len(invlist);
7421 const IV highest_element = high - 1;
7424 PERL_ARGS_ASSERT__INVLIST_SEARCH;
7426 /* If list is empty, return failure. */
7431 /* (We can't get the array unless we know the list is non-empty) */
7432 array = invlist_array(invlist);
7434 mid = invlist_previous_index(invlist);
7435 assert(mid >=0 && mid <= highest_element);
7437 /* <mid> contains the cache of the result of the previous call to this
7438 * function (0 the first time). See if this call is for the same result,
7439 * or if it is for mid-1. This is under the theory that calls to this
7440 * function will often be for related code points that are near each other.
7441 * And benchmarks show that caching gives better results. We also test
7442 * here if the code point is within the bounds of the list. These tests
7443 * replace others that would have had to be made anyway to make sure that
7444 * the array bounds were not exceeded, and these give us extra information
7445 * at the same time */
7446 if (cp >= array[mid]) {
7447 if (cp >= array[highest_element]) {
7448 return highest_element;
7451 /* Here, array[mid] <= cp < array[highest_element]. This means that
7452 * the final element is not the answer, so can exclude it; it also
7453 * means that <mid> is not the final element, so can refer to 'mid + 1'
7455 if (cp < array[mid + 1]) {
7461 else { /* cp < aray[mid] */
7462 if (cp < array[0]) { /* Fail if outside the array */
7466 if (cp >= array[mid - 1]) {
7471 /* Binary search. What we are looking for is <i> such that
7472 * array[i] <= cp < array[i+1]
7473 * The loop below converges on the i+1. Note that there may not be an
7474 * (i+1)th element in the array, and things work nonetheless */
7475 while (low < high) {
7476 mid = (low + high) / 2;
7477 assert(mid <= highest_element);
7478 if (array[mid] <= cp) { /* cp >= array[mid] */
7481 /* We could do this extra test to exit the loop early.
7482 if (cp < array[low]) {
7487 else { /* cp < array[mid] */
7494 invlist_set_previous_index(invlist, high);
7499 Perl__invlist_populate_swatch(pTHX_ SV* const invlist, const UV start, const UV end, U8* swatch)
7501 /* populates a swatch of a swash the same way swatch_get() does in utf8.c,
7502 * but is used when the swash has an inversion list. This makes this much
7503 * faster, as it uses a binary search instead of a linear one. This is
7504 * intimately tied to that function, and perhaps should be in utf8.c,
7505 * except it is intimately tied to inversion lists as well. It assumes
7506 * that <swatch> is all 0's on input */
7509 const IV len = _invlist_len(invlist);
7513 PERL_ARGS_ASSERT__INVLIST_POPULATE_SWATCH;
7515 if (len == 0) { /* Empty inversion list */
7519 array = invlist_array(invlist);
7521 /* Find which element it is */
7522 i = _invlist_search(invlist, start);
7524 /* We populate from <start> to <end> */
7525 while (current < end) {
7528 /* The inversion list gives the results for every possible code point
7529 * after the first one in the list. Only those ranges whose index is
7530 * even are ones that the inversion list matches. For the odd ones,
7531 * and if the initial code point is not in the list, we have to skip
7532 * forward to the next element */
7533 if (i == -1 || ! ELEMENT_RANGE_MATCHES_INVLIST(i)) {
7535 if (i >= len) { /* Finished if beyond the end of the array */
7539 if (current >= end) { /* Finished if beyond the end of what we
7541 if (LIKELY(end < UV_MAX)) {
7545 /* We get here when the upper bound is the maximum
7546 * representable on the machine, and we are looking for just
7547 * that code point. Have to special case it */
7549 goto join_end_of_list;
7552 assert(current >= start);
7554 /* The current range ends one below the next one, except don't go past
7557 upper = (i < len && array[i] < end) ? array[i] : end;
7559 /* Here we are in a range that matches. Populate a bit in the 3-bit U8
7560 * for each code point in it */
7561 for (; current < upper; current++) {
7562 const STRLEN offset = (STRLEN)(current - start);
7563 swatch[offset >> 3] |= 1 << (offset & 7);
7568 /* Quit if at the end of the list */
7571 /* But first, have to deal with the highest possible code point on
7572 * the platform. The previous code assumes that <end> is one
7573 * beyond where we want to populate, but that is impossible at the
7574 * platform's infinity, so have to handle it specially */
7575 if (UNLIKELY(end == UV_MAX && ELEMENT_RANGE_MATCHES_INVLIST(len-1)))
7577 const STRLEN offset = (STRLEN)(end - start);
7578 swatch[offset >> 3] |= 1 << (offset & 7);
7583 /* Advance to the next range, which will be for code points not in the
7592 Perl__invlist_union_maybe_complement_2nd(pTHX_ SV* const a, SV* const b, const bool complement_b, SV** output)
7594 /* Take the union of two inversion lists and point <output> to it. *output
7595 * SHOULD BE DEFINED upon input, and if it points to one of the two lists,
7596 * the reference count to that list will be decremented. The first list,
7597 * <a>, may be NULL, in which case a copy of the second list is returned.
7598 * If <complement_b> is TRUE, the union is taken of the complement
7599 * (inversion) of <b> instead of b itself.
7601 * The basis for this comes from "Unicode Demystified" Chapter 13 by
7602 * Richard Gillam, published by Addison-Wesley, and explained at some
7603 * length there. The preface says to incorporate its examples into your
7604 * code at your own risk.
7606 * The algorithm is like a merge sort.
7608 * XXX A potential performance improvement is to keep track as we go along
7609 * if only one of the inputs contributes to the result, meaning the other
7610 * is a subset of that one. In that case, we can skip the final copy and
7611 * return the larger of the input lists, but then outside code might need
7612 * to keep track of whether to free the input list or not */
7614 const UV* array_a; /* a's array */
7616 UV len_a; /* length of a's array */
7619 SV* u; /* the resulting union */
7623 UV i_a = 0; /* current index into a's array */
7627 /* running count, as explained in the algorithm source book; items are
7628 * stopped accumulating and are output when the count changes to/from 0.
7629 * The count is incremented when we start a range that's in the set, and
7630 * decremented when we start a range that's not in the set. So its range
7631 * is 0 to 2. Only when the count is zero is something not in the set.
7635 PERL_ARGS_ASSERT__INVLIST_UNION_MAYBE_COMPLEMENT_2ND;
7638 /* If either one is empty, the union is the other one */
7639 if (a == NULL || ((len_a = _invlist_len(a)) == 0)) {
7646 *output = invlist_clone(b);
7648 _invlist_invert(*output);
7650 } /* else *output already = b; */
7653 else if ((len_b = _invlist_len(b)) == 0) {
7658 /* The complement of an empty list is a list that has everything in it,
7659 * so the union with <a> includes everything too */
7664 *output = _new_invlist(1);
7665 _append_range_to_invlist(*output, 0, UV_MAX);
7667 else if (*output != a) {
7668 *output = invlist_clone(a);
7670 /* else *output already = a; */
7674 /* Here both lists exist and are non-empty */
7675 array_a = invlist_array(a);
7676 array_b = invlist_array(b);
7678 /* If are to take the union of 'a' with the complement of b, set it
7679 * up so are looking at b's complement. */
7682 /* To complement, we invert: if the first element is 0, remove it. To
7683 * do this, we just pretend the array starts one later */
7684 if (array_b[0] == 0) {
7690 /* But if the first element is not zero, we pretend the list starts
7691 * at the 0 that is always stored immediately before the array. */
7697 /* Size the union for the worst case: that the sets are completely
7699 u = _new_invlist(len_a + len_b);
7701 /* Will contain U+0000 if either component does */
7702 array_u = _invlist_array_init(u, (len_a > 0 && array_a[0] == 0)
7703 || (len_b > 0 && array_b[0] == 0));
7705 /* Go through each list item by item, stopping when exhausted one of
7707 while (i_a < len_a && i_b < len_b) {
7708 UV cp; /* The element to potentially add to the union's array */
7709 bool cp_in_set; /* is it in the the input list's set or not */
7711 /* We need to take one or the other of the two inputs for the union.
7712 * Since we are merging two sorted lists, we take the smaller of the
7713 * next items. In case of a tie, we take the one that is in its set
7714 * first. If we took one not in the set first, it would decrement the
7715 * count, possibly to 0 which would cause it to be output as ending the
7716 * range, and the next time through we would take the same number, and
7717 * output it again as beginning the next range. By doing it the
7718 * opposite way, there is no possibility that the count will be
7719 * momentarily decremented to 0, and thus the two adjoining ranges will
7720 * be seamlessly merged. (In a tie and both are in the set or both not
7721 * in the set, it doesn't matter which we take first.) */
7722 if (array_a[i_a] < array_b[i_b]
7723 || (array_a[i_a] == array_b[i_b]
7724 && ELEMENT_RANGE_MATCHES_INVLIST(i_a)))
7726 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_a);
7730 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_b);
7731 cp = array_b[i_b++];
7734 /* Here, have chosen which of the two inputs to look at. Only output
7735 * if the running count changes to/from 0, which marks the
7736 * beginning/end of a range in that's in the set */
7739 array_u[i_u++] = cp;
7746 array_u[i_u++] = cp;
7751 /* Here, we are finished going through at least one of the lists, which
7752 * means there is something remaining in at most one. We check if the list
7753 * that hasn't been exhausted is positioned such that we are in the middle
7754 * of a range in its set or not. (i_a and i_b point to the element beyond
7755 * the one we care about.) If in the set, we decrement 'count'; if 0, there
7756 * is potentially more to output.
7757 * There are four cases:
7758 * 1) Both weren't in their sets, count is 0, and remains 0. What's left
7759 * in the union is entirely from the non-exhausted set.
7760 * 2) Both were in their sets, count is 2. Nothing further should
7761 * be output, as everything that remains will be in the exhausted
7762 * list's set, hence in the union; decrementing to 1 but not 0 insures
7764 * 3) the exhausted was in its set, non-exhausted isn't, count is 1.
7765 * Nothing further should be output because the union includes
7766 * everything from the exhausted set. Not decrementing ensures that.
7767 * 4) the exhausted wasn't in its set, non-exhausted is, count is 1;
7768 * decrementing to 0 insures that we look at the remainder of the
7769 * non-exhausted set */
7770 if ((i_a != len_a && PREV_RANGE_MATCHES_INVLIST(i_a))
7771 || (i_b != len_b && PREV_RANGE_MATCHES_INVLIST(i_b)))
7776 /* The final length is what we've output so far, plus what else is about to
7777 * be output. (If 'count' is non-zero, then the input list we exhausted
7778 * has everything remaining up to the machine's limit in its set, and hence
7779 * in the union, so there will be no further output. */
7782 /* At most one of the subexpressions will be non-zero */
7783 len_u += (len_a - i_a) + (len_b - i_b);
7786 /* Set result to final length, which can change the pointer to array_u, so
7788 if (len_u != _invlist_len(u)) {
7789 invlist_set_len(u, len_u, *get_invlist_offset_addr(u));
7791 array_u = invlist_array(u);
7794 /* When 'count' is 0, the list that was exhausted (if one was shorter than
7795 * the other) ended with everything above it not in its set. That means
7796 * that the remaining part of the union is precisely the same as the
7797 * non-exhausted list, so can just copy it unchanged. (If both list were
7798 * exhausted at the same time, then the operations below will be both 0.)
7801 IV copy_count; /* At most one will have a non-zero copy count */
7802 if ((copy_count = len_a - i_a) > 0) {
7803 Copy(array_a + i_a, array_u + i_u, copy_count, UV);
7805 else if ((copy_count = len_b - i_b) > 0) {
7806 Copy(array_b + i_b, array_u + i_u, copy_count, UV);
7810 /* We may be removing a reference to one of the inputs */
7811 if (a == *output || b == *output) {
7812 assert(! invlist_is_iterating(*output));
7813 SvREFCNT_dec_NN(*output);
7821 Perl__invlist_intersection_maybe_complement_2nd(pTHX_ SV* const a, SV* const b, const bool complement_b, SV** i)
7823 /* Take the intersection of two inversion lists and point <i> to it. *i
7824 * SHOULD BE DEFINED upon input, and if it points to one of the two lists,
7825 * the reference count to that list will be decremented.
7826 * If <complement_b> is TRUE, the result will be the intersection of <a>
7827 * and the complement (or inversion) of <b> instead of <b> directly.
7829 * The basis for this comes from "Unicode Demystified" Chapter 13 by
7830 * Richard Gillam, published by Addison-Wesley, and explained at some
7831 * length there. The preface says to incorporate its examples into your
7832 * code at your own risk. In fact, it had bugs
7834 * The algorithm is like a merge sort, and is essentially the same as the
7838 const UV* array_a; /* a's array */
7840 UV len_a; /* length of a's array */
7843 SV* r; /* the resulting intersection */
7847 UV i_a = 0; /* current index into a's array */
7851 /* running count, as explained in the algorithm source book; items are
7852 * stopped accumulating and are output when the count changes to/from 2.
7853 * The count is incremented when we start a range that's in the set, and
7854 * decremented when we start a range that's not in the set. So its range
7855 * is 0 to 2. Only when the count is 2 is something in the intersection.
7859 PERL_ARGS_ASSERT__INVLIST_INTERSECTION_MAYBE_COMPLEMENT_2ND;
7862 /* Special case if either one is empty */
7863 len_a = (a == NULL) ? 0 : _invlist_len(a);
7864 if ((len_a == 0) || ((len_b = _invlist_len(b)) == 0)) {
7866 if (len_a != 0 && complement_b) {
7868 /* Here, 'a' is not empty, therefore from the above 'if', 'b' must
7869 * be empty. Here, also we are using 'b's complement, which hence
7870 * must be every possible code point. Thus the intersection is
7877 *i = invlist_clone(a);
7879 /* else *i is already 'a' */
7883 /* Here, 'a' or 'b' is empty and not using the complement of 'b'. The
7884 * intersection must be empty */
7891 *i = _new_invlist(0);
7895 /* Here both lists exist and are non-empty */
7896 array_a = invlist_array(a);
7897 array_b = invlist_array(b);
7899 /* If are to take the intersection of 'a' with the complement of b, set it
7900 * up so are looking at b's complement. */
7903 /* To complement, we invert: if the first element is 0, remove it. To
7904 * do this, we just pretend the array starts one later */
7905 if (array_b[0] == 0) {
7911 /* But if the first element is not zero, we pretend the list starts
7912 * at the 0 that is always stored immediately before the array. */
7918 /* Size the intersection for the worst case: that the intersection ends up
7919 * fragmenting everything to be completely disjoint */
7920 r= _new_invlist(len_a + len_b);
7922 /* Will contain U+0000 iff both components do */
7923 array_r = _invlist_array_init(r, len_a > 0 && array_a[0] == 0
7924 && len_b > 0 && array_b[0] == 0);
7926 /* Go through each list item by item, stopping when exhausted one of
7928 while (i_a < len_a && i_b < len_b) {
7929 UV cp; /* The element to potentially add to the intersection's
7931 bool cp_in_set; /* Is it in the input list's set or not */
7933 /* We need to take one or the other of the two inputs for the
7934 * intersection. Since we are merging two sorted lists, we take the
7935 * smaller of the next items. In case of a tie, we take the one that
7936 * is not in its set first (a difference from the union algorithm). If
7937 * we took one in the set first, it would increment the count, possibly
7938 * to 2 which would cause it to be output as starting a range in the
7939 * intersection, and the next time through we would take that same
7940 * number, and output it again as ending the set. By doing it the
7941 * opposite of this, there is no possibility that the count will be
7942 * momentarily incremented to 2. (In a tie and both are in the set or
7943 * both not in the set, it doesn't matter which we take first.) */
7944 if (array_a[i_a] < array_b[i_b]
7945 || (array_a[i_a] == array_b[i_b]
7946 && ! ELEMENT_RANGE_MATCHES_INVLIST(i_a)))
7948 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_a);
7952 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_b);
7956 /* Here, have chosen which of the two inputs to look at. Only output
7957 * if the running count changes to/from 2, which marks the
7958 * beginning/end of a range that's in the intersection */
7962 array_r[i_r++] = cp;
7967 array_r[i_r++] = cp;
7973 /* Here, we are finished going through at least one of the lists, which
7974 * means there is something remaining in at most one. We check if the list
7975 * that has been exhausted is positioned such that we are in the middle
7976 * of a range in its set or not. (i_a and i_b point to elements 1 beyond
7977 * the ones we care about.) There are four cases:
7978 * 1) Both weren't in their sets, count is 0, and remains 0. There's
7979 * nothing left in the intersection.
7980 * 2) Both were in their sets, count is 2 and perhaps is incremented to
7981 * above 2. What should be output is exactly that which is in the
7982 * non-exhausted set, as everything it has is also in the intersection
7983 * set, and everything it doesn't have can't be in the intersection
7984 * 3) The exhausted was in its set, non-exhausted isn't, count is 1, and
7985 * gets incremented to 2. Like the previous case, the intersection is
7986 * everything that remains in the non-exhausted set.
7987 * 4) the exhausted wasn't in its set, non-exhausted is, count is 1, and
7988 * remains 1. And the intersection has nothing more. */
7989 if ((i_a == len_a && PREV_RANGE_MATCHES_INVLIST(i_a))
7990 || (i_b == len_b && PREV_RANGE_MATCHES_INVLIST(i_b)))
7995 /* The final length is what we've output so far plus what else is in the
7996 * intersection. At most one of the subexpressions below will be non-zero */
7999 len_r += (len_a - i_a) + (len_b - i_b);
8002 /* Set result to final length, which can change the pointer to array_r, so
8004 if (len_r != _invlist_len(r)) {
8005 invlist_set_len(r, len_r, *get_invlist_offset_addr(r));
8007 array_r = invlist_array(r);
8010 /* Finish outputting any remaining */
8011 if (count >= 2) { /* At most one will have a non-zero copy count */
8013 if ((copy_count = len_a - i_a) > 0) {
8014 Copy(array_a + i_a, array_r + i_r, copy_count, UV);
8016 else if ((copy_count = len_b - i_b) > 0) {
8017 Copy(array_b + i_b, array_r + i_r, copy_count, UV);
8021 /* We may be removing a reference to one of the inputs */
8022 if (a == *i || b == *i) {
8023 assert(! invlist_is_iterating(*i));
8024 SvREFCNT_dec_NN(*i);
8032 Perl__add_range_to_invlist(pTHX_ SV* invlist, const UV start, const UV end)
8034 /* Add the range from 'start' to 'end' inclusive to the inversion list's
8035 * set. A pointer to the inversion list is returned. This may actually be
8036 * a new list, in which case the passed in one has been destroyed. The
8037 * passed in inversion list can be NULL, in which case a new one is created
8038 * with just the one range in it */
8043 if (invlist == NULL) {
8044 invlist = _new_invlist(2);
8048 len = _invlist_len(invlist);
8051 /* If comes after the final entry actually in the list, can just append it
8054 || (! ELEMENT_RANGE_MATCHES_INVLIST(len - 1)
8055 && start >= invlist_array(invlist)[len - 1]))
8057 _append_range_to_invlist(invlist, start, end);
8061 /* Here, can't just append things, create and return a new inversion list
8062 * which is the union of this range and the existing inversion list */
8063 range_invlist = _new_invlist(2);
8064 _append_range_to_invlist(range_invlist, start, end);
8066 _invlist_union(invlist, range_invlist, &invlist);
8068 /* The temporary can be freed */
8069 SvREFCNT_dec_NN(range_invlist);
8076 PERL_STATIC_INLINE SV*
8077 S_add_cp_to_invlist(pTHX_ SV* invlist, const UV cp) {
8078 return _add_range_to_invlist(invlist, cp, cp);
8081 #ifndef PERL_IN_XSUB_RE
8083 Perl__invlist_invert(pTHX_ SV* const invlist)
8085 /* Complement the input inversion list. This adds a 0 if the list didn't
8086 * have a zero; removes it otherwise. As described above, the data
8087 * structure is set up so that this is very efficient */
8089 PERL_ARGS_ASSERT__INVLIST_INVERT;
8091 assert(! invlist_is_iterating(invlist));
8093 /* The inverse of matching nothing is matching everything */
8094 if (_invlist_len(invlist) == 0) {
8095 _append_range_to_invlist(invlist, 0, UV_MAX);
8099 *get_invlist_offset_addr(invlist) = ! *get_invlist_offset_addr(invlist);
8103 Perl__invlist_invert_prop(pTHX_ SV* const invlist)
8105 /* Complement the input inversion list (which must be a Unicode property,
8106 * all of which don't match above the Unicode maximum code point.) And
8107 * Perl has chosen to not have the inversion match above that either. This
8108 * adds a 0x110000 if the list didn't end with it, and removes it if it did
8114 PERL_ARGS_ASSERT__INVLIST_INVERT_PROP;
8116 _invlist_invert(invlist);
8118 len = _invlist_len(invlist);
8120 if (len != 0) { /* If empty do nothing */
8121 array = invlist_array(invlist);
8122 if (array[len - 1] != PERL_UNICODE_MAX + 1) {
8123 /* Add 0x110000. First, grow if necessary */
8125 if (invlist_max(invlist) < len) {
8126 invlist_extend(invlist, len);
8127 array = invlist_array(invlist);
8129 invlist_set_len(invlist, len, *get_invlist_offset_addr(invlist));
8130 array[len - 1] = PERL_UNICODE_MAX + 1;
8132 else { /* Remove the 0x110000 */
8133 invlist_set_len(invlist, len - 1, *get_invlist_offset_addr(invlist));
8141 PERL_STATIC_INLINE SV*
8142 S_invlist_clone(pTHX_ SV* const invlist)
8145 /* Return a new inversion list that is a copy of the input one, which is
8148 /* Need to allocate extra space to accommodate Perl's addition of a
8149 * trailing NUL to SvPV's, since it thinks they are always strings */
8150 SV* new_invlist = _new_invlist(_invlist_len(invlist) + 1);
8151 STRLEN physical_length = SvCUR(invlist);
8152 bool offset = *(get_invlist_offset_addr(invlist));
8154 PERL_ARGS_ASSERT_INVLIST_CLONE;
8156 *(get_invlist_offset_addr(new_invlist)) = offset;
8157 invlist_set_len(new_invlist, _invlist_len(invlist), offset);
8158 Copy(SvPVX(invlist), SvPVX(new_invlist), physical_length, char);
8163 PERL_STATIC_INLINE STRLEN*
8164 S_get_invlist_iter_addr(pTHX_ SV* invlist)
8166 /* Return the address of the UV that contains the current iteration
8169 PERL_ARGS_ASSERT_GET_INVLIST_ITER_ADDR;
8171 assert(SvTYPE(invlist) == SVt_INVLIST);
8173 return &(((XINVLIST*) SvANY(invlist))->iterator);
8176 PERL_STATIC_INLINE void
8177 S_invlist_iterinit(pTHX_ SV* invlist) /* Initialize iterator for invlist */
8179 PERL_ARGS_ASSERT_INVLIST_ITERINIT;
8181 *get_invlist_iter_addr(invlist) = 0;
8184 PERL_STATIC_INLINE void
8185 S_invlist_iterfinish(pTHX_ SV* invlist)
8187 /* Terminate iterator for invlist. This is to catch development errors.
8188 * Any iteration that is interrupted before completed should call this
8189 * function. Functions that add code points anywhere else but to the end
8190 * of an inversion list assert that they are not in the middle of an
8191 * iteration. If they were, the addition would make the iteration
8192 * problematical: if the iteration hadn't reached the place where things
8193 * were being added, it would be ok */
8195 PERL_ARGS_ASSERT_INVLIST_ITERFINISH;
8197 *get_invlist_iter_addr(invlist) = (STRLEN) UV_MAX;
8201 S_invlist_iternext(pTHX_ SV* invlist, UV* start, UV* end)
8203 /* An C<invlist_iterinit> call on <invlist> must be used to set this up.
8204 * This call sets in <*start> and <*end>, the next range in <invlist>.
8205 * Returns <TRUE> if successful and the next call will return the next
8206 * range; <FALSE> if was already at the end of the list. If the latter,
8207 * <*start> and <*end> are unchanged, and the next call to this function
8208 * will start over at the beginning of the list */
8210 STRLEN* pos = get_invlist_iter_addr(invlist);
8211 UV len = _invlist_len(invlist);
8214 PERL_ARGS_ASSERT_INVLIST_ITERNEXT;
8217 *pos = (STRLEN) UV_MAX; /* Force iterinit() to be required next time */
8221 array = invlist_array(invlist);
8223 *start = array[(*pos)++];
8229 *end = array[(*pos)++] - 1;
8235 PERL_STATIC_INLINE bool
8236 S_invlist_is_iterating(pTHX_ SV* const invlist)
8238 PERL_ARGS_ASSERT_INVLIST_IS_ITERATING;
8240 return *(get_invlist_iter_addr(invlist)) < (STRLEN) UV_MAX;
8243 PERL_STATIC_INLINE UV
8244 S_invlist_highest(pTHX_ SV* const invlist)
8246 /* Returns the highest code point that matches an inversion list. This API
8247 * has an ambiguity, as it returns 0 under either the highest is actually
8248 * 0, or if the list is empty. If this distinction matters to you, check
8249 * for emptiness before calling this function */
8251 UV len = _invlist_len(invlist);
8254 PERL_ARGS_ASSERT_INVLIST_HIGHEST;
8260 array = invlist_array(invlist);
8262 /* The last element in the array in the inversion list always starts a
8263 * range that goes to infinity. That range may be for code points that are
8264 * matched in the inversion list, or it may be for ones that aren't
8265 * matched. In the latter case, the highest code point in the set is one
8266 * less than the beginning of this range; otherwise it is the final element
8267 * of this range: infinity */
8268 return (ELEMENT_RANGE_MATCHES_INVLIST(len - 1))
8270 : array[len - 1] - 1;
8273 #ifndef PERL_IN_XSUB_RE
8275 Perl__invlist_contents(pTHX_ SV* const invlist)
8277 /* Get the contents of an inversion list into a string SV so that they can
8278 * be printed out. It uses the format traditionally done for debug tracing
8282 SV* output = newSVpvs("\n");
8284 PERL_ARGS_ASSERT__INVLIST_CONTENTS;
8286 assert(! invlist_is_iterating(invlist));
8288 invlist_iterinit(invlist);
8289 while (invlist_iternext(invlist, &start, &end)) {
8290 if (end == UV_MAX) {
8291 Perl_sv_catpvf(aTHX_ output, "%04"UVXf"\tINFINITY\n", start);
8293 else if (end != start) {
8294 Perl_sv_catpvf(aTHX_ output, "%04"UVXf"\t%04"UVXf"\n",
8298 Perl_sv_catpvf(aTHX_ output, "%04"UVXf"\n", start);
8306 #ifndef PERL_IN_XSUB_RE
8308 Perl__invlist_dump(pTHX_ PerlIO *file, I32 level, const char * const indent, SV* const invlist)
8310 /* Designed to be called only by do_sv_dump(). Dumps out the ranges of the
8311 * inversion list 'invlist' to 'file' at 'level' Each line is prefixed by
8312 * the string 'indent'. The output looks like this:
8313 [0] 0x000A .. 0x000D
8315 [4] 0x2028 .. 0x2029
8316 [6] 0x3104 .. INFINITY
8317 * This means that the first range of code points matched by the list are
8318 * 0xA through 0xD; the second range contains only the single code point
8319 * 0x85, etc. An inversion list is an array of UVs. Two array elements
8320 * are used to define each range (except if the final range extends to
8321 * infinity, only a single element is needed). The array index of the
8322 * first element for the corresponding range is given in brackets. */
8327 PERL_ARGS_ASSERT__INVLIST_DUMP;
8329 if (invlist_is_iterating(invlist)) {
8330 Perl_dump_indent(aTHX_ level, file,
8331 "%sCan't dump inversion list because is in middle of iterating\n",
8336 invlist_iterinit(invlist);
8337 while (invlist_iternext(invlist, &start, &end)) {
8338 if (end == UV_MAX) {
8339 Perl_dump_indent(aTHX_ level, file,
8340 "%s[%"UVuf"] 0x%04"UVXf" .. INFINITY\n",
8341 indent, (UV)count, start);
8343 else if (end != start) {
8344 Perl_dump_indent(aTHX_ level, file,
8345 "%s[%"UVuf"] 0x%04"UVXf" .. 0x%04"UVXf"\n",
8346 indent, (UV)count, start, end);
8349 Perl_dump_indent(aTHX_ level, file, "%s[%"UVuf"] 0x%04"UVXf"\n",
8350 indent, (UV)count, start);
8357 #ifdef PERL_ARGS_ASSERT__INVLISTEQ
8359 S__invlistEQ(pTHX_ SV* const a, SV* const b, const bool complement_b)
8361 /* Return a boolean as to if the two passed in inversion lists are
8362 * identical. The final argument, if TRUE, says to take the complement of
8363 * the second inversion list before doing the comparison */
8365 const UV* array_a = invlist_array(a);
8366 const UV* array_b = invlist_array(b);
8367 UV len_a = _invlist_len(a);
8368 UV len_b = _invlist_len(b);
8370 UV i = 0; /* current index into the arrays */
8371 bool retval = TRUE; /* Assume are identical until proven otherwise */
8373 PERL_ARGS_ASSERT__INVLISTEQ;
8375 /* If are to compare 'a' with the complement of b, set it
8376 * up so are looking at b's complement. */
8379 /* The complement of nothing is everything, so <a> would have to have
8380 * just one element, starting at zero (ending at infinity) */
8382 return (len_a == 1 && array_a[0] == 0);
8384 else if (array_b[0] == 0) {
8386 /* Otherwise, to complement, we invert. Here, the first element is
8387 * 0, just remove it. To do this, we just pretend the array starts
8395 /* But if the first element is not zero, we pretend the list starts
8396 * at the 0 that is always stored immediately before the array. */
8402 /* Make sure that the lengths are the same, as well as the final element
8403 * before looping through the remainder. (Thus we test the length, final,
8404 * and first elements right off the bat) */
8405 if (len_a != len_b || array_a[len_a-1] != array_b[len_a-1]) {
8408 else for (i = 0; i < len_a - 1; i++) {
8409 if (array_a[i] != array_b[i]) {
8419 #undef HEADER_LENGTH
8420 #undef TO_INTERNAL_SIZE
8421 #undef FROM_INTERNAL_SIZE
8422 #undef INVLIST_VERSION_ID
8424 /* End of inversion list object */
8427 S_parse_lparen_question_flags(pTHX_ struct RExC_state_t *pRExC_state)
8429 /* This parses the flags that are in either the '(?foo)' or '(?foo:bar)'
8430 * constructs, and updates RExC_flags with them. On input, RExC_parse
8431 * should point to the first flag; it is updated on output to point to the
8432 * final ')' or ':'. There needs to be at least one flag, or this will
8435 /* for (?g), (?gc), and (?o) warnings; warning
8436 about (?c) will warn about (?g) -- japhy */
8438 #define WASTED_O 0x01
8439 #define WASTED_G 0x02
8440 #define WASTED_C 0x04
8441 #define WASTED_GC (WASTED_G|WASTED_C)
8442 I32 wastedflags = 0x00;
8443 U32 posflags = 0, negflags = 0;
8444 U32 *flagsp = &posflags;
8445 char has_charset_modifier = '\0';
8447 bool has_use_defaults = FALSE;
8448 const char* const seqstart = RExC_parse - 1; /* Point to the '?' */
8450 PERL_ARGS_ASSERT_PARSE_LPAREN_QUESTION_FLAGS;
8452 /* '^' as an initial flag sets certain defaults */
8453 if (UCHARAT(RExC_parse) == '^') {
8455 has_use_defaults = TRUE;
8456 STD_PMMOD_FLAGS_CLEAR(&RExC_flags);
8457 set_regex_charset(&RExC_flags, (RExC_utf8 || RExC_uni_semantics)
8458 ? REGEX_UNICODE_CHARSET
8459 : REGEX_DEPENDS_CHARSET);
8462 cs = get_regex_charset(RExC_flags);
8463 if (cs == REGEX_DEPENDS_CHARSET
8464 && (RExC_utf8 || RExC_uni_semantics))
8466 cs = REGEX_UNICODE_CHARSET;
8469 while (*RExC_parse) {
8470 /* && strchr("iogcmsx", *RExC_parse) */
8471 /* (?g), (?gc) and (?o) are useless here
8472 and must be globally applied -- japhy */
8473 switch (*RExC_parse) {
8475 /* Code for the imsx flags */
8476 CASE_STD_PMMOD_FLAGS_PARSE_SET(flagsp);
8478 case LOCALE_PAT_MOD:
8479 if (has_charset_modifier) {
8480 goto excess_modifier;
8482 else if (flagsp == &negflags) {
8485 cs = REGEX_LOCALE_CHARSET;
8486 has_charset_modifier = LOCALE_PAT_MOD;
8487 RExC_contains_locale = 1;
8489 case UNICODE_PAT_MOD:
8490 if (has_charset_modifier) {
8491 goto excess_modifier;
8493 else if (flagsp == &negflags) {
8496 cs = REGEX_UNICODE_CHARSET;
8497 has_charset_modifier = UNICODE_PAT_MOD;
8499 case ASCII_RESTRICT_PAT_MOD:
8500 if (flagsp == &negflags) {
8503 if (has_charset_modifier) {
8504 if (cs != REGEX_ASCII_RESTRICTED_CHARSET) {
8505 goto excess_modifier;
8507 /* Doubled modifier implies more restricted */
8508 cs = REGEX_ASCII_MORE_RESTRICTED_CHARSET;
8511 cs = REGEX_ASCII_RESTRICTED_CHARSET;
8513 has_charset_modifier = ASCII_RESTRICT_PAT_MOD;
8515 case DEPENDS_PAT_MOD:
8516 if (has_use_defaults) {
8517 goto fail_modifiers;
8519 else if (flagsp == &negflags) {
8522 else if (has_charset_modifier) {
8523 goto excess_modifier;
8526 /* The dual charset means unicode semantics if the
8527 * pattern (or target, not known until runtime) are
8528 * utf8, or something in the pattern indicates unicode
8530 cs = (RExC_utf8 || RExC_uni_semantics)
8531 ? REGEX_UNICODE_CHARSET
8532 : REGEX_DEPENDS_CHARSET;
8533 has_charset_modifier = DEPENDS_PAT_MOD;
8537 if (has_charset_modifier == ASCII_RESTRICT_PAT_MOD) {
8538 vFAIL2("Regexp modifier \"%c\" may appear a maximum of twice", ASCII_RESTRICT_PAT_MOD);
8540 else if (has_charset_modifier == *(RExC_parse - 1)) {
8541 vFAIL2("Regexp modifier \"%c\" may not appear twice", *(RExC_parse - 1));
8544 vFAIL3("Regexp modifiers \"%c\" and \"%c\" are mutually exclusive", has_charset_modifier, *(RExC_parse - 1));
8549 vFAIL2("Regexp modifier \"%c\" may not appear after the \"-\"", *(RExC_parse - 1));
8551 case ONCE_PAT_MOD: /* 'o' */
8552 case GLOBAL_PAT_MOD: /* 'g' */
8553 if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
8554 const I32 wflagbit = *RExC_parse == 'o' ? WASTED_O : WASTED_G;
8555 if (! (wastedflags & wflagbit) ) {
8556 wastedflags |= wflagbit;
8557 /* diag_listed_as: Useless (?-%s) - don't use /%s modifier in regex; marked by <-- HERE in m/%s/ */
8560 "Useless (%s%c) - %suse /%c modifier",
8561 flagsp == &negflags ? "?-" : "?",
8563 flagsp == &negflags ? "don't " : "",
8570 case CONTINUE_PAT_MOD: /* 'c' */
8571 if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
8572 if (! (wastedflags & WASTED_C) ) {
8573 wastedflags |= WASTED_GC;
8574 /* diag_listed_as: Useless (?-%s) - don't use /%s modifier in regex; marked by <-- HERE in m/%s/ */
8577 "Useless (%sc) - %suse /gc modifier",
8578 flagsp == &negflags ? "?-" : "?",
8579 flagsp == &negflags ? "don't " : ""
8584 case KEEPCOPY_PAT_MOD: /* 'p' */
8585 if (flagsp == &negflags) {
8587 ckWARNreg(RExC_parse + 1,"Useless use of (?-p)");
8589 *flagsp |= RXf_PMf_KEEPCOPY;
8593 /* A flag is a default iff it is following a minus, so
8594 * if there is a minus, it means will be trying to
8595 * re-specify a default which is an error */
8596 if (has_use_defaults || flagsp == &negflags) {
8597 goto fail_modifiers;
8600 wastedflags = 0; /* reset so (?g-c) warns twice */
8604 RExC_flags |= posflags;
8605 RExC_flags &= ~negflags;
8606 set_regex_charset(&RExC_flags, cs);
8612 vFAIL3("Sequence (%.*s...) not recognized",
8613 RExC_parse-seqstart, seqstart);
8622 - reg - regular expression, i.e. main body or parenthesized thing
8624 * Caller must absorb opening parenthesis.
8626 * Combining parenthesis handling with the base level of regular expression
8627 * is a trifle forced, but the need to tie the tails of the branches to what
8628 * follows makes it hard to avoid.
8630 #define REGTAIL(x,y,z) regtail((x),(y),(z),depth+1)
8632 #define REGTAIL_STUDY(x,y,z) regtail_study((x),(y),(z),depth+1)
8634 #define REGTAIL_STUDY(x,y,z) regtail((x),(y),(z),depth+1)
8637 /* Returns NULL, setting *flagp to TRYAGAIN at the end of (?) that only sets
8638 flags. Returns NULL, setting *flagp to RESTART_UTF8 if the sizing scan
8639 needs to be restarted.
8640 Otherwise would only return NULL if regbranch() returns NULL, which
8643 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
8644 /* paren: Parenthesized? 0=top; 1,2=inside '(': changed to letter.
8645 * 2 is like 1, but indicates that nextchar() has been called to advance
8646 * RExC_parse beyond the '('. Things like '(?' are indivisible tokens, and
8647 * this flag alerts us to the need to check for that */
8650 regnode *ret; /* Will be the head of the group. */
8653 regnode *ender = NULL;
8656 U32 oregflags = RExC_flags;
8657 bool have_branch = 0;
8659 I32 freeze_paren = 0;
8660 I32 after_freeze = 0;
8662 char * parse_start = RExC_parse; /* MJD */
8663 char * const oregcomp_parse = RExC_parse;
8665 GET_RE_DEBUG_FLAGS_DECL;
8667 PERL_ARGS_ASSERT_REG;
8668 DEBUG_PARSE("reg ");
8670 *flagp = 0; /* Tentatively. */
8673 /* Make an OPEN node, if parenthesized. */
8676 /* Under /x, space and comments can be gobbled up between the '(' and
8677 * here (if paren ==2). The forms '(*VERB' and '(?...' disallow such
8678 * intervening space, as the sequence is a token, and a token should be
8680 bool has_intervening_patws = paren == 2 && *(RExC_parse - 1) != '(';
8682 if ( *RExC_parse == '*') { /* (*VERB:ARG) */
8683 char *start_verb = RExC_parse;
8684 STRLEN verb_len = 0;
8685 char *start_arg = NULL;
8686 unsigned char op = 0;
8688 int internal_argval = 0; /* internal_argval is only useful if !argok */
8690 if (has_intervening_patws && SIZE_ONLY) {
8691 ckWARNregdep(RExC_parse + 1, "In '(*VERB...)', splitting the initial '(*' is deprecated");
8693 while ( *RExC_parse && *RExC_parse != ')' ) {
8694 if ( *RExC_parse == ':' ) {
8695 start_arg = RExC_parse + 1;
8701 verb_len = RExC_parse - start_verb;
8704 while ( *RExC_parse && *RExC_parse != ')' )
8706 if ( *RExC_parse != ')' )
8707 vFAIL("Unterminated verb pattern argument");
8708 if ( RExC_parse == start_arg )
8711 if ( *RExC_parse != ')' )
8712 vFAIL("Unterminated verb pattern");
8715 switch ( *start_verb ) {
8716 case 'A': /* (*ACCEPT) */
8717 if ( memEQs(start_verb,verb_len,"ACCEPT") ) {
8719 internal_argval = RExC_nestroot;
8722 case 'C': /* (*COMMIT) */
8723 if ( memEQs(start_verb,verb_len,"COMMIT") )
8726 case 'F': /* (*FAIL) */
8727 if ( verb_len==1 || memEQs(start_verb,verb_len,"FAIL") ) {
8732 case ':': /* (*:NAME) */
8733 case 'M': /* (*MARK:NAME) */
8734 if ( verb_len==0 || memEQs(start_verb,verb_len,"MARK") ) {
8739 case 'P': /* (*PRUNE) */
8740 if ( memEQs(start_verb,verb_len,"PRUNE") )
8743 case 'S': /* (*SKIP) */
8744 if ( memEQs(start_verb,verb_len,"SKIP") )
8747 case 'T': /* (*THEN) */
8748 /* [19:06] <TimToady> :: is then */
8749 if ( memEQs(start_verb,verb_len,"THEN") ) {
8751 RExC_seen |= REG_SEEN_CUTGROUP;
8757 vFAIL3("Unknown verb pattern '%.*s'",
8758 verb_len, start_verb);
8761 if ( start_arg && internal_argval ) {
8762 vFAIL3("Verb pattern '%.*s' may not have an argument",
8763 verb_len, start_verb);
8764 } else if ( argok < 0 && !start_arg ) {
8765 vFAIL3("Verb pattern '%.*s' has a mandatory argument",
8766 verb_len, start_verb);
8768 ret = reganode(pRExC_state, op, internal_argval);
8769 if ( ! internal_argval && ! SIZE_ONLY ) {
8771 SV *sv = newSVpvn( start_arg, RExC_parse - start_arg);
8772 ARG(ret) = add_data( pRExC_state, 1, "S" );
8773 RExC_rxi->data->data[ARG(ret)]=(void*)sv;
8780 if (!internal_argval)
8781 RExC_seen |= REG_SEEN_VERBARG;
8782 } else if ( start_arg ) {
8783 vFAIL3("Verb pattern '%.*s' may not have an argument",
8784 verb_len, start_verb);
8786 ret = reg_node(pRExC_state, op);
8788 nextchar(pRExC_state);
8791 else if (*RExC_parse == '?') { /* (?...) */
8792 bool is_logical = 0;
8793 const char * const seqstart = RExC_parse;
8794 if (has_intervening_patws && SIZE_ONLY) {
8795 ckWARNregdep(RExC_parse + 1, "In '(?...)', splitting the initial '(?' is deprecated");
8799 paren = *RExC_parse++;
8800 ret = NULL; /* For look-ahead/behind. */
8803 case 'P': /* (?P...) variants for those used to PCRE/Python */
8804 paren = *RExC_parse++;
8805 if ( paren == '<') /* (?P<...>) named capture */
8807 else if (paren == '>') { /* (?P>name) named recursion */
8808 goto named_recursion;
8810 else if (paren == '=') { /* (?P=...) named backref */
8811 /* this pretty much dupes the code for \k<NAME> in regatom(), if
8812 you change this make sure you change that */
8813 char* name_start = RExC_parse;
8815 SV *sv_dat = reg_scan_name(pRExC_state,
8816 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
8817 if (RExC_parse == name_start || *RExC_parse != ')')
8818 vFAIL2("Sequence %.3s... not terminated",parse_start);
8821 num = add_data( pRExC_state, 1, "S" );
8822 RExC_rxi->data->data[num]=(void*)sv_dat;
8823 SvREFCNT_inc_simple_void(sv_dat);
8826 ret = reganode(pRExC_state,
8829 : (ASCII_FOLD_RESTRICTED)
8831 : (AT_LEAST_UNI_SEMANTICS)
8839 Set_Node_Offset(ret, parse_start+1);
8840 Set_Node_Cur_Length(ret, parse_start);
8842 nextchar(pRExC_state);
8846 vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
8848 case '<': /* (?<...) */
8849 if (*RExC_parse == '!')
8851 else if (*RExC_parse != '=')
8857 case '\'': /* (?'...') */
8858 name_start= RExC_parse;
8859 svname = reg_scan_name(pRExC_state,
8860 SIZE_ONLY ? /* reverse test from the others */
8861 REG_RSN_RETURN_NAME :
8862 REG_RSN_RETURN_NULL);
8863 if (RExC_parse == name_start) {
8865 vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
8868 if (*RExC_parse != paren)
8869 vFAIL2("Sequence (?%c... not terminated",
8870 paren=='>' ? '<' : paren);
8874 if (!svname) /* shouldn't happen */
8876 "panic: reg_scan_name returned NULL");
8877 if (!RExC_paren_names) {
8878 RExC_paren_names= newHV();
8879 sv_2mortal(MUTABLE_SV(RExC_paren_names));
8881 RExC_paren_name_list= newAV();
8882 sv_2mortal(MUTABLE_SV(RExC_paren_name_list));
8885 he_str = hv_fetch_ent( RExC_paren_names, svname, 1, 0 );
8887 sv_dat = HeVAL(he_str);
8889 /* croak baby croak */
8891 "panic: paren_name hash element allocation failed");
8892 } else if ( SvPOK(sv_dat) ) {
8893 /* (?|...) can mean we have dupes so scan to check
8894 its already been stored. Maybe a flag indicating
8895 we are inside such a construct would be useful,
8896 but the arrays are likely to be quite small, so
8897 for now we punt -- dmq */
8898 IV count = SvIV(sv_dat);
8899 I32 *pv = (I32*)SvPVX(sv_dat);
8901 for ( i = 0 ; i < count ; i++ ) {
8902 if ( pv[i] == RExC_npar ) {
8908 pv = (I32*)SvGROW(sv_dat, SvCUR(sv_dat) + sizeof(I32)+1);
8909 SvCUR_set(sv_dat, SvCUR(sv_dat) + sizeof(I32));
8910 pv[count] = RExC_npar;
8911 SvIV_set(sv_dat, SvIVX(sv_dat) + 1);
8914 (void)SvUPGRADE(sv_dat,SVt_PVNV);
8915 sv_setpvn(sv_dat, (char *)&(RExC_npar), sizeof(I32));
8917 SvIV_set(sv_dat, 1);
8920 /* Yes this does cause a memory leak in debugging Perls */
8921 if (!av_store(RExC_paren_name_list, RExC_npar, SvREFCNT_inc(svname)))
8922 SvREFCNT_dec_NN(svname);
8925 /*sv_dump(sv_dat);*/
8927 nextchar(pRExC_state);
8929 goto capturing_parens;
8931 RExC_seen |= REG_SEEN_LOOKBEHIND;
8932 RExC_in_lookbehind++;
8934 case '=': /* (?=...) */
8935 RExC_seen_zerolen++;
8937 case '!': /* (?!...) */
8938 RExC_seen_zerolen++;
8939 if (*RExC_parse == ')') {
8940 ret=reg_node(pRExC_state, OPFAIL);
8941 nextchar(pRExC_state);
8945 case '|': /* (?|...) */
8946 /* branch reset, behave like a (?:...) except that
8947 buffers in alternations share the same numbers */
8949 after_freeze = freeze_paren = RExC_npar;
8951 case ':': /* (?:...) */
8952 case '>': /* (?>...) */
8954 case '$': /* (?$...) */
8955 case '@': /* (?@...) */
8956 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
8958 case '#': /* (?#...) */
8959 /* XXX As soon as we disallow separating the '?' and '*' (by
8960 * spaces or (?#...) comment), it is believed that this case
8961 * will be unreachable and can be removed. See
8963 while (*RExC_parse && *RExC_parse != ')')
8965 if (*RExC_parse != ')')
8966 FAIL("Sequence (?#... not terminated");
8967 nextchar(pRExC_state);
8970 case '0' : /* (?0) */
8971 case 'R' : /* (?R) */
8972 if (*RExC_parse != ')')
8973 FAIL("Sequence (?R) not terminated");
8974 ret = reg_node(pRExC_state, GOSTART);
8975 *flagp |= POSTPONED;
8976 nextchar(pRExC_state);
8979 { /* named and numeric backreferences */
8981 case '&': /* (?&NAME) */
8982 parse_start = RExC_parse - 1;
8985 SV *sv_dat = reg_scan_name(pRExC_state,
8986 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
8987 num = sv_dat ? *((I32 *)SvPVX(sv_dat)) : 0;
8989 goto gen_recurse_regop;
8990 assert(0); /* NOT REACHED */
8992 if (!(RExC_parse[0] >= '1' && RExC_parse[0] <= '9')) {
8994 vFAIL("Illegal pattern");
8996 goto parse_recursion;
8998 case '-': /* (?-1) */
8999 if (!(RExC_parse[0] >= '1' && RExC_parse[0] <= '9')) {
9000 RExC_parse--; /* rewind to let it be handled later */
9004 case '1': case '2': case '3': case '4': /* (?1) */
9005 case '5': case '6': case '7': case '8': case '9':
9008 num = atoi(RExC_parse);
9009 parse_start = RExC_parse - 1; /* MJD */
9010 if (*RExC_parse == '-')
9012 while (isDIGIT(*RExC_parse))
9014 if (*RExC_parse!=')')
9015 vFAIL("Expecting close bracket");
9018 if ( paren == '-' ) {
9020 Diagram of capture buffer numbering.
9021 Top line is the normal capture buffer numbers
9022 Bottom line is the negative indexing as from
9026 /(a(x)y)(a(b(c(?-2)d)e)f)(g(h))/
9030 num = RExC_npar + num;
9033 vFAIL("Reference to nonexistent group");
9035 } else if ( paren == '+' ) {
9036 num = RExC_npar + num - 1;
9039 ret = reganode(pRExC_state, GOSUB, num);
9041 if (num > (I32)RExC_rx->nparens) {
9043 vFAIL("Reference to nonexistent group");
9045 ARG2L_SET( ret, RExC_recurse_count++);
9047 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
9048 "Recurse #%"UVuf" to %"IVdf"\n", (UV)ARG(ret), (IV)ARG2L(ret)));
9052 RExC_seen |= REG_SEEN_RECURSE;
9053 Set_Node_Length(ret, 1 + regarglen[OP(ret)]); /* MJD */
9054 Set_Node_Offset(ret, parse_start); /* MJD */
9056 *flagp |= POSTPONED;
9057 nextchar(pRExC_state);
9059 } /* named and numeric backreferences */
9060 assert(0); /* NOT REACHED */
9062 case '?': /* (??...) */
9064 if (*RExC_parse != '{') {
9066 vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
9069 *flagp |= POSTPONED;
9070 paren = *RExC_parse++;
9072 case '{': /* (?{...}) */
9075 struct reg_code_block *cb;
9077 RExC_seen_zerolen++;
9079 if ( !pRExC_state->num_code_blocks
9080 || pRExC_state->code_index >= pRExC_state->num_code_blocks
9081 || pRExC_state->code_blocks[pRExC_state->code_index].start
9082 != (STRLEN)((RExC_parse -3 - (is_logical ? 1 : 0))
9085 if (RExC_pm_flags & PMf_USE_RE_EVAL)
9086 FAIL("panic: Sequence (?{...}): no code block found\n");
9087 FAIL("Eval-group not allowed at runtime, use re 'eval'");
9089 /* this is a pre-compiled code block (?{...}) */
9090 cb = &pRExC_state->code_blocks[pRExC_state->code_index];
9091 RExC_parse = RExC_start + cb->end;
9094 if (cb->src_regex) {
9095 n = add_data(pRExC_state, 2, "rl");
9096 RExC_rxi->data->data[n] =
9097 (void*)SvREFCNT_inc((SV*)cb->src_regex);
9098 RExC_rxi->data->data[n+1] = (void*)o;
9101 n = add_data(pRExC_state, 1,
9102 (RExC_pm_flags & PMf_HAS_CV) ? "L" : "l");
9103 RExC_rxi->data->data[n] = (void*)o;
9106 pRExC_state->code_index++;
9107 nextchar(pRExC_state);
9111 ret = reg_node(pRExC_state, LOGICAL);
9112 eval = reganode(pRExC_state, EVAL, n);
9115 /* for later propagation into (??{}) return value */
9116 eval->flags = (U8) (RExC_flags & RXf_PMf_COMPILETIME);
9118 REGTAIL(pRExC_state, ret, eval);
9119 /* deal with the length of this later - MJD */
9122 ret = reganode(pRExC_state, EVAL, n);
9123 Set_Node_Length(ret, RExC_parse - parse_start + 1);
9124 Set_Node_Offset(ret, parse_start);
9127 case '(': /* (?(?{...})...) and (?(?=...)...) */
9130 if (RExC_parse[0] == '?') { /* (?(?...)) */
9131 if (RExC_parse[1] == '=' || RExC_parse[1] == '!'
9132 || RExC_parse[1] == '<'
9133 || RExC_parse[1] == '{') { /* Lookahead or eval. */
9137 ret = reg_node(pRExC_state, LOGICAL);
9141 tail = reg(pRExC_state, 1, &flag, depth+1);
9142 if (flag & RESTART_UTF8) {
9143 *flagp = RESTART_UTF8;
9146 REGTAIL(pRExC_state, ret, tail);
9150 else if ( RExC_parse[0] == '<' /* (?(<NAME>)...) */
9151 || RExC_parse[0] == '\'' ) /* (?('NAME')...) */
9153 char ch = RExC_parse[0] == '<' ? '>' : '\'';
9154 char *name_start= RExC_parse++;
9156 SV *sv_dat=reg_scan_name(pRExC_state,
9157 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
9158 if (RExC_parse == name_start || *RExC_parse != ch)
9159 vFAIL2("Sequence (?(%c... not terminated",
9160 (ch == '>' ? '<' : ch));
9163 num = add_data( pRExC_state, 1, "S" );
9164 RExC_rxi->data->data[num]=(void*)sv_dat;
9165 SvREFCNT_inc_simple_void(sv_dat);
9167 ret = reganode(pRExC_state,NGROUPP,num);
9168 goto insert_if_check_paren;
9170 else if (RExC_parse[0] == 'D' &&
9171 RExC_parse[1] == 'E' &&
9172 RExC_parse[2] == 'F' &&
9173 RExC_parse[3] == 'I' &&
9174 RExC_parse[4] == 'N' &&
9175 RExC_parse[5] == 'E')
9177 ret = reganode(pRExC_state,DEFINEP,0);
9180 goto insert_if_check_paren;
9182 else if (RExC_parse[0] == 'R') {
9185 if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
9186 parno = atoi(RExC_parse++);
9187 while (isDIGIT(*RExC_parse))
9189 } else if (RExC_parse[0] == '&') {
9192 sv_dat = reg_scan_name(pRExC_state,
9193 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
9194 parno = sv_dat ? *((I32 *)SvPVX(sv_dat)) : 0;
9196 ret = reganode(pRExC_state,INSUBP,parno);
9197 goto insert_if_check_paren;
9199 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
9202 parno = atoi(RExC_parse++);
9204 while (isDIGIT(*RExC_parse))
9206 ret = reganode(pRExC_state, GROUPP, parno);
9208 insert_if_check_paren:
9209 if ((c = *nextchar(pRExC_state)) != ')')
9210 vFAIL("Switch condition not recognized");
9212 REGTAIL(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
9213 br = regbranch(pRExC_state, &flags, 1,depth+1);
9215 if (flags & RESTART_UTF8) {
9216 *flagp = RESTART_UTF8;
9219 FAIL2("panic: regbranch returned NULL, flags=%#"UVxf"",
9222 REGTAIL(pRExC_state, br, reganode(pRExC_state, LONGJMP, 0));
9223 c = *nextchar(pRExC_state);
9228 vFAIL("(?(DEFINE)....) does not allow branches");
9229 lastbr = reganode(pRExC_state, IFTHEN, 0); /* Fake one for optimizer. */
9230 if (!regbranch(pRExC_state, &flags, 1,depth+1)) {
9231 if (flags & RESTART_UTF8) {
9232 *flagp = RESTART_UTF8;
9235 FAIL2("panic: regbranch returned NULL, flags=%#"UVxf"",
9238 REGTAIL(pRExC_state, ret, lastbr);
9241 c = *nextchar(pRExC_state);
9246 vFAIL("Switch (?(condition)... contains too many branches");
9247 ender = reg_node(pRExC_state, TAIL);
9248 REGTAIL(pRExC_state, br, ender);
9250 REGTAIL(pRExC_state, lastbr, ender);
9251 REGTAIL(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
9254 REGTAIL(pRExC_state, ret, ender);
9255 RExC_size++; /* XXX WHY do we need this?!!
9256 For large programs it seems to be required
9257 but I can't figure out why. -- dmq*/
9261 vFAIL2("Unknown switch condition (?(%.2s", RExC_parse);
9264 case '[': /* (?[ ... ]) */
9265 return handle_regex_sets(pRExC_state, NULL, flagp, depth,
9268 RExC_parse--; /* for vFAIL to print correctly */
9269 vFAIL("Sequence (? incomplete");
9271 default: /* e.g., (?i) */
9274 parse_lparen_question_flags(pRExC_state);
9275 if (UCHARAT(RExC_parse) != ':') {
9276 nextchar(pRExC_state);
9281 nextchar(pRExC_state);
9291 ret = reganode(pRExC_state, OPEN, parno);
9294 RExC_nestroot = parno;
9295 if (RExC_seen & REG_SEEN_RECURSE
9296 && !RExC_open_parens[parno-1])
9298 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
9299 "Setting open paren #%"IVdf" to %d\n",
9300 (IV)parno, REG_NODE_NUM(ret)));
9301 RExC_open_parens[parno-1]= ret;
9304 Set_Node_Length(ret, 1); /* MJD */
9305 Set_Node_Offset(ret, RExC_parse); /* MJD */
9313 /* Pick up the branches, linking them together. */
9314 parse_start = RExC_parse; /* MJD */
9315 br = regbranch(pRExC_state, &flags, 1,depth+1);
9317 /* branch_len = (paren != 0); */
9320 if (flags & RESTART_UTF8) {
9321 *flagp = RESTART_UTF8;
9324 FAIL2("panic: regbranch returned NULL, flags=%#"UVxf"", (UV) flags);
9326 if (*RExC_parse == '|') {
9327 if (!SIZE_ONLY && RExC_extralen) {
9328 reginsert(pRExC_state, BRANCHJ, br, depth+1);
9331 reginsert(pRExC_state, BRANCH, br, depth+1);
9332 Set_Node_Length(br, paren != 0);
9333 Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
9337 RExC_extralen += 1; /* For BRANCHJ-BRANCH. */
9339 else if (paren == ':') {
9340 *flagp |= flags&SIMPLE;
9342 if (is_open) { /* Starts with OPEN. */
9343 REGTAIL(pRExC_state, ret, br); /* OPEN -> first. */
9345 else if (paren != '?') /* Not Conditional */
9347 *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
9349 while (*RExC_parse == '|') {
9350 if (!SIZE_ONLY && RExC_extralen) {
9351 ender = reganode(pRExC_state, LONGJMP,0);
9352 REGTAIL(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender); /* Append to the previous. */
9355 RExC_extralen += 2; /* Account for LONGJMP. */
9356 nextchar(pRExC_state);
9358 if (RExC_npar > after_freeze)
9359 after_freeze = RExC_npar;
9360 RExC_npar = freeze_paren;
9362 br = regbranch(pRExC_state, &flags, 0, depth+1);
9365 if (flags & RESTART_UTF8) {
9366 *flagp = RESTART_UTF8;
9369 FAIL2("panic: regbranch returned NULL, flags=%#"UVxf"", (UV) flags);
9371 REGTAIL(pRExC_state, lastbr, br); /* BRANCH -> BRANCH. */
9373 *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
9376 if (have_branch || paren != ':') {
9377 /* Make a closing node, and hook it on the end. */
9380 ender = reg_node(pRExC_state, TAIL);
9383 ender = reganode(pRExC_state, CLOSE, parno);
9384 if (!SIZE_ONLY && RExC_seen & REG_SEEN_RECURSE) {
9385 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
9386 "Setting close paren #%"IVdf" to %d\n",
9387 (IV)parno, REG_NODE_NUM(ender)));
9388 RExC_close_parens[parno-1]= ender;
9389 if (RExC_nestroot == parno)
9392 Set_Node_Offset(ender,RExC_parse+1); /* MJD */
9393 Set_Node_Length(ender,1); /* MJD */
9399 *flagp &= ~HASWIDTH;
9402 ender = reg_node(pRExC_state, SUCCEED);
9405 ender = reg_node(pRExC_state, END);
9407 assert(!RExC_opend); /* there can only be one! */
9412 DEBUG_PARSE_r(if (!SIZE_ONLY) {
9413 SV * const mysv_val1=sv_newmortal();
9414 SV * const mysv_val2=sv_newmortal();
9415 DEBUG_PARSE_MSG("lsbr");
9416 regprop(RExC_rx, mysv_val1, lastbr);
9417 regprop(RExC_rx, mysv_val2, ender);
9418 PerlIO_printf(Perl_debug_log, "~ tying lastbr %s (%"IVdf") to ender %s (%"IVdf") offset %"IVdf"\n",
9419 SvPV_nolen_const(mysv_val1),
9420 (IV)REG_NODE_NUM(lastbr),
9421 SvPV_nolen_const(mysv_val2),
9422 (IV)REG_NODE_NUM(ender),
9423 (IV)(ender - lastbr)
9426 REGTAIL(pRExC_state, lastbr, ender);
9428 if (have_branch && !SIZE_ONLY) {
9431 RExC_seen |= REG_TOP_LEVEL_BRANCHES;
9433 /* Hook the tails of the branches to the closing node. */
9434 for (br = ret; br; br = regnext(br)) {
9435 const U8 op = PL_regkind[OP(br)];
9437 REGTAIL_STUDY(pRExC_state, NEXTOPER(br), ender);
9438 if (OP(NEXTOPER(br)) != NOTHING || regnext(NEXTOPER(br)) != ender)
9441 else if (op == BRANCHJ) {
9442 REGTAIL_STUDY(pRExC_state, NEXTOPER(NEXTOPER(br)), ender);
9443 /* for now we always disable this optimisation * /
9444 if (OP(NEXTOPER(NEXTOPER(br))) != NOTHING || regnext(NEXTOPER(NEXTOPER(br))) != ender)
9450 br= PL_regkind[OP(ret)] != BRANCH ? regnext(ret) : ret;
9451 DEBUG_PARSE_r(if (!SIZE_ONLY) {
9452 SV * const mysv_val1=sv_newmortal();
9453 SV * const mysv_val2=sv_newmortal();
9454 DEBUG_PARSE_MSG("NADA");
9455 regprop(RExC_rx, mysv_val1, ret);
9456 regprop(RExC_rx, mysv_val2, ender);
9457 PerlIO_printf(Perl_debug_log, "~ converting ret %s (%"IVdf") to ender %s (%"IVdf") offset %"IVdf"\n",
9458 SvPV_nolen_const(mysv_val1),
9459 (IV)REG_NODE_NUM(ret),
9460 SvPV_nolen_const(mysv_val2),
9461 (IV)REG_NODE_NUM(ender),
9466 if (OP(ender) == TAIL) {
9471 for ( opt= br + 1; opt < ender ; opt++ )
9473 NEXT_OFF(br)= ender - br;
9481 static const char parens[] = "=!<,>";
9483 if (paren && (p = strchr(parens, paren))) {
9484 U8 node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
9485 int flag = (p - parens) > 1;
9488 node = SUSPEND, flag = 0;
9489 reginsert(pRExC_state, node,ret, depth+1);
9490 Set_Node_Cur_Length(ret, parse_start);
9491 Set_Node_Offset(ret, parse_start + 1);
9493 REGTAIL_STUDY(pRExC_state, ret, reg_node(pRExC_state, TAIL));
9497 /* Check for proper termination. */
9499 /* restore original flags, but keep (?p) */
9500 RExC_flags = oregflags | (RExC_flags & RXf_PMf_KEEPCOPY);
9501 if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
9502 RExC_parse = oregcomp_parse;
9503 vFAIL("Unmatched (");
9506 else if (!paren && RExC_parse < RExC_end) {
9507 if (*RExC_parse == ')') {
9509 vFAIL("Unmatched )");
9512 FAIL("Junk on end of regexp"); /* "Can't happen". */
9513 assert(0); /* NOTREACHED */
9516 if (RExC_in_lookbehind) {
9517 RExC_in_lookbehind--;
9519 if (after_freeze > RExC_npar)
9520 RExC_npar = after_freeze;
9525 - regbranch - one alternative of an | operator
9527 * Implements the concatenation operator.
9529 * Returns NULL, setting *flagp to RESTART_UTF8 if the sizing scan needs to be
9533 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first, U32 depth)
9537 regnode *chain = NULL;
9539 I32 flags = 0, c = 0;
9540 GET_RE_DEBUG_FLAGS_DECL;
9542 PERL_ARGS_ASSERT_REGBRANCH;
9544 DEBUG_PARSE("brnc");
9549 if (!SIZE_ONLY && RExC_extralen)
9550 ret = reganode(pRExC_state, BRANCHJ,0);
9552 ret = reg_node(pRExC_state, BRANCH);
9553 Set_Node_Length(ret, 1);
9557 if (!first && SIZE_ONLY)
9558 RExC_extralen += 1; /* BRANCHJ */
9560 *flagp = WORST; /* Tentatively. */
9563 nextchar(pRExC_state);
9564 while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
9566 latest = regpiece(pRExC_state, &flags,depth+1);
9567 if (latest == NULL) {
9568 if (flags & TRYAGAIN)
9570 if (flags & RESTART_UTF8) {
9571 *flagp = RESTART_UTF8;
9574 FAIL2("panic: regpiece returned NULL, flags=%#"UVxf"", (UV) flags);
9576 else if (ret == NULL)
9578 *flagp |= flags&(HASWIDTH|POSTPONED);
9579 if (chain == NULL) /* First piece. */
9580 *flagp |= flags&SPSTART;
9583 REGTAIL(pRExC_state, chain, latest);
9588 if (chain == NULL) { /* Loop ran zero times. */
9589 chain = reg_node(pRExC_state, NOTHING);
9594 *flagp |= flags&SIMPLE;
9601 - regpiece - something followed by possible [*+?]
9603 * Note that the branching code sequences used for ? and the general cases
9604 * of * and + are somewhat optimized: they use the same NOTHING node as
9605 * both the endmarker for their branch list and the body of the last branch.
9606 * It might seem that this node could be dispensed with entirely, but the
9607 * endmarker role is not redundant.
9609 * Returns NULL, setting *flagp to TRYAGAIN if regatom() returns NULL with
9611 * Returns NULL, setting *flagp to RESTART_UTF8 if the sizing scan needs to be
9615 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
9622 const char * const origparse = RExC_parse;
9624 I32 max = REG_INFTY;
9625 #ifdef RE_TRACK_PATTERN_OFFSETS
9628 const char *maxpos = NULL;
9630 /* Save the original in case we change the emitted regop to a FAIL. */
9631 regnode * const orig_emit = RExC_emit;
9633 GET_RE_DEBUG_FLAGS_DECL;
9635 PERL_ARGS_ASSERT_REGPIECE;
9637 DEBUG_PARSE("piec");
9639 ret = regatom(pRExC_state, &flags,depth+1);
9641 if (flags & (TRYAGAIN|RESTART_UTF8))
9642 *flagp |= flags & (TRYAGAIN|RESTART_UTF8);
9644 FAIL2("panic: regatom returned NULL, flags=%#"UVxf"", (UV) flags);
9650 if (op == '{' && regcurly(RExC_parse, FALSE)) {
9652 #ifdef RE_TRACK_PATTERN_OFFSETS
9653 parse_start = RExC_parse; /* MJD */
9655 next = RExC_parse + 1;
9656 while (isDIGIT(*next) || *next == ',') {
9665 if (*next == '}') { /* got one */
9669 min = atoi(RExC_parse);
9673 maxpos = RExC_parse;
9675 if (!max && *maxpos != '0')
9676 max = REG_INFTY; /* meaning "infinity" */
9677 else if (max >= REG_INFTY)
9678 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
9680 nextchar(pRExC_state);
9681 if (max < min) { /* If can't match, warn and optimize to fail
9684 ckWARNreg(RExC_parse, "Quantifier {n,m} with n > m can't match");
9686 /* We can't back off the size because we have to reserve
9687 * enough space for all the things we are about to throw
9688 * away, but we can shrink it by the ammount we are about
9690 RExC_size = PREVOPER(RExC_size) - regarglen[(U8)OPFAIL];
9693 RExC_emit = orig_emit;
9695 ret = reg_node(pRExC_state, OPFAIL);
9700 if ((flags&SIMPLE)) {
9701 RExC_naughty += 2 + RExC_naughty / 2;
9702 reginsert(pRExC_state, CURLY, ret, depth+1);
9703 Set_Node_Offset(ret, parse_start+1); /* MJD */
9704 Set_Node_Cur_Length(ret, parse_start);
9707 regnode * const w = reg_node(pRExC_state, WHILEM);
9710 REGTAIL(pRExC_state, ret, w);
9711 if (!SIZE_ONLY && RExC_extralen) {
9712 reginsert(pRExC_state, LONGJMP,ret, depth+1);
9713 reginsert(pRExC_state, NOTHING,ret, depth+1);
9714 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
9716 reginsert(pRExC_state, CURLYX,ret, depth+1);
9718 Set_Node_Offset(ret, parse_start+1);
9719 Set_Node_Length(ret,
9720 op == '{' ? (RExC_parse - parse_start) : 1);
9722 if (!SIZE_ONLY && RExC_extralen)
9723 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
9724 REGTAIL(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
9726 RExC_whilem_seen++, RExC_extralen += 3;
9727 RExC_naughty += 4 + RExC_naughty; /* compound interest */
9736 ARG1_SET(ret, (U16)min);
9737 ARG2_SET(ret, (U16)max);
9749 #if 0 /* Now runtime fix should be reliable. */
9751 /* if this is reinstated, don't forget to put this back into perldiag:
9753 =item Regexp *+ operand could be empty at {#} in regex m/%s/
9755 (F) The part of the regexp subject to either the * or + quantifier
9756 could match an empty string. The {#} shows in the regular
9757 expression about where the problem was discovered.
9761 if (!(flags&HASWIDTH) && op != '?')
9762 vFAIL("Regexp *+ operand could be empty");
9765 #ifdef RE_TRACK_PATTERN_OFFSETS
9766 parse_start = RExC_parse;
9768 nextchar(pRExC_state);
9770 *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
9772 if (op == '*' && (flags&SIMPLE)) {
9773 reginsert(pRExC_state, STAR, ret, depth+1);
9777 else if (op == '*') {
9781 else if (op == '+' && (flags&SIMPLE)) {
9782 reginsert(pRExC_state, PLUS, ret, depth+1);
9786 else if (op == '+') {
9790 else if (op == '?') {
9795 if (!SIZE_ONLY && !(flags&(HASWIDTH|POSTPONED)) && max > REG_INFTY/3) {
9796 SAVEFREESV(RExC_rx_sv); /* in case of fatal warnings */
9797 ckWARN3reg(RExC_parse,
9798 "%.*s matches null string many times",
9799 (int)(RExC_parse >= origparse ? RExC_parse - origparse : 0),
9801 (void)ReREFCNT_inc(RExC_rx_sv);
9804 if (RExC_parse < RExC_end && *RExC_parse == '?') {
9805 nextchar(pRExC_state);
9806 reginsert(pRExC_state, MINMOD, ret, depth+1);
9807 REGTAIL(pRExC_state, ret, ret + NODE_STEP_REGNODE);
9810 if (RExC_parse < RExC_end && *RExC_parse == '+') {
9812 nextchar(pRExC_state);
9813 ender = reg_node(pRExC_state, SUCCEED);
9814 REGTAIL(pRExC_state, ret, ender);
9815 reginsert(pRExC_state, SUSPEND, ret, depth+1);
9817 ender = reg_node(pRExC_state, TAIL);
9818 REGTAIL(pRExC_state, ret, ender);
9821 if (RExC_parse < RExC_end && ISMULT2(RExC_parse)) {
9823 vFAIL("Nested quantifiers");
9830 S_grok_bslash_N(pTHX_ RExC_state_t *pRExC_state, regnode** node_p, UV *valuep, I32 *flagp, U32 depth, bool in_char_class,
9831 const bool strict /* Apply stricter parsing rules? */
9835 /* This is expected to be called by a parser routine that has recognized '\N'
9836 and needs to handle the rest. RExC_parse is expected to point at the first
9837 char following the N at the time of the call. On successful return,
9838 RExC_parse has been updated to point to just after the sequence identified
9839 by this routine, and <*flagp> has been updated.
9841 The \N may be inside (indicated by the boolean <in_char_class>) or outside a
9844 \N may begin either a named sequence, or if outside a character class, mean
9845 to match a non-newline. For non single-quoted regexes, the tokenizer has
9846 attempted to decide which, and in the case of a named sequence, converted it
9847 into one of the forms: \N{} (if the sequence is null), or \N{U+c1.c2...},
9848 where c1... are the characters in the sequence. For single-quoted regexes,
9849 the tokenizer passes the \N sequence through unchanged; this code will not
9850 attempt to determine this nor expand those, instead raising a syntax error.
9851 The net effect is that if the beginning of the passed-in pattern isn't '{U+'
9852 or there is no '}', it signals that this \N occurrence means to match a
9855 Only the \N{U+...} form should occur in a character class, for the same
9856 reason that '.' inside a character class means to just match a period: it
9857 just doesn't make sense.
9859 The function raises an error (via vFAIL), and doesn't return for various
9860 syntax errors. Otherwise it returns TRUE and sets <node_p> or <valuep> on
9861 success; it returns FALSE otherwise. Returns FALSE, setting *flagp to
9862 RESTART_UTF8 if the sizing scan needs to be restarted. Such a restart is
9863 only possible if node_p is non-NULL.
9866 If <valuep> is non-null, it means the caller can accept an input sequence
9867 consisting of a just a single code point; <*valuep> is set to that value
9868 if the input is such.
9870 If <node_p> is non-null it signifies that the caller can accept any other
9871 legal sequence (i.e., one that isn't just a single code point). <*node_p>
9873 1) \N means not-a-NL: points to a newly created REG_ANY node;
9874 2) \N{}: points to a new NOTHING node;
9875 3) otherwise: points to a new EXACT node containing the resolved
9877 Note that FALSE is returned for single code point sequences if <valuep> is
9881 char * endbrace; /* '}' following the name */
9883 char *endchar; /* Points to '.' or '}' ending cur char in the input
9885 bool has_multiple_chars; /* true if the input stream contains a sequence of
9886 more than one character */
9888 GET_RE_DEBUG_FLAGS_DECL;
9890 PERL_ARGS_ASSERT_GROK_BSLASH_N;
9894 assert(cBOOL(node_p) ^ cBOOL(valuep)); /* Exactly one should be set */
9896 /* The [^\n] meaning of \N ignores spaces and comments under the /x
9897 * modifier. The other meaning does not */
9898 p = (RExC_flags & RXf_PMf_EXTENDED)
9899 ? regwhite( pRExC_state, RExC_parse )
9902 /* Disambiguate between \N meaning a named character versus \N meaning
9903 * [^\n]. The former is assumed when it can't be the latter. */
9904 if (*p != '{' || regcurly(p, FALSE)) {
9907 /* no bare \N in a charclass */
9908 if (in_char_class) {
9909 vFAIL("\\N in a character class must be a named character: \\N{...}");
9913 nextchar(pRExC_state);
9914 *node_p = reg_node(pRExC_state, REG_ANY);
9915 *flagp |= HASWIDTH|SIMPLE;
9918 Set_Node_Length(*node_p, 1); /* MJD */
9922 /* Here, we have decided it should be a named character or sequence */
9924 /* The test above made sure that the next real character is a '{', but
9925 * under the /x modifier, it could be separated by space (or a comment and
9926 * \n) and this is not allowed (for consistency with \x{...} and the
9927 * tokenizer handling of \N{NAME}). */
9928 if (*RExC_parse != '{') {
9929 vFAIL("Missing braces on \\N{}");
9932 RExC_parse++; /* Skip past the '{' */
9934 if (! (endbrace = strchr(RExC_parse, '}')) /* no trailing brace */
9935 || ! (endbrace == RExC_parse /* nothing between the {} */
9936 || (endbrace - RExC_parse >= 2 /* U+ (bad hex is checked below */
9937 && strnEQ(RExC_parse, "U+", 2)))) /* for a better error msg) */
9939 if (endbrace) RExC_parse = endbrace; /* position msg's '<--HERE' */
9940 vFAIL("\\N{NAME} must be resolved by the lexer");
9943 if (endbrace == RExC_parse) { /* empty: \N{} */
9946 *node_p = reg_node(pRExC_state,NOTHING);
9948 else if (in_char_class) {
9949 if (SIZE_ONLY && in_char_class) {
9951 RExC_parse++; /* Position after the "}" */
9952 vFAIL("Zero length \\N{}");
9955 ckWARNreg(RExC_parse,
9956 "Ignoring zero length \\N{} in character class");
9964 nextchar(pRExC_state);
9968 RExC_uni_semantics = 1; /* Unicode named chars imply Unicode semantics */
9969 RExC_parse += 2; /* Skip past the 'U+' */
9971 endchar = RExC_parse + strcspn(RExC_parse, ".}");
9973 /* Code points are separated by dots. If none, there is only one code
9974 * point, and is terminated by the brace */
9975 has_multiple_chars = (endchar < endbrace);
9977 if (valuep && (! has_multiple_chars || in_char_class)) {
9978 /* We only pay attention to the first char of
9979 multichar strings being returned in char classes. I kinda wonder
9980 if this makes sense as it does change the behaviour
9981 from earlier versions, OTOH that behaviour was broken
9982 as well. XXX Solution is to recharacterize as
9983 [rest-of-class]|multi1|multi2... */
9985 STRLEN length_of_hex = (STRLEN)(endchar - RExC_parse);
9986 I32 grok_hex_flags = PERL_SCAN_ALLOW_UNDERSCORES
9987 | PERL_SCAN_DISALLOW_PREFIX
9988 | (SIZE_ONLY ? PERL_SCAN_SILENT_ILLDIGIT : 0);
9990 *valuep = grok_hex(RExC_parse, &length_of_hex, &grok_hex_flags, NULL);
9992 /* The tokenizer should have guaranteed validity, but it's possible to
9993 * bypass it by using single quoting, so check */
9994 if (length_of_hex == 0
9995 || length_of_hex != (STRLEN)(endchar - RExC_parse) )
9997 RExC_parse += length_of_hex; /* Includes all the valid */
9998 RExC_parse += (RExC_orig_utf8) /* point to after 1st invalid */
9999 ? UTF8SKIP(RExC_parse)
10001 /* Guard against malformed utf8 */
10002 if (RExC_parse >= endchar) {
10003 RExC_parse = endchar;
10005 vFAIL("Invalid hexadecimal number in \\N{U+...}");
10008 if (in_char_class && has_multiple_chars) {
10010 RExC_parse = endbrace;
10011 vFAIL("\\N{} in character class restricted to one character");
10014 ckWARNreg(endchar, "Using just the first character returned by \\N{} in character class");
10018 RExC_parse = endbrace + 1;
10020 else if (! node_p || ! has_multiple_chars) {
10022 /* Here, the input is legal, but not according to the caller's
10023 * options. We fail without advancing the parse, so that the
10024 * caller can try again */
10030 /* What is done here is to convert this to a sub-pattern of the form
10031 * (?:\x{char1}\x{char2}...)
10032 * and then call reg recursively. That way, it retains its atomicness,
10033 * while not having to worry about special handling that some code
10034 * points may have. toke.c has converted the original Unicode values
10035 * to native, so that we can just pass on the hex values unchanged. We
10036 * do have to set a flag to keep recoding from happening in the
10039 SV * substitute_parse = newSVpvn_flags("?:", 2, SVf_UTF8|SVs_TEMP);
10041 char *orig_end = RExC_end;
10044 while (RExC_parse < endbrace) {
10046 /* Convert to notation the rest of the code understands */
10047 sv_catpv(substitute_parse, "\\x{");
10048 sv_catpvn(substitute_parse, RExC_parse, endchar - RExC_parse);
10049 sv_catpv(substitute_parse, "}");
10051 /* Point to the beginning of the next character in the sequence. */
10052 RExC_parse = endchar + 1;
10053 endchar = RExC_parse + strcspn(RExC_parse, ".}");
10055 sv_catpv(substitute_parse, ")");
10057 RExC_parse = SvPV(substitute_parse, len);
10059 /* Don't allow empty number */
10061 vFAIL("Invalid hexadecimal number in \\N{U+...}");
10063 RExC_end = RExC_parse + len;
10065 /* The values are Unicode, and therefore not subject to recoding */
10066 RExC_override_recoding = 1;
10068 if (!(*node_p = reg(pRExC_state, 1, &flags, depth+1))) {
10069 if (flags & RESTART_UTF8) {
10070 *flagp = RESTART_UTF8;
10073 FAIL2("panic: reg returned NULL to grok_bslash_N, flags=%#"UVxf"",
10076 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
10078 RExC_parse = endbrace;
10079 RExC_end = orig_end;
10080 RExC_override_recoding = 0;
10082 nextchar(pRExC_state);
10092 * It returns the code point in utf8 for the value in *encp.
10093 * value: a code value in the source encoding
10094 * encp: a pointer to an Encode object
10096 * If the result from Encode is not a single character,
10097 * it returns U+FFFD (Replacement character) and sets *encp to NULL.
10100 S_reg_recode(pTHX_ const char value, SV **encp)
10103 SV * const sv = newSVpvn_flags(&value, numlen, SVs_TEMP);
10104 const char * const s = *encp ? sv_recode_to_utf8(sv, *encp) : SvPVX(sv);
10105 const STRLEN newlen = SvCUR(sv);
10106 UV uv = UNICODE_REPLACEMENT;
10108 PERL_ARGS_ASSERT_REG_RECODE;
10112 ? utf8n_to_uvchr((U8*)s, newlen, &numlen, UTF8_ALLOW_DEFAULT)
10115 if (!newlen || numlen != newlen) {
10116 uv = UNICODE_REPLACEMENT;
10122 PERL_STATIC_INLINE U8
10123 S_compute_EXACTish(pTHX_ RExC_state_t *pRExC_state)
10127 PERL_ARGS_ASSERT_COMPUTE_EXACTISH;
10133 op = get_regex_charset(RExC_flags);
10134 if (op >= REGEX_ASCII_RESTRICTED_CHARSET) {
10135 op--; /* /a is same as /u, and map /aa's offset to what /a's would have
10136 been, so there is no hole */
10139 return op + EXACTF;
10142 PERL_STATIC_INLINE void
10143 S_alloc_maybe_populate_EXACT(pTHX_ RExC_state_t *pRExC_state, regnode *node, I32* flagp, STRLEN len, UV code_point)
10145 /* This knows the details about sizing an EXACTish node, setting flags for
10146 * it (by setting <*flagp>, and potentially populating it with a single
10149 * If <len> (the length in bytes) is non-zero, this function assumes that
10150 * the node has already been populated, and just does the sizing. In this
10151 * case <code_point> should be the final code point that has already been
10152 * placed into the node. This value will be ignored except that under some
10153 * circumstances <*flagp> is set based on it.
10155 * If <len> is zero, the function assumes that the node is to contain only
10156 * the single character given by <code_point> and calculates what <len>
10157 * should be. In pass 1, it sizes the node appropriately. In pass 2, it
10158 * additionally will populate the node's STRING with <code_point>, if <len>
10159 * is 0. In both cases <*flagp> is appropriately set
10161 * It knows that under FOLD, the Latin Sharp S and UTF characters above
10162 * 255, must be folded (the former only when the rules indicate it can
10165 bool len_passed_in = cBOOL(len != 0);
10166 U8 character[UTF8_MAXBYTES_CASE+1];
10168 PERL_ARGS_ASSERT_ALLOC_MAYBE_POPULATE_EXACT;
10170 if (! len_passed_in) {
10172 if (FOLD && (! LOC || code_point > 255)) {
10173 _to_uni_fold_flags(NATIVE_TO_UNI(code_point),
10176 FOLD_FLAGS_FULL | ((LOC)
10177 ? FOLD_FLAGS_LOCALE
10178 : (ASCII_FOLD_RESTRICTED)
10179 ? FOLD_FLAGS_NOMIX_ASCII
10183 uvchr_to_utf8( character, code_point);
10184 len = UTF8SKIP(character);
10188 || code_point != LATIN_SMALL_LETTER_SHARP_S
10189 || ASCII_FOLD_RESTRICTED
10190 || ! AT_LEAST_UNI_SEMANTICS)
10192 *character = (U8) code_point;
10197 *(character + 1) = 's';
10203 RExC_size += STR_SZ(len);
10206 RExC_emit += STR_SZ(len);
10207 STR_LEN(node) = len;
10208 if (! len_passed_in) {
10209 Copy((char *) character, STRING(node), len, char);
10213 *flagp |= HASWIDTH;
10215 /* A single character node is SIMPLE, except for the special-cased SHARP S
10217 if ((len == 1 || (UTF && len == UNISKIP(code_point)))
10218 && (code_point != LATIN_SMALL_LETTER_SHARP_S
10219 || ! FOLD || ! DEPENDS_SEMANTICS))
10226 - regatom - the lowest level
10228 Try to identify anything special at the start of the pattern. If there
10229 is, then handle it as required. This may involve generating a single regop,
10230 such as for an assertion; or it may involve recursing, such as to
10231 handle a () structure.
10233 If the string doesn't start with something special then we gobble up
10234 as much literal text as we can.
10236 Once we have been able to handle whatever type of thing started the
10237 sequence, we return.
10239 Note: we have to be careful with escapes, as they can be both literal
10240 and special, and in the case of \10 and friends, context determines which.
10242 A summary of the code structure is:
10244 switch (first_byte) {
10245 cases for each special:
10246 handle this special;
10249 switch (2nd byte) {
10250 cases for each unambiguous special:
10251 handle this special;
10253 cases for each ambigous special/literal:
10255 if (special) handle here
10257 default: // unambiguously literal:
10260 default: // is a literal char
10263 create EXACTish node for literal;
10264 while (more input and node isn't full) {
10265 switch (input_byte) {
10266 cases for each special;
10267 make sure parse pointer is set so that the next call to
10268 regatom will see this special first
10269 goto loopdone; // EXACTish node terminated by prev. char
10271 append char to EXACTISH node;
10273 get next input byte;
10277 return the generated node;
10279 Specifically there are two separate switches for handling
10280 escape sequences, with the one for handling literal escapes requiring
10281 a dummy entry for all of the special escapes that are actually handled
10284 Returns NULL, setting *flagp to TRYAGAIN if reg() returns NULL with
10286 Returns NULL, setting *flagp to RESTART_UTF8 if the sizing scan needs to be
10288 Otherwise does not return NULL.
10292 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
10295 regnode *ret = NULL;
10297 char *parse_start = RExC_parse;
10301 GET_RE_DEBUG_FLAGS_DECL;
10303 *flagp = WORST; /* Tentatively. */
10305 DEBUG_PARSE("atom");
10307 PERL_ARGS_ASSERT_REGATOM;
10310 switch ((U8)*RExC_parse) {
10312 RExC_seen_zerolen++;
10313 nextchar(pRExC_state);
10314 if (RExC_flags & RXf_PMf_MULTILINE)
10315 ret = reg_node(pRExC_state, MBOL);
10316 else if (RExC_flags & RXf_PMf_SINGLELINE)
10317 ret = reg_node(pRExC_state, SBOL);
10319 ret = reg_node(pRExC_state, BOL);
10320 Set_Node_Length(ret, 1); /* MJD */
10323 nextchar(pRExC_state);
10325 RExC_seen_zerolen++;
10326 if (RExC_flags & RXf_PMf_MULTILINE)
10327 ret = reg_node(pRExC_state, MEOL);
10328 else if (RExC_flags & RXf_PMf_SINGLELINE)
10329 ret = reg_node(pRExC_state, SEOL);
10331 ret = reg_node(pRExC_state, EOL);
10332 Set_Node_Length(ret, 1); /* MJD */
10335 nextchar(pRExC_state);
10336 if (RExC_flags & RXf_PMf_SINGLELINE)
10337 ret = reg_node(pRExC_state, SANY);
10339 ret = reg_node(pRExC_state, REG_ANY);
10340 *flagp |= HASWIDTH|SIMPLE;
10342 Set_Node_Length(ret, 1); /* MJD */
10346 char * const oregcomp_parse = ++RExC_parse;
10347 ret = regclass(pRExC_state, flagp,depth+1,
10348 FALSE, /* means parse the whole char class */
10349 TRUE, /* allow multi-char folds */
10350 FALSE, /* don't silence non-portable warnings. */
10352 if (*RExC_parse != ']') {
10353 RExC_parse = oregcomp_parse;
10354 vFAIL("Unmatched [");
10357 if (*flagp & RESTART_UTF8)
10359 FAIL2("panic: regclass returned NULL to regatom, flags=%#"UVxf"",
10362 nextchar(pRExC_state);
10363 Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
10367 nextchar(pRExC_state);
10368 ret = reg(pRExC_state, 2, &flags,depth+1);
10370 if (flags & TRYAGAIN) {
10371 if (RExC_parse == RExC_end) {
10372 /* Make parent create an empty node if needed. */
10373 *flagp |= TRYAGAIN;
10378 if (flags & RESTART_UTF8) {
10379 *flagp = RESTART_UTF8;
10382 FAIL2("panic: reg returned NULL to regatom, flags=%#"UVxf"", (UV) flags);
10384 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
10388 if (flags & TRYAGAIN) {
10389 *flagp |= TRYAGAIN;
10392 vFAIL("Internal urp");
10393 /* Supposed to be caught earlier. */
10396 if (!regcurly(RExC_parse, FALSE)) {
10405 vFAIL("Quantifier follows nothing");
10410 This switch handles escape sequences that resolve to some kind
10411 of special regop and not to literal text. Escape sequnces that
10412 resolve to literal text are handled below in the switch marked
10415 Every entry in this switch *must* have a corresponding entry
10416 in the literal escape switch. However, the opposite is not
10417 required, as the default for this switch is to jump to the
10418 literal text handling code.
10420 switch ((U8)*++RExC_parse) {
10422 /* Special Escapes */
10424 RExC_seen_zerolen++;
10425 ret = reg_node(pRExC_state, SBOL);
10427 goto finish_meta_pat;
10429 ret = reg_node(pRExC_state, GPOS);
10430 RExC_seen |= REG_SEEN_GPOS;
10432 goto finish_meta_pat;
10434 RExC_seen_zerolen++;
10435 ret = reg_node(pRExC_state, KEEPS);
10437 /* XXX:dmq : disabling in-place substitution seems to
10438 * be necessary here to avoid cases of memory corruption, as
10439 * with: C<$_="x" x 80; s/x\K/y/> -- rgs
10441 RExC_seen |= REG_SEEN_LOOKBEHIND;
10442 goto finish_meta_pat;
10444 ret = reg_node(pRExC_state, SEOL);
10446 RExC_seen_zerolen++; /* Do not optimize RE away */
10447 goto finish_meta_pat;
10449 ret = reg_node(pRExC_state, EOS);
10451 RExC_seen_zerolen++; /* Do not optimize RE away */
10452 goto finish_meta_pat;
10454 ret = reg_node(pRExC_state, CANY);
10455 RExC_seen |= REG_SEEN_CANY;
10456 *flagp |= HASWIDTH|SIMPLE;
10457 goto finish_meta_pat;
10459 ret = reg_node(pRExC_state, CLUMP);
10460 *flagp |= HASWIDTH;
10461 goto finish_meta_pat;
10467 arg = ANYOF_WORDCHAR;
10471 RExC_seen_zerolen++;
10472 RExC_seen |= REG_SEEN_LOOKBEHIND;
10473 op = BOUND + get_regex_charset(RExC_flags);
10474 if (op > BOUNDA) { /* /aa is same as /a */
10477 ret = reg_node(pRExC_state, op);
10478 FLAGS(ret) = get_regex_charset(RExC_flags);
10480 if (! SIZE_ONLY && (U8) *(RExC_parse + 1) == '{') {
10481 ckWARNdep(RExC_parse, "\"\\b{\" is deprecated; use \"\\b\\{\" or \"\\b[{]\" instead");
10483 goto finish_meta_pat;
10485 RExC_seen_zerolen++;
10486 RExC_seen |= REG_SEEN_LOOKBEHIND;
10487 op = NBOUND + get_regex_charset(RExC_flags);
10488 if (op > NBOUNDA) { /* /aa is same as /a */
10491 ret = reg_node(pRExC_state, op);
10492 FLAGS(ret) = get_regex_charset(RExC_flags);
10494 if (! SIZE_ONLY && (U8) *(RExC_parse + 1) == '{') {
10495 ckWARNdep(RExC_parse, "\"\\B{\" is deprecated; use \"\\B\\{\" or \"\\B[{]\" instead");
10497 goto finish_meta_pat;
10507 ret = reg_node(pRExC_state, LNBREAK);
10508 *flagp |= HASWIDTH|SIMPLE;
10509 goto finish_meta_pat;
10517 goto join_posix_op_known;
10523 arg = ANYOF_VERTWS;
10525 goto join_posix_op_known;
10535 op = POSIXD + get_regex_charset(RExC_flags);
10536 if (op > POSIXA) { /* /aa is same as /a */
10540 join_posix_op_known:
10543 op += NPOSIXD - POSIXD;
10546 ret = reg_node(pRExC_state, op);
10548 FLAGS(ret) = namedclass_to_classnum(arg);
10551 *flagp |= HASWIDTH|SIMPLE;
10555 nextchar(pRExC_state);
10556 Set_Node_Length(ret, 2); /* MJD */
10562 char* parse_start = RExC_parse - 2;
10567 ret = regclass(pRExC_state, flagp,depth+1,
10568 TRUE, /* means just parse this element */
10569 FALSE, /* don't allow multi-char folds */
10570 FALSE, /* don't silence non-portable warnings.
10571 It would be a bug if these returned
10574 /* regclass() can only return RESTART_UTF8 if multi-char folds
10577 FAIL2("panic: regclass returned NULL to regatom, flags=%#"UVxf"",
10582 Set_Node_Offset(ret, parse_start + 2);
10583 Set_Node_Cur_Length(ret, parse_start);
10584 nextchar(pRExC_state);
10588 /* Handle \N and \N{NAME} with multiple code points here and not
10589 * below because it can be multicharacter. join_exact() will join
10590 * them up later on. Also this makes sure that things like
10591 * /\N{BLAH}+/ and \N{BLAH} being multi char Just Happen. dmq.
10592 * The options to the grok function call causes it to fail if the
10593 * sequence is just a single code point. We then go treat it as
10594 * just another character in the current EXACT node, and hence it
10595 * gets uniform treatment with all the other characters. The
10596 * special treatment for quantifiers is not needed for such single
10597 * character sequences */
10599 if (! grok_bslash_N(pRExC_state, &ret, NULL, flagp, depth, FALSE,
10600 FALSE /* not strict */ )) {
10601 if (*flagp & RESTART_UTF8)
10607 case 'k': /* Handle \k<NAME> and \k'NAME' */
10610 char ch= RExC_parse[1];
10611 if (ch != '<' && ch != '\'' && ch != '{') {
10613 vFAIL2("Sequence %.2s... not terminated",parse_start);
10615 /* this pretty much dupes the code for (?P=...) in reg(), if
10616 you change this make sure you change that */
10617 char* name_start = (RExC_parse += 2);
10619 SV *sv_dat = reg_scan_name(pRExC_state,
10620 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
10621 ch= (ch == '<') ? '>' : (ch == '{') ? '}' : '\'';
10622 if (RExC_parse == name_start || *RExC_parse != ch)
10623 vFAIL2("Sequence %.3s... not terminated",parse_start);
10626 num = add_data( pRExC_state, 1, "S" );
10627 RExC_rxi->data->data[num]=(void*)sv_dat;
10628 SvREFCNT_inc_simple_void(sv_dat);
10632 ret = reganode(pRExC_state,
10635 : (ASCII_FOLD_RESTRICTED)
10637 : (AT_LEAST_UNI_SEMANTICS)
10643 *flagp |= HASWIDTH;
10645 /* override incorrect value set in reganode MJD */
10646 Set_Node_Offset(ret, parse_start+1);
10647 Set_Node_Cur_Length(ret, parse_start);
10648 nextchar(pRExC_state);
10654 case '1': case '2': case '3': case '4':
10655 case '5': case '6': case '7': case '8': case '9':
10658 bool isg = *RExC_parse == 'g';
10663 if (*RExC_parse == '{') {
10667 if (*RExC_parse == '-') {
10671 if (hasbrace && !isDIGIT(*RExC_parse)) {
10672 if (isrel) RExC_parse--;
10674 goto parse_named_seq;
10676 num = atoi(RExC_parse);
10677 if (isg && num == 0) {
10678 if (*RExC_parse == '0') {
10679 vFAIL("Reference to invalid group 0");
10682 vFAIL("Unterminated \\g... pattern");
10686 num = RExC_npar - num;
10688 vFAIL("Reference to nonexistent or unclosed group");
10690 if (!isg && num > 9 && num >= RExC_npar && *RExC_parse != '8' && *RExC_parse != '9')
10691 /* Probably a character specified in octal, e.g. \35 */
10694 #ifdef RE_TRACK_PATTERN_OFFSETS
10695 char * const parse_start = RExC_parse - 1; /* MJD */
10697 while (isDIGIT(*RExC_parse))
10700 if (*RExC_parse != '}')
10701 vFAIL("Unterminated \\g{...} pattern");
10705 if (num > (I32)RExC_rx->nparens)
10706 vFAIL("Reference to nonexistent group");
10709 ret = reganode(pRExC_state,
10712 : (ASCII_FOLD_RESTRICTED)
10714 : (AT_LEAST_UNI_SEMANTICS)
10720 *flagp |= HASWIDTH;
10722 /* override incorrect value set in reganode MJD */
10723 Set_Node_Offset(ret, parse_start+1);
10724 Set_Node_Cur_Length(ret, parse_start);
10726 nextchar(pRExC_state);
10731 if (RExC_parse >= RExC_end)
10732 FAIL("Trailing \\");
10735 /* Do not generate "unrecognized" warnings here, we fall
10736 back into the quick-grab loop below */
10743 if (RExC_flags & RXf_PMf_EXTENDED) {
10744 if ( reg_skipcomment( pRExC_state ) )
10751 parse_start = RExC_parse - 1;
10760 #define MAX_NODE_STRING_SIZE 127
10761 char foldbuf[MAX_NODE_STRING_SIZE+UTF8_MAXBYTES_CASE];
10763 U8 upper_parse = MAX_NODE_STRING_SIZE;
10765 U8 node_type = compute_EXACTish(pRExC_state);
10766 bool next_is_quantifier;
10767 char * oldp = NULL;
10769 /* We can convert EXACTF nodes to EXACTFU if they contain only
10770 * characters that match identically regardless of the target
10771 * string's UTF8ness. The reason to do this is that EXACTF is not
10772 * trie-able, EXACTFU is. (We don't need to figure this out until
10774 bool maybe_exactfu = node_type == EXACTF && PASS2;
10776 /* If a folding node contains only code points that don't
10777 * participate in folds, it can be changed into an EXACT node,
10778 * which allows the optimizer more things to look for */
10781 ret = reg_node(pRExC_state, node_type);
10783 /* In pass1, folded, we use a temporary buffer instead of the
10784 * actual node, as the node doesn't exist yet */
10785 s = (SIZE_ONLY && FOLD) ? foldbuf : STRING(ret);
10791 /* We do the EXACTFish to EXACT node only if folding, and not if in
10792 * locale, as whether a character folds or not isn't known until
10793 * runtime. (And we don't need to figure this out until pass 2) */
10794 maybe_exact = FOLD && ! LOC && PASS2;
10796 /* XXX The node can hold up to 255 bytes, yet this only goes to
10797 * 127. I (khw) do not know why. Keeping it somewhat less than
10798 * 255 allows us to not have to worry about overflow due to
10799 * converting to utf8 and fold expansion, but that value is
10800 * 255-UTF8_MAXBYTES_CASE. join_exact() may join adjacent nodes
10801 * split up by this limit into a single one using the real max of
10802 * 255. Even at 127, this breaks under rare circumstances. If
10803 * folding, we do not want to split a node at a character that is a
10804 * non-final in a multi-char fold, as an input string could just
10805 * happen to want to match across the node boundary. The join
10806 * would solve that problem if the join actually happens. But a
10807 * series of more than two nodes in a row each of 127 would cause
10808 * the first join to succeed to get to 254, but then there wouldn't
10809 * be room for the next one, which could at be one of those split
10810 * multi-char folds. I don't know of any fool-proof solution. One
10811 * could back off to end with only a code point that isn't such a
10812 * non-final, but it is possible for there not to be any in the
10814 for (p = RExC_parse - 1;
10815 len < upper_parse && p < RExC_end;
10820 if (RExC_flags & RXf_PMf_EXTENDED)
10821 p = regwhite( pRExC_state, p );
10832 /* Literal Escapes Switch
10834 This switch is meant to handle escape sequences that
10835 resolve to a literal character.
10837 Every escape sequence that represents something
10838 else, like an assertion or a char class, is handled
10839 in the switch marked 'Special Escapes' above in this
10840 routine, but also has an entry here as anything that
10841 isn't explicitly mentioned here will be treated as
10842 an unescaped equivalent literal.
10845 switch ((U8)*++p) {
10846 /* These are all the special escapes. */
10847 case 'A': /* Start assertion */
10848 case 'b': case 'B': /* Word-boundary assertion*/
10849 case 'C': /* Single char !DANGEROUS! */
10850 case 'd': case 'D': /* digit class */
10851 case 'g': case 'G': /* generic-backref, pos assertion */
10852 case 'h': case 'H': /* HORIZWS */
10853 case 'k': case 'K': /* named backref, keep marker */
10854 case 'p': case 'P': /* Unicode property */
10855 case 'R': /* LNBREAK */
10856 case 's': case 'S': /* space class */
10857 case 'v': case 'V': /* VERTWS */
10858 case 'w': case 'W': /* word class */
10859 case 'X': /* eXtended Unicode "combining character sequence" */
10860 case 'z': case 'Z': /* End of line/string assertion */
10864 /* Anything after here is an escape that resolves to a
10865 literal. (Except digits, which may or may not)
10871 case 'N': /* Handle a single-code point named character. */
10872 /* The options cause it to fail if a multiple code
10873 * point sequence. Handle those in the switch() above
10875 RExC_parse = p + 1;
10876 if (! grok_bslash_N(pRExC_state, NULL, &ender,
10877 flagp, depth, FALSE,
10878 FALSE /* not strict */ ))
10880 if (*flagp & RESTART_UTF8)
10881 FAIL("panic: grok_bslash_N set RESTART_UTF8");
10882 RExC_parse = p = oldp;
10886 if (ender > 0xff) {
10903 ender = ASCII_TO_NATIVE('\033');
10907 ender = ASCII_TO_NATIVE('\007');
10913 const char* error_msg;
10915 bool valid = grok_bslash_o(&p,
10918 TRUE, /* out warnings */
10919 FALSE, /* not strict */
10920 TRUE, /* Output warnings
10925 RExC_parse = p; /* going to die anyway; point
10926 to exact spot of failure */
10930 if (PL_encoding && ender < 0x100) {
10931 goto recode_encoding;
10933 if (ender > 0xff) {
10940 UV result = UV_MAX; /* initialize to erroneous
10942 const char* error_msg;
10944 bool valid = grok_bslash_x(&p,
10947 TRUE, /* out warnings */
10948 FALSE, /* not strict */
10949 TRUE, /* Output warnings
10954 RExC_parse = p; /* going to die anyway; point
10955 to exact spot of failure */
10960 if (PL_encoding && ender < 0x100) {
10961 goto recode_encoding;
10963 if (ender > 0xff) {
10970 ender = grok_bslash_c(*p++, UTF, SIZE_ONLY);
10972 case '8': case '9': /* must be a backreference */
10975 case '1': case '2': case '3':case '4':
10976 case '5': case '6': case '7':
10977 /* When we parse backslash escapes there is ambiguity between
10978 * backreferences and octal escapes. Any escape from \1 - \9 is
10979 * a backreference, any multi-digit escape which does not start with
10980 * 0 and which when evaluated as decimal could refer to an already
10981 * parsed capture buffer is a backslash. Anything else is octal.
10983 * Note this implies that \118 could be interpreted as 118 OR as
10984 * "\11" . "8" depending on whether there were 118 capture buffers
10985 * defined already in the pattern.
10987 if ( !isDIGIT(p[1]) || atoi(p) <= RExC_npar )
10988 { /* Not to be treated as an octal constant, go
10995 I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
10997 ender = grok_oct(p, &numlen, &flags, NULL);
10998 if (ender > 0xff) {
11002 if (SIZE_ONLY /* like \08, \178 */
11005 && isDIGIT(*p) && ckWARN(WARN_REGEXP))
11007 reg_warn_non_literal_string(
11009 form_short_octal_warning(p, numlen));
11012 if (PL_encoding && ender < 0x100)
11013 goto recode_encoding;
11016 if (! RExC_override_recoding) {
11017 SV* enc = PL_encoding;
11018 ender = reg_recode((const char)(U8)ender, &enc);
11019 if (!enc && SIZE_ONLY)
11020 ckWARNreg(p, "Invalid escape in the specified encoding");
11026 FAIL("Trailing \\");
11029 if (!SIZE_ONLY&& isALPHANUMERIC(*p)) {
11030 /* Include any { following the alpha to emphasize
11031 * that it could be part of an escape at some point
11033 int len = (isALPHA(*p) && *(p + 1) == '{') ? 2 : 1;
11034 ckWARN3reg(p + len, "Unrecognized escape \\%.*s passed through", len, p);
11036 goto normal_default;
11037 } /* End of switch on '\' */
11039 default: /* A literal character */
11042 && RExC_flags & RXf_PMf_EXTENDED
11043 && ckWARN_d(WARN_DEPRECATED)
11044 && is_PATWS_non_low(p, UTF))
11046 vWARN_dep(p + ((UTF) ? UTF8SKIP(p) : 1),
11047 "Escape literal pattern white space under /x");
11051 if (UTF8_IS_START(*p) && UTF) {
11053 ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
11054 &numlen, UTF8_ALLOW_DEFAULT);
11060 } /* End of switch on the literal */
11062 /* Here, have looked at the literal character and <ender>
11063 * contains its ordinal, <p> points to the character after it
11066 if ( RExC_flags & RXf_PMf_EXTENDED)
11067 p = regwhite( pRExC_state, p );
11069 /* If the next thing is a quantifier, it applies to this
11070 * character only, which means that this character has to be in
11071 * its own node and can't just be appended to the string in an
11072 * existing node, so if there are already other characters in
11073 * the node, close the node with just them, and set up to do
11074 * this character again next time through, when it will be the
11075 * only thing in its new node */
11076 if ((next_is_quantifier = (p < RExC_end && ISMULT2(p))) && len)
11084 const STRLEN unilen = reguni(pRExC_state, ender, s);
11090 /* The loop increments <len> each time, as all but this
11091 * path (and one other) through it add a single byte to
11092 * the EXACTish node. But this one has changed len to
11093 * be the correct final value, so subtract one to
11094 * cancel out the increment that follows */
11098 REGC((char)ender, s++);
11103 /* See comments for join_exact() as to why we fold this
11104 * non-UTF at compile time */
11105 || (node_type == EXACTFU
11106 && ender == LATIN_SMALL_LETTER_SHARP_S)))
11108 if (IS_IN_SOME_FOLD_L1(ender)) {
11109 maybe_exact = FALSE;
11111 /* See if the character's fold differs between /d and
11112 * /u. This includes the multi-char fold SHARP S to
11115 && (PL_fold[ender] != PL_fold_latin1[ender]
11116 || ender == LATIN_SMALL_LETTER_SHARP_S
11118 && isARG2_lower_or_UPPER_ARG1('s', ender)
11119 && isARG2_lower_or_UPPER_ARG1('s', *(s-1)))))
11121 maybe_exactfu = FALSE;
11124 *(s++) = (char) ender;
11128 /* Prime the casefolded buffer. Locale rules, which apply
11129 * only to code points < 256, aren't known until execution,
11130 * so for them, just output the original character using
11131 * utf8. If we start to fold non-UTF patterns, be sure to
11132 * update join_exact() */
11133 if (LOC && ender < 256) {
11134 if (UNI_IS_INVARIANT(ender)) {
11138 *s = UTF8_TWO_BYTE_HI(ender);
11139 *(s + 1) = UTF8_TWO_BYTE_LO(ender);
11144 UV folded = _to_uni_fold_flags(
11149 | ((LOC) ? FOLD_FLAGS_LOCALE
11150 : (ASCII_FOLD_RESTRICTED)
11151 ? FOLD_FLAGS_NOMIX_ASCII
11155 /* If this node only contains non-folding code points
11156 * so far, see if this new one is also non-folding */
11158 if (folded != ender) {
11159 maybe_exact = FALSE;
11162 /* Here the fold is the original; we have
11163 * to check further to see if anything
11165 if (! PL_utf8_foldable) {
11166 SV* swash = swash_init("utf8",
11168 &PL_sv_undef, 1, 0);
11170 _get_swash_invlist(swash);
11171 SvREFCNT_dec_NN(swash);
11173 if (_invlist_contains_cp(PL_utf8_foldable,
11176 maybe_exact = FALSE;
11184 /* The loop increments <len> each time, as all but this
11185 * path (and one other) through it add a single byte to the
11186 * EXACTish node. But this one has changed len to be the
11187 * correct final value, so subtract one to cancel out the
11188 * increment that follows */
11189 len += foldlen - 1;
11192 if (next_is_quantifier) {
11194 /* Here, the next input is a quantifier, and to get here,
11195 * the current character is the only one in the node.
11196 * Also, here <len> doesn't include the final byte for this
11202 } /* End of loop through literal characters */
11204 /* Here we have either exhausted the input or ran out of room in
11205 * the node. (If we encountered a character that can't be in the
11206 * node, transfer is made directly to <loopdone>, and so we
11207 * wouldn't have fallen off the end of the loop.) In the latter
11208 * case, we artificially have to split the node into two, because
11209 * we just don't have enough space to hold everything. This
11210 * creates a problem if the final character participates in a
11211 * multi-character fold in the non-final position, as a match that
11212 * should have occurred won't, due to the way nodes are matched,
11213 * and our artificial boundary. So back off until we find a non-
11214 * problematic character -- one that isn't at the beginning or
11215 * middle of such a fold. (Either it doesn't participate in any
11216 * folds, or appears only in the final position of all the folds it
11217 * does participate in.) A better solution with far fewer false
11218 * positives, and that would fill the nodes more completely, would
11219 * be to actually have available all the multi-character folds to
11220 * test against, and to back-off only far enough to be sure that
11221 * this node isn't ending with a partial one. <upper_parse> is set
11222 * further below (if we need to reparse the node) to include just
11223 * up through that final non-problematic character that this code
11224 * identifies, so when it is set to less than the full node, we can
11225 * skip the rest of this */
11226 if (FOLD && p < RExC_end && upper_parse == MAX_NODE_STRING_SIZE) {
11228 const STRLEN full_len = len;
11230 assert(len >= MAX_NODE_STRING_SIZE);
11232 /* Here, <s> points to the final byte of the final character.
11233 * Look backwards through the string until find a non-
11234 * problematic character */
11238 /* These two have no multi-char folds to non-UTF characters
11240 if (ASCII_FOLD_RESTRICTED || LOC) {
11244 while (--s >= s0 && IS_NON_FINAL_FOLD(*s)) { }
11248 if (! PL_NonL1NonFinalFold) {
11249 PL_NonL1NonFinalFold = _new_invlist_C_array(
11250 NonL1_Perl_Non_Final_Folds_invlist);
11253 /* Point to the first byte of the final character */
11254 s = (char *) utf8_hop((U8 *) s, -1);
11256 while (s >= s0) { /* Search backwards until find
11257 non-problematic char */
11258 if (UTF8_IS_INVARIANT(*s)) {
11260 /* There are no ascii characters that participate
11261 * in multi-char folds under /aa. In EBCDIC, the
11262 * non-ascii invariants are all control characters,
11263 * so don't ever participate in any folds. */
11264 if (ASCII_FOLD_RESTRICTED
11265 || ! IS_NON_FINAL_FOLD(*s))
11270 else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
11272 /* No Latin1 characters participate in multi-char
11273 * folds under /l */
11275 || ! IS_NON_FINAL_FOLD(TWO_BYTE_UTF8_TO_UNI(
11281 else if (! _invlist_contains_cp(
11282 PL_NonL1NonFinalFold,
11283 valid_utf8_to_uvchr((U8 *) s, NULL)))
11288 /* Here, the current character is problematic in that
11289 * it does occur in the non-final position of some
11290 * fold, so try the character before it, but have to
11291 * special case the very first byte in the string, so
11292 * we don't read outside the string */
11293 s = (s == s0) ? s -1 : (char *) utf8_hop((U8 *) s, -1);
11294 } /* End of loop backwards through the string */
11296 /* If there were only problematic characters in the string,
11297 * <s> will point to before s0, in which case the length
11298 * should be 0, otherwise include the length of the
11299 * non-problematic character just found */
11300 len = (s < s0) ? 0 : s - s0 + UTF8SKIP(s);
11303 /* Here, have found the final character, if any, that is
11304 * non-problematic as far as ending the node without splitting
11305 * it across a potential multi-char fold. <len> contains the
11306 * number of bytes in the node up-to and including that
11307 * character, or is 0 if there is no such character, meaning
11308 * the whole node contains only problematic characters. In
11309 * this case, give up and just take the node as-is. We can't
11314 /* If the node ends in an 's' we make sure it stays EXACTF,
11315 * as if it turns into an EXACTFU, it could later get
11316 * joined with another 's' that would then wrongly match
11318 if (maybe_exactfu && isARG2_lower_or_UPPER_ARG1('s', ender))
11320 maybe_exactfu = FALSE;
11324 /* Here, the node does contain some characters that aren't
11325 * problematic. If one such is the final character in the
11326 * node, we are done */
11327 if (len == full_len) {
11330 else if (len + ((UTF) ? UTF8SKIP(s) : 1) == full_len) {
11332 /* If the final character is problematic, but the
11333 * penultimate is not, back-off that last character to
11334 * later start a new node with it */
11339 /* Here, the final non-problematic character is earlier
11340 * in the input than the penultimate character. What we do
11341 * is reparse from the beginning, going up only as far as
11342 * this final ok one, thus guaranteeing that the node ends
11343 * in an acceptable character. The reason we reparse is
11344 * that we know how far in the character is, but we don't
11345 * know how to correlate its position with the input parse.
11346 * An alternate implementation would be to build that
11347 * correlation as we go along during the original parse,
11348 * but that would entail extra work for every node, whereas
11349 * this code gets executed only when the string is too
11350 * large for the node, and the final two characters are
11351 * problematic, an infrequent occurrence. Yet another
11352 * possible strategy would be to save the tail of the
11353 * string, and the next time regatom is called, initialize
11354 * with that. The problem with this is that unless you
11355 * back off one more character, you won't be guaranteed
11356 * regatom will get called again, unless regbranch,
11357 * regpiece ... are also changed. If you do back off that
11358 * extra character, so that there is input guaranteed to
11359 * force calling regatom, you can't handle the case where
11360 * just the first character in the node is acceptable. I
11361 * (khw) decided to try this method which doesn't have that
11362 * pitfall; if performance issues are found, we can do a
11363 * combination of the current approach plus that one */
11369 } /* End of verifying node ends with an appropriate char */
11371 loopdone: /* Jumped to when encounters something that shouldn't be in
11374 /* I (khw) don't know if you can get here with zero length, but the
11375 * old code handled this situation by creating a zero-length EXACT
11376 * node. Might as well be NOTHING instead */
11382 /* If 'maybe_exact' is still set here, means there are no
11383 * code points in the node that participate in folds;
11384 * similarly for 'maybe_exactfu' and code points that match
11385 * differently depending on UTF8ness of the target string
11390 else if (maybe_exactfu) {
11394 alloc_maybe_populate_EXACT(pRExC_state, ret, flagp, len, ender);
11397 RExC_parse = p - 1;
11398 Set_Node_Cur_Length(ret, parse_start);
11399 nextchar(pRExC_state);
11401 /* len is STRLEN which is unsigned, need to copy to signed */
11404 vFAIL("Internal disaster");
11407 } /* End of label 'defchar:' */
11409 } /* End of giant switch on input character */
11415 S_regwhite( RExC_state_t *pRExC_state, char *p )
11417 const char *e = RExC_end;
11419 PERL_ARGS_ASSERT_REGWHITE;
11424 else if (*p == '#') {
11427 if (*p++ == '\n') {
11433 RExC_seen |= REG_SEEN_RUN_ON_COMMENT;
11442 S_regpatws( RExC_state_t *pRExC_state, char *p , const bool recognize_comment )
11444 /* Returns the next non-pattern-white space, non-comment character (the
11445 * latter only if 'recognize_comment is true) in the string p, which is
11446 * ended by RExC_end. If there is no line break ending a comment,
11447 * RExC_seen has added the REG_SEEN_RUN_ON_COMMENT flag; */
11448 const char *e = RExC_end;
11450 PERL_ARGS_ASSERT_REGPATWS;
11454 if ((len = is_PATWS_safe(p, e, UTF))) {
11457 else if (recognize_comment && *p == '#') {
11461 if (is_LNBREAK_safe(p, e, UTF)) {
11467 RExC_seen |= REG_SEEN_RUN_ON_COMMENT;
11475 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
11476 Character classes ([:foo:]) can also be negated ([:^foo:]).
11477 Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
11478 Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
11479 but trigger failures because they are currently unimplemented. */
11481 #define POSIXCC_DONE(c) ((c) == ':')
11482 #define POSIXCC_NOTYET(c) ((c) == '=' || (c) == '.')
11483 #define POSIXCC(c) (POSIXCC_DONE(c) || POSIXCC_NOTYET(c))
11485 PERL_STATIC_INLINE I32
11486 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value, const bool strict)
11489 I32 namedclass = OOB_NAMEDCLASS;
11491 PERL_ARGS_ASSERT_REGPPOSIXCC;
11493 if (value == '[' && RExC_parse + 1 < RExC_end &&
11494 /* I smell either [: or [= or [. -- POSIX has been here, right? */
11495 POSIXCC(UCHARAT(RExC_parse)))
11497 const char c = UCHARAT(RExC_parse);
11498 char* const s = RExC_parse++;
11500 while (RExC_parse < RExC_end && UCHARAT(RExC_parse) != c)
11502 if (RExC_parse == RExC_end) {
11505 /* Try to give a better location for the error (than the end of
11506 * the string) by looking for the matching ']' */
11508 while (RExC_parse < RExC_end && UCHARAT(RExC_parse) != ']') {
11511 vFAIL2("Unmatched '%c' in POSIX class", c);
11513 /* Grandfather lone [:, [=, [. */
11517 const char* const t = RExC_parse++; /* skip over the c */
11520 if (UCHARAT(RExC_parse) == ']') {
11521 const char *posixcc = s + 1;
11522 RExC_parse++; /* skip over the ending ] */
11525 const I32 complement = *posixcc == '^' ? *posixcc++ : 0;
11526 const I32 skip = t - posixcc;
11528 /* Initially switch on the length of the name. */
11531 if (memEQ(posixcc, "word", 4)) /* this is not POSIX,
11532 this is the Perl \w
11534 namedclass = ANYOF_WORDCHAR;
11537 /* Names all of length 5. */
11538 /* alnum alpha ascii blank cntrl digit graph lower
11539 print punct space upper */
11540 /* Offset 4 gives the best switch position. */
11541 switch (posixcc[4]) {
11543 if (memEQ(posixcc, "alph", 4)) /* alpha */
11544 namedclass = ANYOF_ALPHA;
11547 if (memEQ(posixcc, "spac", 4)) /* space */
11548 namedclass = ANYOF_PSXSPC;
11551 if (memEQ(posixcc, "grap", 4)) /* graph */
11552 namedclass = ANYOF_GRAPH;
11555 if (memEQ(posixcc, "asci", 4)) /* ascii */
11556 namedclass = ANYOF_ASCII;
11559 if (memEQ(posixcc, "blan", 4)) /* blank */
11560 namedclass = ANYOF_BLANK;
11563 if (memEQ(posixcc, "cntr", 4)) /* cntrl */
11564 namedclass = ANYOF_CNTRL;
11567 if (memEQ(posixcc, "alnu", 4)) /* alnum */
11568 namedclass = ANYOF_ALPHANUMERIC;
11571 if (memEQ(posixcc, "lowe", 4)) /* lower */
11572 namedclass = (FOLD) ? ANYOF_CASED : ANYOF_LOWER;
11573 else if (memEQ(posixcc, "uppe", 4)) /* upper */
11574 namedclass = (FOLD) ? ANYOF_CASED : ANYOF_UPPER;
11577 if (memEQ(posixcc, "digi", 4)) /* digit */
11578 namedclass = ANYOF_DIGIT;
11579 else if (memEQ(posixcc, "prin", 4)) /* print */
11580 namedclass = ANYOF_PRINT;
11581 else if (memEQ(posixcc, "punc", 4)) /* punct */
11582 namedclass = ANYOF_PUNCT;
11587 if (memEQ(posixcc, "xdigit", 6))
11588 namedclass = ANYOF_XDIGIT;
11592 if (namedclass == OOB_NAMEDCLASS)
11593 Simple_vFAIL3("POSIX class [:%.*s:] unknown",
11596 /* The #defines are structured so each complement is +1 to
11597 * the normal one */
11601 assert (posixcc[skip] == ':');
11602 assert (posixcc[skip+1] == ']');
11603 } else if (!SIZE_ONLY) {
11604 /* [[=foo=]] and [[.foo.]] are still future. */
11606 /* adjust RExC_parse so the warning shows after
11607 the class closes */
11608 while (UCHARAT(RExC_parse) && UCHARAT(RExC_parse) != ']')
11610 vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
11613 /* Maternal grandfather:
11614 * "[:" ending in ":" but not in ":]" */
11616 vFAIL("Unmatched '[' in POSIX class");
11619 /* Grandfather lone [:, [=, [. */
11629 S_could_it_be_a_POSIX_class(pTHX_ RExC_state_t *pRExC_state)
11631 /* This applies some heuristics at the current parse position (which should
11632 * be at a '[') to see if what follows might be intended to be a [:posix:]
11633 * class. It returns true if it really is a posix class, of course, but it
11634 * also can return true if it thinks that what was intended was a posix
11635 * class that didn't quite make it.
11637 * It will return true for
11639 * [:alphanumerics] (as long as the ] isn't followed immediately by a
11640 * ')' indicating the end of the (?[
11641 * [:any garbage including %^&$ punctuation:]
11643 * This is designed to be called only from S_handle_regex_sets; it could be
11644 * easily adapted to be called from the spot at the beginning of regclass()
11645 * that checks to see in a normal bracketed class if the surrounding []
11646 * have been omitted ([:word:] instead of [[:word:]]). But doing so would
11647 * change long-standing behavior, so I (khw) didn't do that */
11648 char* p = RExC_parse + 1;
11649 char first_char = *p;
11651 PERL_ARGS_ASSERT_COULD_IT_BE_A_POSIX_CLASS;
11653 assert(*(p - 1) == '[');
11655 if (! POSIXCC(first_char)) {
11660 while (p < RExC_end && isWORDCHAR(*p)) p++;
11662 if (p >= RExC_end) {
11666 if (p - RExC_parse > 2 /* Got at least 1 word character */
11667 && (*p == first_char
11668 || (*p == ']' && p + 1 < RExC_end && *(p + 1) != ')')))
11673 p = (char *) memchr(RExC_parse, ']', RExC_end - RExC_parse);
11676 && p - RExC_parse > 2 /* [:] evaluates to colon;
11677 [::] is a bad posix class. */
11678 && first_char == *(p - 1));
11682 S_handle_regex_sets(pTHX_ RExC_state_t *pRExC_state, SV** return_invlist, I32 *flagp, U32 depth,
11683 char * const oregcomp_parse)
11685 /* Handle the (?[...]) construct to do set operations */
11688 UV start, end; /* End points of code point ranges */
11690 char *save_end, *save_parse;
11695 const bool save_fold = FOLD;
11697 GET_RE_DEBUG_FLAGS_DECL;
11699 PERL_ARGS_ASSERT_HANDLE_REGEX_SETS;
11702 vFAIL("(?[...]) not valid in locale");
11704 RExC_uni_semantics = 1;
11706 /* This will return only an ANYOF regnode, or (unlikely) something smaller
11707 * (such as EXACT). Thus we can skip most everything if just sizing. We
11708 * call regclass to handle '[]' so as to not have to reinvent its parsing
11709 * rules here (throwing away the size it computes each time). And, we exit
11710 * upon an unescaped ']' that isn't one ending a regclass. To do both
11711 * these things, we need to realize that something preceded by a backslash
11712 * is escaped, so we have to keep track of backslashes */
11714 UV depth = 0; /* how many nested (?[...]) constructs */
11716 Perl_ck_warner_d(aTHX_
11717 packWARN(WARN_EXPERIMENTAL__REGEX_SETS),
11718 "The regex_sets feature is experimental" REPORT_LOCATION,
11719 (int) (RExC_parse - RExC_precomp) , RExC_precomp, RExC_parse);
11721 while (RExC_parse < RExC_end) {
11722 SV* current = NULL;
11723 RExC_parse = regpatws(pRExC_state, RExC_parse,
11724 TRUE); /* means recognize comments */
11725 switch (*RExC_parse) {
11727 if (RExC_parse[1] == '[') depth++, RExC_parse++;
11732 /* Skip the next byte (which could cause us to end up in
11733 * the middle of a UTF-8 character, but since none of those
11734 * are confusable with anything we currently handle in this
11735 * switch (invariants all), it's safe. We'll just hit the
11736 * default: case next time and keep on incrementing until
11737 * we find one of the invariants we do handle. */
11742 /* If this looks like it is a [:posix:] class, leave the
11743 * parse pointer at the '[' to fool regclass() into
11744 * thinking it is part of a '[[:posix:]]'. That function
11745 * will use strict checking to force a syntax error if it
11746 * doesn't work out to a legitimate class */
11747 bool is_posix_class
11748 = could_it_be_a_POSIX_class(pRExC_state);
11749 if (! is_posix_class) {
11753 /* regclass() can only return RESTART_UTF8 if multi-char
11754 folds are allowed. */
11755 if (!regclass(pRExC_state, flagp,depth+1,
11756 is_posix_class, /* parse the whole char
11757 class only if not a
11759 FALSE, /* don't allow multi-char folds */
11760 TRUE, /* silence non-portable warnings. */
11762 FAIL2("panic: regclass returned NULL to handle_sets, flags=%#"UVxf"",
11765 /* function call leaves parse pointing to the ']', except
11766 * if we faked it */
11767 if (is_posix_class) {
11771 SvREFCNT_dec(current); /* In case it returned something */
11776 if (depth--) break;
11778 if (RExC_parse < RExC_end
11779 && *RExC_parse == ')')
11781 node = reganode(pRExC_state, ANYOF, 0);
11782 RExC_size += ANYOF_SKIP;
11783 nextchar(pRExC_state);
11784 Set_Node_Length(node,
11785 RExC_parse - oregcomp_parse + 1); /* MJD */
11794 FAIL("Syntax error in (?[...])");
11797 /* Pass 2 only after this. Everything in this construct is a
11798 * metacharacter. Operands begin with either a '\' (for an escape
11799 * sequence), or a '[' for a bracketed character class. Any other
11800 * character should be an operator, or parenthesis for grouping. Both
11801 * types of operands are handled by calling regclass() to parse them. It
11802 * is called with a parameter to indicate to return the computed inversion
11803 * list. The parsing here is implemented via a stack. Each entry on the
11804 * stack is a single character representing one of the operators, or the
11805 * '('; or else a pointer to an operand inversion list. */
11807 #define IS_OPERAND(a) (! SvIOK(a))
11809 /* The stack starts empty. It is a syntax error if the first thing parsed
11810 * is a binary operator; everything else is pushed on the stack. When an
11811 * operand is parsed, the top of the stack is examined. If it is a binary
11812 * operator, the item before it should be an operand, and both are replaced
11813 * by the result of doing that operation on the new operand and the one on
11814 * the stack. Thus a sequence of binary operands is reduced to a single
11815 * one before the next one is parsed.
11817 * A unary operator may immediately follow a binary in the input, for
11820 * When an operand is parsed and the top of the stack is a unary operator,
11821 * the operation is performed, and then the stack is rechecked to see if
11822 * this new operand is part of a binary operation; if so, it is handled as
11825 * A '(' is simply pushed on the stack; it is valid only if the stack is
11826 * empty, or the top element of the stack is an operator or another '('
11827 * (for which the parenthesized expression will become an operand). By the
11828 * time the corresponding ')' is parsed everything in between should have
11829 * been parsed and evaluated to a single operand (or else is a syntax
11830 * error), and is handled as a regular operand */
11832 sv_2mortal((SV *)(stack = newAV()));
11834 while (RExC_parse < RExC_end) {
11835 I32 top_index = av_tindex(stack);
11837 SV* current = NULL;
11839 /* Skip white space */
11840 RExC_parse = regpatws(pRExC_state, RExC_parse,
11841 TRUE); /* means recognize comments */
11842 if (RExC_parse >= RExC_end) {
11843 Perl_croak(aTHX_ "panic: Read past end of '(?[ ])'");
11845 if ((curchar = UCHARAT(RExC_parse)) == ']') {
11852 if (av_tindex(stack) >= 0 /* This makes sure that we can
11853 safely subtract 1 from
11854 RExC_parse in the next clause.
11855 If we have something on the
11856 stack, we have parsed something
11858 && UCHARAT(RExC_parse - 1) == '('
11859 && RExC_parse < RExC_end)
11861 /* If is a '(?', could be an embedded '(?flags:(?[...])'.
11862 * This happens when we have some thing like
11864 * my $thai_or_lao = qr/(?[ \p{Thai} + \p{Lao} ])/;
11866 * qr/(?[ \p{Digit} & $thai_or_lao ])/;
11868 * Here we would be handling the interpolated
11869 * '$thai_or_lao'. We handle this by a recursive call to
11870 * ourselves which returns the inversion list the
11871 * interpolated expression evaluates to. We use the flags
11872 * from the interpolated pattern. */
11873 U32 save_flags = RExC_flags;
11874 const char * const save_parse = ++RExC_parse;
11876 parse_lparen_question_flags(pRExC_state);
11878 if (RExC_parse == save_parse /* Makes sure there was at
11879 least one flag (or this
11880 embedding wasn't compiled)
11882 || RExC_parse >= RExC_end - 4
11883 || UCHARAT(RExC_parse) != ':'
11884 || UCHARAT(++RExC_parse) != '('
11885 || UCHARAT(++RExC_parse) != '?'
11886 || UCHARAT(++RExC_parse) != '[')
11889 /* In combination with the above, this moves the
11890 * pointer to the point just after the first erroneous
11891 * character (or if there are no flags, to where they
11892 * should have been) */
11893 if (RExC_parse >= RExC_end - 4) {
11894 RExC_parse = RExC_end;
11896 else if (RExC_parse != save_parse) {
11897 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
11899 vFAIL("Expecting '(?flags:(?[...'");
11902 (void) handle_regex_sets(pRExC_state, ¤t, flagp,
11903 depth+1, oregcomp_parse);
11905 /* Here, 'current' contains the embedded expression's
11906 * inversion list, and RExC_parse points to the trailing
11907 * ']'; the next character should be the ')' which will be
11908 * paired with the '(' that has been put on the stack, so
11909 * the whole embedded expression reduces to '(operand)' */
11912 RExC_flags = save_flags;
11913 goto handle_operand;
11918 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
11919 vFAIL("Unexpected character");
11922 /* regclass() can only return RESTART_UTF8 if multi-char
11923 folds are allowed. */
11924 if (!regclass(pRExC_state, flagp,depth+1,
11925 TRUE, /* means parse just the next thing */
11926 FALSE, /* don't allow multi-char folds */
11927 FALSE, /* don't silence non-portable warnings. */
11929 FAIL2("panic: regclass returned NULL to handle_sets, flags=%#"UVxf"",
11931 /* regclass() will return with parsing just the \ sequence,
11932 * leaving the parse pointer at the next thing to parse */
11934 goto handle_operand;
11936 case '[': /* Is a bracketed character class */
11938 bool is_posix_class = could_it_be_a_POSIX_class(pRExC_state);
11940 if (! is_posix_class) {
11944 /* regclass() can only return RESTART_UTF8 if multi-char
11945 folds are allowed. */
11946 if(!regclass(pRExC_state, flagp,depth+1,
11947 is_posix_class, /* parse the whole char class
11948 only if not a posix class */
11949 FALSE, /* don't allow multi-char folds */
11950 FALSE, /* don't silence non-portable warnings. */
11952 FAIL2("panic: regclass returned NULL to handle_sets, flags=%#"UVxf"",
11954 /* function call leaves parse pointing to the ']', except if we
11956 if (is_posix_class) {
11960 goto handle_operand;
11969 || ( ! (top_ptr = av_fetch(stack, top_index, FALSE)))
11970 || ! IS_OPERAND(*top_ptr))
11973 vFAIL2("Unexpected binary operator '%c' with no preceding operand", curchar);
11975 av_push(stack, newSVuv(curchar));
11979 av_push(stack, newSVuv(curchar));
11983 if (top_index >= 0) {
11984 top_ptr = av_fetch(stack, top_index, FALSE);
11986 if (IS_OPERAND(*top_ptr)) {
11988 vFAIL("Unexpected '(' with no preceding operator");
11991 av_push(stack, newSVuv(curchar));
11998 || ! (current = av_pop(stack))
11999 || ! IS_OPERAND(current)
12000 || ! (lparen = av_pop(stack))
12001 || IS_OPERAND(lparen)
12002 || SvUV(lparen) != '(')
12004 SvREFCNT_dec(current);
12006 vFAIL("Unexpected ')'");
12009 SvREFCNT_dec_NN(lparen);
12016 /* Here, we have an operand to process, in 'current' */
12018 if (top_index < 0) { /* Just push if stack is empty */
12019 av_push(stack, current);
12022 SV* top = av_pop(stack);
12024 char current_operator;
12026 if (IS_OPERAND(top)) {
12027 SvREFCNT_dec_NN(top);
12028 SvREFCNT_dec_NN(current);
12029 vFAIL("Operand with no preceding operator");
12031 current_operator = (char) SvUV(top);
12032 switch (current_operator) {
12033 case '(': /* Push the '(' back on followed by the new
12035 av_push(stack, top);
12036 av_push(stack, current);
12037 SvREFCNT_inc(top); /* Counters the '_dec' done
12038 just after the 'break', so
12039 it doesn't get wrongly freed
12044 _invlist_invert(current);
12046 /* Unlike binary operators, the top of the stack,
12047 * now that this unary one has been popped off, may
12048 * legally be an operator, and we now have operand
12051 SvREFCNT_dec_NN(top);
12052 goto handle_operand;
12055 prev = av_pop(stack);
12056 _invlist_intersection(prev,
12059 av_push(stack, current);
12064 prev = av_pop(stack);
12065 _invlist_union(prev, current, ¤t);
12066 av_push(stack, current);
12070 prev = av_pop(stack);;
12071 _invlist_subtract(prev, current, ¤t);
12072 av_push(stack, current);
12075 case '^': /* The union minus the intersection */
12081 prev = av_pop(stack);
12082 _invlist_union(prev, current, &u);
12083 _invlist_intersection(prev, current, &i);
12084 /* _invlist_subtract will overwrite current
12085 without freeing what it already contains */
12087 _invlist_subtract(u, i, ¤t);
12088 av_push(stack, current);
12089 SvREFCNT_dec_NN(i);
12090 SvREFCNT_dec_NN(u);
12091 SvREFCNT_dec_NN(element);
12096 Perl_croak(aTHX_ "panic: Unexpected item on '(?[ ])' stack");
12098 SvREFCNT_dec_NN(top);
12099 SvREFCNT_dec(prev);
12103 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
12106 if (av_tindex(stack) < 0 /* Was empty */
12107 || ((final = av_pop(stack)) == NULL)
12108 || ! IS_OPERAND(final)
12109 || av_tindex(stack) >= 0) /* More left on stack */
12111 vFAIL("Incomplete expression within '(?[ ])'");
12114 /* Here, 'final' is the resultant inversion list from evaluating the
12115 * expression. Return it if so requested */
12116 if (return_invlist) {
12117 *return_invlist = final;
12121 /* Otherwise generate a resultant node, based on 'final'. regclass() is
12122 * expecting a string of ranges and individual code points */
12123 invlist_iterinit(final);
12124 result_string = newSVpvs("");
12125 while (invlist_iternext(final, &start, &end)) {
12126 if (start == end) {
12127 Perl_sv_catpvf(aTHX_ result_string, "\\x{%"UVXf"}", start);
12130 Perl_sv_catpvf(aTHX_ result_string, "\\x{%"UVXf"}-\\x{%"UVXf"}",
12135 save_parse = RExC_parse;
12136 RExC_parse = SvPV(result_string, len);
12137 save_end = RExC_end;
12138 RExC_end = RExC_parse + len;
12140 /* We turn off folding around the call, as the class we have constructed
12141 * already has all folding taken into consideration, and we don't want
12142 * regclass() to add to that */
12143 RExC_flags &= ~RXf_PMf_FOLD;
12144 /* regclass() can only return RESTART_UTF8 if multi-char folds are allowed.
12146 node = regclass(pRExC_state, flagp,depth+1,
12147 FALSE, /* means parse the whole char class */
12148 FALSE, /* don't allow multi-char folds */
12149 TRUE, /* silence non-portable warnings. The above may very
12150 well have generated non-portable code points, but
12151 they're valid on this machine */
12154 FAIL2("panic: regclass returned NULL to handle_sets, flags=%#"UVxf,
12157 RExC_flags |= RXf_PMf_FOLD;
12159 RExC_parse = save_parse + 1;
12160 RExC_end = save_end;
12161 SvREFCNT_dec_NN(final);
12162 SvREFCNT_dec_NN(result_string);
12164 nextchar(pRExC_state);
12165 Set_Node_Length(node, RExC_parse - oregcomp_parse + 1); /* MJD */
12170 /* The names of properties whose definitions are not known at compile time are
12171 * stored in this SV, after a constant heading. So if the length has been
12172 * changed since initialization, then there is a run-time definition. */
12173 #define HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION (SvCUR(listsv) != initial_listsv_len)
12176 S_regclass(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth,
12177 const bool stop_at_1, /* Just parse the next thing, don't
12178 look for a full character class */
12179 bool allow_multi_folds,
12180 const bool silence_non_portable, /* Don't output warnings
12183 SV** ret_invlist) /* Return an inversion list, not a node */
12185 /* parse a bracketed class specification. Most of these will produce an
12186 * ANYOF node; but something like [a] will produce an EXACT node; [aA], an
12187 * EXACTFish node; [[:ascii:]], a POSIXA node; etc. It is more complex
12188 * under /i with multi-character folds: it will be rewritten following the
12189 * paradigm of this example, where the <multi-fold>s are characters which
12190 * fold to multiple character sequences:
12191 * /[abc\x{multi-fold1}def\x{multi-fold2}ghi]/i
12192 * gets effectively rewritten as:
12193 * /(?:\x{multi-fold1}|\x{multi-fold2}|[abcdefghi]/i
12194 * reg() gets called (recursively) on the rewritten version, and this
12195 * function will return what it constructs. (Actually the <multi-fold>s
12196 * aren't physically removed from the [abcdefghi], it's just that they are
12197 * ignored in the recursion by means of a flag:
12198 * <RExC_in_multi_char_class>.)
12200 * ANYOF nodes contain a bit map for the first 256 characters, with the
12201 * corresponding bit set if that character is in the list. For characters
12202 * above 255, a range list or swash is used. There are extra bits for \w,
12203 * etc. in locale ANYOFs, as what these match is not determinable at
12206 * Returns NULL, setting *flagp to RESTART_UTF8 if the sizing scan needs
12207 * to be restarted. This can only happen if ret_invlist is non-NULL.
12211 UV prevvalue = OOB_UNICODE, save_prevvalue = OOB_UNICODE;
12213 UV value = OOB_UNICODE, save_value = OOB_UNICODE;
12216 IV namedclass = OOB_NAMEDCLASS;
12217 char *rangebegin = NULL;
12218 bool need_class = 0;
12220 STRLEN initial_listsv_len = 0; /* Kind of a kludge to see if it is more
12221 than just initialized. */
12222 SV* properties = NULL; /* Code points that match \p{} \P{} */
12223 SV* posixes = NULL; /* Code points that match classes like, [:word:],
12224 extended beyond the Latin1 range */
12225 UV element_count = 0; /* Number of distinct elements in the class.
12226 Optimizations may be possible if this is tiny */
12227 AV * multi_char_matches = NULL; /* Code points that fold to more than one
12228 character; used under /i */
12230 char * stop_ptr = RExC_end; /* where to stop parsing */
12231 const bool skip_white = cBOOL(ret_invlist); /* ignore unescaped white
12233 const bool strict = cBOOL(ret_invlist); /* Apply strict parsing rules? */
12235 /* Unicode properties are stored in a swash; this holds the current one
12236 * being parsed. If this swash is the only above-latin1 component of the
12237 * character class, an optimization is to pass it directly on to the
12238 * execution engine. Otherwise, it is set to NULL to indicate that there
12239 * are other things in the class that have to be dealt with at execution
12241 SV* swash = NULL; /* Code points that match \p{} \P{} */
12243 /* Set if a component of this character class is user-defined; just passed
12244 * on to the engine */
12245 bool has_user_defined_property = FALSE;
12247 /* inversion list of code points this node matches only when the target
12248 * string is in UTF-8. (Because is under /d) */
12249 SV* depends_list = NULL;
12251 /* inversion list of code points this node matches. For much of the
12252 * function, it includes only those that match regardless of the utf8ness
12253 * of the target string */
12254 SV* cp_list = NULL;
12257 /* In a range, counts how many 0-2 of the ends of it came from literals,
12258 * not escapes. Thus we can tell if 'A' was input vs \x{C1} */
12259 UV literal_endpoint = 0;
12261 bool invert = FALSE; /* Is this class to be complemented */
12263 /* Is there any thing like \W or [:^digit:] that matches above the legal
12264 * Unicode range? */
12265 bool runtime_posix_matches_above_Unicode = FALSE;
12267 regnode * const orig_emit = RExC_emit; /* Save the original RExC_emit in
12268 case we need to change the emitted regop to an EXACT. */
12269 const char * orig_parse = RExC_parse;
12270 const I32 orig_size = RExC_size;
12271 GET_RE_DEBUG_FLAGS_DECL;
12273 PERL_ARGS_ASSERT_REGCLASS;
12275 PERL_UNUSED_ARG(depth);
12278 DEBUG_PARSE("clas");
12280 /* Assume we are going to generate an ANYOF node. */
12281 ret = reganode(pRExC_state, ANYOF, 0);
12284 RExC_size += ANYOF_SKIP;
12285 listsv = &PL_sv_undef; /* For code scanners: listsv always non-NULL. */
12288 ANYOF_FLAGS(ret) = 0;
12290 RExC_emit += ANYOF_SKIP;
12292 ANYOF_FLAGS(ret) |= ANYOF_LOCALE;
12294 listsv = newSVpvs_flags("# comment\n", SVs_TEMP);
12295 initial_listsv_len = SvCUR(listsv);
12296 SvTEMP_off(listsv); /* Grr, TEMPs and mortals are conflated. */
12300 RExC_parse = regpatws(pRExC_state, RExC_parse,
12301 FALSE /* means don't recognize comments */);
12304 if (UCHARAT(RExC_parse) == '^') { /* Complement of range. */
12307 allow_multi_folds = FALSE;
12310 RExC_parse = regpatws(pRExC_state, RExC_parse,
12311 FALSE /* means don't recognize comments */);
12315 /* Check that they didn't say [:posix:] instead of [[:posix:]] */
12316 if (!SIZE_ONLY && RExC_parse < RExC_end && POSIXCC(UCHARAT(RExC_parse))) {
12317 const char *s = RExC_parse;
12318 const char c = *s++;
12320 while (isWORDCHAR(*s))
12322 if (*s && c == *s && s[1] == ']') {
12323 SAVEFREESV(RExC_rx_sv);
12325 "POSIX syntax [%c %c] belongs inside character classes",
12327 (void)ReREFCNT_inc(RExC_rx_sv);
12331 /* If the caller wants us to just parse a single element, accomplish this
12332 * by faking the loop ending condition */
12333 if (stop_at_1 && RExC_end > RExC_parse) {
12334 stop_ptr = RExC_parse + 1;
12337 /* allow 1st char to be ']' (allowing it to be '-' is dealt with later) */
12338 if (UCHARAT(RExC_parse) == ']')
12339 goto charclassloop;
12343 if (RExC_parse >= stop_ptr) {
12348 RExC_parse = regpatws(pRExC_state, RExC_parse,
12349 FALSE /* means don't recognize comments */);
12352 if (UCHARAT(RExC_parse) == ']') {
12358 namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
12359 save_value = value;
12360 save_prevvalue = prevvalue;
12363 rangebegin = RExC_parse;
12367 value = utf8n_to_uvchr((U8*)RExC_parse,
12368 RExC_end - RExC_parse,
12369 &numlen, UTF8_ALLOW_DEFAULT);
12370 RExC_parse += numlen;
12373 value = UCHARAT(RExC_parse++);
12376 && RExC_parse < RExC_end
12377 && POSIXCC(UCHARAT(RExC_parse)))
12379 namedclass = regpposixcc(pRExC_state, value, strict);
12381 else if (value == '\\') {
12383 value = utf8n_to_uvchr((U8*)RExC_parse,
12384 RExC_end - RExC_parse,
12385 &numlen, UTF8_ALLOW_DEFAULT);
12386 RExC_parse += numlen;
12389 value = UCHARAT(RExC_parse++);
12391 /* Some compilers cannot handle switching on 64-bit integer
12392 * values, therefore value cannot be an UV. Yes, this will
12393 * be a problem later if we want switch on Unicode.
12394 * A similar issue a little bit later when switching on
12395 * namedclass. --jhi */
12397 /* If the \ is escaping white space when white space is being
12398 * skipped, it means that that white space is wanted literally, and
12399 * is already in 'value'. Otherwise, need to translate the escape
12400 * into what it signifies. */
12401 if (! skip_white || ! is_PATWS_cp(value)) switch ((I32)value) {
12403 case 'w': namedclass = ANYOF_WORDCHAR; break;
12404 case 'W': namedclass = ANYOF_NWORDCHAR; break;
12405 case 's': namedclass = ANYOF_SPACE; break;
12406 case 'S': namedclass = ANYOF_NSPACE; break;
12407 case 'd': namedclass = ANYOF_DIGIT; break;
12408 case 'D': namedclass = ANYOF_NDIGIT; break;
12409 case 'v': namedclass = ANYOF_VERTWS; break;
12410 case 'V': namedclass = ANYOF_NVERTWS; break;
12411 case 'h': namedclass = ANYOF_HORIZWS; break;
12412 case 'H': namedclass = ANYOF_NHORIZWS; break;
12413 case 'N': /* Handle \N{NAME} in class */
12415 /* We only pay attention to the first char of
12416 multichar strings being returned. I kinda wonder
12417 if this makes sense as it does change the behaviour
12418 from earlier versions, OTOH that behaviour was broken
12420 if (! grok_bslash_N(pRExC_state, NULL, &value, flagp, depth,
12421 TRUE, /* => charclass */
12424 if (*flagp & RESTART_UTF8)
12425 FAIL("panic: grok_bslash_N set RESTART_UTF8");
12435 /* We will handle any undefined properties ourselves */
12436 U8 swash_init_flags = _CORE_SWASH_INIT_RETURN_IF_UNDEF;
12438 if (RExC_parse >= RExC_end)
12439 vFAIL2("Empty \\%c{}", (U8)value);
12440 if (*RExC_parse == '{') {
12441 const U8 c = (U8)value;
12442 e = strchr(RExC_parse++, '}');
12444 vFAIL2("Missing right brace on \\%c{}", c);
12445 while (isSPACE(UCHARAT(RExC_parse)))
12447 if (e == RExC_parse)
12448 vFAIL2("Empty \\%c{}", c);
12449 n = e - RExC_parse;
12450 while (isSPACE(UCHARAT(RExC_parse + n - 1)))
12461 if (UCHARAT(RExC_parse) == '^') {
12464 /* toggle. (The rhs xor gets the single bit that
12465 * differs between P and p; the other xor inverts just
12467 value ^= 'P' ^ 'p';
12469 while (isSPACE(UCHARAT(RExC_parse))) {
12474 /* Try to get the definition of the property into
12475 * <invlist>. If /i is in effect, the effective property
12476 * will have its name be <__NAME_i>. The design is
12477 * discussed in commit
12478 * 2f833f5208e26b208886e51e09e2c072b5eabb46 */
12479 Newx(name, n + sizeof("_i__\n"), char);
12481 sprintf(name, "%s%.*s%s\n",
12482 (FOLD) ? "__" : "",
12488 /* Look up the property name, and get its swash and
12489 * inversion list, if the property is found */
12491 SvREFCNT_dec_NN(swash);
12493 swash = _core_swash_init("utf8", name, &PL_sv_undef,
12496 NULL, /* No inversion list */
12499 if (! swash || ! (invlist = _get_swash_invlist(swash))) {
12501 SvREFCNT_dec_NN(swash);
12505 /* Here didn't find it. It could be a user-defined
12506 * property that will be available at run-time. If we
12507 * accept only compile-time properties, is an error;
12508 * otherwise add it to the list for run-time look up */
12510 RExC_parse = e + 1;
12511 vFAIL3("Property '%.*s' is unknown", (int) n, name);
12513 Perl_sv_catpvf(aTHX_ listsv, "%cutf8::%s\n",
12514 (value == 'p' ? '+' : '!'),
12516 has_user_defined_property = TRUE;
12518 /* We don't know yet, so have to assume that the
12519 * property could match something in the Latin1 range,
12520 * hence something that isn't utf8. Note that this
12521 * would cause things in <depends_list> to match
12522 * inappropriately, except that any \p{}, including
12523 * this one forces Unicode semantics, which means there
12524 * is <no depends_list> */
12525 ANYOF_FLAGS(ret) |= ANYOF_NONBITMAP_NON_UTF8;
12529 /* Here, did get the swash and its inversion list. If
12530 * the swash is from a user-defined property, then this
12531 * whole character class should be regarded as such */
12532 has_user_defined_property =
12534 & _CORE_SWASH_INIT_USER_DEFINED_PROPERTY);
12536 /* Invert if asking for the complement */
12537 if (value == 'P') {
12538 _invlist_union_complement_2nd(properties,
12542 /* The swash can't be used as-is, because we've
12543 * inverted things; delay removing it to here after
12544 * have copied its invlist above */
12545 SvREFCNT_dec_NN(swash);
12549 _invlist_union(properties, invlist, &properties);
12554 RExC_parse = e + 1;
12555 namedclass = ANYOF_UNIPROP; /* no official name, but it's
12558 /* \p means they want Unicode semantics */
12559 RExC_uni_semantics = 1;
12562 case 'n': value = '\n'; break;
12563 case 'r': value = '\r'; break;
12564 case 't': value = '\t'; break;
12565 case 'f': value = '\f'; break;
12566 case 'b': value = '\b'; break;
12567 case 'e': value = ASCII_TO_NATIVE('\033');break;
12568 case 'a': value = ASCII_TO_NATIVE('\007');break;
12570 RExC_parse--; /* function expects to be pointed at the 'o' */
12572 const char* error_msg;
12573 bool valid = grok_bslash_o(&RExC_parse,
12576 SIZE_ONLY, /* warnings in pass
12579 silence_non_portable,
12585 if (PL_encoding && value < 0x100) {
12586 goto recode_encoding;
12590 RExC_parse--; /* function expects to be pointed at the 'x' */
12592 const char* error_msg;
12593 bool valid = grok_bslash_x(&RExC_parse,
12596 TRUE, /* Output warnings */
12598 silence_non_portable,
12604 if (PL_encoding && value < 0x100)
12605 goto recode_encoding;
12608 value = grok_bslash_c(*RExC_parse++, UTF, SIZE_ONLY);
12610 case '0': case '1': case '2': case '3': case '4':
12611 case '5': case '6': case '7':
12613 /* Take 1-3 octal digits */
12614 I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
12615 numlen = (strict) ? 4 : 3;
12616 value = grok_oct(--RExC_parse, &numlen, &flags, NULL);
12617 RExC_parse += numlen;
12620 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
12621 vFAIL("Need exactly 3 octal digits");
12623 else if (! SIZE_ONLY /* like \08, \178 */
12625 && RExC_parse < RExC_end
12626 && isDIGIT(*RExC_parse)
12627 && ckWARN(WARN_REGEXP))
12629 SAVEFREESV(RExC_rx_sv);
12630 reg_warn_non_literal_string(
12632 form_short_octal_warning(RExC_parse, numlen));
12633 (void)ReREFCNT_inc(RExC_rx_sv);
12636 if (PL_encoding && value < 0x100)
12637 goto recode_encoding;
12641 if (! RExC_override_recoding) {
12642 SV* enc = PL_encoding;
12643 value = reg_recode((const char)(U8)value, &enc);
12646 vFAIL("Invalid escape in the specified encoding");
12648 else if (SIZE_ONLY) {
12649 ckWARNreg(RExC_parse,
12650 "Invalid escape in the specified encoding");
12656 /* Allow \_ to not give an error */
12657 if (!SIZE_ONLY && isWORDCHAR(value) && value != '_') {
12659 vFAIL2("Unrecognized escape \\%c in character class",
12663 SAVEFREESV(RExC_rx_sv);
12664 ckWARN2reg(RExC_parse,
12665 "Unrecognized escape \\%c in character class passed through",
12667 (void)ReREFCNT_inc(RExC_rx_sv);
12671 } /* End of switch on char following backslash */
12672 } /* end of handling backslash escape sequences */
12675 literal_endpoint++;
12678 /* Here, we have the current token in 'value' */
12680 /* What matches in a locale is not known until runtime. This includes
12681 * what the Posix classes (like \w, [:space:]) match. Room must be
12682 * reserved (one time per class) to store such classes, either if Perl
12683 * is compiled so that locale nodes always should have this space, or
12684 * if there is such class info to be stored. The space will contain a
12685 * bit for each named class that is to be matched against. This isn't
12686 * needed for \p{} and pseudo-classes, as they are not affected by
12687 * locale, and hence are dealt with separately */
12690 && (ANYOF_LOCALE == ANYOF_CLASS
12691 || (namedclass > OOB_NAMEDCLASS && namedclass < ANYOF_MAX)))
12695 RExC_size += ANYOF_CLASS_SKIP - ANYOF_SKIP;
12698 RExC_emit += ANYOF_CLASS_SKIP - ANYOF_SKIP;
12699 ANYOF_CLASS_ZERO(ret);
12701 ANYOF_FLAGS(ret) |= ANYOF_CLASS;
12704 if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
12706 /* a bad range like a-\d, a-[:digit:]. The '-' is taken as a
12707 * literal, as is the character that began the false range, i.e.
12708 * the 'a' in the examples */
12711 const int w = (RExC_parse >= rangebegin)
12712 ? RExC_parse - rangebegin
12715 vFAIL4("False [] range \"%*.*s\"", w, w, rangebegin);
12718 SAVEFREESV(RExC_rx_sv); /* in case of fatal warnings */
12719 ckWARN4reg(RExC_parse,
12720 "False [] range \"%*.*s\"",
12722 (void)ReREFCNT_inc(RExC_rx_sv);
12723 cp_list = add_cp_to_invlist(cp_list, '-');
12724 cp_list = add_cp_to_invlist(cp_list, prevvalue);
12728 range = 0; /* this was not a true range */
12729 element_count += 2; /* So counts for three values */
12733 U8 classnum = namedclass_to_classnum(namedclass);
12734 if (namedclass >= ANYOF_MAX) { /* If a special class */
12735 if (namedclass != ANYOF_UNIPROP) { /* UNIPROP = \p and \P */
12737 /* Here, should be \h, \H, \v, or \V. Neither /d nor
12738 * /l make a difference in what these match. There
12739 * would be problems if these characters had folds
12740 * other than themselves, as cp_list is subject to
12742 if (classnum != _CC_VERTSPACE) {
12743 assert( namedclass == ANYOF_HORIZWS
12744 || namedclass == ANYOF_NHORIZWS);
12746 /* It turns out that \h is just a synonym for
12748 classnum = _CC_BLANK;
12751 _invlist_union_maybe_complement_2nd(
12753 PL_XPosix_ptrs[classnum],
12754 cBOOL(namedclass % 2), /* Complement if odd
12755 (NHORIZWS, NVERTWS)
12760 else if (classnum == _CC_ASCII) {
12763 ANYOF_CLASS_SET(ret, namedclass);
12766 #endif /* Not isascii(); just use the hard-coded definition for it */
12767 _invlist_union_maybe_complement_2nd(
12770 cBOOL(namedclass % 2), /* Complement if odd
12774 else { /* Garden variety class */
12776 /* The ascii range inversion list */
12777 SV* ascii_source = PL_Posix_ptrs[classnum];
12779 /* The full Latin1 range inversion list */
12780 SV* l1_source = PL_L1Posix_ptrs[classnum];
12782 /* This code is structured into two major clauses. The
12783 * first is for classes whose complete definitions may not
12784 * already be known. It not, the Latin1 definition
12785 * (guaranteed to already known) is used plus code is
12786 * generated to load the rest at run-time (only if needed).
12787 * If the complete definition is known, it drops down to
12788 * the second clause, where the complete definition is
12791 if (classnum < _FIRST_NON_SWASH_CC) {
12793 /* Here, the class has a swash, which may or not
12794 * already be loaded */
12796 /* The name of the property to use to match the full
12797 * eXtended Unicode range swash for this character
12799 const char *Xname = swash_property_names[classnum];
12801 /* If returning the inversion list, we can't defer
12802 * getting this until runtime */
12803 if (ret_invlist && ! PL_utf8_swash_ptrs[classnum]) {
12804 PL_utf8_swash_ptrs[classnum] =
12805 _core_swash_init("utf8", Xname, &PL_sv_undef,
12808 NULL, /* No inversion list */
12809 NULL /* No flags */
12811 assert(PL_utf8_swash_ptrs[classnum]);
12813 if ( ! PL_utf8_swash_ptrs[classnum]) {
12814 if (namedclass % 2 == 0) { /* A non-complemented
12816 /* If not /a matching, there are code points we
12817 * don't know at compile time. Arrange for the
12818 * unknown matches to be loaded at run-time, if
12820 if (! AT_LEAST_ASCII_RESTRICTED) {
12821 Perl_sv_catpvf(aTHX_ listsv, "+utf8::%s\n",
12824 if (LOC) { /* Under locale, set run-time
12826 ANYOF_CLASS_SET(ret, namedclass);
12829 /* Add the current class's code points to
12830 * the running total */
12831 _invlist_union(posixes,
12832 (AT_LEAST_ASCII_RESTRICTED)
12838 else { /* A complemented class */
12839 if (AT_LEAST_ASCII_RESTRICTED) {
12840 /* Under /a should match everything above
12841 * ASCII, plus the complement of the set's
12843 _invlist_union_complement_2nd(posixes,
12848 /* Arrange for the unknown matches to be
12849 * loaded at run-time, if needed */
12850 Perl_sv_catpvf(aTHX_ listsv, "!utf8::%s\n",
12852 runtime_posix_matches_above_Unicode = TRUE;
12854 ANYOF_CLASS_SET(ret, namedclass);
12858 /* We want to match everything in
12859 * Latin1, except those things that
12860 * l1_source matches */
12861 SV* scratch_list = NULL;
12862 _invlist_subtract(PL_Latin1, l1_source,
12865 /* Add the list from this class to the
12868 posixes = scratch_list;
12871 _invlist_union(posixes,
12874 SvREFCNT_dec_NN(scratch_list);
12876 if (DEPENDS_SEMANTICS) {
12878 |= ANYOF_NON_UTF8_LATIN1_ALL;
12883 goto namedclass_done;
12886 /* Here, there is a swash loaded for the class. If no
12887 * inversion list for it yet, get it */
12888 if (! PL_XPosix_ptrs[classnum]) {
12889 PL_XPosix_ptrs[classnum]
12890 = _swash_to_invlist(PL_utf8_swash_ptrs[classnum]);
12894 /* Here there is an inversion list already loaded for the
12897 if (namedclass % 2 == 0) { /* A non-complemented class,
12898 like ANYOF_PUNCT */
12900 /* For non-locale, just add it to any existing list
12902 _invlist_union(posixes,
12903 (AT_LEAST_ASCII_RESTRICTED)
12905 : PL_XPosix_ptrs[classnum],
12908 else { /* Locale */
12909 SV* scratch_list = NULL;
12911 /* For above Latin1 code points, we use the full
12913 _invlist_intersection(PL_AboveLatin1,
12914 PL_XPosix_ptrs[classnum],
12916 /* And set the output to it, adding instead if
12917 * there already is an output. Checking if
12918 * 'posixes' is NULL first saves an extra clone.
12919 * Its reference count will be decremented at the
12920 * next union, etc, or if this is the only
12921 * instance, at the end of the routine */
12923 posixes = scratch_list;
12926 _invlist_union(posixes, scratch_list, &posixes);
12927 SvREFCNT_dec_NN(scratch_list);
12930 #ifndef HAS_ISBLANK
12931 if (namedclass != ANYOF_BLANK) {
12933 /* Set this class in the node for runtime
12935 ANYOF_CLASS_SET(ret, namedclass);
12936 #ifndef HAS_ISBLANK
12939 /* No isblank(), use the hard-coded ASCII-range
12940 * blanks, adding them to the running total. */
12942 _invlist_union(posixes, ascii_source, &posixes);
12947 else { /* A complemented class, like ANYOF_NPUNCT */
12949 _invlist_union_complement_2nd(
12951 (AT_LEAST_ASCII_RESTRICTED)
12953 : PL_XPosix_ptrs[classnum],
12955 /* Under /d, everything in the upper half of the
12956 * Latin1 range matches this complement */
12957 if (DEPENDS_SEMANTICS) {
12958 ANYOF_FLAGS(ret) |= ANYOF_NON_UTF8_LATIN1_ALL;
12961 else { /* Locale */
12962 SV* scratch_list = NULL;
12963 _invlist_subtract(PL_AboveLatin1,
12964 PL_XPosix_ptrs[classnum],
12967 posixes = scratch_list;
12970 _invlist_union(posixes, scratch_list, &posixes);
12971 SvREFCNT_dec_NN(scratch_list);
12973 #ifndef HAS_ISBLANK
12974 if (namedclass != ANYOF_NBLANK) {
12976 ANYOF_CLASS_SET(ret, namedclass);
12977 #ifndef HAS_ISBLANK
12980 /* Get the list of all code points in Latin1
12981 * that are not ASCII blanks, and add them to
12982 * the running total */
12983 _invlist_subtract(PL_Latin1, ascii_source,
12985 _invlist_union(posixes, scratch_list, &posixes);
12986 SvREFCNT_dec_NN(scratch_list);
12993 continue; /* Go get next character */
12995 } /* end of namedclass \blah */
12997 /* Here, we have a single value. If 'range' is set, it is the ending
12998 * of a range--check its validity. Later, we will handle each
12999 * individual code point in the range. If 'range' isn't set, this
13000 * could be the beginning of a range, so check for that by looking
13001 * ahead to see if the next real character to be processed is the range
13002 * indicator--the minus sign */
13005 RExC_parse = regpatws(pRExC_state, RExC_parse,
13006 FALSE /* means don't recognize comments */);
13010 if (prevvalue > value) /* b-a */ {
13011 const int w = RExC_parse - rangebegin;
13012 Simple_vFAIL4("Invalid [] range \"%*.*s\"", w, w, rangebegin);
13013 range = 0; /* not a valid range */
13017 prevvalue = value; /* save the beginning of the potential range */
13018 if (! stop_at_1 /* Can't be a range if parsing just one thing */
13019 && *RExC_parse == '-')
13021 char* next_char_ptr = RExC_parse + 1;
13022 if (skip_white) { /* Get the next real char after the '-' */
13023 next_char_ptr = regpatws(pRExC_state,
13025 FALSE); /* means don't recognize
13029 /* If the '-' is at the end of the class (just before the ']',
13030 * it is a literal minus; otherwise it is a range */
13031 if (next_char_ptr < RExC_end && *next_char_ptr != ']') {
13032 RExC_parse = next_char_ptr;
13034 /* a bad range like \w-, [:word:]- ? */
13035 if (namedclass > OOB_NAMEDCLASS) {
13036 if (strict || ckWARN(WARN_REGEXP)) {
13038 RExC_parse >= rangebegin ?
13039 RExC_parse - rangebegin : 0;
13041 vFAIL4("False [] range \"%*.*s\"",
13046 "False [] range \"%*.*s\"",
13051 cp_list = add_cp_to_invlist(cp_list, '-');
13055 range = 1; /* yeah, it's a range! */
13056 continue; /* but do it the next time */
13061 /* Here, <prevvalue> is the beginning of the range, if any; or <value>
13064 /* non-Latin1 code point implies unicode semantics. Must be set in
13065 * pass1 so is there for the whole of pass 2 */
13067 RExC_uni_semantics = 1;
13070 /* Ready to process either the single value, or the completed range.
13071 * For single-valued non-inverted ranges, we consider the possibility
13072 * of multi-char folds. (We made a conscious decision to not do this
13073 * for the other cases because it can often lead to non-intuitive
13074 * results. For example, you have the peculiar case that:
13075 * "s s" =~ /^[^\xDF]+$/i => Y
13076 * "ss" =~ /^[^\xDF]+$/i => N
13078 * See [perl #89750] */
13079 if (FOLD && allow_multi_folds && value == prevvalue) {
13080 if (value == LATIN_SMALL_LETTER_SHARP_S
13081 || (value > 255 && _invlist_contains_cp(PL_HasMultiCharFold,
13084 /* Here <value> is indeed a multi-char fold. Get what it is */
13086 U8 foldbuf[UTF8_MAXBYTES_CASE];
13089 UV folded = _to_uni_fold_flags(
13094 | ((LOC) ? FOLD_FLAGS_LOCALE
13095 : (ASCII_FOLD_RESTRICTED)
13096 ? FOLD_FLAGS_NOMIX_ASCII
13100 /* Here, <folded> should be the first character of the
13101 * multi-char fold of <value>, with <foldbuf> containing the
13102 * whole thing. But, if this fold is not allowed (because of
13103 * the flags), <fold> will be the same as <value>, and should
13104 * be processed like any other character, so skip the special
13106 if (folded != value) {
13108 /* Skip if we are recursed, currently parsing the class
13109 * again. Otherwise add this character to the list of
13110 * multi-char folds. */
13111 if (! RExC_in_multi_char_class) {
13112 AV** this_array_ptr;
13114 STRLEN cp_count = utf8_length(foldbuf,
13115 foldbuf + foldlen);
13116 SV* multi_fold = sv_2mortal(newSVpvn("", 0));
13118 Perl_sv_catpvf(aTHX_ multi_fold, "\\x{%"UVXf"}", value);
13121 if (! multi_char_matches) {
13122 multi_char_matches = newAV();
13125 /* <multi_char_matches> is actually an array of arrays.
13126 * There will be one or two top-level elements: [2],
13127 * and/or [3]. The [2] element is an array, each
13128 * element thereof is a character which folds to TWO
13129 * characters; [3] is for folds to THREE characters.
13130 * (Unicode guarantees a maximum of 3 characters in any
13131 * fold.) When we rewrite the character class below,
13132 * we will do so such that the longest folds are
13133 * written first, so that it prefers the longest
13134 * matching strings first. This is done even if it
13135 * turns out that any quantifier is non-greedy, out of
13136 * programmer laziness. Tom Christiansen has agreed
13137 * that this is ok. This makes the test for the
13138 * ligature 'ffi' come before the test for 'ff' */
13139 if (av_exists(multi_char_matches, cp_count)) {
13140 this_array_ptr = (AV**) av_fetch(multi_char_matches,
13142 this_array = *this_array_ptr;
13145 this_array = newAV();
13146 av_store(multi_char_matches, cp_count,
13149 av_push(this_array, multi_fold);
13152 /* This element should not be processed further in this
13155 value = save_value;
13156 prevvalue = save_prevvalue;
13162 /* Deal with this element of the class */
13165 cp_list = _add_range_to_invlist(cp_list, prevvalue, value);
13167 SV* this_range = _new_invlist(1);
13168 _append_range_to_invlist(this_range, prevvalue, value);
13170 /* In EBCDIC, the ranges 'A-Z' and 'a-z' are each not contiguous.
13171 * If this range was specified using something like 'i-j', we want
13172 * to include only the 'i' and the 'j', and not anything in
13173 * between, so exclude non-ASCII, non-alphabetics from it.
13174 * However, if the range was specified with something like
13175 * [\x89-\x91] or [\x89-j], all code points within it should be
13176 * included. literal_endpoint==2 means both ends of the range used
13177 * a literal character, not \x{foo} */
13178 if (literal_endpoint == 2
13179 && (prevvalue >= 'a' && value <= 'z')
13180 || (prevvalue >= 'A' && value <= 'Z'))
13182 _invlist_intersection(this_range, PL_Posix_ptrs[_CC_ALPHA],
13185 _invlist_union(cp_list, this_range, &cp_list);
13186 literal_endpoint = 0;
13190 range = 0; /* this range (if it was one) is done now */
13191 } /* End of loop through all the text within the brackets */
13193 /* If anything in the class expands to more than one character, we have to
13194 * deal with them by building up a substitute parse string, and recursively
13195 * calling reg() on it, instead of proceeding */
13196 if (multi_char_matches) {
13197 SV * substitute_parse = newSVpvn_flags("?:", 2, SVs_TEMP);
13200 char *save_end = RExC_end;
13201 char *save_parse = RExC_parse;
13202 bool first_time = TRUE; /* First multi-char occurrence doesn't get
13207 #if 0 /* Have decided not to deal with multi-char folds in inverted classes,
13208 because too confusing */
13210 sv_catpv(substitute_parse, "(?:");
13214 /* Look at the longest folds first */
13215 for (cp_count = av_len(multi_char_matches); cp_count > 0; cp_count--) {
13217 if (av_exists(multi_char_matches, cp_count)) {
13218 AV** this_array_ptr;
13221 this_array_ptr = (AV**) av_fetch(multi_char_matches,
13223 while ((this_sequence = av_pop(*this_array_ptr)) !=
13226 if (! first_time) {
13227 sv_catpv(substitute_parse, "|");
13229 first_time = FALSE;
13231 sv_catpv(substitute_parse, SvPVX(this_sequence));
13236 /* If the character class contains anything else besides these
13237 * multi-character folds, have to include it in recursive parsing */
13238 if (element_count) {
13239 sv_catpv(substitute_parse, "|[");
13240 sv_catpvn(substitute_parse, orig_parse, RExC_parse - orig_parse);
13241 sv_catpv(substitute_parse, "]");
13244 sv_catpv(substitute_parse, ")");
13247 /* This is a way to get the parse to skip forward a whole named
13248 * sequence instead of matching the 2nd character when it fails the
13250 sv_catpv(substitute_parse, "(*THEN)(*SKIP)(*FAIL)|.)");
13254 RExC_parse = SvPV(substitute_parse, len);
13255 RExC_end = RExC_parse + len;
13256 RExC_in_multi_char_class = 1;
13257 RExC_emit = (regnode *)orig_emit;
13259 ret = reg(pRExC_state, 1, ®_flags, depth+1);
13261 *flagp |= reg_flags&(HASWIDTH|SIMPLE|SPSTART|POSTPONED|RESTART_UTF8);
13263 RExC_parse = save_parse;
13264 RExC_end = save_end;
13265 RExC_in_multi_char_class = 0;
13266 SvREFCNT_dec_NN(multi_char_matches);
13270 /* If the character class contains only a single element, it may be
13271 * optimizable into another node type which is smaller and runs faster.
13272 * Check if this is the case for this class */
13273 if (element_count == 1 && ! ret_invlist) {
13277 if (namedclass > OOB_NAMEDCLASS) { /* this is a named class, like \w or
13278 [:digit:] or \p{foo} */
13280 /* All named classes are mapped into POSIXish nodes, with its FLAG
13281 * argument giving which class it is */
13282 switch ((I32)namedclass) {
13283 case ANYOF_UNIPROP:
13286 /* These don't depend on the charset modifiers. They always
13287 * match under /u rules */
13288 case ANYOF_NHORIZWS:
13289 case ANYOF_HORIZWS:
13290 namedclass = ANYOF_BLANK + namedclass - ANYOF_HORIZWS;
13293 case ANYOF_NVERTWS:
13298 /* The actual POSIXish node for all the rest depends on the
13299 * charset modifier. The ones in the first set depend only on
13300 * ASCII or, if available on this platform, locale */
13304 op = (LOC) ? POSIXL : POSIXA;
13315 /* under /a could be alpha */
13317 if (ASCII_RESTRICTED) {
13318 namedclass = ANYOF_ALPHA + (namedclass % 2);
13326 /* The rest have more possibilities depending on the charset.
13327 * We take advantage of the enum ordering of the charset
13328 * modifiers to get the exact node type, */
13330 op = POSIXD + get_regex_charset(RExC_flags);
13331 if (op > POSIXA) { /* /aa is same as /a */
13334 #ifndef HAS_ISBLANK
13336 && (namedclass == ANYOF_BLANK
13337 || namedclass == ANYOF_NBLANK))
13344 /* The odd numbered ones are the complements of the
13345 * next-lower even number one */
13346 if (namedclass % 2 == 1) {
13350 arg = namedclass_to_classnum(namedclass);
13354 else if (value == prevvalue) {
13356 /* Here, the class consists of just a single code point */
13359 if (! LOC && value == '\n') {
13360 op = REG_ANY; /* Optimize [^\n] */
13361 *flagp |= HASWIDTH|SIMPLE;
13365 else if (value < 256 || UTF) {
13367 /* Optimize a single value into an EXACTish node, but not if it
13368 * would require converting the pattern to UTF-8. */
13369 op = compute_EXACTish(pRExC_state);
13371 } /* Otherwise is a range */
13372 else if (! LOC) { /* locale could vary these */
13373 if (prevvalue == '0') {
13374 if (value == '9') {
13381 /* Here, we have changed <op> away from its initial value iff we found
13382 * an optimization */
13385 /* Throw away this ANYOF regnode, and emit the calculated one,
13386 * which should correspond to the beginning, not current, state of
13388 const char * cur_parse = RExC_parse;
13389 RExC_parse = (char *)orig_parse;
13393 /* To get locale nodes to not use the full ANYOF size would
13394 * require moving the code above that writes the portions
13395 * of it that aren't in other nodes to after this point.
13396 * e.g. ANYOF_CLASS_SET */
13397 RExC_size = orig_size;
13401 RExC_emit = (regnode *)orig_emit;
13402 if (PL_regkind[op] == POSIXD) {
13404 op += NPOSIXD - POSIXD;
13409 ret = reg_node(pRExC_state, op);
13411 if (PL_regkind[op] == POSIXD || PL_regkind[op] == NPOSIXD) {
13415 *flagp |= HASWIDTH|SIMPLE;
13417 else if (PL_regkind[op] == EXACT) {
13418 alloc_maybe_populate_EXACT(pRExC_state, ret, flagp, 0, value);
13421 RExC_parse = (char *) cur_parse;
13423 SvREFCNT_dec(posixes);
13424 SvREFCNT_dec(cp_list);
13431 /****** !SIZE_ONLY (Pass 2) AFTER HERE *********/
13433 /* If folding, we calculate all characters that could fold to or from the
13434 * ones already on the list */
13435 if (FOLD && cp_list) {
13436 UV start, end; /* End points of code point ranges */
13438 SV* fold_intersection = NULL;
13440 /* If the highest code point is within Latin1, we can use the
13441 * compiled-in Alphas list, and not have to go out to disk. This
13442 * yields two false positives, the masculine and feminine ordinal
13443 * indicators, which are weeded out below using the
13444 * IS_IN_SOME_FOLD_L1() macro */
13445 if (invlist_highest(cp_list) < 256) {
13446 _invlist_intersection(PL_L1Posix_ptrs[_CC_ALPHA], cp_list,
13447 &fold_intersection);
13451 /* Here, there are non-Latin1 code points, so we will have to go
13452 * fetch the list of all the characters that participate in folds
13454 if (! PL_utf8_foldable) {
13455 SV* swash = swash_init("utf8", "_Perl_Any_Folds",
13456 &PL_sv_undef, 1, 0);
13457 PL_utf8_foldable = _get_swash_invlist(swash);
13458 SvREFCNT_dec_NN(swash);
13461 /* This is a hash that for a particular fold gives all characters
13462 * that are involved in it */
13463 if (! PL_utf8_foldclosures) {
13465 /* If we were unable to find any folds, then we likely won't be
13466 * able to find the closures. So just create an empty list.
13467 * Folding will effectively be restricted to the non-Unicode
13468 * rules hard-coded into Perl. (This case happens legitimately
13469 * during compilation of Perl itself before the Unicode tables
13470 * are generated) */
13471 if (_invlist_len(PL_utf8_foldable) == 0) {
13472 PL_utf8_foldclosures = newHV();
13475 /* If the folds haven't been read in, call a fold function
13477 if (! PL_utf8_tofold) {
13478 U8 dummy[UTF8_MAXBYTES+1];
13480 /* This string is just a short named one above \xff */
13481 to_utf8_fold((U8*) HYPHEN_UTF8, dummy, NULL);
13482 assert(PL_utf8_tofold); /* Verify that worked */
13484 PL_utf8_foldclosures =
13485 _swash_inversion_hash(PL_utf8_tofold);
13489 /* Only the characters in this class that participate in folds need
13490 * be checked. Get the intersection of this class and all the
13491 * possible characters that are foldable. This can quickly narrow
13492 * down a large class */
13493 _invlist_intersection(PL_utf8_foldable, cp_list,
13494 &fold_intersection);
13497 /* Now look at the foldable characters in this class individually */
13498 invlist_iterinit(fold_intersection);
13499 while (invlist_iternext(fold_intersection, &start, &end)) {
13502 /* Locale folding for Latin1 characters is deferred until runtime */
13503 if (LOC && start < 256) {
13507 /* Look at every character in the range */
13508 for (j = start; j <= end; j++) {
13510 U8 foldbuf[UTF8_MAXBYTES_CASE+1];
13516 /* We have the latin1 folding rules hard-coded here so that
13517 * an innocent-looking character class, like /[ks]/i won't
13518 * have to go out to disk to find the possible matches.
13519 * XXX It would be better to generate these via regen, in
13520 * case a new version of the Unicode standard adds new
13521 * mappings, though that is not really likely, and may be
13522 * caught by the default: case of the switch below. */
13524 if (IS_IN_SOME_FOLD_L1(j)) {
13526 /* ASCII is always matched; non-ASCII is matched only
13527 * under Unicode rules */
13528 if (isASCII(j) || AT_LEAST_UNI_SEMANTICS) {
13530 add_cp_to_invlist(cp_list, PL_fold_latin1[j]);
13534 add_cp_to_invlist(depends_list, PL_fold_latin1[j]);
13538 if (HAS_NONLATIN1_FOLD_CLOSURE(j)
13539 && (! isASCII(j) || ! ASCII_FOLD_RESTRICTED))
13541 /* Certain Latin1 characters have matches outside
13542 * Latin1. To get here, <j> is one of those
13543 * characters. None of these matches is valid for
13544 * ASCII characters under /aa, which is why the 'if'
13545 * just above excludes those. These matches only
13546 * happen when the target string is utf8. The code
13547 * below adds the single fold closures for <j> to the
13548 * inversion list. */
13553 add_cp_to_invlist(cp_list, KELVIN_SIGN);
13557 cp_list = add_cp_to_invlist(cp_list,
13558 LATIN_SMALL_LETTER_LONG_S);
13561 cp_list = add_cp_to_invlist(cp_list,
13562 GREEK_CAPITAL_LETTER_MU);
13563 cp_list = add_cp_to_invlist(cp_list,
13564 GREEK_SMALL_LETTER_MU);
13566 case LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE:
13567 case LATIN_SMALL_LETTER_A_WITH_RING_ABOVE:
13569 add_cp_to_invlist(cp_list, ANGSTROM_SIGN);
13571 case LATIN_SMALL_LETTER_Y_WITH_DIAERESIS:
13572 cp_list = add_cp_to_invlist(cp_list,
13573 LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS);
13575 case LATIN_SMALL_LETTER_SHARP_S:
13576 cp_list = add_cp_to_invlist(cp_list,
13577 LATIN_CAPITAL_LETTER_SHARP_S);
13579 case 'F': case 'f':
13580 case 'I': case 'i':
13581 case 'L': case 'l':
13582 case 'T': case 't':
13583 case 'A': case 'a':
13584 case 'H': case 'h':
13585 case 'J': case 'j':
13586 case 'N': case 'n':
13587 case 'W': case 'w':
13588 case 'Y': case 'y':
13589 /* These all are targets of multi-character
13590 * folds from code points that require UTF8 to
13591 * express, so they can't match unless the
13592 * target string is in UTF-8, so no action here
13593 * is necessary, as regexec.c properly handles
13594 * the general case for UTF-8 matching and
13595 * multi-char folds */
13598 /* Use deprecated warning to increase the
13599 * chances of this being output */
13600 ckWARN2reg_d(RExC_parse, "Perl folding rules are not up-to-date for 0x%"UVXf"; please use the perlbug utility to report;", j);
13607 /* Here is an above Latin1 character. We don't have the rules
13608 * hard-coded for it. First, get its fold. This is the simple
13609 * fold, as the multi-character folds have been handled earlier
13610 * and separated out */
13611 _to_uni_fold_flags(j, foldbuf, &foldlen,
13613 ? FOLD_FLAGS_LOCALE
13614 : (ASCII_FOLD_RESTRICTED)
13615 ? FOLD_FLAGS_NOMIX_ASCII
13618 /* Single character fold of above Latin1. Add everything in
13619 * its fold closure to the list that this node should match.
13620 * The fold closures data structure is a hash with the keys
13621 * being the UTF-8 of every character that is folded to, like
13622 * 'k', and the values each an array of all code points that
13623 * fold to its key. e.g. [ 'k', 'K', KELVIN_SIGN ].
13624 * Multi-character folds are not included */
13625 if ((listp = hv_fetch(PL_utf8_foldclosures,
13626 (char *) foldbuf, foldlen, FALSE)))
13628 AV* list = (AV*) *listp;
13630 for (k = 0; k <= av_len(list); k++) {
13631 SV** c_p = av_fetch(list, k, FALSE);
13634 Perl_croak(aTHX_ "panic: invalid PL_utf8_foldclosures structure");
13638 /* /aa doesn't allow folds between ASCII and non-; /l
13639 * doesn't allow them between above and below 256 */
13640 if ((ASCII_FOLD_RESTRICTED
13641 && (isASCII(c) != isASCII(j)))
13642 || (LOC && c < 256)) {
13646 /* Folds involving non-ascii Latin1 characters
13647 * under /d are added to a separate list */
13648 if (isASCII(c) || c > 255 || AT_LEAST_UNI_SEMANTICS)
13650 cp_list = add_cp_to_invlist(cp_list, c);
13653 depends_list = add_cp_to_invlist(depends_list, c);
13659 SvREFCNT_dec_NN(fold_intersection);
13662 /* And combine the result (if any) with any inversion list from posix
13663 * classes. The lists are kept separate up to now because we don't want to
13664 * fold the classes (folding of those is automatically handled by the swash
13665 * fetching code) */
13667 if (! DEPENDS_SEMANTICS) {
13669 _invlist_union(cp_list, posixes, &cp_list);
13670 SvREFCNT_dec_NN(posixes);
13677 /* Under /d, we put into a separate list the Latin1 things that
13678 * match only when the target string is utf8 */
13679 SV* nonascii_but_latin1_properties = NULL;
13680 _invlist_intersection(posixes, PL_Latin1,
13681 &nonascii_but_latin1_properties);
13682 _invlist_subtract(nonascii_but_latin1_properties, PL_ASCII,
13683 &nonascii_but_latin1_properties);
13684 _invlist_subtract(posixes, nonascii_but_latin1_properties,
13687 _invlist_union(cp_list, posixes, &cp_list);
13688 SvREFCNT_dec_NN(posixes);
13694 if (depends_list) {
13695 _invlist_union(depends_list, nonascii_but_latin1_properties,
13697 SvREFCNT_dec_NN(nonascii_but_latin1_properties);
13700 depends_list = nonascii_but_latin1_properties;
13705 /* And combine the result (if any) with any inversion list from properties.
13706 * The lists are kept separate up to now so that we can distinguish the two
13707 * in regards to matching above-Unicode. A run-time warning is generated
13708 * if a Unicode property is matched against a non-Unicode code point. But,
13709 * we allow user-defined properties to match anything, without any warning,
13710 * and we also suppress the warning if there is a portion of the character
13711 * class that isn't a Unicode property, and which matches above Unicode, \W
13712 * or [\x{110000}] for example.
13713 * (Note that in this case, unlike the Posix one above, there is no
13714 * <depends_list>, because having a Unicode property forces Unicode
13717 bool warn_super = ! has_user_defined_property;
13720 /* If it matters to the final outcome, see if a non-property
13721 * component of the class matches above Unicode. If so, the
13722 * warning gets suppressed. This is true even if just a single
13723 * such code point is specified, as though not strictly correct if
13724 * another such code point is matched against, the fact that they
13725 * are using above-Unicode code points indicates they should know
13726 * the issues involved */
13728 bool non_prop_matches_above_Unicode =
13729 runtime_posix_matches_above_Unicode
13730 | (invlist_highest(cp_list) > PERL_UNICODE_MAX);
13732 non_prop_matches_above_Unicode =
13733 ! non_prop_matches_above_Unicode;
13735 warn_super = ! non_prop_matches_above_Unicode;
13738 _invlist_union(properties, cp_list, &cp_list);
13739 SvREFCNT_dec_NN(properties);
13742 cp_list = properties;
13746 OP(ret) = ANYOF_WARN_SUPER;
13750 /* Here, we have calculated what code points should be in the character
13753 * Now we can see about various optimizations. Fold calculation (which we
13754 * did above) needs to take place before inversion. Otherwise /[^k]/i
13755 * would invert to include K, which under /i would match k, which it
13756 * shouldn't. Therefore we can't invert folded locale now, as it won't be
13757 * folded until runtime */
13759 /* Optimize inverted simple patterns (e.g. [^a-z]) when everything is known
13760 * at compile time. Besides not inverting folded locale now, we can't
13761 * invert if there are things such as \w, which aren't known until runtime
13764 && ! (LOC && (FOLD || (ANYOF_FLAGS(ret) & ANYOF_CLASS)))
13766 && ! HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION)
13768 _invlist_invert(cp_list);
13770 /* Any swash can't be used as-is, because we've inverted things */
13772 SvREFCNT_dec_NN(swash);
13776 /* Clear the invert flag since have just done it here */
13781 *ret_invlist = cp_list;
13782 SvREFCNT_dec(swash);
13784 /* Discard the generated node */
13786 RExC_size = orig_size;
13789 RExC_emit = orig_emit;
13794 /* If we didn't do folding, it's because some information isn't available
13795 * until runtime; set the run-time fold flag for these. (We don't have to
13796 * worry about properties folding, as that is taken care of by the swash
13800 ANYOF_FLAGS(ret) |= ANYOF_LOC_FOLD;
13803 /* Some character classes are equivalent to other nodes. Such nodes take
13804 * up less room and generally fewer operations to execute than ANYOF nodes.
13805 * Above, we checked for and optimized into some such equivalents for
13806 * certain common classes that are easy to test. Getting to this point in
13807 * the code means that the class didn't get optimized there. Since this
13808 * code is only executed in Pass 2, it is too late to save space--it has
13809 * been allocated in Pass 1, and currently isn't given back. But turning
13810 * things into an EXACTish node can allow the optimizer to join it to any
13811 * adjacent such nodes. And if the class is equivalent to things like /./,
13812 * expensive run-time swashes can be avoided. Now that we have more
13813 * complete information, we can find things necessarily missed by the
13814 * earlier code. I (khw) am not sure how much to look for here. It would
13815 * be easy, but perhaps too slow, to check any candidates against all the
13816 * node types they could possibly match using _invlistEQ(). */
13821 && ! (ANYOF_FLAGS(ret) & ANYOF_CLASS)
13822 && ! HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION)
13825 U8 op = END; /* The optimzation node-type */
13826 const char * cur_parse= RExC_parse;
13828 invlist_iterinit(cp_list);
13829 if (! invlist_iternext(cp_list, &start, &end)) {
13831 /* Here, the list is empty. This happens, for example, when a
13832 * Unicode property is the only thing in the character class, and
13833 * it doesn't match anything. (perluniprops.pod notes such
13836 *flagp |= HASWIDTH|SIMPLE;
13838 else if (start == end) { /* The range is a single code point */
13839 if (! invlist_iternext(cp_list, &start, &end)
13841 /* Don't do this optimization if it would require changing
13842 * the pattern to UTF-8 */
13843 && (start < 256 || UTF))
13845 /* Here, the list contains a single code point. Can optimize
13846 * into an EXACT node */
13855 /* A locale node under folding with one code point can be
13856 * an EXACTFL, as its fold won't be calculated until
13862 /* Here, we are generally folding, but there is only one
13863 * code point to match. If we have to, we use an EXACT
13864 * node, but it would be better for joining with adjacent
13865 * nodes in the optimization pass if we used the same
13866 * EXACTFish node that any such are likely to be. We can
13867 * do this iff the code point doesn't participate in any
13868 * folds. For example, an EXACTF of a colon is the same as
13869 * an EXACT one, since nothing folds to or from a colon. */
13871 if (IS_IN_SOME_FOLD_L1(value)) {
13876 if (! PL_utf8_foldable) {
13877 SV* swash = swash_init("utf8", "_Perl_Any_Folds",
13878 &PL_sv_undef, 1, 0);
13879 PL_utf8_foldable = _get_swash_invlist(swash);
13880 SvREFCNT_dec_NN(swash);
13882 if (_invlist_contains_cp(PL_utf8_foldable, value)) {
13887 /* If we haven't found the node type, above, it means we
13888 * can use the prevailing one */
13890 op = compute_EXACTish(pRExC_state);
13895 else if (start == 0) {
13896 if (end == UV_MAX) {
13898 *flagp |= HASWIDTH|SIMPLE;
13901 else if (end == '\n' - 1
13902 && invlist_iternext(cp_list, &start, &end)
13903 && start == '\n' + 1 && end == UV_MAX)
13906 *flagp |= HASWIDTH|SIMPLE;
13910 invlist_iterfinish(cp_list);
13913 RExC_parse = (char *)orig_parse;
13914 RExC_emit = (regnode *)orig_emit;
13916 ret = reg_node(pRExC_state, op);
13918 RExC_parse = (char *)cur_parse;
13920 if (PL_regkind[op] == EXACT) {
13921 alloc_maybe_populate_EXACT(pRExC_state, ret, flagp, 0, value);
13924 SvREFCNT_dec_NN(cp_list);
13929 /* Here, <cp_list> contains all the code points we can determine at
13930 * compile time that match under all conditions. Go through it, and
13931 * for things that belong in the bitmap, put them there, and delete from
13932 * <cp_list>. While we are at it, see if everything above 255 is in the
13933 * list, and if so, set a flag to speed up execution */
13934 ANYOF_BITMAP_ZERO(ret);
13937 /* This gets set if we actually need to modify things */
13938 bool change_invlist = FALSE;
13942 /* Start looking through <cp_list> */
13943 invlist_iterinit(cp_list);
13944 while (invlist_iternext(cp_list, &start, &end)) {
13948 if (end == UV_MAX && start <= 256) {
13949 ANYOF_FLAGS(ret) |= ANYOF_UNICODE_ALL;
13952 /* Quit if are above what we should change */
13957 change_invlist = TRUE;
13959 /* Set all the bits in the range, up to the max that we are doing */
13960 high = (end < 255) ? end : 255;
13961 for (i = start; i <= (int) high; i++) {
13962 if (! ANYOF_BITMAP_TEST(ret, i)) {
13963 ANYOF_BITMAP_SET(ret, i);
13967 invlist_iterfinish(cp_list);
13969 /* Done with loop; remove any code points that are in the bitmap from
13971 if (change_invlist) {
13972 _invlist_subtract(cp_list, PL_Latin1, &cp_list);
13975 /* If have completely emptied it, remove it completely */
13976 if (_invlist_len(cp_list) == 0) {
13977 SvREFCNT_dec_NN(cp_list);
13983 ANYOF_FLAGS(ret) |= ANYOF_INVERT;
13986 /* Here, the bitmap has been populated with all the Latin1 code points that
13987 * always match. Can now add to the overall list those that match only
13988 * when the target string is UTF-8 (<depends_list>). */
13989 if (depends_list) {
13991 _invlist_union(cp_list, depends_list, &cp_list);
13992 SvREFCNT_dec_NN(depends_list);
13995 cp_list = depends_list;
13999 /* If there is a swash and more than one element, we can't use the swash in
14000 * the optimization below. */
14001 if (swash && element_count > 1) {
14002 SvREFCNT_dec_NN(swash);
14007 && ! HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION)
14009 ARG_SET(ret, ANYOF_NONBITMAP_EMPTY);
14012 /* av[0] stores the character class description in its textual form:
14013 * used later (regexec.c:Perl_regclass_swash()) to initialize the
14014 * appropriate swash, and is also useful for dumping the regnode.
14015 * av[1] if NULL, is a placeholder to later contain the swash computed
14016 * from av[0]. But if no further computation need be done, the
14017 * swash is stored there now.
14018 * av[2] stores the cp_list inversion list for use in addition or
14019 * instead of av[0]; used only if av[1] is NULL
14020 * av[3] is set if any component of the class is from a user-defined
14021 * property; used only if av[1] is NULL */
14022 AV * const av = newAV();
14025 av_store(av, 0, (HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION)
14026 ? SvREFCNT_inc(listsv) : &PL_sv_undef);
14028 av_store(av, 1, swash);
14029 SvREFCNT_dec_NN(cp_list);
14032 av_store(av, 1, NULL);
14034 av_store(av, 2, cp_list);
14035 av_store(av, 3, newSVuv(has_user_defined_property));
14039 rv = newRV_noinc(MUTABLE_SV(av));
14040 n = add_data(pRExC_state, 1, "s");
14041 RExC_rxi->data->data[n] = (void*)rv;
14045 *flagp |= HASWIDTH|SIMPLE;
14048 #undef HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION
14051 /* reg_skipcomment()
14053 Absorbs an /x style # comments from the input stream.
14054 Returns true if there is more text remaining in the stream.
14055 Will set the REG_SEEN_RUN_ON_COMMENT flag if the comment
14056 terminates the pattern without including a newline.
14058 Note its the callers responsibility to ensure that we are
14059 actually in /x mode
14064 S_reg_skipcomment(pTHX_ RExC_state_t *pRExC_state)
14068 PERL_ARGS_ASSERT_REG_SKIPCOMMENT;
14070 while (RExC_parse < RExC_end)
14071 if (*RExC_parse++ == '\n') {
14076 /* we ran off the end of the pattern without ending
14077 the comment, so we have to add an \n when wrapping */
14078 RExC_seen |= REG_SEEN_RUN_ON_COMMENT;
14086 Advances the parse position, and optionally absorbs
14087 "whitespace" from the inputstream.
14089 Without /x "whitespace" means (?#...) style comments only,
14090 with /x this means (?#...) and # comments and whitespace proper.
14092 Returns the RExC_parse point from BEFORE the scan occurs.
14094 This is the /x friendly way of saying RExC_parse++.
14098 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
14100 char* const retval = RExC_parse++;
14102 PERL_ARGS_ASSERT_NEXTCHAR;
14105 if (RExC_end - RExC_parse >= 3
14106 && *RExC_parse == '('
14107 && RExC_parse[1] == '?'
14108 && RExC_parse[2] == '#')
14110 while (*RExC_parse != ')') {
14111 if (RExC_parse == RExC_end)
14112 FAIL("Sequence (?#... not terminated");
14118 if (RExC_flags & RXf_PMf_EXTENDED) {
14119 if (isSPACE(*RExC_parse)) {
14123 else if (*RExC_parse == '#') {
14124 if ( reg_skipcomment( pRExC_state ) )
14133 - reg_node - emit a node
14135 STATIC regnode * /* Location. */
14136 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
14140 regnode * const ret = RExC_emit;
14141 GET_RE_DEBUG_FLAGS_DECL;
14143 PERL_ARGS_ASSERT_REG_NODE;
14146 SIZE_ALIGN(RExC_size);
14150 if (RExC_emit >= RExC_emit_bound)
14151 Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d, %p>=%p",
14152 op, RExC_emit, RExC_emit_bound);
14154 NODE_ALIGN_FILL(ret);
14156 FILL_ADVANCE_NODE(ptr, op);
14157 REH_CALL_COMP_NODE_HOOK(pRExC_state->rx, (ptr) - 1);
14158 #ifdef RE_TRACK_PATTERN_OFFSETS
14159 if (RExC_offsets) { /* MJD */
14160 MJD_OFFSET_DEBUG(("%s:%d: (op %s) %s %"UVuf" (len %"UVuf") (max %"UVuf").\n",
14161 "reg_node", __LINE__,
14163 (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0]
14164 ? "Overwriting end of array!\n" : "OK",
14165 (UV)(RExC_emit - RExC_emit_start),
14166 (UV)(RExC_parse - RExC_start),
14167 (UV)RExC_offsets[0]));
14168 Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
14176 - reganode - emit a node with an argument
14178 STATIC regnode * /* Location. */
14179 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
14183 regnode * const ret = RExC_emit;
14184 GET_RE_DEBUG_FLAGS_DECL;
14186 PERL_ARGS_ASSERT_REGANODE;
14189 SIZE_ALIGN(RExC_size);
14194 assert(2==regarglen[op]+1);
14196 Anything larger than this has to allocate the extra amount.
14197 If we changed this to be:
14199 RExC_size += (1 + regarglen[op]);
14201 then it wouldn't matter. Its not clear what side effect
14202 might come from that so its not done so far.
14207 if (RExC_emit >= RExC_emit_bound)
14208 Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d, %p>=%p",
14209 op, RExC_emit, RExC_emit_bound);
14211 NODE_ALIGN_FILL(ret);
14213 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
14214 REH_CALL_COMP_NODE_HOOK(pRExC_state->rx, (ptr) - 2);
14215 #ifdef RE_TRACK_PATTERN_OFFSETS
14216 if (RExC_offsets) { /* MJD */
14217 MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n",
14221 (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0] ?
14222 "Overwriting end of array!\n" : "OK",
14223 (UV)(RExC_emit - RExC_emit_start),
14224 (UV)(RExC_parse - RExC_start),
14225 (UV)RExC_offsets[0]));
14226 Set_Cur_Node_Offset;
14234 - reguni - emit (if appropriate) a Unicode character
14237 S_reguni(pTHX_ const RExC_state_t *pRExC_state, UV uv, char* s)
14241 PERL_ARGS_ASSERT_REGUNI;
14243 return SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
14247 - reginsert - insert an operator in front of already-emitted operand
14249 * Means relocating the operand.
14252 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd, U32 depth)
14258 const int offset = regarglen[(U8)op];
14259 const int size = NODE_STEP_REGNODE + offset;
14260 GET_RE_DEBUG_FLAGS_DECL;
14262 PERL_ARGS_ASSERT_REGINSERT;
14263 PERL_UNUSED_ARG(depth);
14264 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
14265 DEBUG_PARSE_FMT("inst"," - %s",PL_reg_name[op]);
14274 if (RExC_open_parens) {
14276 /*DEBUG_PARSE_FMT("inst"," - %"IVdf, (IV)RExC_npar);*/
14277 for ( paren=0 ; paren < RExC_npar ; paren++ ) {
14278 if ( RExC_open_parens[paren] >= opnd ) {
14279 /*DEBUG_PARSE_FMT("open"," - %d",size);*/
14280 RExC_open_parens[paren] += size;
14282 /*DEBUG_PARSE_FMT("open"," - %s","ok");*/
14284 if ( RExC_close_parens[paren] >= opnd ) {
14285 /*DEBUG_PARSE_FMT("close"," - %d",size);*/
14286 RExC_close_parens[paren] += size;
14288 /*DEBUG_PARSE_FMT("close"," - %s","ok");*/
14293 while (src > opnd) {
14294 StructCopy(--src, --dst, regnode);
14295 #ifdef RE_TRACK_PATTERN_OFFSETS
14296 if (RExC_offsets) { /* MJD 20010112 */
14297 MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s copy %"UVuf" -> %"UVuf" (max %"UVuf").\n",
14301 (UV)(dst - RExC_emit_start) > RExC_offsets[0]
14302 ? "Overwriting end of array!\n" : "OK",
14303 (UV)(src - RExC_emit_start),
14304 (UV)(dst - RExC_emit_start),
14305 (UV)RExC_offsets[0]));
14306 Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
14307 Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
14313 place = opnd; /* Op node, where operand used to be. */
14314 #ifdef RE_TRACK_PATTERN_OFFSETS
14315 if (RExC_offsets) { /* MJD */
14316 MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n",
14320 (UV)(place - RExC_emit_start) > RExC_offsets[0]
14321 ? "Overwriting end of array!\n" : "OK",
14322 (UV)(place - RExC_emit_start),
14323 (UV)(RExC_parse - RExC_start),
14324 (UV)RExC_offsets[0]));
14325 Set_Node_Offset(place, RExC_parse);
14326 Set_Node_Length(place, 1);
14329 src = NEXTOPER(place);
14330 FILL_ADVANCE_NODE(place, op);
14331 REH_CALL_COMP_NODE_HOOK(pRExC_state->rx, (place) - 1);
14332 Zero(src, offset, regnode);
14336 - regtail - set the next-pointer at the end of a node chain of p to val.
14337 - SEE ALSO: regtail_study
14339 /* TODO: All three parms should be const */
14341 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,U32 depth)
14345 GET_RE_DEBUG_FLAGS_DECL;
14347 PERL_ARGS_ASSERT_REGTAIL;
14349 PERL_UNUSED_ARG(depth);
14355 /* Find last node. */
14358 regnode * const temp = regnext(scan);
14360 SV * const mysv=sv_newmortal();
14361 DEBUG_PARSE_MSG((scan==p ? "tail" : ""));
14362 regprop(RExC_rx, mysv, scan);
14363 PerlIO_printf(Perl_debug_log, "~ %s (%d) %s %s\n",
14364 SvPV_nolen_const(mysv), REG_NODE_NUM(scan),
14365 (temp == NULL ? "->" : ""),
14366 (temp == NULL ? PL_reg_name[OP(val)] : "")
14374 if (reg_off_by_arg[OP(scan)]) {
14375 ARG_SET(scan, val - scan);
14378 NEXT_OFF(scan) = val - scan;
14384 - regtail_study - set the next-pointer at the end of a node chain of p to val.
14385 - Look for optimizable sequences at the same time.
14386 - currently only looks for EXACT chains.
14388 This is experimental code. The idea is to use this routine to perform
14389 in place optimizations on branches and groups as they are constructed,
14390 with the long term intention of removing optimization from study_chunk so
14391 that it is purely analytical.
14393 Currently only used when in DEBUG mode. The macro REGTAIL_STUDY() is used
14394 to control which is which.
14397 /* TODO: All four parms should be const */
14400 S_regtail_study(pTHX_ RExC_state_t *pRExC_state, regnode *p, const regnode *val,U32 depth)
14405 #ifdef EXPERIMENTAL_INPLACESCAN
14408 GET_RE_DEBUG_FLAGS_DECL;
14410 PERL_ARGS_ASSERT_REGTAIL_STUDY;
14416 /* Find last node. */
14420 regnode * const temp = regnext(scan);
14421 #ifdef EXPERIMENTAL_INPLACESCAN
14422 if (PL_regkind[OP(scan)] == EXACT) {
14423 bool has_exactf_sharp_s; /* Unexamined in this routine */
14424 if (join_exact(pRExC_state,scan,&min, &has_exactf_sharp_s, 1,val,depth+1))
14429 switch (OP(scan)) {
14435 case EXACTFU_TRICKYFOLD:
14437 if( exact == PSEUDO )
14439 else if ( exact != OP(scan) )
14448 SV * const mysv=sv_newmortal();
14449 DEBUG_PARSE_MSG((scan==p ? "tsdy" : ""));
14450 regprop(RExC_rx, mysv, scan);
14451 PerlIO_printf(Perl_debug_log, "~ %s (%d) -> %s\n",
14452 SvPV_nolen_const(mysv),
14453 REG_NODE_NUM(scan),
14454 PL_reg_name[exact]);
14461 SV * const mysv_val=sv_newmortal();
14462 DEBUG_PARSE_MSG("");
14463 regprop(RExC_rx, mysv_val, val);
14464 PerlIO_printf(Perl_debug_log, "~ attach to %s (%"IVdf") offset to %"IVdf"\n",
14465 SvPV_nolen_const(mysv_val),
14466 (IV)REG_NODE_NUM(val),
14470 if (reg_off_by_arg[OP(scan)]) {
14471 ARG_SET(scan, val - scan);
14474 NEXT_OFF(scan) = val - scan;
14482 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
14487 S_regdump_intflags(pTHX_ const char *lead, const U32 flags)
14492 for (bit=0; bit<32; bit++) {
14493 if (flags & (1<<bit)) {
14494 if (!set++ && lead)
14495 PerlIO_printf(Perl_debug_log, "%s",lead);
14496 PerlIO_printf(Perl_debug_log, "%s ",PL_reg_intflags_name[bit]);
14501 PerlIO_printf(Perl_debug_log, "\n");
14503 PerlIO_printf(Perl_debug_log, "%s[none-set]\n",lead);
14508 S_regdump_extflags(pTHX_ const char *lead, const U32 flags)
14514 for (bit=0; bit<32; bit++) {
14515 if (flags & (1<<bit)) {
14516 if ((1<<bit) & RXf_PMf_CHARSET) { /* Output separately, below */
14519 if (!set++ && lead)
14520 PerlIO_printf(Perl_debug_log, "%s",lead);
14521 PerlIO_printf(Perl_debug_log, "%s ",PL_reg_extflags_name[bit]);
14524 if ((cs = get_regex_charset(flags)) != REGEX_DEPENDS_CHARSET) {
14525 if (!set++ && lead) {
14526 PerlIO_printf(Perl_debug_log, "%s",lead);
14529 case REGEX_UNICODE_CHARSET:
14530 PerlIO_printf(Perl_debug_log, "UNICODE");
14532 case REGEX_LOCALE_CHARSET:
14533 PerlIO_printf(Perl_debug_log, "LOCALE");
14535 case REGEX_ASCII_RESTRICTED_CHARSET:
14536 PerlIO_printf(Perl_debug_log, "ASCII-RESTRICTED");
14538 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
14539 PerlIO_printf(Perl_debug_log, "ASCII-MORE_RESTRICTED");
14542 PerlIO_printf(Perl_debug_log, "UNKNOWN CHARACTER SET");
14548 PerlIO_printf(Perl_debug_log, "\n");
14550 PerlIO_printf(Perl_debug_log, "%s[none-set]\n",lead);
14556 Perl_regdump(pTHX_ const regexp *r)
14560 SV * const sv = sv_newmortal();
14561 SV *dsv= sv_newmortal();
14562 RXi_GET_DECL(r,ri);
14563 GET_RE_DEBUG_FLAGS_DECL;
14565 PERL_ARGS_ASSERT_REGDUMP;
14567 (void)dumpuntil(r, ri->program, ri->program + 1, NULL, NULL, sv, 0, 0);
14569 /* Header fields of interest. */
14570 if (r->anchored_substr) {
14571 RE_PV_QUOTED_DECL(s, 0, dsv, SvPVX_const(r->anchored_substr),
14572 RE_SV_DUMPLEN(r->anchored_substr), 30);
14573 PerlIO_printf(Perl_debug_log,
14574 "anchored %s%s at %"IVdf" ",
14575 s, RE_SV_TAIL(r->anchored_substr),
14576 (IV)r->anchored_offset);
14577 } else if (r->anchored_utf8) {
14578 RE_PV_QUOTED_DECL(s, 1, dsv, SvPVX_const(r->anchored_utf8),
14579 RE_SV_DUMPLEN(r->anchored_utf8), 30);
14580 PerlIO_printf(Perl_debug_log,
14581 "anchored utf8 %s%s at %"IVdf" ",
14582 s, RE_SV_TAIL(r->anchored_utf8),
14583 (IV)r->anchored_offset);
14585 if (r->float_substr) {
14586 RE_PV_QUOTED_DECL(s, 0, dsv, SvPVX_const(r->float_substr),
14587 RE_SV_DUMPLEN(r->float_substr), 30);
14588 PerlIO_printf(Perl_debug_log,
14589 "floating %s%s at %"IVdf"..%"UVuf" ",
14590 s, RE_SV_TAIL(r->float_substr),
14591 (IV)r->float_min_offset, (UV)r->float_max_offset);
14592 } else if (r->float_utf8) {
14593 RE_PV_QUOTED_DECL(s, 1, dsv, SvPVX_const(r->float_utf8),
14594 RE_SV_DUMPLEN(r->float_utf8), 30);
14595 PerlIO_printf(Perl_debug_log,
14596 "floating utf8 %s%s at %"IVdf"..%"UVuf" ",
14597 s, RE_SV_TAIL(r->float_utf8),
14598 (IV)r->float_min_offset, (UV)r->float_max_offset);
14600 if (r->check_substr || r->check_utf8)
14601 PerlIO_printf(Perl_debug_log,
14603 (r->check_substr == r->float_substr
14604 && r->check_utf8 == r->float_utf8
14605 ? "(checking floating" : "(checking anchored"));
14606 if (r->extflags & RXf_NOSCAN)
14607 PerlIO_printf(Perl_debug_log, " noscan");
14608 if (r->extflags & RXf_CHECK_ALL)
14609 PerlIO_printf(Perl_debug_log, " isall");
14610 if (r->check_substr || r->check_utf8)
14611 PerlIO_printf(Perl_debug_log, ") ");
14613 if (ri->regstclass) {
14614 regprop(r, sv, ri->regstclass);
14615 PerlIO_printf(Perl_debug_log, "stclass %s ", SvPVX_const(sv));
14617 if (r->extflags & RXf_ANCH) {
14618 PerlIO_printf(Perl_debug_log, "anchored");
14619 if (r->extflags & RXf_ANCH_BOL)
14620 PerlIO_printf(Perl_debug_log, "(BOL)");
14621 if (r->extflags & RXf_ANCH_MBOL)
14622 PerlIO_printf(Perl_debug_log, "(MBOL)");
14623 if (r->extflags & RXf_ANCH_SBOL)
14624 PerlIO_printf(Perl_debug_log, "(SBOL)");
14625 if (r->extflags & RXf_ANCH_GPOS)
14626 PerlIO_printf(Perl_debug_log, "(GPOS)");
14627 PerlIO_putc(Perl_debug_log, ' ');
14629 if (r->extflags & RXf_GPOS_SEEN)
14630 PerlIO_printf(Perl_debug_log, "GPOS:%"UVuf" ", (UV)r->gofs);
14631 if (r->intflags & PREGf_SKIP)
14632 PerlIO_printf(Perl_debug_log, "plus ");
14633 if (r->intflags & PREGf_IMPLICIT)
14634 PerlIO_printf(Perl_debug_log, "implicit ");
14635 PerlIO_printf(Perl_debug_log, "minlen %"IVdf" ", (IV)r->minlen);
14636 if (r->extflags & RXf_EVAL_SEEN)
14637 PerlIO_printf(Perl_debug_log, "with eval ");
14638 PerlIO_printf(Perl_debug_log, "\n");
14640 regdump_extflags("r->extflags: ",r->extflags);
14641 regdump_intflags("r->intflags: ",r->intflags);
14644 PERL_ARGS_ASSERT_REGDUMP;
14645 PERL_UNUSED_CONTEXT;
14646 PERL_UNUSED_ARG(r);
14647 #endif /* DEBUGGING */
14651 - regprop - printable representation of opcode
14653 #define EMIT_ANYOF_TEST_SEPARATOR(do_sep,sv,flags) \
14656 Perl_sv_catpvf(aTHX_ sv,"%s][%s",PL_colors[1],PL_colors[0]); \
14657 if (flags & ANYOF_INVERT) \
14658 /*make sure the invert info is in each */ \
14659 sv_catpvs(sv, "^"); \
14665 Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o)
14671 /* Should be synchronized with * ANYOF_ #xdefines in regcomp.h */
14672 static const char * const anyofs[] = {
14673 #if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
14674 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
14675 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9 \
14676 || _CC_SPACE != 10 || _CC_BLANK != 11 || _CC_XDIGIT != 12 \
14677 || _CC_PSXSPC != 13 || _CC_CNTRL != 14 || _CC_ASCII != 15 \
14678 || _CC_VERTSPACE != 16
14679 #error Need to adjust order of anyofs[]
14716 RXi_GET_DECL(prog,progi);
14717 GET_RE_DEBUG_FLAGS_DECL;
14719 PERL_ARGS_ASSERT_REGPROP;
14723 if (OP(o) > REGNODE_MAX) /* regnode.type is unsigned */
14724 /* It would be nice to FAIL() here, but this may be called from
14725 regexec.c, and it would be hard to supply pRExC_state. */
14726 Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d", (int)OP(o), (int)REGNODE_MAX);
14727 sv_catpv(sv, PL_reg_name[OP(o)]); /* Take off const! */
14729 k = PL_regkind[OP(o)];
14732 sv_catpvs(sv, " ");
14733 /* Using is_utf8_string() (via PERL_PV_UNI_DETECT)
14734 * is a crude hack but it may be the best for now since
14735 * we have no flag "this EXACTish node was UTF-8"
14737 pv_pretty(sv, STRING(o), STR_LEN(o), 60, PL_colors[0], PL_colors[1],
14738 PERL_PV_ESCAPE_UNI_DETECT |
14739 PERL_PV_ESCAPE_NONASCII |
14740 PERL_PV_PRETTY_ELLIPSES |
14741 PERL_PV_PRETTY_LTGT |
14742 PERL_PV_PRETTY_NOCLEAR
14744 } else if (k == TRIE) {
14745 /* print the details of the trie in dumpuntil instead, as
14746 * progi->data isn't available here */
14747 const char op = OP(o);
14748 const U32 n = ARG(o);
14749 const reg_ac_data * const ac = IS_TRIE_AC(op) ?
14750 (reg_ac_data *)progi->data->data[n] :
14752 const reg_trie_data * const trie
14753 = (reg_trie_data*)progi->data->data[!IS_TRIE_AC(op) ? n : ac->trie];
14755 Perl_sv_catpvf(aTHX_ sv, "-%s",PL_reg_name[o->flags]);
14756 DEBUG_TRIE_COMPILE_r(
14757 Perl_sv_catpvf(aTHX_ sv,
14758 "<S:%"UVuf"/%"IVdf" W:%"UVuf" L:%"UVuf"/%"UVuf" C:%"UVuf"/%"UVuf">",
14759 (UV)trie->startstate,
14760 (IV)trie->statecount-1, /* -1 because of the unused 0 element */
14761 (UV)trie->wordcount,
14764 (UV)TRIE_CHARCOUNT(trie),
14765 (UV)trie->uniquecharcount
14768 if ( IS_ANYOF_TRIE(op) || trie->bitmap ) {
14769 sv_catpvs(sv, "[");
14770 (void) put_latin1_charclass_innards(sv, IS_ANYOF_TRIE(op)
14772 : TRIE_BITMAP(trie));
14773 sv_catpvs(sv, "]");
14776 } else if (k == CURLY) {
14777 if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
14778 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
14779 Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
14781 else if (k == WHILEM && o->flags) /* Ordinal/of */
14782 Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
14783 else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP || OP(o)==ACCEPT) {
14784 Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o)); /* Parenth number */
14785 if ( RXp_PAREN_NAMES(prog) ) {
14786 if ( k != REF || (OP(o) < NREF)) {
14787 AV *list= MUTABLE_AV(progi->data->data[progi->name_list_idx]);
14788 SV **name= av_fetch(list, ARG(o), 0 );
14790 Perl_sv_catpvf(aTHX_ sv, " '%"SVf"'", SVfARG(*name));
14793 AV *list= MUTABLE_AV(progi->data->data[ progi->name_list_idx ]);
14794 SV *sv_dat= MUTABLE_SV(progi->data->data[ ARG( o ) ]);
14795 I32 *nums=(I32*)SvPVX(sv_dat);
14796 SV **name= av_fetch(list, nums[0], 0 );
14799 for ( n=0; n<SvIVX(sv_dat); n++ ) {
14800 Perl_sv_catpvf(aTHX_ sv, "%s%"IVdf,
14801 (n ? "," : ""), (IV)nums[n]);
14803 Perl_sv_catpvf(aTHX_ sv, " '%"SVf"'", SVfARG(*name));
14807 } else if (k == GOSUB)
14808 Perl_sv_catpvf(aTHX_ sv, "%d[%+d]", (int)ARG(o),(int)ARG2L(o)); /* Paren and offset */
14809 else if (k == VERB) {
14811 Perl_sv_catpvf(aTHX_ sv, ":%"SVf,
14812 SVfARG((MUTABLE_SV(progi->data->data[ ARG( o ) ]))));
14813 } else if (k == LOGICAL)
14814 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* 2: embedded, otherwise 1 */
14815 else if (k == ANYOF) {
14816 const U8 flags = ANYOF_FLAGS(o);
14820 if (flags & ANYOF_LOCALE)
14821 sv_catpvs(sv, "{loc}");
14822 if (flags & ANYOF_LOC_FOLD)
14823 sv_catpvs(sv, "{i}");
14824 Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
14825 if (flags & ANYOF_INVERT)
14826 sv_catpvs(sv, "^");
14828 /* output what the standard cp 0-255 bitmap matches */
14829 do_sep = put_latin1_charclass_innards(sv, ANYOF_BITMAP(o));
14831 EMIT_ANYOF_TEST_SEPARATOR(do_sep,sv,flags);
14832 /* output any special charclass tests (used entirely under use locale) */
14833 if (ANYOF_CLASS_TEST_ANY_SET(o)) {
14835 for (i = 0; i < (int)(sizeof(anyofs)/sizeof(char*)); i++) {
14836 if (ANYOF_CLASS_TEST(o,i)) {
14837 sv_catpv(sv, anyofs[i]);
14843 EMIT_ANYOF_TEST_SEPARATOR(do_sep,sv,flags);
14845 if (flags & ANYOF_NON_UTF8_LATIN1_ALL) {
14846 sv_catpvs(sv, "{non-utf8-latin1-all}");
14849 /* output information about the unicode matching */
14850 if (flags & ANYOF_UNICODE_ALL)
14851 sv_catpvs(sv, "{unicode_all}");
14852 else if (ANYOF_NONBITMAP(o)) {
14853 SV *lv; /* Set if there is something outside the bit map. */
14855 bool byte_output = FALSE; /* If something in the bitmap has been
14858 if (flags & ANYOF_NONBITMAP_NON_UTF8) {
14859 sv_catpvs(sv, "{outside bitmap}");
14862 sv_catpvs(sv, "{utf8}");
14865 /* Get the stuff that wasn't in the bitmap */
14866 sw = regclass_swash(prog, o, FALSE, &lv, NULL);
14867 if (lv && lv != &PL_sv_undef) {
14868 char *s = savesvpv(lv);
14869 char * const origs = s;
14871 while (*s && *s != '\n')
14875 const char * const t = ++s;
14878 sv_catpvs(sv, " ");
14884 /* Truncate very long output */
14885 if (s - origs > 256) {
14886 Perl_sv_catpvf(aTHX_ sv,
14888 (int) (s - origs - 1),
14894 else if (*s == '\t') {
14908 SvREFCNT_dec_NN(lv);
14912 Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
14914 else if (k == POSIXD || k == NPOSIXD) {
14915 U8 index = FLAGS(o) * 2;
14916 if (index > (sizeof(anyofs) / sizeof(anyofs[0]))) {
14917 Perl_sv_catpvf(aTHX_ sv, "[illegal type=%d])", index);
14920 sv_catpv(sv, anyofs[index]);
14923 else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
14924 Perl_sv_catpvf(aTHX_ sv, "[%d]", -(o->flags));
14926 PERL_UNUSED_CONTEXT;
14927 PERL_UNUSED_ARG(sv);
14928 PERL_UNUSED_ARG(o);
14929 PERL_UNUSED_ARG(prog);
14930 #endif /* DEBUGGING */
14934 Perl_re_intuit_string(pTHX_ REGEXP * const r)
14935 { /* Assume that RE_INTUIT is set */
14937 struct regexp *const prog = ReANY(r);
14938 GET_RE_DEBUG_FLAGS_DECL;
14940 PERL_ARGS_ASSERT_RE_INTUIT_STRING;
14941 PERL_UNUSED_CONTEXT;
14945 const char * const s = SvPV_nolen_const(prog->check_substr
14946 ? prog->check_substr : prog->check_utf8);
14948 if (!PL_colorset) reginitcolors();
14949 PerlIO_printf(Perl_debug_log,
14950 "%sUsing REx %ssubstr:%s \"%s%.60s%s%s\"\n",
14952 prog->check_substr ? "" : "utf8 ",
14953 PL_colors[5],PL_colors[0],
14956 (strlen(s) > 60 ? "..." : ""));
14959 return prog->check_substr ? prog->check_substr : prog->check_utf8;
14965 handles refcounting and freeing the perl core regexp structure. When
14966 it is necessary to actually free the structure the first thing it
14967 does is call the 'free' method of the regexp_engine associated to
14968 the regexp, allowing the handling of the void *pprivate; member
14969 first. (This routine is not overridable by extensions, which is why
14970 the extensions free is called first.)
14972 See regdupe and regdupe_internal if you change anything here.
14974 #ifndef PERL_IN_XSUB_RE
14976 Perl_pregfree(pTHX_ REGEXP *r)
14982 Perl_pregfree2(pTHX_ REGEXP *rx)
14985 struct regexp *const r = ReANY(rx);
14986 GET_RE_DEBUG_FLAGS_DECL;
14988 PERL_ARGS_ASSERT_PREGFREE2;
14990 if (r->mother_re) {
14991 ReREFCNT_dec(r->mother_re);
14993 CALLREGFREE_PVT(rx); /* free the private data */
14994 SvREFCNT_dec(RXp_PAREN_NAMES(r));
14995 Safefree(r->xpv_len_u.xpvlenu_pv);
14998 SvREFCNT_dec(r->anchored_substr);
14999 SvREFCNT_dec(r->anchored_utf8);
15000 SvREFCNT_dec(r->float_substr);
15001 SvREFCNT_dec(r->float_utf8);
15002 Safefree(r->substrs);
15004 RX_MATCH_COPY_FREE(rx);
15005 #ifdef PERL_ANY_COW
15006 SvREFCNT_dec(r->saved_copy);
15009 SvREFCNT_dec(r->qr_anoncv);
15010 rx->sv_u.svu_rx = 0;
15015 This is a hacky workaround to the structural issue of match results
15016 being stored in the regexp structure which is in turn stored in
15017 PL_curpm/PL_reg_curpm. The problem is that due to qr// the pattern
15018 could be PL_curpm in multiple contexts, and could require multiple
15019 result sets being associated with the pattern simultaneously, such
15020 as when doing a recursive match with (??{$qr})
15022 The solution is to make a lightweight copy of the regexp structure
15023 when a qr// is returned from the code executed by (??{$qr}) this
15024 lightweight copy doesn't actually own any of its data except for
15025 the starp/end and the actual regexp structure itself.
15031 Perl_reg_temp_copy (pTHX_ REGEXP *ret_x, REGEXP *rx)
15033 struct regexp *ret;
15034 struct regexp *const r = ReANY(rx);
15035 const bool islv = ret_x && SvTYPE(ret_x) == SVt_PVLV;
15037 PERL_ARGS_ASSERT_REG_TEMP_COPY;
15040 ret_x = (REGEXP*) newSV_type(SVt_REGEXP);
15042 SvOK_off((SV *)ret_x);
15044 /* For PVLVs, SvANY points to the xpvlv body while sv_u points
15045 to the regexp. (For SVt_REGEXPs, sv_upgrade has already
15046 made both spots point to the same regexp body.) */
15047 REGEXP *temp = (REGEXP *)newSV_type(SVt_REGEXP);
15048 assert(!SvPVX(ret_x));
15049 ret_x->sv_u.svu_rx = temp->sv_any;
15050 temp->sv_any = NULL;
15051 SvFLAGS(temp) = (SvFLAGS(temp) & ~SVTYPEMASK) | SVt_NULL;
15052 SvREFCNT_dec_NN(temp);
15053 /* SvCUR still resides in the xpvlv struct, so the regexp copy-
15054 ing below will not set it. */
15055 SvCUR_set(ret_x, SvCUR(rx));
15058 /* This ensures that SvTHINKFIRST(sv) is true, and hence that
15059 sv_force_normal(sv) is called. */
15061 ret = ReANY(ret_x);
15063 SvFLAGS(ret_x) |= SvUTF8(rx);
15064 /* We share the same string buffer as the original regexp, on which we
15065 hold a reference count, incremented when mother_re is set below.
15066 The string pointer is copied here, being part of the regexp struct.
15068 memcpy(&(ret->xpv_cur), &(r->xpv_cur),
15069 sizeof(regexp) - STRUCT_OFFSET(regexp, xpv_cur));
15071 const I32 npar = r->nparens+1;
15072 Newx(ret->offs, npar, regexp_paren_pair);
15073 Copy(r->offs, ret->offs, npar, regexp_paren_pair);
15076 Newx(ret->substrs, 1, struct reg_substr_data);
15077 StructCopy(r->substrs, ret->substrs, struct reg_substr_data);
15079 SvREFCNT_inc_void(ret->anchored_substr);
15080 SvREFCNT_inc_void(ret->anchored_utf8);
15081 SvREFCNT_inc_void(ret->float_substr);
15082 SvREFCNT_inc_void(ret->float_utf8);
15084 /* check_substr and check_utf8, if non-NULL, point to either their
15085 anchored or float namesakes, and don't hold a second reference. */
15087 RX_MATCH_COPIED_off(ret_x);
15088 #ifdef PERL_ANY_COW
15089 ret->saved_copy = NULL;
15091 ret->mother_re = ReREFCNT_inc(r->mother_re ? r->mother_re : rx);
15092 SvREFCNT_inc_void(ret->qr_anoncv);
15098 /* regfree_internal()
15100 Free the private data in a regexp. This is overloadable by
15101 extensions. Perl takes care of the regexp structure in pregfree(),
15102 this covers the *pprivate pointer which technically perl doesn't
15103 know about, however of course we have to handle the
15104 regexp_internal structure when no extension is in use.
15106 Note this is called before freeing anything in the regexp
15111 Perl_regfree_internal(pTHX_ REGEXP * const rx)
15114 struct regexp *const r = ReANY(rx);
15115 RXi_GET_DECL(r,ri);
15116 GET_RE_DEBUG_FLAGS_DECL;
15118 PERL_ARGS_ASSERT_REGFREE_INTERNAL;
15124 SV *dsv= sv_newmortal();
15125 RE_PV_QUOTED_DECL(s, RX_UTF8(rx),
15126 dsv, RX_PRECOMP(rx), RX_PRELEN(rx), 60);
15127 PerlIO_printf(Perl_debug_log,"%sFreeing REx:%s %s\n",
15128 PL_colors[4],PL_colors[5],s);
15131 #ifdef RE_TRACK_PATTERN_OFFSETS
15133 Safefree(ri->u.offsets); /* 20010421 MJD */
15135 if (ri->code_blocks) {
15137 for (n = 0; n < ri->num_code_blocks; n++)
15138 SvREFCNT_dec(ri->code_blocks[n].src_regex);
15139 Safefree(ri->code_blocks);
15143 int n = ri->data->count;
15146 /* If you add a ->what type here, update the comment in regcomp.h */
15147 switch (ri->data->what[n]) {
15153 SvREFCNT_dec(MUTABLE_SV(ri->data->data[n]));
15156 Safefree(ri->data->data[n]);
15162 { /* Aho Corasick add-on structure for a trie node.
15163 Used in stclass optimization only */
15165 reg_ac_data *aho=(reg_ac_data*)ri->data->data[n];
15167 refcount = --aho->refcount;
15170 PerlMemShared_free(aho->states);
15171 PerlMemShared_free(aho->fail);
15172 /* do this last!!!! */
15173 PerlMemShared_free(ri->data->data[n]);
15174 PerlMemShared_free(ri->regstclass);
15180 /* trie structure. */
15182 reg_trie_data *trie=(reg_trie_data*)ri->data->data[n];
15184 refcount = --trie->refcount;
15187 PerlMemShared_free(trie->charmap);
15188 PerlMemShared_free(trie->states);
15189 PerlMemShared_free(trie->trans);
15191 PerlMemShared_free(trie->bitmap);
15193 PerlMemShared_free(trie->jump);
15194 PerlMemShared_free(trie->wordinfo);
15195 /* do this last!!!! */
15196 PerlMemShared_free(ri->data->data[n]);
15201 Perl_croak(aTHX_ "panic: regfree data code '%c'", ri->data->what[n]);
15204 Safefree(ri->data->what);
15205 Safefree(ri->data);
15211 #define av_dup_inc(s,t) MUTABLE_AV(sv_dup_inc((const SV *)s,t))
15212 #define hv_dup_inc(s,t) MUTABLE_HV(sv_dup_inc((const SV *)s,t))
15213 #define SAVEPVN(p,n) ((p) ? savepvn(p,n) : NULL)
15216 re_dup - duplicate a regexp.
15218 This routine is expected to clone a given regexp structure. It is only
15219 compiled under USE_ITHREADS.
15221 After all of the core data stored in struct regexp is duplicated
15222 the regexp_engine.dupe method is used to copy any private data
15223 stored in the *pprivate pointer. This allows extensions to handle
15224 any duplication it needs to do.
15226 See pregfree() and regfree_internal() if you change anything here.
15228 #if defined(USE_ITHREADS)
15229 #ifndef PERL_IN_XSUB_RE
15231 Perl_re_dup_guts(pTHX_ const REGEXP *sstr, REGEXP *dstr, CLONE_PARAMS *param)
15235 const struct regexp *r = ReANY(sstr);
15236 struct regexp *ret = ReANY(dstr);
15238 PERL_ARGS_ASSERT_RE_DUP_GUTS;
15240 npar = r->nparens+1;
15241 Newx(ret->offs, npar, regexp_paren_pair);
15242 Copy(r->offs, ret->offs, npar, regexp_paren_pair);
15244 if (ret->substrs) {
15245 /* Do it this way to avoid reading from *r after the StructCopy().
15246 That way, if any of the sv_dup_inc()s dislodge *r from the L1
15247 cache, it doesn't matter. */
15248 const bool anchored = r->check_substr
15249 ? r->check_substr == r->anchored_substr
15250 : r->check_utf8 == r->anchored_utf8;
15251 Newx(ret->substrs, 1, struct reg_substr_data);
15252 StructCopy(r->substrs, ret->substrs, struct reg_substr_data);
15254 ret->anchored_substr = sv_dup_inc(ret->anchored_substr, param);
15255 ret->anchored_utf8 = sv_dup_inc(ret->anchored_utf8, param);
15256 ret->float_substr = sv_dup_inc(ret->float_substr, param);
15257 ret->float_utf8 = sv_dup_inc(ret->float_utf8, param);
15259 /* check_substr and check_utf8, if non-NULL, point to either their
15260 anchored or float namesakes, and don't hold a second reference. */
15262 if (ret->check_substr) {
15264 assert(r->check_utf8 == r->anchored_utf8);
15265 ret->check_substr = ret->anchored_substr;
15266 ret->check_utf8 = ret->anchored_utf8;
15268 assert(r->check_substr == r->float_substr);
15269 assert(r->check_utf8 == r->float_utf8);
15270 ret->check_substr = ret->float_substr;
15271 ret->check_utf8 = ret->float_utf8;
15273 } else if (ret->check_utf8) {
15275 ret->check_utf8 = ret->anchored_utf8;
15277 ret->check_utf8 = ret->float_utf8;
15282 RXp_PAREN_NAMES(ret) = hv_dup_inc(RXp_PAREN_NAMES(ret), param);
15283 ret->qr_anoncv = MUTABLE_CV(sv_dup_inc((const SV *)ret->qr_anoncv, param));
15286 RXi_SET(ret,CALLREGDUPE_PVT(dstr,param));
15288 if (RX_MATCH_COPIED(dstr))
15289 ret->subbeg = SAVEPVN(ret->subbeg, ret->sublen);
15291 ret->subbeg = NULL;
15292 #ifdef PERL_ANY_COW
15293 ret->saved_copy = NULL;
15296 /* Whether mother_re be set or no, we need to copy the string. We
15297 cannot refrain from copying it when the storage points directly to
15298 our mother regexp, because that's
15299 1: a buffer in a different thread
15300 2: something we no longer hold a reference on
15301 so we need to copy it locally. */
15302 RX_WRAPPED(dstr) = SAVEPVN(RX_WRAPPED(sstr), SvCUR(sstr)+1);
15303 ret->mother_re = NULL;
15305 #endif /* PERL_IN_XSUB_RE */
15310 This is the internal complement to regdupe() which is used to copy
15311 the structure pointed to by the *pprivate pointer in the regexp.
15312 This is the core version of the extension overridable cloning hook.
15313 The regexp structure being duplicated will be copied by perl prior
15314 to this and will be provided as the regexp *r argument, however
15315 with the /old/ structures pprivate pointer value. Thus this routine
15316 may override any copying normally done by perl.
15318 It returns a pointer to the new regexp_internal structure.
15322 Perl_regdupe_internal(pTHX_ REGEXP * const rx, CLONE_PARAMS *param)
15325 struct regexp *const r = ReANY(rx);
15326 regexp_internal *reti;
15328 RXi_GET_DECL(r,ri);
15330 PERL_ARGS_ASSERT_REGDUPE_INTERNAL;
15334 Newxc(reti, sizeof(regexp_internal) + len*sizeof(regnode), char, regexp_internal);
15335 Copy(ri->program, reti->program, len+1, regnode);
15337 reti->num_code_blocks = ri->num_code_blocks;
15338 if (ri->code_blocks) {
15340 Newxc(reti->code_blocks, ri->num_code_blocks, struct reg_code_block,
15341 struct reg_code_block);
15342 Copy(ri->code_blocks, reti->code_blocks, ri->num_code_blocks,
15343 struct reg_code_block);
15344 for (n = 0; n < ri->num_code_blocks; n++)
15345 reti->code_blocks[n].src_regex = (REGEXP*)
15346 sv_dup_inc((SV*)(ri->code_blocks[n].src_regex), param);
15349 reti->code_blocks = NULL;
15351 reti->regstclass = NULL;
15354 struct reg_data *d;
15355 const int count = ri->data->count;
15358 Newxc(d, sizeof(struct reg_data) + count*sizeof(void *),
15359 char, struct reg_data);
15360 Newx(d->what, count, U8);
15363 for (i = 0; i < count; i++) {
15364 d->what[i] = ri->data->what[i];
15365 switch (d->what[i]) {
15366 /* see also regcomp.h and regfree_internal() */
15367 case 'a': /* actually an AV, but the dup function is identical. */
15371 case 'u': /* actually an HV, but the dup function is identical. */
15372 d->data[i] = sv_dup_inc((const SV *)ri->data->data[i], param);
15375 /* This is cheating. */
15376 Newx(d->data[i], 1, struct regnode_charclass_class);
15377 StructCopy(ri->data->data[i], d->data[i],
15378 struct regnode_charclass_class);
15379 reti->regstclass = (regnode*)d->data[i];
15382 /* Trie stclasses are readonly and can thus be shared
15383 * without duplication. We free the stclass in pregfree
15384 * when the corresponding reg_ac_data struct is freed.
15386 reti->regstclass= ri->regstclass;
15390 ((reg_trie_data*)ri->data->data[i])->refcount++;
15395 d->data[i] = ri->data->data[i];
15398 Perl_croak(aTHX_ "panic: re_dup unknown data code '%c'", ri->data->what[i]);
15407 reti->name_list_idx = ri->name_list_idx;
15409 #ifdef RE_TRACK_PATTERN_OFFSETS
15410 if (ri->u.offsets) {
15411 Newx(reti->u.offsets, 2*len+1, U32);
15412 Copy(ri->u.offsets, reti->u.offsets, 2*len+1, U32);
15415 SetProgLen(reti,len);
15418 return (void*)reti;
15421 #endif /* USE_ITHREADS */
15423 #ifndef PERL_IN_XSUB_RE
15426 - regnext - dig the "next" pointer out of a node
15429 Perl_regnext(pTHX_ regnode *p)
15437 if (OP(p) > REGNODE_MAX) { /* regnode.type is unsigned */
15438 Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d", (int)OP(p), (int)REGNODE_MAX);
15441 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
15450 S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
15453 STRLEN l1 = strlen(pat1);
15454 STRLEN l2 = strlen(pat2);
15457 const char *message;
15459 PERL_ARGS_ASSERT_RE_CROAK2;
15465 Copy(pat1, buf, l1 , char);
15466 Copy(pat2, buf + l1, l2 , char);
15467 buf[l1 + l2] = '\n';
15468 buf[l1 + l2 + 1] = '\0';
15470 /* ANSI variant takes additional second argument */
15471 va_start(args, pat2);
15475 msv = vmess(buf, &args);
15477 message = SvPV_const(msv,l1);
15480 Copy(message, buf, l1 , char);
15481 buf[l1-1] = '\0'; /* Overwrite \n */
15482 Perl_croak(aTHX_ "%s", buf);
15485 /* XXX Here's a total kludge. But we need to re-enter for swash routines. */
15487 #ifndef PERL_IN_XSUB_RE
15489 Perl_save_re_context(pTHX)
15493 /* Save $1..$n (#18107: UTF-8 s/(\w+)/uc($1)/e); AMS 20021106. */
15495 const REGEXP * const rx = PM_GETRE(PL_curpm);
15498 for (i = 1; i <= RX_NPARENS(rx); i++) {
15499 char digits[TYPE_CHARS(long)];
15500 const STRLEN len = my_snprintf(digits, sizeof(digits), "%lu", (long)i);
15501 GV *const *const gvp
15502 = (GV**)hv_fetch(PL_defstash, digits, len, 0);
15505 GV * const gv = *gvp;
15506 if (SvTYPE(gv) == SVt_PVGV && GvSV(gv))
15518 S_put_byte(pTHX_ SV *sv, int c)
15520 PERL_ARGS_ASSERT_PUT_BYTE;
15522 /* Our definition of isPRINT() ignores locales, so only bytes that are
15523 not part of UTF-8 are considered printable. I assume that the same
15524 holds for UTF-EBCDIC.
15525 Also, code point 255 is not printable in either (it's E0 in EBCDIC,
15526 which Wikipedia says:
15528 EO, or Eight Ones, is an 8-bit EBCDIC character code represented as all
15529 ones (binary 1111 1111, hexadecimal FF). It is similar, but not
15530 identical, to the ASCII delete (DEL) or rubout control character. ...
15531 it is typically mapped to hexadecimal code 9F, in order to provide a
15532 unique character mapping in both directions)
15534 So the old condition can be simplified to !isPRINT(c) */
15537 case '\r': Perl_sv_catpvf(aTHX_ sv, "\\r"); break;
15538 case '\n': Perl_sv_catpvf(aTHX_ sv, "\\n"); break;
15539 case '\t': Perl_sv_catpvf(aTHX_ sv, "\\t"); break;
15540 case '\f': Perl_sv_catpvf(aTHX_ sv, "\\f"); break;
15541 case '\a': Perl_sv_catpvf(aTHX_ sv, "\\a"); break;
15544 Perl_sv_catpvf(aTHX_ sv, "\\x{%x}", c);
15549 const char string = c;
15550 if (c == '-' || c == ']' || c == '\\' || c == '^')
15551 sv_catpvs(sv, "\\");
15552 sv_catpvn(sv, &string, 1);
15557 S_put_latin1_charclass_innards(pTHX_ SV *sv, char *bitmap)
15559 /* Appends to 'sv' a displayable version of the innards of the bracketed
15560 * character class whose bitmap is 'bitmap'; Returns 'TRUE' if it actually
15561 * output anything */
15564 int rangestart = -1;
15565 bool has_output_anything = FALSE;
15567 PERL_ARGS_ASSERT_PUT_LATIN1_CHARCLASS_INNARDS;
15569 for (i = 0; i <= 256; i++) {
15570 if (i < 256 && BITMAP_TEST((U8 *) bitmap,i)) {
15571 if (rangestart == -1)
15573 } else if (rangestart != -1) {
15575 if (i <= rangestart + 3) { /* Individual chars in short ranges */
15576 for (; rangestart < i; rangestart++)
15577 put_byte(sv, rangestart);
15580 || ! isALPHANUMERIC(rangestart)
15581 || ! isALPHANUMERIC(j)
15582 || isDIGIT(rangestart) != isDIGIT(j)
15583 || isUPPER(rangestart) != isUPPER(j)
15584 || isLOWER(rangestart) != isLOWER(j)
15586 /* This final test should get optimized out except
15587 * on EBCDIC platforms, where it causes ranges that
15588 * cross discontinuities like i/j to be shown as hex
15589 * instead of the misleading, e.g. H-K (since that
15590 * range includes more than H, I, J, K). */
15591 || (j - rangestart)
15592 != NATIVE_TO_ASCII(j) - NATIVE_TO_ASCII(rangestart))
15594 Perl_sv_catpvf(aTHX_ sv, "\\x{%02x}-\\x{%02x}",
15596 (j < 256) ? j : 255);
15598 else { /* Here, the ends of the range are both digits, or both
15599 uppercase, or both lowercase; and there's no
15600 discontinuity in the range (which could happen on EBCDIC
15602 put_byte(sv, rangestart);
15603 sv_catpvs(sv, "-");
15607 has_output_anything = TRUE;
15611 return has_output_anything;
15614 #define CLEAR_OPTSTART \
15615 if (optstart) STMT_START { \
15616 DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log, " (%"IVdf" nodes)\n", (IV)(node - optstart))); \
15620 #define DUMPUNTIL(b,e) CLEAR_OPTSTART; node=dumpuntil(r,start,(b),(e),last,sv,indent+1,depth+1);
15622 STATIC const regnode *
15623 S_dumpuntil(pTHX_ const regexp *r, const regnode *start, const regnode *node,
15624 const regnode *last, const regnode *plast,
15625 SV* sv, I32 indent, U32 depth)
15628 U8 op = PSEUDO; /* Arbitrary non-END op. */
15629 const regnode *next;
15630 const regnode *optstart= NULL;
15632 RXi_GET_DECL(r,ri);
15633 GET_RE_DEBUG_FLAGS_DECL;
15635 PERL_ARGS_ASSERT_DUMPUNTIL;
15637 #ifdef DEBUG_DUMPUNTIL
15638 PerlIO_printf(Perl_debug_log, "--- %d : %d - %d - %d\n",indent,node-start,
15639 last ? last-start : 0,plast ? plast-start : 0);
15642 if (plast && plast < last)
15645 while (PL_regkind[op] != END && (!last || node < last)) {
15646 /* While that wasn't END last time... */
15649 if (op == CLOSE || op == WHILEM)
15651 next = regnext((regnode *)node);
15654 if (OP(node) == OPTIMIZED) {
15655 if (!optstart && RE_DEBUG_FLAG(RE_DEBUG_COMPILE_OPTIMISE))
15662 regprop(r, sv, node);
15663 PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
15664 (int)(2*indent + 1), "", SvPVX_const(sv));
15666 if (OP(node) != OPTIMIZED) {
15667 if (next == NULL) /* Next ptr. */
15668 PerlIO_printf(Perl_debug_log, " (0)");
15669 else if (PL_regkind[(U8)op] == BRANCH && PL_regkind[OP(next)] != BRANCH )
15670 PerlIO_printf(Perl_debug_log, " (FAIL)");
15672 PerlIO_printf(Perl_debug_log, " (%"IVdf")", (IV)(next - start));
15673 (void)PerlIO_putc(Perl_debug_log, '\n');
15677 if (PL_regkind[(U8)op] == BRANCHJ) {
15680 const regnode *nnode = (OP(next) == LONGJMP
15681 ? regnext((regnode *)next)
15683 if (last && nnode > last)
15685 DUMPUNTIL(NEXTOPER(NEXTOPER(node)), nnode);
15688 else if (PL_regkind[(U8)op] == BRANCH) {
15690 DUMPUNTIL(NEXTOPER(node), next);
15692 else if ( PL_regkind[(U8)op] == TRIE ) {
15693 const regnode *this_trie = node;
15694 const char op = OP(node);
15695 const U32 n = ARG(node);
15696 const reg_ac_data * const ac = op>=AHOCORASICK ?
15697 (reg_ac_data *)ri->data->data[n] :
15699 const reg_trie_data * const trie =
15700 (reg_trie_data*)ri->data->data[op<AHOCORASICK ? n : ac->trie];
15702 AV *const trie_words = MUTABLE_AV(ri->data->data[n + TRIE_WORDS_OFFSET]);
15704 const regnode *nextbranch= NULL;
15707 for (word_idx= 0; word_idx < (I32)trie->wordcount; word_idx++) {
15708 SV ** const elem_ptr = av_fetch(trie_words,word_idx,0);
15710 PerlIO_printf(Perl_debug_log, "%*s%s ",
15711 (int)(2*(indent+3)), "",
15712 elem_ptr ? pv_pretty(sv, SvPV_nolen_const(*elem_ptr), SvCUR(*elem_ptr), 60,
15713 PL_colors[0], PL_colors[1],
15714 (SvUTF8(*elem_ptr) ? PERL_PV_ESCAPE_UNI : 0) |
15715 PERL_PV_PRETTY_ELLIPSES |
15716 PERL_PV_PRETTY_LTGT
15721 U16 dist= trie->jump[word_idx+1];
15722 PerlIO_printf(Perl_debug_log, "(%"UVuf")\n",
15723 (UV)((dist ? this_trie + dist : next) - start));
15726 nextbranch= this_trie + trie->jump[0];
15727 DUMPUNTIL(this_trie + dist, nextbranch);
15729 if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
15730 nextbranch= regnext((regnode *)nextbranch);
15732 PerlIO_printf(Perl_debug_log, "\n");
15735 if (last && next > last)
15740 else if ( op == CURLY ) { /* "next" might be very big: optimizer */
15741 DUMPUNTIL(NEXTOPER(node) + EXTRA_STEP_2ARGS,
15742 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1);
15744 else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
15746 DUMPUNTIL(NEXTOPER(node) + EXTRA_STEP_2ARGS, next);
15748 else if ( op == PLUS || op == STAR) {
15749 DUMPUNTIL(NEXTOPER(node), NEXTOPER(node) + 1);
15751 else if (PL_regkind[(U8)op] == ANYOF) {
15752 /* arglen 1 + class block */
15753 node += 1 + ((ANYOF_FLAGS(node) & ANYOF_CLASS)
15754 ? ANYOF_CLASS_SKIP : ANYOF_SKIP);
15755 node = NEXTOPER(node);
15757 else if (PL_regkind[(U8)op] == EXACT) {
15758 /* Literal string, where present. */
15759 node += NODE_SZ_STR(node) - 1;
15760 node = NEXTOPER(node);
15763 node = NEXTOPER(node);
15764 node += regarglen[(U8)op];
15766 if (op == CURLYX || op == OPEN)
15770 #ifdef DEBUG_DUMPUNTIL
15771 PerlIO_printf(Perl_debug_log, "--- %d\n", (int)indent);
15776 #endif /* DEBUGGING */
15780 * c-indentation-style: bsd
15781 * c-basic-offset: 4
15782 * indent-tabs-mode: nil
15785 * ex: set ts=8 sts=4 sw=4 et: