5 * 'A fair jaw-cracker dwarf-language must be.' --Samwise Gamgee
7 * [p.285 of _The Lord of the Rings_, II/iii: "The Ring Goes South"]
10 /* This file contains functions for compiling a regular expression. See
11 * also regexec.c which funnily enough, contains functions for executing
12 * a regular expression.
14 * This file is also copied at build time to ext/re/re_comp.c, where
15 * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
16 * This causes the main functions to be compiled under new names and with
17 * debugging support added, which makes "use re 'debug'" work.
20 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
21 * confused with the original package (see point 3 below). Thanks, Henry!
24 /* Additional note: this code is very heavily munged from Henry's version
25 * in places. In some spots I've traded clarity for efficiency, so don't
26 * blame Henry for some of the lack of readability.
29 /* The names of the functions have been changed from regcomp and
30 * regexec to pregcomp and pregexec in order to avoid conflicts
31 * with the POSIX routines of the same names.
34 #ifdef PERL_EXT_RE_BUILD
39 * pregcomp and pregexec -- regsub and regerror are not used in perl
41 * Copyright (c) 1986 by University of Toronto.
42 * Written by Henry Spencer. Not derived from licensed software.
44 * Permission is granted to anyone to use this software for any
45 * purpose on any computer system, and to redistribute it freely,
46 * subject to the following restrictions:
48 * 1. The author is not responsible for the consequences of use of
49 * this software, no matter how awful, even if they arise
52 * 2. The origin of this software must not be misrepresented, either
53 * by explicit claim or by omission.
55 * 3. Altered versions must be plainly marked as such, and must not
56 * be misrepresented as being the original software.
59 **** Alterations to Henry's code are...
61 **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
62 **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
63 **** by Larry Wall and others
65 **** You may distribute under the terms of either the GNU General Public
66 **** License or the Artistic License, as specified in the README file.
69 * Beware that some of this code is subtly aware of the way operator
70 * precedence is structured in regular expressions. Serious changes in
71 * regular-expression syntax might require a total rethink.
74 #define PERL_IN_REGCOMP_C
77 #ifndef PERL_IN_XSUB_RE
82 #ifdef PERL_IN_XSUB_RE
84 EXTERN_C const struct regexp_engine my_reg_engine;
89 #include "dquote_static.c"
90 #include "charclass_invlists.h"
91 #include "inline_invlist.c"
92 #include "unicode_constants.h"
94 #define HAS_NONLATIN1_FOLD_CLOSURE(i) \
95 _HAS_NONLATIN1_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(i)
96 #define HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(i) \
97 _HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE_ONLY_FOR_USE_BY_REGCOMP_DOT_C_AND_REGEXEC_DOT_C(i)
98 #define IS_NON_FINAL_FOLD(c) _IS_NON_FINAL_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c)
99 #define IS_IN_SOME_FOLD_L1(c) _IS_IN_SOME_FOLD_ONLY_FOR_USE_BY_REGCOMP_DOT_C(c)
102 #define STATIC static
106 struct RExC_state_t {
107 U32 flags; /* RXf_* are we folding, multilining? */
108 U32 pm_flags; /* PMf_* stuff from the calling PMOP */
109 char *precomp; /* uncompiled string. */
110 REGEXP *rx_sv; /* The SV that is the regexp. */
111 regexp *rx; /* perl core regexp structure */
112 regexp_internal *rxi; /* internal data for regexp object
114 char *start; /* Start of input for compile */
115 char *end; /* End of input for compile */
116 char *parse; /* Input-scan pointer. */
117 SSize_t whilem_seen; /* number of WHILEM in this expr */
118 regnode *emit_start; /* Start of emitted-code area */
119 regnode *emit_bound; /* First regnode outside of the
121 regnode *emit; /* Code-emit pointer; if = &emit_dummy,
122 implies compiling, so don't emit */
123 regnode_ssc emit_dummy; /* placeholder for emit to point to;
124 large enough for the largest
125 non-EXACTish node, so can use it as
127 I32 naughty; /* How bad is this pattern? */
128 I32 sawback; /* Did we see \1, ...? */
130 SSize_t size; /* Code size. */
131 I32 npar; /* Capture buffer count, (OPEN) plus
132 one. ("par" 0 is the whole
134 I32 nestroot; /* root parens we are in - used by
138 regnode **open_parens; /* pointers to open parens */
139 regnode **close_parens; /* pointers to close parens */
140 regnode *opend; /* END node in program */
141 I32 utf8; /* whether the pattern is utf8 or not */
142 I32 orig_utf8; /* whether the pattern was originally in utf8 */
143 /* XXX use this for future optimisation of case
144 * where pattern must be upgraded to utf8. */
145 I32 uni_semantics; /* If a d charset modifier should use unicode
146 rules, even if the pattern is not in
148 HV *paren_names; /* Paren names */
150 regnode **recurse; /* Recurse regops */
151 I32 recurse_count; /* Number of recurse regops */
152 U8 *study_chunk_recursed; /* bitmap of which parens we have moved
154 U32 study_chunk_recursed_bytes; /* bytes in bitmap */
158 I32 override_recoding;
159 I32 in_multi_char_class;
160 struct reg_code_block *code_blocks; /* positions of literal (?{})
162 int num_code_blocks; /* size of code_blocks[] */
163 int code_index; /* next code_blocks[] slot */
164 SSize_t maxlen; /* mininum possible number of chars in string to match */
165 #ifdef ADD_TO_REGEXEC
166 char *starttry; /* -Dr: where regtry was called. */
167 #define RExC_starttry (pRExC_state->starttry)
169 SV *runtime_code_qr; /* qr with the runtime code blocks */
171 const char *lastparse;
173 AV *paren_name_list; /* idx -> name */
174 #define RExC_lastparse (pRExC_state->lastparse)
175 #define RExC_lastnum (pRExC_state->lastnum)
176 #define RExC_paren_name_list (pRExC_state->paren_name_list)
180 #define RExC_flags (pRExC_state->flags)
181 #define RExC_pm_flags (pRExC_state->pm_flags)
182 #define RExC_precomp (pRExC_state->precomp)
183 #define RExC_rx_sv (pRExC_state->rx_sv)
184 #define RExC_rx (pRExC_state->rx)
185 #define RExC_rxi (pRExC_state->rxi)
186 #define RExC_start (pRExC_state->start)
187 #define RExC_end (pRExC_state->end)
188 #define RExC_parse (pRExC_state->parse)
189 #define RExC_whilem_seen (pRExC_state->whilem_seen)
190 #ifdef RE_TRACK_PATTERN_OFFSETS
191 #define RExC_offsets (pRExC_state->rxi->u.offsets) /* I am not like the
194 #define RExC_emit (pRExC_state->emit)
195 #define RExC_emit_dummy (pRExC_state->emit_dummy)
196 #define RExC_emit_start (pRExC_state->emit_start)
197 #define RExC_emit_bound (pRExC_state->emit_bound)
198 #define RExC_naughty (pRExC_state->naughty)
199 #define RExC_sawback (pRExC_state->sawback)
200 #define RExC_seen (pRExC_state->seen)
201 #define RExC_size (pRExC_state->size)
202 #define RExC_maxlen (pRExC_state->maxlen)
203 #define RExC_npar (pRExC_state->npar)
204 #define RExC_nestroot (pRExC_state->nestroot)
205 #define RExC_extralen (pRExC_state->extralen)
206 #define RExC_seen_zerolen (pRExC_state->seen_zerolen)
207 #define RExC_utf8 (pRExC_state->utf8)
208 #define RExC_uni_semantics (pRExC_state->uni_semantics)
209 #define RExC_orig_utf8 (pRExC_state->orig_utf8)
210 #define RExC_open_parens (pRExC_state->open_parens)
211 #define RExC_close_parens (pRExC_state->close_parens)
212 #define RExC_opend (pRExC_state->opend)
213 #define RExC_paren_names (pRExC_state->paren_names)
214 #define RExC_recurse (pRExC_state->recurse)
215 #define RExC_recurse_count (pRExC_state->recurse_count)
216 #define RExC_study_chunk_recursed (pRExC_state->study_chunk_recursed)
217 #define RExC_study_chunk_recursed_bytes \
218 (pRExC_state->study_chunk_recursed_bytes)
219 #define RExC_in_lookbehind (pRExC_state->in_lookbehind)
220 #define RExC_contains_locale (pRExC_state->contains_locale)
221 #define RExC_contains_i (pRExC_state->contains_i)
222 #define RExC_override_recoding (pRExC_state->override_recoding)
223 #define RExC_in_multi_char_class (pRExC_state->in_multi_char_class)
226 #define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
227 #define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
228 ((*s) == '{' && regcurly(s)))
231 * Flags to be passed up and down.
233 #define WORST 0 /* Worst case. */
234 #define HASWIDTH 0x01 /* Known to match non-null strings. */
236 /* Simple enough to be STAR/PLUS operand; in an EXACTish node must be a single
237 * character. (There needs to be a case: in the switch statement in regexec.c
238 * for any node marked SIMPLE.) Note that this is not the same thing as
241 #define SPSTART 0x04 /* Starts with * or + */
242 #define POSTPONED 0x08 /* (?1),(?&name), (??{...}) or similar */
243 #define TRYAGAIN 0x10 /* Weeded out a declaration. */
244 #define RESTART_UTF8 0x20 /* Restart, need to calcuate sizes as UTF-8 */
246 #define REG_NODE_NUM(x) ((x) ? (int)((x)-RExC_emit_start) : -1)
248 /* whether trie related optimizations are enabled */
249 #if PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION
250 #define TRIE_STUDY_OPT
251 #define FULL_TRIE_STUDY
257 #define PBYTE(u8str,paren) ((U8*)(u8str))[(paren) >> 3]
258 #define PBITVAL(paren) (1 << ((paren) & 7))
259 #define PAREN_TEST(u8str,paren) ( PBYTE(u8str,paren) & PBITVAL(paren))
260 #define PAREN_SET(u8str,paren) PBYTE(u8str,paren) |= PBITVAL(paren)
261 #define PAREN_UNSET(u8str,paren) PBYTE(u8str,paren) &= (~PBITVAL(paren))
263 #define REQUIRE_UTF8 STMT_START { \
265 *flagp = RESTART_UTF8; \
270 /* This converts the named class defined in regcomp.h to its equivalent class
271 * number defined in handy.h. */
272 #define namedclass_to_classnum(class) ((int) ((class) / 2))
273 #define classnum_to_namedclass(classnum) ((classnum) * 2)
275 #define _invlist_union_complement_2nd(a, b, output) \
276 _invlist_union_maybe_complement_2nd(a, b, TRUE, output)
277 #define _invlist_intersection_complement_2nd(a, b, output) \
278 _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output)
280 /* About scan_data_t.
282 During optimisation we recurse through the regexp program performing
283 various inplace (keyhole style) optimisations. In addition study_chunk
284 and scan_commit populate this data structure with information about
285 what strings MUST appear in the pattern. We look for the longest
286 string that must appear at a fixed location, and we look for the
287 longest string that may appear at a floating location. So for instance
292 Both 'FOO' and 'A' are fixed strings. Both 'B' and 'BAR' are floating
293 strings (because they follow a .* construct). study_chunk will identify
294 both FOO and BAR as being the longest fixed and floating strings respectively.
296 The strings can be composites, for instance
300 will result in a composite fixed substring 'foo'.
302 For each string some basic information is maintained:
304 - offset or min_offset
305 This is the position the string must appear at, or not before.
306 It also implicitly (when combined with minlenp) tells us how many
307 characters must match before the string we are searching for.
308 Likewise when combined with minlenp and the length of the string it
309 tells us how many characters must appear after the string we have
313 Only used for floating strings. This is the rightmost point that
314 the string can appear at. If set to SSize_t_MAX it indicates that the
315 string can occur infinitely far to the right.
318 A pointer to the minimum number of characters of the pattern that the
319 string was found inside. This is important as in the case of positive
320 lookahead or positive lookbehind we can have multiple patterns
325 The minimum length of the pattern overall is 3, the minimum length
326 of the lookahead part is 3, but the minimum length of the part that
327 will actually match is 1. So 'FOO's minimum length is 3, but the
328 minimum length for the F is 1. This is important as the minimum length
329 is used to determine offsets in front of and behind the string being
330 looked for. Since strings can be composites this is the length of the
331 pattern at the time it was committed with a scan_commit. Note that
332 the length is calculated by study_chunk, so that the minimum lengths
333 are not known until the full pattern has been compiled, thus the
334 pointer to the value.
338 In the case of lookbehind the string being searched for can be
339 offset past the start point of the final matching string.
340 If this value was just blithely removed from the min_offset it would
341 invalidate some of the calculations for how many chars must match
342 before or after (as they are derived from min_offset and minlen and
343 the length of the string being searched for).
344 When the final pattern is compiled and the data is moved from the
345 scan_data_t structure into the regexp structure the information
346 about lookbehind is factored in, with the information that would
347 have been lost precalculated in the end_shift field for the
350 The fields pos_min and pos_delta are used to store the minimum offset
351 and the delta to the maximum offset at the current point in the pattern.
355 typedef struct scan_data_t {
356 /*I32 len_min; unused */
357 /*I32 len_delta; unused */
361 SSize_t last_end; /* min value, <0 unless valid. */
362 SSize_t last_start_min;
363 SSize_t last_start_max;
364 SV **longest; /* Either &l_fixed, or &l_float. */
365 SV *longest_fixed; /* longest fixed string found in pattern */
366 SSize_t offset_fixed; /* offset where it starts */
367 SSize_t *minlen_fixed; /* pointer to the minlen relevant to the string */
368 I32 lookbehind_fixed; /* is the position of the string modfied by LB */
369 SV *longest_float; /* longest floating string found in pattern */
370 SSize_t offset_float_min; /* earliest point in string it can appear */
371 SSize_t offset_float_max; /* latest point in string it can appear */
372 SSize_t *minlen_float; /* pointer to the minlen relevant to the string */
373 SSize_t lookbehind_float; /* is the pos of the string modified by LB */
376 SSize_t *last_closep;
377 regnode_ssc *start_class;
380 /* The below is perhaps overboard, but this allows us to save a test at the
381 * expense of a mask. This is because on both EBCDIC and ASCII machines, 'A'
382 * and 'a' differ by a single bit; the same with the upper and lower case of
383 * all other ASCII-range alphabetics. On ASCII platforms, they are 32 apart;
384 * on EBCDIC, they are 64. This uses an exclusive 'or' to find that bit and
385 * then inverts it to form a mask, with just a single 0, in the bit position
386 * where the upper- and lowercase differ. XXX There are about 40 other
387 * instances in the Perl core where this micro-optimization could be used.
388 * Should decide if maintenance cost is worse, before changing those
390 * Returns a boolean as to whether or not 'v' is either a lowercase or
391 * uppercase instance of 'c', where 'c' is in [A-Za-z]. If 'c' is a
392 * compile-time constant, the generated code is better than some optimizing
393 * compilers figure out, amounting to a mask and test. The results are
394 * meaningless if 'c' is not one of [A-Za-z] */
395 #define isARG2_lower_or_UPPER_ARG1(c, v) \
396 (((v) & ~('A' ^ 'a')) == ((c) & ~('A' ^ 'a')))
399 * Forward declarations for pregcomp()'s friends.
402 static const scan_data_t zero_scan_data =
403 { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ,0};
405 #define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
406 #define SF_BEFORE_SEOL 0x0001
407 #define SF_BEFORE_MEOL 0x0002
408 #define SF_FIX_BEFORE_EOL (SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
409 #define SF_FL_BEFORE_EOL (SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
411 #define SF_FIX_SHIFT_EOL (+2)
412 #define SF_FL_SHIFT_EOL (+4)
414 #define SF_FIX_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
415 #define SF_FIX_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
417 #define SF_FL_BEFORE_SEOL (SF_BEFORE_SEOL << SF_FL_SHIFT_EOL)
418 #define SF_FL_BEFORE_MEOL (SF_BEFORE_MEOL << SF_FL_SHIFT_EOL) /* 0x20 */
419 #define SF_IS_INF 0x0040
420 #define SF_HAS_PAR 0x0080
421 #define SF_IN_PAR 0x0100
422 #define SF_HAS_EVAL 0x0200
423 #define SCF_DO_SUBSTR 0x0400
424 #define SCF_DO_STCLASS_AND 0x0800
425 #define SCF_DO_STCLASS_OR 0x1000
426 #define SCF_DO_STCLASS (SCF_DO_STCLASS_AND|SCF_DO_STCLASS_OR)
427 #define SCF_WHILEM_VISITED_POS 0x2000
429 #define SCF_TRIE_RESTUDY 0x4000 /* Do restudy? */
430 #define SCF_SEEN_ACCEPT 0x8000
431 #define SCF_TRIE_DOING_RESTUDY 0x10000
433 #define UTF cBOOL(RExC_utf8)
435 /* The enums for all these are ordered so things work out correctly */
436 #define LOC (get_regex_charset(RExC_flags) == REGEX_LOCALE_CHARSET)
437 #define DEPENDS_SEMANTICS (get_regex_charset(RExC_flags) \
438 == REGEX_DEPENDS_CHARSET)
439 #define UNI_SEMANTICS (get_regex_charset(RExC_flags) == REGEX_UNICODE_CHARSET)
440 #define AT_LEAST_UNI_SEMANTICS (get_regex_charset(RExC_flags) \
441 >= REGEX_UNICODE_CHARSET)
442 #define ASCII_RESTRICTED (get_regex_charset(RExC_flags) \
443 == REGEX_ASCII_RESTRICTED_CHARSET)
444 #define AT_LEAST_ASCII_RESTRICTED (get_regex_charset(RExC_flags) \
445 >= REGEX_ASCII_RESTRICTED_CHARSET)
446 #define ASCII_FOLD_RESTRICTED (get_regex_charset(RExC_flags) \
447 == REGEX_ASCII_MORE_RESTRICTED_CHARSET)
449 #define FOLD cBOOL(RExC_flags & RXf_PMf_FOLD)
451 /* For programs that want to be strictly Unicode compatible by dying if any
452 * attempt is made to match a non-Unicode code point against a Unicode
454 #define ALWAYS_WARN_SUPER ckDEAD(packWARN(WARN_NON_UNICODE))
456 #define OOB_NAMEDCLASS -1
458 /* There is no code point that is out-of-bounds, so this is problematic. But
459 * its only current use is to initialize a variable that is always set before
461 #define OOB_UNICODE 0xDEADBEEF
463 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
464 #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
467 /* length of regex to show in messages that don't mark a position within */
468 #define RegexLengthToShowInErrorMessages 127
471 * If MARKER[12] are adjusted, be sure to adjust the constants at the top
472 * of t/op/regmesg.t, the tests in t/op/re_tests, and those in
473 * op/pragma/warn/regcomp.
475 #define MARKER1 "<-- HERE" /* marker as it appears in the description */
476 #define MARKER2 " <-- HERE " /* marker as it appears within the regex */
478 #define REPORT_LOCATION " in regex; marked by " MARKER1 \
479 " in m/%"UTF8f MARKER2 "%"UTF8f"/"
481 #define REPORT_LOCATION_ARGS(offset) \
482 UTF8fARG(UTF, offset, RExC_precomp), \
483 UTF8fARG(UTF, RExC_end - RExC_precomp - offset, RExC_precomp + offset)
486 * Calls SAVEDESTRUCTOR_X if needed, then calls Perl_croak with the given
487 * arg. Show regex, up to a maximum length. If it's too long, chop and add
490 #define _FAIL(code) STMT_START { \
491 const char *ellipses = ""; \
492 IV len = RExC_end - RExC_precomp; \
495 SAVEFREESV(RExC_rx_sv); \
496 if (len > RegexLengthToShowInErrorMessages) { \
497 /* chop 10 shorter than the max, to ensure meaning of "..." */ \
498 len = RegexLengthToShowInErrorMessages - 10; \
504 #define FAIL(msg) _FAIL( \
505 Perl_croak(aTHX_ "%s in regex m/%"UTF8f"%s/", \
506 msg, UTF8fARG(UTF, len, RExC_precomp), ellipses))
508 #define FAIL2(msg,arg) _FAIL( \
509 Perl_croak(aTHX_ msg " in regex m/%"UTF8f"%s/", \
510 arg, UTF8fARG(UTF, len, RExC_precomp), ellipses))
513 * Simple_vFAIL -- like FAIL, but marks the current location in the scan
515 #define Simple_vFAIL(m) STMT_START { \
516 const IV offset = RExC_parse - RExC_precomp; \
517 Perl_croak(aTHX_ "%s" REPORT_LOCATION, \
518 m, REPORT_LOCATION_ARGS(offset)); \
522 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL()
524 #define vFAIL(m) STMT_START { \
526 SAVEFREESV(RExC_rx_sv); \
531 * Like Simple_vFAIL(), but accepts two arguments.
533 #define Simple_vFAIL2(m,a1) STMT_START { \
534 const IV offset = RExC_parse - RExC_precomp; \
535 S_re_croak2(aTHX_ UTF, m, REPORT_LOCATION, a1, \
536 REPORT_LOCATION_ARGS(offset)); \
540 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL2().
542 #define vFAIL2(m,a1) STMT_START { \
544 SAVEFREESV(RExC_rx_sv); \
545 Simple_vFAIL2(m, a1); \
550 * Like Simple_vFAIL(), but accepts three arguments.
552 #define Simple_vFAIL3(m, a1, a2) STMT_START { \
553 const IV offset = RExC_parse - RExC_precomp; \
554 S_re_croak2(aTHX_ UTF, m, REPORT_LOCATION, a1, a2, \
555 REPORT_LOCATION_ARGS(offset)); \
559 * Calls SAVEDESTRUCTOR_X if needed, then Simple_vFAIL3().
561 #define vFAIL3(m,a1,a2) STMT_START { \
563 SAVEFREESV(RExC_rx_sv); \
564 Simple_vFAIL3(m, a1, a2); \
568 * Like Simple_vFAIL(), but accepts four arguments.
570 #define Simple_vFAIL4(m, a1, a2, a3) STMT_START { \
571 const IV offset = RExC_parse - RExC_precomp; \
572 S_re_croak2(aTHX_ UTF, m, REPORT_LOCATION, a1, a2, a3, \
573 REPORT_LOCATION_ARGS(offset)); \
576 #define vFAIL4(m,a1,a2,a3) STMT_START { \
578 SAVEFREESV(RExC_rx_sv); \
579 Simple_vFAIL4(m, a1, a2, a3); \
582 /* A specialized version of vFAIL2 that works with UTF8f */
583 #define vFAIL2utf8f(m, a1) STMT_START { \
584 const IV offset = RExC_parse - RExC_precomp; \
586 SAVEFREESV(RExC_rx_sv); \
587 S_re_croak2(aTHX_ UTF, m, REPORT_LOCATION, a1, \
588 REPORT_LOCATION_ARGS(offset)); \
592 /* m is not necessarily a "literal string", in this macro */
593 #define reg_warn_non_literal_string(loc, m) STMT_START { \
594 const IV offset = loc - RExC_precomp; \
595 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s" REPORT_LOCATION, \
596 m, REPORT_LOCATION_ARGS(offset)); \
599 #define ckWARNreg(loc,m) STMT_START { \
600 const IV offset = loc - RExC_precomp; \
601 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
602 REPORT_LOCATION_ARGS(offset)); \
605 #define vWARN_dep(loc, m) STMT_START { \
606 const IV offset = loc - RExC_precomp; \
607 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), m REPORT_LOCATION, \
608 REPORT_LOCATION_ARGS(offset)); \
611 #define ckWARNdep(loc,m) STMT_START { \
612 const IV offset = loc - RExC_precomp; \
613 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED), \
615 REPORT_LOCATION_ARGS(offset)); \
618 #define ckWARNregdep(loc,m) STMT_START { \
619 const IV offset = loc - RExC_precomp; \
620 Perl_ck_warner_d(aTHX_ packWARN2(WARN_DEPRECATED, WARN_REGEXP), \
622 REPORT_LOCATION_ARGS(offset)); \
625 #define ckWARN2reg_d(loc,m, a1) STMT_START { \
626 const IV offset = loc - RExC_precomp; \
627 Perl_ck_warner_d(aTHX_ packWARN(WARN_REGEXP), \
629 a1, REPORT_LOCATION_ARGS(offset)); \
632 #define ckWARN2reg(loc, m, a1) STMT_START { \
633 const IV offset = loc - RExC_precomp; \
634 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
635 a1, REPORT_LOCATION_ARGS(offset)); \
638 #define vWARN3(loc, m, a1, a2) STMT_START { \
639 const IV offset = loc - RExC_precomp; \
640 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
641 a1, a2, REPORT_LOCATION_ARGS(offset)); \
644 #define ckWARN3reg(loc, m, a1, a2) STMT_START { \
645 const IV offset = loc - RExC_precomp; \
646 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
647 a1, a2, REPORT_LOCATION_ARGS(offset)); \
650 #define vWARN4(loc, m, a1, a2, a3) STMT_START { \
651 const IV offset = loc - RExC_precomp; \
652 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
653 a1, a2, a3, REPORT_LOCATION_ARGS(offset)); \
656 #define ckWARN4reg(loc, m, a1, a2, a3) STMT_START { \
657 const IV offset = loc - RExC_precomp; \
658 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
659 a1, a2, a3, REPORT_LOCATION_ARGS(offset)); \
662 #define vWARN5(loc, m, a1, a2, a3, a4) STMT_START { \
663 const IV offset = loc - RExC_precomp; \
664 Perl_warner(aTHX_ packWARN(WARN_REGEXP), m REPORT_LOCATION, \
665 a1, a2, a3, a4, REPORT_LOCATION_ARGS(offset)); \
669 /* Allow for side effects in s */
670 #define REGC(c,s) STMT_START { \
671 if (!SIZE_ONLY) *(s) = (c); else (void)(s); \
674 /* Macros for recording node offsets. 20001227 mjd@plover.com
675 * Nodes are numbered 1, 2, 3, 4. Node #n's position is recorded in
676 * element 2*n-1 of the array. Element #2n holds the byte length node #n.
677 * Element 0 holds the number n.
678 * Position is 1 indexed.
680 #ifndef RE_TRACK_PATTERN_OFFSETS
681 #define Set_Node_Offset_To_R(node,byte)
682 #define Set_Node_Offset(node,byte)
683 #define Set_Cur_Node_Offset
684 #define Set_Node_Length_To_R(node,len)
685 #define Set_Node_Length(node,len)
686 #define Set_Node_Cur_Length(node,start)
687 #define Node_Offset(n)
688 #define Node_Length(n)
689 #define Set_Node_Offset_Length(node,offset,len)
690 #define ProgLen(ri) ri->u.proglen
691 #define SetProgLen(ri,x) ri->u.proglen = x
693 #define ProgLen(ri) ri->u.offsets[0]
694 #define SetProgLen(ri,x) ri->u.offsets[0] = x
695 #define Set_Node_Offset_To_R(node,byte) STMT_START { \
697 MJD_OFFSET_DEBUG(("** (%d) offset of node %d is %d.\n", \
698 __LINE__, (int)(node), (int)(byte))); \
700 Perl_croak(aTHX_ "value of node is %d in Offset macro", \
703 RExC_offsets[2*(node)-1] = (byte); \
708 #define Set_Node_Offset(node,byte) \
709 Set_Node_Offset_To_R((node)-RExC_emit_start, (byte)-RExC_start)
710 #define Set_Cur_Node_Offset Set_Node_Offset(RExC_emit, RExC_parse)
712 #define Set_Node_Length_To_R(node,len) STMT_START { \
714 MJD_OFFSET_DEBUG(("** (%d) size of node %d is %d.\n", \
715 __LINE__, (int)(node), (int)(len))); \
717 Perl_croak(aTHX_ "value of node is %d in Length macro", \
720 RExC_offsets[2*(node)] = (len); \
725 #define Set_Node_Length(node,len) \
726 Set_Node_Length_To_R((node)-RExC_emit_start, len)
727 #define Set_Node_Cur_Length(node, start) \
728 Set_Node_Length(node, RExC_parse - start)
730 /* Get offsets and lengths */
731 #define Node_Offset(n) (RExC_offsets[2*((n)-RExC_emit_start)-1])
732 #define Node_Length(n) (RExC_offsets[2*((n)-RExC_emit_start)])
734 #define Set_Node_Offset_Length(node,offset,len) STMT_START { \
735 Set_Node_Offset_To_R((node)-RExC_emit_start, (offset)); \
736 Set_Node_Length_To_R((node)-RExC_emit_start, (len)); \
740 #if PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS
741 #define EXPERIMENTAL_INPLACESCAN
742 #endif /*PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS*/
744 #define DEBUG_RExC_seen() \
745 DEBUG_OPTIMISE_MORE_r({ \
746 PerlIO_printf(Perl_debug_log,"RExC_seen: "); \
748 if (RExC_seen & REG_ZERO_LEN_SEEN) \
749 PerlIO_printf(Perl_debug_log,"REG_ZERO_LEN_SEEN "); \
751 if (RExC_seen & REG_LOOKBEHIND_SEEN) \
752 PerlIO_printf(Perl_debug_log,"REG_LOOKBEHIND_SEEN "); \
754 if (RExC_seen & REG_GPOS_SEEN) \
755 PerlIO_printf(Perl_debug_log,"REG_GPOS_SEEN "); \
757 if (RExC_seen & REG_CANY_SEEN) \
758 PerlIO_printf(Perl_debug_log,"REG_CANY_SEEN "); \
760 if (RExC_seen & REG_RECURSE_SEEN) \
761 PerlIO_printf(Perl_debug_log,"REG_RECURSE_SEEN "); \
763 if (RExC_seen & REG_TOP_LEVEL_BRANCHES_SEEN) \
764 PerlIO_printf(Perl_debug_log,"REG_TOP_LEVEL_BRANCHES_SEEN "); \
766 if (RExC_seen & REG_VERBARG_SEEN) \
767 PerlIO_printf(Perl_debug_log,"REG_VERBARG_SEEN "); \
769 if (RExC_seen & REG_CUTGROUP_SEEN) \
770 PerlIO_printf(Perl_debug_log,"REG_CUTGROUP_SEEN "); \
772 if (RExC_seen & REG_RUN_ON_COMMENT_SEEN) \
773 PerlIO_printf(Perl_debug_log,"REG_RUN_ON_COMMENT_SEEN "); \
775 if (RExC_seen & REG_UNFOLDED_MULTI_SEEN) \
776 PerlIO_printf(Perl_debug_log,"REG_UNFOLDED_MULTI_SEEN "); \
778 if (RExC_seen & REG_GOSTART_SEEN) \
779 PerlIO_printf(Perl_debug_log,"REG_GOSTART_SEEN "); \
781 if (RExC_seen & REG_UNBOUNDED_QUANTIFIER_SEEN) \
782 PerlIO_printf(Perl_debug_log,"REG_UNBOUNDED_QUANTIFIER_SEEN "); \
784 PerlIO_printf(Perl_debug_log,"\n"); \
787 #define DEBUG_STUDYDATA(str,data,depth) \
788 DEBUG_OPTIMISE_MORE_r(if(data){ \
789 PerlIO_printf(Perl_debug_log, \
790 "%*s" str "Pos:%"IVdf"/%"IVdf \
791 " Flags: 0x%"UVXf" Whilem_c: %"IVdf" Lcp: %"IVdf" %s", \
792 (int)(depth)*2, "", \
793 (IV)((data)->pos_min), \
794 (IV)((data)->pos_delta), \
795 (UV)((data)->flags), \
796 (IV)((data)->whilem_c), \
797 (IV)((data)->last_closep ? *((data)->last_closep) : -1), \
798 is_inf ? "INF " : "" \
800 if ((data)->last_found) \
801 PerlIO_printf(Perl_debug_log, \
802 "Last:'%s' %"IVdf":%"IVdf"/%"IVdf" %sFixed:'%s' @ %"IVdf \
803 " %sFloat: '%s' @ %"IVdf"/%"IVdf"", \
804 SvPVX_const((data)->last_found), \
805 (IV)((data)->last_end), \
806 (IV)((data)->last_start_min), \
807 (IV)((data)->last_start_max), \
808 ((data)->longest && \
809 (data)->longest==&((data)->longest_fixed)) ? "*" : "", \
810 SvPVX_const((data)->longest_fixed), \
811 (IV)((data)->offset_fixed), \
812 ((data)->longest && \
813 (data)->longest==&((data)->longest_float)) ? "*" : "", \
814 SvPVX_const((data)->longest_float), \
815 (IV)((data)->offset_float_min), \
816 (IV)((data)->offset_float_max) \
818 PerlIO_printf(Perl_debug_log,"\n"); \
821 /* Mark that we cannot extend a found fixed substring at this point.
822 Update the longest found anchored substring and the longest found
823 floating substrings if needed. */
826 S_scan_commit(pTHX_ const RExC_state_t *pRExC_state, scan_data_t *data,
827 SSize_t *minlenp, int is_inf)
829 const STRLEN l = CHR_SVLEN(data->last_found);
830 const STRLEN old_l = CHR_SVLEN(*data->longest);
831 GET_RE_DEBUG_FLAGS_DECL;
833 PERL_ARGS_ASSERT_SCAN_COMMIT;
835 if ((l >= old_l) && ((l > old_l) || (data->flags & SF_BEFORE_EOL))) {
836 SvSetMagicSV(*data->longest, data->last_found);
837 if (*data->longest == data->longest_fixed) {
838 data->offset_fixed = l ? data->last_start_min : data->pos_min;
839 if (data->flags & SF_BEFORE_EOL)
841 |= ((data->flags & SF_BEFORE_EOL) << SF_FIX_SHIFT_EOL);
843 data->flags &= ~SF_FIX_BEFORE_EOL;
844 data->minlen_fixed=minlenp;
845 data->lookbehind_fixed=0;
847 else { /* *data->longest == data->longest_float */
848 data->offset_float_min = l ? data->last_start_min : data->pos_min;
849 data->offset_float_max = (l
850 ? data->last_start_max
851 : (data->pos_delta == SSize_t_MAX
853 : data->pos_min + data->pos_delta));
855 || (STRLEN)data->offset_float_max > (STRLEN)SSize_t_MAX)
856 data->offset_float_max = SSize_t_MAX;
857 if (data->flags & SF_BEFORE_EOL)
859 |= ((data->flags & SF_BEFORE_EOL) << SF_FL_SHIFT_EOL);
861 data->flags &= ~SF_FL_BEFORE_EOL;
862 data->minlen_float=minlenp;
863 data->lookbehind_float=0;
866 SvCUR_set(data->last_found, 0);
868 SV * const sv = data->last_found;
869 if (SvUTF8(sv) && SvMAGICAL(sv)) {
870 MAGIC * const mg = mg_find(sv, PERL_MAGIC_utf8);
876 data->flags &= ~SF_BEFORE_EOL;
877 DEBUG_STUDYDATA("commit: ",data,0);
880 /* An SSC is just a regnode_charclass_posix with an extra field: the inversion
881 * list that describes which code points it matches */
884 S_ssc_anything(pTHX_ regnode_ssc *ssc)
886 /* Set the SSC 'ssc' to match an empty string or any code point */
888 PERL_ARGS_ASSERT_SSC_ANYTHING;
890 assert(is_ANYOF_SYNTHETIC(ssc));
892 ssc->invlist = sv_2mortal(_new_invlist(2)); /* mortalize so won't leak */
893 _append_range_to_invlist(ssc->invlist, 0, UV_MAX);
894 ANYOF_FLAGS(ssc) |= ANYOF_EMPTY_STRING; /* Plus match empty string */
898 S_ssc_is_anything(const regnode_ssc *ssc)
900 /* Returns TRUE if the SSC 'ssc' can match the empty string and any code
901 * point; FALSE otherwise. Thus, this is used to see if using 'ssc' buys
902 * us anything: if the function returns TRUE, 'ssc' hasn't been restricted
903 * in any way, so there's no point in using it */
908 PERL_ARGS_ASSERT_SSC_IS_ANYTHING;
910 assert(is_ANYOF_SYNTHETIC(ssc));
912 if (! (ANYOF_FLAGS(ssc) & ANYOF_EMPTY_STRING)) {
916 /* See if the list consists solely of the range 0 - Infinity */
917 invlist_iterinit(ssc->invlist);
918 ret = invlist_iternext(ssc->invlist, &start, &end)
922 invlist_iterfinish(ssc->invlist);
928 /* If e.g., both \w and \W are set, matches everything */
929 if (ANYOF_POSIXL_SSC_TEST_ANY_SET(ssc)) {
931 for (i = 0; i < ANYOF_POSIXL_MAX; i += 2) {
932 if (ANYOF_POSIXL_TEST(ssc, i) && ANYOF_POSIXL_TEST(ssc, i+1)) {
942 S_ssc_init(pTHX_ const RExC_state_t *pRExC_state, regnode_ssc *ssc)
944 /* Initializes the SSC 'ssc'. This includes setting it to match an empty
945 * string, any code point, or any posix class under locale */
947 PERL_ARGS_ASSERT_SSC_INIT;
949 Zero(ssc, 1, regnode_ssc);
950 set_ANYOF_SYNTHETIC(ssc);
951 ARG_SET(ssc, ANYOF_NONBITMAP_EMPTY);
954 /* If any portion of the regex is to operate under locale rules,
955 * initialization includes it. The reason this isn't done for all regexes
956 * is that the optimizer was written under the assumption that locale was
957 * all-or-nothing. Given the complexity and lack of documentation in the
958 * optimizer, and that there are inadequate test cases for locale, many
959 * parts of it may not work properly, it is safest to avoid locale unless
961 if (RExC_contains_locale) {
962 ANYOF_POSIXL_SETALL(ssc);
965 ANYOF_POSIXL_ZERO(ssc);
970 S_ssc_is_cp_posixl_init(const RExC_state_t *pRExC_state,
971 const regnode_ssc *ssc)
973 /* Returns TRUE if the SSC 'ssc' is in its initial state with regard only
974 * to the list of code points matched, and locale posix classes; hence does
975 * not check its flags) */
980 PERL_ARGS_ASSERT_SSC_IS_CP_POSIXL_INIT;
982 assert(is_ANYOF_SYNTHETIC(ssc));
984 invlist_iterinit(ssc->invlist);
985 ret = invlist_iternext(ssc->invlist, &start, &end)
989 invlist_iterfinish(ssc->invlist);
995 if (RExC_contains_locale && ! ANYOF_POSIXL_SSC_TEST_ALL_SET(ssc)) {
1003 S_get_ANYOF_cp_list_for_ssc(pTHX_ const RExC_state_t *pRExC_state,
1004 const regnode_charclass* const node)
1006 /* Returns a mortal inversion list defining which code points are matched
1007 * by 'node', which is of type ANYOF. Handles complementing the result if
1008 * appropriate. If some code points aren't knowable at this time, the
1009 * returned list must, and will, contain every code point that is a
1012 SV* invlist = sv_2mortal(_new_invlist(0));
1013 SV* only_utf8_locale_invlist = NULL;
1015 const U32 n = ARG(node);
1016 bool new_node_has_latin1 = FALSE;
1018 PERL_ARGS_ASSERT_GET_ANYOF_CP_LIST_FOR_SSC;
1020 /* Look at the data structure created by S_set_ANYOF_arg() */
1021 if (n != ANYOF_NONBITMAP_EMPTY) {
1022 SV * const rv = MUTABLE_SV(RExC_rxi->data->data[n]);
1023 AV * const av = MUTABLE_AV(SvRV(rv));
1024 SV **const ary = AvARRAY(av);
1025 assert(RExC_rxi->data->what[n] == 's');
1027 if (ary[1] && ary[1] != &PL_sv_undef) { /* Has compile-time swash */
1028 invlist = sv_2mortal(invlist_clone(_get_swash_invlist(ary[1])));
1030 else if (ary[0] && ary[0] != &PL_sv_undef) {
1032 /* Here, no compile-time swash, and there are things that won't be
1033 * known until runtime -- we have to assume it could be anything */
1034 return _add_range_to_invlist(invlist, 0, UV_MAX);
1036 else if (ary[3] && ary[3] != &PL_sv_undef) {
1038 /* Here no compile-time swash, and no run-time only data. Use the
1039 * node's inversion list */
1040 invlist = sv_2mortal(invlist_clone(ary[3]));
1043 /* Get the code points valid only under UTF-8 locales */
1044 if ((ANYOF_FLAGS(node) & ANYOF_LOC_FOLD)
1045 && ary[2] && ary[2] != &PL_sv_undef)
1047 only_utf8_locale_invlist = ary[2];
1051 /* An ANYOF node contains a bitmap for the first 256 code points, and an
1052 * inversion list for the others, but if there are code points that should
1053 * match only conditionally on the target string being UTF-8, those are
1054 * placed in the inversion list, and not the bitmap. Since there are
1055 * circumstances under which they could match, they are included in the
1056 * SSC. But if the ANYOF node is to be inverted, we have to exclude them
1057 * here, so that when we invert below, the end result actually does include
1058 * them. (Think about "\xe0" =~ /[^\xc0]/di;). We have to do this here
1059 * before we add the unconditionally matched code points */
1060 if (ANYOF_FLAGS(node) & ANYOF_INVERT) {
1061 _invlist_intersection_complement_2nd(invlist,
1066 /* Add in the points from the bit map */
1067 for (i = 0; i < 256; i++) {
1068 if (ANYOF_BITMAP_TEST(node, i)) {
1069 invlist = add_cp_to_invlist(invlist, i);
1070 new_node_has_latin1 = TRUE;
1074 /* If this can match all upper Latin1 code points, have to add them
1076 if (ANYOF_FLAGS(node) & ANYOF_NON_UTF8_NON_ASCII_ALL) {
1077 _invlist_union(invlist, PL_UpperLatin1, &invlist);
1080 /* Similarly for these */
1081 if (ANYOF_FLAGS(node) & ANYOF_ABOVE_LATIN1_ALL) {
1082 invlist = _add_range_to_invlist(invlist, 256, UV_MAX);
1085 if (ANYOF_FLAGS(node) & ANYOF_INVERT) {
1086 _invlist_invert(invlist);
1088 else if (new_node_has_latin1 && ANYOF_FLAGS(node) & ANYOF_LOC_FOLD) {
1090 /* Under /li, any 0-255 could fold to any other 0-255, depending on the
1091 * locale. We can skip this if there are no 0-255 at all. */
1092 _invlist_union(invlist, PL_Latin1, &invlist);
1095 /* Similarly add the UTF-8 locale possible matches. These have to be
1096 * deferred until after the non-UTF-8 locale ones are taken care of just
1097 * above, or it leads to wrong results under ANYOF_INVERT */
1098 if (only_utf8_locale_invlist) {
1099 _invlist_union_maybe_complement_2nd(invlist,
1100 only_utf8_locale_invlist,
1101 ANYOF_FLAGS(node) & ANYOF_INVERT,
1108 /* These two functions currently do the exact same thing */
1109 #define ssc_init_zero ssc_init
1111 #define ssc_add_cp(ssc, cp) ssc_add_range((ssc), (cp), (cp))
1112 #define ssc_match_all_cp(ssc) ssc_add_range(ssc, 0, UV_MAX)
1114 /* 'AND' a given class with another one. Can create false positives. 'ssc'
1115 * should not be inverted. 'and_with->flags & ANYOF_POSIXL' should be 0 if
1116 * 'and_with' is a regnode_charclass instead of a regnode_ssc. */
1119 S_ssc_and(pTHX_ const RExC_state_t *pRExC_state, regnode_ssc *ssc,
1120 const regnode_charclass *and_with)
1122 /* Accumulate into SSC 'ssc' its 'AND' with 'and_with', which is either
1123 * another SSC or a regular ANYOF class. Can create false positives. */
1128 PERL_ARGS_ASSERT_SSC_AND;
1130 assert(is_ANYOF_SYNTHETIC(ssc));
1132 /* 'and_with' is used as-is if it too is an SSC; otherwise have to extract
1133 * the code point inversion list and just the relevant flags */
1134 if (is_ANYOF_SYNTHETIC(and_with)) {
1135 anded_cp_list = ((regnode_ssc *)and_with)->invlist;
1136 anded_flags = ANYOF_FLAGS(and_with);
1138 /* XXX This is a kludge around what appears to be deficiencies in the
1139 * optimizer. If we make S_ssc_anything() add in the WARN_SUPER flag,
1140 * there are paths through the optimizer where it doesn't get weeded
1141 * out when it should. And if we don't make some extra provision for
1142 * it like the code just below, it doesn't get added when it should.
1143 * This solution is to add it only when AND'ing, which is here, and
1144 * only when what is being AND'ed is the pristine, original node
1145 * matching anything. Thus it is like adding it to ssc_anything() but
1146 * only when the result is to be AND'ed. Probably the same solution
1147 * could be adopted for the same problem we have with /l matching,
1148 * which is solved differently in S_ssc_init(), and that would lead to
1149 * fewer false positives than that solution has. But if this solution
1150 * creates bugs, the consequences are only that a warning isn't raised
1151 * that should be; while the consequences for having /l bugs is
1152 * incorrect matches */
1153 if (ssc_is_anything((regnode_ssc *)and_with)) {
1154 anded_flags |= ANYOF_WARN_SUPER;
1158 anded_cp_list = get_ANYOF_cp_list_for_ssc(pRExC_state, and_with);
1159 anded_flags = ANYOF_FLAGS(and_with) & ANYOF_COMMON_FLAGS;
1162 ANYOF_FLAGS(ssc) &= anded_flags;
1164 /* Below, C1 is the list of code points in 'ssc'; P1, its posix classes.
1165 * C2 is the list of code points in 'and-with'; P2, its posix classes.
1166 * 'and_with' may be inverted. When not inverted, we have the situation of
1168 * (C1 | P1) & (C2 | P2)
1169 * = (C1 & (C2 | P2)) | (P1 & (C2 | P2))
1170 * = ((C1 & C2) | (C1 & P2)) | ((P1 & C2) | (P1 & P2))
1171 * <= ((C1 & C2) | P2)) | ( P1 | (P1 & P2))
1172 * <= ((C1 & C2) | P1 | P2)
1173 * Alternatively, the last few steps could be:
1174 * = ((C1 & C2) | (C1 & P2)) | ((P1 & C2) | (P1 & P2))
1175 * <= ((C1 & C2) | C1 ) | ( C2 | (P1 & P2))
1176 * <= (C1 | C2 | (P1 & P2))
1177 * We favor the second approach if either P1 or P2 is non-empty. This is
1178 * because these components are a barrier to doing optimizations, as what
1179 * they match cannot be known until the moment of matching as they are
1180 * dependent on the current locale, 'AND"ing them likely will reduce or
1182 * But we can do better if we know that C1,P1 are in their initial state (a
1183 * frequent occurrence), each matching everything:
1184 * (<everything>) & (C2 | P2) = C2 | P2
1185 * Similarly, if C2,P2 are in their initial state (again a frequent
1186 * occurrence), the result is a no-op
1187 * (C1 | P1) & (<everything>) = C1 | P1
1190 * (C1 | P1) & ~(C2 | P2) = (C1 | P1) & (~C2 & ~P2)
1191 * = (C1 & (~C2 & ~P2)) | (P1 & (~C2 & ~P2))
1192 * <= (C1 & ~C2) | (P1 & ~P2)
1195 if ((ANYOF_FLAGS(and_with) & ANYOF_INVERT)
1196 && ! is_ANYOF_SYNTHETIC(and_with))
1200 ssc_intersection(ssc,
1202 FALSE /* Has already been inverted */
1205 /* If either P1 or P2 is empty, the intersection will be also; can skip
1207 if (! (ANYOF_FLAGS(and_with) & ANYOF_POSIXL)) {
1208 ANYOF_POSIXL_ZERO(ssc);
1210 else if (ANYOF_POSIXL_SSC_TEST_ANY_SET(ssc)) {
1212 /* Note that the Posix class component P from 'and_with' actually
1214 * P = Pa | Pb | ... | Pn
1215 * where each component is one posix class, such as in [\w\s].
1217 * ~P = ~(Pa | Pb | ... | Pn)
1218 * = ~Pa & ~Pb & ... & ~Pn
1219 * <= ~Pa | ~Pb | ... | ~Pn
1220 * The last is something we can easily calculate, but unfortunately
1221 * is likely to have many false positives. We could do better
1222 * in some (but certainly not all) instances if two classes in
1223 * P have known relationships. For example
1224 * :lower: <= :alpha: <= :alnum: <= \w <= :graph: <= :print:
1226 * :lower: & :print: = :lower:
1227 * And similarly for classes that must be disjoint. For example,
1228 * since \s and \w can have no elements in common based on rules in
1229 * the POSIX standard,
1230 * \w & ^\S = nothing
1231 * Unfortunately, some vendor locales do not meet the Posix
1232 * standard, in particular almost everything by Microsoft.
1233 * The loop below just changes e.g., \w into \W and vice versa */
1235 regnode_charclass_posixl temp;
1236 int add = 1; /* To calculate the index of the complement */
1238 ANYOF_POSIXL_ZERO(&temp);
1239 for (i = 0; i < ANYOF_MAX; i++) {
1241 || ! ANYOF_POSIXL_TEST((regnode_charclass_posixl*) and_with, i)
1242 || ! ANYOF_POSIXL_TEST((regnode_charclass_posixl*) and_with, i + 1));
1244 if (ANYOF_POSIXL_TEST((regnode_charclass_posixl*) and_with, i)) {
1245 ANYOF_POSIXL_SET(&temp, i + add);
1247 add = 0 - add; /* 1 goes to -1; -1 goes to 1 */
1249 ANYOF_POSIXL_AND(&temp, ssc);
1251 } /* else ssc already has no posixes */
1252 } /* else: Not inverted. This routine is a no-op if 'and_with' is an SSC
1253 in its initial state */
1254 else if (! is_ANYOF_SYNTHETIC(and_with)
1255 || ! ssc_is_cp_posixl_init(pRExC_state, (regnode_ssc *)and_with))
1257 /* But if 'ssc' is in its initial state, the result is just 'and_with';
1258 * copy it over 'ssc' */
1259 if (ssc_is_cp_posixl_init(pRExC_state, ssc)) {
1260 if (is_ANYOF_SYNTHETIC(and_with)) {
1261 StructCopy(and_with, ssc, regnode_ssc);
1264 ssc->invlist = anded_cp_list;
1265 ANYOF_POSIXL_ZERO(ssc);
1266 if (ANYOF_FLAGS(and_with) & ANYOF_POSIXL) {
1267 ANYOF_POSIXL_OR((regnode_charclass_posixl*) and_with, ssc);
1271 else if (ANYOF_POSIXL_SSC_TEST_ANY_SET(ssc)
1272 || (ANYOF_FLAGS(and_with) & ANYOF_POSIXL))
1274 /* One or the other of P1, P2 is non-empty. */
1275 if (ANYOF_FLAGS(and_with) & ANYOF_POSIXL) {
1276 ANYOF_POSIXL_AND((regnode_charclass_posixl*) and_with, ssc);
1278 ssc_union(ssc, anded_cp_list, FALSE);
1280 else { /* P1 = P2 = empty */
1281 ssc_intersection(ssc, anded_cp_list, FALSE);
1287 S_ssc_or(pTHX_ const RExC_state_t *pRExC_state, regnode_ssc *ssc,
1288 const regnode_charclass *or_with)
1290 /* Accumulate into SSC 'ssc' its 'OR' with 'or_with', which is either
1291 * another SSC or a regular ANYOF class. Can create false positives if
1292 * 'or_with' is to be inverted. */
1297 PERL_ARGS_ASSERT_SSC_OR;
1299 assert(is_ANYOF_SYNTHETIC(ssc));
1301 /* 'or_with' is used as-is if it too is an SSC; otherwise have to extract
1302 * the code point inversion list and just the relevant flags */
1303 if (is_ANYOF_SYNTHETIC(or_with)) {
1304 ored_cp_list = ((regnode_ssc*) or_with)->invlist;
1305 ored_flags = ANYOF_FLAGS(or_with);
1308 ored_cp_list = get_ANYOF_cp_list_for_ssc(pRExC_state, or_with);
1309 ored_flags = ANYOF_FLAGS(or_with) & ANYOF_COMMON_FLAGS;
1312 ANYOF_FLAGS(ssc) |= ored_flags;
1314 /* Below, C1 is the list of code points in 'ssc'; P1, its posix classes.
1315 * C2 is the list of code points in 'or-with'; P2, its posix classes.
1316 * 'or_with' may be inverted. When not inverted, we have the simple
1317 * situation of computing:
1318 * (C1 | P1) | (C2 | P2) = (C1 | C2) | (P1 | P2)
1319 * If P1|P2 yields a situation with both a class and its complement are
1320 * set, like having both \w and \W, this matches all code points, and we
1321 * can delete these from the P component of the ssc going forward. XXX We
1322 * might be able to delete all the P components, but I (khw) am not certain
1323 * about this, and it is better to be safe.
1326 * (C1 | P1) | ~(C2 | P2) = (C1 | P1) | (~C2 & ~P2)
1327 * <= (C1 | P1) | ~C2
1328 * <= (C1 | ~C2) | P1
1329 * (which results in actually simpler code than the non-inverted case)
1332 if ((ANYOF_FLAGS(or_with) & ANYOF_INVERT)
1333 && ! is_ANYOF_SYNTHETIC(or_with))
1335 /* We ignore P2, leaving P1 going forward */
1336 } /* else Not inverted */
1337 else if (ANYOF_FLAGS(or_with) & ANYOF_POSIXL) {
1338 ANYOF_POSIXL_OR((regnode_charclass_posixl*)or_with, ssc);
1339 if (ANYOF_POSIXL_SSC_TEST_ANY_SET(ssc)) {
1341 for (i = 0; i < ANYOF_MAX; i += 2) {
1342 if (ANYOF_POSIXL_TEST(ssc, i) && ANYOF_POSIXL_TEST(ssc, i + 1))
1344 ssc_match_all_cp(ssc);
1345 ANYOF_POSIXL_CLEAR(ssc, i);
1346 ANYOF_POSIXL_CLEAR(ssc, i+1);
1354 FALSE /* Already has been inverted */
1358 PERL_STATIC_INLINE void
1359 S_ssc_union(pTHX_ regnode_ssc *ssc, SV* const invlist, const bool invert2nd)
1361 PERL_ARGS_ASSERT_SSC_UNION;
1363 assert(is_ANYOF_SYNTHETIC(ssc));
1365 _invlist_union_maybe_complement_2nd(ssc->invlist,
1371 PERL_STATIC_INLINE void
1372 S_ssc_intersection(pTHX_ regnode_ssc *ssc,
1374 const bool invert2nd)
1376 PERL_ARGS_ASSERT_SSC_INTERSECTION;
1378 assert(is_ANYOF_SYNTHETIC(ssc));
1380 _invlist_intersection_maybe_complement_2nd(ssc->invlist,
1386 PERL_STATIC_INLINE void
1387 S_ssc_add_range(pTHX_ regnode_ssc *ssc, const UV start, const UV end)
1389 PERL_ARGS_ASSERT_SSC_ADD_RANGE;
1391 assert(is_ANYOF_SYNTHETIC(ssc));
1393 ssc->invlist = _add_range_to_invlist(ssc->invlist, start, end);
1396 PERL_STATIC_INLINE void
1397 S_ssc_cp_and(pTHX_ regnode_ssc *ssc, const UV cp)
1399 /* AND just the single code point 'cp' into the SSC 'ssc' */
1401 SV* cp_list = _new_invlist(2);
1403 PERL_ARGS_ASSERT_SSC_CP_AND;
1405 assert(is_ANYOF_SYNTHETIC(ssc));
1407 cp_list = add_cp_to_invlist(cp_list, cp);
1408 ssc_intersection(ssc, cp_list,
1409 FALSE /* Not inverted */
1411 SvREFCNT_dec_NN(cp_list);
1414 PERL_STATIC_INLINE void
1415 S_ssc_clear_locale(regnode_ssc *ssc)
1417 /* Set the SSC 'ssc' to not match any locale things */
1418 PERL_ARGS_ASSERT_SSC_CLEAR_LOCALE;
1420 assert(is_ANYOF_SYNTHETIC(ssc));
1422 ANYOF_POSIXL_ZERO(ssc);
1423 ANYOF_FLAGS(ssc) &= ~ANYOF_LOCALE_FLAGS;
1427 S_ssc_finalize(pTHX_ RExC_state_t *pRExC_state, regnode_ssc *ssc)
1429 /* The inversion list in the SSC is marked mortal; now we need a more
1430 * permanent copy, which is stored the same way that is done in a regular
1431 * ANYOF node, with the first 256 code points in a bit map */
1433 SV* invlist = invlist_clone(ssc->invlist);
1435 PERL_ARGS_ASSERT_SSC_FINALIZE;
1437 assert(is_ANYOF_SYNTHETIC(ssc));
1439 /* The code in this file assumes that all but these flags aren't relevant
1440 * to the SSC, except ANYOF_EMPTY_STRING, which should be cleared by the
1441 * time we reach here */
1442 assert(! (ANYOF_FLAGS(ssc) & ~ANYOF_COMMON_FLAGS));
1444 populate_ANYOF_from_invlist( (regnode *) ssc, &invlist);
1446 set_ANYOF_arg(pRExC_state, (regnode *) ssc, invlist,
1447 NULL, NULL, NULL, FALSE);
1449 /* Make sure is clone-safe */
1450 ssc->invlist = NULL;
1452 if (ANYOF_POSIXL_SSC_TEST_ANY_SET(ssc)) {
1453 ANYOF_FLAGS(ssc) |= ANYOF_POSIXL;
1456 assert(! (ANYOF_FLAGS(ssc) & ANYOF_LOCALE_FLAGS) || RExC_contains_locale);
1459 #define TRIE_LIST_ITEM(state,idx) (trie->states[state].trans.list)[ idx ]
1460 #define TRIE_LIST_CUR(state) ( TRIE_LIST_ITEM( state, 0 ).forid )
1461 #define TRIE_LIST_LEN(state) ( TRIE_LIST_ITEM( state, 0 ).newstate )
1462 #define TRIE_LIST_USED(idx) ( trie->states[state].trans.list \
1463 ? (TRIE_LIST_CUR( idx ) - 1) \
1469 dump_trie(trie,widecharmap,revcharmap)
1470 dump_trie_interim_list(trie,widecharmap,revcharmap,next_alloc)
1471 dump_trie_interim_table(trie,widecharmap,revcharmap,next_alloc)
1473 These routines dump out a trie in a somewhat readable format.
1474 The _interim_ variants are used for debugging the interim
1475 tables that are used to generate the final compressed
1476 representation which is what dump_trie expects.
1478 Part of the reason for their existence is to provide a form
1479 of documentation as to how the different representations function.
1484 Dumps the final compressed table form of the trie to Perl_debug_log.
1485 Used for debugging make_trie().
1489 S_dump_trie(pTHX_ const struct _reg_trie_data *trie, HV *widecharmap,
1490 AV *revcharmap, U32 depth)
1493 SV *sv=sv_newmortal();
1494 int colwidth= widecharmap ? 6 : 4;
1496 GET_RE_DEBUG_FLAGS_DECL;
1498 PERL_ARGS_ASSERT_DUMP_TRIE;
1500 PerlIO_printf( Perl_debug_log, "%*sChar : %-6s%-6s%-4s ",
1501 (int)depth * 2 + 2,"",
1502 "Match","Base","Ofs" );
1504 for( state = 0 ; state < trie->uniquecharcount ; state++ ) {
1505 SV ** const tmp = av_fetch( revcharmap, state, 0);
1507 PerlIO_printf( Perl_debug_log, "%*s",
1509 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
1510 PL_colors[0], PL_colors[1],
1511 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
1512 PERL_PV_ESCAPE_FIRSTCHAR
1517 PerlIO_printf( Perl_debug_log, "\n%*sState|-----------------------",
1518 (int)depth * 2 + 2,"");
1520 for( state = 0 ; state < trie->uniquecharcount ; state++ )
1521 PerlIO_printf( Perl_debug_log, "%.*s", colwidth, "--------");
1522 PerlIO_printf( Perl_debug_log, "\n");
1524 for( state = 1 ; state < trie->statecount ; state++ ) {
1525 const U32 base = trie->states[ state ].trans.base;
1527 PerlIO_printf( Perl_debug_log, "%*s#%4"UVXf"|",
1528 (int)depth * 2 + 2,"", (UV)state);
1530 if ( trie->states[ state ].wordnum ) {
1531 PerlIO_printf( Perl_debug_log, " W%4X",
1532 trie->states[ state ].wordnum );
1534 PerlIO_printf( Perl_debug_log, "%6s", "" );
1537 PerlIO_printf( Perl_debug_log, " @%4"UVXf" ", (UV)base );
1542 while( ( base + ofs < trie->uniquecharcount ) ||
1543 ( base + ofs - trie->uniquecharcount < trie->lasttrans
1544 && trie->trans[ base + ofs - trie->uniquecharcount ].check
1548 PerlIO_printf( Perl_debug_log, "+%2"UVXf"[ ", (UV)ofs);
1550 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
1551 if ( ( base + ofs >= trie->uniquecharcount )
1552 && ( base + ofs - trie->uniquecharcount
1554 && trie->trans[ base + ofs
1555 - trie->uniquecharcount ].check == state )
1557 PerlIO_printf( Perl_debug_log, "%*"UVXf,
1559 (UV)trie->trans[ base + ofs
1560 - trie->uniquecharcount ].next );
1562 PerlIO_printf( Perl_debug_log, "%*s",colwidth," ." );
1566 PerlIO_printf( Perl_debug_log, "]");
1569 PerlIO_printf( Perl_debug_log, "\n" );
1571 PerlIO_printf(Perl_debug_log, "%*sword_info N:(prev,len)=",
1573 for (word=1; word <= trie->wordcount; word++) {
1574 PerlIO_printf(Perl_debug_log, " %d:(%d,%d)",
1575 (int)word, (int)(trie->wordinfo[word].prev),
1576 (int)(trie->wordinfo[word].len));
1578 PerlIO_printf(Perl_debug_log, "\n" );
1581 Dumps a fully constructed but uncompressed trie in list form.
1582 List tries normally only are used for construction when the number of
1583 possible chars (trie->uniquecharcount) is very high.
1584 Used for debugging make_trie().
1587 S_dump_trie_interim_list(pTHX_ const struct _reg_trie_data *trie,
1588 HV *widecharmap, AV *revcharmap, U32 next_alloc,
1592 SV *sv=sv_newmortal();
1593 int colwidth= widecharmap ? 6 : 4;
1594 GET_RE_DEBUG_FLAGS_DECL;
1596 PERL_ARGS_ASSERT_DUMP_TRIE_INTERIM_LIST;
1598 /* print out the table precompression. */
1599 PerlIO_printf( Perl_debug_log, "%*sState :Word | Transition Data\n%*s%s",
1600 (int)depth * 2 + 2,"", (int)depth * 2 + 2,"",
1601 "------:-----+-----------------\n" );
1603 for( state=1 ; state < next_alloc ; state ++ ) {
1606 PerlIO_printf( Perl_debug_log, "%*s %4"UVXf" :",
1607 (int)depth * 2 + 2,"", (UV)state );
1608 if ( ! trie->states[ state ].wordnum ) {
1609 PerlIO_printf( Perl_debug_log, "%5s| ","");
1611 PerlIO_printf( Perl_debug_log, "W%4x| ",
1612 trie->states[ state ].wordnum
1615 for( charid = 1 ; charid <= TRIE_LIST_USED( state ) ; charid++ ) {
1616 SV ** const tmp = av_fetch( revcharmap,
1617 TRIE_LIST_ITEM(state,charid).forid, 0);
1619 PerlIO_printf( Perl_debug_log, "%*s:%3X=%4"UVXf" | ",
1621 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp),
1623 PL_colors[0], PL_colors[1],
1624 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)
1625 | PERL_PV_ESCAPE_FIRSTCHAR
1627 TRIE_LIST_ITEM(state,charid).forid,
1628 (UV)TRIE_LIST_ITEM(state,charid).newstate
1631 PerlIO_printf(Perl_debug_log, "\n%*s| ",
1632 (int)((depth * 2) + 14), "");
1635 PerlIO_printf( Perl_debug_log, "\n");
1640 Dumps a fully constructed but uncompressed trie in table form.
1641 This is the normal DFA style state transition table, with a few
1642 twists to facilitate compression later.
1643 Used for debugging make_trie().
1646 S_dump_trie_interim_table(pTHX_ const struct _reg_trie_data *trie,
1647 HV *widecharmap, AV *revcharmap, U32 next_alloc,
1652 SV *sv=sv_newmortal();
1653 int colwidth= widecharmap ? 6 : 4;
1654 GET_RE_DEBUG_FLAGS_DECL;
1656 PERL_ARGS_ASSERT_DUMP_TRIE_INTERIM_TABLE;
1659 print out the table precompression so that we can do a visual check
1660 that they are identical.
1663 PerlIO_printf( Perl_debug_log, "%*sChar : ",(int)depth * 2 + 2,"" );
1665 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
1666 SV ** const tmp = av_fetch( revcharmap, charid, 0);
1668 PerlIO_printf( Perl_debug_log, "%*s",
1670 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), colwidth,
1671 PL_colors[0], PL_colors[1],
1672 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
1673 PERL_PV_ESCAPE_FIRSTCHAR
1679 PerlIO_printf( Perl_debug_log, "\n%*sState+-",(int)depth * 2 + 2,"" );
1681 for( charid=0 ; charid < trie->uniquecharcount ; charid++ ) {
1682 PerlIO_printf( Perl_debug_log, "%.*s", colwidth,"--------");
1685 PerlIO_printf( Perl_debug_log, "\n" );
1687 for( state=1 ; state < next_alloc ; state += trie->uniquecharcount ) {
1689 PerlIO_printf( Perl_debug_log, "%*s%4"UVXf" : ",
1690 (int)depth * 2 + 2,"",
1691 (UV)TRIE_NODENUM( state ) );
1693 for( charid = 0 ; charid < trie->uniquecharcount ; charid++ ) {
1694 UV v=(UV)SAFE_TRIE_NODENUM( trie->trans[ state + charid ].next );
1696 PerlIO_printf( Perl_debug_log, "%*"UVXf, colwidth, v );
1698 PerlIO_printf( Perl_debug_log, "%*s", colwidth, "." );
1700 if ( ! trie->states[ TRIE_NODENUM( state ) ].wordnum ) {
1701 PerlIO_printf( Perl_debug_log, " (%4"UVXf")\n",
1702 (UV)trie->trans[ state ].check );
1704 PerlIO_printf( Perl_debug_log, " (%4"UVXf") W%4X\n",
1705 (UV)trie->trans[ state ].check,
1706 trie->states[ TRIE_NODENUM( state ) ].wordnum );
1714 /* make_trie(startbranch,first,last,tail,word_count,flags,depth)
1715 startbranch: the first branch in the whole branch sequence
1716 first : start branch of sequence of branch-exact nodes.
1717 May be the same as startbranch
1718 last : Thing following the last branch.
1719 May be the same as tail.
1720 tail : item following the branch sequence
1721 count : words in the sequence
1722 flags : currently the OP() type we will be building one of /EXACT(|F|FA|FU|FU_SS)/
1723 depth : indent depth
1725 Inplace optimizes a sequence of 2 or more Branch-Exact nodes into a TRIE node.
1727 A trie is an N'ary tree where the branches are determined by digital
1728 decomposition of the key. IE, at the root node you look up the 1st character and
1729 follow that branch repeat until you find the end of the branches. Nodes can be
1730 marked as "accepting" meaning they represent a complete word. Eg:
1734 would convert into the following structure. Numbers represent states, letters
1735 following numbers represent valid transitions on the letter from that state, if
1736 the number is in square brackets it represents an accepting state, otherwise it
1737 will be in parenthesis.
1739 +-h->+-e->[3]-+-r->(8)-+-s->[9]
1743 (1) +-i->(6)-+-s->[7]
1745 +-s->(3)-+-h->(4)-+-e->[5]
1747 Accept Word Mapping: 3=>1 (he),5=>2 (she), 7=>3 (his), 9=>4 (hers)
1749 This shows that when matching against the string 'hers' we will begin at state 1
1750 read 'h' and move to state 2, read 'e' and move to state 3 which is accepting,
1751 then read 'r' and go to state 8 followed by 's' which takes us to state 9 which
1752 is also accepting. Thus we know that we can match both 'he' and 'hers' with a
1753 single traverse. We store a mapping from accepting to state to which word was
1754 matched, and then when we have multiple possibilities we try to complete the
1755 rest of the regex in the order in which they occured in the alternation.
1757 The only prior NFA like behaviour that would be changed by the TRIE support is
1758 the silent ignoring of duplicate alternations which are of the form:
1760 / (DUPE|DUPE) X? (?{ ... }) Y /x
1762 Thus EVAL blocks following a trie may be called a different number of times with
1763 and without the optimisation. With the optimisations dupes will be silently
1764 ignored. This inconsistent behaviour of EVAL type nodes is well established as
1765 the following demonstrates:
1767 'words'=~/(word|word|word)(?{ print $1 })[xyz]/
1769 which prints out 'word' three times, but
1771 'words'=~/(word|word|word)(?{ print $1 })S/
1773 which doesnt print it out at all. This is due to other optimisations kicking in.
1775 Example of what happens on a structural level:
1777 The regexp /(ac|ad|ab)+/ will produce the following debug output:
1779 1: CURLYM[1] {1,32767}(18)
1790 This would be optimizable with startbranch=5, first=5, last=16, tail=16
1791 and should turn into:
1793 1: CURLYM[1] {1,32767}(18)
1795 [Words:3 Chars Stored:6 Unique Chars:4 States:5 NCP:1]
1803 Cases where tail != last would be like /(?foo|bar)baz/:
1813 which would be optimizable with startbranch=1, first=1, last=7, tail=8
1814 and would end up looking like:
1817 [Words:2 Chars Stored:6 Unique Chars:5 States:7 NCP:1]
1824 d = uvchr_to_utf8_flags(d, uv, 0);
1826 is the recommended Unicode-aware way of saying
1831 #define TRIE_STORE_REVCHAR(val) \
1834 SV *zlopp = newSV(7); /* XXX: optimize me */ \
1835 unsigned char *flrbbbbb = (unsigned char *) SvPVX(zlopp); \
1836 unsigned const char *const kapow = uvchr_to_utf8(flrbbbbb, val); \
1837 SvCUR_set(zlopp, kapow - flrbbbbb); \
1840 av_push(revcharmap, zlopp); \
1842 char ooooff = (char)val; \
1843 av_push(revcharmap, newSVpvn(&ooooff, 1)); \
1847 /* This gets the next character from the input, folding it if not already
1849 #define TRIE_READ_CHAR STMT_START { \
1852 /* if it is UTF then it is either already folded, or does not need \
1854 uvc = valid_utf8_to_uvchr( (const U8*) uc, &len); \
1856 else if (folder == PL_fold_latin1) { \
1857 /* This folder implies Unicode rules, which in the range expressible \
1858 * by not UTF is the lower case, with the two exceptions, one of \
1859 * which should have been taken care of before calling this */ \
1860 assert(*uc != LATIN_SMALL_LETTER_SHARP_S); \
1861 uvc = toLOWER_L1(*uc); \
1862 if (UNLIKELY(uvc == MICRO_SIGN)) uvc = GREEK_SMALL_LETTER_MU; \
1865 /* raw data, will be folded later if needed */ \
1873 #define TRIE_LIST_PUSH(state,fid,ns) STMT_START { \
1874 if ( TRIE_LIST_CUR( state ) >=TRIE_LIST_LEN( state ) ) { \
1875 U32 ging = TRIE_LIST_LEN( state ) *= 2; \
1876 Renew( trie->states[ state ].trans.list, ging, reg_trie_trans_le ); \
1878 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).forid = fid; \
1879 TRIE_LIST_ITEM( state, TRIE_LIST_CUR( state ) ).newstate = ns; \
1880 TRIE_LIST_CUR( state )++; \
1883 #define TRIE_LIST_NEW(state) STMT_START { \
1884 Newxz( trie->states[ state ].trans.list, \
1885 4, reg_trie_trans_le ); \
1886 TRIE_LIST_CUR( state ) = 1; \
1887 TRIE_LIST_LEN( state ) = 4; \
1890 #define TRIE_HANDLE_WORD(state) STMT_START { \
1891 U16 dupe= trie->states[ state ].wordnum; \
1892 regnode * const noper_next = regnext( noper ); \
1895 /* store the word for dumping */ \
1897 if (OP(noper) != NOTHING) \
1898 tmp = newSVpvn_utf8(STRING(noper), STR_LEN(noper), UTF); \
1900 tmp = newSVpvn_utf8( "", 0, UTF ); \
1901 av_push( trie_words, tmp ); \
1905 trie->wordinfo[curword].prev = 0; \
1906 trie->wordinfo[curword].len = wordlen; \
1907 trie->wordinfo[curword].accept = state; \
1909 if ( noper_next < tail ) { \
1911 trie->jump = (U16 *) PerlMemShared_calloc( word_count + 1, \
1913 trie->jump[curword] = (U16)(noper_next - convert); \
1915 jumper = noper_next; \
1917 nextbranch= regnext(cur); \
1921 /* It's a dupe. Pre-insert into the wordinfo[].prev */\
1922 /* chain, so that when the bits of chain are later */\
1923 /* linked together, the dups appear in the chain */\
1924 trie->wordinfo[curword].prev = trie->wordinfo[dupe].prev; \
1925 trie->wordinfo[dupe].prev = curword; \
1927 /* we haven't inserted this word yet. */ \
1928 trie->states[ state ].wordnum = curword; \
1933 #define TRIE_TRANS_STATE(state,base,ucharcount,charid,special) \
1934 ( ( base + charid >= ucharcount \
1935 && base + charid < ubound \
1936 && state == trie->trans[ base - ucharcount + charid ].check \
1937 && trie->trans[ base - ucharcount + charid ].next ) \
1938 ? trie->trans[ base - ucharcount + charid ].next \
1939 : ( state==1 ? special : 0 ) \
1943 #define MADE_JUMP_TRIE 2
1944 #define MADE_EXACT_TRIE 4
1947 S_make_trie(pTHX_ RExC_state_t *pRExC_state, regnode *startbranch,
1948 regnode *first, regnode *last, regnode *tail,
1949 U32 word_count, U32 flags, U32 depth)
1951 /* first pass, loop through and scan words */
1952 reg_trie_data *trie;
1953 HV *widecharmap = NULL;
1954 AV *revcharmap = newAV();
1960 regnode *jumper = NULL;
1961 regnode *nextbranch = NULL;
1962 regnode *convert = NULL;
1963 U32 *prev_states; /* temp array mapping each state to previous one */
1964 /* we just use folder as a flag in utf8 */
1965 const U8 * folder = NULL;
1968 const U32 data_slot = add_data( pRExC_state, STR_WITH_LEN("tuuu"));
1969 AV *trie_words = NULL;
1970 /* along with revcharmap, this only used during construction but both are
1971 * useful during debugging so we store them in the struct when debugging.
1974 const U32 data_slot = add_data( pRExC_state, STR_WITH_LEN("tu"));
1975 STRLEN trie_charcount=0;
1977 SV *re_trie_maxbuff;
1978 GET_RE_DEBUG_FLAGS_DECL;
1980 PERL_ARGS_ASSERT_MAKE_TRIE;
1982 PERL_UNUSED_ARG(depth);
1989 case EXACTFU: folder = PL_fold_latin1; break;
1990 case EXACTF: folder = PL_fold; break;
1991 default: Perl_croak( aTHX_ "panic! In trie construction, unknown node type %u %s", (unsigned) flags, PL_reg_name[flags] );
1994 trie = (reg_trie_data *) PerlMemShared_calloc( 1, sizeof(reg_trie_data) );
1996 trie->startstate = 1;
1997 trie->wordcount = word_count;
1998 RExC_rxi->data->data[ data_slot ] = (void*)trie;
1999 trie->charmap = (U16 *) PerlMemShared_calloc( 256, sizeof(U16) );
2001 trie->bitmap = (char *) PerlMemShared_calloc( ANYOF_BITMAP_SIZE, 1 );
2002 trie->wordinfo = (reg_trie_wordinfo *) PerlMemShared_calloc(
2003 trie->wordcount+1, sizeof(reg_trie_wordinfo));
2006 trie_words = newAV();
2009 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
2010 assert(re_trie_maxbuff);
2011 if (!SvIOK(re_trie_maxbuff)) {
2012 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
2014 DEBUG_TRIE_COMPILE_r({
2015 PerlIO_printf( Perl_debug_log,
2016 "%*smake_trie start==%d, first==%d, last==%d, tail==%d depth=%d\n",
2017 (int)depth * 2 + 2, "",
2018 REG_NODE_NUM(startbranch),REG_NODE_NUM(first),
2019 REG_NODE_NUM(last), REG_NODE_NUM(tail), (int)depth);
2022 /* Find the node we are going to overwrite */
2023 if ( first == startbranch && OP( last ) != BRANCH ) {
2024 /* whole branch chain */
2027 /* branch sub-chain */
2028 convert = NEXTOPER( first );
2031 /* -- First loop and Setup --
2033 We first traverse the branches and scan each word to determine if it
2034 contains widechars, and how many unique chars there are, this is
2035 important as we have to build a table with at least as many columns as we
2038 We use an array of integers to represent the character codes 0..255
2039 (trie->charmap) and we use a an HV* to store Unicode characters. We use
2040 the native representation of the character value as the key and IV's for
2043 *TODO* If we keep track of how many times each character is used we can
2044 remap the columns so that the table compression later on is more
2045 efficient in terms of memory by ensuring the most common value is in the
2046 middle and the least common are on the outside. IMO this would be better
2047 than a most to least common mapping as theres a decent chance the most
2048 common letter will share a node with the least common, meaning the node
2049 will not be compressible. With a middle is most common approach the worst
2050 case is when we have the least common nodes twice.
2054 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
2055 regnode *noper = NEXTOPER( cur );
2056 const U8 *uc = (U8*)STRING( noper );
2057 const U8 *e = uc + STR_LEN( noper );
2059 U32 wordlen = 0; /* required init */
2060 STRLEN minchars = 0;
2061 STRLEN maxchars = 0;
2062 bool set_bit = trie->bitmap ? 1 : 0; /*store the first char in the
2065 if (OP(noper) == NOTHING) {
2066 regnode *noper_next= regnext(noper);
2067 if (noper_next != tail && OP(noper_next) == flags) {
2069 uc= (U8*)STRING(noper);
2070 e= uc + STR_LEN(noper);
2071 trie->minlen= STR_LEN(noper);
2078 if ( set_bit ) { /* bitmap only alloced when !(UTF&&Folding) */
2079 TRIE_BITMAP_SET(trie,*uc); /* store the raw first byte
2080 regardless of encoding */
2081 if (OP( noper ) == EXACTFU_SS) {
2082 /* false positives are ok, so just set this */
2083 TRIE_BITMAP_SET(trie, LATIN_SMALL_LETTER_SHARP_S);
2086 for ( ; uc < e ; uc += len ) { /* Look at each char in the current
2088 TRIE_CHARCOUNT(trie)++;
2091 /* TRIE_READ_CHAR returns the current character, or its fold if /i
2092 * is in effect. Under /i, this character can match itself, or
2093 * anything that folds to it. If not under /i, it can match just
2094 * itself. Most folds are 1-1, for example k, K, and KELVIN SIGN
2095 * all fold to k, and all are single characters. But some folds
2096 * expand to more than one character, so for example LATIN SMALL
2097 * LIGATURE FFI folds to the three character sequence 'ffi'. If
2098 * the string beginning at 'uc' is 'ffi', it could be matched by
2099 * three characters, or just by the one ligature character. (It
2100 * could also be matched by two characters: LATIN SMALL LIGATURE FF
2101 * followed by 'i', or by 'f' followed by LATIN SMALL LIGATURE FI).
2102 * (Of course 'I' and/or 'F' instead of 'i' and 'f' can also
2103 * match.) The trie needs to know the minimum and maximum number
2104 * of characters that could match so that it can use size alone to
2105 * quickly reject many match attempts. The max is simple: it is
2106 * the number of folded characters in this branch (since a fold is
2107 * never shorter than what folds to it. */
2111 /* And the min is equal to the max if not under /i (indicated by
2112 * 'folder' being NULL), or there are no multi-character folds. If
2113 * there is a multi-character fold, the min is incremented just
2114 * once, for the character that folds to the sequence. Each
2115 * character in the sequence needs to be added to the list below of
2116 * characters in the trie, but we count only the first towards the
2117 * min number of characters needed. This is done through the
2118 * variable 'foldlen', which is returned by the macros that look
2119 * for these sequences as the number of bytes the sequence
2120 * occupies. Each time through the loop, we decrement 'foldlen' by
2121 * how many bytes the current char occupies. Only when it reaches
2122 * 0 do we increment 'minchars' or look for another multi-character
2124 if (folder == NULL) {
2127 else if (foldlen > 0) {
2128 foldlen -= (UTF) ? UTF8SKIP(uc) : 1;
2133 /* See if *uc is the beginning of a multi-character fold. If
2134 * so, we decrement the length remaining to look at, to account
2135 * for the current character this iteration. (We can use 'uc'
2136 * instead of the fold returned by TRIE_READ_CHAR because for
2137 * non-UTF, the latin1_safe macro is smart enough to account
2138 * for all the unfolded characters, and because for UTF, the
2139 * string will already have been folded earlier in the
2140 * compilation process */
2142 if ((foldlen = is_MULTI_CHAR_FOLD_utf8_safe(uc, e))) {
2143 foldlen -= UTF8SKIP(uc);
2146 else if ((foldlen = is_MULTI_CHAR_FOLD_latin1_safe(uc, e))) {
2151 /* The current character (and any potential folds) should be added
2152 * to the possible matching characters for this position in this
2156 U8 folded= folder[ (U8) uvc ];
2157 if ( !trie->charmap[ folded ] ) {
2158 trie->charmap[ folded ]=( ++trie->uniquecharcount );
2159 TRIE_STORE_REVCHAR( folded );
2162 if ( !trie->charmap[ uvc ] ) {
2163 trie->charmap[ uvc ]=( ++trie->uniquecharcount );
2164 TRIE_STORE_REVCHAR( uvc );
2167 /* store the codepoint in the bitmap, and its folded
2169 TRIE_BITMAP_SET(trie, uvc);
2171 /* store the folded codepoint */
2172 if ( folder ) TRIE_BITMAP_SET(trie, folder[(U8) uvc ]);
2175 /* store first byte of utf8 representation of
2176 variant codepoints */
2177 if (! UVCHR_IS_INVARIANT(uvc)) {
2178 TRIE_BITMAP_SET(trie, UTF8_TWO_BYTE_HI(uvc));
2181 set_bit = 0; /* We've done our bit :-) */
2185 /* XXX We could come up with the list of code points that fold
2186 * to this using PL_utf8_foldclosures, except not for
2187 * multi-char folds, as there may be multiple combinations
2188 * there that could work, which needs to wait until runtime to
2189 * resolve (The comment about LIGATURE FFI above is such an
2194 widecharmap = newHV();
2196 svpp = hv_fetch( widecharmap, (char*)&uvc, sizeof( UV ), 1 );
2199 Perl_croak( aTHX_ "error creating/fetching widecharmap entry for 0x%"UVXf, uvc );
2201 if ( !SvTRUE( *svpp ) ) {
2202 sv_setiv( *svpp, ++trie->uniquecharcount );
2203 TRIE_STORE_REVCHAR(uvc);
2206 } /* end loop through characters in this branch of the trie */
2208 /* We take the min and max for this branch and combine to find the min
2209 * and max for all branches processed so far */
2210 if( cur == first ) {
2211 trie->minlen = minchars;
2212 trie->maxlen = maxchars;
2213 } else if (minchars < trie->minlen) {
2214 trie->minlen = minchars;
2215 } else if (maxchars > trie->maxlen) {
2216 trie->maxlen = maxchars;
2218 } /* end first pass */
2219 DEBUG_TRIE_COMPILE_r(
2220 PerlIO_printf( Perl_debug_log,
2221 "%*sTRIE(%s): W:%d C:%d Uq:%d Min:%d Max:%d\n",
2222 (int)depth * 2 + 2,"",
2223 ( widecharmap ? "UTF8" : "NATIVE" ), (int)word_count,
2224 (int)TRIE_CHARCOUNT(trie), trie->uniquecharcount,
2225 (int)trie->minlen, (int)trie->maxlen )
2229 We now know what we are dealing with in terms of unique chars and
2230 string sizes so we can calculate how much memory a naive
2231 representation using a flat table will take. If it's over a reasonable
2232 limit (as specified by ${^RE_TRIE_MAXBUF}) we use a more memory
2233 conservative but potentially much slower representation using an array
2236 At the end we convert both representations into the same compressed
2237 form that will be used in regexec.c for matching with. The latter
2238 is a form that cannot be used to construct with but has memory
2239 properties similar to the list form and access properties similar
2240 to the table form making it both suitable for fast searches and
2241 small enough that its feasable to store for the duration of a program.
2243 See the comment in the code where the compressed table is produced
2244 inplace from the flat tabe representation for an explanation of how
2245 the compression works.
2250 Newx(prev_states, TRIE_CHARCOUNT(trie) + 2, U32);
2253 if ( (IV)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount + 1)
2254 > SvIV(re_trie_maxbuff) )
2257 Second Pass -- Array Of Lists Representation
2259 Each state will be represented by a list of charid:state records
2260 (reg_trie_trans_le) the first such element holds the CUR and LEN
2261 points of the allocated array. (See defines above).
2263 We build the initial structure using the lists, and then convert
2264 it into the compressed table form which allows faster lookups
2265 (but cant be modified once converted).
2268 STRLEN transcount = 1;
2270 DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log,
2271 "%*sCompiling trie using list compiler\n",
2272 (int)depth * 2 + 2, ""));
2274 trie->states = (reg_trie_state *)
2275 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
2276 sizeof(reg_trie_state) );
2280 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
2282 regnode *noper = NEXTOPER( cur );
2283 U8 *uc = (U8*)STRING( noper );
2284 const U8 *e = uc + STR_LEN( noper );
2285 U32 state = 1; /* required init */
2286 U16 charid = 0; /* sanity init */
2287 U32 wordlen = 0; /* required init */
2289 if (OP(noper) == NOTHING) {
2290 regnode *noper_next= regnext(noper);
2291 if (noper_next != tail && OP(noper_next) == flags) {
2293 uc= (U8*)STRING(noper);
2294 e= uc + STR_LEN(noper);
2298 if (OP(noper) != NOTHING) {
2299 for ( ; uc < e ; uc += len ) {
2304 charid = trie->charmap[ uvc ];
2306 SV** const svpp = hv_fetch( widecharmap,
2313 charid=(U16)SvIV( *svpp );
2316 /* charid is now 0 if we dont know the char read, or
2317 * nonzero if we do */
2324 if ( !trie->states[ state ].trans.list ) {
2325 TRIE_LIST_NEW( state );
2328 check <= TRIE_LIST_USED( state );
2331 if ( TRIE_LIST_ITEM( state, check ).forid
2334 newstate = TRIE_LIST_ITEM( state, check ).newstate;
2339 newstate = next_alloc++;
2340 prev_states[newstate] = state;
2341 TRIE_LIST_PUSH( state, charid, newstate );
2346 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %"IVdf, uvc );
2350 TRIE_HANDLE_WORD(state);
2352 } /* end second pass */
2354 /* next alloc is the NEXT state to be allocated */
2355 trie->statecount = next_alloc;
2356 trie->states = (reg_trie_state *)
2357 PerlMemShared_realloc( trie->states,
2359 * sizeof(reg_trie_state) );
2361 /* and now dump it out before we compress it */
2362 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_list(trie, widecharmap,
2363 revcharmap, next_alloc,
2367 trie->trans = (reg_trie_trans *)
2368 PerlMemShared_calloc( transcount, sizeof(reg_trie_trans) );
2375 for( state=1 ; state < next_alloc ; state ++ ) {
2379 DEBUG_TRIE_COMPILE_MORE_r(
2380 PerlIO_printf( Perl_debug_log, "tp: %d zp: %d ",tp,zp)
2384 if (trie->states[state].trans.list) {
2385 U16 minid=TRIE_LIST_ITEM( state, 1).forid;
2389 for( idx = 2 ; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
2390 const U16 forid = TRIE_LIST_ITEM( state, idx).forid;
2391 if ( forid < minid ) {
2393 } else if ( forid > maxid ) {
2397 if ( transcount < tp + maxid - minid + 1) {
2399 trie->trans = (reg_trie_trans *)
2400 PerlMemShared_realloc( trie->trans,
2402 * sizeof(reg_trie_trans) );
2403 Zero( trie->trans + (transcount / 2),
2407 base = trie->uniquecharcount + tp - minid;
2408 if ( maxid == minid ) {
2410 for ( ; zp < tp ; zp++ ) {
2411 if ( ! trie->trans[ zp ].next ) {
2412 base = trie->uniquecharcount + zp - minid;
2413 trie->trans[ zp ].next = TRIE_LIST_ITEM( state,
2415 trie->trans[ zp ].check = state;
2421 trie->trans[ tp ].next = TRIE_LIST_ITEM( state,
2423 trie->trans[ tp ].check = state;
2428 for ( idx=1; idx <= TRIE_LIST_USED( state ) ; idx++ ) {
2429 const U32 tid = base
2430 - trie->uniquecharcount
2431 + TRIE_LIST_ITEM( state, idx ).forid;
2432 trie->trans[ tid ].next = TRIE_LIST_ITEM( state,
2434 trie->trans[ tid ].check = state;
2436 tp += ( maxid - minid + 1 );
2438 Safefree(trie->states[ state ].trans.list);
2441 DEBUG_TRIE_COMPILE_MORE_r(
2442 PerlIO_printf( Perl_debug_log, " base: %d\n",base);
2445 trie->states[ state ].trans.base=base;
2447 trie->lasttrans = tp + 1;
2451 Second Pass -- Flat Table Representation.
2453 we dont use the 0 slot of either trans[] or states[] so we add 1 to
2454 each. We know that we will need Charcount+1 trans at most to store
2455 the data (one row per char at worst case) So we preallocate both
2456 structures assuming worst case.
2458 We then construct the trie using only the .next slots of the entry
2461 We use the .check field of the first entry of the node temporarily
2462 to make compression both faster and easier by keeping track of how
2463 many non zero fields are in the node.
2465 Since trans are numbered from 1 any 0 pointer in the table is a FAIL
2468 There are two terms at use here: state as a TRIE_NODEIDX() which is
2469 a number representing the first entry of the node, and state as a
2470 TRIE_NODENUM() which is the trans number. state 1 is TRIE_NODEIDX(1)
2471 and TRIE_NODENUM(1), state 2 is TRIE_NODEIDX(2) and TRIE_NODENUM(3)
2472 if there are 2 entrys per node. eg:
2480 The table is internally in the right hand, idx form. However as we
2481 also have to deal with the states array which is indexed by nodenum
2482 we have to use TRIE_NODENUM() to convert.
2485 DEBUG_TRIE_COMPILE_MORE_r( PerlIO_printf( Perl_debug_log,
2486 "%*sCompiling trie using table compiler\n",
2487 (int)depth * 2 + 2, ""));
2489 trie->trans = (reg_trie_trans *)
2490 PerlMemShared_calloc( ( TRIE_CHARCOUNT(trie) + 1 )
2491 * trie->uniquecharcount + 1,
2492 sizeof(reg_trie_trans) );
2493 trie->states = (reg_trie_state *)
2494 PerlMemShared_calloc( TRIE_CHARCOUNT(trie) + 2,
2495 sizeof(reg_trie_state) );
2496 next_alloc = trie->uniquecharcount + 1;
2499 for ( cur = first ; cur < last ; cur = regnext( cur ) ) {
2501 regnode *noper = NEXTOPER( cur );
2502 const U8 *uc = (U8*)STRING( noper );
2503 const U8 *e = uc + STR_LEN( noper );
2505 U32 state = 1; /* required init */
2507 U16 charid = 0; /* sanity init */
2508 U32 accept_state = 0; /* sanity init */
2510 U32 wordlen = 0; /* required init */
2512 if (OP(noper) == NOTHING) {
2513 regnode *noper_next= regnext(noper);
2514 if (noper_next != tail && OP(noper_next) == flags) {
2516 uc= (U8*)STRING(noper);
2517 e= uc + STR_LEN(noper);
2521 if ( OP(noper) != NOTHING ) {
2522 for ( ; uc < e ; uc += len ) {
2527 charid = trie->charmap[ uvc ];
2529 SV* const * const svpp = hv_fetch( widecharmap,
2533 charid = svpp ? (U16)SvIV(*svpp) : 0;
2537 if ( !trie->trans[ state + charid ].next ) {
2538 trie->trans[ state + charid ].next = next_alloc;
2539 trie->trans[ state ].check++;
2540 prev_states[TRIE_NODENUM(next_alloc)]
2541 = TRIE_NODENUM(state);
2542 next_alloc += trie->uniquecharcount;
2544 state = trie->trans[ state + charid ].next;
2546 Perl_croak( aTHX_ "panic! In trie construction, no char mapping for %"IVdf, uvc );
2548 /* charid is now 0 if we dont know the char read, or
2549 * nonzero if we do */
2552 accept_state = TRIE_NODENUM( state );
2553 TRIE_HANDLE_WORD(accept_state);
2555 } /* end second pass */
2557 /* and now dump it out before we compress it */
2558 DEBUG_TRIE_COMPILE_MORE_r(dump_trie_interim_table(trie, widecharmap,
2560 next_alloc, depth+1));
2564 * Inplace compress the table.*
2566 For sparse data sets the table constructed by the trie algorithm will
2567 be mostly 0/FAIL transitions or to put it another way mostly empty.
2568 (Note that leaf nodes will not contain any transitions.)
2570 This algorithm compresses the tables by eliminating most such
2571 transitions, at the cost of a modest bit of extra work during lookup:
2573 - Each states[] entry contains a .base field which indicates the
2574 index in the state[] array wheres its transition data is stored.
2576 - If .base is 0 there are no valid transitions from that node.
2578 - If .base is nonzero then charid is added to it to find an entry in
2581 -If trans[states[state].base+charid].check!=state then the
2582 transition is taken to be a 0/Fail transition. Thus if there are fail
2583 transitions at the front of the node then the .base offset will point
2584 somewhere inside the previous nodes data (or maybe even into a node
2585 even earlier), but the .check field determines if the transition is
2589 The following process inplace converts the table to the compressed
2590 table: We first do not compress the root node 1,and mark all its
2591 .check pointers as 1 and set its .base pointer as 1 as well. This
2592 allows us to do a DFA construction from the compressed table later,
2593 and ensures that any .base pointers we calculate later are greater
2596 - We set 'pos' to indicate the first entry of the second node.
2598 - We then iterate over the columns of the node, finding the first and
2599 last used entry at l and m. We then copy l..m into pos..(pos+m-l),
2600 and set the .check pointers accordingly, and advance pos
2601 appropriately and repreat for the next node. Note that when we copy
2602 the next pointers we have to convert them from the original
2603 NODEIDX form to NODENUM form as the former is not valid post
2606 - If a node has no transitions used we mark its base as 0 and do not
2607 advance the pos pointer.
2609 - If a node only has one transition we use a second pointer into the
2610 structure to fill in allocated fail transitions from other states.
2611 This pointer is independent of the main pointer and scans forward
2612 looking for null transitions that are allocated to a state. When it
2613 finds one it writes the single transition into the "hole". If the
2614 pointer doesnt find one the single transition is appended as normal.
2616 - Once compressed we can Renew/realloc the structures to release the
2619 See "Table-Compression Methods" in sec 3.9 of the Red Dragon,
2620 specifically Fig 3.47 and the associated pseudocode.
2624 const U32 laststate = TRIE_NODENUM( next_alloc );
2627 trie->statecount = laststate;
2629 for ( state = 1 ; state < laststate ; state++ ) {
2631 const U32 stateidx = TRIE_NODEIDX( state );
2632 const U32 o_used = trie->trans[ stateidx ].check;
2633 U32 used = trie->trans[ stateidx ].check;
2634 trie->trans[ stateidx ].check = 0;
2637 used && charid < trie->uniquecharcount;
2640 if ( flag || trie->trans[ stateidx + charid ].next ) {
2641 if ( trie->trans[ stateidx + charid ].next ) {
2643 for ( ; zp < pos ; zp++ ) {
2644 if ( ! trie->trans[ zp ].next ) {
2648 trie->states[ state ].trans.base
2650 + trie->uniquecharcount
2652 trie->trans[ zp ].next
2653 = SAFE_TRIE_NODENUM( trie->trans[ stateidx
2655 trie->trans[ zp ].check = state;
2656 if ( ++zp > pos ) pos = zp;
2663 trie->states[ state ].trans.base
2664 = pos + trie->uniquecharcount - charid ;
2666 trie->trans[ pos ].next
2667 = SAFE_TRIE_NODENUM(
2668 trie->trans[ stateidx + charid ].next );
2669 trie->trans[ pos ].check = state;
2674 trie->lasttrans = pos + 1;
2675 trie->states = (reg_trie_state *)
2676 PerlMemShared_realloc( trie->states, laststate
2677 * sizeof(reg_trie_state) );
2678 DEBUG_TRIE_COMPILE_MORE_r(
2679 PerlIO_printf( Perl_debug_log,
2680 "%*sAlloc: %d Orig: %"IVdf" elements, Final:%"IVdf". Savings of %%%5.2f\n",
2681 (int)depth * 2 + 2,"",
2682 (int)( ( TRIE_CHARCOUNT(trie) + 1 ) * trie->uniquecharcount
2686 ( ( next_alloc - pos ) * 100 ) / (double)next_alloc );
2689 } /* end table compress */
2691 DEBUG_TRIE_COMPILE_MORE_r(
2692 PerlIO_printf(Perl_debug_log,
2693 "%*sStatecount:%"UVxf" Lasttrans:%"UVxf"\n",
2694 (int)depth * 2 + 2, "",
2695 (UV)trie->statecount,
2696 (UV)trie->lasttrans)
2698 /* resize the trans array to remove unused space */
2699 trie->trans = (reg_trie_trans *)
2700 PerlMemShared_realloc( trie->trans, trie->lasttrans
2701 * sizeof(reg_trie_trans) );
2703 { /* Modify the program and insert the new TRIE node */
2704 U8 nodetype =(U8)(flags & 0xFF);
2708 regnode *optimize = NULL;
2709 #ifdef RE_TRACK_PATTERN_OFFSETS
2712 U32 mjd_nodelen = 0;
2713 #endif /* RE_TRACK_PATTERN_OFFSETS */
2714 #endif /* DEBUGGING */
2716 This means we convert either the first branch or the first Exact,
2717 depending on whether the thing following (in 'last') is a branch
2718 or not and whther first is the startbranch (ie is it a sub part of
2719 the alternation or is it the whole thing.)
2720 Assuming its a sub part we convert the EXACT otherwise we convert
2721 the whole branch sequence, including the first.
2723 /* Find the node we are going to overwrite */
2724 if ( first != startbranch || OP( last ) == BRANCH ) {
2725 /* branch sub-chain */
2726 NEXT_OFF( first ) = (U16)(last - first);
2727 #ifdef RE_TRACK_PATTERN_OFFSETS
2729 mjd_offset= Node_Offset((convert));
2730 mjd_nodelen= Node_Length((convert));
2733 /* whole branch chain */
2735 #ifdef RE_TRACK_PATTERN_OFFSETS
2738 const regnode *nop = NEXTOPER( convert );
2739 mjd_offset= Node_Offset((nop));
2740 mjd_nodelen= Node_Length((nop));
2744 PerlIO_printf(Perl_debug_log,
2745 "%*sMJD offset:%"UVuf" MJD length:%"UVuf"\n",
2746 (int)depth * 2 + 2, "",
2747 (UV)mjd_offset, (UV)mjd_nodelen)
2750 /* But first we check to see if there is a common prefix we can
2751 split out as an EXACT and put in front of the TRIE node. */
2752 trie->startstate= 1;
2753 if ( trie->bitmap && !widecharmap && !trie->jump ) {
2755 for ( state = 1 ; state < trie->statecount-1 ; state++ ) {
2759 const U32 base = trie->states[ state ].trans.base;
2761 if ( trie->states[state].wordnum )
2764 for ( ofs = 0 ; ofs < trie->uniquecharcount ; ofs++ ) {
2765 if ( ( base + ofs >= trie->uniquecharcount ) &&
2766 ( base + ofs - trie->uniquecharcount < trie->lasttrans ) &&
2767 trie->trans[ base + ofs - trie->uniquecharcount ].check == state )
2769 if ( ++count > 1 ) {
2770 SV **tmp = av_fetch( revcharmap, ofs, 0);
2771 const U8 *ch = (U8*)SvPV_nolen_const( *tmp );
2772 if ( state == 1 ) break;
2774 Zero(trie->bitmap, ANYOF_BITMAP_SIZE, char);
2776 PerlIO_printf(Perl_debug_log,
2777 "%*sNew Start State=%"UVuf" Class: [",
2778 (int)depth * 2 + 2, "",
2781 SV ** const tmp = av_fetch( revcharmap, idx, 0);
2782 const U8 * const ch = (U8*)SvPV_nolen_const( *tmp );
2784 TRIE_BITMAP_SET(trie,*ch);
2786 TRIE_BITMAP_SET(trie, folder[ *ch ]);
2788 PerlIO_printf(Perl_debug_log, "%s", (char*)ch)
2792 TRIE_BITMAP_SET(trie,*ch);
2794 TRIE_BITMAP_SET(trie,folder[ *ch ]);
2795 DEBUG_OPTIMISE_r(PerlIO_printf( Perl_debug_log,"%s", ch));
2801 SV **tmp = av_fetch( revcharmap, idx, 0);
2803 char *ch = SvPV( *tmp, len );
2805 SV *sv=sv_newmortal();
2806 PerlIO_printf( Perl_debug_log,
2807 "%*sPrefix State: %"UVuf" Idx:%"UVuf" Char='%s'\n",
2808 (int)depth * 2 + 2, "",
2810 pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 6,
2811 PL_colors[0], PL_colors[1],
2812 (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) |
2813 PERL_PV_ESCAPE_FIRSTCHAR
2818 OP( convert ) = nodetype;
2819 str=STRING(convert);
2822 STR_LEN(convert) += len;
2828 DEBUG_OPTIMISE_r(PerlIO_printf( Perl_debug_log,"]\n"));
2833 trie->prefixlen = (state-1);
2835 regnode *n = convert+NODE_SZ_STR(convert);
2836 NEXT_OFF(convert) = NODE_SZ_STR(convert);
2837 trie->startstate = state;
2838 trie->minlen -= (state - 1);
2839 trie->maxlen -= (state - 1);
2841 /* At least the UNICOS C compiler choked on this
2842 * being argument to DEBUG_r(), so let's just have
2845 #ifdef PERL_EXT_RE_BUILD
2851 regnode *fix = convert;
2852 U32 word = trie->wordcount;
2854 Set_Node_Offset_Length(convert, mjd_offset, state - 1);
2855 while( ++fix < n ) {
2856 Set_Node_Offset_Length(fix, 0, 0);
2859 SV ** const tmp = av_fetch( trie_words, word, 0 );
2861 if ( STR_LEN(convert) <= SvCUR(*tmp) )
2862 sv_chop(*tmp, SvPV_nolen(*tmp) + STR_LEN(convert));
2864 sv_chop(*tmp, SvPV_nolen(*tmp) + SvCUR(*tmp));
2872 NEXT_OFF(convert) = (U16)(tail - convert);
2873 DEBUG_r(optimize= n);
2879 if ( trie->maxlen ) {
2880 NEXT_OFF( convert ) = (U16)(tail - convert);
2881 ARG_SET( convert, data_slot );
2882 /* Store the offset to the first unabsorbed branch in
2883 jump[0], which is otherwise unused by the jump logic.
2884 We use this when dumping a trie and during optimisation. */
2886 trie->jump[0] = (U16)(nextbranch - convert);
2888 /* If the start state is not accepting (meaning there is no empty string/NOTHING)
2889 * and there is a bitmap
2890 * and the first "jump target" node we found leaves enough room
2891 * then convert the TRIE node into a TRIEC node, with the bitmap
2892 * embedded inline in the opcode - this is hypothetically faster.
2894 if ( !trie->states[trie->startstate].wordnum
2896 && ( (char *)jumper - (char *)convert) >= (int)sizeof(struct regnode_charclass) )
2898 OP( convert ) = TRIEC;
2899 Copy(trie->bitmap, ((struct regnode_charclass *)convert)->bitmap, ANYOF_BITMAP_SIZE, char);
2900 PerlMemShared_free(trie->bitmap);
2903 OP( convert ) = TRIE;
2905 /* store the type in the flags */
2906 convert->flags = nodetype;
2910 + regarglen[ OP( convert ) ];
2912 /* XXX We really should free up the resource in trie now,
2913 as we won't use them - (which resources?) dmq */
2915 /* needed for dumping*/
2916 DEBUG_r(if (optimize) {
2917 regnode *opt = convert;
2919 while ( ++opt < optimize) {
2920 Set_Node_Offset_Length(opt,0,0);
2923 Try to clean up some of the debris left after the
2926 while( optimize < jumper ) {
2927 mjd_nodelen += Node_Length((optimize));
2928 OP( optimize ) = OPTIMIZED;
2929 Set_Node_Offset_Length(optimize,0,0);
2932 Set_Node_Offset_Length(convert,mjd_offset,mjd_nodelen);
2934 } /* end node insert */
2935 REH_CALL_COMP_NODE_HOOK(pRExC_state->rx, convert);
2937 /* Finish populating the prev field of the wordinfo array. Walk back
2938 * from each accept state until we find another accept state, and if
2939 * so, point the first word's .prev field at the second word. If the
2940 * second already has a .prev field set, stop now. This will be the
2941 * case either if we've already processed that word's accept state,
2942 * or that state had multiple words, and the overspill words were
2943 * already linked up earlier.
2950 for (word=1; word <= trie->wordcount; word++) {
2952 if (trie->wordinfo[word].prev)
2954 state = trie->wordinfo[word].accept;
2956 state = prev_states[state];
2959 prev = trie->states[state].wordnum;
2963 trie->wordinfo[word].prev = prev;
2965 Safefree(prev_states);
2969 /* and now dump out the compressed format */
2970 DEBUG_TRIE_COMPILE_r(dump_trie(trie, widecharmap, revcharmap, depth+1));
2972 RExC_rxi->data->data[ data_slot + 1 ] = (void*)widecharmap;
2974 RExC_rxi->data->data[ data_slot + TRIE_WORDS_OFFSET ] = (void*)trie_words;
2975 RExC_rxi->data->data[ data_slot + 3 ] = (void*)revcharmap;
2977 SvREFCNT_dec_NN(revcharmap);
2981 : trie->startstate>1
2987 S_construct_ahocorasick_from_trie(pTHX_ RExC_state_t *pRExC_state, regnode *source, U32 depth)
2989 /* The Trie is constructed and compressed now so we can build a fail array if
2992 This is basically the Aho-Corasick algorithm. Its from exercise 3.31 and
2994 "Red Dragon" -- Compilers, principles, techniques, and tools. Aho, Sethi,
2998 We find the fail state for each state in the trie, this state is the longest
2999 proper suffix of the current state's 'word' that is also a proper prefix of
3000 another word in our trie. State 1 represents the word '' and is thus the
3001 default fail state. This allows the DFA not to have to restart after its
3002 tried and failed a word at a given point, it simply continues as though it
3003 had been matching the other word in the first place.
3005 'abcdgu'=~/abcdefg|cdgu/
3006 When we get to 'd' we are still matching the first word, we would encounter
3007 'g' which would fail, which would bring us to the state representing 'd' in
3008 the second word where we would try 'g' and succeed, proceeding to match
3011 /* add a fail transition */
3012 const U32 trie_offset = ARG(source);
3013 reg_trie_data *trie=(reg_trie_data *)RExC_rxi->data->data[trie_offset];
3015 const U32 ucharcount = trie->uniquecharcount;
3016 const U32 numstates = trie->statecount;
3017 const U32 ubound = trie->lasttrans + ucharcount;
3021 U32 base = trie->states[ 1 ].trans.base;
3024 const U32 data_slot = add_data( pRExC_state, STR_WITH_LEN("T"));
3026 GET_RE_DEBUG_FLAGS_DECL;
3028 PERL_ARGS_ASSERT_CONSTRUCT_AHOCORASICK_FROM_TRIE;
3029 PERL_UNUSED_CONTEXT;
3031 PERL_UNUSED_ARG(depth);
3034 if ( OP(source) == TRIE ) {
3035 struct regnode_1 *op = (struct regnode_1 *)
3036 PerlMemShared_calloc(1, sizeof(struct regnode_1));
3037 StructCopy(source,op,struct regnode_1);
3038 stclass = (regnode *)op;
3040 struct regnode_charclass *op = (struct regnode_charclass *)
3041 PerlMemShared_calloc(1, sizeof(struct regnode_charclass));
3042 StructCopy(source,op,struct regnode_charclass);
3043 stclass = (regnode *)op;
3045 OP(stclass)+=2; /* covert the TRIE type to its AHO-CORASICK equivalent */
3047 ARG_SET( stclass, data_slot );
3048 aho = (reg_ac_data *) PerlMemShared_calloc( 1, sizeof(reg_ac_data) );
3049 RExC_rxi->data->data[ data_slot ] = (void*)aho;
3050 aho->trie=trie_offset;
3051 aho->states=(reg_trie_state *)PerlMemShared_malloc( numstates * sizeof(reg_trie_state) );
3052 Copy( trie->states, aho->states, numstates, reg_trie_state );
3053 Newxz( q, numstates, U32);
3054 aho->fail = (U32 *) PerlMemShared_calloc( numstates, sizeof(U32) );
3057 /* initialize fail[0..1] to be 1 so that we always have
3058 a valid final fail state */
3059 fail[ 0 ] = fail[ 1 ] = 1;
3061 for ( charid = 0; charid < ucharcount ; charid++ ) {
3062 const U32 newstate = TRIE_TRANS_STATE( 1, base, ucharcount, charid, 0 );
3064 q[ q_write ] = newstate;
3065 /* set to point at the root */
3066 fail[ q[ q_write++ ] ]=1;
3069 while ( q_read < q_write) {
3070 const U32 cur = q[ q_read++ % numstates ];
3071 base = trie->states[ cur ].trans.base;
3073 for ( charid = 0 ; charid < ucharcount ; charid++ ) {
3074 const U32 ch_state = TRIE_TRANS_STATE( cur, base, ucharcount, charid, 1 );
3076 U32 fail_state = cur;
3079 fail_state = fail[ fail_state ];
3080 fail_base = aho->states[ fail_state ].trans.base;
3081 } while ( !TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 ) );
3083 fail_state = TRIE_TRANS_STATE( fail_state, fail_base, ucharcount, charid, 1 );
3084 fail[ ch_state ] = fail_state;
3085 if ( !aho->states[ ch_state ].wordnum && aho->states[ fail_state ].wordnum )
3087 aho->states[ ch_state ].wordnum = aho->states[ fail_state ].wordnum;
3089 q[ q_write++ % numstates] = ch_state;
3093 /* restore fail[0..1] to 0 so that we "fall out" of the AC loop
3094 when we fail in state 1, this allows us to use the
3095 charclass scan to find a valid start char. This is based on the principle
3096 that theres a good chance the string being searched contains lots of stuff
3097 that cant be a start char.
3099 fail[ 0 ] = fail[ 1 ] = 0;
3100 DEBUG_TRIE_COMPILE_r({
3101 PerlIO_printf(Perl_debug_log,
3102 "%*sStclass Failtable (%"UVuf" states): 0",
3103 (int)(depth * 2), "", (UV)numstates
3105 for( q_read=1; q_read<numstates; q_read++ ) {
3106 PerlIO_printf(Perl_debug_log, ", %"UVuf, (UV)fail[q_read]);
3108 PerlIO_printf(Perl_debug_log, "\n");
3111 /*RExC_seen |= REG_TRIEDFA_SEEN;*/
3116 #define DEBUG_PEEP(str,scan,depth) \
3117 DEBUG_OPTIMISE_r({if (scan){ \
3118 SV * const mysv=sv_newmortal(); \
3119 regnode *Next = regnext(scan); \
3120 regprop(RExC_rx, mysv, scan, NULL); \
3121 PerlIO_printf(Perl_debug_log, "%*s" str ">%3d: %s (%d)\n", \
3122 (int)depth*2, "", REG_NODE_NUM(scan), SvPV_nolen_const(mysv),\
3123 Next ? (REG_NODE_NUM(Next)) : 0 ); \
3127 /* The below joins as many adjacent EXACTish nodes as possible into a single
3128 * one. The regop may be changed if the node(s) contain certain sequences that
3129 * require special handling. The joining is only done if:
3130 * 1) there is room in the current conglomerated node to entirely contain the
3132 * 2) they are the exact same node type
3134 * The adjacent nodes actually may be separated by NOTHING-kind nodes, and
3135 * these get optimized out
3137 * If a node is to match under /i (folded), the number of characters it matches
3138 * can be different than its character length if it contains a multi-character
3139 * fold. *min_subtract is set to the total delta number of characters of the
3142 * And *unfolded_multi_char is set to indicate whether or not the node contains
3143 * an unfolded multi-char fold. This happens when whether the fold is valid or
3144 * not won't be known until runtime; namely for EXACTF nodes that contain LATIN
3145 * SMALL LETTER SHARP S, as only if the target string being matched against
3146 * turns out to be UTF-8 is that fold valid; and also for EXACTFL nodes whose
3147 * folding rules depend on the locale in force at runtime. (Multi-char folds
3148 * whose components are all above the Latin1 range are not run-time locale
3149 * dependent, and have already been folded by the time this function is
3152 * This is as good a place as any to discuss the design of handling these
3153 * multi-character fold sequences. It's been wrong in Perl for a very long
3154 * time. There are three code points in Unicode whose multi-character folds
3155 * were long ago discovered to mess things up. The previous designs for
3156 * dealing with these involved assigning a special node for them. This
3157 * approach doesn't always work, as evidenced by this example:
3158 * "\xDFs" =~ /s\xDF/ui # Used to fail before these patches
3159 * Both sides fold to "sss", but if the pattern is parsed to create a node that
3160 * would match just the \xDF, it won't be able to handle the case where a
3161 * successful match would have to cross the node's boundary. The new approach
3162 * that hopefully generally solves the problem generates an EXACTFU_SS node
3163 * that is "sss" in this case.
3165 * It turns out that there are problems with all multi-character folds, and not
3166 * just these three. Now the code is general, for all such cases. The
3167 * approach taken is:
3168 * 1) This routine examines each EXACTFish node that could contain multi-
3169 * character folded sequences. Since a single character can fold into
3170 * such a sequence, the minimum match length for this node is less than
3171 * the number of characters in the node. This routine returns in
3172 * *min_subtract how many characters to subtract from the the actual
3173 * length of the string to get a real minimum match length; it is 0 if
3174 * there are no multi-char foldeds. This delta is used by the caller to
3175 * adjust the min length of the match, and the delta between min and max,
3176 * so that the optimizer doesn't reject these possibilities based on size
3178 * 2) For the sequence involving the Sharp s (\xDF), the node type EXACTFU_SS
3179 * is used for an EXACTFU node that contains at least one "ss" sequence in
3180 * it. For non-UTF-8 patterns and strings, this is the only case where
3181 * there is a possible fold length change. That means that a regular
3182 * EXACTFU node without UTF-8 involvement doesn't have to concern itself
3183 * with length changes, and so can be processed faster. regexec.c takes
3184 * advantage of this. Generally, an EXACTFish node that is in UTF-8 is
3185 * pre-folded by regcomp.c (except EXACTFL, some of whose folds aren't
3186 * known until runtime). This saves effort in regex matching. However,
3187 * the pre-folding isn't done for non-UTF8 patterns because the fold of
3188 * the MICRO SIGN requires UTF-8, and we don't want to slow things down by
3189 * forcing the pattern into UTF8 unless necessary. Also what EXACTF (and,
3190 * again, EXACTFL) nodes fold to isn't known until runtime. The fold
3191 * possibilities for the non-UTF8 patterns are quite simple, except for
3192 * the sharp s. All the ones that don't involve a UTF-8 target string are
3193 * members of a fold-pair, and arrays are set up for all of them so that
3194 * the other member of the pair can be found quickly. Code elsewhere in
3195 * this file makes sure that in EXACTFU nodes, the sharp s gets folded to
3196 * 'ss', even if the pattern isn't UTF-8. This avoids the issues
3197 * described in the next item.
3198 * 3) A problem remains for unfolded multi-char folds. (These occur when the
3199 * validity of the fold won't be known until runtime, and so must remain
3200 * unfolded for now. This happens for the sharp s in EXACTF and EXACTFA
3201 * nodes when the pattern isn't in UTF-8. (Note, BTW, that there cannot
3202 * be an EXACTF node with a UTF-8 pattern.) They also occur for various
3203 * folds in EXACTFL nodes, regardless of the UTF-ness of the pattern.)
3204 * The reason this is a problem is that the optimizer part of regexec.c
3205 * (probably unwittingly, in Perl_regexec_flags()) makes an assumption
3206 * that a character in the pattern corresponds to at most a single
3207 * character in the target string. (And I do mean character, and not byte
3208 * here, unlike other parts of the documentation that have never been
3209 * updated to account for multibyte Unicode.) sharp s in EXACTF and
3210 * EXACTFL nodes can match the two character string 'ss'; in EXACTFA nodes
3211 * it can match "\x{17F}\x{17F}". These, along with other ones in EXACTFL
3212 * nodes, violate the assumption, and they are the only instances where it
3213 * is violated. I'm reluctant to try to change the assumption, as the
3214 * code involved is impenetrable to me (khw), so instead the code here
3215 * punts. This routine examines EXACTFL nodes, and (when the pattern
3216 * isn't UTF-8) EXACTF and EXACTFA for such unfolded folds, and returns a
3217 * boolean indicating whether or not the node contains such a fold. When
3218 * it is true, the caller sets a flag that later causes the optimizer in
3219 * this file to not set values for the floating and fixed string lengths,
3220 * and thus avoids the optimizer code in regexec.c that makes the invalid
3221 * assumption. Thus, there is no optimization based on string lengths for
3222 * EXACTFL nodes that contain these few folds, nor for non-UTF8-pattern
3223 * EXACTF and EXACTFA nodes that contain the sharp s. (The reason the
3224 * assumption is wrong only in these cases is that all other non-UTF-8
3225 * folds are 1-1; and, for UTF-8 patterns, we pre-fold all other folds to
3226 * their expanded versions. (Again, we can't prefold sharp s to 'ss' in
3227 * EXACTF nodes because we don't know at compile time if it actually
3228 * matches 'ss' or not. For EXACTF nodes it will match iff the target
3229 * string is in UTF-8. This is in contrast to EXACTFU nodes, where it
3230 * always matches; and EXACTFA where it never does. In an EXACTFA node in
3231 * a UTF-8 pattern, sharp s is folded to "\x{17F}\x{17F}, avoiding the
3232 * problem; but in a non-UTF8 pattern, folding it to that above-Latin1
3233 * string would require the pattern to be forced into UTF-8, the overhead
3234 * of which we want to avoid. Similarly the unfolded multi-char folds in
3235 * EXACTFL nodes will match iff the locale at the time of match is a UTF-8
3238 * Similarly, the code that generates tries doesn't currently handle
3239 * not-already-folded multi-char folds, and it looks like a pain to change
3240 * that. Therefore, trie generation of EXACTFA nodes with the sharp s
3241 * doesn't work. Instead, such an EXACTFA is turned into a new regnode,
3242 * EXACTFA_NO_TRIE, which the trie code knows not to handle. Most people
3243 * using /iaa matching will be doing so almost entirely with ASCII
3244 * strings, so this should rarely be encountered in practice */
3246 #define JOIN_EXACT(scan,min_subtract,unfolded_multi_char, flags) \
3247 if (PL_regkind[OP(scan)] == EXACT) \
3248 join_exact(pRExC_state,(scan),(min_subtract),unfolded_multi_char, (flags),NULL,depth+1)
3251 S_join_exact(pTHX_ RExC_state_t *pRExC_state, regnode *scan,
3252 UV *min_subtract, bool *unfolded_multi_char,
3253 U32 flags,regnode *val, U32 depth)
3255 /* Merge several consecutive EXACTish nodes into one. */
3256 regnode *n = regnext(scan);
3258 regnode *next = scan + NODE_SZ_STR(scan);
3262 regnode *stop = scan;
3263 GET_RE_DEBUG_FLAGS_DECL;
3265 PERL_UNUSED_ARG(depth);
3268 PERL_ARGS_ASSERT_JOIN_EXACT;
3269 #ifndef EXPERIMENTAL_INPLACESCAN
3270 PERL_UNUSED_ARG(flags);
3271 PERL_UNUSED_ARG(val);
3273 DEBUG_PEEP("join",scan,depth);
3275 /* Look through the subsequent nodes in the chain. Skip NOTHING, merge
3276 * EXACT ones that are mergeable to the current one. */
3278 && (PL_regkind[OP(n)] == NOTHING
3279 || (stringok && OP(n) == OP(scan)))
3281 && NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX)
3284 if (OP(n) == TAIL || n > next)
3286 if (PL_regkind[OP(n)] == NOTHING) {
3287 DEBUG_PEEP("skip:",n,depth);
3288 NEXT_OFF(scan) += NEXT_OFF(n);
3289 next = n + NODE_STEP_REGNODE;
3296 else if (stringok) {
3297 const unsigned int oldl = STR_LEN(scan);
3298 regnode * const nnext = regnext(n);
3300 /* XXX I (khw) kind of doubt that this works on platforms (should
3301 * Perl ever run on one) where U8_MAX is above 255 because of lots
3302 * of other assumptions */
3303 /* Don't join if the sum can't fit into a single node */
3304 if (oldl + STR_LEN(n) > U8_MAX)
3307 DEBUG_PEEP("merg",n,depth);
3310 NEXT_OFF(scan) += NEXT_OFF(n);
3311 STR_LEN(scan) += STR_LEN(n);
3312 next = n + NODE_SZ_STR(n);
3313 /* Now we can overwrite *n : */
3314 Move(STRING(n), STRING(scan) + oldl, STR_LEN(n), char);
3322 #ifdef EXPERIMENTAL_INPLACESCAN
3323 if (flags && !NEXT_OFF(n)) {
3324 DEBUG_PEEP("atch", val, depth);
3325 if (reg_off_by_arg[OP(n)]) {
3326 ARG_SET(n, val - n);
3329 NEXT_OFF(n) = val - n;
3337 *unfolded_multi_char = FALSE;
3339 /* Here, all the adjacent mergeable EXACTish nodes have been merged. We
3340 * can now analyze for sequences of problematic code points. (Prior to
3341 * this final joining, sequences could have been split over boundaries, and
3342 * hence missed). The sequences only happen in folding, hence for any
3343 * non-EXACT EXACTish node */
3344 if (OP(scan) != EXACT) {
3345 U8* s0 = (U8*) STRING(scan);
3347 U8* s_end = s0 + STR_LEN(scan);
3349 int total_count_delta = 0; /* Total delta number of characters that
3350 multi-char folds expand to */
3352 /* One pass is made over the node's string looking for all the
3353 * possibilities. To avoid some tests in the loop, there are two main
3354 * cases, for UTF-8 patterns (which can't have EXACTF nodes) and
3359 if (OP(scan) == EXACTFL) {
3362 /* An EXACTFL node would already have been changed to another
3363 * node type unless there is at least one character in it that
3364 * is problematic; likely a character whose fold definition
3365 * won't be known until runtime, and so has yet to be folded.
3366 * For all but the UTF-8 locale, folds are 1-1 in length, but
3367 * to handle the UTF-8 case, we need to create a temporary
3368 * folded copy using UTF-8 locale rules in order to analyze it.
3369 * This is because our macros that look to see if a sequence is
3370 * a multi-char fold assume everything is folded (otherwise the
3371 * tests in those macros would be too complicated and slow).
3372 * Note that here, the non-problematic folds will have already
3373 * been done, so we can just copy such characters. We actually
3374 * don't completely fold the EXACTFL string. We skip the
3375 * unfolded multi-char folds, as that would just create work
3376 * below to figure out the size they already are */
3378 Newx(folded, UTF8_MAX_FOLD_CHAR_EXPAND * STR_LEN(scan) + 1, U8);
3381 STRLEN s_len = UTF8SKIP(s);
3382 if (! is_PROBLEMATIC_LOCALE_FOLD_utf8(s)) {
3383 Copy(s, d, s_len, U8);
3386 else if (is_FOLDS_TO_MULTI_utf8(s)) {
3387 *unfolded_multi_char = TRUE;
3388 Copy(s, d, s_len, U8);
3391 else if (isASCII(*s)) {
3392 *(d++) = toFOLD(*s);
3396 _to_utf8_fold_flags(s, d, &len, FOLD_FLAGS_FULL);
3402 /* Point the remainder of the routine to look at our temporary
3406 } /* End of creating folded copy of EXACTFL string */
3408 /* Examine the string for a multi-character fold sequence. UTF-8
3409 * patterns have all characters pre-folded by the time this code is
3411 while (s < s_end - 1) /* Can stop 1 before the end, as minimum
3412 length sequence we are looking for is 2 */
3414 int count = 0; /* How many characters in a multi-char fold */
3415 int len = is_MULTI_CHAR_FOLD_utf8_safe(s, s_end);
3416 if (! len) { /* Not a multi-char fold: get next char */
3421 /* Nodes with 'ss' require special handling, except for
3422 * EXACTFA-ish for which there is no multi-char fold to this */
3423 if (len == 2 && *s == 's' && *(s+1) == 's'
3424 && OP(scan) != EXACTFA
3425 && OP(scan) != EXACTFA_NO_TRIE)
3428 if (OP(scan) != EXACTFL) {
3429 OP(scan) = EXACTFU_SS;
3433 else { /* Here is a generic multi-char fold. */
3434 U8* multi_end = s + len;
3436 /* Count how many characters are in it. In the case of
3437 * /aa, no folds which contain ASCII code points are
3438 * allowed, so check for those, and skip if found. */
3439 if (OP(scan) != EXACTFA && OP(scan) != EXACTFA_NO_TRIE) {
3440 count = utf8_length(s, multi_end);
3444 while (s < multi_end) {
3447 goto next_iteration;
3457 /* The delta is how long the sequence is minus 1 (1 is how long
3458 * the character that folds to the sequence is) */
3459 total_count_delta += count - 1;
3463 /* We created a temporary folded copy of the string in EXACTFL
3464 * nodes. Therefore we need to be sure it doesn't go below zero,
3465 * as the real string could be shorter */
3466 if (OP(scan) == EXACTFL) {
3467 int total_chars = utf8_length((U8*) STRING(scan),
3468 (U8*) STRING(scan) + STR_LEN(scan));
3469 if (total_count_delta > total_chars) {
3470 total_count_delta = total_chars;
3474 *min_subtract += total_count_delta;
3477 else if (OP(scan) == EXACTFA) {
3479 /* Non-UTF-8 pattern, EXACTFA node. There can't be a multi-char
3480 * fold to the ASCII range (and there are no existing ones in the
3481 * upper latin1 range). But, as outlined in the comments preceding
3482 * this function, we need to flag any occurrences of the sharp s.
3483 * This character forbids trie formation (because of added
3486 if (*s == LATIN_SMALL_LETTER_SHARP_S) {
3487 OP(scan) = EXACTFA_NO_TRIE;
3488 *unfolded_multi_char = TRUE;
3497 /* Non-UTF-8 pattern, not EXACTFA node. Look for the multi-char
3498 * folds that are all Latin1. As explained in the comments
3499 * preceding this function, we look also for the sharp s in EXACTF
3500 * and EXACTFL nodes; it can be in the final position. Otherwise
3501 * we can stop looking 1 byte earlier because have to find at least
3502 * two characters for a multi-fold */
3503 const U8* upper = (OP(scan) == EXACTF || OP(scan) == EXACTFL)
3508 int len = is_MULTI_CHAR_FOLD_latin1_safe(s, s_end);
3509 if (! len) { /* Not a multi-char fold. */
3510 if (*s == LATIN_SMALL_LETTER_SHARP_S
3511 && (OP(scan) == EXACTF || OP(scan) == EXACTFL))
3513 *unfolded_multi_char = TRUE;
3520 && isARG2_lower_or_UPPER_ARG1('s', *s)
3521 && isARG2_lower_or_UPPER_ARG1('s', *(s+1)))
3524 /* EXACTF nodes need to know that the minimum length
3525 * changed so that a sharp s in the string can match this
3526 * ss in the pattern, but they remain EXACTF nodes, as they
3527 * won't match this unless the target string is is UTF-8,
3528 * which we don't know until runtime. EXACTFL nodes can't
3529 * transform into EXACTFU nodes */
3530 if (OP(scan) != EXACTF && OP(scan) != EXACTFL) {
3531 OP(scan) = EXACTFU_SS;
3535 *min_subtract += len - 1;
3542 /* Allow dumping but overwriting the collection of skipped
3543 * ops and/or strings with fake optimized ops */
3544 n = scan + NODE_SZ_STR(scan);
3552 DEBUG_OPTIMISE_r(if (merged){DEBUG_PEEP("finl",scan,depth)});
3556 /* REx optimizer. Converts nodes into quicker variants "in place".
3557 Finds fixed substrings. */
3559 /* Stops at toplevel WHILEM as well as at "last". At end *scanp is set
3560 to the position after last scanned or to NULL. */
3562 #define INIT_AND_WITHP \
3563 assert(!and_withp); \
3564 Newx(and_withp,1, regnode_ssc); \
3565 SAVEFREEPV(and_withp)
3567 /* this is a chain of data about sub patterns we are processing that
3568 need to be handled separately/specially in study_chunk. Its so
3569 we can simulate recursion without losing state. */
3571 typedef struct scan_frame {
3572 regnode *last; /* last node to process in this frame */
3573 regnode *next; /* next node to process when last is reached */
3574 struct scan_frame *prev; /*previous frame*/
3575 U32 prev_recursed_depth;
3576 I32 stop; /* what stopparen do we use */
3581 S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
3582 SSize_t *minlenp, SSize_t *deltap,
3587 regnode_ssc *and_withp,
3588 U32 flags, U32 depth)
3589 /* scanp: Start here (read-write). */
3590 /* deltap: Write maxlen-minlen here. */
3591 /* last: Stop before this one. */
3592 /* data: string data about the pattern */
3593 /* stopparen: treat close N as END */
3594 /* recursed: which subroutines have we recursed into */
3595 /* and_withp: Valid if flags & SCF_DO_STCLASS_OR */
3597 /* There must be at least this number of characters to match */
3600 regnode *scan = *scanp, *next;
3602 int is_inf = (flags & SCF_DO_SUBSTR) && (data->flags & SF_IS_INF);
3603 int is_inf_internal = 0; /* The studied chunk is infinite */
3604 I32 is_par = OP(scan) == OPEN ? ARG(scan) : 0;
3605 scan_data_t data_fake;
3606 SV *re_trie_maxbuff = NULL;
3607 regnode *first_non_open = scan;
3608 SSize_t stopmin = SSize_t_MAX;
3609 scan_frame *frame = NULL;
3610 GET_RE_DEBUG_FLAGS_DECL;
3612 PERL_ARGS_ASSERT_STUDY_CHUNK;
3615 StructCopy(&zero_scan_data, &data_fake, scan_data_t);
3618 while (first_non_open && OP(first_non_open) == OPEN)
3619 first_non_open=regnext(first_non_open);
3624 while ( scan && OP(scan) != END && scan < last ){
3625 UV min_subtract = 0; /* How mmany chars to subtract from the minimum
3626 node length to get a real minimum (because
3627 the folded version may be shorter) */
3628 bool unfolded_multi_char = FALSE;
3629 /* Peephole optimizer: */
3630 DEBUG_OPTIMISE_MORE_r(
3632 PerlIO_printf(Perl_debug_log,
3633 "%*sstudy_chunk stopparen=%ld depth=%lu recursed_depth=%lu ",
3634 ((int) depth*2), "", (long)stopparen,
3635 (unsigned long)depth, (unsigned long)recursed_depth);
3636 if (recursed_depth) {
3639 for ( j = 0 ; j < recursed_depth ; j++ ) {
3640 PerlIO_printf(Perl_debug_log,"[");
3641 for ( i = 0 ; i < (U32)RExC_npar ; i++ )
3642 PerlIO_printf(Perl_debug_log,"%d",
3643 PAREN_TEST(RExC_study_chunk_recursed +
3644 (j * RExC_study_chunk_recursed_bytes), i)
3647 PerlIO_printf(Perl_debug_log,"]");
3650 PerlIO_printf(Perl_debug_log,"\n");
3653 DEBUG_STUDYDATA("Peep:", data, depth);
3654 DEBUG_PEEP("Peep", scan, depth);
3657 /* The reason we do this here we need to deal with things like /(?:f)(?:o)(?:o)/
3658 * which cant be dealt with by the normal EXACT parsing code, as each (?:..) is handled
3659 * by a different invocation of reg() -- Yves
3661 JOIN_EXACT(scan,&min_subtract, &unfolded_multi_char, 0);
3663 /* Follow the next-chain of the current node and optimize
3664 away all the NOTHINGs from it. */
3665 if (OP(scan) != CURLYX) {
3666 const int max = (reg_off_by_arg[OP(scan)]
3668 /* I32 may be smaller than U16 on CRAYs! */
3669 : (I32_MAX < U16_MAX ? I32_MAX : U16_MAX));
3670 int off = (reg_off_by_arg[OP(scan)] ? ARG(scan) : NEXT_OFF(scan));
3674 /* Skip NOTHING and LONGJMP. */
3675 while ((n = regnext(n))
3676 && ((PL_regkind[OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
3677 || ((OP(n) == LONGJMP) && (noff = ARG(n))))
3678 && off + noff < max)
3680 if (reg_off_by_arg[OP(scan)])
3683 NEXT_OFF(scan) = off;
3688 /* The principal pseudo-switch. Cannot be a switch, since we
3689 look into several different things. */
3690 if (OP(scan) == BRANCH || OP(scan) == BRANCHJ
3691 || OP(scan) == IFTHEN) {
3692 next = regnext(scan);
3694 /* demq: the op(next)==code check is to see if we have
3695 * "branch-branch" AFAICT */
3697 if (OP(next) == code || code == IFTHEN) {
3698 /* NOTE - There is similar code to this block below for
3699 * handling TRIE nodes on a re-study. If you change stuff here
3700 * check there too. */
3701 SSize_t max1 = 0, min1 = SSize_t_MAX, num = 0;
3703 regnode * const startbranch=scan;
3705 if (flags & SCF_DO_SUBSTR) {
3706 /* Cannot merge strings after this. */
3707 scan_commit(pRExC_state, data, minlenp, is_inf);
3710 if (flags & SCF_DO_STCLASS)
3711 ssc_init_zero(pRExC_state, &accum);
3713 while (OP(scan) == code) {
3714 SSize_t deltanext, minnext, fake;
3716 regnode_ssc this_class;
3719 data_fake.flags = 0;
3721 data_fake.whilem_c = data->whilem_c;
3722 data_fake.last_closep = data->last_closep;
3725 data_fake.last_closep = &fake;
3727 data_fake.pos_delta = delta;
3728 next = regnext(scan);
3729 scan = NEXTOPER(scan);
3731 scan = NEXTOPER(scan);
3732 if (flags & SCF_DO_STCLASS) {
3733 ssc_init(pRExC_state, &this_class);
3734 data_fake.start_class = &this_class;
3735 f = SCF_DO_STCLASS_AND;
3737 if (flags & SCF_WHILEM_VISITED_POS)
3738 f |= SCF_WHILEM_VISITED_POS;
3740 /* we suppose the run is continuous, last=next...*/
3741 minnext = study_chunk(pRExC_state, &scan, minlenp,
3742 &deltanext, next, &data_fake, stopparen,
3743 recursed_depth, NULL, f,depth+1);
3746 if (deltanext == SSize_t_MAX) {
3747 is_inf = is_inf_internal = 1;
3749 } else if (max1 < minnext + deltanext)
3750 max1 = minnext + deltanext;
3752 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
3754 if (data_fake.flags & SCF_SEEN_ACCEPT) {
3755 if ( stopmin > minnext)
3756 stopmin = min + min1;
3757 flags &= ~SCF_DO_SUBSTR;
3759 data->flags |= SCF_SEEN_ACCEPT;
3762 if (data_fake.flags & SF_HAS_EVAL)
3763 data->flags |= SF_HAS_EVAL;
3764 data->whilem_c = data_fake.whilem_c;
3766 if (flags & SCF_DO_STCLASS)
3767 ssc_or(pRExC_state, &accum, (regnode_charclass*)&this_class);
3769 if (code == IFTHEN && num < 2) /* Empty ELSE branch */
3771 if (flags & SCF_DO_SUBSTR) {
3772 data->pos_min += min1;
3773 if (data->pos_delta >= SSize_t_MAX - (max1 - min1))
3774 data->pos_delta = SSize_t_MAX;
3776 data->pos_delta += max1 - min1;
3777 if (max1 != min1 || is_inf)
3778 data->longest = &(data->longest_float);
3781 if (delta == SSize_t_MAX
3782 || SSize_t_MAX - delta - (max1 - min1) < 0)
3783 delta = SSize_t_MAX;
3785 delta += max1 - min1;
3786 if (flags & SCF_DO_STCLASS_OR) {
3787 ssc_or(pRExC_state, data->start_class, (regnode_charclass*) &accum);
3789 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
3790 flags &= ~SCF_DO_STCLASS;
3793 else if (flags & SCF_DO_STCLASS_AND) {
3795 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) &accum);
3796 flags &= ~SCF_DO_STCLASS;
3799 /* Switch to OR mode: cache the old value of
3800 * data->start_class */
3802 StructCopy(data->start_class, and_withp, regnode_ssc);
3803 flags &= ~SCF_DO_STCLASS_AND;
3804 StructCopy(&accum, data->start_class, regnode_ssc);
3805 flags |= SCF_DO_STCLASS_OR;
3809 if (PERL_ENABLE_TRIE_OPTIMISATION &&
3810 OP( startbranch ) == BRANCH )
3814 Assuming this was/is a branch we are dealing with: 'scan'
3815 now points at the item that follows the branch sequence,
3816 whatever it is. We now start at the beginning of the
3817 sequence and look for subsequences of
3823 which would be constructed from a pattern like
3826 If we can find such a subsequence we need to turn the first
3827 element into a trie and then add the subsequent branch exact
3828 strings to the trie.
3832 1. patterns where the whole set of branches can be
3835 2. patterns where only a subset can be converted.
3837 In case 1 we can replace the whole set with a single regop
3838 for the trie. In case 2 we need to keep the start and end
3841 'BRANCH EXACT; BRANCH EXACT; BRANCH X'
3842 becomes BRANCH TRIE; BRANCH X;
3844 There is an additional case, that being where there is a
3845 common prefix, which gets split out into an EXACT like node
3846 preceding the TRIE node.
3848 If x(1..n)==tail then we can do a simple trie, if not we make
3849 a "jump" trie, such that when we match the appropriate word
3850 we "jump" to the appropriate tail node. Essentially we turn
3851 a nested if into a case structure of sorts.
3856 if (!re_trie_maxbuff) {
3857 re_trie_maxbuff = get_sv(RE_TRIE_MAXBUF_NAME, 1);
3858 if (!SvIOK(re_trie_maxbuff))
3859 sv_setiv(re_trie_maxbuff, RE_TRIE_MAXBUF_INIT);
3861 if ( SvIV(re_trie_maxbuff)>=0 ) {
3863 regnode *first = (regnode *)NULL;
3864 regnode *last = (regnode *)NULL;
3865 regnode *tail = scan;
3870 SV * const mysv = sv_newmortal(); /* for dumping */
3872 /* var tail is used because there may be a TAIL
3873 regop in the way. Ie, the exacts will point to the
3874 thing following the TAIL, but the last branch will
3875 point at the TAIL. So we advance tail. If we
3876 have nested (?:) we may have to move through several
3880 while ( OP( tail ) == TAIL ) {
3881 /* this is the TAIL generated by (?:) */
3882 tail = regnext( tail );
3886 DEBUG_TRIE_COMPILE_r({
3887 regprop(RExC_rx, mysv, tail, NULL);
3888 PerlIO_printf( Perl_debug_log, "%*s%s%s\n",
3889 (int)depth * 2 + 2, "",
3890 "Looking for TRIE'able sequences. Tail node is: ",
3891 SvPV_nolen_const( mysv )
3897 Step through the branches
3898 cur represents each branch,
3899 noper is the first thing to be matched as part
3901 noper_next is the regnext() of that node.
3903 We normally handle a case like this
3904 /FOO[xyz]|BAR[pqr]/ via a "jump trie" but we also
3905 support building with NOJUMPTRIE, which restricts
3906 the trie logic to structures like /FOO|BAR/.
3908 If noper is a trieable nodetype then the branch is
3909 a possible optimization target. If we are building
3910 under NOJUMPTRIE then we require that noper_next is
3911 the same as scan (our current position in the regex
3914 Once we have two or more consecutive such branches
3915 we can create a trie of the EXACT's contents and
3916 stitch it in place into the program.
3918 If the sequence represents all of the branches in
3919 the alternation we replace the entire thing with a
3922 Otherwise when it is a subsequence we need to
3923 stitch it in place and replace only the relevant
3924 branches. This means the first branch has to remain
3925 as it is used by the alternation logic, and its
3926 next pointer, and needs to be repointed at the item
3927 on the branch chain following the last branch we
3928 have optimized away.
3930 This could be either a BRANCH, in which case the
3931 subsequence is internal, or it could be the item
3932 following the branch sequence in which case the
3933 subsequence is at the end (which does not
3934 necessarily mean the first node is the start of the
3937 TRIE_TYPE(X) is a define which maps the optype to a
3941 ----------------+-----------
3945 EXACTFU_SS | EXACTFU
3950 #define TRIE_TYPE(X) ( ( NOTHING == (X) ) ? NOTHING : \
3951 ( EXACT == (X) ) ? EXACT : \
3952 ( EXACTFU == (X) || EXACTFU_SS == (X) ) ? EXACTFU : \
3953 ( EXACTFA == (X) ) ? EXACTFA : \
3956 /* dont use tail as the end marker for this traverse */
3957 for ( cur = startbranch ; cur != scan ; cur = regnext( cur ) ) {
3958 regnode * const noper = NEXTOPER( cur );
3959 U8 noper_type = OP( noper );
3960 U8 noper_trietype = TRIE_TYPE( noper_type );
3961 #if defined(DEBUGGING) || defined(NOJUMPTRIE)
3962 regnode * const noper_next = regnext( noper );
3963 U8 noper_next_type = (noper_next && noper_next != tail) ? OP(noper_next) : 0;
3964 U8 noper_next_trietype = (noper_next && noper_next != tail) ? TRIE_TYPE( noper_next_type ) :0;
3967 DEBUG_TRIE_COMPILE_r({
3968 regprop(RExC_rx, mysv, cur, NULL);
3969 PerlIO_printf( Perl_debug_log, "%*s- %s (%d)",
3970 (int)depth * 2 + 2,"", SvPV_nolen_const( mysv ), REG_NODE_NUM(cur) );
3972 regprop(RExC_rx, mysv, noper, NULL);
3973 PerlIO_printf( Perl_debug_log, " -> %s",
3974 SvPV_nolen_const(mysv));
3977 regprop(RExC_rx, mysv, noper_next, NULL);
3978 PerlIO_printf( Perl_debug_log,"\t=> %s\t",
3979 SvPV_nolen_const(mysv));
3981 PerlIO_printf( Perl_debug_log, "(First==%d,Last==%d,Cur==%d,tt==%s,nt==%s,nnt==%s)\n",
3982 REG_NODE_NUM(first), REG_NODE_NUM(last), REG_NODE_NUM(cur),
3983 PL_reg_name[trietype], PL_reg_name[noper_trietype], PL_reg_name[noper_next_trietype]
3987 /* Is noper a trieable nodetype that can be merged
3988 * with the current trie (if there is one)? */
3992 ( noper_trietype == NOTHING)
3993 || ( trietype == NOTHING )
3994 || ( trietype == noper_trietype )
3997 && noper_next == tail
4001 /* Handle mergable triable node Either we are
4002 * the first node in a new trieable sequence,
4003 * in which case we do some bookkeeping,
4004 * otherwise we update the end pointer. */
4007 if ( noper_trietype == NOTHING ) {
4008 #if !defined(DEBUGGING) && !defined(NOJUMPTRIE)
4009 regnode * const noper_next = regnext( noper );
4010 U8 noper_next_type = (noper_next && noper_next!=tail) ? OP(noper_next) : 0;
4011 U8 noper_next_trietype = noper_next_type ? TRIE_TYPE( noper_next_type ) :0;
4014 if ( noper_next_trietype ) {
4015 trietype = noper_next_trietype;
4016 } else if (noper_next_type) {
4017 /* a NOTHING regop is 1 regop wide.
4018 * We need at least two for a trie
4019 * so we can't merge this in */
4023 trietype = noper_trietype;
4026 if ( trietype == NOTHING )
4027 trietype = noper_trietype;
4032 } /* end handle mergable triable node */
4034 /* handle unmergable node -
4035 * noper may either be a triable node which can
4036 * not be tried together with the current trie,
4037 * or a non triable node */
4039 /* If last is set and trietype is not
4040 * NOTHING then we have found at least two
4041 * triable branch sequences in a row of a
4042 * similar trietype so we can turn them
4043 * into a trie. If/when we allow NOTHING to
4044 * start a trie sequence this condition
4045 * will be required, and it isn't expensive
4046 * so we leave it in for now. */
4047 if ( trietype && trietype != NOTHING )
4048 make_trie( pRExC_state,
4049 startbranch, first, cur, tail,
4050 count, trietype, depth+1 );
4051 last = NULL; /* note: we clear/update
4052 first, trietype etc below,
4053 so we dont do it here */
4057 && noper_next == tail
4060 /* noper is triable, so we can start a new
4064 trietype = noper_trietype;
4066 /* if we already saw a first but the
4067 * current node is not triable then we have
4068 * to reset the first information. */
4073 } /* end handle unmergable node */
4074 } /* loop over branches */
4075 DEBUG_TRIE_COMPILE_r({
4076 regprop(RExC_rx, mysv, cur, NULL);
4077 PerlIO_printf( Perl_debug_log,
4078 "%*s- %s (%d) <SCAN FINISHED>\n",
4080 "", SvPV_nolen_const( mysv ),REG_NODE_NUM(cur));
4083 if ( last && trietype ) {
4084 if ( trietype != NOTHING ) {
4085 /* the last branch of the sequence was part of
4086 * a trie, so we have to construct it here
4087 * outside of the loop */
4088 made= make_trie( pRExC_state, startbranch,
4089 first, scan, tail, count,
4090 trietype, depth+1 );
4091 #ifdef TRIE_STUDY_OPT
4092 if ( ((made == MADE_EXACT_TRIE &&
4093 startbranch == first)
4094 || ( first_non_open == first )) &&
4096 flags |= SCF_TRIE_RESTUDY;
4097 if ( startbranch == first
4100 RExC_seen &=~REG_TOP_LEVEL_BRANCHES_SEEN;
4105 /* at this point we know whatever we have is a
4106 * NOTHING sequence/branch AND if 'startbranch'
4107 * is 'first' then we can turn the whole thing
4110 if ( startbranch == first ) {
4112 /* the entire thing is a NOTHING sequence,
4113 * something like this: (?:|) So we can
4114 * turn it into a plain NOTHING op. */
4115 DEBUG_TRIE_COMPILE_r({
4116 regprop(RExC_rx, mysv, cur, NULL);
4117 PerlIO_printf( Perl_debug_log,
4118 "%*s- %s (%d) <NOTHING BRANCH SEQUENCE>\n", (int)depth * 2 + 2,
4119 "", SvPV_nolen_const( mysv ),REG_NODE_NUM(cur));
4122 OP(startbranch)= NOTHING;
4123 NEXT_OFF(startbranch)= tail - startbranch;
4124 for ( opt= startbranch + 1; opt < tail ; opt++ )
4128 } /* end if ( last) */
4129 } /* TRIE_MAXBUF is non zero */
4134 else if ( code == BRANCHJ ) { /* single branch is optimized. */
4135 scan = NEXTOPER(NEXTOPER(scan));
4136 } else /* single branch is optimized. */
4137 scan = NEXTOPER(scan);
4139 } else if (OP(scan) == SUSPEND || OP(scan) == GOSUB || OP(scan) == GOSTART) {
4140 scan_frame *newframe = NULL;
4144 U32 my_recursed_depth= recursed_depth;
4146 if (OP(scan) != SUSPEND) {
4147 /* set the pointer */
4148 if (OP(scan) == GOSUB) {
4150 RExC_recurse[ARG2L(scan)] = scan;
4151 start = RExC_open_parens[paren-1];
4152 end = RExC_close_parens[paren-1];
4155 start = RExC_rxi->program + 1;
4160 !PAREN_TEST(RExC_study_chunk_recursed + ((recursed_depth-1) * RExC_study_chunk_recursed_bytes), paren)
4162 if (!recursed_depth) {
4163 Zero(RExC_study_chunk_recursed, RExC_study_chunk_recursed_bytes, U8);
4165 Copy(RExC_study_chunk_recursed + ((recursed_depth-1) * RExC_study_chunk_recursed_bytes),
4166 RExC_study_chunk_recursed + (recursed_depth * RExC_study_chunk_recursed_bytes),
4167 RExC_study_chunk_recursed_bytes, U8);
4169 /* we havent recursed into this paren yet, so recurse into it */
4170 DEBUG_STUDYDATA("set:", data,depth);
4171 PAREN_SET(RExC_study_chunk_recursed + (recursed_depth * RExC_study_chunk_recursed_bytes), paren);
4172 my_recursed_depth= recursed_depth + 1;
4173 Newx(newframe,1,scan_frame);
4175 DEBUG_STUDYDATA("inf:", data,depth);
4176 /* some form of infinite recursion, assume infinite length
4178 if (flags & SCF_DO_SUBSTR) {
4179 scan_commit(pRExC_state, data, minlenp, is_inf);
4180 data->longest = &(data->longest_float);
4182 is_inf = is_inf_internal = 1;
4183 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
4184 ssc_anything(data->start_class);
4185 flags &= ~SCF_DO_STCLASS;
4188 Newx(newframe,1,scan_frame);
4191 end = regnext(scan);
4196 SAVEFREEPV(newframe);
4197 newframe->next = regnext(scan);
4198 newframe->last = last;
4199 newframe->stop = stopparen;
4200 newframe->prev = frame;
4201 newframe->prev_recursed_depth = recursed_depth;
4203 DEBUG_STUDYDATA("frame-new:",data,depth);
4204 DEBUG_PEEP("fnew", scan, depth);
4211 recursed_depth= my_recursed_depth;
4216 else if (OP(scan) == EXACT) {
4217 SSize_t l = STR_LEN(scan);
4220 const U8 * const s = (U8*)STRING(scan);
4221 uc = utf8_to_uvchr_buf(s, s + l, NULL);
4222 l = utf8_length(s, s + l);
4224 uc = *((U8*)STRING(scan));
4227 if (flags & SCF_DO_SUBSTR) { /* Update longest substr. */
4228 /* The code below prefers earlier match for fixed
4229 offset, later match for variable offset. */
4230 if (data->last_end == -1) { /* Update the start info. */
4231 data->last_start_min = data->pos_min;
4232 data->last_start_max = is_inf
4233 ? SSize_t_MAX : data->pos_min + data->pos_delta;
4235 sv_catpvn(data->last_found, STRING(scan), STR_LEN(scan));
4237 SvUTF8_on(data->last_found);
4239 SV * const sv = data->last_found;
4240 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
4241 mg_find(sv, PERL_MAGIC_utf8) : NULL;
4242 if (mg && mg->mg_len >= 0)
4243 mg->mg_len += utf8_length((U8*)STRING(scan),
4244 (U8*)STRING(scan)+STR_LEN(scan));
4246 data->last_end = data->pos_min + l;
4247 data->pos_min += l; /* As in the first entry. */
4248 data->flags &= ~SF_BEFORE_EOL;
4251 /* ANDing the code point leaves at most it, and not in locale, and
4252 * can't match null string */
4253 if (flags & SCF_DO_STCLASS_AND) {
4254 ssc_cp_and(data->start_class, uc);
4255 ANYOF_FLAGS(data->start_class) &= ~ANYOF_EMPTY_STRING;
4256 ssc_clear_locale(data->start_class);
4258 else if (flags & SCF_DO_STCLASS_OR) {
4259 ssc_add_cp(data->start_class, uc);
4260 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
4262 /* See commit msg 749e076fceedeb708a624933726e7989f2302f6a */
4263 ANYOF_FLAGS(data->start_class) &= ~ANYOF_EMPTY_STRING;
4265 flags &= ~SCF_DO_STCLASS;
4267 else if (PL_regkind[OP(scan)] == EXACT) { /* But OP != EXACT!, so is
4269 SSize_t l = STR_LEN(scan);
4270 UV uc = *((U8*)STRING(scan));
4271 SV* EXACTF_invlist = _new_invlist(4); /* Start out big enough for 2
4272 separate code points */
4273 const U8 * s = (U8*)STRING(scan);
4275 /* Search for fixed substrings supports EXACT only. */
4276 if (flags & SCF_DO_SUBSTR) {
4278 scan_commit(pRExC_state, data, minlenp, is_inf);
4281 uc = utf8_to_uvchr_buf(s, s + l, NULL);
4282 l = utf8_length(s, s + l);
4284 if (unfolded_multi_char) {
4285 RExC_seen |= REG_UNFOLDED_MULTI_SEEN;
4287 min += l - min_subtract;
4289 delta += min_subtract;
4290 if (flags & SCF_DO_SUBSTR) {
4291 data->pos_min += l - min_subtract;
4292 if (data->pos_min < 0) {
4295 data->pos_delta += min_subtract;
4297 data->longest = &(data->longest_float);
4301 if (OP(scan) != EXACTFL && flags & SCF_DO_STCLASS_AND) {
4302 ssc_clear_locale(data->start_class);
4307 /* We punt and assume can match anything if the node begins
4308 * with a multi-character fold. Things are complicated. For
4309 * example, /ffi/i could match any of:
4310 * "\N{LATIN SMALL LIGATURE FFI}"
4311 * "\N{LATIN SMALL LIGATURE FF}I"
4312 * "F\N{LATIN SMALL LIGATURE FI}"
4313 * plus several other things; and making sure we have all the
4314 * possibilities is hard. */
4315 if (is_MULTI_CHAR_FOLD_latin1_safe(s, s + STR_LEN(scan))) {
4317 _add_range_to_invlist(EXACTF_invlist, 0, UV_MAX);
4321 /* Any Latin1 range character can potentially match any
4322 * other depending on the locale */
4323 if (OP(scan) == EXACTFL) {
4324 _invlist_union(EXACTF_invlist, PL_Latin1,
4328 /* But otherwise, it matches at least itself. We can
4329 * quickly tell if it has a distinct fold, and if so,
4330 * it matches that as well */
4331 EXACTF_invlist = add_cp_to_invlist(EXACTF_invlist, uc);
4332 if (IS_IN_SOME_FOLD_L1(uc)) {
4333 EXACTF_invlist = add_cp_to_invlist(EXACTF_invlist,
4334 PL_fold_latin1[uc]);
4338 /* Some characters match above-Latin1 ones under /i. This
4339 * is true of EXACTFL ones when the locale is UTF-8 */
4340 if (HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(uc)
4341 && (! isASCII(uc) || (OP(scan) != EXACTFA
4342 && OP(scan) != EXACTFA_NO_TRIE)))
4344 add_above_Latin1_folds(pRExC_state,
4350 else { /* Pattern is UTF-8 */
4351 U8 folded[UTF8_MAX_FOLD_CHAR_EXPAND * UTF8_MAXBYTES_CASE + 1] = { '\0' };
4352 STRLEN foldlen = UTF8SKIP(s);
4353 const U8* e = s + STR_LEN(scan);
4356 /* The only code points that aren't folded in a UTF EXACTFish
4357 * node are are the problematic ones in EXACTFL nodes */
4358 if (OP(scan) == EXACTFL
4359 && is_PROBLEMATIC_LOCALE_FOLDEDS_START_cp(uc))
4361 /* We need to check for the possibility that this EXACTFL
4362 * node begins with a multi-char fold. Therefore we fold
4363 * the first few characters of it so that we can make that
4368 for (i = 0; i < UTF8_MAX_FOLD_CHAR_EXPAND && s < e; i++) {
4370 *(d++) = (U8) toFOLD(*s);
4375 to_utf8_fold(s, d, &len);
4381 /* And set up so the code below that looks in this folded
4382 * buffer instead of the node's string */
4384 foldlen = UTF8SKIP(folded);
4388 /* When we reach here 's' points to the fold of the first
4389 * character(s) of the node; and 'e' points to far enough along
4390 * the folded string to be just past any possible multi-char
4391 * fold. 'foldlen' is the length in bytes of the first
4394 * Unlike the non-UTF-8 case, the macro for determining if a
4395 * string is a multi-char fold requires all the characters to
4396 * already be folded. This is because of all the complications
4397 * if not. Note that they are folded anyway, except in EXACTFL
4398 * nodes. Like the non-UTF case above, we punt if the node
4399 * begins with a multi-char fold */
4401 if (is_MULTI_CHAR_FOLD_utf8_safe(s, e)) {
4403 _add_range_to_invlist(EXACTF_invlist, 0, UV_MAX);
4405 else { /* Single char fold */
4407 /* It matches all the things that fold to it, which are
4408 * found in PL_utf8_foldclosures (including itself) */
4409 EXACTF_invlist = add_cp_to_invlist(EXACTF_invlist, uc);
4410 if (! PL_utf8_foldclosures) {
4411 _load_PL_utf8_foldclosures();
4413 if ((listp = hv_fetch(PL_utf8_foldclosures,
4414 (char *) s, foldlen, FALSE)))
4416 AV* list = (AV*) *listp;
4418 for (k = 0; k <= av_tindex(list); k++) {
4419 SV** c_p = av_fetch(list, k, FALSE);
4425 /* /aa doesn't allow folds between ASCII and non- */
4426 if ((OP(scan) == EXACTFA || OP(scan) == EXACTFA_NO_TRIE)
4427 && isASCII(c) != isASCII(uc))
4432 EXACTF_invlist = add_cp_to_invlist(EXACTF_invlist, c);
4437 if (flags & SCF_DO_STCLASS_AND) {
4438 ANYOF_FLAGS(data->start_class) &= ~ANYOF_EMPTY_STRING;
4439 ANYOF_POSIXL_ZERO(data->start_class);
4440 ssc_intersection(data->start_class, EXACTF_invlist, FALSE);
4442 else if (flags & SCF_DO_STCLASS_OR) {
4443 ssc_union(data->start_class, EXACTF_invlist, FALSE);
4444 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
4446 /* See commit msg 749e076fceedeb708a624933726e7989f2302f6a */
4447 ANYOF_FLAGS(data->start_class) &= ~ANYOF_EMPTY_STRING;
4449 flags &= ~SCF_DO_STCLASS;
4450 SvREFCNT_dec(EXACTF_invlist);
4452 else if (REGNODE_VARIES(OP(scan))) {
4453 SSize_t mincount, maxcount, minnext, deltanext, pos_before = 0;
4454 I32 fl = 0, f = flags;
4455 regnode * const oscan = scan;
4456 regnode_ssc this_class;
4457 regnode_ssc *oclass = NULL;
4458 I32 next_is_eval = 0;
4460 switch (PL_regkind[OP(scan)]) {
4461 case WHILEM: /* End of (?:...)* . */
4462 scan = NEXTOPER(scan);
4465 if (flags & (SCF_DO_SUBSTR | SCF_DO_STCLASS)) {
4466 next = NEXTOPER(scan);
4467 if (OP(next) == EXACT || (flags & SCF_DO_STCLASS)) {
4469 maxcount = REG_INFTY;
4470 next = regnext(scan);
4471 scan = NEXTOPER(scan);
4475 if (flags & SCF_DO_SUBSTR)
4480 if (flags & SCF_DO_STCLASS) {
4482 maxcount = REG_INFTY;
4483 next = regnext(scan);
4484 scan = NEXTOPER(scan);
4487 if (flags & SCF_DO_SUBSTR) {
4488 scan_commit(pRExC_state, data, minlenp, is_inf);
4489 /* Cannot extend fixed substrings */
4490 data->longest = &(data->longest_float);
4492 is_inf = is_inf_internal = 1;
4493 scan = regnext(scan);
4494 goto optimize_curly_tail;
4496 if (stopparen>0 && (OP(scan)==CURLYN || OP(scan)==CURLYM)
4497 && (scan->flags == stopparen))
4502 mincount = ARG1(scan);
4503 maxcount = ARG2(scan);
4505 next = regnext(scan);
4506 if (OP(scan) == CURLYX) {
4507 I32 lp = (data ? *(data->last_closep) : 0);
4508 scan->flags = ((lp <= (I32)U8_MAX) ? (U8)lp : U8_MAX);
4510 scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
4511 next_is_eval = (OP(scan) == EVAL);
4513 if (flags & SCF_DO_SUBSTR) {
4515 scan_commit(pRExC_state, data, minlenp, is_inf);
4516 /* Cannot extend fixed substrings */
4517 pos_before = data->pos_min;
4521 data->flags &= ~(SF_HAS_PAR|SF_IN_PAR|SF_HAS_EVAL);
4523 data->flags |= SF_IS_INF;
4525 if (flags & SCF_DO_STCLASS) {
4526 ssc_init(pRExC_state, &this_class);
4527 oclass = data->start_class;
4528 data->start_class = &this_class;
4529 f |= SCF_DO_STCLASS_AND;
4530 f &= ~SCF_DO_STCLASS_OR;
4532 /* Exclude from super-linear cache processing any {n,m}
4533 regops for which the combination of input pos and regex
4534 pos is not enough information to determine if a match
4537 For example, in the regex /foo(bar\s*){4,8}baz/ with the
4538 regex pos at the \s*, the prospects for a match depend not
4539 only on the input position but also on how many (bar\s*)
4540 repeats into the {4,8} we are. */
4541 if ((mincount > 1) || (maxcount > 1 && maxcount != REG_INFTY))
4542 f &= ~SCF_WHILEM_VISITED_POS;
4544 /* This will finish on WHILEM, setting scan, or on NULL: */
4545 minnext = study_chunk(pRExC_state, &scan, minlenp, &deltanext,
4546 last, data, stopparen, recursed_depth, NULL,
4548 ? (f & ~SCF_DO_SUBSTR)
4552 if (flags & SCF_DO_STCLASS)
4553 data->start_class = oclass;
4554 if (mincount == 0 || minnext == 0) {
4555 if (flags & SCF_DO_STCLASS_OR) {
4556 ssc_or(pRExC_state, data->start_class, (regnode_charclass *) &this_class);
4558 else if (flags & SCF_DO_STCLASS_AND) {
4559 /* Switch to OR mode: cache the old value of
4560 * data->start_class */
4562 StructCopy(data->start_class, and_withp, regnode_ssc);
4563 flags &= ~SCF_DO_STCLASS_AND;
4564 StructCopy(&this_class, data->start_class, regnode_ssc);
4565 flags |= SCF_DO_STCLASS_OR;
4566 ANYOF_FLAGS(data->start_class) |= ANYOF_EMPTY_STRING;
4568 } else { /* Non-zero len */
4569 if (flags & SCF_DO_STCLASS_OR) {
4570 ssc_or(pRExC_state, data->start_class, (regnode_charclass *) &this_class);
4571 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
4573 else if (flags & SCF_DO_STCLASS_AND)
4574 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) &this_class);
4575 flags &= ~SCF_DO_STCLASS;
4577 if (!scan) /* It was not CURLYX, but CURLY. */
4579 if (!(flags & SCF_TRIE_DOING_RESTUDY)
4580 /* ? quantifier ok, except for (?{ ... }) */
4581 && (next_is_eval || !(mincount == 0 && maxcount == 1))
4582 && (minnext == 0) && (deltanext == 0)
4583 && data && !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
4584 && maxcount <= REG_INFTY/3) /* Complement check for big
4587 /* Fatal warnings may leak the regexp without this: */
4588 SAVEFREESV(RExC_rx_sv);
4589 ckWARNreg(RExC_parse,
4590 "Quantifier unexpected on zero-length expression");
4591 (void)ReREFCNT_inc(RExC_rx_sv);
4594 min += minnext * mincount;
4595 is_inf_internal |= deltanext == SSize_t_MAX
4596 || (maxcount == REG_INFTY && minnext + deltanext > 0);
4597 is_inf |= is_inf_internal;
4599 delta = SSize_t_MAX;
4601 delta += (minnext + deltanext) * maxcount
4602 - minnext * mincount;
4604 /* Try powerful optimization CURLYX => CURLYN. */
4605 if ( OP(oscan) == CURLYX && data
4606 && data->flags & SF_IN_PAR
4607 && !(data->flags & SF_HAS_EVAL)
4608 && !deltanext && minnext == 1 ) {
4609 /* Try to optimize to CURLYN. */
4610 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS;
4611 regnode * const nxt1 = nxt;
4618 if (!REGNODE_SIMPLE(OP(nxt))
4619 && !(PL_regkind[OP(nxt)] == EXACT
4620 && STR_LEN(nxt) == 1))
4626 if (OP(nxt) != CLOSE)
4628 if (RExC_open_parens) {
4629 RExC_open_parens[ARG(nxt1)-1]=oscan; /*open->CURLYM*/
4630 RExC_close_parens[ARG(nxt1)-1]=nxt+2; /*close->while*/
4632 /* Now we know that nxt2 is the only contents: */
4633 oscan->flags = (U8)ARG(nxt);
4635 OP(nxt1) = NOTHING; /* was OPEN. */
4638 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
4639 NEXT_OFF(nxt1+ 1) = 0; /* just for consistency. */
4640 NEXT_OFF(nxt2) = 0; /* just for consistency with CURLY. */
4641 OP(nxt) = OPTIMIZED; /* was CLOSE. */
4642 OP(nxt + 1) = OPTIMIZED; /* was count. */
4643 NEXT_OFF(nxt+ 1) = 0; /* just for consistency. */
4648 /* Try optimization CURLYX => CURLYM. */
4649 if ( OP(oscan) == CURLYX && data
4650 && !(data->flags & SF_HAS_PAR)
4651 && !(data->flags & SF_HAS_EVAL)
4652 && !deltanext /* atom is fixed width */
4653 && minnext != 0 /* CURLYM can't handle zero width */
4655 /* Nor characters whose fold at run-time may be
4656 * multi-character */
4657 && ! (RExC_seen & REG_UNFOLDED_MULTI_SEEN)
4659 /* XXXX How to optimize if data == 0? */
4660 /* Optimize to a simpler form. */
4661 regnode *nxt = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN */
4665 while ( (nxt2 = regnext(nxt)) /* skip over embedded stuff*/
4666 && (OP(nxt2) != WHILEM))
4668 OP(nxt2) = SUCCEED; /* Whas WHILEM */
4669 /* Need to optimize away parenths. */
4670 if ((data->flags & SF_IN_PAR) && OP(nxt) == CLOSE) {
4671 /* Set the parenth number. */
4672 regnode *nxt1 = NEXTOPER(oscan) + EXTRA_STEP_2ARGS; /* OPEN*/
4674 oscan->flags = (U8)ARG(nxt);
4675 if (RExC_open_parens) {
4676 RExC_open_parens[ARG(nxt1)-1]=oscan; /*open->CURLYM*/
4677 RExC_close_parens[ARG(nxt1)-1]=nxt2+1; /*close->NOTHING*/
4679 OP(nxt1) = OPTIMIZED; /* was OPEN. */
4680 OP(nxt) = OPTIMIZED; /* was CLOSE. */
4683 OP(nxt1 + 1) = OPTIMIZED; /* was count. */
4684 OP(nxt + 1) = OPTIMIZED; /* was count. */
4685 NEXT_OFF(nxt1 + 1) = 0; /* just for consistency. */
4686 NEXT_OFF(nxt + 1) = 0; /* just for consistency. */
4689 while ( nxt1 && (OP(nxt1) != WHILEM)) {
4690 regnode *nnxt = regnext(nxt1);
4692 if (reg_off_by_arg[OP(nxt1)])
4693 ARG_SET(nxt1, nxt2 - nxt1);
4694 else if (nxt2 - nxt1 < U16_MAX)
4695 NEXT_OFF(nxt1) = nxt2 - nxt1;
4697 OP(nxt) = NOTHING; /* Cannot beautify */
4702 /* Optimize again: */
4703 study_chunk(pRExC_state, &nxt1, minlenp, &deltanext, nxt,
4704 NULL, stopparen, recursed_depth, NULL, 0,depth+1);
4709 else if ((OP(oscan) == CURLYX)
4710 && (flags & SCF_WHILEM_VISITED_POS)
4711 /* See the comment on a similar expression above.
4712 However, this time it's not a subexpression
4713 we care about, but the expression itself. */
4714 && (maxcount == REG_INFTY)
4715 && data && ++data->whilem_c < 16) {
4716 /* This stays as CURLYX, we can put the count/of pair. */
4717 /* Find WHILEM (as in regexec.c) */
4718 regnode *nxt = oscan + NEXT_OFF(oscan);
4720 if (OP(PREVOPER(nxt)) == NOTHING) /* LONGJMP */
4722 PREVOPER(nxt)->flags = (U8)(data->whilem_c
4723 | (RExC_whilem_seen << 4)); /* On WHILEM */
4725 if (data && fl & (SF_HAS_PAR|SF_IN_PAR))
4727 if (flags & SCF_DO_SUBSTR) {
4728 SV *last_str = NULL;
4729 STRLEN last_chrs = 0;
4730 int counted = mincount != 0;
4732 if (data->last_end > 0 && mincount != 0) { /* Ends with a
4734 SSize_t b = pos_before >= data->last_start_min
4735 ? pos_before : data->last_start_min;
4737 const char * const s = SvPV_const(data->last_found, l);
4738 SSize_t old = b - data->last_start_min;
4741 old = utf8_hop((U8*)s, old) - (U8*)s;
4743 /* Get the added string: */
4744 last_str = newSVpvn_utf8(s + old, l, UTF);
4745 last_chrs = UTF ? utf8_length((U8*)(s + old),
4746 (U8*)(s + old + l)) : l;
4747 if (deltanext == 0 && pos_before == b) {
4748 /* What was added is a constant string */
4751 SvGROW(last_str, (mincount * l) + 1);
4752 repeatcpy(SvPVX(last_str) + l,
4753 SvPVX_const(last_str), l,
4755 SvCUR_set(last_str, SvCUR(last_str) * mincount);
4756 /* Add additional parts. */
4757 SvCUR_set(data->last_found,
4758 SvCUR(data->last_found) - l);
4759 sv_catsv(data->last_found, last_str);
4761 SV * sv = data->last_found;
4763 SvUTF8(sv) && SvMAGICAL(sv) ?
4764 mg_find(sv, PERL_MAGIC_utf8) : NULL;
4765 if (mg && mg->mg_len >= 0)
4766 mg->mg_len += last_chrs * (mincount-1);
4768 last_chrs *= mincount;
4769 data->last_end += l * (mincount - 1);
4772 /* start offset must point into the last copy */
4773 data->last_start_min += minnext * (mincount - 1);
4774 data->last_start_max += is_inf ? SSize_t_MAX
4775 : (maxcount - 1) * (minnext + data->pos_delta);
4778 /* It is counted once already... */
4779 data->pos_min += minnext * (mincount - counted);
4781 PerlIO_printf(Perl_debug_log, "counted=%"UVuf" deltanext=%"UVuf
4782 " SSize_t_MAX=%"UVuf" minnext=%"UVuf
4783 " maxcount=%"UVuf" mincount=%"UVuf"\n",
4784 (UV)counted, (UV)deltanext, (UV)SSize_t_MAX, (UV)minnext, (UV)maxcount,
4786 if (deltanext != SSize_t_MAX)
4787 PerlIO_printf(Perl_debug_log, "LHS=%"UVuf" RHS=%"UVuf"\n",
4788 (UV)(-counted * deltanext + (minnext + deltanext) * maxcount
4789 - minnext * mincount), (UV)(SSize_t_MAX - data->pos_delta));
4791 if (deltanext == SSize_t_MAX
4792 || -counted * deltanext + (minnext + deltanext) * maxcount - minnext * mincount >= SSize_t_MAX - data->pos_delta)
4793 data->pos_delta = SSize_t_MAX;
4795 data->pos_delta += - counted * deltanext +
4796 (minnext + deltanext) * maxcount - minnext * mincount;
4797 if (mincount != maxcount) {
4798 /* Cannot extend fixed substrings found inside
4800 scan_commit(pRExC_state, data, minlenp, is_inf);
4801 if (mincount && last_str) {
4802 SV * const sv = data->last_found;
4803 MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
4804 mg_find(sv, PERL_MAGIC_utf8) : NULL;
4808 sv_setsv(sv, last_str);
4809 data->last_end = data->pos_min;
4810 data->last_start_min = data->pos_min - last_chrs;
4811 data->last_start_max = is_inf
4813 : data->pos_min + data->pos_delta - last_chrs;
4815 data->longest = &(data->longest_float);
4817 SvREFCNT_dec(last_str);
4819 if (data && (fl & SF_HAS_EVAL))
4820 data->flags |= SF_HAS_EVAL;
4821 optimize_curly_tail:
4822 if (OP(oscan) != CURLYX) {
4823 while (PL_regkind[OP(next = regnext(oscan))] == NOTHING
4825 NEXT_OFF(oscan) += NEXT_OFF(next);
4831 Perl_croak(aTHX_ "panic: unexpected varying REx opcode %d",
4836 if (flags & SCF_DO_SUBSTR) {
4837 /* Cannot expect anything... */
4838 scan_commit(pRExC_state, data, minlenp, is_inf);
4839 data->longest = &(data->longest_float);
4841 is_inf = is_inf_internal = 1;
4842 if (flags & SCF_DO_STCLASS_OR) {
4843 if (OP(scan) == CLUMP) {
4844 /* Actually is any start char, but very few code points
4845 * aren't start characters */
4846 ssc_match_all_cp(data->start_class);
4849 ssc_anything(data->start_class);
4852 flags &= ~SCF_DO_STCLASS;
4856 else if (OP(scan) == LNBREAK) {
4857 if (flags & SCF_DO_STCLASS) {
4858 if (flags & SCF_DO_STCLASS_AND) {
4859 ssc_intersection(data->start_class,
4860 PL_XPosix_ptrs[_CC_VERTSPACE], FALSE);
4861 ssc_clear_locale(data->start_class);
4862 ANYOF_FLAGS(data->start_class) &= ~ANYOF_EMPTY_STRING;
4864 else if (flags & SCF_DO_STCLASS_OR) {
4865 ssc_union(data->start_class,
4866 PL_XPosix_ptrs[_CC_VERTSPACE],
4868 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
4870 /* See commit msg for
4871 * 749e076fceedeb708a624933726e7989f2302f6a */
4872 ANYOF_FLAGS(data->start_class) &= ~ANYOF_EMPTY_STRING;
4874 flags &= ~SCF_DO_STCLASS;
4877 delta++; /* Because of the 2 char string cr-lf */
4878 if (flags & SCF_DO_SUBSTR) {
4879 /* Cannot expect anything... */
4880 scan_commit(pRExC_state, data, minlenp, is_inf);
4882 data->pos_delta += 1;
4883 data->longest = &(data->longest_float);
4886 else if (REGNODE_SIMPLE(OP(scan))) {
4888 if (flags & SCF_DO_SUBSTR) {
4889 scan_commit(pRExC_state, data, minlenp, is_inf);
4893 if (flags & SCF_DO_STCLASS) {
4895 SV* my_invlist = sv_2mortal(_new_invlist(0));
4898 /* See commit msg 749e076fceedeb708a624933726e7989f2302f6a */
4899 ANYOF_FLAGS(data->start_class) &= ~ANYOF_EMPTY_STRING;
4901 /* Some of the logic below assumes that switching
4902 locale on will only add false positives. */
4907 Perl_croak(aTHX_ "panic: unexpected simple REx opcode %d",
4912 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
4913 ssc_match_all_cp(data->start_class);
4918 SV* REG_ANY_invlist = _new_invlist(2);
4919 REG_ANY_invlist = add_cp_to_invlist(REG_ANY_invlist,
4921 if (flags & SCF_DO_STCLASS_OR) {
4922 ssc_union(data->start_class,
4924 TRUE /* TRUE => invert, hence all but \n
4928 else if (flags & SCF_DO_STCLASS_AND) {
4929 ssc_intersection(data->start_class,
4931 TRUE /* TRUE => invert */
4933 ssc_clear_locale(data->start_class);
4935 SvREFCNT_dec_NN(REG_ANY_invlist);
4940 if (flags & SCF_DO_STCLASS_AND)
4941 ssc_and(pRExC_state, data->start_class,
4942 (regnode_charclass *) scan);
4944 ssc_or(pRExC_state, data->start_class,
4945 (regnode_charclass *) scan);
4953 namedclass = classnum_to_namedclass(FLAGS(scan)) + invert;
4954 if (flags & SCF_DO_STCLASS_AND) {
4955 bool was_there = cBOOL(
4956 ANYOF_POSIXL_TEST(data->start_class,
4958 ANYOF_POSIXL_ZERO(data->start_class);
4959 if (was_there) { /* Do an AND */
4960 ANYOF_POSIXL_SET(data->start_class, namedclass);
4962 /* No individual code points can now match */
4963 data->start_class->invlist
4964 = sv_2mortal(_new_invlist(0));
4967 int complement = namedclass + ((invert) ? -1 : 1);
4969 assert(flags & SCF_DO_STCLASS_OR);
4971 /* If the complement of this class was already there,
4972 * the result is that they match all code points,
4973 * (\d + \D == everything). Remove the classes from
4974 * future consideration. Locale is not relevant in
4976 if (ANYOF_POSIXL_TEST(data->start_class, complement)) {
4977 ssc_match_all_cp(data->start_class);
4978 ANYOF_POSIXL_CLEAR(data->start_class, namedclass);
4979 ANYOF_POSIXL_CLEAR(data->start_class, complement);
4981 else { /* The usual case; just add this class to the
4983 ANYOF_POSIXL_SET(data->start_class, namedclass);
4988 case NPOSIXA: /* For these, we always know the exact set of
4993 if (FLAGS(scan) == _CC_ASCII) {
4994 my_invlist = PL_XPosix_ptrs[_CC_ASCII];
4997 _invlist_intersection(PL_XPosix_ptrs[FLAGS(scan)],
4998 PL_XPosix_ptrs[_CC_ASCII],
5009 my_invlist = invlist_clone(PL_XPosix_ptrs[FLAGS(scan)]);
5011 /* NPOSIXD matches all upper Latin1 code points unless the
5012 * target string being matched is UTF-8, which is
5013 * unknowable until match time. Since we are going to
5014 * invert, we want to get rid of all of them so that the
5015 * inversion will match all */
5016 if (OP(scan) == NPOSIXD) {
5017 _invlist_subtract(my_invlist, PL_UpperLatin1,
5023 if (flags & SCF_DO_STCLASS_AND) {
5024 ssc_intersection(data->start_class, my_invlist, invert);
5025 ssc_clear_locale(data->start_class);
5028 assert(flags & SCF_DO_STCLASS_OR);
5029 ssc_union(data->start_class, my_invlist, invert);
5032 if (flags & SCF_DO_STCLASS_OR)
5033 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
5034 flags &= ~SCF_DO_STCLASS;
5037 else if (PL_regkind[OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
5038 data->flags |= (OP(scan) == MEOL
5041 scan_commit(pRExC_state, data, minlenp, is_inf);
5044 else if ( PL_regkind[OP(scan)] == BRANCHJ
5045 /* Lookbehind, or need to calculate parens/evals/stclass: */
5046 && (scan->flags || data || (flags & SCF_DO_STCLASS))
5047 && (OP(scan) == IFMATCH || OP(scan) == UNLESSM))
5049 if ( OP(scan) == UNLESSM &&
5051 OP(NEXTOPER(NEXTOPER(scan))) == NOTHING &&
5052 OP(regnext(NEXTOPER(NEXTOPER(scan)))) == SUCCEED
5055 regnode *upto= regnext(scan);
5057 SV * const mysv_val=sv_newmortal();
5058 DEBUG_STUDYDATA("OPFAIL",data,depth);
5060 /*DEBUG_PARSE_MSG("opfail");*/
5061 regprop(RExC_rx, mysv_val, upto, NULL);
5062 PerlIO_printf(Perl_debug_log,
5063 "~ replace with OPFAIL pointed at %s (%"IVdf") offset %"IVdf"\n",
5064 SvPV_nolen_const(mysv_val),
5065 (IV)REG_NODE_NUM(upto),
5070 NEXT_OFF(scan) = upto - scan;
5071 for (opt= scan + 1; opt < upto ; opt++)
5072 OP(opt) = OPTIMIZED;
5076 if ( !PERL_ENABLE_POSITIVE_ASSERTION_STUDY
5077 || OP(scan) == UNLESSM )
5079 /* Negative Lookahead/lookbehind
5080 In this case we can't do fixed string optimisation.
5083 SSize_t deltanext, minnext, fake = 0;
5088 data_fake.flags = 0;
5090 data_fake.whilem_c = data->whilem_c;
5091 data_fake.last_closep = data->last_closep;
5094 data_fake.last_closep = &fake;
5095 data_fake.pos_delta = delta;
5096 if ( flags & SCF_DO_STCLASS && !scan->flags
5097 && OP(scan) == IFMATCH ) { /* Lookahead */
5098 ssc_init(pRExC_state, &intrnl);
5099 data_fake.start_class = &intrnl;
5100 f |= SCF_DO_STCLASS_AND;
5102 if (flags & SCF_WHILEM_VISITED_POS)
5103 f |= SCF_WHILEM_VISITED_POS;
5104 next = regnext(scan);
5105 nscan = NEXTOPER(NEXTOPER(scan));
5106 minnext = study_chunk(pRExC_state, &nscan, minlenp, &deltanext,
5107 last, &data_fake, stopparen,
5108 recursed_depth, NULL, f, depth+1);
5111 FAIL("Variable length lookbehind not implemented");
5113 else if (minnext > (I32)U8_MAX) {
5114 FAIL2("Lookbehind longer than %"UVuf" not implemented",
5117 scan->flags = (U8)minnext;
5120 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
5122 if (data_fake.flags & SF_HAS_EVAL)
5123 data->flags |= SF_HAS_EVAL;
5124 data->whilem_c = data_fake.whilem_c;
5126 if (f & SCF_DO_STCLASS_AND) {
5127 if (flags & SCF_DO_STCLASS_OR) {
5128 /* OR before, AND after: ideally we would recurse with
5129 * data_fake to get the AND applied by study of the
5130 * remainder of the pattern, and then derecurse;
5131 * *** HACK *** for now just treat as "no information".
5132 * See [perl #56690].
5134 ssc_init(pRExC_state, data->start_class);
5136 /* AND before and after: combine and continue. These
5137 * assertions are zero-length, so can match an EMPTY
5139 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) &intrnl);
5140 ANYOF_FLAGS(data->start_class) |= ANYOF_EMPTY_STRING;
5144 #if PERL_ENABLE_POSITIVE_ASSERTION_STUDY
5146 /* Positive Lookahead/lookbehind
5147 In this case we can do fixed string optimisation,
5148 but we must be careful about it. Note in the case of
5149 lookbehind the positions will be offset by the minimum
5150 length of the pattern, something we won't know about
5151 until after the recurse.
5153 SSize_t deltanext, fake = 0;
5157 /* We use SAVEFREEPV so that when the full compile
5158 is finished perl will clean up the allocated
5159 minlens when it's all done. This way we don't
5160 have to worry about freeing them when we know
5161 they wont be used, which would be a pain.
5164 Newx( minnextp, 1, SSize_t );
5165 SAVEFREEPV(minnextp);
5168 StructCopy(data, &data_fake, scan_data_t);
5169 if ((flags & SCF_DO_SUBSTR) && data->last_found) {
5172 scan_commit(pRExC_state, &data_fake, minlenp, is_inf);
5173 data_fake.last_found=newSVsv(data->last_found);
5177 data_fake.last_closep = &fake;
5178 data_fake.flags = 0;
5179 data_fake.pos_delta = delta;
5181 data_fake.flags |= SF_IS_INF;
5182 if ( flags & SCF_DO_STCLASS && !scan->flags
5183 && OP(scan) == IFMATCH ) { /* Lookahead */
5184 ssc_init(pRExC_state, &intrnl);
5185 data_fake.start_class = &intrnl;
5186 f |= SCF_DO_STCLASS_AND;
5188 if (flags & SCF_WHILEM_VISITED_POS)
5189 f |= SCF_WHILEM_VISITED_POS;
5190 next = regnext(scan);
5191 nscan = NEXTOPER(NEXTOPER(scan));
5193 *minnextp = study_chunk(pRExC_state, &nscan, minnextp,
5194 &deltanext, last, &data_fake,
5195 stopparen, recursed_depth, NULL,
5199 FAIL("Variable length lookbehind not implemented");
5201 else if (*minnextp > (I32)U8_MAX) {
5202 FAIL2("Lookbehind longer than %"UVuf" not implemented",
5205 scan->flags = (U8)*minnextp;
5210 if (f & SCF_DO_STCLASS_AND) {
5211 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) &intrnl);
5212 ANYOF_FLAGS(data->start_class) |= ANYOF_EMPTY_STRING;
5215 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
5217 if (data_fake.flags & SF_HAS_EVAL)
5218 data->flags |= SF_HAS_EVAL;
5219 data->whilem_c = data_fake.whilem_c;
5220 if ((flags & SCF_DO_SUBSTR) && data_fake.last_found) {
5221 if (RExC_rx->minlen<*minnextp)
5222 RExC_rx->minlen=*minnextp;
5223 scan_commit(pRExC_state, &data_fake, minnextp, is_inf);
5224 SvREFCNT_dec_NN(data_fake.last_found);
5226 if ( data_fake.minlen_fixed != minlenp )
5228 data->offset_fixed= data_fake.offset_fixed;
5229 data->minlen_fixed= data_fake.minlen_fixed;
5230 data->lookbehind_fixed+= scan->flags;
5232 if ( data_fake.minlen_float != minlenp )
5234 data->minlen_float= data_fake.minlen_float;
5235 data->offset_float_min=data_fake.offset_float_min;
5236 data->offset_float_max=data_fake.offset_float_max;
5237 data->lookbehind_float+= scan->flags;
5244 else if (OP(scan) == OPEN) {
5245 if (stopparen != (I32)ARG(scan))
5248 else if (OP(scan) == CLOSE) {
5249 if (stopparen == (I32)ARG(scan)) {
5252 if ((I32)ARG(scan) == is_par) {
5253 next = regnext(scan);
5255 if ( next && (OP(next) != WHILEM) && next < last)
5256 is_par = 0; /* Disable optimization */
5259 *(data->last_closep) = ARG(scan);
5261 else if (OP(scan) == EVAL) {
5263 data->flags |= SF_HAS_EVAL;
5265 else if ( PL_regkind[OP(scan)] == ENDLIKE ) {
5266 if (flags & SCF_DO_SUBSTR) {
5267 scan_commit(pRExC_state, data, minlenp, is_inf);
5268 flags &= ~SCF_DO_SUBSTR;
5270 if (data && OP(scan)==ACCEPT) {
5271 data->flags |= SCF_SEEN_ACCEPT;
5276 else if (OP(scan) == LOGICAL && scan->flags == 2) /* Embedded follows */
5278 if (flags & SCF_DO_SUBSTR) {
5279 scan_commit(pRExC_state, data, minlenp, is_inf);
5280 data->longest = &(data->longest_float);
5282 is_inf = is_inf_internal = 1;
5283 if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
5284 ssc_anything(data->start_class);
5285 flags &= ~SCF_DO_STCLASS;
5287 else if (OP(scan) == GPOS) {
5288 if (!(RExC_rx->intflags & PREGf_GPOS_FLOAT) &&
5289 !(delta || is_inf || (data && data->pos_delta)))
5291 if (!(RExC_rx->intflags & PREGf_ANCH) && (flags & SCF_DO_SUBSTR))
5292 RExC_rx->intflags |= PREGf_ANCH_GPOS;
5293 if (RExC_rx->gofs < (STRLEN)min)
5294 RExC_rx->gofs = min;
5296 RExC_rx->intflags |= PREGf_GPOS_FLOAT;
5300 #ifdef TRIE_STUDY_OPT
5301 #ifdef FULL_TRIE_STUDY
5302 else if (PL_regkind[OP(scan)] == TRIE) {
5303 /* NOTE - There is similar code to this block above for handling
5304 BRANCH nodes on the initial study. If you change stuff here
5306 regnode *trie_node= scan;
5307 regnode *tail= regnext(scan);
5308 reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
5309 SSize_t max1 = 0, min1 = SSize_t_MAX;
5312 if (flags & SCF_DO_SUBSTR) { /* XXXX Add !SUSPEND? */
5313 /* Cannot merge strings after this. */
5314 scan_commit(pRExC_state, data, minlenp, is_inf);
5316 if (flags & SCF_DO_STCLASS)
5317 ssc_init_zero(pRExC_state, &accum);
5323 const regnode *nextbranch= NULL;
5326 for ( word=1 ; word <= trie->wordcount ; word++)
5328 SSize_t deltanext=0, minnext=0, f = 0, fake;
5329 regnode_ssc this_class;
5331 data_fake.flags = 0;
5333 data_fake.whilem_c = data->whilem_c;
5334 data_fake.last_closep = data->last_closep;
5337 data_fake.last_closep = &fake;
5338 data_fake.pos_delta = delta;
5339 if (flags & SCF_DO_STCLASS) {
5340 ssc_init(pRExC_state, &this_class);
5341 data_fake.start_class = &this_class;
5342 f = SCF_DO_STCLASS_AND;
5344 if (flags & SCF_WHILEM_VISITED_POS)
5345 f |= SCF_WHILEM_VISITED_POS;
5347 if (trie->jump[word]) {
5349 nextbranch = trie_node + trie->jump[0];
5350 scan= trie_node + trie->jump[word];
5351 /* We go from the jump point to the branch that follows
5352 it. Note this means we need the vestigal unused
5353 branches even though they arent otherwise used. */
5354 minnext = study_chunk(pRExC_state, &scan, minlenp,
5355 &deltanext, (regnode *)nextbranch, &data_fake,
5356 stopparen, recursed_depth, NULL, f,depth+1);
5358 if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
5359 nextbranch= regnext((regnode*)nextbranch);
5361 if (min1 > (SSize_t)(minnext + trie->minlen))
5362 min1 = minnext + trie->minlen;
5363 if (deltanext == SSize_t_MAX) {
5364 is_inf = is_inf_internal = 1;
5366 } else if (max1 < (SSize_t)(minnext + deltanext + trie->maxlen))
5367 max1 = minnext + deltanext + trie->maxlen;
5369 if (data_fake.flags & (SF_HAS_PAR|SF_IN_PAR))
5371 if (data_fake.flags & SCF_SEEN_ACCEPT) {
5372 if ( stopmin > min + min1)
5373 stopmin = min + min1;
5374 flags &= ~SCF_DO_SUBSTR;
5376 data->flags |= SCF_SEEN_ACCEPT;
5379 if (data_fake.flags & SF_HAS_EVAL)
5380 data->flags |= SF_HAS_EVAL;
5381 data->whilem_c = data_fake.whilem_c;
5383 if (flags & SCF_DO_STCLASS)
5384 ssc_or(pRExC_state, &accum, (regnode_charclass *) &this_class);
5387 if (flags & SCF_DO_SUBSTR) {
5388 data->pos_min += min1;
5389 data->pos_delta += max1 - min1;
5390 if (max1 != min1 || is_inf)
5391 data->longest = &(data->longest_float);
5394 delta += max1 - min1;
5395 if (flags & SCF_DO_STCLASS_OR) {
5396 ssc_or(pRExC_state, data->start_class, (regnode_charclass *) &accum);
5398 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
5399 flags &= ~SCF_DO_STCLASS;
5402 else if (flags & SCF_DO_STCLASS_AND) {
5404 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) &accum);
5405 flags &= ~SCF_DO_STCLASS;
5408 /* Switch to OR mode: cache the old value of
5409 * data->start_class */
5411 StructCopy(data->start_class, and_withp, regnode_ssc);
5412 flags &= ~SCF_DO_STCLASS_AND;
5413 StructCopy(&accum, data->start_class, regnode_ssc);
5414 flags |= SCF_DO_STCLASS_OR;
5421 else if (PL_regkind[OP(scan)] == TRIE) {
5422 reg_trie_data *trie = (reg_trie_data*)RExC_rxi->data->data[ ARG(scan) ];
5425 min += trie->minlen;
5426 delta += (trie->maxlen - trie->minlen);
5427 flags &= ~SCF_DO_STCLASS; /* xxx */
5428 if (flags & SCF_DO_SUBSTR) {
5429 /* Cannot expect anything... */
5430 scan_commit(pRExC_state, data, minlenp, is_inf);
5431 data->pos_min += trie->minlen;
5432 data->pos_delta += (trie->maxlen - trie->minlen);
5433 if (trie->maxlen != trie->minlen)
5434 data->longest = &(data->longest_float);
5436 if (trie->jump) /* no more substrings -- for now /grr*/
5437 flags &= ~SCF_DO_SUBSTR;
5439 #endif /* old or new */
5440 #endif /* TRIE_STUDY_OPT */
5442 /* Else: zero-length, ignore. */
5443 scan = regnext(scan);
5445 /* If we are exiting a recursion we can unset its recursed bit
5446 * and allow ourselves to enter it again - no danger of an
5447 * infinite loop there.
5448 if (stopparen > -1 && recursed) {
5449 DEBUG_STUDYDATA("unset:", data,depth);
5450 PAREN_UNSET( recursed, stopparen);
5454 DEBUG_STUDYDATA("frame-end:",data,depth);
5455 DEBUG_PEEP("fend", scan, depth);
5456 /* restore previous context */
5459 stopparen = frame->stop;
5460 recursed_depth = frame->prev_recursed_depth;
5463 frame = frame->prev;
5464 goto fake_study_recurse;
5469 DEBUG_STUDYDATA("pre-fin:",data,depth);
5472 *deltap = is_inf_internal ? SSize_t_MAX : delta;
5474 if (flags & SCF_DO_SUBSTR && is_inf)
5475 data->pos_delta = SSize_t_MAX - data->pos_min;
5476 if (is_par > (I32)U8_MAX)
5478 if (is_par && pars==1 && data) {
5479 data->flags |= SF_IN_PAR;
5480 data->flags &= ~SF_HAS_PAR;
5482 else if (pars && data) {
5483 data->flags |= SF_HAS_PAR;
5484 data->flags &= ~SF_IN_PAR;
5486 if (flags & SCF_DO_STCLASS_OR)
5487 ssc_and(pRExC_state, data->start_class, (regnode_charclass *) and_withp);
5488 if (flags & SCF_TRIE_RESTUDY)
5489 data->flags |= SCF_TRIE_RESTUDY;
5491 DEBUG_STUDYDATA("post-fin:",data,depth);
5494 SSize_t final_minlen= min < stopmin ? min : stopmin;
5496 if (!(RExC_seen & REG_UNBOUNDED_QUANTIFIER_SEEN) && (RExC_maxlen < final_minlen + delta)) {
5497 RExC_maxlen = final_minlen + delta;
5499 return final_minlen;
5505 S_add_data(RExC_state_t* const pRExC_state, const char* const s, const U32 n)
5507 U32 count = RExC_rxi->data ? RExC_rxi->data->count : 0;
5509 PERL_ARGS_ASSERT_ADD_DATA;
5511 Renewc(RExC_rxi->data,
5512 sizeof(*RExC_rxi->data) + sizeof(void*) * (count + n - 1),
5513 char, struct reg_data);
5515 Renew(RExC_rxi->data->what, count + n, U8);
5517 Newx(RExC_rxi->data->what, n, U8);
5518 RExC_rxi->data->count = count + n;
5519 Copy(s, RExC_rxi->data->what + count, n, U8);
5523 /*XXX: todo make this not included in a non debugging perl, but appears to be
5524 * used anyway there, in 'use re' */
5525 #ifndef PERL_IN_XSUB_RE
5527 Perl_reginitcolors(pTHX)
5529 const char * const s = PerlEnv_getenv("PERL_RE_COLORS");
5531 char *t = savepv(s);
5535 t = strchr(t, '\t');
5541 PL_colors[i] = t = (char *)"";
5546 PL_colors[i++] = (char *)"";
5553 #ifdef TRIE_STUDY_OPT
5554 #define CHECK_RESTUDY_GOTO_butfirst(dOsomething) \
5557 (data.flags & SCF_TRIE_RESTUDY) \
5565 #define CHECK_RESTUDY_GOTO_butfirst
5569 * pregcomp - compile a regular expression into internal code
5571 * Decides which engine's compiler to call based on the hint currently in
5575 #ifndef PERL_IN_XSUB_RE
5577 /* return the currently in-scope regex engine (or the default if none) */
5579 regexp_engine const *
5580 Perl_current_re_engine(pTHX)
5582 if (IN_PERL_COMPILETIME) {
5583 HV * const table = GvHV(PL_hintgv);
5586 if (!table || !(PL_hints & HINT_LOCALIZE_HH))
5587 return &reh_regexp_engine;
5588 ptr = hv_fetchs(table, "regcomp", FALSE);
5589 if ( !(ptr && SvIOK(*ptr) && SvIV(*ptr)))
5590 return &reh_regexp_engine;
5591 return INT2PTR(regexp_engine*,SvIV(*ptr));
5595 if (!PL_curcop->cop_hints_hash)
5596 return &reh_regexp_engine;
5597 ptr = cop_hints_fetch_pvs(PL_curcop, "regcomp", 0);
5598 if ( !(ptr && SvIOK(ptr) && SvIV(ptr)))
5599 return &reh_regexp_engine;
5600 return INT2PTR(regexp_engine*,SvIV(ptr));
5606 Perl_pregcomp(pTHX_ SV * const pattern, const U32 flags)
5608 regexp_engine const *eng = current_re_engine();
5609 GET_RE_DEBUG_FLAGS_DECL;
5611 PERL_ARGS_ASSERT_PREGCOMP;
5613 /* Dispatch a request to compile a regexp to correct regexp engine. */
5615 PerlIO_printf(Perl_debug_log, "Using engine %"UVxf"\n",
5618 return CALLREGCOMP_ENG(eng, pattern, flags);
5622 /* public(ish) entry point for the perl core's own regex compiling code.
5623 * It's actually a wrapper for Perl_re_op_compile that only takes an SV
5624 * pattern rather than a list of OPs, and uses the internal engine rather
5625 * than the current one */
5628 Perl_re_compile(pTHX_ SV * const pattern, U32 rx_flags)
5630 SV *pat = pattern; /* defeat constness! */
5631 PERL_ARGS_ASSERT_RE_COMPILE;
5632 return Perl_re_op_compile(aTHX_ &pat, 1, NULL,
5633 #ifdef PERL_IN_XSUB_RE
5638 NULL, NULL, rx_flags, 0);
5642 /* upgrade pattern pat_p of length plen_p to UTF8, and if there are code
5643 * blocks, recalculate the indices. Update pat_p and plen_p in-place to
5644 * point to the realloced string and length.
5646 * This is essentially a copy of Perl_bytes_to_utf8() with the code index
5650 S_pat_upgrade_to_utf8(pTHX_ RExC_state_t * const pRExC_state,
5651 char **pat_p, STRLEN *plen_p, int num_code_blocks)
5653 U8 *const src = (U8*)*pat_p;
5656 STRLEN s = 0, d = 0;
5658 GET_RE_DEBUG_FLAGS_DECL;
5660 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log,
5661 "UTF8 mismatch! Converting to utf8 for resizing and compile\n"));
5663 Newx(dst, *plen_p * 2 + 1, U8);
5665 while (s < *plen_p) {
5666 if (NATIVE_BYTE_IS_INVARIANT(src[s]))
5669 dst[d++] = UTF8_EIGHT_BIT_HI(src[s]);
5670 dst[d] = UTF8_EIGHT_BIT_LO(src[s]);
5672 if (n < num_code_blocks) {
5673 if (!do_end && pRExC_state->code_blocks[n].start == s) {
5674 pRExC_state->code_blocks[n].start = d;
5675 assert(dst[d] == '(');
5678 else if (do_end && pRExC_state->code_blocks[n].end == s) {
5679 pRExC_state->code_blocks[n].end = d;
5680 assert(dst[d] == ')');
5690 *pat_p = (char*) dst;
5692 RExC_orig_utf8 = RExC_utf8 = 1;
5697 /* S_concat_pat(): concatenate a list of args to the pattern string pat,
5698 * while recording any code block indices, and handling overloading,
5699 * nested qr// objects etc. If pat is null, it will allocate a new
5700 * string, or just return the first arg, if there's only one.
5702 * Returns the malloced/updated pat.
5703 * patternp and pat_count is the array of SVs to be concatted;
5704 * oplist is the optional list of ops that generated the SVs;
5705 * recompile_p is a pointer to a boolean that will be set if
5706 * the regex will need to be recompiled.
5707 * delim, if non-null is an SV that will be inserted between each element
5711 S_concat_pat(pTHX_ RExC_state_t * const pRExC_state,
5712 SV *pat, SV ** const patternp, int pat_count,
5713 OP *oplist, bool *recompile_p, SV *delim)
5717 bool use_delim = FALSE;
5718 bool alloced = FALSE;
5720 /* if we know we have at least two args, create an empty string,
5721 * then concatenate args to that. For no args, return an empty string */
5722 if (!pat && pat_count != 1) {
5728 for (svp = patternp; svp < patternp + pat_count; svp++) {
5731 STRLEN orig_patlen = 0;
5733 SV *msv = use_delim ? delim : *svp;
5734 if (!msv) msv = &PL_sv_undef;
5736 /* if we've got a delimiter, we go round the loop twice for each
5737 * svp slot (except the last), using the delimiter the second
5746 if (SvTYPE(msv) == SVt_PVAV) {
5747 /* we've encountered an interpolated array within
5748 * the pattern, e.g. /...@a..../. Expand the list of elements,
5749 * then recursively append elements.
5750 * The code in this block is based on S_pushav() */
5752 AV *const av = (AV*)msv;
5753 const SSize_t maxarg = AvFILL(av) + 1;
5757 assert(oplist->op_type == OP_PADAV
5758 || oplist->op_type == OP_RV2AV);
5759 oplist = OP_SIBLING(oplist);
5762 if (SvRMAGICAL(av)) {
5765 Newx(array, maxarg, SV*);
5767 for (i=0; i < maxarg; i++) {
5768 SV ** const svp = av_fetch(av, i, FALSE);
5769 array[i] = svp ? *svp : &PL_sv_undef;
5773 array = AvARRAY(av);
5775 pat = S_concat_pat(aTHX_ pRExC_state, pat,
5776 array, maxarg, NULL, recompile_p,
5778 GvSV((gv_fetchpvs("\"", GV_ADDMULTI, SVt_PV))));
5784 /* we make the assumption here that each op in the list of
5785 * op_siblings maps to one SV pushed onto the stack,
5786 * except for code blocks, with have both an OP_NULL and
5788 * This allows us to match up the list of SVs against the
5789 * list of OPs to find the next code block.
5791 * Note that PUSHMARK PADSV PADSV ..
5793 * PADRANGE PADSV PADSV ..
5794 * so the alignment still works. */
5797 if (oplist->op_type == OP_NULL
5798 && (oplist->op_flags & OPf_SPECIAL))
5800 assert(n < pRExC_state->num_code_blocks);
5801 pRExC_state->code_blocks[n].start = pat ? SvCUR(pat) : 0;
5802 pRExC_state->code_blocks[n].block = oplist;
5803 pRExC_state->code_blocks[n].src_regex = NULL;
5806 oplist = OP_SIBLING(oplist); /* skip CONST */
5809 oplist = OP_SIBLING(oplist);;
5812 /* apply magic and QR overloading to arg */
5815 if (SvROK(msv) && SvAMAGIC(msv)) {
5816 SV *sv = AMG_CALLunary(msv, regexp_amg);
5820 if (SvTYPE(sv) != SVt_REGEXP)
5821 Perl_croak(aTHX_ "Overloaded qr did not return a REGEXP");
5826 /* try concatenation overload ... */
5827 if (pat && (SvAMAGIC(pat) || SvAMAGIC(msv)) &&
5828 (sv = amagic_call(pat, msv, concat_amg, AMGf_assign)))
5831 /* overloading involved: all bets are off over literal
5832 * code. Pretend we haven't seen it */
5833 pRExC_state->num_code_blocks -= n;
5837 /* ... or failing that, try "" overload */
5838 while (SvAMAGIC(msv)
5839 && (sv = AMG_CALLunary(msv, string_amg))
5843 && SvRV(msv) == SvRV(sv))
5848 if (SvROK(msv) && SvTYPE(SvRV(msv)) == SVt_REGEXP)
5852 /* this is a partially unrolled
5853 * sv_catsv_nomg(pat, msv);
5854 * that allows us to adjust code block indices if
5857 char *dst = SvPV_force_nomg(pat, dlen);
5859 if (SvUTF8(msv) && !SvUTF8(pat)) {
5860 S_pat_upgrade_to_utf8(aTHX_ pRExC_state, &dst, &dlen, n);
5861 sv_setpvn(pat, dst, dlen);
5864 sv_catsv_nomg(pat, msv);
5871 pRExC_state->code_blocks[n-1].end = SvCUR(pat)-1;
5874 /* extract any code blocks within any embedded qr//'s */
5875 if (rx && SvTYPE(rx) == SVt_REGEXP
5876 && RX_ENGINE((REGEXP*)rx)->op_comp)
5879 RXi_GET_DECL(ReANY((REGEXP *)rx), ri);
5880 if (ri->num_code_blocks) {
5882 /* the presence of an embedded qr// with code means
5883 * we should always recompile: the text of the
5884 * qr// may not have changed, but it may be a
5885 * different closure than last time */
5887 Renew(pRExC_state->code_blocks,
5888 pRExC_state->num_code_blocks + ri->num_code_blocks,
5889 struct reg_code_block);
5890 pRExC_state->num_code_blocks += ri->num_code_blocks;
5892 for (i=0; i < ri->num_code_blocks; i++) {
5893 struct reg_code_block *src, *dst;
5894 STRLEN offset = orig_patlen
5895 + ReANY((REGEXP *)rx)->pre_prefix;
5896 assert(n < pRExC_state->num_code_blocks);
5897 src = &ri->code_blocks[i];
5898 dst = &pRExC_state->code_blocks[n];
5899 dst->start = src->start + offset;
5900 dst->end = src->end + offset;
5901 dst->block = src->block;
5902 dst->src_regex = (REGEXP*) SvREFCNT_inc( (SV*)
5911 /* avoid calling magic multiple times on a single element e.g. =~ $qr */
5920 /* see if there are any run-time code blocks in the pattern.
5921 * False positives are allowed */
5924 S_has_runtime_code(pTHX_ RExC_state_t * const pRExC_state,
5925 char *pat, STRLEN plen)
5930 PERL_UNUSED_CONTEXT;
5932 for (s = 0; s < plen; s++) {
5933 if (n < pRExC_state->num_code_blocks
5934 && s == pRExC_state->code_blocks[n].start)
5936 s = pRExC_state->code_blocks[n].end;
5940 /* TODO ideally should handle [..], (#..), /#.../x to reduce false
5942 if (pat[s] == '(' && s+2 <= plen && pat[s+1] == '?' &&
5944 || (s + 2 <= plen && pat[s+2] == '?' && pat[s+3] == '{'))
5951 /* Handle run-time code blocks. We will already have compiled any direct
5952 * or indirect literal code blocks. Now, take the pattern 'pat' and make a
5953 * copy of it, but with any literal code blocks blanked out and
5954 * appropriate chars escaped; then feed it into
5956 * eval "qr'modified_pattern'"
5960 * a\bc(?{"this was literal"})def'ghi\\jkl(?{"this is runtime"})mno
5964 * qr'a\\bc_______________________def\'ghi\\\\jkl(?{"this is runtime"})mno'
5966 * After eval_sv()-ing that, grab any new code blocks from the returned qr
5967 * and merge them with any code blocks of the original regexp.
5969 * If the pat is non-UTF8, while the evalled qr is UTF8, don't merge;
5970 * instead, just save the qr and return FALSE; this tells our caller that
5971 * the original pattern needs upgrading to utf8.
5975 S_compile_runtime_code(pTHX_ RExC_state_t * const pRExC_state,
5976 char *pat, STRLEN plen)
5980 GET_RE_DEBUG_FLAGS_DECL;
5982 if (pRExC_state->runtime_code_qr) {
5983 /* this is the second time we've been called; this should
5984 * only happen if the main pattern got upgraded to utf8
5985 * during compilation; re-use the qr we compiled first time
5986 * round (which should be utf8 too)
5988 qr = pRExC_state->runtime_code_qr;
5989 pRExC_state->runtime_code_qr = NULL;
5990 assert(RExC_utf8 && SvUTF8(qr));
5996 int newlen = plen + 6; /* allow for "qr''x\0" extra chars */
6000 /* determine how many extra chars we need for ' and \ escaping */
6001 for (s = 0; s < plen; s++) {
6002 if (pat[s] == '\'' || pat[s] == '\\')
6006 Newx(newpat, newlen, char);
6008 *p++ = 'q'; *p++ = 'r'; *p++ = '\'';
6010 for (s = 0; s < plen; s++) {
6011 if (n < pRExC_state->num_code_blocks
6012 && s == pRExC_state->code_blocks[n].start)
6014 /* blank out literal code block */
6015 assert(pat[s] == '(');
6016 while (s <= pRExC_state->code_blocks[n].end) {
6024 if (pat[s] == '\'' || pat[s] == '\\')
6029 if (pRExC_state->pm_flags & RXf_PMf_EXTENDED)
6033 PerlIO_printf(Perl_debug_log,
6034 "%sre-parsing pattern for runtime code:%s %s\n",
6035 PL_colors[4],PL_colors[5],newpat);
6038 sv = newSVpvn_flags(newpat, p-newpat-1, RExC_utf8 ? SVf_UTF8 : 0);
6044 PUSHSTACKi(PERLSI_REQUIRE);
6045 /* G_RE_REPARSING causes the toker to collapse \\ into \ when
6046 * parsing qr''; normally only q'' does this. It also alters
6048 eval_sv(sv, G_SCALAR|G_RE_REPARSING);
6049 SvREFCNT_dec_NN(sv);
6054 SV * const errsv = ERRSV;
6055 if (SvTRUE_NN(errsv))
6057 Safefree(pRExC_state->code_blocks);
6058 /* use croak_sv ? */
6059 Perl_croak_nocontext("%"SVf, SVfARG(errsv));
6062 assert(SvROK(qr_ref));
6064 assert(SvTYPE(qr) == SVt_REGEXP && RX_ENGINE((REGEXP*)qr)->op_comp);
6065 /* the leaving below frees the tmp qr_ref.
6066 * Give qr a life of its own */
6074 if (!RExC_utf8 && SvUTF8(qr)) {
6075 /* first time through; the pattern got upgraded; save the
6076 * qr for the next time through */
6077 assert(!pRExC_state->runtime_code_qr);
6078 pRExC_state->runtime_code_qr = qr;
6083 /* extract any code blocks within the returned qr// */
6086 /* merge the main (r1) and run-time (r2) code blocks into one */
6088 RXi_GET_DECL(ReANY((REGEXP *)qr), r2);
6089 struct reg_code_block *new_block, *dst;
6090 RExC_state_t * const r1 = pRExC_state; /* convenient alias */
6093 if (!r2->num_code_blocks) /* we guessed wrong */
6095 SvREFCNT_dec_NN(qr);
6100 r1->num_code_blocks + r2->num_code_blocks,
6101 struct reg_code_block);
6104 while ( i1 < r1->num_code_blocks
6105 || i2 < r2->num_code_blocks)
6107 struct reg_code_block *src;
6110 if (i1 == r1->num_code_blocks) {
6111 src = &r2->code_blocks[i2++];
6114 else if (i2 == r2->num_code_blocks)
6115 src = &r1->code_blocks[i1++];
6116 else if ( r1->code_blocks[i1].start
6117 < r2->code_blocks[i2].start)
6119 src = &r1->code_blocks[i1++];
6120 assert(src->end < r2->code_blocks[i2].start);
6123 assert( r1->code_blocks[i1].start
6124 > r2->code_blocks[i2].start);
6125 src = &r2->code_blocks[i2++];
6127 assert(src->end < r1->code_blocks[i1].start);
6130 assert(pat[src->start] == '(');
6131 assert(pat[src->end] == ')');
6132 dst->start = src->start;
6133 dst->end = src->end;
6134 dst->block = src->block;
6135 dst->src_regex = is_qr ? (REGEXP*) SvREFCNT_inc( (SV*) qr)
6139 r1->num_code_blocks += r2->num_code_blocks;
6140 Safefree(r1->code_blocks);
6141 r1->code_blocks = new_block;
6144 SvREFCNT_dec_NN(qr);
6150 S_setup_longest(pTHX_ RExC_state_t *pRExC_state, SV* sv_longest,
6151 SV** rx_utf8, SV** rx_substr, SSize_t* rx_end_shift,
6152 SSize_t lookbehind, SSize_t offset, SSize_t *minlen,
6153 STRLEN longest_length, bool eol, bool meol)
6155 /* This is the common code for setting up the floating and fixed length
6156 * string data extracted from Perl_re_op_compile() below. Returns a boolean
6157 * as to whether succeeded or not */
6162 if (! (longest_length
6163 || (eol /* Can't have SEOL and MULTI */
6164 && (! meol || (RExC_flags & RXf_PMf_MULTILINE)))
6166 /* See comments for join_exact for why REG_UNFOLDED_MULTI_SEEN */
6167 || (RExC_seen & REG_UNFOLDED_MULTI_SEEN))
6172 /* copy the information about the longest from the reg_scan_data
6173 over to the program. */
6174 if (SvUTF8(sv_longest)) {
6175 *rx_utf8 = sv_longest;
6178 *rx_substr = sv_longest;
6181 /* end_shift is how many chars that must be matched that
6182 follow this item. We calculate it ahead of time as once the
6183 lookbehind offset is added in we lose the ability to correctly
6185 ml = minlen ? *(minlen) : (SSize_t)longest_length;
6186 *rx_end_shift = ml - offset
6187 - longest_length + (SvTAIL(sv_longest) != 0)
6190 t = (eol/* Can't have SEOL and MULTI */
6191 && (! meol || (RExC_flags & RXf_PMf_MULTILINE)));
6192 fbm_compile(sv_longest, t ? FBMcf_TAIL : 0);
6198 * Perl_re_op_compile - the perl internal RE engine's function to compile a
6199 * regular expression into internal code.
6200 * The pattern may be passed either as:
6201 * a list of SVs (patternp plus pat_count)
6202 * a list of OPs (expr)
6203 * If both are passed, the SV list is used, but the OP list indicates
6204 * which SVs are actually pre-compiled code blocks
6206 * The SVs in the list have magic and qr overloading applied to them (and
6207 * the list may be modified in-place with replacement SVs in the latter
6210 * If the pattern hasn't changed from old_re, then old_re will be
6213 * eng is the current engine. If that engine has an op_comp method, then
6214 * handle directly (i.e. we assume that op_comp was us); otherwise, just
6215 * do the initial concatenation of arguments and pass on to the external
6218 * If is_bare_re is not null, set it to a boolean indicating whether the
6219 * arg list reduced (after overloading) to a single bare regex which has
6220 * been returned (i.e. /$qr/).
6222 * orig_rx_flags contains RXf_* flags. See perlreapi.pod for more details.
6224 * pm_flags contains the PMf_* flags, typically based on those from the
6225 * pm_flags field of the related PMOP. Currently we're only interested in
6226 * PMf_HAS_CV, PMf_IS_QR, PMf_USE_RE_EVAL.
6228 * We can't allocate space until we know how big the compiled form will be,
6229 * but we can't compile it (and thus know how big it is) until we've got a
6230 * place to put the code. So we cheat: we compile it twice, once with code
6231 * generation turned off and size counting turned on, and once "for real".
6232 * This also means that we don't allocate space until we are sure that the
6233 * thing really will compile successfully, and we never have to move the
6234 * code and thus invalidate pointers into it. (Note that it has to be in
6235 * one piece because free() must be able to free it all.) [NB: not true in perl]
6237 * Beware that the optimization-preparation code in here knows about some
6238 * of the structure of the compiled regexp. [I'll say.]
6242 Perl_re_op_compile(pTHX_ SV ** const patternp, int pat_count,
6243 OP *expr, const regexp_engine* eng, REGEXP *old_re,
6244 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags)
6248 regexp_internal *ri;
6256 SV *code_blocksv = NULL;
6257 SV** new_patternp = patternp;
6259 /* these are all flags - maybe they should be turned
6260 * into a single int with different bit masks */
6261 I32 sawlookahead = 0;
6266 regex_charset initial_charset = get_regex_charset(orig_rx_flags);
6268 bool runtime_code = 0;
6270 RExC_state_t RExC_state;
6271 RExC_state_t * const pRExC_state = &RExC_state;
6272 #ifdef TRIE_STUDY_OPT
6274 RExC_state_t copyRExC_state;
6276 GET_RE_DEBUG_FLAGS_DECL;
6278 PERL_ARGS_ASSERT_RE_OP_COMPILE;
6280 DEBUG_r(if (!PL_colorset) reginitcolors());
6282 #ifndef PERL_IN_XSUB_RE
6283 /* Initialize these here instead of as-needed, as is quick and avoids
6284 * having to test them each time otherwise */
6285 if (! PL_AboveLatin1) {
6286 PL_AboveLatin1 = _new_invlist_C_array(AboveLatin1_invlist);
6287 PL_Latin1 = _new_invlist_C_array(Latin1_invlist);
6288 PL_UpperLatin1 = _new_invlist_C_array(UpperLatin1_invlist);
6289 PL_utf8_foldable = _new_invlist_C_array(_Perl_Any_Folds_invlist);
6290 PL_HasMultiCharFold =
6291 _new_invlist_C_array(_Perl_Folds_To_Multi_Char_invlist);
6295 pRExC_state->code_blocks = NULL;
6296 pRExC_state->num_code_blocks = 0;
6299 *is_bare_re = FALSE;
6301 if (expr && (expr->op_type == OP_LIST ||
6302 (expr->op_type == OP_NULL && expr->op_targ == OP_LIST))) {
6303 /* allocate code_blocks if needed */
6307 for (o = cLISTOPx(expr)->op_first; o; o = OP_SIBLING(o))
6308 if (o->op_type == OP_NULL && (o->op_flags & OPf_SPECIAL))
6309 ncode++; /* count of DO blocks */
6311 pRExC_state->num_code_blocks = ncode;
6312 Newx(pRExC_state->code_blocks, ncode, struct reg_code_block);
6317 /* compile-time pattern with just OP_CONSTs and DO blocks */
6322 /* find how many CONSTs there are */
6325 if (expr->op_type == OP_CONST)
6328 for (o = cLISTOPx(expr)->op_first; o; o = OP_SIBLING(o)) {
6329 if (o->op_type == OP_CONST)
6333 /* fake up an SV array */
6335 assert(!new_patternp);
6336 Newx(new_patternp, n, SV*);
6337 SAVEFREEPV(new_patternp);
6341 if (expr->op_type == OP_CONST)
6342 new_patternp[n] = cSVOPx_sv(expr);
6344 for (o = cLISTOPx(expr)->op_first; o; o = OP_SIBLING(o)) {
6345 if (o->op_type == OP_CONST)
6346 new_patternp[n++] = cSVOPo_sv;
6351 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log,
6352 "Assembling pattern from %d elements%s\n", pat_count,
6353 orig_rx_flags & RXf_SPLIT ? " for split" : ""));
6355 /* set expr to the first arg op */
6357 if (pRExC_state->num_code_blocks
6358 && expr->op_type != OP_CONST)
6360 expr = cLISTOPx(expr)->op_first;
6361 assert( expr->op_type == OP_PUSHMARK
6362 || (expr->op_type == OP_NULL && expr->op_targ == OP_PUSHMARK)
6363 || expr->op_type == OP_PADRANGE);
6364 expr = OP_SIBLING(expr);
6367 pat = S_concat_pat(aTHX_ pRExC_state, NULL, new_patternp, pat_count,
6368 expr, &recompile, NULL);
6370 /* handle bare (possibly after overloading) regex: foo =~ $re */
6375 if (SvTYPE(re) == SVt_REGEXP) {
6379 Safefree(pRExC_state->code_blocks);
6380 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log,
6381 "Precompiled pattern%s\n",
6382 orig_rx_flags & RXf_SPLIT ? " for split" : ""));
6388 exp = SvPV_nomg(pat, plen);
6390 if (!eng->op_comp) {
6391 if ((SvUTF8(pat) && IN_BYTES)
6392 || SvGMAGICAL(pat) || SvAMAGIC(pat))
6394 /* make a temporary copy; either to convert to bytes,
6395 * or to avoid repeating get-magic / overloaded stringify */
6396 pat = newSVpvn_flags(exp, plen, SVs_TEMP |
6397 (IN_BYTES ? 0 : SvUTF8(pat)));
6399 Safefree(pRExC_state->code_blocks);
6400 return CALLREGCOMP_ENG(eng, pat, orig_rx_flags);
6403 /* ignore the utf8ness if the pattern is 0 length */
6404 RExC_utf8 = RExC_orig_utf8 = (plen == 0 || IN_BYTES) ? 0 : SvUTF8(pat);
6405 RExC_uni_semantics = 0;
6406 RExC_contains_locale = 0;
6407 RExC_contains_i = 0;
6408 pRExC_state->runtime_code_qr = NULL;
6411 SV *dsv= sv_newmortal();
6412 RE_PV_QUOTED_DECL(s, RExC_utf8, dsv, exp, plen, 60);
6413 PerlIO_printf(Perl_debug_log, "%sCompiling REx%s %s\n",
6414 PL_colors[4],PL_colors[5],s);
6418 /* we jump here if we upgrade the pattern to utf8 and have to
6421 if ((pm_flags & PMf_USE_RE_EVAL)
6422 /* this second condition covers the non-regex literal case,
6423 * i.e. $foo =~ '(?{})'. */
6424 || (IN_PERL_COMPILETIME && (PL_hints & HINT_RE_EVAL))
6426 runtime_code = S_has_runtime_code(aTHX_ pRExC_state, exp, plen);
6428 /* return old regex if pattern hasn't changed */
6429 /* XXX: note in the below we have to check the flags as well as the
6432 * Things get a touch tricky as we have to compare the utf8 flag
6433 * independently from the compile flags. */
6437 && !!RX_UTF8(old_re) == !!RExC_utf8
6438 && ( RX_COMPFLAGS(old_re) == ( orig_rx_flags & RXf_PMf_FLAGCOPYMASK ) )
6439 && RX_PRECOMP(old_re)
6440 && RX_PRELEN(old_re) == plen
6441 && memEQ(RX_PRECOMP(old_re), exp, plen)
6442 && !runtime_code /* with runtime code, always recompile */ )
6444 Safefree(pRExC_state->code_blocks);
6448 rx_flags = orig_rx_flags;
6450 if (rx_flags & PMf_FOLD) {
6451 RExC_contains_i = 1;
6453 if (RExC_utf8 && initial_charset == REGEX_DEPENDS_CHARSET) {
6455 /* Set to use unicode semantics if the pattern is in utf8 and has the
6456 * 'depends' charset specified, as it means unicode when utf8 */
6457 set_regex_charset(&rx_flags, REGEX_UNICODE_CHARSET);
6461 RExC_flags = rx_flags;
6462 RExC_pm_flags = pm_flags;
6465 if (TAINTING_get && TAINT_get)
6466 Perl_croak(aTHX_ "Eval-group in insecure regular expression");
6468 if (!S_compile_runtime_code(aTHX_ pRExC_state, exp, plen)) {
6469 /* whoops, we have a non-utf8 pattern, whilst run-time code
6470 * got compiled as utf8. Try again with a utf8 pattern */
6471 S_pat_upgrade_to_utf8(aTHX_ pRExC_state, &exp, &plen,
6472 pRExC_state->num_code_blocks);
6473 goto redo_first_pass;
6476 assert(!pRExC_state->runtime_code_qr);
6482 RExC_in_lookbehind = 0;
6483 RExC_seen_zerolen = *exp == '^' ? -1 : 0;
6485 RExC_override_recoding = 0;
6486 RExC_in_multi_char_class = 0;
6488 /* First pass: determine size, legality. */
6491 RExC_end = exp + plen;
6496 RExC_emit = (regnode *) &RExC_emit_dummy;
6497 RExC_whilem_seen = 0;
6498 RExC_open_parens = NULL;
6499 RExC_close_parens = NULL;
6501 RExC_paren_names = NULL;
6503 RExC_paren_name_list = NULL;
6505 RExC_recurse = NULL;
6506 RExC_study_chunk_recursed = NULL;
6507 RExC_study_chunk_recursed_bytes= 0;
6508 RExC_recurse_count = 0;
6509 pRExC_state->code_index = 0;
6511 #if 0 /* REGC() is (currently) a NOP at the first pass.
6512 * Clever compilers notice this and complain. --jhi */
6513 REGC((U8)REG_MAGIC, (char*)RExC_emit);
6516 PerlIO_printf(Perl_debug_log, "Starting first pass (sizing)\n");
6518 RExC_lastparse=NULL;
6520 /* reg may croak on us, not giving us a chance to free
6521 pRExC_state->code_blocks. We cannot SAVEFREEPV it now, as we may
6522 need it to survive as long as the regexp (qr/(?{})/).
6523 We must check that code_blocksv is not already set, because we may
6524 have jumped back to restart the sizing pass. */
6525 if (pRExC_state->code_blocks && !code_blocksv) {
6526 code_blocksv = newSV_type(SVt_PV);
6527 SAVEFREESV(code_blocksv);
6528 SvPV_set(code_blocksv, (char *)pRExC_state->code_blocks);
6529 SvLEN_set(code_blocksv, 1); /*sufficient to make sv_clear free it*/
6531 if (reg(pRExC_state, 0, &flags,1) == NULL) {
6532 /* It's possible to write a regexp in ascii that represents Unicode
6533 codepoints outside of the byte range, such as via \x{100}. If we
6534 detect such a sequence we have to convert the entire pattern to utf8
6535 and then recompile, as our sizing calculation will have been based
6536 on 1 byte == 1 character, but we will need to use utf8 to encode
6537 at least some part of the pattern, and therefore must convert the whole
6540 if (flags & RESTART_UTF8) {
6541 S_pat_upgrade_to_utf8(aTHX_ pRExC_state, &exp, &plen,
6542 pRExC_state->num_code_blocks);
6543 goto redo_first_pass;
6545 Perl_croak(aTHX_ "panic: reg returned NULL to re_op_compile for sizing pass, flags=%#"UVxf"", (UV) flags);
6548 SvLEN_set(code_blocksv,0); /* no you can't have it, sv_clear */
6551 PerlIO_printf(Perl_debug_log,
6552 "Required size %"IVdf" nodes\n"
6553 "Starting second pass (creation)\n",
6556 RExC_lastparse=NULL;
6559 /* The first pass could have found things that force Unicode semantics */
6560 if ((RExC_utf8 || RExC_uni_semantics)
6561 && get_regex_charset(rx_flags) == REGEX_DEPENDS_CHARSET)
6563 set_regex_charset(&rx_flags, REGEX_UNICODE_CHARSET);
6566 /* Small enough for pointer-storage convention?
6567 If extralen==0, this means that we will not need long jumps. */
6568 if (RExC_size >= 0x10000L && RExC_extralen)
6569 RExC_size += RExC_extralen;
6572 if (RExC_whilem_seen > 15)
6573 RExC_whilem_seen = 15;
6575 /* Allocate space and zero-initialize. Note, the two step process
6576 of zeroing when in debug mode, thus anything assigned has to
6577 happen after that */
6578 rx = (REGEXP*) newSV_type(SVt_REGEXP);
6580 Newxc(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode),
6581 char, regexp_internal);
6582 if ( r == NULL || ri == NULL )
6583 FAIL("Regexp out of space");
6585 /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
6586 Zero(ri, sizeof(regexp_internal) + (unsigned)RExC_size * sizeof(regnode),
6589 /* bulk initialize base fields with 0. */
6590 Zero(ri, sizeof(regexp_internal), char);
6593 /* non-zero initialization begins here */
6596 r->extflags = rx_flags;
6597 RXp_COMPFLAGS(r) = orig_rx_flags & RXf_PMf_FLAGCOPYMASK;
6599 if (pm_flags & PMf_IS_QR) {
6600 ri->code_blocks = pRExC_state->code_blocks;
6601 ri->num_code_blocks = pRExC_state->num_code_blocks;
6606 for (n = 0; n < pRExC_state->num_code_blocks; n++)
6607 if (pRExC_state->code_blocks[n].src_regex)
6608 SAVEFREESV(pRExC_state->code_blocks[n].src_regex);
6609 SAVEFREEPV(pRExC_state->code_blocks);
6613 bool has_p = ((r->extflags & RXf_PMf_KEEPCOPY) == RXf_PMf_KEEPCOPY);
6614 bool has_charset = (get_regex_charset(r->extflags)
6615 != REGEX_DEPENDS_CHARSET);
6617 /* The caret is output if there are any defaults: if not all the STD
6618 * flags are set, or if no character set specifier is needed */
6620 (((r->extflags & RXf_PMf_STD_PMMOD) != RXf_PMf_STD_PMMOD)
6622 bool has_runon = ((RExC_seen & REG_RUN_ON_COMMENT_SEEN)
6623 == REG_RUN_ON_COMMENT_SEEN);
6624 U16 reganch = (U16)((r->extflags & RXf_PMf_STD_PMMOD)
6625 >> RXf_PMf_STD_PMMOD_SHIFT);
6626 const char *fptr = STD_PAT_MODS; /*"msix"*/
6628 /* Allocate for the worst case, which is all the std flags are turned
6629 * on. If more precision is desired, we could do a population count of
6630 * the flags set. This could be done with a small lookup table, or by
6631 * shifting, masking and adding, or even, when available, assembly
6632 * language for a machine-language population count.
6633 * We never output a minus, as all those are defaults, so are
6634 * covered by the caret */
6635 const STRLEN wraplen = plen + has_p + has_runon
6636 + has_default /* If needs a caret */
6638 /* If needs a character set specifier */
6639 + ((has_charset) ? MAX_CHARSET_NAME_LENGTH : 0)
6640 + (sizeof(STD_PAT_MODS) - 1)
6641 + (sizeof("(?:)") - 1);
6643 Newx(p, wraplen + 1, char); /* +1 for the ending NUL */
6644 r->xpv_len_u.xpvlenu_pv = p;
6646 SvFLAGS(rx) |= SVf_UTF8;
6649 /* If a default, cover it using the caret */
6651 *p++= DEFAULT_PAT_MOD;
6655 const char* const name = get_regex_charset_name(r->extflags, &len);
6656 Copy(name, p, len, char);
6660 *p++ = KEEPCOPY_PAT_MOD; /*'p'*/
6663 while((ch = *fptr++)) {
6671 Copy(RExC_precomp, p, plen, char);
6672 assert ((RX_WRAPPED(rx) - p) < 16);
6673 r->pre_prefix = p - RX_WRAPPED(rx);
6679 SvCUR_set(rx, p - RX_WRAPPED(rx));
6683 r->nparens = RExC_npar - 1; /* set early to validate backrefs */
6685 /* setup various meta data about recursion, this all requires
6686 * RExC_npar to be correctly set, and a bit later on we clear it */
6687 if (RExC_seen & REG_RECURSE_SEEN) {
6688 Newxz(RExC_open_parens, RExC_npar,regnode *);
6689 SAVEFREEPV(RExC_open_parens);
6690 Newxz(RExC_close_parens,RExC_npar,regnode *);
6691 SAVEFREEPV(RExC_close_parens);
6693 if (RExC_seen & (REG_RECURSE_SEEN | REG_GOSTART_SEEN)) {
6694 /* Note, RExC_npar is 1 + the number of parens in a pattern.
6695 * So its 1 if there are no parens. */
6696 RExC_study_chunk_recursed_bytes= (RExC_npar >> 3) +
6697 ((RExC_npar & 0x07) != 0);
6698 Newx(RExC_study_chunk_recursed,
6699 RExC_study_chunk_recursed_bytes * RExC_npar, U8);
6700 SAVEFREEPV(RExC_study_chunk_recursed);
6703 /* Useful during FAIL. */
6704 #ifdef RE_TRACK_PATTERN_OFFSETS
6705 Newxz(ri->u.offsets, 2*RExC_size+1, U32); /* MJD 20001228 */
6706 DEBUG_OFFSETS_r(PerlIO_printf(Perl_debug_log,
6707 "%s %"UVuf" bytes for offset annotations.\n",
6708 ri->u.offsets ? "Got" : "Couldn't get",
6709 (UV)((2*RExC_size+1) * sizeof(U32))));
6711 SetProgLen(ri,RExC_size);
6715 REH_CALL_COMP_BEGIN_HOOK(pRExC_state->rx);
6717 /* Second pass: emit code. */
6718 RExC_flags = rx_flags; /* don't let top level (?i) bleed */
6719 RExC_pm_flags = pm_flags;
6721 RExC_end = exp + plen;
6724 RExC_emit_start = ri->program;
6725 RExC_emit = ri->program;
6726 RExC_emit_bound = ri->program + RExC_size + 1;
6727 pRExC_state->code_index = 0;
6729 REGC((U8)REG_MAGIC, (char*) RExC_emit++);
6730 if (reg(pRExC_state, 0, &flags,1) == NULL) {
6732 Perl_croak(aTHX_ "panic: reg returned NULL to re_op_compile for generation pass, flags=%#"UVxf"", (UV) flags);
6734 /* XXXX To minimize changes to RE engine we always allocate
6735 3-units-long substrs field. */
6736 Newx(r->substrs, 1, struct reg_substr_data);
6737 if (RExC_recurse_count) {
6738 Newxz(RExC_recurse,RExC_recurse_count,regnode *);
6739 SAVEFREEPV(RExC_recurse);
6743 r->minlen = minlen = sawlookahead = sawplus = sawopen = sawminmod = 0;
6744 Zero(r->substrs, 1, struct reg_substr_data);
6745 if (RExC_study_chunk_recursed)
6746 Zero(RExC_study_chunk_recursed,
6747 RExC_study_chunk_recursed_bytes * RExC_npar, U8);
6749 #ifdef TRIE_STUDY_OPT
6751 StructCopy(&zero_scan_data, &data, scan_data_t);
6752 copyRExC_state = RExC_state;
6755 DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log,"Restudying\n"));
6757 RExC_state = copyRExC_state;
6758 if (seen & REG_TOP_LEVEL_BRANCHES_SEEN)
6759 RExC_seen |= REG_TOP_LEVEL_BRANCHES_SEEN;
6761 RExC_seen &= ~REG_TOP_LEVEL_BRANCHES_SEEN;
6762 StructCopy(&zero_scan_data, &data, scan_data_t);
6765 StructCopy(&zero_scan_data, &data, scan_data_t);
6768 /* Dig out information for optimizations. */
6769 r->extflags = RExC_flags; /* was pm_op */
6770 /*dmq: removed as part of de-PMOP: pm->op_pmflags = RExC_flags; */
6773 SvUTF8_on(rx); /* Unicode in it? */
6774 ri->regstclass = NULL;
6775 if (RExC_naughty >= 10) /* Probably an expensive pattern. */
6776 r->intflags |= PREGf_NAUGHTY;
6777 scan = ri->program + 1; /* First BRANCH. */
6779 /* testing for BRANCH here tells us whether there is "must appear"
6780 data in the pattern. If there is then we can use it for optimisations */
6781 if (!(RExC_seen & REG_TOP_LEVEL_BRANCHES_SEEN)) { /* Only one top-level choice.
6784 STRLEN longest_float_length, longest_fixed_length;
6785 regnode_ssc ch_class; /* pointed to by data */
6787 SSize_t last_close = 0; /* pointed to by data */
6788 regnode *first= scan;
6789 regnode *first_next= regnext(first);
6791 * Skip introductions and multiplicators >= 1
6792 * so that we can extract the 'meat' of the pattern that must
6793 * match in the large if() sequence following.
6794 * NOTE that EXACT is NOT covered here, as it is normally
6795 * picked up by the optimiser separately.
6797 * This is unfortunate as the optimiser isnt handling lookahead
6798 * properly currently.
6801 while ((OP(first) == OPEN && (sawopen = 1)) ||
6802 /* An OR of *one* alternative - should not happen now. */
6803 (OP(first) == BRANCH && OP(first_next) != BRANCH) ||
6804 /* for now we can't handle lookbehind IFMATCH*/
6805 (OP(first) == IFMATCH && !first->flags && (sawlookahead = 1)) ||
6806 (OP(first) == PLUS) ||
6807 (OP(first) == MINMOD) ||
6808 /* An {n,m} with n>0 */
6809 (PL_regkind[OP(first)] == CURLY && ARG1(first) > 0) ||
6810 (OP(first) == NOTHING && PL_regkind[OP(first_next)] != END ))
6813 * the only op that could be a regnode is PLUS, all the rest
6814 * will be regnode_1 or regnode_2.
6816 * (yves doesn't think this is true)
6818 if (OP(first) == PLUS)
6821 if (OP(first) == MINMOD)
6823 first += regarglen[OP(first)];
6825 first = NEXTOPER(first);
6826 first_next= regnext(first);
6829 /* Starting-point info. */
6831 DEBUG_PEEP("first:",first,0);
6832 /* Ignore EXACT as we deal with it later. */
6833 if (PL_regkind[OP(first)] == EXACT) {
6834 if (OP(first) == EXACT)
6835 NOOP; /* Empty, get anchored substr later. */
6837 ri->regstclass = first;
6840 else if (PL_regkind[OP(first)] == TRIE &&
6841 ((reg_trie_data *)ri->data->data[ ARG(first) ])->minlen>0)
6843 /* this can happen only on restudy */
6844 ri->regstclass = construct_ahocorasick_from_trie(pRExC_state, (regnode *)first, 0);
6847 else if (REGNODE_SIMPLE(OP(first)))
6848 ri->regstclass = first;
6849 else if (PL_regkind[OP(first)] == BOUND ||
6850 PL_regkind[OP(first)] == NBOUND)
6851 ri->regstclass = first;
6852 else if (PL_regkind[OP(first)] == BOL) {
6853 r->intflags |= (OP(first) == MBOL
6855 : (OP(first) == SBOL
6858 first = NEXTOPER(first);
6861 else if (OP(first) == GPOS) {
6862 r->intflags |= PREGf_ANCH_GPOS;
6863 first = NEXTOPER(first);
6866 else if ((!sawopen || !RExC_sawback) &&
6868 (OP(first) == STAR &&
6869 PL_regkind[OP(NEXTOPER(first))] == REG_ANY) &&
6870 !(r->intflags & PREGf_ANCH) && !pRExC_state->num_code_blocks)
6872 /* turn .* into ^.* with an implied $*=1 */
6874 (OP(NEXTOPER(first)) == REG_ANY)
6877 r->intflags |= (type | PREGf_IMPLICIT);
6878 first = NEXTOPER(first);
6881 if (sawplus && !sawminmod && !sawlookahead
6882 && (!sawopen || !RExC_sawback)
6883 && !pRExC_state->num_code_blocks) /* May examine pos and $& */
6884 /* x+ must match at the 1st pos of run of x's */
6885 r->intflags |= PREGf_SKIP;
6887 /* Scan is after the zeroth branch, first is atomic matcher. */
6888 #ifdef TRIE_STUDY_OPT
6891 PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
6892 (IV)(first - scan + 1))
6896 PerlIO_printf(Perl_debug_log, "first at %"IVdf"\n",
6897 (IV)(first - scan + 1))
6903 * If there's something expensive in the r.e., find the
6904 * longest literal string that must appear and make it the
6905 * regmust. Resolve ties in favor of later strings, since
6906 * the regstart check works with the beginning of the r.e.
6907 * and avoiding duplication strengthens checking. Not a
6908 * strong reason, but sufficient in the absence of others.
6909 * [Now we resolve ties in favor of the earlier string if
6910 * it happens that c_offset_min has been invalidated, since the
6911 * earlier string may buy us something the later one won't.]
6914 data.longest_fixed = newSVpvs("");
6915 data.longest_float = newSVpvs("");
6916 data.last_found = newSVpvs("");
6917 data.longest = &(data.longest_fixed);
6918 ENTER_with_name("study_chunk");
6919 SAVEFREESV(data.longest_fixed);
6920 SAVEFREESV(data.longest_float);
6921 SAVEFREESV(data.last_found);
6923 if (!ri->regstclass) {
6924 ssc_init(pRExC_state, &ch_class);
6925 data.start_class = &ch_class;
6926 stclass_flag = SCF_DO_STCLASS_AND;
6927 } else /* XXXX Check for BOUND? */
6929 data.last_closep = &last_close;
6932 minlen = study_chunk(pRExC_state, &first, &minlen, &fake,
6933 scan + RExC_size, /* Up to end */
6935 SCF_DO_SUBSTR | SCF_WHILEM_VISITED_POS | stclass_flag
6936 | (restudied ? SCF_TRIE_DOING_RESTUDY : 0),
6940 CHECK_RESTUDY_GOTO_butfirst(LEAVE_with_name("study_chunk"));
6943 if ( RExC_npar == 1 && data.longest == &(data.longest_fixed)
6944 && data.last_start_min == 0 && data.last_end > 0
6945 && !RExC_seen_zerolen
6946 && !(RExC_seen & REG_VERBARG_SEEN)
6947 && !(RExC_seen & REG_GPOS_SEEN)
6949 r->extflags |= RXf_CHECK_ALL;
6951 scan_commit(pRExC_state, &data,&minlen,0);
6953 longest_float_length = CHR_SVLEN(data.longest_float);
6955 if (! ((SvCUR(data.longest_fixed) /* ok to leave SvCUR */
6956 && data.offset_fixed == data.offset_float_min
6957 && SvCUR(data.longest_fixed) == SvCUR(data.longest_float)))
6958 && S_setup_longest (aTHX_ pRExC_state,
6962 &(r->float_end_shift),
6963 data.lookbehind_float,
6964 data.offset_float_min,
6966 longest_float_length,
6967 cBOOL(data.flags & SF_FL_BEFORE_EOL),
6968 cBOOL(data.flags & SF_FL_BEFORE_MEOL)))
6970 r->float_min_offset = data.offset_float_min - data.lookbehind_float;
6971 r->float_max_offset = data.offset_float_max;
6972 if (data.offset_float_max < SSize_t_MAX) /* Don't offset infinity */
6973 r->float_max_offset -= data.lookbehind_float;
6974 SvREFCNT_inc_simple_void_NN(data.longest_float);
6977 r->float_substr = r->float_utf8 = NULL;
6978 longest_float_length = 0;
6981 longest_fixed_length = CHR_SVLEN(data.longest_fixed);
6983 if (S_setup_longest (aTHX_ pRExC_state,
6985 &(r->anchored_utf8),
6986 &(r->anchored_substr),
6987 &(r->anchored_end_shift),
6988 data.lookbehind_fixed,
6991 longest_fixed_length,
6992 cBOOL(data.flags & SF_FIX_BEFORE_EOL),
6993 cBOOL(data.flags & SF_FIX_BEFORE_MEOL)))
6995 r->anchored_offset = data.offset_fixed - data.lookbehind_fixed;
6996 SvREFCNT_inc_simple_void_NN(data.longest_fixed);
6999 r->anchored_substr = r->anchored_utf8 = NULL;
7000 longest_fixed_length = 0;
7002 LEAVE_with_name("study_chunk");
7005 && (OP(ri->regstclass) == REG_ANY || OP(ri->regstclass) == SANY))
7006 ri->regstclass = NULL;
7008 if ((!(r->anchored_substr || r->anchored_utf8) || r->anchored_offset)
7010 && ! (ANYOF_FLAGS(data.start_class) & ANYOF_EMPTY_STRING)
7011 && !ssc_is_anything(data.start_class))
7013 const U32 n = add_data(pRExC_state, STR_WITH_LEN("f"));
7015 ssc_finalize(pRExC_state, data.start_class);
7017 Newx(RExC_rxi->data->data[n], 1, regnode_ssc);
7018 StructCopy(data.start_class,
7019 (regnode_ssc*)RExC_rxi->data->data[n],
7021 ri->regstclass = (regnode*)RExC_rxi->data->data[n];
7022 r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
7023 DEBUG_COMPILE_r({ SV *sv = sv_newmortal();
7024 regprop(r, sv, (regnode*)data.start_class, NULL);
7025 PerlIO_printf(Perl_debug_log,
7026 "synthetic stclass \"%s\".\n",
7027 SvPVX_const(sv));});
7028 data.start_class = NULL;
7031 /* A temporary algorithm prefers floated substr to fixed one to dig
7033 if (longest_fixed_length > longest_float_length) {
7034 r->substrs->check_ix = 0;
7035 r->check_end_shift = r->anchored_end_shift;
7036 r->check_substr = r->anchored_substr;
7037 r->check_utf8 = r->anchored_utf8;
7038 r->check_offset_min = r->check_offset_max = r->anchored_offset;
7039 if (r->intflags & (PREGf_ANCH_SBOL|PREGf_ANCH_GPOS))
7040 r->intflags |= PREGf_NOSCAN;
7043 r->substrs->check_ix = 1;
7044 r->check_end_shift = r->float_end_shift;
7045 r->check_substr = r->float_substr;
7046 r->check_utf8 = r->float_utf8;
7047 r->check_offset_min = r->float_min_offset;
7048 r->check_offset_max = r->float_max_offset;
7050 if ((r->check_substr || r->check_utf8) ) {
7051 r->extflags |= RXf_USE_INTUIT;
7052 if (SvTAIL(r->check_substr ? r->check_substr : r->check_utf8))
7053 r->extflags |= RXf_INTUIT_TAIL;
7055 r->substrs->data[0].max_offset = r->substrs->data[0].min_offset;
7057 /* XXX Unneeded? dmq (shouldn't as this is handled elsewhere)
7058 if ( (STRLEN)minlen < longest_float_length )
7059 minlen= longest_float_length;
7060 if ( (STRLEN)minlen < longest_fixed_length )
7061 minlen= longest_fixed_length;
7065 /* Several toplevels. Best we can is to set minlen. */
7067 regnode_ssc ch_class;
7068 SSize_t last_close = 0;
7070 DEBUG_PARSE_r(PerlIO_printf(Perl_debug_log, "\nMulti Top Level\n"));
7072 scan = ri->program + 1;
7073 ssc_init(pRExC_state, &ch_class);
7074 data.start_class = &ch_class;
7075 data.last_closep = &last_close;
7078 minlen = study_chunk(pRExC_state,
7079 &scan, &minlen, &fake, scan + RExC_size, &data, -1, 0, NULL,
7080 SCF_DO_STCLASS_AND|SCF_WHILEM_VISITED_POS|(restudied
7081 ? SCF_TRIE_DOING_RESTUDY
7085 CHECK_RESTUDY_GOTO_butfirst(NOOP);
7087 r->check_substr = r->check_utf8 = r->anchored_substr = r->anchored_utf8
7088 = r->float_substr = r->float_utf8 = NULL;
7090 if (! (ANYOF_FLAGS(data.start_class) & ANYOF_EMPTY_STRING)
7091 && ! ssc_is_anything(data.start_class))
7093 const U32 n = add_data(pRExC_state, STR_WITH_LEN("f"));
7095 ssc_finalize(pRExC_state, data.start_class);
7097 Newx(RExC_rxi->data->data[n], 1, regnode_ssc);
7098 StructCopy(data.start_class,
7099 (regnode_ssc*)RExC_rxi->data->data[n],
7101 ri->regstclass = (regnode*)RExC_rxi->data->data[n];
7102 r->intflags &= ~PREGf_SKIP; /* Used in find_byclass(). */
7103 DEBUG_COMPILE_r({ SV* sv = sv_newmortal();
7104 regprop(r, sv, (regnode*)data.start_class, NULL);
7105 PerlIO_printf(Perl_debug_log,
7106 "synthetic stclass \"%s\".\n",
7107 SvPVX_const(sv));});
7108 data.start_class = NULL;
7112 if (RExC_seen & REG_UNBOUNDED_QUANTIFIER_SEEN) {
7113 r->extflags |= RXf_UNBOUNDED_QUANTIFIER_SEEN;
7114 r->maxlen = REG_INFTY;
7117 r->maxlen = RExC_maxlen;
7120 /* Guard against an embedded (?=) or (?<=) with a longer minlen than
7121 the "real" pattern. */
7123 PerlIO_printf(Perl_debug_log,"minlen: %"IVdf" r->minlen:%"IVdf" maxlen:%ld\n",
7124 (IV)minlen, (IV)r->minlen, RExC_maxlen);
7126 r->minlenret = minlen;
7127 if (r->minlen < minlen)
7130 if (RExC_seen & REG_GPOS_SEEN)
7131 r->intflags |= PREGf_GPOS_SEEN;
7132 if (RExC_seen & REG_LOOKBEHIND_SEEN)
7133 r->extflags |= RXf_NO_INPLACE_SUBST; /* inplace might break the
7135 if (pRExC_state->num_code_blocks)
7136 r->extflags |= RXf_EVAL_SEEN;
7137 if (RExC_seen & REG_CANY_SEEN)
7138 r->intflags |= PREGf_CANY_SEEN;
7139 if (RExC_seen & REG_VERBARG_SEEN)
7141 r->intflags |= PREGf_VERBARG_SEEN;
7142 r->extflags |= RXf_NO_INPLACE_SUBST; /* don't understand this! Yves */
7144 if (RExC_seen & REG_CUTGROUP_SEEN)
7145 r->intflags |= PREGf_CUTGROUP_SEEN;
7146 if (pm_flags & PMf_USE_RE_EVAL)
7147 r->intflags |= PREGf_USE_RE_EVAL;
7148 if (RExC_paren_names)
7149 RXp_PAREN_NAMES(r) = MUTABLE_HV(SvREFCNT_inc(RExC_paren_names));
7151 RXp_PAREN_NAMES(r) = NULL;
7153 /* If we have seen an anchor in our pattern then we set the extflag RXf_IS_ANCHORED
7154 * so it can be used in pp.c */
7155 if (r->intflags & PREGf_ANCH)
7156 r->extflags |= RXf_IS_ANCHORED;
7160 /* this is used to identify "special" patterns that might result
7161 * in Perl NOT calling the regex engine and instead doing the match "itself",
7162 * particularly special cases in split//. By having the regex compiler
7163 * do this pattern matching at a regop level (instead of by inspecting the pattern)
7164 * we avoid weird issues with equivalent patterns resulting in different behavior,
7165 * AND we allow non Perl engines to get the same optimizations by the setting the
7166 * flags appropriately - Yves */
7167 regnode *first = ri->program + 1;
7169 regnode *next = NEXTOPER(first);
7172 if (PL_regkind[fop] == NOTHING && nop == END)
7173 r->extflags |= RXf_NULL;
7174 else if (PL_regkind[fop] == BOL && nop == END)
7175 r->extflags |= RXf_START_ONLY;
7176 else if (fop == PLUS
7177 && PL_regkind[nop] == POSIXD && FLAGS(next) == _CC_SPACE
7178 && OP(regnext(first)) == END)
7179 r->extflags |= RXf_WHITE;
7180 else if ( r->extflags & RXf_SPLIT
7182 && STR_LEN(first) == 1
7183 && *(STRING(first)) == ' '
7184 && OP(regnext(first)) == END )
7185 r->extflags |= (RXf_SKIPWHITE|RXf_WHITE);
7189 if (RExC_contains_locale) {
7190 RXp_EXTFLAGS(r) |= RXf_TAINTED;
7194 if (RExC_paren_names) {
7195 ri->name_list_idx = add_data( pRExC_state, STR_WITH_LEN("a"));
7196 ri->data->data[ri->name_list_idx]
7197 = (void*)SvREFCNT_inc(RExC_paren_name_list);
7200 ri->name_list_idx = 0;
7202 if (RExC_recurse_count) {
7203 for ( ; RExC_recurse_count ; RExC_recurse_count-- ) {
7204 const regnode *scan = RExC_recurse[RExC_recurse_count-1];
7205 ARG2L_SET( scan, RExC_open_parens[ARG(scan)-1] - scan );
7208 Newxz(r->offs, RExC_npar, regexp_paren_pair);
7209 /* assume we don't need to swap parens around before we match */
7213 PerlIO_printf(Perl_debug_log,"Final program:\n");
7216 #ifdef RE_TRACK_PATTERN_OFFSETS
7217 DEBUG_OFFSETS_r(if (ri->u.offsets) {
7218 const STRLEN len = ri->u.offsets[0];
7220 GET_RE_DEBUG_FLAGS_DECL;
7221 PerlIO_printf(Perl_debug_log,
7222 "Offsets: [%"UVuf"]\n\t", (UV)ri->u.offsets[0]);
7223 for (i = 1; i <= len; i++) {
7224 if (ri->u.offsets[i*2-1] || ri->u.offsets[i*2])
7225 PerlIO_printf(Perl_debug_log, "%"UVuf":%"UVuf"[%"UVuf"] ",
7226 (UV)i, (UV)ri->u.offsets[i*2-1], (UV)ri->u.offsets[i*2]);
7228 PerlIO_printf(Perl_debug_log, "\n");
7233 /* under ithreads the ?pat? PMf_USED flag on the pmop is simulated
7234 * by setting the regexp SV to readonly-only instead. If the
7235 * pattern's been recompiled, the USEDness should remain. */
7236 if (old_re && SvREADONLY(old_re))
7244 Perl_reg_named_buff(pTHX_ REGEXP * const rx, SV * const key, SV * const value,
7247 PERL_ARGS_ASSERT_REG_NAMED_BUFF;
7249 PERL_UNUSED_ARG(value);
7251 if (flags & RXapif_FETCH) {
7252 return reg_named_buff_fetch(rx, key, flags);
7253 } else if (flags & (RXapif_STORE | RXapif_DELETE | RXapif_CLEAR)) {
7254 Perl_croak_no_modify();
7256 } else if (flags & RXapif_EXISTS) {
7257 return reg_named_buff_exists(rx, key, flags)
7260 } else if (flags & RXapif_REGNAMES) {
7261 return reg_named_buff_all(rx, flags);
7262 } else if (flags & (RXapif_SCALAR | RXapif_REGNAMES_COUNT)) {
7263 return reg_named_buff_scalar(rx, flags);
7265 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff", (int)flags);
7271 Perl_reg_named_buff_iter(pTHX_ REGEXP * const rx, const SV * const lastkey,
7274 PERL_ARGS_ASSERT_REG_NAMED_BUFF_ITER;
7275 PERL_UNUSED_ARG(lastkey);
7277 if (flags & RXapif_FIRSTKEY)
7278 return reg_named_buff_firstkey(rx, flags);
7279 else if (flags & RXapif_NEXTKEY)
7280 return reg_named_buff_nextkey(rx, flags);
7282 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_iter",
7289 Perl_reg_named_buff_fetch(pTHX_ REGEXP * const r, SV * const namesv,
7292 AV *retarray = NULL;
7294 struct regexp *const rx = ReANY(r);
7296 PERL_ARGS_ASSERT_REG_NAMED_BUFF_FETCH;
7298 if (flags & RXapif_ALL)
7301 if (rx && RXp_PAREN_NAMES(rx)) {
7302 HE *he_str = hv_fetch_ent( RXp_PAREN_NAMES(rx), namesv, 0, 0 );
7305 SV* sv_dat=HeVAL(he_str);
7306 I32 *nums=(I32*)SvPVX(sv_dat);
7307 for ( i=0; i<SvIVX(sv_dat); i++ ) {
7308 if ((I32)(rx->nparens) >= nums[i]
7309 && rx->offs[nums[i]].start != -1
7310 && rx->offs[nums[i]].end != -1)
7313 CALLREG_NUMBUF_FETCH(r,nums[i],ret);
7318 ret = newSVsv(&PL_sv_undef);
7321 av_push(retarray, ret);
7324 return newRV_noinc(MUTABLE_SV(retarray));
7331 Perl_reg_named_buff_exists(pTHX_ REGEXP * const r, SV * const key,
7334 struct regexp *const rx = ReANY(r);
7336 PERL_ARGS_ASSERT_REG_NAMED_BUFF_EXISTS;
7338 if (rx && RXp_PAREN_NAMES(rx)) {
7339 if (flags & RXapif_ALL) {
7340 return hv_exists_ent(RXp_PAREN_NAMES(rx), key, 0);
7342 SV *sv = CALLREG_NAMED_BUFF_FETCH(r, key, flags);
7344 SvREFCNT_dec_NN(sv);
7356 Perl_reg_named_buff_firstkey(pTHX_ REGEXP * const r, const U32 flags)
7358 struct regexp *const rx = ReANY(r);
7360 PERL_ARGS_ASSERT_REG_NAMED_BUFF_FIRSTKEY;
7362 if ( rx && RXp_PAREN_NAMES(rx) ) {
7363 (void)hv_iterinit(RXp_PAREN_NAMES(rx));
7365 return CALLREG_NAMED_BUFF_NEXTKEY(r, NULL, flags & ~RXapif_FIRSTKEY);
7372 Perl_reg_named_buff_nextkey(pTHX_ REGEXP * const r, const U32 flags)
7374 struct regexp *const rx = ReANY(r);
7375 GET_RE_DEBUG_FLAGS_DECL;
7377 PERL_ARGS_ASSERT_REG_NAMED_BUFF_NEXTKEY;
7379 if (rx && RXp_PAREN_NAMES(rx)) {
7380 HV *hv = RXp_PAREN_NAMES(rx);
7382 while ( (temphe = hv_iternext_flags(hv,0)) ) {
7385 SV* sv_dat = HeVAL(temphe);
7386 I32 *nums = (I32*)SvPVX(sv_dat);
7387 for ( i = 0; i < SvIVX(sv_dat); i++ ) {
7388 if ((I32)(rx->lastparen) >= nums[i] &&
7389 rx->offs[nums[i]].start != -1 &&
7390 rx->offs[nums[i]].end != -1)
7396 if (parno || flags & RXapif_ALL) {
7397 return newSVhek(HeKEY_hek(temphe));
7405 Perl_reg_named_buff_scalar(pTHX_ REGEXP * const r, const U32 flags)
7410 struct regexp *const rx = ReANY(r);
7412 PERL_ARGS_ASSERT_REG_NAMED_BUFF_SCALAR;
7414 if (rx && RXp_PAREN_NAMES(rx)) {
7415 if (flags & (RXapif_ALL | RXapif_REGNAMES_COUNT)) {
7416 return newSViv(HvTOTALKEYS(RXp_PAREN_NAMES(rx)));
7417 } else if (flags & RXapif_ONE) {
7418 ret = CALLREG_NAMED_BUFF_ALL(r, (flags | RXapif_REGNAMES));
7419 av = MUTABLE_AV(SvRV(ret));
7420 length = av_tindex(av);
7421 SvREFCNT_dec_NN(ret);
7422 return newSViv(length + 1);
7424 Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_scalar",
7429 return &PL_sv_undef;
7433 Perl_reg_named_buff_all(pTHX_ REGEXP * const r, const U32 flags)
7435 struct regexp *const rx = ReANY(r);
7438 PERL_ARGS_ASSERT_REG_NAMED_BUFF_ALL;
7440 if (rx && RXp_PAREN_NAMES(rx)) {
7441 HV *hv= RXp_PAREN_NAMES(rx);
7443 (void)hv_iterinit(hv);
7444 while ( (temphe = hv_iternext_flags(hv,0)) ) {
7447 SV* sv_dat = HeVAL(temphe);
7448 I32 *nums = (I32*)SvPVX(sv_dat);
7449 for ( i = 0; i < SvIVX(sv_dat); i++ ) {
7450 if ((I32)(rx->lastparen) >= nums[i] &&
7451 rx->offs[nums[i]].start != -1 &&
7452 rx->offs[nums[i]].end != -1)
7458 if (parno || flags & RXapif_ALL) {
7459 av_push(av, newSVhek(HeKEY_hek(temphe)));
7464 return newRV_noinc(MUTABLE_SV(av));
7468 Perl_reg_numbered_buff_fetch(pTHX_ REGEXP * const r, const I32 paren,
7471 struct regexp *const rx = ReANY(r);
7477 PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_FETCH;
7479 if ( n == RX_BUFF_IDX_CARET_PREMATCH
7480 || n == RX_BUFF_IDX_CARET_FULLMATCH
7481 || n == RX_BUFF_IDX_CARET_POSTMATCH
7484 bool keepcopy = cBOOL(rx->extflags & RXf_PMf_KEEPCOPY);
7486 /* on something like
7489 * the KEEPCOPY is set on the PMOP rather than the regex */
7490 if (PL_curpm && r == PM_GETRE(PL_curpm))
7491 keepcopy = cBOOL(PL_curpm->op_pmflags & PMf_KEEPCOPY);
7500 if (n == RX_BUFF_IDX_CARET_FULLMATCH)
7501 /* no need to distinguish between them any more */
7502 n = RX_BUFF_IDX_FULLMATCH;
7504 if ((n == RX_BUFF_IDX_PREMATCH || n == RX_BUFF_IDX_CARET_PREMATCH)
7505 && rx->offs[0].start != -1)
7507 /* $`, ${^PREMATCH} */
7508 i = rx->offs[0].start;
7512 if ((n == RX_BUFF_IDX_POSTMATCH || n == RX_BUFF_IDX_CARET_POSTMATCH)
7513 && rx->offs[0].end != -1)
7515 /* $', ${^POSTMATCH} */
7516 s = rx->subbeg - rx->suboffset + rx->offs[0].end;
7517 i = rx->sublen + rx->suboffset - rx->offs[0].end;
7520 if ( 0 <= n && n <= (I32)rx->nparens &&
7521 (s1 = rx->offs[n].start) != -1 &&
7522 (t1 = rx->offs[n].end) != -1)
7524 /* $&, ${^MATCH}, $1 ... */
7526 s = rx->subbeg + s1 - rx->suboffset;
7531 assert(s >= rx->subbeg);
7532 assert((STRLEN)rx->sublen >= (STRLEN)((s - rx->subbeg) + i) );
7534 #ifdef NO_TAINT_SUPPORT
7535 sv_setpvn(sv, s, i);
7537 const int oldtainted = TAINT_get;
7539 sv_setpvn(sv, s, i);
7540 TAINT_set(oldtainted);
7542 if ( (rx->intflags & PREGf_CANY_SEEN)
7543 ? (RXp_MATCH_UTF8(rx)
7544 && (!i || is_utf8_string((U8*)s, i)))
7545 : (RXp_MATCH_UTF8(rx)) )
7552 if (RXp_MATCH_TAINTED(rx)) {
7553 if (SvTYPE(sv) >= SVt_PVMG) {
7554 MAGIC* const mg = SvMAGIC(sv);
7557 SvMAGIC_set(sv, mg->mg_moremagic);
7559 if ((mgt = SvMAGIC(sv))) {
7560 mg->mg_moremagic = mgt;
7561 SvMAGIC_set(sv, mg);
7572 sv_setsv(sv,&PL_sv_undef);
7578 Perl_reg_numbered_buff_store(pTHX_ REGEXP * const rx, const I32 paren,
7579 SV const * const value)
7581 PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_STORE;
7583 PERL_UNUSED_ARG(rx);
7584 PERL_UNUSED_ARG(paren);
7585 PERL_UNUSED_ARG(value);
7588 Perl_croak_no_modify();
7592 Perl_reg_numbered_buff_length(pTHX_ REGEXP * const r, const SV * const sv,
7595 struct regexp *const rx = ReANY(r);
7599 PERL_ARGS_ASSERT_REG_NUMBERED_BUFF_LENGTH;
7601 if ( paren == RX_BUFF_IDX_CARET_PREMATCH
7602 || paren == RX_BUFF_IDX_CARET_FULLMATCH
7603 || paren == RX_BUFF_IDX_CARET_POSTMATCH
7606 bool keepcopy = cBOOL(rx->extflags & RXf_PMf_KEEPCOPY);
7608 /* on something like
7611 * the KEEPCOPY is set on the PMOP rather than the regex */
7612 if (PL_curpm && r == PM_GETRE(PL_curpm))
7613 keepcopy = cBOOL(PL_curpm->op_pmflags & PMf_KEEPCOPY);
7619 /* Some of this code was originally in C<Perl_magic_len> in F<mg.c> */
7621 case RX_BUFF_IDX_CARET_PREMATCH: /* ${^PREMATCH} */
7622 case RX_BUFF_IDX_PREMATCH: /* $` */
7623 if (rx->offs[0].start != -1) {
7624 i = rx->offs[0].start;
7633 case RX_BUFF_IDX_CARET_POSTMATCH: /* ${^POSTMATCH} */
7634 case RX_BUFF_IDX_POSTMATCH: /* $' */
7635 if (rx->offs[0].end != -1) {
7636 i = rx->sublen - rx->offs[0].end;
7638 s1 = rx->offs[0].end;
7645 default: /* $& / ${^MATCH}, $1, $2, ... */
7646 if (paren <= (I32)rx->nparens &&
7647 (s1 = rx->offs[paren].start) != -1 &&
7648 (t1 = rx->offs[paren].end) != -1)
7654 if (ckWARN(WARN_UNINITIALIZED))
7655 report_uninit((const SV *)sv);
7660 if (i > 0 && RXp_MATCH_UTF8(rx)) {
7661 const char * const s = rx->subbeg - rx->suboffset + s1;
7666 if (is_utf8_string_loclen((U8*)s, i, &ep, &el))
7673 Perl_reg_qr_package(pTHX_ REGEXP * const rx)
7675 PERL_ARGS_ASSERT_REG_QR_PACKAGE;
7676 PERL_UNUSED_ARG(rx);
7680 return newSVpvs("Regexp");
7683 /* Scans the name of a named buffer from the pattern.
7684 * If flags is REG_RSN_RETURN_NULL returns null.
7685 * If flags is REG_RSN_RETURN_NAME returns an SV* containing the name
7686 * If flags is REG_RSN_RETURN_DATA returns the data SV* corresponding
7687 * to the parsed name as looked up in the RExC_paren_names hash.
7688 * If there is an error throws a vFAIL().. type exception.
7691 #define REG_RSN_RETURN_NULL 0
7692 #define REG_RSN_RETURN_NAME 1
7693 #define REG_RSN_RETURN_DATA 2
7696 S_reg_scan_name(pTHX_ RExC_state_t *pRExC_state, U32 flags)
7698 char *name_start = RExC_parse;
7700 PERL_ARGS_ASSERT_REG_SCAN_NAME;
7702 assert (RExC_parse <= RExC_end);
7703 if (RExC_parse == RExC_end) NOOP;
7704 else if (isIDFIRST_lazy_if(RExC_parse, UTF)) {
7705 /* skip IDFIRST by using do...while */
7708 RExC_parse += UTF8SKIP(RExC_parse);
7709 } while (isWORDCHAR_utf8((U8*)RExC_parse));
7713 } while (isWORDCHAR(*RExC_parse));
7715 RExC_parse++; /* so the <- from the vFAIL is after the offending
7717 vFAIL("Group name must start with a non-digit word character");
7721 = newSVpvn_flags(name_start, (int)(RExC_parse - name_start),
7722 SVs_TEMP | (UTF ? SVf_UTF8 : 0));
7723 if ( flags == REG_RSN_RETURN_NAME)
7725 else if (flags==REG_RSN_RETURN_DATA) {
7728 if ( ! sv_name ) /* should not happen*/
7729 Perl_croak(aTHX_ "panic: no svname in reg_scan_name");
7730 if (RExC_paren_names)
7731 he_str = hv_fetch_ent( RExC_paren_names, sv_name, 0, 0 );
7733 sv_dat = HeVAL(he_str);
7735 vFAIL("Reference to nonexistent named group");
7739 Perl_croak(aTHX_ "panic: bad flag %lx in reg_scan_name",
7740 (unsigned long) flags);
7742 assert(0); /* NOT REACHED */
7747 #define DEBUG_PARSE_MSG(funcname) DEBUG_PARSE_r({ \
7748 int rem=(int)(RExC_end - RExC_parse); \
7757 if (RExC_lastparse!=RExC_parse) \
7758 PerlIO_printf(Perl_debug_log," >%.*s%-*s", \
7761 iscut ? "..." : "<" \
7764 PerlIO_printf(Perl_debug_log,"%16s",""); \
7767 num = RExC_size + 1; \
7769 num=REG_NODE_NUM(RExC_emit); \
7770 if (RExC_lastnum!=num) \
7771 PerlIO_printf(Perl_debug_log,"|%4d",num); \
7773 PerlIO_printf(Perl_debug_log,"|%4s",""); \
7774 PerlIO_printf(Perl_debug_log,"|%*s%-4s", \
7775 (int)((depth*2)), "", \
7779 RExC_lastparse=RExC_parse; \
7784 #define DEBUG_PARSE(funcname) DEBUG_PARSE_r({ \
7785 DEBUG_PARSE_MSG((funcname)); \
7786 PerlIO_printf(Perl_debug_log,"%4s","\n"); \
7788 #define DEBUG_PARSE_FMT(funcname,fmt,args) DEBUG_PARSE_r({ \
7789 DEBUG_PARSE_MSG((funcname)); \
7790 PerlIO_printf(Perl_debug_log,fmt "\n",args); \
7793 /* This section of code defines the inversion list object and its methods. The
7794 * interfaces are highly subject to change, so as much as possible is static to
7795 * this file. An inversion list is here implemented as a malloc'd C UV array
7796 * as an SVt_INVLIST scalar.
7798 * An inversion list for Unicode is an array of code points, sorted by ordinal
7799 * number. The zeroth element is the first code point in the list. The 1th
7800 * element is the first element beyond that not in the list. In other words,
7801 * the first range is
7802 * invlist[0]..(invlist[1]-1)
7803 * The other ranges follow. Thus every element whose index is divisible by two
7804 * marks the beginning of a range that is in the list, and every element not
7805 * divisible by two marks the beginning of a range not in the list. A single
7806 * element inversion list that contains the single code point N generally
7807 * consists of two elements
7810 * (The exception is when N is the highest representable value on the
7811 * machine, in which case the list containing just it would be a single
7812 * element, itself. By extension, if the last range in the list extends to
7813 * infinity, then the first element of that range will be in the inversion list
7814 * at a position that is divisible by two, and is the final element in the
7816 * Taking the complement (inverting) an inversion list is quite simple, if the
7817 * first element is 0, remove it; otherwise add a 0 element at the beginning.
7818 * This implementation reserves an element at the beginning of each inversion
7819 * list to always contain 0; there is an additional flag in the header which
7820 * indicates if the list begins at the 0, or is offset to begin at the next
7823 * More about inversion lists can be found in "Unicode Demystified"
7824 * Chapter 13 by Richard Gillam, published by Addison-Wesley.
7825 * More will be coming when functionality is added later.
7827 * The inversion list data structure is currently implemented as an SV pointing
7828 * to an array of UVs that the SV thinks are bytes. This allows us to have an
7829 * array of UV whose memory management is automatically handled by the existing
7830 * facilities for SV's.
7832 * Some of the methods should always be private to the implementation, and some
7833 * should eventually be made public */
7835 /* The header definitions are in F<inline_invlist.c> */
7837 PERL_STATIC_INLINE UV*
7838 S__invlist_array_init(SV* const invlist, const bool will_have_0)
7840 /* Returns a pointer to the first element in the inversion list's array.
7841 * This is called upon initialization of an inversion list. Where the
7842 * array begins depends on whether the list has the code point U+0000 in it
7843 * or not. The other parameter tells it whether the code that follows this
7844 * call is about to put a 0 in the inversion list or not. The first
7845 * element is either the element reserved for 0, if TRUE, or the element
7846 * after it, if FALSE */
7848 bool* offset = get_invlist_offset_addr(invlist);
7849 UV* zero_addr = (UV *) SvPVX(invlist);
7851 PERL_ARGS_ASSERT__INVLIST_ARRAY_INIT;
7854 assert(! _invlist_len(invlist));
7858 /* 1^1 = 0; 1^0 = 1 */
7859 *offset = 1 ^ will_have_0;
7860 return zero_addr + *offset;
7863 PERL_STATIC_INLINE UV*
7864 S_invlist_array(SV* const invlist)
7866 /* Returns the pointer to the inversion list's array. Every time the
7867 * length changes, this needs to be called in case malloc or realloc moved
7870 PERL_ARGS_ASSERT_INVLIST_ARRAY;
7872 /* Must not be empty. If these fail, you probably didn't check for <len>
7873 * being non-zero before trying to get the array */
7874 assert(_invlist_len(invlist));
7876 /* The very first element always contains zero, The array begins either
7877 * there, or if the inversion list is offset, at the element after it.
7878 * The offset header field determines which; it contains 0 or 1 to indicate
7879 * how much additionally to add */
7880 assert(0 == *(SvPVX(invlist)));
7881 return ((UV *) SvPVX(invlist) + *get_invlist_offset_addr(invlist));
7884 PERL_STATIC_INLINE void
7885 S_invlist_set_len(pTHX_ SV* const invlist, const UV len, const bool offset)
7887 /* Sets the current number of elements stored in the inversion list.
7888 * Updates SvCUR correspondingly */
7889 PERL_UNUSED_CONTEXT;
7890 PERL_ARGS_ASSERT_INVLIST_SET_LEN;
7892 assert(SvTYPE(invlist) == SVt_INVLIST);
7897 : TO_INTERNAL_SIZE(len + offset));
7898 assert(SvLEN(invlist) == 0 || SvCUR(invlist) <= SvLEN(invlist));
7901 PERL_STATIC_INLINE IV*
7902 S_get_invlist_previous_index_addr(SV* invlist)
7904 /* Return the address of the IV that is reserved to hold the cached index
7906 PERL_ARGS_ASSERT_GET_INVLIST_PREVIOUS_INDEX_ADDR;
7908 assert(SvTYPE(invlist) == SVt_INVLIST);
7910 return &(((XINVLIST*) SvANY(invlist))->prev_index);
7913 PERL_STATIC_INLINE IV
7914 S_invlist_previous_index(SV* const invlist)
7916 /* Returns cached index of previous search */
7918 PERL_ARGS_ASSERT_INVLIST_PREVIOUS_INDEX;
7920 return *get_invlist_previous_index_addr(invlist);
7923 PERL_STATIC_INLINE void
7924 S_invlist_set_previous_index(SV* const invlist, const IV index)
7926 /* Caches <index> for later retrieval */
7928 PERL_ARGS_ASSERT_INVLIST_SET_PREVIOUS_INDEX;
7930 assert(index == 0 || index < (int) _invlist_len(invlist));
7932 *get_invlist_previous_index_addr(invlist) = index;
7935 PERL_STATIC_INLINE UV
7936 S_invlist_max(SV* const invlist)
7938 /* Returns the maximum number of elements storable in the inversion list's
7939 * array, without having to realloc() */
7941 PERL_ARGS_ASSERT_INVLIST_MAX;
7943 assert(SvTYPE(invlist) == SVt_INVLIST);
7945 /* Assumes worst case, in which the 0 element is not counted in the
7946 * inversion list, so subtracts 1 for that */
7947 return SvLEN(invlist) == 0 /* This happens under _new_invlist_C_array */
7948 ? FROM_INTERNAL_SIZE(SvCUR(invlist)) - 1
7949 : FROM_INTERNAL_SIZE(SvLEN(invlist)) - 1;
7952 #ifndef PERL_IN_XSUB_RE
7954 Perl__new_invlist(pTHX_ IV initial_size)
7957 /* Return a pointer to a newly constructed inversion list, with enough
7958 * space to store 'initial_size' elements. If that number is negative, a
7959 * system default is used instead */
7963 if (initial_size < 0) {
7967 /* Allocate the initial space */
7968 new_list = newSV_type(SVt_INVLIST);
7970 /* First 1 is in case the zero element isn't in the list; second 1 is for
7972 SvGROW(new_list, TO_INTERNAL_SIZE(initial_size + 1) + 1);
7973 invlist_set_len(new_list, 0, 0);
7975 /* Force iterinit() to be used to get iteration to work */
7976 *get_invlist_iter_addr(new_list) = (STRLEN) UV_MAX;
7978 *get_invlist_previous_index_addr(new_list) = 0;
7984 Perl__new_invlist_C_array(pTHX_ const UV* const list)
7986 /* Return a pointer to a newly constructed inversion list, initialized to
7987 * point to <list>, which has to be in the exact correct inversion list
7988 * form, including internal fields. Thus this is a dangerous routine that
7989 * should not be used in the wrong hands. The passed in 'list' contains
7990 * several header fields at the beginning that are not part of the
7991 * inversion list body proper */
7993 const STRLEN length = (STRLEN) list[0];
7994 const UV version_id = list[1];
7995 const bool offset = cBOOL(list[2]);
7996 #define HEADER_LENGTH 3
7997 /* If any of the above changes in any way, you must change HEADER_LENGTH
7998 * (if appropriate) and regenerate INVLIST_VERSION_ID by running
7999 * perl -E 'say int(rand 2**31-1)'
8001 #define INVLIST_VERSION_ID 148565664 /* This is a combination of a version and
8002 data structure type, so that one being
8003 passed in can be validated to be an
8004 inversion list of the correct vintage.
8007 SV* invlist = newSV_type(SVt_INVLIST);
8009 PERL_ARGS_ASSERT__NEW_INVLIST_C_ARRAY;
8011 if (version_id != INVLIST_VERSION_ID) {
8012 Perl_croak(aTHX_ "panic: Incorrect version for previously generated inversion list");
8015 /* The generated array passed in includes header elements that aren't part
8016 * of the list proper, so start it just after them */
8017 SvPV_set(invlist, (char *) (list + HEADER_LENGTH));
8019 SvLEN_set(invlist, 0); /* Means we own the contents, and the system
8020 shouldn't touch it */
8022 *(get_invlist_offset_addr(invlist)) = offset;
8024 /* The 'length' passed to us is the physical number of elements in the
8025 * inversion list. But if there is an offset the logical number is one
8027 invlist_set_len(invlist, length - offset, offset);
8029 invlist_set_previous_index(invlist, 0);
8031 /* Initialize the iteration pointer. */
8032 invlist_iterfinish(invlist);
8034 SvREADONLY_on(invlist);
8038 #endif /* ifndef PERL_IN_XSUB_RE */
8041 S_invlist_extend(pTHX_ SV* const invlist, const UV new_max)
8043 /* Grow the maximum size of an inversion list */
8045 PERL_ARGS_ASSERT_INVLIST_EXTEND;
8047 assert(SvTYPE(invlist) == SVt_INVLIST);
8049 /* Add one to account for the zero element at the beginning which may not
8050 * be counted by the calling parameters */
8051 SvGROW((SV *)invlist, TO_INTERNAL_SIZE(new_max + 1));
8054 PERL_STATIC_INLINE void
8055 S_invlist_trim(SV* const invlist)
8057 PERL_ARGS_ASSERT_INVLIST_TRIM;
8059 assert(SvTYPE(invlist) == SVt_INVLIST);
8061 /* Change the length of the inversion list to how many entries it currently
8063 SvPV_shrink_to_cur((SV *) invlist);
8067 S__append_range_to_invlist(pTHX_ SV* const invlist,
8068 const UV start, const UV end)
8070 /* Subject to change or removal. Append the range from 'start' to 'end' at
8071 * the end of the inversion list. The range must be above any existing
8075 UV max = invlist_max(invlist);
8076 UV len = _invlist_len(invlist);
8079 PERL_ARGS_ASSERT__APPEND_RANGE_TO_INVLIST;
8081 if (len == 0) { /* Empty lists must be initialized */
8082 offset = start != 0;
8083 array = _invlist_array_init(invlist, ! offset);
8086 /* Here, the existing list is non-empty. The current max entry in the
8087 * list is generally the first value not in the set, except when the
8088 * set extends to the end of permissible values, in which case it is
8089 * the first entry in that final set, and so this call is an attempt to
8090 * append out-of-order */
8092 UV final_element = len - 1;
8093 array = invlist_array(invlist);
8094 if (array[final_element] > start
8095 || ELEMENT_RANGE_MATCHES_INVLIST(final_element))
8097 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",
8098 array[final_element], start,
8099 ELEMENT_RANGE_MATCHES_INVLIST(final_element) ? 't' : 'f');
8102 /* Here, it is a legal append. If the new range begins with the first
8103 * value not in the set, it is extending the set, so the new first
8104 * value not in the set is one greater than the newly extended range.
8106 offset = *get_invlist_offset_addr(invlist);
8107 if (array[final_element] == start) {
8108 if (end != UV_MAX) {
8109 array[final_element] = end + 1;
8112 /* But if the end is the maximum representable on the machine,
8113 * just let the range that this would extend to have no end */
8114 invlist_set_len(invlist, len - 1, offset);
8120 /* Here the new range doesn't extend any existing set. Add it */
8122 len += 2; /* Includes an element each for the start and end of range */
8124 /* If wll overflow the existing space, extend, which may cause the array to
8127 invlist_extend(invlist, len);
8129 /* Have to set len here to avoid assert failure in invlist_array() */
8130 invlist_set_len(invlist, len, offset);
8132 array = invlist_array(invlist);
8135 invlist_set_len(invlist, len, offset);
8138 /* The next item on the list starts the range, the one after that is
8139 * one past the new range. */
8140 array[len - 2] = start;
8141 if (end != UV_MAX) {
8142 array[len - 1] = end + 1;
8145 /* But if the end is the maximum representable on the machine, just let
8146 * the range have no end */
8147 invlist_set_len(invlist, len - 1, offset);
8151 #ifndef PERL_IN_XSUB_RE
8154 Perl__invlist_search(SV* const invlist, const UV cp)
8156 /* Searches the inversion list for the entry that contains the input code
8157 * point <cp>. If <cp> is not in the list, -1 is returned. Otherwise, the
8158 * return value is the index into the list's array of the range that
8163 IV high = _invlist_len(invlist);
8164 const IV highest_element = high - 1;
8167 PERL_ARGS_ASSERT__INVLIST_SEARCH;
8169 /* If list is empty, return failure. */
8174 /* (We can't get the array unless we know the list is non-empty) */
8175 array = invlist_array(invlist);
8177 mid = invlist_previous_index(invlist);
8178 assert(mid >=0 && mid <= highest_element);
8180 /* <mid> contains the cache of the result of the previous call to this
8181 * function (0 the first time). See if this call is for the same result,
8182 * or if it is for mid-1. This is under the theory that calls to this
8183 * function will often be for related code points that are near each other.
8184 * And benchmarks show that caching gives better results. We also test
8185 * here if the code point is within the bounds of the list. These tests
8186 * replace others that would have had to be made anyway to make sure that
8187 * the array bounds were not exceeded, and these give us extra information
8188 * at the same time */
8189 if (cp >= array[mid]) {
8190 if (cp >= array[highest_element]) {
8191 return highest_element;
8194 /* Here, array[mid] <= cp < array[highest_element]. This means that
8195 * the final element is not the answer, so can exclude it; it also
8196 * means that <mid> is not the final element, so can refer to 'mid + 1'
8198 if (cp < array[mid + 1]) {
8204 else { /* cp < aray[mid] */
8205 if (cp < array[0]) { /* Fail if outside the array */
8209 if (cp >= array[mid - 1]) {
8214 /* Binary search. What we are looking for is <i> such that
8215 * array[i] <= cp < array[i+1]
8216 * The loop below converges on the i+1. Note that there may not be an
8217 * (i+1)th element in the array, and things work nonetheless */
8218 while (low < high) {
8219 mid = (low + high) / 2;
8220 assert(mid <= highest_element);
8221 if (array[mid] <= cp) { /* cp >= array[mid] */
8224 /* We could do this extra test to exit the loop early.
8225 if (cp < array[low]) {
8230 else { /* cp < array[mid] */
8237 invlist_set_previous_index(invlist, high);
8242 Perl__invlist_populate_swatch(SV* const invlist,
8243 const UV start, const UV end, U8* swatch)
8245 /* populates a swatch of a swash the same way swatch_get() does in utf8.c,
8246 * but is used when the swash has an inversion list. This makes this much
8247 * faster, as it uses a binary search instead of a linear one. This is
8248 * intimately tied to that function, and perhaps should be in utf8.c,
8249 * except it is intimately tied to inversion lists as well. It assumes
8250 * that <swatch> is all 0's on input */
8253 const IV len = _invlist_len(invlist);
8257 PERL_ARGS_ASSERT__INVLIST_POPULATE_SWATCH;
8259 if (len == 0) { /* Empty inversion list */
8263 array = invlist_array(invlist);
8265 /* Find which element it is */
8266 i = _invlist_search(invlist, start);
8268 /* We populate from <start> to <end> */
8269 while (current < end) {
8272 /* The inversion list gives the results for every possible code point
8273 * after the first one in the list. Only those ranges whose index is
8274 * even are ones that the inversion list matches. For the odd ones,
8275 * and if the initial code point is not in the list, we have to skip
8276 * forward to the next element */
8277 if (i == -1 || ! ELEMENT_RANGE_MATCHES_INVLIST(i)) {
8279 if (i >= len) { /* Finished if beyond the end of the array */
8283 if (current >= end) { /* Finished if beyond the end of what we
8285 if (LIKELY(end < UV_MAX)) {
8289 /* We get here when the upper bound is the maximum
8290 * representable on the machine, and we are looking for just
8291 * that code point. Have to special case it */
8293 goto join_end_of_list;
8296 assert(current >= start);
8298 /* The current range ends one below the next one, except don't go past
8301 upper = (i < len && array[i] < end) ? array[i] : end;
8303 /* Here we are in a range that matches. Populate a bit in the 3-bit U8
8304 * for each code point in it */
8305 for (; current < upper; current++) {
8306 const STRLEN offset = (STRLEN)(current - start);
8307 swatch[offset >> 3] |= 1 << (offset & 7);
8312 /* Quit if at the end of the list */
8315 /* But first, have to deal with the highest possible code point on
8316 * the platform. The previous code assumes that <end> is one
8317 * beyond where we want to populate, but that is impossible at the
8318 * platform's infinity, so have to handle it specially */
8319 if (UNLIKELY(end == UV_MAX && ELEMENT_RANGE_MATCHES_INVLIST(len-1)))
8321 const STRLEN offset = (STRLEN)(end - start);
8322 swatch[offset >> 3] |= 1 << (offset & 7);
8327 /* Advance to the next range, which will be for code points not in the
8336 Perl__invlist_union_maybe_complement_2nd(pTHX_ SV* const a, SV* const b,
8337 const bool complement_b, SV** output)
8339 /* Take the union of two inversion lists and point <output> to it. *output
8340 * SHOULD BE DEFINED upon input, and if it points to one of the two lists,
8341 * the reference count to that list will be decremented if not already a
8342 * temporary (mortal); otherwise *output will be made correspondingly
8343 * mortal. The first list, <a>, may be NULL, in which case a copy of the
8344 * second list is returned. If <complement_b> is TRUE, the union is taken
8345 * of the complement (inversion) of <b> instead of b itself.
8347 * The basis for this comes from "Unicode Demystified" Chapter 13 by
8348 * Richard Gillam, published by Addison-Wesley, and explained at some
8349 * length there. The preface says to incorporate its examples into your
8350 * code at your own risk.
8352 * The algorithm is like a merge sort.
8354 * XXX A potential performance improvement is to keep track as we go along
8355 * if only one of the inputs contributes to the result, meaning the other
8356 * is a subset of that one. In that case, we can skip the final copy and
8357 * return the larger of the input lists, but then outside code might need
8358 * to keep track of whether to free the input list or not */
8360 const UV* array_a; /* a's array */
8362 UV len_a; /* length of a's array */
8365 SV* u; /* the resulting union */
8369 UV i_a = 0; /* current index into a's array */
8373 /* running count, as explained in the algorithm source book; items are
8374 * stopped accumulating and are output when the count changes to/from 0.
8375 * The count is incremented when we start a range that's in the set, and
8376 * decremented when we start a range that's not in the set. So its range
8377 * is 0 to 2. Only when the count is zero is something not in the set.
8381 PERL_ARGS_ASSERT__INVLIST_UNION_MAYBE_COMPLEMENT_2ND;
8384 /* If either one is empty, the union is the other one */
8385 if (a == NULL || ((len_a = _invlist_len(a)) == 0)) {
8386 bool make_temp = FALSE; /* Should we mortalize the result? */
8390 if (! (make_temp = cBOOL(SvTEMP(a)))) {
8396 *output = invlist_clone(b);
8398 _invlist_invert(*output);
8400 } /* else *output already = b; */
8403 sv_2mortal(*output);
8407 else if ((len_b = _invlist_len(b)) == 0) {
8408 bool make_temp = FALSE;
8410 if (! (make_temp = cBOOL(SvTEMP(b)))) {
8415 /* The complement of an empty list is a list that has everything in it,
8416 * so the union with <a> includes everything too */
8419 if (! (make_temp = cBOOL(SvTEMP(a)))) {
8423 *output = _new_invlist(1);
8424 _append_range_to_invlist(*output, 0, UV_MAX);
8426 else if (*output != a) {
8427 *output = invlist_clone(a);
8429 /* else *output already = a; */
8432 sv_2mortal(*output);
8437 /* Here both lists exist and are non-empty */
8438 array_a = invlist_array(a);
8439 array_b = invlist_array(b);
8441 /* If are to take the union of 'a' with the complement of b, set it
8442 * up so are looking at b's complement. */
8445 /* To complement, we invert: if the first element is 0, remove it. To
8446 * do this, we just pretend the array starts one later */
8447 if (array_b[0] == 0) {
8453 /* But if the first element is not zero, we pretend the list starts
8454 * at the 0 that is always stored immediately before the array. */
8460 /* Size the union for the worst case: that the sets are completely
8462 u = _new_invlist(len_a + len_b);
8464 /* Will contain U+0000 if either component does */
8465 array_u = _invlist_array_init(u, (len_a > 0 && array_a[0] == 0)
8466 || (len_b > 0 && array_b[0] == 0));
8468 /* Go through each list item by item, stopping when exhausted one of
8470 while (i_a < len_a && i_b < len_b) {
8471 UV cp; /* The element to potentially add to the union's array */
8472 bool cp_in_set; /* is it in the the input list's set or not */
8474 /* We need to take one or the other of the two inputs for the union.
8475 * Since we are merging two sorted lists, we take the smaller of the
8476 * next items. In case of a tie, we take the one that is in its set
8477 * first. If we took one not in the set first, it would decrement the
8478 * count, possibly to 0 which would cause it to be output as ending the
8479 * range, and the next time through we would take the same number, and
8480 * output it again as beginning the next range. By doing it the
8481 * opposite way, there is no possibility that the count will be
8482 * momentarily decremented to 0, and thus the two adjoining ranges will
8483 * be seamlessly merged. (In a tie and both are in the set or both not
8484 * in the set, it doesn't matter which we take first.) */
8485 if (array_a[i_a] < array_b[i_b]
8486 || (array_a[i_a] == array_b[i_b]
8487 && ELEMENT_RANGE_MATCHES_INVLIST(i_a)))
8489 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_a);
8493 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_b);
8494 cp = array_b[i_b++];
8497 /* Here, have chosen which of the two inputs to look at. Only output
8498 * if the running count changes to/from 0, which marks the
8499 * beginning/end of a range in that's in the set */
8502 array_u[i_u++] = cp;
8509 array_u[i_u++] = cp;
8514 /* Here, we are finished going through at least one of the lists, which
8515 * means there is something remaining in at most one. We check if the list
8516 * that hasn't been exhausted is positioned such that we are in the middle
8517 * of a range in its set or not. (i_a and i_b point to the element beyond
8518 * the one we care about.) If in the set, we decrement 'count'; if 0, there
8519 * is potentially more to output.
8520 * There are four cases:
8521 * 1) Both weren't in their sets, count is 0, and remains 0. What's left
8522 * in the union is entirely from the non-exhausted set.
8523 * 2) Both were in their sets, count is 2. Nothing further should
8524 * be output, as everything that remains will be in the exhausted
8525 * list's set, hence in the union; decrementing to 1 but not 0 insures
8527 * 3) the exhausted was in its set, non-exhausted isn't, count is 1.
8528 * Nothing further should be output because the union includes
8529 * everything from the exhausted set. Not decrementing ensures that.
8530 * 4) the exhausted wasn't in its set, non-exhausted is, count is 1;
8531 * decrementing to 0 insures that we look at the remainder of the
8532 * non-exhausted set */
8533 if ((i_a != len_a && PREV_RANGE_MATCHES_INVLIST(i_a))
8534 || (i_b != len_b && PREV_RANGE_MATCHES_INVLIST(i_b)))
8539 /* The final length is what we've output so far, plus what else is about to
8540 * be output. (If 'count' is non-zero, then the input list we exhausted
8541 * has everything remaining up to the machine's limit in its set, and hence
8542 * in the union, so there will be no further output. */
8545 /* At most one of the subexpressions will be non-zero */
8546 len_u += (len_a - i_a) + (len_b - i_b);
8549 /* Set result to final length, which can change the pointer to array_u, so
8551 if (len_u != _invlist_len(u)) {
8552 invlist_set_len(u, len_u, *get_invlist_offset_addr(u));
8554 array_u = invlist_array(u);
8557 /* When 'count' is 0, the list that was exhausted (if one was shorter than
8558 * the other) ended with everything above it not in its set. That means
8559 * that the remaining part of the union is precisely the same as the
8560 * non-exhausted list, so can just copy it unchanged. (If both list were
8561 * exhausted at the same time, then the operations below will be both 0.)
8564 IV copy_count; /* At most one will have a non-zero copy count */
8565 if ((copy_count = len_a - i_a) > 0) {
8566 Copy(array_a + i_a, array_u + i_u, copy_count, UV);
8568 else if ((copy_count = len_b - i_b) > 0) {
8569 Copy(array_b + i_b, array_u + i_u, copy_count, UV);
8573 /* We may be removing a reference to one of the inputs. If so, the output
8574 * is made mortal if the input was. (Mortal SVs shouldn't have their ref
8575 * count decremented) */
8576 if (a == *output || b == *output) {
8577 assert(! invlist_is_iterating(*output));
8578 if ((SvTEMP(*output))) {
8582 SvREFCNT_dec_NN(*output);
8592 Perl__invlist_intersection_maybe_complement_2nd(pTHX_ SV* const a, SV* const b,
8593 const bool complement_b, SV** i)
8595 /* Take the intersection of two inversion lists and point <i> to it. *i
8596 * SHOULD BE DEFINED upon input, and if it points to one of the two lists,
8597 * the reference count to that list will be decremented if not already a
8598 * temporary (mortal); otherwise *i will be made correspondingly mortal.
8599 * The first list, <a>, may be NULL, in which case an empty list is
8600 * returned. If <complement_b> is TRUE, the result will be the
8601 * intersection of <a> and the complement (or inversion) of <b> instead of
8604 * The basis for this comes from "Unicode Demystified" Chapter 13 by
8605 * Richard Gillam, published by Addison-Wesley, and explained at some
8606 * length there. The preface says to incorporate its examples into your
8607 * code at your own risk. In fact, it had bugs
8609 * The algorithm is like a merge sort, and is essentially the same as the
8613 const UV* array_a; /* a's array */
8615 UV len_a; /* length of a's array */
8618 SV* r; /* the resulting intersection */
8622 UV i_a = 0; /* current index into a's array */
8626 /* running count, as explained in the algorithm source book; items are
8627 * stopped accumulating and are output when the count changes to/from 2.
8628 * The count is incremented when we start a range that's in the set, and
8629 * decremented when we start a range that's not in the set. So its range
8630 * is 0 to 2. Only when the count is 2 is something in the intersection.
8634 PERL_ARGS_ASSERT__INVLIST_INTERSECTION_MAYBE_COMPLEMENT_2ND;
8637 /* Special case if either one is empty */
8638 len_a = (a == NULL) ? 0 : _invlist_len(a);
8639 if ((len_a == 0) || ((len_b = _invlist_len(b)) == 0)) {
8640 bool make_temp = FALSE;
8642 if (len_a != 0 && complement_b) {
8644 /* Here, 'a' is not empty, therefore from the above 'if', 'b' must
8645 * be empty. Here, also we are using 'b's complement, which hence
8646 * must be every possible code point. Thus the intersection is
8650 if (! (make_temp = cBOOL(SvTEMP(b)))) {
8655 *i = invlist_clone(a);
8657 /* else *i is already 'a' */
8665 /* Here, 'a' or 'b' is empty and not using the complement of 'b'. The
8666 * intersection must be empty */
8668 if (! (make_temp = cBOOL(SvTEMP(a)))) {
8673 if (! (make_temp = cBOOL(SvTEMP(b)))) {
8677 *i = _new_invlist(0);
8685 /* Here both lists exist and are non-empty */
8686 array_a = invlist_array(a);
8687 array_b = invlist_array(b);
8689 /* If are to take the intersection of 'a' with the complement of b, set it
8690 * up so are looking at b's complement. */
8693 /* To complement, we invert: if the first element is 0, remove it. To
8694 * do this, we just pretend the array starts one later */
8695 if (array_b[0] == 0) {
8701 /* But if the first element is not zero, we pretend the list starts
8702 * at the 0 that is always stored immediately before the array. */
8708 /* Size the intersection for the worst case: that the intersection ends up
8709 * fragmenting everything to be completely disjoint */
8710 r= _new_invlist(len_a + len_b);
8712 /* Will contain U+0000 iff both components do */
8713 array_r = _invlist_array_init(r, len_a > 0 && array_a[0] == 0
8714 && len_b > 0 && array_b[0] == 0);
8716 /* Go through each list item by item, stopping when exhausted one of
8718 while (i_a < len_a && i_b < len_b) {
8719 UV cp; /* The element to potentially add to the intersection's
8721 bool cp_in_set; /* Is it in the input list's set or not */
8723 /* We need to take one or the other of the two inputs for the
8724 * intersection. Since we are merging two sorted lists, we take the
8725 * smaller of the next items. In case of a tie, we take the one that
8726 * is not in its set first (a difference from the union algorithm). If
8727 * we took one in the set first, it would increment the count, possibly
8728 * to 2 which would cause it to be output as starting a range in the
8729 * intersection, and the next time through we would take that same
8730 * number, and output it again as ending the set. By doing it the
8731 * opposite of this, there is no possibility that the count will be
8732 * momentarily incremented to 2. (In a tie and both are in the set or
8733 * both not in the set, it doesn't matter which we take first.) */
8734 if (array_a[i_a] < array_b[i_b]
8735 || (array_a[i_a] == array_b[i_b]
8736 && ! ELEMENT_RANGE_MATCHES_INVLIST(i_a)))
8738 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_a);
8742 cp_in_set = ELEMENT_RANGE_MATCHES_INVLIST(i_b);
8746 /* Here, have chosen which of the two inputs to look at. Only output
8747 * if the running count changes to/from 2, which marks the
8748 * beginning/end of a range that's in the intersection */
8752 array_r[i_r++] = cp;
8757 array_r[i_r++] = cp;
8763 /* Here, we are finished going through at least one of the lists, which
8764 * means there is something remaining in at most one. We check if the list
8765 * that has been exhausted is positioned such that we are in the middle
8766 * of a range in its set or not. (i_a and i_b point to elements 1 beyond
8767 * the ones we care about.) There are four cases:
8768 * 1) Both weren't in their sets, count is 0, and remains 0. There's
8769 * nothing left in the intersection.
8770 * 2) Both were in their sets, count is 2 and perhaps is incremented to
8771 * above 2. What should be output is exactly that which is in the
8772 * non-exhausted set, as everything it has is also in the intersection
8773 * set, and everything it doesn't have can't be in the intersection
8774 * 3) The exhausted was in its set, non-exhausted isn't, count is 1, and
8775 * gets incremented to 2. Like the previous case, the intersection is
8776 * everything that remains in the non-exhausted set.
8777 * 4) the exhausted wasn't in its set, non-exhausted is, count is 1, and
8778 * remains 1. And the intersection has nothing more. */
8779 if ((i_a == len_a && PREV_RANGE_MATCHES_INVLIST(i_a))
8780 || (i_b == len_b && PREV_RANGE_MATCHES_INVLIST(i_b)))
8785 /* The final length is what we've output so far plus what else is in the
8786 * intersection. At most one of the subexpressions below will be non-zero
8790 len_r += (len_a - i_a) + (len_b - i_b);
8793 /* Set result to final length, which can change the pointer to array_r, so
8795 if (len_r != _invlist_len(r)) {
8796 invlist_set_len(r, len_r, *get_invlist_offset_addr(r));
8798 array_r = invlist_array(r);
8801 /* Finish outputting any remaining */
8802 if (count >= 2) { /* At most one will have a non-zero copy count */
8804 if ((copy_count = len_a - i_a) > 0) {
8805 Copy(array_a + i_a, array_r + i_r, copy_count, UV);
8807 else if ((copy_count = len_b - i_b) > 0) {
8808 Copy(array_b + i_b, array_r + i_r, copy_count, UV);
8812 /* We may be removing a reference to one of the inputs. If so, the output
8813 * is made mortal if the input was. (Mortal SVs shouldn't have their ref
8814 * count decremented) */
8815 if (a == *i || b == *i) {
8816 assert(! invlist_is_iterating(*i));
8821 SvREFCNT_dec_NN(*i);
8831 Perl__add_range_to_invlist(pTHX_ SV* invlist, const UV start, const UV end)
8833 /* Add the range from 'start' to 'end' inclusive to the inversion list's
8834 * set. A pointer to the inversion list is returned. This may actually be
8835 * a new list, in which case the passed in one has been destroyed. The
8836 * passed in inversion list can be NULL, in which case a new one is created
8837 * with just the one range in it */
8842 if (invlist == NULL) {
8843 invlist = _new_invlist(2);
8847 len = _invlist_len(invlist);
8850 /* If comes after the final entry actually in the list, can just append it
8853 || (! ELEMENT_RANGE_MATCHES_INVLIST(len - 1)
8854 && start >= invlist_array(invlist)[len - 1]))
8856 _append_range_to_invlist(invlist, start, end);
8860 /* Here, can't just append things, create and return a new inversion list
8861 * which is the union of this range and the existing inversion list */
8862 range_invlist = _new_invlist(2);
8863 _append_range_to_invlist(range_invlist, start, end);
8865 _invlist_union(invlist, range_invlist, &invlist);
8867 /* The temporary can be freed */
8868 SvREFCNT_dec_NN(range_invlist);
8874 Perl__setup_canned_invlist(pTHX_ const STRLEN size, const UV element0,
8875 UV** other_elements_ptr)
8877 /* Create and return an inversion list whose contents are to be populated
8878 * by the caller. The caller gives the number of elements (in 'size') and
8879 * the very first element ('element0'). This function will set
8880 * '*other_elements_ptr' to an array of UVs, where the remaining elements
8883 * Obviously there is some trust involved that the caller will properly
8884 * fill in the other elements of the array.
8886 * (The first element needs to be passed in, as the underlying code does
8887 * things differently depending on whether it is zero or non-zero) */
8889 SV* invlist = _new_invlist(size);
8892 PERL_ARGS_ASSERT__SETUP_CANNED_INVLIST;
8894 _append_range_to_invlist(invlist, element0, element0);
8895 offset = *get_invlist_offset_addr(invlist);
8897 invlist_set_len(invlist, size, offset);
8898 *other_elements_ptr = invlist_array(invlist) + 1;
8904 PERL_STATIC_INLINE SV*
8905 S_add_cp_to_invlist(pTHX_ SV* invlist, const UV cp) {
8906 return _add_range_to_invlist(invlist, cp, cp);
8909 #ifndef PERL_IN_XSUB_RE
8911 Perl__invlist_invert(pTHX_ SV* const invlist)
8913 /* Complement the input inversion list. This adds a 0 if the list didn't
8914 * have a zero; removes it otherwise. As described above, the data
8915 * structure is set up so that this is very efficient */
8917 PERL_ARGS_ASSERT__INVLIST_INVERT;
8919 assert(! invlist_is_iterating(invlist));
8921 /* The inverse of matching nothing is matching everything */
8922 if (_invlist_len(invlist) == 0) {
8923 _append_range_to_invlist(invlist, 0, UV_MAX);
8927 *get_invlist_offset_addr(invlist) = ! *get_invlist_offset_addr(invlist);
8932 PERL_STATIC_INLINE SV*
8933 S_invlist_clone(pTHX_ SV* const invlist)
8936 /* Return a new inversion list that is a copy of the input one, which is
8937 * unchanged. The new list will not be mortal even if the old one was. */
8939 /* Need to allocate extra space to accommodate Perl's addition of a
8940 * trailing NUL to SvPV's, since it thinks they are always strings */
8941 SV* new_invlist = _new_invlist(_invlist_len(invlist) + 1);
8942 STRLEN physical_length = SvCUR(invlist);
8943 bool offset = *(get_invlist_offset_addr(invlist));
8945 PERL_ARGS_ASSERT_INVLIST_CLONE;
8947 *(get_invlist_offset_addr(new_invlist)) = offset;
8948 invlist_set_len(new_invlist, _invlist_len(invlist), offset);
8949 Copy(SvPVX(invlist), SvPVX(new_invlist), physical_length, char);
8954 PERL_STATIC_INLINE STRLEN*
8955 S_get_invlist_iter_addr(SV* invlist)
8957 /* Return the address of the UV that contains the current iteration
8960 PERL_ARGS_ASSERT_GET_INVLIST_ITER_ADDR;
8962 assert(SvTYPE(invlist) == SVt_INVLIST);
8964 return &(((XINVLIST*) SvANY(invlist))->iterator);
8967 PERL_STATIC_INLINE void
8968 S_invlist_iterinit(SV* invlist) /* Initialize iterator for invlist */
8970 PERL_ARGS_ASSERT_INVLIST_ITERINIT;
8972 *get_invlist_iter_addr(invlist) = 0;
8975 PERL_STATIC_INLINE void
8976 S_invlist_iterfinish(SV* invlist)
8978 /* Terminate iterator for invlist. This is to catch development errors.
8979 * Any iteration that is interrupted before completed should call this
8980 * function. Functions that add code points anywhere else but to the end
8981 * of an inversion list assert that they are not in the middle of an
8982 * iteration. If they were, the addition would make the iteration
8983 * problematical: if the iteration hadn't reached the place where things
8984 * were being added, it would be ok */
8986 PERL_ARGS_ASSERT_INVLIST_ITERFINISH;
8988 *get_invlist_iter_addr(invlist) = (STRLEN) UV_MAX;
8992 S_invlist_iternext(SV* invlist, UV* start, UV* end)
8994 /* An C<invlist_iterinit> call on <invlist> must be used to set this up.
8995 * This call sets in <*start> and <*end>, the next range in <invlist>.
8996 * Returns <TRUE> if successful and the next call will return the next
8997 * range; <FALSE> if was already at the end of the list. If the latter,
8998 * <*start> and <*end> are unchanged, and the next call to this function
8999 * will start over at the beginning of the list */
9001 STRLEN* pos = get_invlist_iter_addr(invlist);
9002 UV len = _invlist_len(invlist);
9005 PERL_ARGS_ASSERT_INVLIST_ITERNEXT;
9008 *pos = (STRLEN) UV_MAX; /* Force iterinit() to be required next time */
9012 array = invlist_array(invlist);
9014 *start = array[(*pos)++];
9020 *end = array[(*pos)++] - 1;
9026 PERL_STATIC_INLINE bool
9027 S_invlist_is_iterating(SV* const invlist)
9029 PERL_ARGS_ASSERT_INVLIST_IS_ITERATING;
9031 return *(get_invlist_iter_addr(invlist)) < (STRLEN) UV_MAX;
9034 PERL_STATIC_INLINE UV
9035 S_invlist_highest(SV* const invlist)
9037 /* Returns the highest code point that matches an inversion list. This API
9038 * has an ambiguity, as it returns 0 under either the highest is actually
9039 * 0, or if the list is empty. If this distinction matters to you, check
9040 * for emptiness before calling this function */
9042 UV len = _invlist_len(invlist);
9045 PERL_ARGS_ASSERT_INVLIST_HIGHEST;
9051 array = invlist_array(invlist);
9053 /* The last element in the array in the inversion list always starts a
9054 * range that goes to infinity. That range may be for code points that are
9055 * matched in the inversion list, or it may be for ones that aren't
9056 * matched. In the latter case, the highest code point in the set is one
9057 * less than the beginning of this range; otherwise it is the final element
9058 * of this range: infinity */
9059 return (ELEMENT_RANGE_MATCHES_INVLIST(len - 1))
9061 : array[len - 1] - 1;
9064 #ifndef PERL_IN_XSUB_RE
9066 Perl__invlist_contents(pTHX_ SV* const invlist)
9068 /* Get the contents of an inversion list into a string SV so that they can
9069 * be printed out. It uses the format traditionally done for debug tracing
9073 SV* output = newSVpvs("\n");
9075 PERL_ARGS_ASSERT__INVLIST_CONTENTS;
9077 assert(! invlist_is_iterating(invlist));
9079 invlist_iterinit(invlist);
9080 while (invlist_iternext(invlist, &start, &end)) {
9081 if (end == UV_MAX) {
9082 Perl_sv_catpvf(aTHX_ output, "%04"UVXf"\tINFINITY\n", start);
9084 else if (end != start) {
9085 Perl_sv_catpvf(aTHX_ output, "%04"UVXf"\t%04"UVXf"\n",
9089 Perl_sv_catpvf(aTHX_ output, "%04"UVXf"\n", start);
9097 #ifndef PERL_IN_XSUB_RE
9099 Perl__invlist_dump(pTHX_ PerlIO *file, I32 level,
9100 const char * const indent, SV* const invlist)
9102 /* Designed to be called only by do_sv_dump(). Dumps out the ranges of the
9103 * inversion list 'invlist' to 'file' at 'level' Each line is prefixed by
9104 * the string 'indent'. The output looks like this:
9105 [0] 0x000A .. 0x000D
9107 [4] 0x2028 .. 0x2029
9108 [6] 0x3104 .. INFINITY
9109 * This means that the first range of code points matched by the list are
9110 * 0xA through 0xD; the second range contains only the single code point
9111 * 0x85, etc. An inversion list is an array of UVs. Two array elements
9112 * are used to define each range (except if the final range extends to
9113 * infinity, only a single element is needed). The array index of the
9114 * first element for the corresponding range is given in brackets. */
9119 PERL_ARGS_ASSERT__INVLIST_DUMP;
9121 if (invlist_is_iterating(invlist)) {
9122 Perl_dump_indent(aTHX_ level, file,
9123 "%sCan't dump inversion list because is in middle of iterating\n",
9128 invlist_iterinit(invlist);
9129 while (invlist_iternext(invlist, &start, &end)) {
9130 if (end == UV_MAX) {
9131 Perl_dump_indent(aTHX_ level, file,
9132 "%s[%"UVuf"] 0x%04"UVXf" .. INFINITY\n",
9133 indent, (UV)count, start);
9135 else if (end != start) {
9136 Perl_dump_indent(aTHX_ level, file,
9137 "%s[%"UVuf"] 0x%04"UVXf" .. 0x%04"UVXf"\n",
9138 indent, (UV)count, start, end);
9141 Perl_dump_indent(aTHX_ level, file, "%s[%"UVuf"] 0x%04"UVXf"\n",
9142 indent, (UV)count, start);
9149 Perl__load_PL_utf8_foldclosures (pTHX)
9151 assert(! PL_utf8_foldclosures);
9153 /* If the folds haven't been read in, call a fold function
9155 if (! PL_utf8_tofold) {
9156 U8 dummy[UTF8_MAXBYTES_CASE+1];
9158 /* This string is just a short named one above \xff */
9159 to_utf8_fold((U8*) HYPHEN_UTF8, dummy, NULL);
9160 assert(PL_utf8_tofold); /* Verify that worked */
9162 PL_utf8_foldclosures = _swash_inversion_hash(PL_utf8_tofold);
9166 #ifdef PERL_ARGS_ASSERT__INVLISTEQ
9168 S__invlistEQ(pTHX_ SV* const a, SV* const b, const bool complement_b)
9170 /* Return a boolean as to if the two passed in inversion lists are
9171 * identical. The final argument, if TRUE, says to take the complement of
9172 * the second inversion list before doing the comparison */
9174 const UV* array_a = invlist_array(a);
9175 const UV* array_b = invlist_array(b);
9176 UV len_a = _invlist_len(a);
9177 UV len_b = _invlist_len(b);
9179 UV i = 0; /* current index into the arrays */
9180 bool retval = TRUE; /* Assume are identical until proven otherwise */
9182 PERL_ARGS_ASSERT__INVLISTEQ;
9184 /* If are to compare 'a' with the complement of b, set it
9185 * up so are looking at b's complement. */
9188 /* The complement of nothing is everything, so <a> would have to have
9189 * just one element, starting at zero (ending at infinity) */
9191 return (len_a == 1 && array_a[0] == 0);
9193 else if (array_b[0] == 0) {
9195 /* Otherwise, to complement, we invert. Here, the first element is
9196 * 0, just remove it. To do this, we just pretend the array starts
9204 /* But if the first element is not zero, we pretend the list starts
9205 * at the 0 that is always stored immediately before the array. */
9211 /* Make sure that the lengths are the same, as well as the final element
9212 * before looping through the remainder. (Thus we test the length, final,
9213 * and first elements right off the bat) */
9214 if (len_a != len_b || array_a[len_a-1] != array_b[len_a-1]) {
9217 else for (i = 0; i < len_a - 1; i++) {
9218 if (array_a[i] != array_b[i]) {
9228 #undef HEADER_LENGTH
9229 #undef TO_INTERNAL_SIZE
9230 #undef FROM_INTERNAL_SIZE
9231 #undef INVLIST_VERSION_ID
9233 /* End of inversion list object */
9236 S_parse_lparen_question_flags(pTHX_ RExC_state_t *pRExC_state)
9238 /* This parses the flags that are in either the '(?foo)' or '(?foo:bar)'
9239 * constructs, and updates RExC_flags with them. On input, RExC_parse
9240 * should point to the first flag; it is updated on output to point to the
9241 * final ')' or ':'. There needs to be at least one flag, or this will
9244 /* for (?g), (?gc), and (?o) warnings; warning
9245 about (?c) will warn about (?g) -- japhy */
9247 #define WASTED_O 0x01
9248 #define WASTED_G 0x02
9249 #define WASTED_C 0x04
9250 #define WASTED_GC (WASTED_G|WASTED_C)
9251 I32 wastedflags = 0x00;
9252 U32 posflags = 0, negflags = 0;
9253 U32 *flagsp = &posflags;
9254 char has_charset_modifier = '\0';
9256 bool has_use_defaults = FALSE;
9257 const char* const seqstart = RExC_parse - 1; /* Point to the '?' */
9259 PERL_ARGS_ASSERT_PARSE_LPAREN_QUESTION_FLAGS;
9261 /* '^' as an initial flag sets certain defaults */
9262 if (UCHARAT(RExC_parse) == '^') {
9264 has_use_defaults = TRUE;
9265 STD_PMMOD_FLAGS_CLEAR(&RExC_flags);
9266 set_regex_charset(&RExC_flags, (RExC_utf8 || RExC_uni_semantics)
9267 ? REGEX_UNICODE_CHARSET
9268 : REGEX_DEPENDS_CHARSET);
9271 cs = get_regex_charset(RExC_flags);
9272 if (cs == REGEX_DEPENDS_CHARSET
9273 && (RExC_utf8 || RExC_uni_semantics))
9275 cs = REGEX_UNICODE_CHARSET;
9278 while (*RExC_parse) {
9279 /* && strchr("iogcmsx", *RExC_parse) */
9280 /* (?g), (?gc) and (?o) are useless here
9281 and must be globally applied -- japhy */
9282 switch (*RExC_parse) {
9284 /* Code for the imsx flags */
9285 CASE_STD_PMMOD_FLAGS_PARSE_SET(flagsp);
9287 case LOCALE_PAT_MOD:
9288 if (has_charset_modifier) {
9289 goto excess_modifier;
9291 else if (flagsp == &negflags) {
9294 cs = REGEX_LOCALE_CHARSET;
9295 has_charset_modifier = LOCALE_PAT_MOD;
9297 case UNICODE_PAT_MOD:
9298 if (has_charset_modifier) {
9299 goto excess_modifier;
9301 else if (flagsp == &negflags) {
9304 cs = REGEX_UNICODE_CHARSET;
9305 has_charset_modifier = UNICODE_PAT_MOD;
9307 case ASCII_RESTRICT_PAT_MOD:
9308 if (flagsp == &negflags) {
9311 if (has_charset_modifier) {
9312 if (cs != REGEX_ASCII_RESTRICTED_CHARSET) {
9313 goto excess_modifier;
9315 /* Doubled modifier implies more restricted */
9316 cs = REGEX_ASCII_MORE_RESTRICTED_CHARSET;
9319 cs = REGEX_ASCII_RESTRICTED_CHARSET;
9321 has_charset_modifier = ASCII_RESTRICT_PAT_MOD;
9323 case DEPENDS_PAT_MOD:
9324 if (has_use_defaults) {
9325 goto fail_modifiers;
9327 else if (flagsp == &negflags) {
9330 else if (has_charset_modifier) {
9331 goto excess_modifier;
9334 /* The dual charset means unicode semantics if the
9335 * pattern (or target, not known until runtime) are
9336 * utf8, or something in the pattern indicates unicode
9338 cs = (RExC_utf8 || RExC_uni_semantics)
9339 ? REGEX_UNICODE_CHARSET
9340 : REGEX_DEPENDS_CHARSET;
9341 has_charset_modifier = DEPENDS_PAT_MOD;
9345 if (has_charset_modifier == ASCII_RESTRICT_PAT_MOD) {
9346 vFAIL2("Regexp modifier \"%c\" may appear a maximum of twice", ASCII_RESTRICT_PAT_MOD);
9348 else if (has_charset_modifier == *(RExC_parse - 1)) {
9349 vFAIL2("Regexp modifier \"%c\" may not appear twice",
9353 vFAIL3("Regexp modifiers \"%c\" and \"%c\" are mutually exclusive", has_charset_modifier, *(RExC_parse - 1));
9358 vFAIL2("Regexp modifier \"%c\" may not appear after the \"-\"",
9361 case ONCE_PAT_MOD: /* 'o' */
9362 case GLOBAL_PAT_MOD: /* 'g' */
9363 if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
9364 const I32 wflagbit = *RExC_parse == 'o'
9367 if (! (wastedflags & wflagbit) ) {
9368 wastedflags |= wflagbit;
9369 /* diag_listed_as: Useless (?-%s) - don't use /%s modifier in regex; marked by <-- HERE in m/%s/ */
9372 "Useless (%s%c) - %suse /%c modifier",
9373 flagsp == &negflags ? "?-" : "?",
9375 flagsp == &negflags ? "don't " : "",
9382 case CONTINUE_PAT_MOD: /* 'c' */
9383 if (SIZE_ONLY && ckWARN(WARN_REGEXP)) {
9384 if (! (wastedflags & WASTED_C) ) {
9385 wastedflags |= WASTED_GC;
9386 /* diag_listed_as: Useless (?-%s) - don't use /%s modifier in regex; marked by <-- HERE in m/%s/ */
9389 "Useless (%sc) - %suse /gc modifier",
9390 flagsp == &negflags ? "?-" : "?",
9391 flagsp == &negflags ? "don't " : ""
9396 case KEEPCOPY_PAT_MOD: /* 'p' */
9397 if (flagsp == &negflags) {
9399 ckWARNreg(RExC_parse + 1,"Useless use of (?-p)");
9401 *flagsp |= RXf_PMf_KEEPCOPY;
9405 /* A flag is a default iff it is following a minus, so
9406 * if there is a minus, it means will be trying to
9407 * re-specify a default which is an error */
9408 if (has_use_defaults || flagsp == &negflags) {
9409 goto fail_modifiers;
9412 wastedflags = 0; /* reset so (?g-c) warns twice */
9416 RExC_flags |= posflags;
9417 RExC_flags &= ~negflags;
9418 set_regex_charset(&RExC_flags, cs);
9419 if (RExC_flags & RXf_PMf_FOLD) {
9420 RExC_contains_i = 1;
9426 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
9427 /* diag_listed_as: Sequence (?%s...) not recognized in regex; marked by <-- HERE in m/%s/ */
9428 vFAIL2utf8f("Sequence (%"UTF8f"...) not recognized",
9429 UTF8fARG(UTF, RExC_parse-seqstart, seqstart));
9438 - reg - regular expression, i.e. main body or parenthesized thing
9440 * Caller must absorb opening parenthesis.
9442 * Combining parenthesis handling with the base level of regular expression
9443 * is a trifle forced, but the need to tie the tails of the branches to what
9444 * follows makes it hard to avoid.
9446 #define REGTAIL(x,y,z) regtail((x),(y),(z),depth+1)
9448 #define REGTAIL_STUDY(x,y,z) regtail_study((x),(y),(z),depth+1)
9450 #define REGTAIL_STUDY(x,y,z) regtail((x),(y),(z),depth+1)
9453 /* Returns NULL, setting *flagp to TRYAGAIN at the end of (?) that only sets
9454 flags. Returns NULL, setting *flagp to RESTART_UTF8 if the sizing scan
9455 needs to be restarted.
9456 Otherwise would only return NULL if regbranch() returns NULL, which
9459 S_reg(pTHX_ RExC_state_t *pRExC_state, I32 paren, I32 *flagp,U32 depth)
9460 /* paren: Parenthesized? 0=top; 1,2=inside '(': changed to letter.
9461 * 2 is like 1, but indicates that nextchar() has been called to advance
9462 * RExC_parse beyond the '('. Things like '(?' are indivisible tokens, and
9463 * this flag alerts us to the need to check for that */
9465 regnode *ret; /* Will be the head of the group. */
9468 regnode *ender = NULL;
9471 U32 oregflags = RExC_flags;
9472 bool have_branch = 0;
9474 I32 freeze_paren = 0;
9475 I32 after_freeze = 0;
9476 I32 num; /* numeric backreferences */
9478 char * parse_start = RExC_parse; /* MJD */
9479 char * const oregcomp_parse = RExC_parse;
9481 GET_RE_DEBUG_FLAGS_DECL;
9483 PERL_ARGS_ASSERT_REG;
9484 DEBUG_PARSE("reg ");
9486 *flagp = 0; /* Tentatively. */
9489 /* Make an OPEN node, if parenthesized. */
9492 /* Under /x, space and comments can be gobbled up between the '(' and
9493 * here (if paren ==2). The forms '(*VERB' and '(?...' disallow such
9494 * intervening space, as the sequence is a token, and a token should be
9496 bool has_intervening_patws = paren == 2 && *(RExC_parse - 1) != '(';
9498 if ( *RExC_parse == '*') { /* (*VERB:ARG) */
9499 char *start_verb = RExC_parse;
9500 STRLEN verb_len = 0;
9501 char *start_arg = NULL;
9502 unsigned char op = 0;
9504 int internal_argval = 0; /* internal_argval is only useful if
9507 if (has_intervening_patws) {
9509 vFAIL("In '(*VERB...)', the '(' and '*' must be adjacent");
9511 while ( *RExC_parse && *RExC_parse != ')' ) {
9512 if ( *RExC_parse == ':' ) {
9513 start_arg = RExC_parse + 1;
9519 verb_len = RExC_parse - start_verb;
9522 while ( *RExC_parse && *RExC_parse != ')' )
9524 if ( *RExC_parse != ')' )
9525 vFAIL("Unterminated verb pattern argument");
9526 if ( RExC_parse == start_arg )
9529 if ( *RExC_parse != ')' )
9530 vFAIL("Unterminated verb pattern");
9533 switch ( *start_verb ) {
9534 case 'A': /* (*ACCEPT) */
9535 if ( memEQs(start_verb,verb_len,"ACCEPT") ) {
9537 internal_argval = RExC_nestroot;
9540 case 'C': /* (*COMMIT) */
9541 if ( memEQs(start_verb,verb_len,"COMMIT") )
9544 case 'F': /* (*FAIL) */
9545 if ( verb_len==1 || memEQs(start_verb,verb_len,"FAIL") ) {
9550 case ':': /* (*:NAME) */
9551 case 'M': /* (*MARK:NAME) */
9552 if ( verb_len==0 || memEQs(start_verb,verb_len,"MARK") ) {
9557 case 'P': /* (*PRUNE) */
9558 if ( memEQs(start_verb,verb_len,"PRUNE") )
9561 case 'S': /* (*SKIP) */
9562 if ( memEQs(start_verb,verb_len,"SKIP") )
9565 case 'T': /* (*THEN) */
9566 /* [19:06] <TimToady> :: is then */
9567 if ( memEQs(start_verb,verb_len,"THEN") ) {
9569 RExC_seen |= REG_CUTGROUP_SEEN;
9574 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
9576 "Unknown verb pattern '%"UTF8f"'",
9577 UTF8fARG(UTF, verb_len, start_verb));
9580 if ( start_arg && internal_argval ) {
9581 vFAIL3("Verb pattern '%.*s' may not have an argument",
9582 verb_len, start_verb);
9583 } else if ( argok < 0 && !start_arg ) {
9584 vFAIL3("Verb pattern '%.*s' has a mandatory argument",
9585 verb_len, start_verb);
9587 ret = reganode(pRExC_state, op, internal_argval);
9588 if ( ! internal_argval && ! SIZE_ONLY ) {
9590 SV *sv = newSVpvn( start_arg,
9591 RExC_parse - start_arg);
9592 ARG(ret) = add_data( pRExC_state,
9594 RExC_rxi->data->data[ARG(ret)]=(void*)sv;
9601 if (!internal_argval)
9602 RExC_seen |= REG_VERBARG_SEEN;
9603 } else if ( start_arg ) {
9604 vFAIL3("Verb pattern '%.*s' may not have an argument",
9605 verb_len, start_verb);
9607 ret = reg_node(pRExC_state, op);
9609 nextchar(pRExC_state);
9612 else if (*RExC_parse == '?') { /* (?...) */
9613 bool is_logical = 0;
9614 const char * const seqstart = RExC_parse;
9615 const char * endptr;
9616 if (has_intervening_patws) {
9618 vFAIL("In '(?...)', the '(' and '?' must be adjacent");
9622 paren = *RExC_parse++;
9623 ret = NULL; /* For look-ahead/behind. */
9626 case 'P': /* (?P...) variants for those used to PCRE/Python */
9627 paren = *RExC_parse++;
9628 if ( paren == '<') /* (?P<...>) named capture */
9630 else if (paren == '>') { /* (?P>name) named recursion */
9631 goto named_recursion;
9633 else if (paren == '=') { /* (?P=...) named backref */
9634 /* this pretty much dupes the code for \k<NAME> in
9635 * regatom(), if you change this make sure you change that
9637 char* name_start = RExC_parse;
9639 SV *sv_dat = reg_scan_name(pRExC_state,
9640 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
9641 if (RExC_parse == name_start || *RExC_parse != ')')
9642 /* diag_listed_as: Sequence ?P=... not terminated in regex; marked by <-- HERE in m/%s/ */
9643 vFAIL2("Sequence %.3s... not terminated",parse_start);
9646 num = add_data( pRExC_state, STR_WITH_LEN("S"));
9647 RExC_rxi->data->data[num]=(void*)sv_dat;
9648 SvREFCNT_inc_simple_void(sv_dat);
9651 ret = reganode(pRExC_state,
9654 : (ASCII_FOLD_RESTRICTED)
9656 : (AT_LEAST_UNI_SEMANTICS)
9664 Set_Node_Offset(ret, parse_start+1);
9665 Set_Node_Cur_Length(ret, parse_start);
9667 nextchar(pRExC_state);
9671 /* diag_listed_as: Sequence (?%s...) not recognized in regex; marked by <-- HERE in m/%s/ */
9672 vFAIL3("Sequence (%.*s...) not recognized",
9673 RExC_parse-seqstart, seqstart);
9675 case '<': /* (?<...) */
9676 if (*RExC_parse == '!')
9678 else if (*RExC_parse != '=')
9684 case '\'': /* (?'...') */
9685 name_start= RExC_parse;
9686 svname = reg_scan_name(pRExC_state,
9687 SIZE_ONLY /* reverse test from the others */
9688 ? REG_RSN_RETURN_NAME
9689 : REG_RSN_RETURN_NULL);
9690 if (RExC_parse == name_start || *RExC_parse != paren)
9691 vFAIL2("Sequence (?%c... not terminated",
9692 paren=='>' ? '<' : paren);
9696 if (!svname) /* shouldn't happen */
9698 "panic: reg_scan_name returned NULL");
9699 if (!RExC_paren_names) {
9700 RExC_paren_names= newHV();
9701 sv_2mortal(MUTABLE_SV(RExC_paren_names));
9703 RExC_paren_name_list= newAV();
9704 sv_2mortal(MUTABLE_SV(RExC_paren_name_list));
9707 he_str = hv_fetch_ent( RExC_paren_names, svname, 1, 0 );
9709 sv_dat = HeVAL(he_str);
9711 /* croak baby croak */
9713 "panic: paren_name hash element allocation failed");
9714 } else if ( SvPOK(sv_dat) ) {
9715 /* (?|...) can mean we have dupes so scan to check
9716 its already been stored. Maybe a flag indicating
9717 we are inside such a construct would be useful,
9718 but the arrays are likely to be quite small, so
9719 for now we punt -- dmq */
9720 IV count = SvIV(sv_dat);
9721 I32 *pv = (I32*)SvPVX(sv_dat);
9723 for ( i = 0 ; i < count ; i++ ) {
9724 if ( pv[i] == RExC_npar ) {
9730 pv = (I32*)SvGROW(sv_dat,
9731 SvCUR(sv_dat) + sizeof(I32)+1);
9732 SvCUR_set(sv_dat, SvCUR(sv_dat) + sizeof(I32));
9733 pv[count] = RExC_npar;
9734 SvIV_set(sv_dat, SvIVX(sv_dat) + 1);
9737 (void)SvUPGRADE(sv_dat,SVt_PVNV);
9738 sv_setpvn(sv_dat, (char *)&(RExC_npar),
9741 SvIV_set(sv_dat, 1);
9744 /* Yes this does cause a memory leak in debugging Perls
9746 if (!av_store(RExC_paren_name_list,
9747 RExC_npar, SvREFCNT_inc(svname)))
9748 SvREFCNT_dec_NN(svname);
9751 /*sv_dump(sv_dat);*/
9753 nextchar(pRExC_state);
9755 goto capturing_parens;
9757 RExC_seen |= REG_LOOKBEHIND_SEEN;
9758 RExC_in_lookbehind++;
9761 case '=': /* (?=...) */
9762 RExC_seen_zerolen++;
9764 case '!': /* (?!...) */
9765 RExC_seen_zerolen++;
9766 if (*RExC_parse == ')') {
9767 ret=reg_node(pRExC_state, OPFAIL);
9768 nextchar(pRExC_state);
9772 case '|': /* (?|...) */
9773 /* branch reset, behave like a (?:...) except that
9774 buffers in alternations share the same numbers */
9776 after_freeze = freeze_paren = RExC_npar;
9778 case ':': /* (?:...) */
9779 case '>': /* (?>...) */
9781 case '$': /* (?$...) */
9782 case '@': /* (?@...) */
9783 vFAIL2("Sequence (?%c...) not implemented", (int)paren);
9785 case '0' : /* (?0) */
9786 case 'R' : /* (?R) */
9787 if (*RExC_parse != ')')
9788 FAIL("Sequence (?R) not terminated");
9789 ret = reg_node(pRExC_state, GOSTART);
9790 RExC_seen |= REG_GOSTART_SEEN;
9791 *flagp |= POSTPONED;
9792 nextchar(pRExC_state);
9795 /* named and numeric backreferences */
9796 case '&': /* (?&NAME) */
9797 parse_start = RExC_parse - 1;
9800 SV *sv_dat = reg_scan_name(pRExC_state,
9801 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
9802 num = sv_dat ? *((I32 *)SvPVX(sv_dat)) : 0;
9804 if (RExC_parse == RExC_end || *RExC_parse != ')')
9805 vFAIL("Sequence (?&... not terminated");
9806 goto gen_recurse_regop;
9807 assert(0); /* NOT REACHED */
9809 if (!(RExC_parse[0] >= '1' && RExC_parse[0] <= '9')) {
9811 vFAIL("Illegal pattern");
9813 goto parse_recursion;
9815 case '-': /* (?-1) */
9816 if (!(RExC_parse[0] >= '1' && RExC_parse[0] <= '9')) {
9817 RExC_parse--; /* rewind to let it be handled later */
9821 case '1': case '2': case '3': case '4': /* (?1) */
9822 case '5': case '6': case '7': case '8': case '9':
9826 bool is_neg = FALSE;
9827 parse_start = RExC_parse - 1; /* MJD */
9828 if (*RExC_parse == '-') {
9832 num = grok_atou(RExC_parse, &endptr);
9834 RExC_parse = (char*)endptr;
9836 /* Some limit for num? */
9840 if (*RExC_parse!=')')
9841 vFAIL("Expecting close bracket");
9844 if ( paren == '-' ) {
9846 Diagram of capture buffer numbering.
9847 Top line is the normal capture buffer numbers
9848 Bottom line is the negative indexing as from
9852 /(a(x)y)(a(b(c(?-2)d)e)f)(g(h))/
9856 num = RExC_npar + num;
9859 vFAIL("Reference to nonexistent group");
9861 } else if ( paren == '+' ) {
9862 num = RExC_npar + num - 1;
9865 ret = reganode(pRExC_state, GOSUB, num);
9867 if (num > (I32)RExC_rx->nparens) {
9869 vFAIL("Reference to nonexistent group");
9871 ARG2L_SET( ret, RExC_recurse_count++);
9873 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
9874 "Recurse #%"UVuf" to %"IVdf"\n",
9875 (UV)ARG(ret), (IV)ARG2L(ret)));
9879 RExC_seen |= REG_RECURSE_SEEN;
9880 Set_Node_Length(ret, 1 + regarglen[OP(ret)]); /* MJD */
9881 Set_Node_Offset(ret, parse_start); /* MJD */
9883 *flagp |= POSTPONED;
9884 nextchar(pRExC_state);
9887 assert(0); /* NOT REACHED */
9889 case '?': /* (??...) */
9891 if (*RExC_parse != '{') {
9893 /* diag_listed_as: Sequence (?%s...) not recognized in regex; marked by <-- HERE in m/%s/ */
9895 "Sequence (%"UTF8f"...) not recognized",
9896 UTF8fARG(UTF, RExC_parse-seqstart, seqstart));
9899 *flagp |= POSTPONED;
9900 paren = *RExC_parse++;
9902 case '{': /* (?{...}) */
9905 struct reg_code_block *cb;
9907 RExC_seen_zerolen++;
9909 if ( !pRExC_state->num_code_blocks
9910 || pRExC_state->code_index >= pRExC_state->num_code_blocks
9911 || pRExC_state->code_blocks[pRExC_state->code_index].start
9912 != (STRLEN)((RExC_parse -3 - (is_logical ? 1 : 0))
9915 if (RExC_pm_flags & PMf_USE_RE_EVAL)
9916 FAIL("panic: Sequence (?{...}): no code block found\n");
9917 FAIL("Eval-group not allowed at runtime, use re 'eval'");
9919 /* this is a pre-compiled code block (?{...}) */
9920 cb = &pRExC_state->code_blocks[pRExC_state->code_index];
9921 RExC_parse = RExC_start + cb->end;
9924 if (cb->src_regex) {
9925 n = add_data(pRExC_state, STR_WITH_LEN("rl"));
9926 RExC_rxi->data->data[n] =
9927 (void*)SvREFCNT_inc((SV*)cb->src_regex);
9928 RExC_rxi->data->data[n+1] = (void*)o;
9931 n = add_data(pRExC_state,
9932 (RExC_pm_flags & PMf_HAS_CV) ? "L" : "l", 1);
9933 RExC_rxi->data->data[n] = (void*)o;
9936 pRExC_state->code_index++;
9937 nextchar(pRExC_state);
9941 ret = reg_node(pRExC_state, LOGICAL);
9942 eval = reganode(pRExC_state, EVAL, n);
9945 /* for later propagation into (??{}) return value */
9946 eval->flags = (U8) (RExC_flags & RXf_PMf_COMPILETIME);
9948 REGTAIL(pRExC_state, ret, eval);
9949 /* deal with the length of this later - MJD */
9952 ret = reganode(pRExC_state, EVAL, n);
9953 Set_Node_Length(ret, RExC_parse - parse_start + 1);
9954 Set_Node_Offset(ret, parse_start);
9957 case '(': /* (?(?{...})...) and (?(?=...)...) */
9960 if (RExC_parse[0] == '?') { /* (?(?...)) */
9961 if (RExC_parse[1] == '=' || RExC_parse[1] == '!'
9962 || RExC_parse[1] == '<'
9963 || RExC_parse[1] == '{') { /* Lookahead or eval. */
9967 ret = reg_node(pRExC_state, LOGICAL);
9971 tail = reg(pRExC_state, 1, &flag, depth+1);
9972 if (flag & RESTART_UTF8) {
9973 *flagp = RESTART_UTF8;
9976 REGTAIL(pRExC_state, ret, tail);
9979 /* Fall through to ‘Unknown switch condition’ at the
9980 end of the if/else chain. */
9982 else if ( RExC_parse[0] == '<' /* (?(<NAME>)...) */
9983 || RExC_parse[0] == '\'' ) /* (?('NAME')...) */
9985 char ch = RExC_parse[0] == '<' ? '>' : '\'';
9986 char *name_start= RExC_parse++;
9988 SV *sv_dat=reg_scan_name(pRExC_state,
9989 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
9990 if (RExC_parse == name_start || *RExC_parse != ch)
9991 vFAIL2("Sequence (?(%c... not terminated",
9992 (ch == '>' ? '<' : ch));
9995 num = add_data( pRExC_state, STR_WITH_LEN("S"));
9996 RExC_rxi->data->data[num]=(void*)sv_dat;
9997 SvREFCNT_inc_simple_void(sv_dat);
9999 ret = reganode(pRExC_state,NGROUPP,num);
10000 goto insert_if_check_paren;
10002 else if (RExC_parse[0] == 'D' &&
10003 RExC_parse[1] == 'E' &&
10004 RExC_parse[2] == 'F' &&
10005 RExC_parse[3] == 'I' &&
10006 RExC_parse[4] == 'N' &&
10007 RExC_parse[5] == 'E')
10009 ret = reganode(pRExC_state,DEFINEP,0);
10012 goto insert_if_check_paren;
10014 else if (RExC_parse[0] == 'R') {
10017 if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
10018 parno = grok_atou(RExC_parse, &endptr);
10020 RExC_parse = (char*)endptr;
10021 } else if (RExC_parse[0] == '&') {
10024 sv_dat = reg_scan_name(pRExC_state,
10026 ? REG_RSN_RETURN_NULL
10027 : REG_RSN_RETURN_DATA);
10028 parno = sv_dat ? *((I32 *)SvPVX(sv_dat)) : 0;
10030 ret = reganode(pRExC_state,INSUBP,parno);
10031 goto insert_if_check_paren;
10033 else if (RExC_parse[0] >= '1' && RExC_parse[0] <= '9' ) {
10037 parno = grok_atou(RExC_parse, &endptr);
10039 RExC_parse = (char*)endptr;
10040 ret = reganode(pRExC_state, GROUPP, parno);
10042 insert_if_check_paren:
10043 if (*(tmp = nextchar(pRExC_state)) != ')') {
10044 /* nextchar also skips comments, so undo its work
10045 * and skip over the the next character.
10048 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
10049 vFAIL("Switch condition not recognized");
10052 REGTAIL(pRExC_state, ret, reganode(pRExC_state, IFTHEN, 0));
10053 br = regbranch(pRExC_state, &flags, 1,depth+1);
10055 if (flags & RESTART_UTF8) {
10056 *flagp = RESTART_UTF8;
10059 FAIL2("panic: regbranch returned NULL, flags=%#"UVxf"",
10062 REGTAIL(pRExC_state, br, reganode(pRExC_state,
10064 c = *nextchar(pRExC_state);
10065 if (flags&HASWIDTH)
10066 *flagp |= HASWIDTH;
10069 vFAIL("(?(DEFINE)....) does not allow branches");
10071 /* Fake one for optimizer. */
10072 lastbr = reganode(pRExC_state, IFTHEN, 0);
10074 if (!regbranch(pRExC_state, &flags, 1,depth+1)) {
10075 if (flags & RESTART_UTF8) {
10076 *flagp = RESTART_UTF8;
10079 FAIL2("panic: regbranch returned NULL, flags=%#"UVxf"",
10082 REGTAIL(pRExC_state, ret, lastbr);
10083 if (flags&HASWIDTH)
10084 *flagp |= HASWIDTH;
10085 c = *nextchar(pRExC_state);
10090 vFAIL("Switch (?(condition)... contains too many branches");
10091 ender = reg_node(pRExC_state, TAIL);
10092 REGTAIL(pRExC_state, br, ender);
10094 REGTAIL(pRExC_state, lastbr, ender);
10095 REGTAIL(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
10098 REGTAIL(pRExC_state, ret, ender);
10099 RExC_size++; /* XXX WHY do we need this?!!
10100 For large programs it seems to be required
10101 but I can't figure out why. -- dmq*/
10104 RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
10105 vFAIL("Unknown switch condition (?(...))");
10107 case '[': /* (?[ ... ]) */
10108 return handle_regex_sets(pRExC_state, NULL, flagp, depth,
10111 RExC_parse--; /* for vFAIL to print correctly */
10112 vFAIL("Sequence (? incomplete");
10114 default: /* e.g., (?i) */
10117 parse_lparen_question_flags(pRExC_state);
10118 if (UCHARAT(RExC_parse) != ':') {
10119 nextchar(pRExC_state);
10124 nextchar(pRExC_state);
10134 ret = reganode(pRExC_state, OPEN, parno);
10136 if (!RExC_nestroot)
10137 RExC_nestroot = parno;
10138 if (RExC_seen & REG_RECURSE_SEEN
10139 && !RExC_open_parens[parno-1])
10141 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
10142 "Setting open paren #%"IVdf" to %d\n",
10143 (IV)parno, REG_NODE_NUM(ret)));
10144 RExC_open_parens[parno-1]= ret;
10147 Set_Node_Length(ret, 1); /* MJD */
10148 Set_Node_Offset(ret, RExC_parse); /* MJD */
10156 /* Pick up the branches, linking them together. */
10157 parse_start = RExC_parse; /* MJD */
10158 br = regbranch(pRExC_state, &flags, 1,depth+1);
10160 /* branch_len = (paren != 0); */
10163 if (flags & RESTART_UTF8) {
10164 *flagp = RESTART_UTF8;
10167 FAIL2("panic: regbranch returned NULL, flags=%#"UVxf"", (UV) flags);
10169 if (*RExC_parse == '|') {
10170 if (!SIZE_ONLY && RExC_extralen) {
10171 reginsert(pRExC_state, BRANCHJ, br, depth+1);
10174 reginsert(pRExC_state, BRANCH, br, depth+1);
10175 Set_Node_Length(br, paren != 0);
10176 Set_Node_Offset_To_R(br-RExC_emit_start, parse_start-RExC_start);
10180 RExC_extralen += 1; /* For BRANCHJ-BRANCH. */
10182 else if (paren == ':') {
10183 *flagp |= flags&SIMPLE;
10185 if (is_open) { /* Starts with OPEN. */
10186 REGTAIL(pRExC_state, ret, br); /* OPEN -> first. */
10188 else if (paren != '?') /* Not Conditional */
10190 *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
10192 while (*RExC_parse == '|') {
10193 if (!SIZE_ONLY && RExC_extralen) {
10194 ender = reganode(pRExC_state, LONGJMP,0);
10196 /* Append to the previous. */
10197 REGTAIL(pRExC_state, NEXTOPER(NEXTOPER(lastbr)), ender);
10200 RExC_extralen += 2; /* Account for LONGJMP. */
10201 nextchar(pRExC_state);
10202 if (freeze_paren) {
10203 if (RExC_npar > after_freeze)
10204 after_freeze = RExC_npar;
10205 RExC_npar = freeze_paren;
10207 br = regbranch(pRExC_state, &flags, 0, depth+1);
10210 if (flags & RESTART_UTF8) {
10211 *flagp = RESTART_UTF8;
10214 FAIL2("panic: regbranch returned NULL, flags=%#"UVxf"", (UV) flags);
10216 REGTAIL(pRExC_state, lastbr, br); /* BRANCH -> BRANCH. */
10218 *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
10221 if (have_branch || paren != ':') {
10222 /* Make a closing node, and hook it on the end. */
10225 ender = reg_node(pRExC_state, TAIL);
10228 ender = reganode(pRExC_state, CLOSE, parno);
10229 if (!SIZE_ONLY && RExC_seen & REG_RECURSE_SEEN) {
10230 DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
10231 "Setting close paren #%"IVdf" to %d\n",
10232 (IV)parno, REG_NODE_NUM(ender)));
10233 RExC_close_parens[parno-1]= ender;
10234 if (RExC_nestroot == parno)
10237 Set_Node_Offset(ender,RExC_parse+1); /* MJD */
10238 Set_Node_Length(ender,1); /* MJD */
10244 *flagp &= ~HASWIDTH;
10247 ender = reg_node(pRExC_state, SUCCEED);
10250 ender = reg_node(pRExC_state, END);
10252 assert(!RExC_opend); /* there can only be one! */
10253 RExC_opend = ender;
10257 DEBUG_PARSE_r(if (!SIZE_ONLY) {
10258 SV * const mysv_val1=sv_newmortal();
10259 SV * const mysv_val2=sv_newmortal();
10260 DEBUG_PARSE_MSG("lsbr");
10261 regprop(RExC_rx, mysv_val1, lastbr, NULL);
10262 regprop(RExC_rx, mysv_val2, ender, NULL);
10263 PerlIO_printf(Perl_debug_log, "~ tying lastbr %s (%"IVdf") to ender %s (%"IVdf") offset %"IVdf"\n",
10264 SvPV_nolen_const(mysv_val1),
10265 (IV)REG_NODE_NUM(lastbr),
10266 SvPV_nolen_const(mysv_val2),
10267 (IV)REG_NODE_NUM(ender),
10268 (IV)(ender - lastbr)
10271 REGTAIL(pRExC_state, lastbr, ender);
10273 if (have_branch && !SIZE_ONLY) {
10274 char is_nothing= 1;
10276 RExC_seen |= REG_TOP_LEVEL_BRANCHES_SEEN;
10278 /* Hook the tails of the branches to the closing node. */
10279 for (br = ret; br; br = regnext(br)) {
10280 const U8 op = PL_regkind[OP(br)];
10281 if (op == BRANCH) {
10282 REGTAIL_STUDY(pRExC_state, NEXTOPER(br), ender);
10283 if ( OP(NEXTOPER(br)) != NOTHING
10284 || regnext(NEXTOPER(br)) != ender)
10287 else if (op == BRANCHJ) {
10288 REGTAIL_STUDY(pRExC_state, NEXTOPER(NEXTOPER(br)), ender);
10289 /* for now we always disable this optimisation * /
10290 if ( OP(NEXTOPER(NEXTOPER(br))) != NOTHING
10291 || regnext(NEXTOPER(NEXTOPER(br))) != ender)
10297 br= PL_regkind[OP(ret)] != BRANCH ? regnext(ret) : ret;
10298 DEBUG_PARSE_r(if (!SIZE_ONLY) {
10299 SV * const mysv_val1=sv_newmortal();
10300 SV * const mysv_val2=sv_newmortal();
10301 DEBUG_PARSE_MSG("NADA");
10302 regprop(RExC_rx, mysv_val1, ret, NULL);
10303 regprop(RExC_rx, mysv_val2, ender, NULL);
10304 PerlIO_printf(Perl_debug_log, "~ converting ret %s (%"IVdf") to ender %s (%"IVdf") offset %"IVdf"\n",
10305 SvPV_nolen_const(mysv_val1),
10306 (IV)REG_NODE_NUM(ret),
10307 SvPV_nolen_const(mysv_val2),
10308 (IV)REG_NODE_NUM(ender),
10313 if (OP(ender) == TAIL) {
10318 for ( opt= br + 1; opt < ender ; opt++ )
10319 OP(opt)= OPTIMIZED;
10320 NEXT_OFF(br)= ender - br;
10328 static const char parens[] = "=!<,>";
10330 if (paren && (p = strchr(parens, paren))) {
10331 U8 node = ((p - parens) % 2) ? UNLESSM : IFMATCH;
10332 int flag = (p - parens) > 1;
10335 node = SUSPEND, flag = 0;
10336 reginsert(pRExC_state, node,ret, depth+1);
10337 Set_Node_Cur_Length(ret, parse_start);
10338 Set_Node_Offset(ret, parse_start + 1);
10340 REGTAIL_STUDY(pRExC_state, ret, reg_node(pRExC_state, TAIL));
10344 /* Check for proper termination. */
10346 /* restore original flags, but keep (?p) */
10347 RExC_flags = oregflags | (RExC_flags & RXf_PMf_KEEPCOPY);
10348 if (RExC_parse >= RExC_end || *nextchar(pRExC_state) != ')') {
10349 RExC_parse = oregcomp_parse;
10350 vFAIL("Unmatched (");
10353 else if (!paren && RExC_parse < RExC_end) {
10354 if (*RExC_parse == ')') {
10356 vFAIL("Unmatched )");
10359 FAIL("Junk on end of regexp"); /* "Can't happen". */
10360 assert(0); /* NOTREACHED */
10363 if (RExC_in_lookbehind) {
10364 RExC_in_lookbehind--;
10366 if (after_freeze > RExC_npar)
10367 RExC_npar = after_freeze;
10372 - regbranch - one alternative of an | operator
10374 * Implements the concatenation operator.
10376 * Returns NULL, setting *flagp to RESTART_UTF8 if the sizing scan needs to be
10380 S_regbranch(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, I32 first, U32 depth)
10383 regnode *chain = NULL;
10385 I32 flags = 0, c = 0;
10386 GET_RE_DEBUG_FLAGS_DECL;
10388 PERL_ARGS_ASSERT_REGBRANCH;
10390 DEBUG_PARSE("brnc");
10395 if (!SIZE_ONLY && RExC_extralen)
10396 ret = reganode(pRExC_state, BRANCHJ,0);
10398 ret = reg_node(pRExC_state, BRANCH);
10399 Set_Node_Length(ret, 1);
10403 if (!first && SIZE_ONLY)
10404 RExC_extralen += 1; /* BRANCHJ */
10406 *flagp = WORST; /* Tentatively. */
10409 nextchar(pRExC_state);
10410 while (RExC_parse < RExC_end && *RExC_parse != '|' && *RExC_parse != ')') {
10411 flags &= ~TRYAGAIN;
10412 latest = regpiece(pRExC_state, &flags,depth+1);
10413 if (latest == NULL) {
10414 if (flags & TRYAGAIN)
10416 if (flags & RESTART_UTF8) {
10417 *flagp = RESTART_UTF8;
10420 FAIL2("panic: regpiece returned NULL, flags=%#"UVxf"", (UV) flags);
10422 else if (ret == NULL)
10424 *flagp |= flags&(HASWIDTH|POSTPONED);
10425 if (chain == NULL) /* First piece. */
10426 *flagp |= flags&SPSTART;
10429 REGTAIL(pRExC_state, chain, latest);
10434 if (chain == NULL) { /* Loop ran zero times. */
10435 chain = reg_node(pRExC_state, NOTHING);
10440 *flagp |= flags&SIMPLE;
10447 - regpiece - something followed by possible [*+?]
10449 * Note that the branching code sequences used for ? and the general cases
10450 * of * and + are somewhat optimized: they use the same NOTHING node as
10451 * both the endmarker for their branch list and the body of the last branch.
10452 * It might seem that this node could be dispensed with entirely, but the
10453 * endmarker role is not redundant.
10455 * Returns NULL, setting *flagp to TRYAGAIN if regatom() returns NULL with
10457 * Returns NULL, setting *flagp to RESTART_UTF8 if the sizing scan needs to be
10461 S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
10467 const char * const origparse = RExC_parse;
10469 I32 max = REG_INFTY;
10470 #ifdef RE_TRACK_PATTERN_OFFSETS
10473 const char *maxpos = NULL;
10475 /* Save the original in case we change the emitted regop to a FAIL. */
10476 regnode * const orig_emit = RExC_emit;
10478 GET_RE_DEBUG_FLAGS_DECL;
10480 PERL_ARGS_ASSERT_REGPIECE;
10482 DEBUG_PARSE("piec");
10484 ret = regatom(pRExC_state, &flags,depth+1);
10486 if (flags & (TRYAGAIN|RESTART_UTF8))
10487 *flagp |= flags & (TRYAGAIN|RESTART_UTF8);
10489 FAIL2("panic: regatom returned NULL, flags=%#"UVxf"", (UV) flags);
10495 if (op == '{' && regcurly(RExC_parse)) {
10497 #ifdef RE_TRACK_PATTERN_OFFSETS
10498 parse_start = RExC_parse; /* MJD */
10500 next = RExC_parse + 1;
10501 while (isDIGIT(*next) || *next == ',') {
10502 if (*next == ',') {
10510 if (*next == '}') { /* got one */
10511 const char* endptr;
10515 min = grok_atou(RExC_parse, &endptr);
10516 if (*maxpos == ',')
10519 maxpos = RExC_parse;
10520 max = grok_atou(maxpos, &endptr);
10521 if (!max && *maxpos != '0')
10522 max = REG_INFTY; /* meaning "infinity" */
10523 else if (max >= REG_INFTY)
10524 vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
10526 nextchar(pRExC_state);
10527 if (max < min) { /* If can't match, warn and optimize to fail
10530 ckWARNreg(RExC_parse, "Quantifier {n,m} with n > m can't match");
10532 /* We can't back off the size because we have to reserve
10533 * enough space for all the things we are about to throw
10534 * away, but we can shrink it by the ammount we are about
10535 * to re-use here */
10536 RExC_size = PREVOPER(RExC_size) - regarglen[(U8)OPFAIL];
10539 RExC_emit = orig_emit;
10541 ret = reg_node(pRExC_state, OPFAIL);
10544 else if (min == max
10545 && RExC_parse < RExC_end
10546 && (*RExC_parse == '?' || *RExC_parse == '+'))
10549 ckWARN2reg(RExC_parse + 1,
10550 "Useless use of greediness modifier '%c'",
10553 /* Absorb the modifier, so later code doesn't see nor use
10555 nextchar(pRExC_state);
10559 if ((flags&SIMPLE)) {
10560 RExC_naughty += 2 + RExC_naughty / 2;
10561 reginsert(pRExC_state, CURLY, ret, depth+1);
10562 Set_Node_Offset(ret, parse_start+1); /* MJD */
10563 Set_Node_Cur_Length(ret, parse_start);
10566 regnode * const w = reg_node(pRExC_state, WHILEM);
10569 REGTAIL(pRExC_state, ret, w);
10570 if (!SIZE_ONLY && RExC_extralen) {
10571 reginsert(pRExC_state, LONGJMP,ret, depth+1);
10572 reginsert(pRExC_state, NOTHING,ret, depth+1);
10573 NEXT_OFF(ret) = 3; /* Go over LONGJMP. */
10575 reginsert(pRExC_state, CURLYX,ret, depth+1);
10577 Set_Node_Offset(ret, parse_start+1);
10578 Set_Node_Length(ret,
10579 op == '{' ? (RExC_parse - parse_start) : 1);
10581 if (!SIZE_ONLY && RExC_extralen)
10582 NEXT_OFF(ret) = 3; /* Go over NOTHING to LONGJMP. */
10583 REGTAIL(pRExC_state, ret, reg_node(pRExC_state, NOTHING));
10585 RExC_whilem_seen++, RExC_extralen += 3;
10586 RExC_naughty += 4 + RExC_naughty; /* compound interest */
10593 *flagp |= HASWIDTH;
10595 ARG1_SET(ret, (U16)min);
10596 ARG2_SET(ret, (U16)max);
10598 if (max == REG_INFTY)
10599 RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
10605 if (!ISMULT1(op)) {
10610 #if 0 /* Now runtime fix should be reliable. */
10612 /* if this is reinstated, don't forget to put this back into perldiag:
10614 =item Regexp *+ operand could be empty at {#} in regex m/%s/
10616 (F) The part of the regexp subject to either the * or + quantifier
10617 could match an empty string. The {#} shows in the regular
10618 expression about where the problem was discovered.
10622 if (!(flags&HASWIDTH) && op != '?')
10623 vFAIL("Regexp *+ operand could be empty");
10626 #ifdef RE_TRACK_PATTERN_OFFSETS
10627 parse_start = RExC_parse;
10629 nextchar(pRExC_state);
10631 *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
10633 if (op == '*' && (flags&SIMPLE)) {
10634 reginsert(pRExC_state, STAR, ret, depth+1);
10637 RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
10639 else if (op == '*') {
10643 else if (op == '+' && (flags&SIMPLE)) {
10644 reginsert(pRExC_state, PLUS, ret, depth+1);
10647 RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
10649 else if (op == '+') {
10653 else if (op == '?') {
10658 if (!SIZE_ONLY && !(flags&(HASWIDTH|POSTPONED)) && max > REG_INFTY/3) {
10659 SAVEFREESV(RExC_rx_sv); /* in case of fatal warnings */
10660 ckWARN2reg(RExC_parse,
10661 "%"UTF8f" matches null string many times",
10662 UTF8fARG(UTF, (RExC_parse >= origparse
10663 ? RExC_parse - origparse
10666 (void)ReREFCNT_inc(RExC_rx_sv);
10669 if (RExC_parse < RExC_end && *RExC_parse == '?') {
10670 nextchar(pRExC_state);
10671 reginsert(pRExC_state, MINMOD, ret, depth+1);
10672 REGTAIL(pRExC_state, ret, ret + NODE_STEP_REGNODE);
10675 if (RExC_parse < RExC_end && *RExC_parse == '+') {
10677 nextchar(pRExC_state);
10678 ender = reg_node(pRExC_state, SUCCEED);
10679 REGTAIL(pRExC_state, ret, ender);
10680 reginsert(pRExC_state, SUSPEND, ret, depth+1);
10682 ender = reg_node(pRExC_state, TAIL);
10683 REGTAIL(pRExC_state, ret, ender);
10686 if (RExC_parse < RExC_end && ISMULT2(RExC_parse)) {
10688 vFAIL("Nested quantifiers");
10695 S_grok_bslash_N(pTHX_ RExC_state_t *pRExC_state, regnode** node_p,
10696 UV *valuep, I32 *flagp, U32 depth, bool in_char_class,
10697 const bool strict /* Apply stricter parsing rules? */
10701 /* This is expected to be called by a parser routine that has recognized '\N'
10702 and needs to handle the rest. RExC_parse is expected to point at the first
10703 char following the N at the time of the call. On successful return,
10704 RExC_parse has been updated to point to just after the sequence identified
10705 by this routine, and <*flagp> has been updated.
10707 The \N may be inside (indicated by the boolean <in_char_class>) or outside a
10710 \N may begin either a named sequence, or if outside a character class, mean
10711 to match a non-newline. For non single-quoted regexes, the tokenizer has
10712 attempted to decide which, and in the case of a named sequence, converted it
10713 into one of the forms: \N{} (if the sequence is null), or \N{U+c1.c2...},
10714 where c1... are the characters in the sequence. For single-quoted regexes,
10715 the tokenizer passes the \N sequence through unchanged; this code will not
10716 attempt to determine this nor expand those, instead raising a syntax error.
10717 The net effect is that if the beginning of the passed-in pattern isn't '{U+'
10718 or there is no '}', it signals that this \N occurrence means to match a
10721 Only the \N{U+...} form should occur in a character class, for the same
10722 reason that '.' inside a character class means to just match a period: it
10723 just doesn't make sense.
10725 The function raises an error (via vFAIL), and doesn't return for various
10726 syntax errors. Otherwise it returns TRUE and sets <node_p> or <valuep> on
10727 success; it returns FALSE otherwise. Returns FALSE, setting *flagp to
10728 RESTART_UTF8 if the sizing scan needs to be restarted. Such a restart is
10729 only possible if node_p is non-NULL.
10732 If <valuep> is non-null, it means the caller can accept an input sequence
10733 consisting of a just a single code point; <*valuep> is set to that value
10734 if the input is such.
10736 If <node_p> is non-null it signifies that the caller can accept any other
10737 legal sequence (i.e., one that isn't just a single code point). <*node_p>
10739 1) \N means not-a-NL: points to a newly created REG_ANY node;
10740 2) \N{}: points to a new NOTHING node;
10741 3) otherwise: points to a new EXACT node containing the resolved
10743 Note that FALSE is returned for single code point sequences if <valuep> is
10747 char * endbrace; /* '}' following the name */
10749 char *endchar; /* Points to '.' or '}' ending cur char in the input
10751 bool has_multiple_chars; /* true if the input stream contains a sequence of
10752 more than one character */
10754 GET_RE_DEBUG_FLAGS_DECL;
10756 PERL_ARGS_ASSERT_GROK_BSLASH_N;
10758 GET_RE_DEBUG_FLAGS;
10760 assert(cBOOL(node_p) ^ cBOOL(valuep)); /* Exactly one should be set */
10762 /* The [^\n] meaning of \N ignores spaces and comments under the /x
10763 * modifier. The other meaning does not, so use a temporary until we find
10764 * out which we are being called with */
10765 p = (RExC_flags & RXf_PMf_EXTENDED)
10766 ? regpatws(pRExC_state, RExC_parse,
10767 TRUE) /* means recognize comments */
10770 /* Disambiguate between \N meaning a named character versus \N meaning
10771 * [^\n]. The former is assumed when it can't be the latter. */
10772 if (*p != '{' || regcurly(p)) {
10775 /* no bare \N allowed in a charclass */
10776 if (in_char_class) {
10777 vFAIL("\\N in a character class must be a named character: \\N{...}");
10781 RExC_parse--; /* Need to back off so nextchar() doesn't skip the
10783 nextchar(pRExC_state);
10784 *node_p = reg_node(pRExC_state, REG_ANY);
10785 *flagp |= HASWIDTH|SIMPLE;
10787 Set_Node_Length(*node_p, 1); /* MJD */
10791 /* Here, we have decided it should be a named character or sequence */
10793 /* The test above made sure that the next real character is a '{', but
10794 * under the /x modifier, it could be separated by space (or a comment and
10795 * \n) and this is not allowed (for consistency with \x{...} and the
10796 * tokenizer handling of \N{NAME}). */
10797 if (*RExC_parse != '{') {
10798 vFAIL("Missing braces on \\N{}");
10801 RExC_parse++; /* Skip past the '{' */
10803 if (! (endbrace = strchr(RExC_parse, '}')) /* no trailing brace */
10804 || ! (endbrace == RExC_parse /* nothing between the {} */
10805 || (endbrace - RExC_parse >= 2 /* U+ (bad hex is checked below
10807 && strnEQ(RExC_parse, "U+", 2)))) /* for a better error msg)
10810 if (endbrace) RExC_parse = endbrace; /* position msg's '<--HERE' */
10811 vFAIL("\\N{NAME} must be resolved by the lexer");
10814 if (endbrace == RExC_parse) { /* empty: \N{} */
10817 *node_p = reg_node(pRExC_state,NOTHING);
10819 else if (in_char_class) {
10820 if (SIZE_ONLY && in_char_class) {
10822 RExC_parse++; /* Position after the "}" */
10823 vFAIL("Zero length \\N{}");
10826 ckWARNreg(RExC_parse,
10827 "Ignoring zero length \\N{} in character class");
10835 nextchar(pRExC_state);
10839 RExC_uni_semantics = 1; /* Unicode named chars imply Unicode semantics */
10840 RExC_parse += 2; /* Skip past the 'U+' */
10842 endchar = RExC_parse + strcspn(RExC_parse, ".}");
10844 /* Code points are separated by dots. If none, there is only one code
10845 * point, and is terminated by the brace */
10846 has_multiple_chars = (endchar < endbrace);
10848 if (valuep && (! has_multiple_chars || in_char_class)) {
10849 /* We only pay attention to the first char of
10850 multichar strings being returned in char classes. I kinda wonder
10851 if this makes sense as it does change the behaviour
10852 from earlier versions, OTOH that behaviour was broken
10853 as well. XXX Solution is to recharacterize as
10854 [rest-of-class]|multi1|multi2... */
10856 STRLEN length_of_hex = (STRLEN)(endchar - RExC_parse);
10857 I32 grok_hex_flags = PERL_SCAN_ALLOW_UNDERSCORES
10858 | PERL_SCAN_DISALLOW_PREFIX
10859 | (SIZE_ONLY ? PERL_SCAN_SILENT_ILLDIGIT : 0);
10861 *valuep = grok_hex(RExC_parse, &length_of_hex, &grok_hex_flags, NULL);
10863 /* The tokenizer should have guaranteed validity, but it's possible to
10864 * bypass it by using single quoting, so check */
10865 if (length_of_hex == 0
10866 || length_of_hex != (STRLEN)(endchar - RExC_parse) )
10868 RExC_parse += length_of_hex; /* Includes all the valid */
10869 RExC_parse += (RExC_orig_utf8) /* point to after 1st invalid */
10870 ? UTF8SKIP(RExC_parse)
10872 /* Guard against malformed utf8 */
10873 if (RExC_parse >= endchar) {
10874 RExC_parse = endchar;
10876 vFAIL("Invalid hexadecimal number in \\N{U+...}");
10879 if (in_char_class && has_multiple_chars) {
10881 RExC_parse = endbrace;
10882 vFAIL("\\N{} in character class restricted to one character");
10885 ckWARNreg(endchar, "Using just the first character returned by \\N{} in character class");
10889 RExC_parse = endbrace + 1;
10891 else if (! node_p || ! has_multiple_chars) {
10893 /* Here, the input is legal, but not according to the caller's
10894 * options. We fail without advancing the parse, so that the
10895 * caller can try again */
10901 /* What is done here is to convert this to a sub-pattern of the form
10902 * (?:\x{char1}\x{char2}...)
10903 * and then call reg recursively. That way, it retains its atomicness,
10904 * while not having to worry about special handling that some code
10905 * points may have. toke.c has converted the original Unicode values
10906 * to native, so that we can just pass on the hex values unchanged. We
10907 * do have to set a flag to keep recoding from happening in the
10910 SV * substitute_parse = newSVpvn_flags("?:", 2, SVf_UTF8|SVs_TEMP);
10912 char *orig_end = RExC_end;
10915 while (RExC_parse < endbrace) {
10917 /* Convert to notation the rest of the code understands */
10918 sv_catpv(substitute_parse, "\\x{");
10919 sv_catpvn(substitute_parse, RExC_parse, endchar - RExC_parse);
10920 sv_catpv(substitute_parse, "}");
10922 /* Point to the beginning of the next character in the sequence. */
10923 RExC_parse = endchar + 1;
10924 endchar = RExC_parse + strcspn(RExC_parse, ".}");
10926 sv_catpv(substitute_parse, ")");
10928 RExC_parse = SvPV(substitute_parse, len);
10930 /* Don't allow empty number */
10932 vFAIL("Invalid hexadecimal number in \\N{U+...}");
10934 RExC_end = RExC_parse + len;
10936 /* The values are Unicode, and therefore not subject to recoding */
10937 RExC_override_recoding = 1;
10939 if (!(*node_p = reg(pRExC_state, 1, &flags, depth+1))) {
10940 if (flags & RESTART_UTF8) {
10941 *flagp = RESTART_UTF8;
10944 FAIL2("panic: reg returned NULL to grok_bslash_N, flags=%#"UVxf"",
10947 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
10949 RExC_parse = endbrace;
10950 RExC_end = orig_end;
10951 RExC_override_recoding = 0;
10953 nextchar(pRExC_state);
10963 * It returns the code point in utf8 for the value in *encp.
10964 * value: a code value in the source encoding
10965 * encp: a pointer to an Encode object
10967 * If the result from Encode is not a single character,
10968 * it returns U+FFFD (Replacement character) and sets *encp to NULL.
10971 S_reg_recode(pTHX_ const char value, SV **encp)
10974 SV * const sv = newSVpvn_flags(&value, numlen, SVs_TEMP);
10975 const char * const s = *encp ? sv_recode_to_utf8(sv, *encp) : SvPVX(sv);
10976 const STRLEN newlen = SvCUR(sv);
10977 UV uv = UNICODE_REPLACEMENT;
10979 PERL_ARGS_ASSERT_REG_RECODE;
10983 ? utf8n_to_uvchr((U8*)s, newlen, &numlen, UTF8_ALLOW_DEFAULT)
10986 if (!newlen || numlen != newlen) {
10987 uv = UNICODE_REPLACEMENT;
10993 PERL_STATIC_INLINE U8
10994 S_compute_EXACTish(RExC_state_t *pRExC_state)
10998 PERL_ARGS_ASSERT_COMPUTE_EXACTISH;
11004 op = get_regex_charset(RExC_flags);
11005 if (op >= REGEX_ASCII_RESTRICTED_CHARSET) {
11006 op--; /* /a is same as /u, and map /aa's offset to what /a's would have
11007 been, so there is no hole */
11010 return op + EXACTF;
11013 PERL_STATIC_INLINE void
11014 S_alloc_maybe_populate_EXACT(pTHX_ RExC_state_t *pRExC_state,
11015 regnode *node, I32* flagp, STRLEN len, UV code_point,
11018 /* This knows the details about sizing an EXACTish node, setting flags for
11019 * it (by setting <*flagp>, and potentially populating it with a single
11022 * If <len> (the length in bytes) is non-zero, this function assumes that
11023 * the node has already been populated, and just does the sizing. In this
11024 * case <code_point> should be the final code point that has already been
11025 * placed into the node. This value will be ignored except that under some
11026 * circumstances <*flagp> is set based on it.
11028 * If <len> is zero, the function assumes that the node is to contain only
11029 * the single character given by <code_point> and calculates what <len>
11030 * should be. In pass 1, it sizes the node appropriately. In pass 2, it
11031 * additionally will populate the node's STRING with <code_point> or its
11034 * In both cases <*flagp> is appropriately set
11036 * It knows that under FOLD, the Latin Sharp S and UTF characters above
11037 * 255, must be folded (the former only when the rules indicate it can
11040 * When it does the populating, it looks at the flag 'downgradable'. If
11041 * true with a node that folds, it checks if the single code point
11042 * participates in a fold, and if not downgrades the node to an EXACT.
11043 * This helps the optimizer */
11045 bool len_passed_in = cBOOL(len != 0);
11046 U8 character[UTF8_MAXBYTES_CASE+1];
11048 PERL_ARGS_ASSERT_ALLOC_MAYBE_POPULATE_EXACT;
11050 /* Don't bother to check for downgrading in PASS1, as it doesn't make any
11051 * sizing difference, and is extra work that is thrown away */
11052 if (downgradable && ! PASS2) {
11053 downgradable = FALSE;
11056 if (! len_passed_in) {
11058 if (UNI_IS_INVARIANT(code_point)) {
11059 if (LOC || ! FOLD) { /* /l defers folding until runtime */
11060 *character = (U8) code_point;
11062 else { /* Here is /i and not /l (toFOLD() is defined on just
11063 ASCII, which isn't the same thing as INVARIANT on
11064 EBCDIC, but it works there, as the extra invariants
11065 fold to themselves) */
11066 *character = toFOLD((U8) code_point);
11068 && *character == code_point
11069 && ! HAS_NONLATIN1_FOLD_CLOSURE(code_point))
11076 else if (FOLD && (! LOC
11077 || ! is_PROBLEMATIC_LOCALE_FOLD_cp(code_point)))
11078 { /* Folding, and ok to do so now */
11079 UV folded = _to_uni_fold_flags(
11083 FOLD_FLAGS_FULL | ((ASCII_FOLD_RESTRICTED)
11084 ? FOLD_FLAGS_NOMIX_ASCII
11087 && folded == code_point
11088 && ! _invlist_contains_cp(PL_utf8_foldable, code_point))
11093 else if (code_point <= MAX_UTF8_TWO_BYTE) {
11095 /* Not folding this cp, and can output it directly */
11096 *character = UTF8_TWO_BYTE_HI(code_point);
11097 *(character + 1) = UTF8_TWO_BYTE_LO(code_point);
11101 uvchr_to_utf8( character, code_point);
11102 len = UTF8SKIP(character);
11104 } /* Else pattern isn't UTF8. */
11106 *character = (U8) code_point;
11108 } /* Else is folded non-UTF8 */
11109 else if (LIKELY(code_point != LATIN_SMALL_LETTER_SHARP_S)) {
11111 /* We don't fold any non-UTF8 except possibly the Sharp s (see
11112 * comments at join_exact()); */
11113 *character = (U8) code_point;
11116 /* Can turn into an EXACT node if we know the fold at compile time,
11117 * and it folds to itself and doesn't particpate in other folds */
11120 && PL_fold_latin1[code_point] == code_point
11121 && (! HAS_NONLATIN1_FOLD_CLOSURE(code_point)
11122 || (isASCII(code_point) && ASCII_FOLD_RESTRICTED)))
11126 } /* else is Sharp s. May need to fold it */
11127 else if (AT_LEAST_UNI_SEMANTICS && ! ASCII_FOLD_RESTRICTED) {
11129 *(character + 1) = 's';
11133 *character = LATIN_SMALL_LETTER_SHARP_S;
11139 RExC_size += STR_SZ(len);
11142 RExC_emit += STR_SZ(len);
11143 STR_LEN(node) = len;
11144 if (! len_passed_in) {
11145 Copy((char *) character, STRING(node), len, char);
11149 *flagp |= HASWIDTH;
11151 /* A single character node is SIMPLE, except for the special-cased SHARP S
11153 if ((len == 1 || (UTF && len == UNISKIP(code_point)))
11154 && (code_point != LATIN_SMALL_LETTER_SHARP_S
11155 || ! FOLD || ! DEPENDS_SEMANTICS))
11160 /* The OP may not be well defined in PASS1 */
11161 if (PASS2 && OP(node) == EXACTFL) {
11162 RExC_contains_locale = 1;
11167 /* Parse backref decimal value, unless it's too big to sensibly be a backref,
11168 * in which case return I32_MAX (rather than possibly 32-bit wrapping) */
11171 S_backref_value(char *p)
11173 const char* endptr;
11174 UV val = grok_atou(p, &endptr);
11175 if (endptr == p || endptr == NULL || val > I32_MAX)
11182 - regatom - the lowest level
11184 Try to identify anything special at the start of the pattern. If there
11185 is, then handle it as required. This may involve generating a single regop,
11186 such as for an assertion; or it may involve recursing, such as to
11187 handle a () structure.
11189 If the string doesn't start with something special then we gobble up
11190 as much literal text as we can.
11192 Once we have been able to handle whatever type of thing started the
11193 sequence, we return.
11195 Note: we have to be careful with escapes, as they can be both literal
11196 and special, and in the case of \10 and friends, context determines which.
11198 A summary of the code structure is:
11200 switch (first_byte) {
11201 cases for each special:
11202 handle this special;
11205 switch (2nd byte) {
11206 cases for each unambiguous special:
11207 handle this special;
11209 cases for each ambigous special/literal:
11211 if (special) handle here
11213 default: // unambiguously literal:
11216 default: // is a literal char
11219 create EXACTish node for literal;
11220 while (more input and node isn't full) {
11221 switch (input_byte) {
11222 cases for each special;
11223 make sure parse pointer is set so that the next call to
11224 regatom will see this special first
11225 goto loopdone; // EXACTish node terminated by prev. char
11227 append char to EXACTISH node;
11229 get next input byte;
11233 return the generated node;
11235 Specifically there are two separate switches for handling
11236 escape sequences, with the one for handling literal escapes requiring
11237 a dummy entry for all of the special escapes that are actually handled
11240 Returns NULL, setting *flagp to TRYAGAIN if reg() returns NULL with
11242 Returns NULL, setting *flagp to RESTART_UTF8 if the sizing scan needs to be
11244 Otherwise does not return NULL.
11248 S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
11250 regnode *ret = NULL;
11252 char *parse_start = RExC_parse;
11257 GET_RE_DEBUG_FLAGS_DECL;
11259 *flagp = WORST; /* Tentatively. */
11261 DEBUG_PARSE("atom");
11263 PERL_ARGS_ASSERT_REGATOM;
11266 switch ((U8)*RExC_parse) {
11268 RExC_seen_zerolen++;
11269 nextchar(pRExC_state);
11270 if (RExC_flags & RXf_PMf_MULTILINE)
11271 ret = reg_node(pRExC_state, MBOL);
11272 else if (RExC_flags & RXf_PMf_SINGLELINE)
11273 ret = reg_node(pRExC_state, SBOL);
11275 ret = reg_node(pRExC_state, BOL);
11276 Set_Node_Length(ret, 1); /* MJD */
11279 nextchar(pRExC_state);
11281 RExC_seen_zerolen++;
11282 if (RExC_flags & RXf_PMf_MULTILINE)
11283 ret = reg_node(pRExC_state, MEOL);
11284 else if (RExC_flags & RXf_PMf_SINGLELINE)
11285 ret = reg_node(pRExC_state, SEOL);
11287 ret = reg_node(pRExC_state, EOL);
11288 Set_Node_Length(ret, 1); /* MJD */
11291 nextchar(pRExC_state);
11292 if (RExC_flags & RXf_PMf_SINGLELINE)
11293 ret = reg_node(pRExC_state, SANY);
11295 ret = reg_node(pRExC_state, REG_ANY);
11296 *flagp |= HASWIDTH|SIMPLE;
11298 Set_Node_Length(ret, 1); /* MJD */
11302 char * const oregcomp_parse = ++RExC_parse;
11303 ret = regclass(pRExC_state, flagp,depth+1,
11304 FALSE, /* means parse the whole char class */
11305 TRUE, /* allow multi-char folds */
11306 FALSE, /* don't silence non-portable warnings. */
11308 if (*RExC_parse != ']') {
11309 RExC_parse = oregcomp_parse;
11310 vFAIL("Unmatched [");
11313 if (*flagp & RESTART_UTF8)
11315 FAIL2("panic: regclass returned NULL to regatom, flags=%#"UVxf"",
11318 nextchar(pRExC_state);
11319 Set_Node_Length(ret, RExC_parse - oregcomp_parse + 1); /* MJD */
11323 nextchar(pRExC_state);
11324 ret = reg(pRExC_state, 2, &flags,depth+1);
11326 if (flags & TRYAGAIN) {
11327 if (RExC_parse == RExC_end) {
11328 /* Make parent create an empty node if needed. */
11329 *flagp |= TRYAGAIN;
11334 if (flags & RESTART_UTF8) {
11335 *flagp = RESTART_UTF8;
11338 FAIL2("panic: reg returned NULL to regatom, flags=%#"UVxf"",
11341 *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
11345 if (flags & TRYAGAIN) {
11346 *flagp |= TRYAGAIN;
11349 vFAIL("Internal urp");
11350 /* Supposed to be caught earlier. */
11356 vFAIL("Quantifier follows nothing");
11361 This switch handles escape sequences that resolve to some kind
11362 of special regop and not to literal text. Escape sequnces that
11363 resolve to literal text are handled below in the switch marked
11366 Every entry in this switch *must* have a corresponding entry
11367 in the literal escape switch. However, the opposite is not
11368 required, as the default for this switch is to jump to the
11369 literal text handling code.
11371 switch ((U8)*++RExC_parse) {
11372 /* Special Escapes */
11374 RExC_seen_zerolen++;
11375 ret = reg_node(pRExC_state, SBOL);
11377 goto finish_meta_pat;
11379 ret = reg_node(pRExC_state, GPOS);
11380 RExC_seen |= REG_GPOS_SEEN;
11382 goto finish_meta_pat;
11384 RExC_seen_zerolen++;
11385 ret = reg_node(pRExC_state, KEEPS);
11387 /* XXX:dmq : disabling in-place substitution seems to
11388 * be necessary here to avoid cases of memory corruption, as
11389 * with: C<$_="x" x 80; s/x\K/y/> -- rgs
11391 RExC_seen |= REG_LOOKBEHIND_SEEN;
11392 goto finish_meta_pat;
11394 ret = reg_node(pRExC_state, SEOL);
11396 RExC_seen_zerolen++; /* Do not optimize RE away */
11397 goto finish_meta_pat;
11399 ret = reg_node(pRExC_state, EOS);
11401 RExC_seen_zerolen++; /* Do not optimize RE away */
11402 goto finish_meta_pat;
11404 ret = reg_node(pRExC_state, CANY);
11405 RExC_seen |= REG_CANY_SEEN;
11406 *flagp |= HASWIDTH|SIMPLE;
11408 ckWARNdep(RExC_parse+1, "\\C is deprecated");
11410 goto finish_meta_pat;
11412 ret = reg_node(pRExC_state, CLUMP);
11413 *flagp |= HASWIDTH;
11414 goto finish_meta_pat;
11420 arg = ANYOF_WORDCHAR;
11424 RExC_seen_zerolen++;
11425 RExC_seen |= REG_LOOKBEHIND_SEEN;
11426 op = BOUND + get_regex_charset(RExC_flags);
11427 if (op > BOUNDA) { /* /aa is same as /a */
11430 else if (op == BOUNDL) {
11431 RExC_contains_locale = 1;
11433 ret = reg_node(pRExC_state, op);
11434 FLAGS(ret) = get_regex_charset(RExC_flags);
11436 if (! SIZE_ONLY && (U8) *(RExC_parse + 1) == '{') {
11437 /* diag_listed_as: Use "%s" instead of "%s" */
11438 vFAIL("Use \"\\b\\{\" instead of \"\\b{\"");
11440 goto finish_meta_pat;
11442 RExC_seen_zerolen++;
11443 RExC_seen |= REG_LOOKBEHIND_SEEN;
11444 op = NBOUND + get_regex_charset(RExC_flags);
11445 if (op > NBOUNDA) { /* /aa is same as /a */
11448 else if (op == NBOUNDL) {
11449 RExC_contains_locale = 1;
11451 ret = reg_node(pRExC_state, op);
11452 FLAGS(ret) = get_regex_charset(RExC_flags);
11454 if (! SIZE_ONLY && (U8) *(RExC_parse + 1) == '{') {
11455 /* diag_listed_as: Use "%s" instead of "%s" */
11456 vFAIL("Use \"\\B\\{\" instead of \"\\B{\"");
11458 goto finish_meta_pat;
11468 ret = reg_node(pRExC_state, LNBREAK);
11469 *flagp |= HASWIDTH|SIMPLE;
11470 goto finish_meta_pat;
11478 goto join_posix_op_known;
11484 arg = ANYOF_VERTWS;
11486 goto join_posix_op_known;
11496 op = POSIXD + get_regex_charset(RExC_flags);
11497 if (op > POSIXA) { /* /aa is same as /a */
11500 else if (op == POSIXL) {
11501 RExC_contains_locale = 1;
11504 join_posix_op_known:
11507 op += NPOSIXD - POSIXD;
11510 ret = reg_node(pRExC_state, op);
11512 FLAGS(ret) = namedclass_to_classnum(arg);
11515 *flagp |= HASWIDTH|SIMPLE;
11519 nextchar(pRExC_state);
11520 Set_Node_Length(ret, 2); /* MJD */
11526 char* parse_start = RExC_parse - 2;
11531 ret = regclass(pRExC_state, flagp,depth+1,
11532 TRUE, /* means just parse this element */
11533 FALSE, /* don't allow multi-char folds */
11534 FALSE, /* don't silence non-portable warnings.
11535 It would be a bug if these returned
11538 /* regclass() can only return RESTART_UTF8 if multi-char folds
11541 FAIL2("panic: regclass returned NULL to regatom, flags=%#"UVxf"",
11546 Set_Node_Offset(ret, parse_start + 2);
11547 Set_Node_Cur_Length(ret, parse_start);
11548 nextchar(pRExC_state);
11552 /* Handle \N and \N{NAME} with multiple code points here and not
11553 * below because it can be multicharacter. join_exact() will join
11554 * them up later on. Also this makes sure that things like
11555 * /\N{BLAH}+/ and \N{BLAH} being multi char Just Happen. dmq.
11556 * The options to the grok function call causes it to fail if the
11557 * sequence is just a single code point. We then go treat it as
11558 * just another character in the current EXACT node, and hence it
11559 * gets uniform treatment with all the other characters. The
11560 * special treatment for quantifiers is not needed for such single
11561 * character sequences */
11563 if (! grok_bslash_N(pRExC_state, &ret, NULL, flagp, depth, FALSE,
11564 FALSE /* not strict */ )) {
11565 if (*flagp & RESTART_UTF8)
11571 case 'k': /* Handle \k<NAME> and \k'NAME' */
11574 char ch= RExC_parse[1];
11575 if (ch != '<' && ch != '\'' && ch != '{') {
11577 /* diag_listed_as: Sequence \%s... not terminated in regex; marked by <-- HERE in m/%s/ */
11578 vFAIL2("Sequence %.2s... not terminated",parse_start);
11580 /* this pretty much dupes the code for (?P=...) in reg(), if
11581 you change this make sure you change that */
11582 char* name_start = (RExC_parse += 2);
11584 SV *sv_dat = reg_scan_name(pRExC_state,
11585 SIZE_ONLY ? REG_RSN_RETURN_NULL : REG_RSN_RETURN_DATA);
11586 ch= (ch == '<') ? '>' : (ch == '{') ? '}' : '\'';
11587 if (RExC_parse == name_start || *RExC_parse != ch)
11588 /* diag_listed_as: Sequence \%s... not terminated in regex; marked by <-- HERE in m/%s/ */
11589 vFAIL2("Sequence %.3s... not terminated",parse_start);
11592 num = add_data( pRExC_state, STR_WITH_LEN("S"));
11593 RExC_rxi->data->data[num]=(void*)sv_dat;
11594 SvREFCNT_inc_simple_void(sv_dat);
11598 ret = reganode(pRExC_state,
11601 : (ASCII_FOLD_RESTRICTED)
11603 : (AT_LEAST_UNI_SEMANTICS)
11609 *flagp |= HASWIDTH;
11611 /* override incorrect value set in reganode MJD */
11612 Set_Node_Offset(ret, parse_start+1);
11613 Set_Node_Cur_Length(ret, parse_start);
11614 nextchar(pRExC_state);
11620 case '1': case '2': case '3': case '4':
11621 case '5': case '6': case '7': case '8': case '9':
11626 if (*RExC_parse == 'g') {
11630 if (*RExC_parse == '{') {
11634 if (*RExC_parse == '-') {
11638 if (hasbrace && !isDIGIT(*RExC_parse)) {
11639 if (isrel) RExC_parse--;
11641 goto parse_named_seq;
11644 num = S_backref_value(RExC_parse);
11646 vFAIL("Reference to invalid group 0");
11647 else if (num == I32_MAX) {
11648 if (isDIGIT(*RExC_parse))
11649 vFAIL("Reference to nonexistent group");
11651 vFAIL("Unterminated \\g... pattern");
11655 num = RExC_npar - num;
11657 vFAIL("Reference to nonexistent or unclosed group");
11661 num = S_backref_value(RExC_parse);
11662 /* bare \NNN might be backref or octal - if it is larger than or equal
11663 * RExC_npar then it is assumed to be and octal escape.
11664 * Note RExC_npar is +1 from the actual number of parens*/
11665 if (num == I32_MAX || (num > 9 && num >= RExC_npar
11666 && *RExC_parse != '8' && *RExC_parse != '9'))
11668 /* Probably a character specified in octal, e.g. \35 */
11673 /* at this point RExC_parse definitely points to a backref
11676 #ifdef RE_TRACK_PATTERN_OFFSETS
11677 char * const parse_start = RExC_parse - 1; /* MJD */
11679 while (isDIGIT(*RExC_parse))
11682 if (*RExC_parse != '}')
11683 vFAIL("Unterminated \\g{...} pattern");
11687 if (num > (I32)RExC_rx->nparens)
11688 vFAIL("Reference to nonexistent group");
11691 ret = reganode(pRExC_state,
11694 : (ASCII_FOLD_RESTRICTED)
11696 : (AT_LEAST_UNI_SEMANTICS)
11702 *flagp |= HASWIDTH;
11704 /* override incorrect value set in reganode MJD */
11705 Set_Node_Offset(ret, parse_start+1);
11706 Set_Node_Cur_Length(ret, parse_start);
11708 nextchar(pRExC_state);
11713 if (RExC_parse >= RExC_end)
11714 FAIL("Trailing \\");
11717 /* Do not generate "unrecognized" warnings here, we fall
11718 back into the quick-grab loop below */
11725 if (RExC_flags & RXf_PMf_EXTENDED) {
11726 RExC_parse = reg_skipcomment( pRExC_state, RExC_parse );
11727 if (RExC_parse < RExC_end)
11734 parse_start = RExC_parse - 1;
11743 #define MAX_NODE_STRING_SIZE 127
11744 char foldbuf[MAX_NODE_STRING_SIZE+UTF8_MAXBYTES_CASE];
11746 U8 upper_parse = MAX_NODE_STRING_SIZE;
11747 U8 node_type = compute_EXACTish(pRExC_state);
11748 bool next_is_quantifier;
11749 char * oldp = NULL;
11751 /* We can convert EXACTF nodes to EXACTFU if they contain only
11752 * characters that match identically regardless of the target
11753 * string's UTF8ness. The reason to do this is that EXACTF is not
11754 * trie-able, EXACTFU is.
11756 * Similarly, we can convert EXACTFL nodes to EXACTFU if they
11757 * contain only above-Latin1 characters (hence must be in UTF8),
11758 * which don't participate in folds with Latin1-range characters,
11759 * as the latter's folds aren't known until runtime. (We don't
11760 * need to figure this out until pass 2) */
11761 bool maybe_exactfu = PASS2
11762 && (node_type == EXACTF || node_type == EXACTFL);
11764 /* If a folding node contains only code points that don't
11765 * participate in folds, it can be changed into an EXACT node,
11766 * which allows the optimizer more things to look for */
11769 ret = reg_node(pRExC_state, node_type);
11771 /* In pass1, folded, we use a temporary buffer instead of the
11772 * actual node, as the node doesn't exist yet */
11773 s = (SIZE_ONLY && FOLD) ? foldbuf : STRING(ret);
11779 /* We do the EXACTFish to EXACT node only if folding. (And we
11780 * don't need to figure this out until pass 2) */
11781 maybe_exact = FOLD && PASS2;
11783 /* XXX The node can hold up to 255 bytes, yet this only goes to
11784 * 127. I (khw) do not know why. Keeping it somewhat less than
11785 * 255 allows us to not have to worry about overflow due to
11786 * converting to utf8 and fold expansion, but that value is
11787 * 255-UTF8_MAXBYTES_CASE. join_exact() may join adjacent nodes
11788 * split up by this limit into a single one using the real max of
11789 * 255. Even at 127, this breaks under rare circumstances. If
11790 * folding, we do not want to split a node at a character that is a
11791 * non-final in a multi-char fold, as an input string could just
11792 * happen to want to match across the node boundary. The join
11793 * would solve that problem if the join actually happens. But a
11794 * series of more than two nodes in a row each of 127 would cause
11795 * the first join to succeed to get to 254, but then there wouldn't
11796 * be room for the next one, which could at be one of those split
11797 * multi-char folds. I don't know of any fool-proof solution. One
11798 * could back off to end with only a code point that isn't such a
11799 * non-final, but it is possible for there not to be any in the
11801 for (p = RExC_parse - 1;
11802 len < upper_parse && p < RExC_end;
11807 if (RExC_flags & RXf_PMf_EXTENDED)
11808 p = regpatws(pRExC_state, p,
11809 TRUE); /* means recognize comments */
11820 /* Literal Escapes Switch
11822 This switch is meant to handle escape sequences that
11823 resolve to a literal character.
11825 Every escape sequence that represents something
11826 else, like an assertion or a char class, is handled
11827 in the switch marked 'Special Escapes' above in this
11828 routine, but also has an entry here as anything that
11829 isn't explicitly mentioned here will be treated as
11830 an unescaped equivalent literal.
11833 switch ((U8)*++p) {
11834 /* These are all the special escapes. */
11835 case 'A': /* Start assertion */
11836 case 'b': case 'B': /* Word-boundary assertion*/
11837 case 'C': /* Single char !DANGEROUS! */
11838 case 'd': case 'D': /* digit class */
11839 case 'g': case 'G': /* generic-backref, pos assertion */
11840 case 'h': case 'H': /* HORIZWS */
11841 case 'k': case 'K': /* named backref, keep marker */
11842 case 'p': case 'P': /* Unicode property */
11843 case 'R': /* LNBREAK */
11844 case 's': case 'S': /* space class */
11845 case 'v': case 'V': /* VERTWS */
11846 case 'w': case 'W': /* word class */
11847 case 'X': /* eXtended Unicode "combining
11848 character sequence" */
11849 case 'z': case 'Z': /* End of line/string assertion */
11853 /* Anything after here is an escape that resolves to a
11854 literal. (Except digits, which may or may not)
11860 case 'N': /* Handle a single-code point named character. */
11861 /* The options cause it to fail if a multiple code
11862 * point sequence. Handle those in the switch() above
11864 RExC_parse = p + 1;
11865 if (! grok_bslash_N(pRExC_state, NULL, &ender,
11866 flagp, depth, FALSE,
11867 FALSE /* not strict */ ))
11869 if (*flagp & RESTART_UTF8)
11870 FAIL("panic: grok_bslash_N set RESTART_UTF8");
11871 RExC_parse = p = oldp;
11875 if (ender > 0xff) {
11892 ender = ASCII_TO_NATIVE('\033');
11902 const char* error_msg;
11904 bool valid = grok_bslash_o(&p,
11907 TRUE, /* out warnings */
11908 FALSE, /* not strict */
11909 TRUE, /* Output warnings
11914 RExC_parse = p; /* going to die anyway; point
11915 to exact spot of failure */
11919 if (PL_encoding && ender < 0x100) {
11920 goto recode_encoding;
11922 if (ender > 0xff) {
11929 UV result = UV_MAX; /* initialize to erroneous
11931 const char* error_msg;
11933 bool valid = grok_bslash_x(&p,
11936 TRUE, /* out warnings */
11937 FALSE, /* not strict */
11938 TRUE, /* Output warnings
11943 RExC_parse = p; /* going to die anyway; point
11944 to exact spot of failure */
11949 if (PL_encoding && ender < 0x100) {
11950 goto recode_encoding;
11952 if (ender > 0xff) {
11959 ender = grok_bslash_c(*p++, SIZE_ONLY);
11961 case '8': case '9': /* must be a backreference */
11964 case '1': case '2': case '3':case '4':
11965 case '5': case '6': case '7':
11966 /* When we parse backslash escapes there is ambiguity
11967 * between backreferences and octal escapes. Any escape
11968 * from \1 - \9 is a backreference, any multi-digit
11969 * escape which does not start with 0 and which when
11970 * evaluated as decimal could refer to an already
11971 * parsed capture buffer is a backslash. Anything else
11974 * Note this implies that \118 could be interpreted as
11975 * 118 OR as "\11" . "8" depending on whether there
11976 * were 118 capture buffers defined already in the
11979 /* NOTE, RExC_npar is 1 more than the actual number of
11980 * parens we have seen so far, hence the < RExC_npar below. */
11982 if ( !isDIGIT(p[1]) || S_backref_value(p) < RExC_npar)
11983 { /* Not to be treated as an octal constant, go
11991 I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
11993 ender = grok_oct(p, &numlen, &flags, NULL);
11994 if (ender > 0xff) {
11998 if (SIZE_ONLY /* like \08, \178 */
12001 && isDIGIT(*p) && ckWARN(WARN_REGEXP))
12003 reg_warn_non_literal_string(
12005 form_short_octal_warning(p, numlen));
12008 if (PL_encoding && ender < 0x100)
12009 goto recode_encoding;
12012 if (! RExC_override_recoding) {
12013 SV* enc = PL_encoding;
12014 ender = reg_recode((const char)(U8)ender, &enc);
12015 if (!enc && SIZE_ONLY)
12016 ckWARNreg(p, "Invalid escape in the specified encoding");
12022 FAIL("Trailing \\");
12025 if (!SIZE_ONLY&& isALPHANUMERIC(*p)) {
12026 /* Include any { following the alpha to emphasize
12027 * that it could be part of an escape at some point
12029 int len = (isALPHA(*p) && *(p + 1) == '{') ? 2 : 1;
12030 ckWARN3reg(p + len, "Unrecognized escape \\%.*s passed through", len, p);
12032 goto normal_default;
12033 } /* End of switch on '\' */
12036 /* Currently we don't warn when the lbrace is at the start
12037 * of a construct. This catches it in the middle of a
12038 * literal string, or when its the first thing after
12039 * something like "\b" */
12041 && (len || (p > RExC_start && isALPHA_A(*(p -1)))))
12043 ckWARNregdep(p + 1, "Unescaped left brace in regex is deprecated, passed through");
12046 default: /* A literal character */
12048 if (UTF8_IS_START(*p) && UTF) {
12050 ender = utf8n_to_uvchr((U8*)p, RExC_end - p,
12051 &numlen, UTF8_ALLOW_DEFAULT);
12057 } /* End of switch on the literal */
12059 /* Here, have looked at the literal character and <ender>
12060 * contains its ordinal, <p> points to the character after it
12063 if ( RExC_flags & RXf_PMf_EXTENDED)
12064 p = regpatws(pRExC_state, p,
12065 TRUE); /* means recognize comments */
12067 /* If the next thing is a quantifier, it applies to this
12068 * character only, which means that this character has to be in
12069 * its own node and can't just be appended to the string in an
12070 * existing node, so if there are already other characters in
12071 * the node, close the node with just them, and set up to do
12072 * this character again next time through, when it will be the
12073 * only thing in its new node */
12074 if ((next_is_quantifier = (p < RExC_end && ISMULT2(p))) && len)
12080 if (! FOLD /* The simple case, just append the literal */
12081 || (LOC /* Also don't fold for tricky chars under /l */
12082 && is_PROBLEMATIC_LOCALE_FOLD_cp(ender)))
12085 const STRLEN unilen = reguni(pRExC_state, ender, s);
12091 /* The loop increments <len> each time, as all but this
12092 * path (and one other) through it add a single byte to
12093 * the EXACTish node. But this one has changed len to
12094 * be the correct final value, so subtract one to
12095 * cancel out the increment that follows */
12099 REGC((char)ender, s++);
12102 /* Can get here if folding only if is one of the /l
12103 * characters whose fold depends on the locale. The
12104 * occurrence of any of these indicate that we can't
12105 * simplify things */
12107 maybe_exact = FALSE;
12108 maybe_exactfu = FALSE;
12113 /* See comments for join_exact() as to why we fold this
12114 * non-UTF at compile time */
12115 || (node_type == EXACTFU
12116 && ender == LATIN_SMALL_LETTER_SHARP_S)))
12118 /* Here, are folding and are not UTF-8 encoded; therefore
12119 * the character must be in the range 0-255, and is not /l
12120 * (Not /l because we already handled these under /l in
12121 * is_PROBLEMATIC_LOCALE_FOLD_cp */
12122 if (IS_IN_SOME_FOLD_L1(ender)) {
12123 maybe_exact = FALSE;
12125 /* See if the character's fold differs between /d and
12126 * /u. This includes the multi-char fold SHARP S to
12129 && (PL_fold[ender] != PL_fold_latin1[ender]
12130 || ender == LATIN_SMALL_LETTER_SHARP_S
12132 && isARG2_lower_or_UPPER_ARG1('s', ender)
12133 && isARG2_lower_or_UPPER_ARG1('s',
12136 maybe_exactfu = FALSE;
12140 /* Even when folding, we store just the input character, as
12141 * we have an array that finds its fold quickly */
12142 *(s++) = (char) ender;
12144 else { /* FOLD and UTF */
12145 /* Unlike the non-fold case, we do actually have to
12146 * calculate the results here in pass 1. This is for two
12147 * reasons, the folded length may be longer than the
12148 * unfolded, and we have to calculate how many EXACTish
12149 * nodes it will take; and we may run out of room in a node
12150 * in the middle of a potential multi-char fold, and have
12151 * to back off accordingly. (Hence we can't use REGC for
12152 * the simple case just below.) */
12155 if (isASCII(ender)) {
12156 folded = toFOLD(ender);
12157 *(s)++ = (U8) folded;
12162 folded = _to_uni_fold_flags(
12166 FOLD_FLAGS_FULL | ((ASCII_FOLD_RESTRICTED)
12167 ? FOLD_FLAGS_NOMIX_ASCII
12171 /* The loop increments <len> each time, as all but this
12172 * path (and one other) through it add a single byte to
12173 * the EXACTish node. But this one has changed len to
12174 * be the correct final value, so subtract one to
12175 * cancel out the increment that follows */
12176 len += foldlen - 1;
12178 /* If this node only contains non-folding code points so
12179 * far, see if this new one is also non-folding */
12181 if (folded != ender) {
12182 maybe_exact = FALSE;
12185 /* Here the fold is the original; we have to check
12186 * further to see if anything folds to it */
12187 if (_invlist_contains_cp(PL_utf8_foldable,
12190 maybe_exact = FALSE;
12197 if (next_is_quantifier) {
12199 /* Here, the next input is a quantifier, and to get here,
12200 * the current character is the only one in the node.
12201 * Also, here <len> doesn't include the final byte for this
12207 } /* End of loop through literal characters */
12209 /* Here we have either exhausted the input or ran out of room in
12210 * the node. (If we encountered a character that can't be in the
12211 * node, transfer is made directly to <loopdone>, and so we
12212 * wouldn't have fallen off the end of the loop.) In the latter
12213 * case, we artificially have to split the node into two, because
12214 * we just don't have enough space to hold everything. This
12215 * creates a problem if the final character participates in a
12216 * multi-character fold in the non-final position, as a match that
12217 * should have occurred won't, due to the way nodes are matched,
12218 * and our artificial boundary. So back off until we find a non-
12219 * problematic character -- one that isn't at the beginning or
12220 * middle of such a fold. (Either it doesn't participate in any
12221 * folds, or appears only in the final position of all the folds it
12222 * does participate in.) A better solution with far fewer false
12223 * positives, and that would fill the nodes more completely, would
12224 * be to actually have available all the multi-character folds to
12225 * test against, and to back-off only far enough to be sure that
12226 * this node isn't ending with a partial one. <upper_parse> is set
12227 * further below (if we need to reparse the node) to include just
12228 * up through that final non-problematic character that this code
12229 * identifies, so when it is set to less than the full node, we can
12230 * skip the rest of this */
12231 if (FOLD && p < RExC_end && upper_parse == MAX_NODE_STRING_SIZE) {
12233 const STRLEN full_len = len;
12235 assert(len >= MAX_NODE_STRING_SIZE);
12237 /* Here, <s> points to the final byte of the final character.
12238 * Look backwards through the string until find a non-
12239 * problematic character */
12243 /* This has no multi-char folds to non-UTF characters */
12244 if (ASCII_FOLD_RESTRICTED) {
12248 while (--s >= s0 && IS_NON_FINAL_FOLD(*s)) { }
12252 if (! PL_NonL1NonFinalFold) {
12253 PL_NonL1NonFinalFold = _new_invlist_C_array(
12254 NonL1_Perl_Non_Final_Folds_invlist);
12257 /* Point to the first byte of the final character */
12258 s = (char *) utf8_hop((U8 *) s, -1);
12260 while (s >= s0) { /* Search backwards until find
12261 non-problematic char */
12262 if (UTF8_IS_INVARIANT(*s)) {
12264 /* There are no ascii characters that participate
12265 * in multi-char folds under /aa. In EBCDIC, the
12266 * non-ascii invariants are all control characters,
12267 * so don't ever participate in any folds. */
12268 if (ASCII_FOLD_RESTRICTED
12269 || ! IS_NON_FINAL_FOLD(*s))
12274 else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
12275 if (! IS_NON_FINAL_FOLD(TWO_BYTE_UTF8_TO_NATIVE(
12281 else if (! _invlist_contains_cp(
12282 PL_NonL1NonFinalFold,
12283 valid_utf8_to_uvchr((U8 *) s, NULL)))
12288 /* Here, the current character is problematic in that
12289 * it does occur in the non-final position of some
12290 * fold, so try the character before it, but have to
12291 * special case the very first byte in the string, so
12292 * we don't read outside the string */
12293 s = (s == s0) ? s -1 : (char *) utf8_hop((U8 *) s, -1);
12294 } /* End of loop backwards through the string */
12296 /* If there were only problematic characters in the string,
12297 * <s> will point to before s0, in which case the length
12298 * should be 0, otherwise include the length of the
12299 * non-problematic character just found */
12300 len = (s < s0) ? 0 : s - s0 + UTF8SKIP(s);
12303 /* Here, have found the final character, if any, that is
12304 * non-problematic as far as ending the node without splitting
12305 * it across a potential multi-char fold. <len> contains the
12306 * number of bytes in the node up-to and including that
12307 * character, or is 0 if there is no such character, meaning
12308 * the whole node contains only problematic characters. In
12309 * this case, give up and just take the node as-is. We can't
12314 /* If the node ends in an 's' we make sure it stays EXACTF,
12315 * as if it turns into an EXACTFU, it could later get
12316 * joined with another 's' that would then wrongly match
12318 if (maybe_exactfu && isARG2_lower_or_UPPER_ARG1('s', ender))
12320 maybe_exactfu = FALSE;
12324 /* Here, the node does contain some characters that aren't
12325 * problematic. If one such is the final character in the
12326 * node, we are done */
12327 if (len == full_len) {
12330 else if (len + ((UTF) ? UTF8SKIP(s) : 1) == full_len) {
12332 /* If the final character is problematic, but the
12333 * penultimate is not, back-off that last character to
12334 * later start a new node with it */
12339 /* Here, the final non-problematic character is earlier
12340 * in the input than the penultimate character. What we do
12341 * is reparse from the beginning, going up only as far as
12342 * this final ok one, thus guaranteeing that the node ends
12343 * in an acceptable character. The reason we reparse is
12344 * that we know how far in the character is, but we don't
12345 * know how to correlate its position with the input parse.
12346 * An alternate implementation would be to build that
12347 * correlation as we go along during the original parse,
12348 * but that would entail extra work for every node, whereas
12349 * this code gets executed only when the string is too
12350 * large for the node, and the final two characters are
12351 * problematic, an infrequent occurrence. Yet another
12352 * possible strategy would be to save the tail of the
12353 * string, and the next time regatom is called, initialize
12354 * with that. The problem with this is that unless you
12355 * back off one more character, you won't be guaranteed
12356 * regatom will get called again, unless regbranch,
12357 * regpiece ... are also changed. If you do back off that
12358 * extra character, so that there is input guaranteed to
12359 * force calling regatom, you can't handle the case where
12360 * just the first character in the node is acceptable. I
12361 * (khw) decided to try this method which doesn't have that
12362 * pitfall; if performance issues are found, we can do a
12363 * combination of the current approach plus that one */
12369 } /* End of verifying node ends with an appropriate char */
12371 loopdone: /* Jumped to when encounters something that shouldn't be in
12374 /* I (khw) don't know if you can get here with zero length, but the
12375 * old code handled this situation by creating a zero-length EXACT
12376 * node. Might as well be NOTHING instead */
12382 /* If 'maybe_exact' is still set here, means there are no
12383 * code points in the node that participate in folds;
12384 * similarly for 'maybe_exactfu' and code points that match
12385 * differently depending on UTF8ness of the target string
12386 * (for /u), or depending on locale for /l */
12390 else if (maybe_exactfu) {
12394 alloc_maybe_populate_EXACT(pRExC_state, ret, flagp, len, ender,
12395 FALSE /* Don't look to see if could
12396 be turned into an EXACT
12397 node, as we have already
12402 RExC_parse = p - 1;
12403 Set_Node_Cur_Length(ret, parse_start);
12404 nextchar(pRExC_state);
12406 /* len is STRLEN which is unsigned, need to copy to signed */
12409 vFAIL("Internal disaster");
12412 } /* End of label 'defchar:' */
12414 } /* End of giant switch on input character */
12420 S_regpatws(RExC_state_t *pRExC_state, char *p , const bool recognize_comment )
12422 /* Returns the next non-pattern-white space, non-comment character (the
12423 * latter only if 'recognize_comment is true) in the string p, which is
12424 * ended by RExC_end. See also reg_skipcomment */
12425 const char *e = RExC_end;
12427 PERL_ARGS_ASSERT_REGPATWS;
12431 if ((len = is_PATWS_safe(p, e, UTF))) {
12434 else if (recognize_comment && *p == '#') {
12435 p = reg_skipcomment(pRExC_state, p);
12444 S_populate_ANYOF_from_invlist(pTHX_ regnode *node, SV** invlist_ptr)
12446 /* Uses the inversion list '*invlist_ptr' to populate the ANYOF 'node'. It
12447 * sets up the bitmap and any flags, removing those code points from the
12448 * inversion list, setting it to NULL should it become completely empty */
12450 PERL_ARGS_ASSERT_POPULATE_ANYOF_FROM_INVLIST;
12451 assert(PL_regkind[OP(node)] == ANYOF);
12453 ANYOF_BITMAP_ZERO(node);
12454 if (*invlist_ptr) {
12456 /* This gets set if we actually need to modify things */
12457 bool change_invlist = FALSE;
12461 /* Start looking through *invlist_ptr */
12462 invlist_iterinit(*invlist_ptr);
12463 while (invlist_iternext(*invlist_ptr, &start, &end)) {
12467 if (end == UV_MAX && start <= 256) {
12468 ANYOF_FLAGS(node) |= ANYOF_ABOVE_LATIN1_ALL;
12470 else if (end >= 256) {
12471 ANYOF_FLAGS(node) |= ANYOF_UTF8;
12474 /* Quit if are above what we should change */
12479 change_invlist = TRUE;
12481 /* Set all the bits in the range, up to the max that we are doing */
12482 high = (end < 255) ? end : 255;
12483 for (i = start; i <= (int) high; i++) {
12484 if (! ANYOF_BITMAP_TEST(node, i)) {
12485 ANYOF_BITMAP_SET(node, i);
12489 invlist_iterfinish(*invlist_ptr);
12491 /* Done with loop; remove any code points that are in the bitmap from
12492 * *invlist_ptr; similarly for code points above latin1 if we have a
12493 * flag to match all of them anyways */
12494 if (change_invlist) {
12495 _invlist_subtract(*invlist_ptr, PL_Latin1, invlist_ptr);
12497 if (ANYOF_FLAGS(node) & ANYOF_ABOVE_LATIN1_ALL) {
12498 _invlist_intersection(*invlist_ptr, PL_Latin1, invlist_ptr);
12501 /* If have completely emptied it, remove it completely */
12502 if (_invlist_len(*invlist_ptr) == 0) {
12503 SvREFCNT_dec_NN(*invlist_ptr);
12504 *invlist_ptr = NULL;
12509 /* Parse POSIX character classes: [[:foo:]], [[=foo=]], [[.foo.]].
12510 Character classes ([:foo:]) can also be negated ([:^foo:]).
12511 Returns a named class id (ANYOF_XXX) if successful, -1 otherwise.
12512 Equivalence classes ([=foo=]) and composites ([.foo.]) are parsed,
12513 but trigger failures because they are currently unimplemented. */
12515 #define POSIXCC_DONE(c) ((c) == ':')
12516 #define POSIXCC_NOTYET(c) ((c) == '=' || (c) == '.')
12517 #define POSIXCC(c) (POSIXCC_DONE(c) || POSIXCC_NOTYET(c))
12519 PERL_STATIC_INLINE I32
12520 S_regpposixcc(pTHX_ RExC_state_t *pRExC_state, I32 value, const bool strict)
12522 I32 namedclass = OOB_NAMEDCLASS;
12524 PERL_ARGS_ASSERT_REGPPOSIXCC;
12526 if (value == '[' && RExC_parse + 1 < RExC_end &&
12527 /* I smell either [: or [= or [. -- POSIX has been here, right? */
12528 POSIXCC(UCHARAT(RExC_parse)))
12530 const char c = UCHARAT(RExC_parse);
12531 char* const s = RExC_parse++;
12533 while (RExC_parse < RExC_end && UCHARAT(RExC_parse) != c)
12535 if (RExC_parse == RExC_end) {
12538 /* Try to give a better location for the error (than the end of
12539 * the string) by looking for the matching ']' */
12541 while (RExC_parse < RExC_end && UCHARAT(RExC_parse) != ']') {
12544 vFAIL2("Unmatched '%c' in POSIX class", c);
12546 /* Grandfather lone [:, [=, [. */
12550 const char* const t = RExC_parse++; /* skip over the c */
12553 if (UCHARAT(RExC_parse) == ']') {
12554 const char *posixcc = s + 1;
12555 RExC_parse++; /* skip over the ending ] */
12558 const I32 complement = *posixcc == '^' ? *posixcc++ : 0;
12559 const I32 skip = t - posixcc;
12561 /* Initially switch on the length of the name. */
12564 if (memEQ(posixcc, "word", 4)) /* this is not POSIX,
12565 this is the Perl \w
12567 namedclass = ANYOF_WORDCHAR;
12570 /* Names all of length 5. */
12571 /* alnum alpha ascii blank cntrl digit graph lower
12572 print punct space upper */
12573 /* Offset 4 gives the best switch position. */
12574 switch (posixcc[4]) {
12576 if (memEQ(posixcc, "alph", 4)) /* alpha */
12577 namedclass = ANYOF_ALPHA;
12580 if (memEQ(posixcc, "spac", 4)) /* space */
12581 namedclass = ANYOF_PSXSPC;
12584 if (memEQ(posixcc, "grap", 4)) /* graph */
12585 namedclass = ANYOF_GRAPH;
12588 if (memEQ(posixcc, "asci", 4)) /* ascii */
12589 namedclass = ANYOF_ASCII;
12592 if (memEQ(posixcc, "blan", 4)) /* blank */
12593 namedclass = ANYOF_BLANK;
12596 if (memEQ(posixcc, "cntr", 4)) /* cntrl */
12597 namedclass = ANYOF_CNTRL;
12600 if (memEQ(posixcc, "alnu", 4)) /* alnum */
12601 namedclass = ANYOF_ALPHANUMERIC;
12604 if (memEQ(posixcc, "lowe", 4)) /* lower */
12605 namedclass = (FOLD) ? ANYOF_CASED : ANYOF_LOWER;
12606 else if (memEQ(posixcc, "uppe", 4)) /* upper */
12607 namedclass = (FOLD) ? ANYOF_CASED : ANYOF_UPPER;
12610 if (memEQ(posixcc, "digi", 4)) /* digit */
12611 namedclass = ANYOF_DIGIT;
12612 else if (memEQ(posixcc, "prin", 4)) /* print */
12613 namedclass = ANYOF_PRINT;
12614 else if (memEQ(posixcc, "punc", 4)) /* punct */
12615 namedclass = ANYOF_PUNCT;
12620 if (memEQ(posixcc, "xdigit", 6))
12621 namedclass = ANYOF_XDIGIT;
12625 if (namedclass == OOB_NAMEDCLASS)
12627 "POSIX class [:%"UTF8f":] unknown",
12628 UTF8fARG(UTF, t - s - 1, s + 1));
12630 /* The #defines are structured so each complement is +1 to
12631 * the normal one */
12635 assert (posixcc[skip] == ':');
12636 assert (posixcc[skip+1] == ']');
12637 } else if (!SIZE_ONLY) {
12638 /* [[=foo=]] and [[.foo.]] are still future. */
12640 /* adjust RExC_parse so the warning shows after
12641 the class closes */
12642 while (UCHARAT(RExC_parse) && UCHARAT(RExC_parse) != ']')
12644 vFAIL3("POSIX syntax [%c %c] is reserved for future extensions", c, c);
12647 /* Maternal grandfather:
12648 * "[:" ending in ":" but not in ":]" */
12650 vFAIL("Unmatched '[' in POSIX class");
12653 /* Grandfather lone [:, [=, [. */
12663 S_could_it_be_a_POSIX_class(RExC_state_t *pRExC_state)
12665 /* This applies some heuristics at the current parse position (which should
12666 * be at a '[') to see if what follows might be intended to be a [:posix:]
12667 * class. It returns true if it really is a posix class, of course, but it
12668 * also can return true if it thinks that what was intended was a posix
12669 * class that didn't quite make it.
12671 * It will return true for
12673 * [:alphanumerics] (as long as the ] isn't followed immediately by a
12674 * ')' indicating the end of the (?[
12675 * [:any garbage including %^&$ punctuation:]
12677 * This is designed to be called only from S_handle_regex_sets; it could be
12678 * easily adapted to be called from the spot at the beginning of regclass()
12679 * that checks to see in a normal bracketed class if the surrounding []
12680 * have been omitted ([:word:] instead of [[:word:]]). But doing so would
12681 * change long-standing behavior, so I (khw) didn't do that */
12682 char* p = RExC_parse + 1;
12683 char first_char = *p;
12685 PERL_ARGS_ASSERT_COULD_IT_BE_A_POSIX_CLASS;
12687 assert(*(p - 1) == '[');
12689 if (! POSIXCC(first_char)) {
12694 while (p < RExC_end && isWORDCHAR(*p)) p++;
12696 if (p >= RExC_end) {
12700 if (p - RExC_parse > 2 /* Got at least 1 word character */
12701 && (*p == first_char
12702 || (*p == ']' && p + 1 < RExC_end && *(p + 1) != ')')))
12707 p = (char *) memchr(RExC_parse, ']', RExC_end - RExC_parse);
12710 && p - RExC_parse > 2 /* [:] evaluates to colon;
12711 [::] is a bad posix class. */
12712 && first_char == *(p - 1));
12716 S_handle_regex_sets(pTHX_ RExC_state_t *pRExC_state, SV** return_invlist,
12717 I32 *flagp, U32 depth,
12718 char * const oregcomp_parse)
12720 /* Handle the (?[...]) construct to do set operations */
12723 UV start, end; /* End points of code point ranges */
12725 char *save_end, *save_parse;
12730 const bool save_fold = FOLD;
12732 GET_RE_DEBUG_FLAGS_DECL;
12734 PERL_ARGS_ASSERT_HANDLE_REGEX_SETS;
12737 vFAIL("(?[...]) not valid in locale");
12739 RExC_uni_semantics = 1;
12741 /* This will return only an ANYOF regnode, or (unlikely) something smaller
12742 * (such as EXACT). Thus we can skip most everything if just sizing. We
12743 * call regclass to handle '[]' so as to not have to reinvent its parsing
12744 * rules here (throwing away the size it computes each time). And, we exit
12745 * upon an unescaped ']' that isn't one ending a regclass. To do both
12746 * these things, we need to realize that something preceded by a backslash
12747 * is escaped, so we have to keep track of backslashes */
12749 UV depth = 0; /* how many nested (?[...]) constructs */
12751 Perl_ck_warner_d(aTHX_
12752 packWARN(WARN_EXPERIMENTAL__REGEX_SETS),
12753 "The regex_sets feature is experimental" REPORT_LOCATION,
12754 UTF8fARG(UTF, (RExC_parse - RExC_precomp), RExC_precomp),
12756 RExC_end - RExC_start - (RExC_parse - RExC_precomp),
12757 RExC_precomp + (RExC_parse - RExC_precomp)));
12759 while (RExC_parse < RExC_end) {
12760 SV* current = NULL;
12761 RExC_parse = regpatws(pRExC_state, RExC_parse,
12762 TRUE); /* means recognize comments */
12763 switch (*RExC_parse) {
12765 if (RExC_parse[1] == '[') depth++, RExC_parse++;
12770 /* Skip the next byte (which could cause us to end up in
12771 * the middle of a UTF-8 character, but since none of those
12772 * are confusable with anything we currently handle in this
12773 * switch (invariants all), it's safe. We'll just hit the
12774 * default: case next time and keep on incrementing until
12775 * we find one of the invariants we do handle. */
12780 /* If this looks like it is a [:posix:] class, leave the
12781 * parse pointer at the '[' to fool regclass() into
12782 * thinking it is part of a '[[:posix:]]'. That function
12783 * will use strict checking to force a syntax error if it
12784 * doesn't work out to a legitimate class */
12785 bool is_posix_class
12786 = could_it_be_a_POSIX_class(pRExC_state);
12787 if (! is_posix_class) {
12791 /* regclass() can only return RESTART_UTF8 if multi-char
12792 folds are allowed. */
12793 if (!regclass(pRExC_state, flagp,depth+1,
12794 is_posix_class, /* parse the whole char
12795 class only if not a
12797 FALSE, /* don't allow multi-char folds */
12798 TRUE, /* silence non-portable warnings. */
12800 FAIL2("panic: regclass returned NULL to handle_sets, flags=%#"UVxf"",
12803 /* function call leaves parse pointing to the ']', except
12804 * if we faked it */
12805 if (is_posix_class) {
12809 SvREFCNT_dec(current); /* In case it returned something */
12814 if (depth--) break;
12816 if (RExC_parse < RExC_end
12817 && *RExC_parse == ')')
12819 node = reganode(pRExC_state, ANYOF, 0);
12820 RExC_size += ANYOF_SKIP;
12821 nextchar(pRExC_state);
12822 Set_Node_Length(node,
12823 RExC_parse - oregcomp_parse + 1); /* MJD */
12832 FAIL("Syntax error in (?[...])");
12835 /* Pass 2 only after this. Everything in this construct is a
12836 * metacharacter. Operands begin with either a '\' (for an escape
12837 * sequence), or a '[' for a bracketed character class. Any other
12838 * character should be an operator, or parenthesis for grouping. Both
12839 * types of operands are handled by calling regclass() to parse them. It
12840 * is called with a parameter to indicate to return the computed inversion
12841 * list. The parsing here is implemented via a stack. Each entry on the
12842 * stack is a single character representing one of the operators, or the
12843 * '('; or else a pointer to an operand inversion list. */
12845 #define IS_OPERAND(a) (! SvIOK(a))
12847 /* The stack starts empty. It is a syntax error if the first thing parsed
12848 * is a binary operator; everything else is pushed on the stack. When an
12849 * operand is parsed, the top of the stack is examined. If it is a binary
12850 * operator, the item before it should be an operand, and both are replaced
12851 * by the result of doing that operation on the new operand and the one on
12852 * the stack. Thus a sequence of binary operands is reduced to a single
12853 * one before the next one is parsed.
12855 * A unary operator may immediately follow a binary in the input, for
12858 * When an operand is parsed and the top of the stack is a unary operator,
12859 * the operation is performed, and then the stack is rechecked to see if
12860 * this new operand is part of a binary operation; if so, it is handled as
12863 * A '(' is simply pushed on the stack; it is valid only if the stack is
12864 * empty, or the top element of the stack is an operator or another '('
12865 * (for which the parenthesized expression will become an operand). By the
12866 * time the corresponding ')' is parsed everything in between should have
12867 * been parsed and evaluated to a single operand (or else is a syntax
12868 * error), and is handled as a regular operand */
12870 sv_2mortal((SV *)(stack = newAV()));
12872 while (RExC_parse < RExC_end) {
12873 I32 top_index = av_tindex(stack);
12875 SV* current = NULL;
12877 /* Skip white space */
12878 RExC_parse = regpatws(pRExC_state, RExC_parse,
12879 TRUE /* means recognize comments */ );
12880 if (RExC_parse >= RExC_end) {
12881 Perl_croak(aTHX_ "panic: Read past end of '(?[ ])'");
12883 if ((curchar = UCHARAT(RExC_parse)) == ']') {
12890 if (av_tindex(stack) >= 0 /* This makes sure that we can
12891 safely subtract 1 from
12892 RExC_parse in the next clause.
12893 If we have something on the
12894 stack, we have parsed something
12896 && UCHARAT(RExC_parse - 1) == '('
12897 && RExC_parse < RExC_end)
12899 /* If is a '(?', could be an embedded '(?flags:(?[...])'.
12900 * This happens when we have some thing like
12902 * my $thai_or_lao = qr/(?[ \p{Thai} + \p{Lao} ])/;
12904 * qr/(?[ \p{Digit} & $thai_or_lao ])/;
12906 * Here we would be handling the interpolated
12907 * '$thai_or_lao'. We handle this by a recursive call to
12908 * ourselves which returns the inversion list the
12909 * interpolated expression evaluates to. We use the flags
12910 * from the interpolated pattern. */
12911 U32 save_flags = RExC_flags;
12912 const char * const save_parse = ++RExC_parse;
12914 parse_lparen_question_flags(pRExC_state);
12916 if (RExC_parse == save_parse /* Makes sure there was at
12917 least one flag (or this
12918 embedding wasn't compiled)
12920 || RExC_parse >= RExC_end - 4
12921 || UCHARAT(RExC_parse) != ':'
12922 || UCHARAT(++RExC_parse) != '('
12923 || UCHARAT(++RExC_parse) != '?'
12924 || UCHARAT(++RExC_parse) != '[')
12927 /* In combination with the above, this moves the
12928 * pointer to the point just after the first erroneous
12929 * character (or if there are no flags, to where they
12930 * should have been) */
12931 if (RExC_parse >= RExC_end - 4) {
12932 RExC_parse = RExC_end;
12934 else if (RExC_parse != save_parse) {
12935 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
12937 vFAIL("Expecting '(?flags:(?[...'");
12940 (void) handle_regex_sets(pRExC_state, ¤t, flagp,
12941 depth+1, oregcomp_parse);
12943 /* Here, 'current' contains the embedded expression's
12944 * inversion list, and RExC_parse points to the trailing
12945 * ']'; the next character should be the ')' which will be
12946 * paired with the '(' that has been put on the stack, so
12947 * the whole embedded expression reduces to '(operand)' */
12950 RExC_flags = save_flags;
12951 goto handle_operand;
12956 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
12957 vFAIL("Unexpected character");
12960 /* regclass() can only return RESTART_UTF8 if multi-char
12961 folds are allowed. */
12962 if (!regclass(pRExC_state, flagp,depth+1,
12963 TRUE, /* means parse just the next thing */
12964 FALSE, /* don't allow multi-char folds */
12965 FALSE, /* don't silence non-portable warnings. */
12967 FAIL2("panic: regclass returned NULL to handle_sets, flags=%#"UVxf"",
12969 /* regclass() will return with parsing just the \ sequence,
12970 * leaving the parse pointer at the next thing to parse */
12972 goto handle_operand;
12974 case '[': /* Is a bracketed character class */
12976 bool is_posix_class = could_it_be_a_POSIX_class(pRExC_state);
12978 if (! is_posix_class) {
12982 /* regclass() can only return RESTART_UTF8 if multi-char
12983 folds are allowed. */
12984 if(!regclass(pRExC_state, flagp,depth+1,
12985 is_posix_class, /* parse the whole char class
12986 only if not a posix class */
12987 FALSE, /* don't allow multi-char folds */
12988 FALSE, /* don't silence non-portable warnings. */
12990 FAIL2("panic: regclass returned NULL to handle_sets, flags=%#"UVxf"",
12992 /* function call leaves parse pointing to the ']', except if we
12994 if (is_posix_class) {
12998 goto handle_operand;
13007 || ( ! (top_ptr = av_fetch(stack, top_index, FALSE)))
13008 || ! IS_OPERAND(*top_ptr))
13011 vFAIL2("Unexpected binary operator '%c' with no preceding operand", curchar);
13013 av_push(stack, newSVuv(curchar));
13017 av_push(stack, newSVuv(curchar));
13021 if (top_index >= 0) {
13022 top_ptr = av_fetch(stack, top_index, FALSE);
13024 if (IS_OPERAND(*top_ptr)) {
13026 vFAIL("Unexpected '(' with no preceding operator");
13029 av_push(stack, newSVuv(curchar));
13036 || ! (current = av_pop(stack))
13037 || ! IS_OPERAND(current)
13038 || ! (lparen = av_pop(stack))
13039 || IS_OPERAND(lparen)
13040 || SvUV(lparen) != '(')
13042 SvREFCNT_dec(current);
13044 vFAIL("Unexpected ')'");
13047 SvREFCNT_dec_NN(lparen);
13054 /* Here, we have an operand to process, in 'current' */
13056 if (top_index < 0) { /* Just push if stack is empty */
13057 av_push(stack, current);
13060 SV* top = av_pop(stack);
13062 char current_operator;
13064 if (IS_OPERAND(top)) {
13065 SvREFCNT_dec_NN(top);
13066 SvREFCNT_dec_NN(current);
13067 vFAIL("Operand with no preceding operator");
13069 current_operator = (char) SvUV(top);
13070 switch (current_operator) {
13071 case '(': /* Push the '(' back on followed by the new
13073 av_push(stack, top);
13074 av_push(stack, current);
13075 SvREFCNT_inc(top); /* Counters the '_dec' done
13076 just after the 'break', so
13077 it doesn't get wrongly freed
13082 _invlist_invert(current);
13084 /* Unlike binary operators, the top of the stack,
13085 * now that this unary one has been popped off, may
13086 * legally be an operator, and we now have operand
13089 SvREFCNT_dec_NN(top);
13090 goto handle_operand;
13093 prev = av_pop(stack);
13094 _invlist_intersection(prev,
13097 av_push(stack, current);
13102 prev = av_pop(stack);
13103 _invlist_union(prev, current, ¤t);
13104 av_push(stack, current);
13108 prev = av_pop(stack);;
13109 _invlist_subtract(prev, current, ¤t);
13110 av_push(stack, current);
13113 case '^': /* The union minus the intersection */
13119 prev = av_pop(stack);
13120 _invlist_union(prev, current, &u);
13121 _invlist_intersection(prev, current, &i);
13122 /* _invlist_subtract will overwrite current
13123 without freeing what it already contains */
13125 _invlist_subtract(u, i, ¤t);
13126 av_push(stack, current);
13127 SvREFCNT_dec_NN(i);
13128 SvREFCNT_dec_NN(u);
13129 SvREFCNT_dec_NN(element);
13134 Perl_croak(aTHX_ "panic: Unexpected item on '(?[ ])' stack");
13136 SvREFCNT_dec_NN(top);
13137 SvREFCNT_dec(prev);
13141 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
13144 if (av_tindex(stack) < 0 /* Was empty */
13145 || ((final = av_pop(stack)) == NULL)
13146 || ! IS_OPERAND(final)
13147 || av_tindex(stack) >= 0) /* More left on stack */
13149 vFAIL("Incomplete expression within '(?[ ])'");
13152 /* Here, 'final' is the resultant inversion list from evaluating the
13153 * expression. Return it if so requested */
13154 if (return_invlist) {
13155 *return_invlist = final;
13159 /* Otherwise generate a resultant node, based on 'final'. regclass() is
13160 * expecting a string of ranges and individual code points */
13161 invlist_iterinit(final);
13162 result_string = newSVpvs("");
13163 while (invlist_iternext(final, &start, &end)) {
13164 if (start == end) {
13165 Perl_sv_catpvf(aTHX_ result_string, "\\x{%"UVXf"}", start);
13168 Perl_sv_catpvf(aTHX_ result_string, "\\x{%"UVXf"}-\\x{%"UVXf"}",
13173 save_parse = RExC_parse;
13174 RExC_parse = SvPV(result_string, len);
13175 save_end = RExC_end;
13176 RExC_end = RExC_parse + len;
13178 /* We turn off folding around the call, as the class we have constructed
13179 * already has all folding taken into consideration, and we don't want
13180 * regclass() to add to that */
13181 RExC_flags &= ~RXf_PMf_FOLD;
13182 /* regclass() can only return RESTART_UTF8 if multi-char folds are allowed.
13184 node = regclass(pRExC_state, flagp,depth+1,
13185 FALSE, /* means parse the whole char class */
13186 FALSE, /* don't allow multi-char folds */
13187 TRUE, /* silence non-portable warnings. The above may very
13188 well have generated non-portable code points, but
13189 they're valid on this machine */
13192 FAIL2("panic: regclass returned NULL to handle_sets, flags=%#"UVxf,
13195 RExC_flags |= RXf_PMf_FOLD;
13197 RExC_parse = save_parse + 1;
13198 RExC_end = save_end;
13199 SvREFCNT_dec_NN(final);
13200 SvREFCNT_dec_NN(result_string);
13202 nextchar(pRExC_state);
13203 Set_Node_Length(node, RExC_parse - oregcomp_parse + 1); /* MJD */
13209 S_add_above_Latin1_folds(pTHX_ RExC_state_t *pRExC_state, const U8 cp, SV** invlist)
13211 /* This hard-codes the Latin1/above-Latin1 folding rules, so that an
13212 * innocent-looking character class, like /[ks]/i won't have to go out to
13213 * disk to find the possible matches.
13215 * This should be called only for a Latin1-range code points, cp, which is
13216 * known to be involved in a simple fold with other code points above
13217 * Latin1. It would give false results if /aa has been specified.
13218 * Multi-char folds are outside the scope of this, and must be handled
13221 * XXX It would be better to generate these via regen, in case a new
13222 * version of the Unicode standard adds new mappings, though that is not
13223 * really likely, and may be caught by the default: case of the switch
13226 PERL_ARGS_ASSERT_ADD_ABOVE_LATIN1_FOLDS;
13228 assert(HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(cp));
13234 add_cp_to_invlist(*invlist, KELVIN_SIGN);
13238 *invlist = add_cp_to_invlist(*invlist, LATIN_SMALL_LETTER_LONG_S);
13241 *invlist = add_cp_to_invlist(*invlist, GREEK_CAPITAL_LETTER_MU);
13242 *invlist = add_cp_to_invlist(*invlist, GREEK_SMALL_LETTER_MU);
13244 case LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE:
13245 case LATIN_SMALL_LETTER_A_WITH_RING_ABOVE:
13246 *invlist = add_cp_to_invlist(*invlist, ANGSTROM_SIGN);
13248 case LATIN_SMALL_LETTER_Y_WITH_DIAERESIS:
13249 *invlist = add_cp_to_invlist(*invlist,
13250 LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS);
13252 case LATIN_SMALL_LETTER_SHARP_S:
13253 *invlist = add_cp_to_invlist(*invlist, LATIN_CAPITAL_LETTER_SHARP_S);
13256 /* Use deprecated warning to increase the chances of this being
13258 ckWARN2reg_d(RExC_parse, "Perl folding rules are not up-to-date for 0x%02X; please use the perlbug utility to report;", cp);
13263 /* The names of properties whose definitions are not known at compile time are
13264 * stored in this SV, after a constant heading. So if the length has been
13265 * changed since initialization, then there is a run-time definition. */
13266 #define HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION \
13267 (SvCUR(listsv) != initial_listsv_len)
13270 S_regclass(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth,
13271 const bool stop_at_1, /* Just parse the next thing, don't
13272 look for a full character class */
13273 bool allow_multi_folds,
13274 const bool silence_non_portable, /* Don't output warnings
13277 SV** ret_invlist) /* Return an inversion list, not a node */
13279 /* parse a bracketed class specification. Most of these will produce an
13280 * ANYOF node; but something like [a] will produce an EXACT node; [aA], an
13281 * EXACTFish node; [[:ascii:]], a POSIXA node; etc. It is more complex
13282 * under /i with multi-character folds: it will be rewritten following the
13283 * paradigm of this example, where the <multi-fold>s are characters which
13284 * fold to multiple character sequences:
13285 * /[abc\x{multi-fold1}def\x{multi-fold2}ghi]/i
13286 * gets effectively rewritten as:
13287 * /(?:\x{multi-fold1}|\x{multi-fold2}|[abcdefghi]/i
13288 * reg() gets called (recursively) on the rewritten version, and this
13289 * function will return what it constructs. (Actually the <multi-fold>s
13290 * aren't physically removed from the [abcdefghi], it's just that they are
13291 * ignored in the recursion by means of a flag:
13292 * <RExC_in_multi_char_class>.)
13294 * ANYOF nodes contain a bit map for the first 256 characters, with the
13295 * corresponding bit set if that character is in the list. For characters
13296 * above 255, a range list or swash is used. There are extra bits for \w,
13297 * etc. in locale ANYOFs, as what these match is not determinable at
13300 * Returns NULL, setting *flagp to RESTART_UTF8 if the sizing scan needs
13301 * to be restarted. This can only happen if ret_invlist is non-NULL.
13304 UV prevvalue = OOB_UNICODE, save_prevvalue = OOB_UNICODE;
13306 UV value = OOB_UNICODE, save_value = OOB_UNICODE;
13309 IV namedclass = OOB_NAMEDCLASS;
13310 char *rangebegin = NULL;
13311 bool need_class = 0;
13313 STRLEN initial_listsv_len = 0; /* Kind of a kludge to see if it is more
13314 than just initialized. */
13315 SV* properties = NULL; /* Code points that match \p{} \P{} */
13316 SV* posixes = NULL; /* Code points that match classes like [:word:],
13317 extended beyond the Latin1 range. These have to
13318 be kept separate from other code points for much
13319 of this function because their handling is
13320 different under /i, and for most classes under
13322 SV* nposixes = NULL; /* Similarly for [:^word:]. These are kept
13323 separate for a while from the non-complemented
13324 versions because of complications with /d
13326 UV element_count = 0; /* Number of distinct elements in the class.
13327 Optimizations may be possible if this is tiny */
13328 AV * multi_char_matches = NULL; /* Code points that fold to more than one
13329 character; used under /i */
13331 char * stop_ptr = RExC_end; /* where to stop parsing */
13332 const bool skip_white = cBOOL(ret_invlist); /* ignore unescaped white
13334 const bool strict = cBOOL(ret_invlist); /* Apply strict parsing rules? */
13336 /* Unicode properties are stored in a swash; this holds the current one
13337 * being parsed. If this swash is the only above-latin1 component of the
13338 * character class, an optimization is to pass it directly on to the
13339 * execution engine. Otherwise, it is set to NULL to indicate that there
13340 * are other things in the class that have to be dealt with at execution
13342 SV* swash = NULL; /* Code points that match \p{} \P{} */
13344 /* Set if a component of this character class is user-defined; just passed
13345 * on to the engine */
13346 bool has_user_defined_property = FALSE;
13348 /* inversion list of code points this node matches only when the target
13349 * string is in UTF-8. (Because is under /d) */
13350 SV* depends_list = NULL;
13352 /* Inversion list of code points this node matches regardless of things
13353 * like locale, folding, utf8ness of the target string */
13354 SV* cp_list = NULL;
13356 /* Like cp_list, but code points on this list need to be checked for things
13357 * that fold to/from them under /i */
13358 SV* cp_foldable_list = NULL;
13360 /* Like cp_list, but code points on this list are valid only when the
13361 * runtime locale is UTF-8 */
13362 SV* only_utf8_locale_list = NULL;
13365 /* In a range, counts how many 0-2 of the ends of it came from literals,
13366 * not escapes. Thus we can tell if 'A' was input vs \x{C1} */
13367 UV literal_endpoint = 0;
13369 bool invert = FALSE; /* Is this class to be complemented */
13371 bool warn_super = ALWAYS_WARN_SUPER;
13373 regnode * const orig_emit = RExC_emit; /* Save the original RExC_emit in
13374 case we need to change the emitted regop to an EXACT. */
13375 const char * orig_parse = RExC_parse;
13376 const SSize_t orig_size = RExC_size;
13377 bool posixl_matches_all = FALSE; /* Does /l class have both e.g. \W,\w ? */
13378 GET_RE_DEBUG_FLAGS_DECL;
13380 PERL_ARGS_ASSERT_REGCLASS;
13382 PERL_UNUSED_ARG(depth);
13385 DEBUG_PARSE("clas");
13387 /* Assume we are going to generate an ANYOF node. */
13388 ret = reganode(pRExC_state, ANYOF, 0);
13391 RExC_size += ANYOF_SKIP;
13392 listsv = &PL_sv_undef; /* For code scanners: listsv always non-NULL. */
13395 ANYOF_FLAGS(ret) = 0;
13397 RExC_emit += ANYOF_SKIP;
13398 listsv = newSVpvs_flags("# comment\n", SVs_TEMP);
13399 initial_listsv_len = SvCUR(listsv);
13400 SvTEMP_off(listsv); /* Grr, TEMPs and mortals are conflated. */
13404 RExC_parse = regpatws(pRExC_state, RExC_parse,
13405 FALSE /* means don't recognize comments */ );
13408 if (UCHARAT(RExC_parse) == '^') { /* Complement of range. */
13411 allow_multi_folds = FALSE;
13414 RExC_parse = regpatws(pRExC_state, RExC_parse,
13415 FALSE /* means don't recognize comments */ );
13419 /* Check that they didn't say [:posix:] instead of [[:posix:]] */
13420 if (!SIZE_ONLY && RExC_parse < RExC_end && POSIXCC(UCHARAT(RExC_parse))) {
13421 const char *s = RExC_parse;
13422 const char c = *s++;
13424 while (isWORDCHAR(*s))
13426 if (*s && c == *s && s[1] == ']') {
13427 SAVEFREESV(RExC_rx_sv);
13429 "POSIX syntax [%c %c] belongs inside character classes",
13431 (void)ReREFCNT_inc(RExC_rx_sv);
13435 /* If the caller wants us to just parse a single element, accomplish this
13436 * by faking the loop ending condition */
13437 if (stop_at_1 && RExC_end > RExC_parse) {
13438 stop_ptr = RExC_parse + 1;
13441 /* allow 1st char to be ']' (allowing it to be '-' is dealt with later) */
13442 if (UCHARAT(RExC_parse) == ']')
13443 goto charclassloop;
13447 if (RExC_parse >= stop_ptr) {
13452 RExC_parse = regpatws(pRExC_state, RExC_parse,
13453 FALSE /* means don't recognize comments */ );
13456 if (UCHARAT(RExC_parse) == ']') {
13462 namedclass = OOB_NAMEDCLASS; /* initialize as illegal */
13463 save_value = value;
13464 save_prevvalue = prevvalue;
13467 rangebegin = RExC_parse;
13471 value = utf8n_to_uvchr((U8*)RExC_parse,
13472 RExC_end - RExC_parse,
13473 &numlen, UTF8_ALLOW_DEFAULT);
13474 RExC_parse += numlen;
13477 value = UCHARAT(RExC_parse++);
13480 && RExC_parse < RExC_end
13481 && POSIXCC(UCHARAT(RExC_parse)))
13483 namedclass = regpposixcc(pRExC_state, value, strict);
13485 else if (value == '\\') {
13487 value = utf8n_to_uvchr((U8*)RExC_parse,
13488 RExC_end - RExC_parse,
13489 &numlen, UTF8_ALLOW_DEFAULT);
13490 RExC_parse += numlen;
13493 value = UCHARAT(RExC_parse++);
13495 /* Some compilers cannot handle switching on 64-bit integer
13496 * values, therefore value cannot be an UV. Yes, this will
13497 * be a problem later if we want switch on Unicode.
13498 * A similar issue a little bit later when switching on
13499 * namedclass. --jhi */
13501 /* If the \ is escaping white space when white space is being
13502 * skipped, it means that that white space is wanted literally, and
13503 * is already in 'value'. Otherwise, need to translate the escape
13504 * into what it signifies. */
13505 if (! skip_white || ! is_PATWS_cp(value)) switch ((I32)value) {
13507 case 'w': namedclass = ANYOF_WORDCHAR; break;
13508 case 'W': namedclass = ANYOF_NWORDCHAR; break;
13509 case 's': namedclass = ANYOF_SPACE; break;
13510 case 'S': namedclass = ANYOF_NSPACE; break;
13511 case 'd': namedclass = ANYOF_DIGIT; break;
13512 case 'D': namedclass = ANYOF_NDIGIT; break;
13513 case 'v': namedclass = ANYOF_VERTWS; break;
13514 case 'V': namedclass = ANYOF_NVERTWS; break;
13515 case 'h': namedclass = ANYOF_HORIZWS; break;
13516 case 'H': namedclass = ANYOF_NHORIZWS; break;
13517 case 'N': /* Handle \N{NAME} in class */
13519 /* We only pay attention to the first char of
13520 multichar strings being returned. I kinda wonder
13521 if this makes sense as it does change the behaviour
13522 from earlier versions, OTOH that behaviour was broken
13524 if (! grok_bslash_N(pRExC_state, NULL, &value, flagp, depth,
13525 TRUE, /* => charclass */
13528 if (*flagp & RESTART_UTF8)
13529 FAIL("panic: grok_bslash_N set RESTART_UTF8");
13539 /* We will handle any undefined properties ourselves */
13540 U8 swash_init_flags = _CORE_SWASH_INIT_RETURN_IF_UNDEF
13541 /* And we actually would prefer to get
13542 * the straight inversion list of the
13543 * swash, since we will be accessing it
13544 * anyway, to save a little time */
13545 |_CORE_SWASH_INIT_ACCEPT_INVLIST;
13547 if (RExC_parse >= RExC_end)
13548 vFAIL2("Empty \\%c{}", (U8)value);
13549 if (*RExC_parse == '{') {
13550 const U8 c = (U8)value;
13551 e = strchr(RExC_parse++, '}');
13553 vFAIL2("Missing right brace on \\%c{}", c);
13554 while (isSPACE(*RExC_parse))
13556 if (e == RExC_parse)
13557 vFAIL2("Empty \\%c{}", c);
13558 n = e - RExC_parse;
13559 while (isSPACE(*(RExC_parse + n - 1)))
13570 if (UCHARAT(RExC_parse) == '^') {
13573 /* toggle. (The rhs xor gets the single bit that
13574 * differs between P and p; the other xor inverts just
13576 value ^= 'P' ^ 'p';
13578 while (isSPACE(*RExC_parse)) {
13583 /* Try to get the definition of the property into
13584 * <invlist>. If /i is in effect, the effective property
13585 * will have its name be <__NAME_i>. The design is
13586 * discussed in commit
13587 * 2f833f5208e26b208886e51e09e2c072b5eabb46 */
13588 name = savepv(Perl_form(aTHX_
13590 (FOLD) ? "__" : "",
13596 /* Look up the property name, and get its swash and
13597 * inversion list, if the property is found */
13599 SvREFCNT_dec_NN(swash);
13601 swash = _core_swash_init("utf8", name, &PL_sv_undef,
13604 NULL, /* No inversion list */
13607 if (! swash || ! (invlist = _get_swash_invlist(swash))) {
13608 HV* curpkg = (IN_PERL_COMPILETIME)
13610 : CopSTASH(PL_curcop);
13612 SvREFCNT_dec_NN(swash);
13616 /* Here didn't find it. It could be a user-defined
13617 * property that will be available at run-time. If we
13618 * accept only compile-time properties, is an error;
13619 * otherwise add it to the list for run-time look up */
13621 RExC_parse = e + 1;
13623 "Property '%"UTF8f"' is unknown",
13624 UTF8fARG(UTF, n, name));
13627 /* If the property name doesn't already have a package
13628 * name, add the current one to it so that it can be
13629 * referred to outside it. [perl #121777] */
13630 if (curpkg && ! instr(name, "::")) {
13631 char* pkgname = HvNAME(curpkg);
13632 if (strNE(pkgname, "main")) {
13633 char* full_name = Perl_form(aTHX_
13637 n = strlen(full_name);
13639 name = savepvn(full_name, n);
13642 Perl_sv_catpvf(aTHX_ listsv, "%cutf8::%"UTF8f"\n",
13643 (value == 'p' ? '+' : '!'),
13644 UTF8fARG(UTF, n, name));
13645 has_user_defined_property = TRUE;
13647 /* We don't know yet, so have to assume that the
13648 * property could match something in the Latin1 range,
13649 * hence something that isn't utf8. Note that this
13650 * would cause things in <depends_list> to match
13651 * inappropriately, except that any \p{}, including
13652 * this one forces Unicode semantics, which means there
13653 * is no <depends_list> */
13654 ANYOF_FLAGS(ret) |= ANYOF_NONBITMAP_NON_UTF8;
13658 /* Here, did get the swash and its inversion list. If
13659 * the swash is from a user-defined property, then this
13660 * whole character class should be regarded as such */
13661 if (swash_init_flags
13662 & _CORE_SWASH_INIT_USER_DEFINED_PROPERTY)
13664 has_user_defined_property = TRUE;
13667 /* We warn on matching an above-Unicode code point
13668 * if the match would return true, except don't
13669 * warn for \p{All}, which has exactly one element
13671 (_invlist_contains_cp(invlist, 0x110000)
13672 && (! (_invlist_len(invlist) == 1
13673 && *invlist_array(invlist) == 0)))
13679 /* Invert if asking for the complement */
13680 if (value == 'P') {
13681 _invlist_union_complement_2nd(properties,
13685 /* The swash can't be used as-is, because we've
13686 * inverted things; delay removing it to here after
13687 * have copied its invlist above */
13688 SvREFCNT_dec_NN(swash);
13692 _invlist_union(properties, invlist, &properties);
13697 RExC_parse = e + 1;
13698 namedclass = ANYOF_UNIPROP; /* no official name, but it's
13701 /* \p means they want Unicode semantics */
13702 RExC_uni_semantics = 1;
13705 case 'n': value = '\n'; break;
13706 case 'r': value = '\r'; break;
13707 case 't': value = '\t'; break;
13708 case 'f': value = '\f'; break;
13709 case 'b': value = '\b'; break;
13710 case 'e': value = ASCII_TO_NATIVE('\033');break;
13711 case 'a': value = '\a'; break;
13713 RExC_parse--; /* function expects to be pointed at the 'o' */
13715 const char* error_msg;
13716 bool valid = grok_bslash_o(&RExC_parse,
13719 SIZE_ONLY, /* warnings in pass
13722 silence_non_portable,
13728 if (PL_encoding && value < 0x100) {
13729 goto recode_encoding;
13733 RExC_parse--; /* function expects to be pointed at the 'x' */
13735 const char* error_msg;
13736 bool valid = grok_bslash_x(&RExC_parse,
13739 TRUE, /* Output warnings */
13741 silence_non_portable,
13747 if (PL_encoding && value < 0x100)
13748 goto recode_encoding;
13751 value = grok_bslash_c(*RExC_parse++, SIZE_ONLY);
13753 case '0': case '1': case '2': case '3': case '4':
13754 case '5': case '6': case '7':
13756 /* Take 1-3 octal digits */
13757 I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
13758 numlen = (strict) ? 4 : 3;
13759 value = grok_oct(--RExC_parse, &numlen, &flags, NULL);
13760 RExC_parse += numlen;
13763 RExC_parse += (UTF) ? UTF8SKIP(RExC_parse) : 1;
13764 vFAIL("Need exactly 3 octal digits");
13766 else if (! SIZE_ONLY /* like \08, \178 */
13768 && RExC_parse < RExC_end
13769 && isDIGIT(*RExC_parse)
13770 && ckWARN(WARN_REGEXP))
13772 SAVEFREESV(RExC_rx_sv);
13773 reg_warn_non_literal_string(
13775 form_short_octal_warning(RExC_parse, numlen));
13776 (void)ReREFCNT_inc(RExC_rx_sv);
13779 if (PL_encoding && value < 0x100)
13780 goto recode_encoding;
13784 if (! RExC_override_recoding) {
13785 SV* enc = PL_encoding;
13786 value = reg_recode((const char)(U8)value, &enc);
13789 vFAIL("Invalid escape in the specified encoding");
13791 else if (SIZE_ONLY) {
13792 ckWARNreg(RExC_parse,
13793 "Invalid escape in the specified encoding");
13799 /* Allow \_ to not give an error */
13800 if (!SIZE_ONLY && isWORDCHAR(value) && value != '_') {
13802 vFAIL2("Unrecognized escape \\%c in character class",
13806 SAVEFREESV(RExC_rx_sv);
13807 ckWARN2reg(RExC_parse,
13808 "Unrecognized escape \\%c in character class passed through",
13810 (void)ReREFCNT_inc(RExC_rx_sv);
13814 } /* End of switch on char following backslash */
13815 } /* end of handling backslash escape sequences */
13818 literal_endpoint++;
13821 /* Here, we have the current token in 'value' */
13823 if (namedclass > OOB_NAMEDCLASS) { /* this is a named class \blah */
13826 /* a bad range like a-\d, a-[:digit:]. The '-' is taken as a
13827 * literal, as is the character that began the false range, i.e.
13828 * the 'a' in the examples */
13831 const int w = (RExC_parse >= rangebegin)
13832 ? RExC_parse - rangebegin
13836 "False [] range \"%"UTF8f"\"",
13837 UTF8fARG(UTF, w, rangebegin));
13840 SAVEFREESV(RExC_rx_sv); /* in case of fatal warnings */
13841 ckWARN2reg(RExC_parse,
13842 "False [] range \"%"UTF8f"\"",
13843 UTF8fARG(UTF, w, rangebegin));
13844 (void)ReREFCNT_inc(RExC_rx_sv);
13845 cp_list = add_cp_to_invlist(cp_list, '-');
13846 cp_foldable_list = add_cp_to_invlist(cp_foldable_list,
13851 range = 0; /* this was not a true range */
13852 element_count += 2; /* So counts for three values */
13855 classnum = namedclass_to_classnum(namedclass);
13857 if (LOC && namedclass < ANYOF_POSIXL_MAX
13858 #ifndef HAS_ISASCII
13859 && classnum != _CC_ASCII
13862 /* What the Posix classes (like \w, [:space:]) match in locale
13863 * isn't knowable under locale until actual match time. Room
13864 * must be reserved (one time per outer bracketed class) to
13865 * store such classes. The space will contain a bit for each
13866 * named class that is to be matched against. This isn't
13867 * needed for \p{} and pseudo-classes, as they are not affected
13868 * by locale, and hence are dealt with separately */
13869 if (! need_class) {
13872 RExC_size += ANYOF_POSIXL_SKIP - ANYOF_SKIP;
13875 RExC_emit += ANYOF_POSIXL_SKIP - ANYOF_SKIP;
13877 ANYOF_FLAGS(ret) |= ANYOF_POSIXL;
13878 ANYOF_POSIXL_ZERO(ret);
13881 /* Coverity thinks it is possible for this to be negative; both
13882 * jhi and khw think it's not, but be safer */
13883 assert(! (ANYOF_FLAGS(ret) & ANYOF_POSIXL)
13884 || (namedclass + ((namedclass % 2) ? -1 : 1)) >= 0);
13886 /* See if it already matches the complement of this POSIX
13888 if ((ANYOF_FLAGS(ret) & ANYOF_POSIXL)
13889 && ANYOF_POSIXL_TEST(ret, namedclass + ((namedclass % 2)
13893 posixl_matches_all = TRUE;
13894 break; /* No need to continue. Since it matches both
13895 e.g., \w and \W, it matches everything, and the
13896 bracketed class can be optimized into qr/./s */
13899 /* Add this class to those that should be checked at runtime */
13900 ANYOF_POSIXL_SET(ret, namedclass);
13902 /* The above-Latin1 characters are not subject to locale rules.
13903 * Just add them, in the second pass, to the
13904 * unconditionally-matched list */
13906 SV* scratch_list = NULL;
13908 /* Get the list of the above-Latin1 code points this
13910 _invlist_intersection_maybe_complement_2nd(PL_AboveLatin1,
13911 PL_XPosix_ptrs[classnum],
13913 /* Odd numbers are complements, like
13914 * NDIGIT, NASCII, ... */
13915 namedclass % 2 != 0,
13917 /* Checking if 'cp_list' is NULL first saves an extra
13918 * clone. Its reference count will be decremented at the
13919 * next union, etc, or if this is the only instance, at the
13920 * end of the routine */
13922 cp_list = scratch_list;
13925 _invlist_union(cp_list, scratch_list, &cp_list);
13926 SvREFCNT_dec_NN(scratch_list);
13928 continue; /* Go get next character */
13931 else if (! SIZE_ONLY) {
13933 /* Here, not in pass1 (in that pass we skip calculating the
13934 * contents of this class), and is /l, or is a POSIX class for
13935 * which /l doesn't matter (or is a Unicode property, which is
13936 * skipped here). */
13937 if (namedclass >= ANYOF_POSIXL_MAX) { /* If a special class */
13938 if (namedclass != ANYOF_UNIPROP) { /* UNIPROP = \p and \P */
13940 /* Here, should be \h, \H, \v, or \V. None of /d, /i
13941 * nor /l make a difference in what these match,
13942 * therefore we just add what they match to cp_list. */
13943 if (classnum != _CC_VERTSPACE) {
13944 assert( namedclass == ANYOF_HORIZWS
13945 || namedclass == ANYOF_NHORIZWS);
13947 /* It turns out that \h is just a synonym for
13949 classnum = _CC_BLANK;
13952 _invlist_union_maybe_complement_2nd(
13954 PL_XPosix_ptrs[classnum],
13955 namedclass % 2 != 0, /* Complement if odd
13956 (NHORIZWS, NVERTWS)
13961 else { /* Garden variety class. If is NASCII, NDIGIT, ...
13962 complement and use nposixes */
13963 SV** posixes_ptr = namedclass % 2 == 0
13966 SV** source_ptr = &PL_XPosix_ptrs[classnum];
13967 _invlist_union_maybe_complement_2nd(
13970 namedclass % 2 != 0,
13973 continue; /* Go get next character */
13975 } /* end of namedclass \blah */
13977 /* Here, we have a single value. If 'range' is set, it is the ending
13978 * of a range--check its validity. Later, we will handle each
13979 * individual code point in the range. If 'range' isn't set, this
13980 * could be the beginning of a range, so check for that by looking
13981 * ahead to see if the next real character to be processed is the range
13982 * indicator--the minus sign */
13985 RExC_parse = regpatws(pRExC_state, RExC_parse,
13986 FALSE /* means don't recognize comments */ );
13990 if (prevvalue > value) /* b-a */ {
13991 const int w = RExC_parse - rangebegin;
13993 "Invalid [] range \"%"UTF8f"\"",
13994 UTF8fARG(UTF, w, rangebegin));
13995 range = 0; /* not a valid range */
13999 prevvalue = value; /* save the beginning of the potential range */
14000 if (! stop_at_1 /* Can't be a range if parsing just one thing */
14001 && *RExC_parse == '-')
14003 char* next_char_ptr = RExC_parse + 1;
14004 if (skip_white) { /* Get the next real char after the '-' */
14005 next_char_ptr = regpatws(pRExC_state,
14007 FALSE); /* means don't recognize
14011 /* If the '-' is at the end of the class (just before the ']',
14012 * it is a literal minus; otherwise it is a range */
14013 if (next_char_ptr < RExC_end && *next_char_ptr != ']') {
14014 RExC_parse = next_char_ptr;
14016 /* a bad range like \w-, [:word:]- ? */
14017 if (namedclass > OOB_NAMEDCLASS) {
14018 if (strict || ckWARN(WARN_REGEXP)) {
14020 RExC_parse >= rangebegin ?
14021 RExC_parse - rangebegin : 0;
14023 vFAIL4("False [] range \"%*.*s\"",
14028 "False [] range \"%*.*s\"",
14033 cp_list = add_cp_to_invlist(cp_list, '-');
14037 range = 1; /* yeah, it's a range! */
14038 continue; /* but do it the next time */
14043 /* Here, <prevvalue> is the beginning of the range, if any; or <value>
14046 /* non-Latin1 code point implies unicode semantics. Must be set in
14047 * pass1 so is there for the whole of pass 2 */
14049 RExC_uni_semantics = 1;
14052 /* Ready to process either the single value, or the completed range.
14053 * For single-valued non-inverted ranges, we consider the possibility
14054 * of multi-char folds. (We made a conscious decision to not do this
14055 * for the other cases because it can often lead to non-intuitive
14056 * results. For example, you have the peculiar case that:
14057 * "s s" =~ /^[^\xDF]+$/i => Y
14058 * "ss" =~ /^[^\xDF]+$/i => N
14060 * See [perl #89750] */
14061 if (FOLD && allow_multi_folds && value == prevvalue) {
14062 if (value == LATIN_SMALL_LETTER_SHARP_S
14063 || (value > 255 && _invlist_contains_cp(PL_HasMultiCharFold,
14066 /* Here <value> is indeed a multi-char fold. Get what it is */
14068 U8 foldbuf[UTF8_MAXBYTES_CASE];
14071 UV folded = _to_uni_fold_flags(
14075 FOLD_FLAGS_FULL | (ASCII_FOLD_RESTRICTED
14076 ? FOLD_FLAGS_NOMIX_ASCII
14080 /* Here, <folded> should be the first character of the
14081 * multi-char fold of <value>, with <foldbuf> containing the
14082 * whole thing. But, if this fold is not allowed (because of
14083 * the flags), <fold> will be the same as <value>, and should
14084 * be processed like any other character, so skip the special
14086 if (folded != value) {
14088 /* Skip if we are recursed, currently parsing the class
14089 * again. Otherwise add this character to the list of
14090 * multi-char folds. */
14091 if (! RExC_in_multi_char_class) {
14092 AV** this_array_ptr;
14094 STRLEN cp_count = utf8_length(foldbuf,
14095 foldbuf + foldlen);
14096 SV* multi_fold = sv_2mortal(newSVpvs(""));
14098 Perl_sv_catpvf(aTHX_ multi_fold, "\\x{%"UVXf"}", value);
14101 if (! multi_char_matches) {
14102 multi_char_matches = newAV();
14105 /* <multi_char_matches> is actually an array of arrays.
14106 * There will be one or two top-level elements: [2],
14107 * and/or [3]. The [2] element is an array, each
14108 * element thereof is a character which folds to TWO
14109 * characters; [3] is for folds to THREE characters.
14110 * (Unicode guarantees a maximum of 3 characters in any
14111 * fold.) When we rewrite the character class below,
14112 * we will do so such that the longest folds are
14113 * written first, so that it prefers the longest
14114 * matching strings first. This is done even if it
14115 * turns out that any quantifier is non-greedy, out of
14116 * programmer laziness. Tom Christiansen has agreed
14117 * that this is ok. This makes the test for the
14118 * ligature 'ffi' come before the test for 'ff' */
14119 if (av_exists(multi_char_matches, cp_count)) {
14120 this_array_ptr = (AV**) av_fetch(multi_char_matches,
14122 this_array = *this_array_ptr;
14125 this_array = newAV();
14126 av_store(multi_char_matches, cp_count,
14129 av_push(this_array, multi_fold);
14132 /* This element should not be processed further in this
14135 value = save_value;
14136 prevvalue = save_prevvalue;
14142 /* Deal with this element of the class */
14145 cp_foldable_list = _add_range_to_invlist(cp_foldable_list,
14148 SV* this_range = _new_invlist(1);
14149 _append_range_to_invlist(this_range, prevvalue, value);
14151 /* In EBCDIC, the ranges 'A-Z' and 'a-z' are each not contiguous.
14152 * If this range was specified using something like 'i-j', we want
14153 * to include only the 'i' and the 'j', and not anything in
14154 * between, so exclude non-ASCII, non-alphabetics from it.
14155 * However, if the range was specified with something like
14156 * [\x89-\x91] or [\x89-j], all code points within it should be
14157 * included. literal_endpoint==2 means both ends of the range used
14158 * a literal character, not \x{foo} */
14159 if (literal_endpoint == 2
14160 && ((prevvalue >= 'a' && value <= 'z')
14161 || (prevvalue >= 'A' && value <= 'Z')))
14163 _invlist_intersection(this_range, PL_XPosix_ptrs[_CC_ASCII],
14166 /* Since this above only contains ascii, the intersection of it
14167 * with anything will still yield only ascii */
14168 _invlist_intersection(this_range, PL_XPosix_ptrs[_CC_ALPHA],
14171 _invlist_union(cp_foldable_list, this_range, &cp_foldable_list);
14172 literal_endpoint = 0;
14176 range = 0; /* this range (if it was one) is done now */
14177 } /* End of loop through all the text within the brackets */
14179 /* If anything in the class expands to more than one character, we have to
14180 * deal with them by building up a substitute parse string, and recursively
14181 * calling reg() on it, instead of proceeding */
14182 if (multi_char_matches) {
14183 SV * substitute_parse = newSVpvn_flags("?:", 2, SVs_TEMP);
14186 char *save_end = RExC_end;
14187 char *save_parse = RExC_parse;
14188 bool first_time = TRUE; /* First multi-char occurrence doesn't get
14193 #if 0 /* Have decided not to deal with multi-char folds in inverted classes,
14194 because too confusing */
14196 sv_catpv(substitute_parse, "(?:");
14200 /* Look at the longest folds first */
14201 for (cp_count = av_tindex(multi_char_matches); cp_count > 0; cp_count--) {
14203 if (av_exists(multi_char_matches, cp_count)) {
14204 AV** this_array_ptr;
14207 this_array_ptr = (AV**) av_fetch(multi_char_matches,
14209 while ((this_sequence = av_pop(*this_array_ptr)) !=
14212 if (! first_time) {
14213 sv_catpv(substitute_parse, "|");
14215 first_time = FALSE;
14217 sv_catpv(substitute_parse, SvPVX(this_sequence));
14222 /* If the character class contains anything else besides these
14223 * multi-character folds, have to include it in recursive parsing */
14224 if (element_count) {
14225 sv_catpv(substitute_parse, "|[");
14226 sv_catpvn(substitute_parse, orig_parse, RExC_parse - orig_parse);
14227 sv_catpv(substitute_parse, "]");
14230 sv_catpv(substitute_parse, ")");
14233 /* This is a way to get the parse to skip forward a whole named
14234 * sequence instead of matching the 2nd character when it fails the
14236 sv_catpv(substitute_parse, "(*THEN)(*SKIP)(*FAIL)|.)");
14240 RExC_parse = SvPV(substitute_parse, len);
14241 RExC_end = RExC_parse + len;
14242 RExC_in_multi_char_class = 1;
14243 RExC_emit = (regnode *)orig_emit;
14245 ret = reg(pRExC_state, 1, ®_flags, depth+1);
14247 *flagp |= reg_flags&(HASWIDTH|SIMPLE|SPSTART|POSTPONED|RESTART_UTF8);
14249 RExC_parse = save_parse;
14250 RExC_end = save_end;
14251 RExC_in_multi_char_class = 0;
14252 SvREFCNT_dec_NN(multi_char_matches);
14256 /* Here, we've gone through the entire class and dealt with multi-char
14257 * folds. We are now in a position that we can do some checks to see if we
14258 * can optimize this ANYOF node into a simpler one, even in Pass 1.
14259 * Currently we only do two checks:
14260 * 1) is in the unlikely event that the user has specified both, eg. \w and
14261 * \W under /l, then the class matches everything. (This optimization
14262 * is done only to make the optimizer code run later work.)
14263 * 2) if the character class contains only a single element (including a
14264 * single range), we see if there is an equivalent node for it.
14265 * Other checks are possible */
14266 if (! ret_invlist /* Can't optimize if returning the constructed
14268 && (UNLIKELY(posixl_matches_all) || element_count == 1))
14273 if (UNLIKELY(posixl_matches_all)) {
14276 else if (namedclass > OOB_NAMEDCLASS) { /* this is a named class, like
14277 \w or [:digit:] or \p{foo}
14280 /* All named classes are mapped into POSIXish nodes, with its FLAG
14281 * argument giving which class it is */
14282 switch ((I32)namedclass) {
14283 case ANYOF_UNIPROP:
14286 /* These don't depend on the charset modifiers. They always
14287 * match under /u rules */
14288 case ANYOF_NHORIZWS:
14289 case ANYOF_HORIZWS:
14290 namedclass = ANYOF_BLANK + namedclass - ANYOF_HORIZWS;
14293 case ANYOF_NVERTWS:
14298 /* The actual POSIXish node for all the rest depends on the
14299 * charset modifier. The ones in the first set depend only on
14300 * ASCII or, if available on this platform, locale */
14304 op = (LOC) ? POSIXL : POSIXA;
14315 /* under /a could be alpha */
14317 if (ASCII_RESTRICTED) {
14318 namedclass = ANYOF_ALPHA + (namedclass % 2);
14326 /* The rest have more possibilities depending on the charset.
14327 * We take advantage of the enum ordering of the charset
14328 * modifiers to get the exact node type, */
14330 op = POSIXD + get_regex_charset(RExC_flags);
14331 if (op > POSIXA) { /* /aa is same as /a */
14336 /* The odd numbered ones are the complements of the
14337 * next-lower even number one */
14338 if (namedclass % 2 == 1) {
14342 arg = namedclass_to_classnum(namedclass);
14346 else if (value == prevvalue) {
14348 /* Here, the class consists of just a single code point */
14351 if (! LOC && value == '\n') {
14352 op = REG_ANY; /* Optimize [^\n] */
14353 *flagp |= HASWIDTH|SIMPLE;
14357 else if (value < 256 || UTF) {
14359 /* Optimize a single value into an EXACTish node, but not if it
14360 * would require converting the pattern to UTF-8. */
14361 op = compute_EXACTish(pRExC_state);
14363 } /* Otherwise is a range */
14364 else if (! LOC) { /* locale could vary these */
14365 if (prevvalue == '0') {
14366 if (value == '9') {
14371 else if (prevvalue == 'A') {
14374 && literal_endpoint == 2
14377 arg = (FOLD) ? _CC_ALPHA : _CC_UPPER;
14381 else if (prevvalue == 'a') {
14384 && literal_endpoint == 2
14387 arg = (FOLD) ? _CC_ALPHA : _CC_LOWER;
14393 /* Here, we have changed <op> away from its initial value iff we found
14394 * an optimization */
14397 /* Throw away this ANYOF regnode, and emit the calculated one,
14398 * which should correspond to the beginning, not current, state of
14400 const char * cur_parse = RExC_parse;
14401 RExC_parse = (char *)orig_parse;
14405 /* To get locale nodes to not use the full ANYOF size would
14406 * require moving the code above that writes the portions
14407 * of it that aren't in other nodes to after this point.
14408 * e.g. ANYOF_POSIXL_SET */
14409 RExC_size = orig_size;
14413 RExC_emit = (regnode *)orig_emit;
14414 if (PL_regkind[op] == POSIXD) {
14415 if (op == POSIXL) {
14416 RExC_contains_locale = 1;
14419 op += NPOSIXD - POSIXD;
14424 ret = reg_node(pRExC_state, op);
14426 if (PL_regkind[op] == POSIXD || PL_regkind[op] == NPOSIXD) {
14430 *flagp |= HASWIDTH|SIMPLE;
14432 else if (PL_regkind[op] == EXACT) {
14433 alloc_maybe_populate_EXACT(pRExC_state, ret, flagp, 0, value,
14434 TRUE /* downgradable to EXACT */
14438 RExC_parse = (char *) cur_parse;
14440 SvREFCNT_dec(posixes);
14441 SvREFCNT_dec(nposixes);
14442 SvREFCNT_dec(cp_list);
14443 SvREFCNT_dec(cp_foldable_list);
14450 /****** !SIZE_ONLY (Pass 2) AFTER HERE *********/
14452 /* If folding, we calculate all characters that could fold to or from the
14453 * ones already on the list */
14454 if (cp_foldable_list) {
14456 UV start, end; /* End points of code point ranges */
14458 SV* fold_intersection = NULL;
14461 /* Our calculated list will be for Unicode rules. For locale
14462 * matching, we have to keep a separate list that is consulted at
14463 * runtime only when the locale indicates Unicode rules. For
14464 * non-locale, we just use to the general list */
14466 use_list = &only_utf8_locale_list;
14469 use_list = &cp_list;
14472 /* Only the characters in this class that participate in folds need
14473 * be checked. Get the intersection of this class and all the
14474 * possible characters that are foldable. This can quickly narrow
14475 * down a large class */
14476 _invlist_intersection(PL_utf8_foldable, cp_foldable_list,
14477 &fold_intersection);
14479 /* The folds for all the Latin1 characters are hard-coded into this
14480 * program, but we have to go out to disk to get the others. */
14481 if (invlist_highest(cp_foldable_list) >= 256) {
14483 /* This is a hash that for a particular fold gives all
14484 * characters that are involved in it */
14485 if (! PL_utf8_foldclosures) {
14486 _load_PL_utf8_foldclosures();
14490 /* Now look at the foldable characters in this class individually */
14491 invlist_iterinit(fold_intersection);
14492 while (invlist_iternext(fold_intersection, &start, &end)) {
14495 /* Look at every character in the range */
14496 for (j = start; j <= end; j++) {
14497 U8 foldbuf[UTF8_MAXBYTES_CASE+1];
14503 if (IS_IN_SOME_FOLD_L1(j)) {
14505 /* ASCII is always matched; non-ASCII is matched
14506 * only under Unicode rules (which could happen
14507 * under /l if the locale is a UTF-8 one */
14508 if (isASCII(j) || ! DEPENDS_SEMANTICS) {
14509 *use_list = add_cp_to_invlist(*use_list,
14510 PL_fold_latin1[j]);
14514 add_cp_to_invlist(depends_list,
14515 PL_fold_latin1[j]);
14519 if (HAS_NONLATIN1_SIMPLE_FOLD_CLOSURE(j)
14520 && (! isASCII(j) || ! ASCII_FOLD_RESTRICTED))
14522 add_above_Latin1_folds(pRExC_state,
14529 /* Here is an above Latin1 character. We don't have the
14530 * rules hard-coded for it. First, get its fold. This is
14531 * the simple fold, as the multi-character folds have been
14532 * handled earlier and separated out */
14533 _to_uni_fold_flags(j, foldbuf, &foldlen,
14534 (ASCII_FOLD_RESTRICTED)
14535 ? FOLD_FLAGS_NOMIX_ASCII
14538 /* Single character fold of above Latin1. Add everything in
14539 * its fold closure to the list that this node should match.
14540 * The fold closures data structure is a hash with the keys
14541 * being the UTF-8 of every character that is folded to, like
14542 * 'k', and the values each an array of all code points that
14543 * fold to its key. e.g. [ 'k', 'K', KELVIN_SIGN ].
14544 * Multi-character folds are not included */
14545 if ((listp = hv_fetch(PL_utf8_foldclosures,
14546 (char *) foldbuf, foldlen, FALSE)))
14548 AV* list = (AV*) *listp;
14550 for (k = 0; k <= av_tindex(list); k++) {
14551 SV** c_p = av_fetch(list, k, FALSE);
14557 /* /aa doesn't allow folds between ASCII and non- */
14558 if ((ASCII_FOLD_RESTRICTED
14559 && (isASCII(c) != isASCII(j))))
14564 /* Folds under /l which cross the 255/256 boundary
14565 * are added to a separate list. (These are valid
14566 * only when the locale is UTF-8.) */
14567 if (c < 256 && LOC) {
14568 *use_list = add_cp_to_invlist(*use_list, c);
14572 if (isASCII(c) || c > 255 || AT_LEAST_UNI_SEMANTICS)
14574 cp_list = add_cp_to_invlist(cp_list, c);
14577 /* Similarly folds involving non-ascii Latin1
14578 * characters under /d are added to their list */
14579 depends_list = add_cp_to_invlist(depends_list,
14586 SvREFCNT_dec_NN(fold_intersection);
14589 /* Now that we have finished adding all the folds, there is no reason
14590 * to keep the foldable list separate */
14591 _invlist_union(cp_list, cp_foldable_list, &cp_list);
14592 SvREFCNT_dec_NN(cp_foldable_list);
14595 /* And combine the result (if any) with any inversion list from posix
14596 * classes. The lists are kept separate up to now because we don't want to
14597 * fold the classes (folding of those is automatically handled by the swash
14598 * fetching code) */
14599 if (posixes || nposixes) {
14600 if (posixes && AT_LEAST_ASCII_RESTRICTED) {
14601 /* Under /a and /aa, nothing above ASCII matches these */
14602 _invlist_intersection(posixes,
14603 PL_XPosix_ptrs[_CC_ASCII],
14607 if (DEPENDS_SEMANTICS) {
14608 /* Under /d, everything in the upper half of the Latin1 range
14609 * matches these complements */
14610 ANYOF_FLAGS(ret) |= ANYOF_NON_UTF8_NON_ASCII_ALL;
14612 else if (AT_LEAST_ASCII_RESTRICTED) {
14613 /* Under /a and /aa, everything above ASCII matches these
14615 _invlist_union_complement_2nd(nposixes,
14616 PL_XPosix_ptrs[_CC_ASCII],
14620 _invlist_union(posixes, nposixes, &posixes);
14621 SvREFCNT_dec_NN(nposixes);
14624 posixes = nposixes;
14627 if (! DEPENDS_SEMANTICS) {
14629 _invlist_union(cp_list, posixes, &cp_list);
14630 SvREFCNT_dec_NN(posixes);
14637 /* Under /d, we put into a separate list the Latin1 things that
14638 * match only when the target string is utf8 */
14639 SV* nonascii_but_latin1_properties = NULL;
14640 _invlist_intersection(posixes, PL_UpperLatin1,
14641 &nonascii_but_latin1_properties);
14642 _invlist_subtract(posixes, nonascii_but_latin1_properties,
14645 _invlist_union(cp_list, posixes, &cp_list);
14646 SvREFCNT_dec_NN(posixes);
14652 if (depends_list) {
14653 _invlist_union(depends_list, nonascii_but_latin1_properties,
14655 SvREFCNT_dec_NN(nonascii_but_latin1_properties);
14658 depends_list = nonascii_but_latin1_properties;
14663 /* And combine the result (if any) with any inversion list from properties.
14664 * The lists are kept separate up to now so that we can distinguish the two
14665 * in regards to matching above-Unicode. A run-time warning is generated
14666 * if a Unicode property is matched against a non-Unicode code point. But,
14667 * we allow user-defined properties to match anything, without any warning,
14668 * and we also suppress the warning if there is a portion of the character
14669 * class that isn't a Unicode property, and which matches above Unicode, \W
14670 * or [\x{110000}] for example.
14671 * (Note that in this case, unlike the Posix one above, there is no
14672 * <depends_list>, because having a Unicode property forces Unicode
14677 /* If it matters to the final outcome, see if a non-property
14678 * component of the class matches above Unicode. If so, the
14679 * warning gets suppressed. This is true even if just a single
14680 * such code point is specified, as though not strictly correct if
14681 * another such code point is matched against, the fact that they
14682 * are using above-Unicode code points indicates they should know
14683 * the issues involved */
14685 warn_super = ! (invert
14686 ^ (invlist_highest(cp_list) > PERL_UNICODE_MAX));
14689 _invlist_union(properties, cp_list, &cp_list);
14690 SvREFCNT_dec_NN(properties);
14693 cp_list = properties;
14697 ANYOF_FLAGS(ret) |= ANYOF_WARN_SUPER;
14701 /* Here, we have calculated what code points should be in the character
14704 * Now we can see about various optimizations. Fold calculation (which we
14705 * did above) needs to take place before inversion. Otherwise /[^k]/i
14706 * would invert to include K, which under /i would match k, which it
14707 * shouldn't. Therefore we can't invert folded locale now, as it won't be
14708 * folded until runtime */
14710 /* If we didn't do folding, it's because some information isn't available
14711 * until runtime; set the run-time fold flag for these. (We don't have to
14712 * worry about properties folding, as that is taken care of by the swash
14713 * fetching). We know to set the flag if we have a non-NULL list for UTF-8
14714 * locales, or the class matches at least one 0-255 range code point */
14716 if (only_utf8_locale_list) {
14717 ANYOF_FLAGS(ret) |= ANYOF_LOC_FOLD;
14719 else if (cp_list) { /* Look to see if there a 0-255 code point is in
14722 invlist_iterinit(cp_list);
14723 if (invlist_iternext(cp_list, &start, &end) && start < 256) {
14724 ANYOF_FLAGS(ret) |= ANYOF_LOC_FOLD;
14726 invlist_iterfinish(cp_list);
14730 /* Optimize inverted simple patterns (e.g. [^a-z]) when everything is known
14731 * at compile time. Besides not inverting folded locale now, we can't
14732 * invert if there are things such as \w, which aren't known until runtime
14736 && ! (ANYOF_FLAGS(ret) & (ANYOF_LOCALE_FLAGS))
14738 && ! HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION)
14740 _invlist_invert(cp_list);
14742 /* Any swash can't be used as-is, because we've inverted things */
14744 SvREFCNT_dec_NN(swash);
14748 /* Clear the invert flag since have just done it here */
14753 *ret_invlist = cp_list;
14754 SvREFCNT_dec(swash);
14756 /* Discard the generated node */
14758 RExC_size = orig_size;
14761 RExC_emit = orig_emit;
14766 /* Some character classes are equivalent to other nodes. Such nodes take
14767 * up less room and generally fewer operations to execute than ANYOF nodes.
14768 * Above, we checked for and optimized into some such equivalents for
14769 * certain common classes that are easy to test. Getting to this point in
14770 * the code means that the class didn't get optimized there. Since this
14771 * code is only executed in Pass 2, it is too late to save space--it has
14772 * been allocated in Pass 1, and currently isn't given back. But turning
14773 * things into an EXACTish node can allow the optimizer to join it to any
14774 * adjacent such nodes. And if the class is equivalent to things like /./,
14775 * expensive run-time swashes can be avoided. Now that we have more
14776 * complete information, we can find things necessarily missed by the
14777 * earlier code. I (khw) am not sure how much to look for here. It would
14778 * be easy, but perhaps too slow, to check any candidates against all the
14779 * node types they could possibly match using _invlistEQ(). */
14784 && ! (ANYOF_FLAGS(ret) & (ANYOF_LOCALE_FLAGS))
14785 && ! HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION
14787 /* We don't optimize if we are supposed to make sure all non-Unicode
14788 * code points raise a warning, as only ANYOF nodes have this check.
14790 && ! ((ANYOF_FLAGS(ret) & ANYOF_WARN_SUPER) && ALWAYS_WARN_SUPER))
14793 U8 op = END; /* The optimzation node-type */
14794 const char * cur_parse= RExC_parse;
14796 invlist_iterinit(cp_list);
14797 if (! invlist_iternext(cp_list, &start, &end)) {
14799 /* Here, the list is empty. This happens, for example, when a
14800 * Unicode property is the only thing in the character class, and
14801 * it doesn't match anything. (perluniprops.pod notes such
14804 *flagp |= HASWIDTH|SIMPLE;
14806 else if (start == end) { /* The range is a single code point */
14807 if (! invlist_iternext(cp_list, &start, &end)
14809 /* Don't do this optimization if it would require changing
14810 * the pattern to UTF-8 */
14811 && (start < 256 || UTF))
14813 /* Here, the list contains a single code point. Can optimize
14814 * into an EXACTish node */
14823 /* A locale node under folding with one code point can be
14824 * an EXACTFL, as its fold won't be calculated until
14830 /* Here, we are generally folding, but there is only one
14831 * code point to match. If we have to, we use an EXACT
14832 * node, but it would be better for joining with adjacent
14833 * nodes in the optimization pass if we used the same
14834 * EXACTFish node that any such are likely to be. We can
14835 * do this iff the code point doesn't participate in any
14836 * folds. For example, an EXACTF of a colon is the same as
14837 * an EXACT one, since nothing folds to or from a colon. */
14839 if (IS_IN_SOME_FOLD_L1(value)) {
14844 if (_invlist_contains_cp(PL_utf8_foldable, value)) {
14849 /* If we haven't found the node type, above, it means we
14850 * can use the prevailing one */
14852 op = compute_EXACTish(pRExC_state);
14857 else if (start == 0) {
14858 if (end == UV_MAX) {
14860 *flagp |= HASWIDTH|SIMPLE;
14863 else if (end == '\n' - 1
14864 && invlist_iternext(cp_list, &start, &end)
14865 && start == '\n' + 1 && end == UV_MAX)
14868 *flagp |= HASWIDTH|SIMPLE;
14872 invlist_iterfinish(cp_list);
14875 RExC_parse = (char *)orig_parse;
14876 RExC_emit = (regnode *)orig_emit;
14878 ret = reg_node(pRExC_state, op);
14880 RExC_parse = (char *)cur_parse;
14882 if (PL_regkind[op] == EXACT) {
14883 alloc_maybe_populate_EXACT(pRExC_state, ret, flagp, 0, value,
14884 TRUE /* downgradable to EXACT */
14888 SvREFCNT_dec_NN(cp_list);
14893 /* Here, <cp_list> contains all the code points we can determine at
14894 * compile time that match under all conditions. Go through it, and
14895 * for things that belong in the bitmap, put them there, and delete from
14896 * <cp_list>. While we are at it, see if everything above 255 is in the
14897 * list, and if so, set a flag to speed up execution */
14899 populate_ANYOF_from_invlist(ret, &cp_list);
14902 ANYOF_FLAGS(ret) |= ANYOF_INVERT;
14905 /* Here, the bitmap has been populated with all the Latin1 code points that
14906 * always match. Can now add to the overall list those that match only
14907 * when the target string is UTF-8 (<depends_list>). */
14908 if (depends_list) {
14910 _invlist_union(cp_list, depends_list, &cp_list);
14911 SvREFCNT_dec_NN(depends_list);
14914 cp_list = depends_list;
14916 ANYOF_FLAGS(ret) |= ANYOF_UTF8;
14919 /* If there is a swash and more than one element, we can't use the swash in
14920 * the optimization below. */
14921 if (swash && element_count > 1) {
14922 SvREFCNT_dec_NN(swash);
14926 set_ANYOF_arg(pRExC_state, ret, cp_list,
14927 (HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION)
14929 only_utf8_locale_list,
14930 swash, has_user_defined_property);
14932 *flagp |= HASWIDTH|SIMPLE;
14934 if (ANYOF_FLAGS(ret) & ANYOF_LOCALE_FLAGS) {
14935 RExC_contains_locale = 1;
14941 #undef HAS_NONLOCALE_RUNTIME_PROPERTY_DEFINITION
14944 S_set_ANYOF_arg(pTHX_ RExC_state_t* const pRExC_state,
14945 regnode* const node,
14947 SV* const runtime_defns,
14948 SV* const only_utf8_locale_list,
14950 const bool has_user_defined_property)
14952 /* Sets the arg field of an ANYOF-type node 'node', using information about
14953 * the node passed-in. If there is nothing outside the node's bitmap, the
14954 * arg is set to ANYOF_NONBITMAP_EMPTY. Otherwise, it sets the argument to
14955 * the count returned by add_data(), having allocated and stored an array,
14956 * av, that that count references, as follows:
14957 * av[0] stores the character class description in its textual form.
14958 * This is used later (regexec.c:Perl_regclass_swash()) to
14959 * initialize the appropriate swash, and is also useful for dumping
14960 * the regnode. This is set to &PL_sv_undef if the textual
14961 * description is not needed at run-time (as happens if the other
14962 * elements completely define the class)
14963 * av[1] if &PL_sv_undef, is a placeholder to later contain the swash
14964 * computed from av[0]. But if no further computation need be done,
14965 * the swash is stored here now (and av[0] is &PL_sv_undef).
14966 * av[2] stores the inversion list of code points that match only if the
14967 * current locale is UTF-8
14968 * av[3] stores the cp_list inversion list for use in addition or instead
14969 * of av[0]; used only if cp_list exists and av[1] is &PL_sv_undef.
14970 * (Otherwise everything needed is already in av[0] and av[1])
14971 * av[4] is set if any component of the class is from a user-defined
14972 * property; used only if av[3] exists */
14976 PERL_ARGS_ASSERT_SET_ANYOF_ARG;
14978 if (! cp_list && ! runtime_defns && ! only_utf8_locale_list) {
14979 assert(! (ANYOF_FLAGS(node)
14980 & (ANYOF_UTF8|ANYOF_NONBITMAP_NON_UTF8)));
14981 ARG_SET(node, ANYOF_NONBITMAP_EMPTY);
14984 AV * const av = newAV();
14987 assert(ANYOF_FLAGS(node)
14988 & (ANYOF_UTF8|ANYOF_NONBITMAP_NON_UTF8|ANYOF_LOC_FOLD));
14990 av_store(av, 0, (runtime_defns)
14991 ? SvREFCNT_inc(runtime_defns) : &PL_sv_undef);
14994 av_store(av, 1, swash);
14995 SvREFCNT_dec_NN(cp_list);
14998 av_store(av, 1, &PL_sv_undef);
15000 av_store(av, 3, cp_list);
15001 av_store(av, 4, newSVuv(has_user_defined_property));
15005 if (only_utf8_locale_list) {
15006 av_store(av, 2, only_utf8_locale_list);
15009 av_store(av, 2, &PL_sv_undef);
15012 rv = newRV_noinc(MUTABLE_SV(av));
15013 n = add_data(pRExC_state, STR_WITH_LEN("s"));
15014 RExC_rxi->data->data[n] = (void*)rv;
15020 /* reg_skipcomment()
15022 Absorbs an /x style # comment from the input stream,
15023 returning a pointer to the first character beyond the comment, or if the
15024 comment terminates the pattern without anything following it, this returns
15025 one past the final character of the pattern (in other words, RExC_end) and
15026 sets the REG_RUN_ON_COMMENT_SEEN flag.
15028 Note it's the callers responsibility to ensure that we are
15029 actually in /x mode
15033 PERL_STATIC_INLINE char*
15034 S_reg_skipcomment(RExC_state_t *pRExC_state, char* p)
15036 PERL_ARGS_ASSERT_REG_SKIPCOMMENT;
15040 while (p < RExC_end) {
15041 if (*(++p) == '\n') {
15046 /* we ran off the end of the pattern without ending the comment, so we have
15047 * to add an \n when wrapping */
15048 RExC_seen |= REG_RUN_ON_COMMENT_SEEN;
15054 Advances the parse position, and optionally absorbs
15055 "whitespace" from the inputstream.
15057 Without /x "whitespace" means (?#...) style comments only,
15058 with /x this means (?#...) and # comments and whitespace proper.
15060 Returns the RExC_parse point from BEFORE the scan occurs.
15062 This is the /x friendly way of saying RExC_parse++.
15066 S_nextchar(pTHX_ RExC_state_t *pRExC_state)
15068 char* const retval = RExC_parse++;
15070 PERL_ARGS_ASSERT_NEXTCHAR;
15073 if (RExC_end - RExC_parse >= 3
15074 && *RExC_parse == '('
15075 && RExC_parse[1] == '?'
15076 && RExC_parse[2] == '#')
15078 while (*RExC_parse != ')') {
15079 if (RExC_parse == RExC_end)
15080 FAIL("Sequence (?#... not terminated");
15086 if (RExC_flags & RXf_PMf_EXTENDED) {
15087 char * p = regpatws(pRExC_state, RExC_parse,
15088 TRUE); /* means recognize comments */
15089 if (p != RExC_parse) {
15099 - reg_node - emit a node
15101 STATIC regnode * /* Location. */
15102 S_reg_node(pTHX_ RExC_state_t *pRExC_state, U8 op)
15105 regnode * const ret = RExC_emit;
15106 GET_RE_DEBUG_FLAGS_DECL;
15108 PERL_ARGS_ASSERT_REG_NODE;
15111 SIZE_ALIGN(RExC_size);
15115 if (RExC_emit >= RExC_emit_bound)
15116 Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d, %p>=%p",
15117 op, (void*)RExC_emit, (void*)RExC_emit_bound);
15119 NODE_ALIGN_FILL(ret);
15121 FILL_ADVANCE_NODE(ptr, op);
15122 REH_CALL_COMP_NODE_HOOK(pRExC_state->rx, (ptr) - 1);
15123 #ifdef RE_TRACK_PATTERN_OFFSETS
15124 if (RExC_offsets) { /* MJD */
15126 ("%s:%d: (op %s) %s %"UVuf" (len %"UVuf") (max %"UVuf").\n",
15127 "reg_node", __LINE__,
15129 (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0]
15130 ? "Overwriting end of array!\n" : "OK",
15131 (UV)(RExC_emit - RExC_emit_start),
15132 (UV)(RExC_parse - RExC_start),
15133 (UV)RExC_offsets[0]));
15134 Set_Node_Offset(RExC_emit, RExC_parse + (op == END));
15142 - reganode - emit a node with an argument
15144 STATIC regnode * /* Location. */
15145 S_reganode(pTHX_ RExC_state_t *pRExC_state, U8 op, U32 arg)
15148 regnode * const ret = RExC_emit;
15149 GET_RE_DEBUG_FLAGS_DECL;
15151 PERL_ARGS_ASSERT_REGANODE;
15154 SIZE_ALIGN(RExC_size);
15159 assert(2==regarglen[op]+1);
15161 Anything larger than this has to allocate the extra amount.
15162 If we changed this to be:
15164 RExC_size += (1 + regarglen[op]);
15166 then it wouldn't matter. Its not clear what side effect
15167 might come from that so its not done so far.
15172 if (RExC_emit >= RExC_emit_bound)
15173 Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d, %p>=%p",
15174 op, (void*)RExC_emit, (void*)RExC_emit_bound);
15176 NODE_ALIGN_FILL(ret);
15178 FILL_ADVANCE_NODE_ARG(ptr, op, arg);
15179 REH_CALL_COMP_NODE_HOOK(pRExC_state->rx, (ptr) - 2);
15180 #ifdef RE_TRACK_PATTERN_OFFSETS
15181 if (RExC_offsets) { /* MJD */
15183 ("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n",
15187 (UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0] ?
15188 "Overwriting end of array!\n" : "OK",
15189 (UV)(RExC_emit - RExC_emit_start),
15190 (UV)(RExC_parse - RExC_start),
15191 (UV)RExC_offsets[0]));
15192 Set_Cur_Node_Offset;
15200 - reguni - emit (if appropriate) a Unicode character
15202 PERL_STATIC_INLINE STRLEN
15203 S_reguni(pTHX_ const RExC_state_t *pRExC_state, UV uv, char* s)
15205 PERL_ARGS_ASSERT_REGUNI;
15207 return SIZE_ONLY ? UNISKIP(uv) : (uvchr_to_utf8((U8*)s, uv) - (U8*)s);
15211 - reginsert - insert an operator in front of already-emitted operand
15213 * Means relocating the operand.
15216 S_reginsert(pTHX_ RExC_state_t *pRExC_state, U8 op, regnode *opnd, U32 depth)
15221 const int offset = regarglen[(U8)op];
15222 const int size = NODE_STEP_REGNODE + offset;
15223 GET_RE_DEBUG_FLAGS_DECL;
15225 PERL_ARGS_ASSERT_REGINSERT;
15226 PERL_UNUSED_CONTEXT;
15227 PERL_UNUSED_ARG(depth);
15228 /* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
15229 DEBUG_PARSE_FMT("inst"," - %s",PL_reg_name[op]);
15238 if (RExC_open_parens) {
15240 /*DEBUG_PARSE_FMT("inst"," - %"IVdf, (IV)RExC_npar);*/
15241 for ( paren=0 ; paren < RExC_npar ; paren++ ) {
15242 if ( RExC_open_parens[paren] >= opnd ) {
15243 /*DEBUG_PARSE_FMT("open"," - %d",size);*/
15244 RExC_open_parens[paren] += size;
15246 /*DEBUG_PARSE_FMT("open"," - %s","ok");*/
15248 if ( RExC_close_parens[paren] >= opnd ) {
15249 /*DEBUG_PARSE_FMT("close"," - %d",size);*/
15250 RExC_close_parens[paren] += size;
15252 /*DEBUG_PARSE_FMT("close"," - %s","ok");*/
15257 while (src > opnd) {
15258 StructCopy(--src, --dst, regnode);
15259 #ifdef RE_TRACK_PATTERN_OFFSETS
15260 if (RExC_offsets) { /* MJD 20010112 */
15262 ("%s(%d): (op %s) %s copy %"UVuf" -> %"UVuf" (max %"UVuf").\n",
15266 (UV)(dst - RExC_emit_start) > RExC_offsets[0]
15267 ? "Overwriting end of array!\n" : "OK",
15268 (UV)(src - RExC_emit_start),
15269 (UV)(dst - RExC_emit_start),
15270 (UV)RExC_offsets[0]));
15271 Set_Node_Offset_To_R(dst-RExC_emit_start, Node_Offset(src));
15272 Set_Node_Length_To_R(dst-RExC_emit_start, Node_Length(src));
15278 place = opnd; /* Op node, where operand used to be. */
15279 #ifdef RE_TRACK_PATTERN_OFFSETS
15280 if (RExC_offsets) { /* MJD */
15282 ("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n",
15286 (UV)(place - RExC_emit_start) > RExC_offsets[0]
15287 ? "Overwriting end of array!\n" : "OK",
15288 (UV)(place - RExC_emit_start),
15289 (UV)(RExC_parse - RExC_start),
15290 (UV)RExC_offsets[0]));
15291 Set_Node_Offset(place, RExC_parse);
15292 Set_Node_Length(place, 1);
15295 src = NEXTOPER(place);
15296 FILL_ADVANCE_NODE(place, op);
15297 REH_CALL_COMP_NODE_HOOK(pRExC_state->rx, (place) - 1);
15298 Zero(src, offset, regnode);
15302 - regtail - set the next-pointer at the end of a node chain of p to val.
15303 - SEE ALSO: regtail_study
15305 /* TODO: All three parms should be const */
15307 S_regtail(pTHX_ RExC_state_t *pRExC_state, regnode *p,
15308 const regnode *val,U32 depth)
15311 GET_RE_DEBUG_FLAGS_DECL;
15313 PERL_ARGS_ASSERT_REGTAIL;
15315 PERL_UNUSED_ARG(depth);
15321 /* Find last node. */
15324 regnode * const temp = regnext(scan);
15326 SV * const mysv=sv_newmortal();
15327 DEBUG_PARSE_MSG((scan==p ? "tail" : ""));
15328 regprop(RExC_rx, mysv, scan, NULL);
15329 PerlIO_printf(Perl_debug_log, "~ %s (%d) %s %s\n",
15330 SvPV_nolen_const(mysv), REG_NODE_NUM(scan),
15331 (temp == NULL ? "->" : ""),
15332 (temp == NULL ? PL_reg_name[OP(val)] : "")
15340 if (reg_off_by_arg[OP(scan)]) {
15341 ARG_SET(scan, val - scan);
15344 NEXT_OFF(scan) = val - scan;
15350 - regtail_study - set the next-pointer at the end of a node chain of p to val.
15351 - Look for optimizable sequences at the same time.
15352 - currently only looks for EXACT chains.
15354 This is experimental code. The idea is to use this routine to perform
15355 in place optimizations on branches and groups as they are constructed,
15356 with the long term intention of removing optimization from study_chunk so
15357 that it is purely analytical.
15359 Currently only used when in DEBUG mode. The macro REGTAIL_STUDY() is used
15360 to control which is which.
15363 /* TODO: All four parms should be const */
15366 S_regtail_study(pTHX_ RExC_state_t *pRExC_state, regnode *p,
15367 const regnode *val,U32 depth)
15371 #ifdef EXPERIMENTAL_INPLACESCAN
15374 GET_RE_DEBUG_FLAGS_DECL;
15376 PERL_ARGS_ASSERT_REGTAIL_STUDY;
15382 /* Find last node. */
15386 regnode * const temp = regnext(scan);
15387 #ifdef EXPERIMENTAL_INPLACESCAN
15388 if (PL_regkind[OP(scan)] == EXACT) {
15389 bool unfolded_multi_char; /* Unexamined in this routine */
15390 if (join_exact(pRExC_state, scan, &min,
15391 &unfolded_multi_char, 1, val, depth+1))
15396 switch (OP(scan)) {
15399 case EXACTFA_NO_TRIE:
15404 if( exact == PSEUDO )
15406 else if ( exact != OP(scan) )
15415 SV * const mysv=sv_newmortal();
15416 DEBUG_PARSE_MSG((scan==p ? "tsdy" : ""));
15417 regprop(RExC_rx, mysv, scan, NULL);
15418 PerlIO_printf(Perl_debug_log, "~ %s (%d) -> %s\n",
15419 SvPV_nolen_const(mysv),
15420 REG_NODE_NUM(scan),
15421 PL_reg_name[exact]);
15428 SV * const mysv_val=sv_newmortal();
15429 DEBUG_PARSE_MSG("");
15430 regprop(RExC_rx, mysv_val, val, NULL);
15431 PerlIO_printf(Perl_debug_log,
15432 "~ attach to %s (%"IVdf") offset to %"IVdf"\n",
15433 SvPV_nolen_const(mysv_val),
15434 (IV)REG_NODE_NUM(val),
15438 if (reg_off_by_arg[OP(scan)]) {
15439 ARG_SET(scan, val - scan);
15442 NEXT_OFF(scan) = val - scan;
15450 - regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
15455 S_regdump_intflags(pTHX_ const char *lead, const U32 flags)
15460 ASSUME(REG_INTFLAGS_NAME_SIZE <= sizeof(flags)*8);
15462 for (bit=0; bit<REG_INTFLAGS_NAME_SIZE; bit++) {
15463 if (flags & (1<<bit)) {
15464 if (!set++ && lead)
15465 PerlIO_printf(Perl_debug_log, "%s",lead);
15466 PerlIO_printf(Perl_debug_log, "%s ",PL_reg_intflags_name[bit]);
15471 PerlIO_printf(Perl_debug_log, "\n");
15473 PerlIO_printf(Perl_debug_log, "%s[none-set]\n",lead);
15478 S_regdump_extflags(pTHX_ const char *lead, const U32 flags)
15484 ASSUME(REG_EXTFLAGS_NAME_SIZE <= sizeof(flags)*8);
15486 for (bit=0; bit<REG_EXTFLAGS_NAME_SIZE; bit++) {
15487 if (flags & (1<<bit)) {
15488 if ((1<<bit) & RXf_PMf_CHARSET) { /* Output separately, below */
15491 if (!set++ && lead)
15492 PerlIO_printf(Perl_debug_log, "%s",lead);
15493 PerlIO_printf(Perl_debug_log, "%s ",PL_reg_extflags_name[bit]);
15496 if ((cs = get_regex_charset(flags)) != REGEX_DEPENDS_CHARSET) {
15497 if (!set++ && lead) {
15498 PerlIO_printf(Perl_debug_log, "%s",lead);
15501 case REGEX_UNICODE_CHARSET:
15502 PerlIO_printf(Perl_debug_log, "UNICODE");
15504 case REGEX_LOCALE_CHARSET:
15505 PerlIO_printf(Perl_debug_log, "LOCALE");
15507 case REGEX_ASCII_RESTRICTED_CHARSET:
15508 PerlIO_printf(Perl_debug_log, "ASCII-RESTRICTED");
15510 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
15511 PerlIO_printf(Perl_debug_log, "ASCII-MORE_RESTRICTED");
15514 PerlIO_printf(Perl_debug_log, "UNKNOWN CHARACTER SET");
15520 PerlIO_printf(Perl_debug_log, "\n");
15522 PerlIO_printf(Perl_debug_log, "%s[none-set]\n",lead);
15528 Perl_regdump(pTHX_ const regexp *r)
15531 SV * const sv = sv_newmortal();
15532 SV *dsv= sv_newmortal();
15533 RXi_GET_DECL(r,ri);
15534 GET_RE_DEBUG_FLAGS_DECL;
15536 PERL_ARGS_ASSERT_REGDUMP;
15538 (void)dumpuntil(r, ri->program, ri->program + 1, NULL, NULL, sv, 0, 0);
15540 /* Header fields of interest. */
15541 if (r->anchored_substr) {
15542 RE_PV_QUOTED_DECL(s, 0, dsv, SvPVX_const(r->anchored_substr),
15543 RE_SV_DUMPLEN(r->anchored_substr), 30);
15544 PerlIO_printf(Perl_debug_log,
15545 "anchored %s%s at %"IVdf" ",
15546 s, RE_SV_TAIL(r->anchored_substr),
15547 (IV)r->anchored_offset);
15548 } else if (r->anchored_utf8) {
15549 RE_PV_QUOTED_DECL(s, 1, dsv, SvPVX_const(r->anchored_utf8),
15550 RE_SV_DUMPLEN(r->anchored_utf8), 30);
15551 PerlIO_printf(Perl_debug_log,
15552 "anchored utf8 %s%s at %"IVdf" ",
15553 s, RE_SV_TAIL(r->anchored_utf8),
15554 (IV)r->anchored_offset);
15556 if (r->float_substr) {
15557 RE_PV_QUOTED_DECL(s, 0, dsv, SvPVX_const(r->float_substr),
15558 RE_SV_DUMPLEN(r->float_substr), 30);
15559 PerlIO_printf(Perl_debug_log,
15560 "floating %s%s at %"IVdf"..%"UVuf" ",
15561 s, RE_SV_TAIL(r->float_substr),
15562 (IV)r->float_min_offset, (UV)r->float_max_offset);
15563 } else if (r->float_utf8) {
15564 RE_PV_QUOTED_DECL(s, 1, dsv, SvPVX_const(r->float_utf8),
15565 RE_SV_DUMPLEN(r->float_utf8), 30);
15566 PerlIO_printf(Perl_debug_log,
15567 "floating utf8 %s%s at %"IVdf"..%"UVuf" ",
15568 s, RE_SV_TAIL(r->float_utf8),
15569 (IV)r->float_min_offset, (UV)r->float_max_offset);
15571 if (r->check_substr || r->check_utf8)
15572 PerlIO_printf(Perl_debug_log,
15574 (r->check_substr == r->float_substr
15575 && r->check_utf8 == r->float_utf8
15576 ? "(checking floating" : "(checking anchored"));
15577 if (r->intflags & PREGf_NOSCAN)
15578 PerlIO_printf(Perl_debug_log, " noscan");
15579 if (r->extflags & RXf_CHECK_ALL)
15580 PerlIO_printf(Perl_debug_log, " isall");
15581 if (r->check_substr || r->check_utf8)
15582 PerlIO_printf(Perl_debug_log, ") ");
15584 if (ri->regstclass) {
15585 regprop(r, sv, ri->regstclass, NULL);
15586 PerlIO_printf(Perl_debug_log, "stclass %s ", SvPVX_const(sv));
15588 if (r->intflags & PREGf_ANCH) {
15589 PerlIO_printf(Perl_debug_log, "anchored");
15590 if (r->intflags & PREGf_ANCH_BOL)
15591 PerlIO_printf(Perl_debug_log, "(BOL)");
15592 if (r->intflags & PREGf_ANCH_MBOL)
15593 PerlIO_printf(Perl_debug_log, "(MBOL)");
15594 if (r->intflags & PREGf_ANCH_SBOL)
15595 PerlIO_printf(Perl_debug_log, "(SBOL)");
15596 if (r->intflags & PREGf_ANCH_GPOS)
15597 PerlIO_printf(Perl_debug_log, "(GPOS)");
15598 PerlIO_putc(Perl_debug_log, ' ');
15600 if (r->intflags & PREGf_GPOS_SEEN)
15601 PerlIO_printf(Perl_debug_log, "GPOS:%"UVuf" ", (UV)r->gofs);
15602 if (r->intflags & PREGf_SKIP)
15603 PerlIO_printf(Perl_debug_log, "plus ");
15604 if (r->intflags & PREGf_IMPLICIT)
15605 PerlIO_printf(Perl_debug_log, "implicit ");
15606 PerlIO_printf(Perl_debug_log, "minlen %"IVdf" ", (IV)r->minlen);
15607 if (r->extflags & RXf_EVAL_SEEN)
15608 PerlIO_printf(Perl_debug_log, "with eval ");
15609 PerlIO_printf(Perl_debug_log, "\n");
15611 regdump_extflags("r->extflags: ",r->extflags);
15612 regdump_intflags("r->intflags: ",r->intflags);
15615 PERL_ARGS_ASSERT_REGDUMP;
15616 PERL_UNUSED_CONTEXT;
15617 PERL_UNUSED_ARG(r);
15618 #endif /* DEBUGGING */
15622 - regprop - printable representation of opcode, with run time support
15626 Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o, const regmatch_info *reginfo)
15631 /* Should be synchronized with * ANYOF_ #xdefines in regcomp.h */
15632 static const char * const anyofs[] = {
15633 #if _CC_WORDCHAR != 0 || _CC_DIGIT != 1 || _CC_ALPHA != 2 || _CC_LOWER != 3 \
15634 || _CC_UPPER != 4 || _CC_PUNCT != 5 || _CC_PRINT != 6 \
15635 || _CC_ALPHANUMERIC != 7 || _CC_GRAPH != 8 || _CC_CASED != 9 \
15636 || _CC_SPACE != 10 || _CC_BLANK != 11 || _CC_XDIGIT != 12 \
15637 || _CC_PSXSPC != 13 || _CC_CNTRL != 14 || _CC_ASCII != 15 \
15638 || _CC_VERTSPACE != 16
15639 #error Need to adjust order of anyofs[]
15676 RXi_GET_DECL(prog,progi);
15677 GET_RE_DEBUG_FLAGS_DECL;
15679 PERL_ARGS_ASSERT_REGPROP;
15683 if (OP(o) > REGNODE_MAX) /* regnode.type is unsigned */
15684 /* It would be nice to FAIL() here, but this may be called from
15685 regexec.c, and it would be hard to supply pRExC_state. */
15686 Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d",
15687 (int)OP(o), (int)REGNODE_MAX);
15688 sv_catpv(sv, PL_reg_name[OP(o)]); /* Take off const! */
15690 k = PL_regkind[OP(o)];
15693 sv_catpvs(sv, " ");
15694 /* Using is_utf8_string() (via PERL_PV_UNI_DETECT)
15695 * is a crude hack but it may be the best for now since
15696 * we have no flag "this EXACTish node was UTF-8"
15698 pv_pretty(sv, STRING(o), STR_LEN(o), 60, PL_colors[0], PL_colors[1],
15699 PERL_PV_ESCAPE_UNI_DETECT |
15700 PERL_PV_ESCAPE_NONASCII |
15701 PERL_PV_PRETTY_ELLIPSES |
15702 PERL_PV_PRETTY_LTGT |
15703 PERL_PV_PRETTY_NOCLEAR
15705 } else if (k == TRIE) {
15706 /* print the details of the trie in dumpuntil instead, as
15707 * progi->data isn't available here */
15708 const char op = OP(o);
15709 const U32 n = ARG(o);
15710 const reg_ac_data * const ac = IS_TRIE_AC(op) ?
15711 (reg_ac_data *)progi->data->data[n] :
15713 const reg_trie_data * const trie
15714 = (reg_trie_data*)progi->data->data[!IS_TRIE_AC(op) ? n : ac->trie];
15716 Perl_sv_catpvf(aTHX_ sv, "-%s",PL_reg_name[o->flags]);
15717 DEBUG_TRIE_COMPILE_r(
15718 Perl_sv_catpvf(aTHX_ sv,
15719 "<S:%"UVuf"/%"IVdf" W:%"UVuf" L:%"UVuf"/%"UVuf" C:%"UVuf"/%"UVuf">",
15720 (UV)trie->startstate,
15721 (IV)trie->statecount-1, /* -1 because of the unused 0 element */
15722 (UV)trie->wordcount,
15725 (UV)TRIE_CHARCOUNT(trie),
15726 (UV)trie->uniquecharcount
15729 if ( IS_ANYOF_TRIE(op) || trie->bitmap ) {
15730 sv_catpvs(sv, "[");
15731 (void) put_latin1_charclass_innards(sv, IS_ANYOF_TRIE(op)
15733 : TRIE_BITMAP(trie));
15734 sv_catpvs(sv, "]");
15737 } else if (k == CURLY) {
15738 if (OP(o) == CURLYM || OP(o) == CURLYN || OP(o) == CURLYX)
15739 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
15740 Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
15742 else if (k == WHILEM && o->flags) /* Ordinal/of */
15743 Perl_sv_catpvf(aTHX_ sv, "[%d/%d]", o->flags & 0xf, o->flags>>4);
15744 else if (k == REF || k == OPEN || k == CLOSE
15745 || k == GROUPP || OP(o)==ACCEPT)
15747 Perl_sv_catpvf(aTHX_ sv, "%d", (int)ARG(o)); /* Parenth number */
15748 if ( RXp_PAREN_NAMES(prog) ) {
15749 if ( k != REF || (OP(o) < NREF)) {
15750 AV *list= MUTABLE_AV(progi->data->data[progi->name_list_idx]);
15751 SV **name= av_fetch(list, ARG(o), 0 );
15753 Perl_sv_catpvf(aTHX_ sv, " '%"SVf"'", SVfARG(*name));
15756 AV *list= MUTABLE_AV(progi->data->data[ progi->name_list_idx ]);
15757 SV *sv_dat= MUTABLE_SV(progi->data->data[ ARG( o ) ]);
15758 I32 *nums=(I32*)SvPVX(sv_dat);
15759 SV **name= av_fetch(list, nums[0], 0 );
15762 for ( n=0; n<SvIVX(sv_dat); n++ ) {
15763 Perl_sv_catpvf(aTHX_ sv, "%s%"IVdf,
15764 (n ? "," : ""), (IV)nums[n]);
15766 Perl_sv_catpvf(aTHX_ sv, " '%"SVf"'", SVfARG(*name));
15770 if ( k == REF && reginfo) {
15771 U32 n = ARG(o); /* which paren pair */
15772 I32 ln = prog->offs[n].start;
15773 if (prog->lastparen < n || ln == -1)
15774 Perl_sv_catpvf(aTHX_ sv, ": FAIL");
15775 else if (ln == prog->offs[n].end)
15776 Perl_sv_catpvf(aTHX_ sv, ": ACCEPT - EMPTY STRING");
15778 const char *s = reginfo->strbeg + ln;
15779 Perl_sv_catpvf(aTHX_ sv, ": ");
15780 Perl_pv_pretty( aTHX_ sv, s, prog->offs[n].end - prog->offs[n].start, 32, 0, 0,
15781 PERL_PV_ESCAPE_UNI_DETECT|PERL_PV_PRETTY_NOCLEAR|PERL_PV_PRETTY_ELLIPSES|PERL_PV_PRETTY_QUOTE );
15784 } else if (k == GOSUB)
15785 /* Paren and offset */
15786 Perl_sv_catpvf(aTHX_ sv, "%d[%+d]", (int)ARG(o),(int)ARG2L(o));
15787 else if (k == VERB) {
15789 Perl_sv_catpvf(aTHX_ sv, ":%"SVf,
15790 SVfARG((MUTABLE_SV(progi->data->data[ ARG( o ) ]))));
15791 } else if (k == LOGICAL)
15792 /* 2: embedded, otherwise 1 */
15793 Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags);
15794 else if (k == ANYOF) {
15795 const U8 flags = ANYOF_FLAGS(o);
15799 if (flags & ANYOF_LOCALE_FLAGS)
15800 sv_catpvs(sv, "{loc}");
15801 if (flags & ANYOF_LOC_FOLD)
15802 sv_catpvs(sv, "{i}");
15803 Perl_sv_catpvf(aTHX_ sv, "[%s", PL_colors[0]);
15804 if (flags & ANYOF_INVERT)
15805 sv_catpvs(sv, "^");
15807 /* output what the standard cp 0-255 bitmap matches */
15808 do_sep = put_latin1_charclass_innards(sv, ANYOF_BITMAP(o));
15810 /* output any special charclass tests (used entirely under use
15812 if (ANYOF_POSIXL_TEST_ANY_SET(o)) {
15814 for (i = 0; i < ANYOF_POSIXL_MAX; i++) {
15815 if (ANYOF_POSIXL_TEST(o,i)) {
15816 sv_catpv(sv, anyofs[i]);
15822 if ((flags & (ANYOF_ABOVE_LATIN1_ALL
15824 |ANYOF_NONBITMAP_NON_UTF8
15828 Perl_sv_catpvf(aTHX_ sv,"%s][%s",PL_colors[1],PL_colors[0]);
15829 if (flags & ANYOF_INVERT)
15830 /*make sure the invert info is in each */
15831 sv_catpvs(sv, "^");
15834 if (flags & ANYOF_NON_UTF8_NON_ASCII_ALL) {
15835 sv_catpvs(sv, "{non-utf8-latin1-all}");
15838 /* output information about the unicode matching */
15839 if (flags & ANYOF_ABOVE_LATIN1_ALL)
15840 sv_catpvs(sv, "{unicode_all}");
15841 else if (ARG(o) != ANYOF_NONBITMAP_EMPTY) {
15842 SV *lv; /* Set if there is something outside the bit map. */
15843 bool byte_output = FALSE; /* If something in the bitmap has
15845 SV *only_utf8_locale;
15847 /* Get the stuff that wasn't in the bitmap */
15848 (void) _get_regclass_nonbitmap_data(prog, o, FALSE,
15849 &lv, &only_utf8_locale);
15850 if (lv && lv != &PL_sv_undef) {
15851 char *s = savesvpv(lv);
15852 char * const origs = s;
15854 while (*s && *s != '\n')
15858 const char * const t = ++s;
15860 if (flags & ANYOF_NONBITMAP_NON_UTF8) {
15861 sv_catpvs(sv, "{outside bitmap}");
15864 sv_catpvs(sv, "{utf8}");
15868 sv_catpvs(sv, " ");
15874 /* Truncate very long output */
15875 if (s - origs > 256) {
15876 Perl_sv_catpvf(aTHX_ sv,
15878 (int) (s - origs - 1),
15884 else if (*s == '\t') {
15898 SvREFCNT_dec_NN(lv);
15901 if ((flags & ANYOF_LOC_FOLD)
15902 && only_utf8_locale
15903 && only_utf8_locale != &PL_sv_undef)
15906 int max_entries = 256;
15908 sv_catpvs(sv, "{utf8 locale}");
15909 invlist_iterinit(only_utf8_locale);
15910 while (invlist_iternext(only_utf8_locale,
15912 put_range(sv, start, end);
15914 if (max_entries < 0) {
15915 sv_catpvs(sv, "...");
15919 invlist_iterfinish(only_utf8_locale);
15924 Perl_sv_catpvf(aTHX_ sv, "%s]", PL_colors[1]);
15926 else if (k == POSIXD || k == NPOSIXD) {
15927 U8 index = FLAGS(o) * 2;
15928 if (index < C_ARRAY_LENGTH(anyofs)) {
15929 if (*anyofs[index] != '[') {
15932 sv_catpv(sv, anyofs[index]);
15933 if (*anyofs[index] != '[') {
15938 Perl_sv_catpvf(aTHX_ sv, "[illegal type=%d])", index);
15941 else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
15942 Perl_sv_catpvf(aTHX_ sv, "[%d]", -(o->flags));
15944 PERL_UNUSED_CONTEXT;
15945 PERL_UNUSED_ARG(sv);
15946 PERL_UNUSED_ARG(o);
15947 PERL_UNUSED_ARG(prog);
15948 PERL_UNUSED_ARG(reginfo);
15949 #endif /* DEBUGGING */
15955 Perl_re_intuit_string(pTHX_ REGEXP * const r)
15956 { /* Assume that RE_INTUIT is set */
15957 struct regexp *const prog = ReANY(r);
15958 GET_RE_DEBUG_FLAGS_DECL;
15960 PERL_ARGS_ASSERT_RE_INTUIT_STRING;
15961 PERL_UNUSED_CONTEXT;
15965 const char * const s = SvPV_nolen_const(prog->check_substr
15966 ? prog->check_substr : prog->check_utf8);
15968 if (!PL_colorset) reginitcolors();
15969 PerlIO_printf(Perl_debug_log,
15970 "%sUsing REx %ssubstr:%s \"%s%.60s%s%s\"\n",
15972 prog->check_substr ? "" : "utf8 ",
15973 PL_colors[5],PL_colors[0],
15976 (strlen(s) > 60 ? "..." : ""));
15979 return prog->check_substr ? prog->check_substr : prog->check_utf8;
15985 handles refcounting and freeing the perl core regexp structure. When
15986 it is necessary to actually free the structure the first thing it
15987 does is call the 'free' method of the regexp_engine associated to
15988 the regexp, allowing the handling of the void *pprivate; member
15989 first. (This routine is not overridable by extensions, which is why
15990 the extensions free is called first.)
15992 See regdupe and regdupe_internal if you change anything here.
15994 #ifndef PERL_IN_XSUB_RE
15996 Perl_pregfree(pTHX_ REGEXP *r)
16002 Perl_pregfree2(pTHX_ REGEXP *rx)
16004 struct regexp *const r = ReANY(rx);
16005 GET_RE_DEBUG_FLAGS_DECL;
16007 PERL_ARGS_ASSERT_PREGFREE2;
16009 if (r->mother_re) {
16010 ReREFCNT_dec(r->mother_re);
16012 CALLREGFREE_PVT(rx); /* free the private data */
16013 SvREFCNT_dec(RXp_PAREN_NAMES(r));
16014 Safefree(r->xpv_len_u.xpvlenu_pv);
16017 SvREFCNT_dec(r->anchored_substr);
16018 SvREFCNT_dec(r->anchored_utf8);
16019 SvREFCNT_dec(r->float_substr);
16020 SvREFCNT_dec(r->float_utf8);
16021 Safefree(r->substrs);
16023 RX_MATCH_COPY_FREE(rx);
16024 #ifdef PERL_ANY_COW
16025 SvREFCNT_dec(r->saved_copy);
16028 SvREFCNT_dec(r->qr_anoncv);
16029 rx->sv_u.svu_rx = 0;
16034 This is a hacky workaround to the structural issue of match results
16035 being stored in the regexp structure which is in turn stored in
16036 PL_curpm/PL_reg_curpm. The problem is that due to qr// the pattern
16037 could be PL_curpm in multiple contexts, and could require multiple
16038 result sets being associated with the pattern simultaneously, such
16039 as when doing a recursive match with (??{$qr})
16041 The solution is to make a lightweight copy of the regexp structure
16042 when a qr// is returned from the code executed by (??{$qr}) this
16043 lightweight copy doesn't actually own any of its data except for
16044 the starp/end and the actual regexp structure itself.
16050 Perl_reg_temp_copy (pTHX_ REGEXP *ret_x, REGEXP *rx)
16052 struct regexp *ret;
16053 struct regexp *const r = ReANY(rx);
16054 const bool islv = ret_x && SvTYPE(ret_x) == SVt_PVLV;
16056 PERL_ARGS_ASSERT_REG_TEMP_COPY;
16059 ret_x = (REGEXP*) newSV_type(SVt_REGEXP);
16061 SvOK_off((SV *)ret_x);
16063 /* For PVLVs, SvANY points to the xpvlv body while sv_u points
16064 to the regexp. (For SVt_REGEXPs, sv_upgrade has already
16065 made both spots point to the same regexp body.) */
16066 REGEXP *temp = (REGEXP *)newSV_type(SVt_REGEXP);
16067 assert(!SvPVX(ret_x));
16068 ret_x->sv_u.svu_rx = temp->sv_any;
16069 temp->sv_any = NULL;
16070 SvFLAGS(temp) = (SvFLAGS(temp) & ~SVTYPEMASK) | SVt_NULL;
16071 SvREFCNT_dec_NN(temp);
16072 /* SvCUR still resides in the xpvlv struct, so the regexp copy-
16073 ing below will not set it. */
16074 SvCUR_set(ret_x, SvCUR(rx));
16077 /* This ensures that SvTHINKFIRST(sv) is true, and hence that
16078 sv_force_normal(sv) is called. */
16080 ret = ReANY(ret_x);
16082 SvFLAGS(ret_x) |= SvUTF8(rx);
16083 /* We share the same string buffer as the original regexp, on which we
16084 hold a reference count, incremented when mother_re is set below.
16085 The string pointer is copied here, being part of the regexp struct.
16087 memcpy(&(ret->xpv_cur), &(r->xpv_cur),
16088 sizeof(regexp) - STRUCT_OFFSET(regexp, xpv_cur));
16090 const I32 npar = r->nparens+1;
16091 Newx(ret->offs, npar, regexp_paren_pair);
16092 Copy(r->offs, ret->offs, npar, regexp_paren_pair);
16095 Newx(ret->substrs, 1, struct reg_substr_data);
16096 StructCopy(r->substrs, ret->substrs, struct reg_substr_data);
16098 SvREFCNT_inc_void(ret->anchored_substr);
16099 SvREFCNT_inc_void(ret->anchored_utf8);
16100 SvREFCNT_inc_void(ret->float_substr);
16101 SvREFCNT_inc_void(ret->float_utf8);
16103 /* check_substr and check_utf8, if non-NULL, point to either their
16104 anchored or float namesakes, and don't hold a second reference. */
16106 RX_MATCH_COPIED_off(ret_x);
16107 #ifdef PERL_ANY_COW
16108 ret->saved_copy = NULL;
16110 ret->mother_re = ReREFCNT_inc(r->mother_re ? r->mother_re : rx);
16111 SvREFCNT_inc_void(ret->qr_anoncv);
16117 /* regfree_internal()
16119 Free the private data in a regexp. This is overloadable by
16120 extensions. Perl takes care of the regexp structure in pregfree(),
16121 this covers the *pprivate pointer which technically perl doesn't
16122 know about, however of course we have to handle the
16123 regexp_internal structure when no extension is in use.
16125 Note this is called before freeing anything in the regexp
16130 Perl_regfree_internal(pTHX_ REGEXP * const rx)
16132 struct regexp *const r = ReANY(rx);
16133 RXi_GET_DECL(r,ri);
16134 GET_RE_DEBUG_FLAGS_DECL;
16136 PERL_ARGS_ASSERT_REGFREE_INTERNAL;
16142 SV *dsv= sv_newmortal();
16143 RE_PV_QUOTED_DECL(s, RX_UTF8(rx),
16144 dsv, RX_PRECOMP(rx), RX_PRELEN(rx), 60);
16145 PerlIO_printf(Perl_debug_log,"%sFreeing REx:%s %s\n",
16146 PL_colors[4],PL_colors[5],s);
16149 #ifdef RE_TRACK_PATTERN_OFFSETS
16151 Safefree(ri->u.offsets); /* 20010421 MJD */
16153 if (ri->code_blocks) {
16155 for (n = 0; n < ri->num_code_blocks; n++)
16156 SvREFCNT_dec(ri->code_blocks[n].src_regex);
16157 Safefree(ri->code_blocks);
16161 int n = ri->data->count;
16164 /* If you add a ->what type here, update the comment in regcomp.h */
16165 switch (ri->data->what[n]) {
16171 SvREFCNT_dec(MUTABLE_SV(ri->data->data[n]));
16174 Safefree(ri->data->data[n]);
16180 { /* Aho Corasick add-on structure for a trie node.
16181 Used in stclass optimization only */
16183 reg_ac_data *aho=(reg_ac_data*)ri->data->data[n];
16184 #ifdef USE_ITHREADS
16188 refcount = --aho->refcount;
16191 PerlMemShared_free(aho->states);
16192 PerlMemShared_free(aho->fail);
16193 /* do this last!!!! */
16194 PerlMemShared_free(ri->data->data[n]);
16195 /* we should only ever get called once, so
16196 * assert as much, and also guard the free
16197 * which /might/ happen twice. At the least
16198 * it will make code anlyzers happy and it
16199 * doesn't cost much. - Yves */
16200 assert(ri->regstclass);
16201 if (ri->regstclass) {
16202 PerlMemShared_free(ri->regstclass);
16203 ri->regstclass = 0;
16210 /* trie structure. */
16212 reg_trie_data *trie=(reg_trie_data*)ri->data->data[n];
16213 #ifdef USE_ITHREADS
16217 refcount = --trie->refcount;
16220 PerlMemShared_free(trie->charmap);
16221 PerlMemShared_free(trie->states);
16222 PerlMemShared_free(trie->trans);
16224 PerlMemShared_free(trie->bitmap);
16226 PerlMemShared_free(trie->jump);
16227 PerlMemShared_free(trie->wordinfo);
16228 /* do this last!!!! */
16229 PerlMemShared_free(ri->data->data[n]);
16234 Perl_croak(aTHX_ "panic: regfree data code '%c'",
16235 ri->data->what[n]);
16238 Safefree(ri->data->what);
16239 Safefree(ri->data);
16245 #define av_dup_inc(s,t) MUTABLE_AV(sv_dup_inc((const SV *)s,t))
16246 #define hv_dup_inc(s,t) MUTABLE_HV(sv_dup_inc((const SV *)s,t))
16247 #define SAVEPVN(p,n) ((p) ? savepvn(p,n) : NULL)
16250 re_dup - duplicate a regexp.
16252 This routine is expected to clone a given regexp structure. It is only
16253 compiled under USE_ITHREADS.
16255 After all of the core data stored in struct regexp is duplicated
16256 the regexp_engine.dupe method is used to copy any private data
16257 stored in the *pprivate pointer. This allows extensions to handle
16258 any duplication it needs to do.
16260 See pregfree() and regfree_internal() if you change anything here.
16262 #if defined(USE_ITHREADS)
16263 #ifndef PERL_IN_XSUB_RE
16265 Perl_re_dup_guts(pTHX_ const REGEXP *sstr, REGEXP *dstr, CLONE_PARAMS *param)
16269 const struct regexp *r = ReANY(sstr);
16270 struct regexp *ret = ReANY(dstr);
16272 PERL_ARGS_ASSERT_RE_DUP_GUTS;
16274 npar = r->nparens+1;
16275 Newx(ret->offs, npar, regexp_paren_pair);
16276 Copy(r->offs, ret->offs, npar, regexp_paren_pair);
16278 if (ret->substrs) {
16279 /* Do it this way to avoid reading from *r after the StructCopy().
16280 That way, if any of the sv_dup_inc()s dislodge *r from the L1
16281 cache, it doesn't matter. */
16282 const bool anchored = r->check_substr
16283 ? r->check_substr == r->anchored_substr
16284 : r->check_utf8 == r->anchored_utf8;
16285 Newx(ret->substrs, 1, struct reg_substr_data);
16286 StructCopy(r->substrs, ret->substrs, struct reg_substr_data);
16288 ret->anchored_substr = sv_dup_inc(ret->anchored_substr, param);
16289 ret->anchored_utf8 = sv_dup_inc(ret->anchored_utf8, param);
16290 ret->float_substr = sv_dup_inc(ret->float_substr, param);
16291 ret->float_utf8 = sv_dup_inc(ret->float_utf8, param);
16293 /* check_substr and check_utf8, if non-NULL, point to either their
16294 anchored or float namesakes, and don't hold a second reference. */
16296 if (ret->check_substr) {
16298 assert(r->check_utf8 == r->anchored_utf8);
16299 ret->check_substr = ret->anchored_substr;
16300 ret->check_utf8 = ret->anchored_utf8;
16302 assert(r->check_substr == r->float_substr);
16303 assert(r->check_utf8 == r->float_utf8);
16304 ret->check_substr = ret->float_substr;
16305 ret->check_utf8 = ret->float_utf8;
16307 } else if (ret->check_utf8) {
16309 ret->check_utf8 = ret->anchored_utf8;
16311 ret->check_utf8 = ret->float_utf8;
16316 RXp_PAREN_NAMES(ret) = hv_dup_inc(RXp_PAREN_NAMES(ret), param);
16317 ret->qr_anoncv = MUTABLE_CV(sv_dup_inc((const SV *)ret->qr_anoncv, param));
16320 RXi_SET(ret,CALLREGDUPE_PVT(dstr,param));
16322 if (RX_MATCH_COPIED(dstr))
16323 ret->subbeg = SAVEPVN(ret->subbeg, ret->sublen);
16325 ret->subbeg = NULL;
16326 #ifdef PERL_ANY_COW
16327 ret->saved_copy = NULL;
16330 /* Whether mother_re be set or no, we need to copy the string. We
16331 cannot refrain from copying it when the storage points directly to
16332 our mother regexp, because that's
16333 1: a buffer in a different thread
16334 2: something we no longer hold a reference on
16335 so we need to copy it locally. */
16336 RX_WRAPPED(dstr) = SAVEPVN(RX_WRAPPED(sstr), SvCUR(sstr)+1);
16337 ret->mother_re = NULL;
16339 #endif /* PERL_IN_XSUB_RE */
16344 This is the internal complement to regdupe() which is used to copy
16345 the structure pointed to by the *pprivate pointer in the regexp.
16346 This is the core version of the extension overridable cloning hook.
16347 The regexp structure being duplicated will be copied by perl prior
16348 to this and will be provided as the regexp *r argument, however
16349 with the /old/ structures pprivate pointer value. Thus this routine
16350 may override any copying normally done by perl.
16352 It returns a pointer to the new regexp_internal structure.
16356 Perl_regdupe_internal(pTHX_ REGEXP * const rx, CLONE_PARAMS *param)
16359 struct regexp *const r = ReANY(rx);
16360 regexp_internal *reti;
16362 RXi_GET_DECL(r,ri);
16364 PERL_ARGS_ASSERT_REGDUPE_INTERNAL;
16368 Newxc(reti, sizeof(regexp_internal) + len*sizeof(regnode),
16369 char, regexp_internal);
16370 Copy(ri->program, reti->program, len+1, regnode);
16372 reti->num_code_blocks = ri->num_code_blocks;
16373 if (ri->code_blocks) {
16375 Newxc(reti->code_blocks, ri->num_code_blocks, struct reg_code_block,
16376 struct reg_code_block);
16377 Copy(ri->code_blocks, reti->code_blocks, ri->num_code_blocks,
16378 struct reg_code_block);
16379 for (n = 0; n < ri->num_code_blocks; n++)
16380 reti->code_blocks[n].src_regex = (REGEXP*)
16381 sv_dup_inc((SV*)(ri->code_blocks[n].src_regex), param);
16384 reti->code_blocks = NULL;
16386 reti->regstclass = NULL;
16389 struct reg_data *d;
16390 const int count = ri->data->count;
16393 Newxc(d, sizeof(struct reg_data) + count*sizeof(void *),
16394 char, struct reg_data);
16395 Newx(d->what, count, U8);
16398 for (i = 0; i < count; i++) {
16399 d->what[i] = ri->data->what[i];
16400 switch (d->what[i]) {
16401 /* see also regcomp.h and regfree_internal() */
16402 case 'a': /* actually an AV, but the dup function is identical. */
16406 case 'u': /* actually an HV, but the dup function is identical. */
16407 d->data[i] = sv_dup_inc((const SV *)ri->data->data[i], param);
16410 /* This is cheating. */
16411 Newx(d->data[i], 1, regnode_ssc);
16412 StructCopy(ri->data->data[i], d->data[i], regnode_ssc);
16413 reti->regstclass = (regnode*)d->data[i];
16416 /* Trie stclasses are readonly and can thus be shared
16417 * without duplication. We free the stclass in pregfree
16418 * when the corresponding reg_ac_data struct is freed.
16420 reti->regstclass= ri->regstclass;
16424 ((reg_trie_data*)ri->data->data[i])->refcount++;
16429 d->data[i] = ri->data->data[i];
16432 Perl_croak(aTHX_ "panic: re_dup unknown data code '%c'",
16433 ri->data->what[i]);
16442 reti->name_list_idx = ri->name_list_idx;
16444 #ifdef RE_TRACK_PATTERN_OFFSETS
16445 if (ri->u.offsets) {
16446 Newx(reti->u.offsets, 2*len+1, U32);
16447 Copy(ri->u.offsets, reti->u.offsets, 2*len+1, U32);
16450 SetProgLen(reti,len);
16453 return (void*)reti;
16456 #endif /* USE_ITHREADS */
16458 #ifndef PERL_IN_XSUB_RE
16461 - regnext - dig the "next" pointer out of a node
16464 Perl_regnext(pTHX_ regnode *p)
16471 if (OP(p) > REGNODE_MAX) { /* regnode.type is unsigned */
16472 Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d",
16473 (int)OP(p), (int)REGNODE_MAX);
16476 offset = (reg_off_by_arg[OP(p)] ? ARG(p) : NEXT_OFF(p));
16485 S_re_croak2(pTHX_ bool utf8, const char* pat1,const char* pat2,...)
16488 STRLEN l1 = strlen(pat1);
16489 STRLEN l2 = strlen(pat2);
16492 const char *message;
16494 PERL_ARGS_ASSERT_RE_CROAK2;
16500 Copy(pat1, buf, l1 , char);
16501 Copy(pat2, buf + l1, l2 , char);
16502 buf[l1 + l2] = '\n';
16503 buf[l1 + l2 + 1] = '\0';
16504 va_start(args, pat2);
16505 msv = vmess(buf, &args);
16507 message = SvPV_const(msv,l1);
16510 Copy(message, buf, l1 , char);
16511 /* l1-1 to avoid \n */
16512 Perl_croak(aTHX_ "%"UTF8f, UTF8fARG(utf8, l1-1, buf));
16515 /* XXX Here's a total kludge. But we need to re-enter for swash routines. */
16517 #ifndef PERL_IN_XSUB_RE
16519 Perl_save_re_context(pTHX)
16521 /* Save $1..$n (#18107: UTF-8 s/(\w+)/uc($1)/e); AMS 20021106. */
16523 const REGEXP * const rx = PM_GETRE(PL_curpm);
16526 for (i = 1; i <= RX_NPARENS(rx); i++) {
16527 char digits[TYPE_CHARS(long)];
16528 const STRLEN len = my_snprintf(digits, sizeof(digits),
16530 GV *const *const gvp
16531 = (GV**)hv_fetch(PL_defstash, digits, len, 0);
16534 GV * const gv = *gvp;
16535 if (SvTYPE(gv) == SVt_PVGV && GvSV(gv))
16547 S_put_byte(pTHX_ SV *sv, int c)
16549 PERL_ARGS_ASSERT_PUT_BYTE;
16553 case '\r': Perl_sv_catpvf(aTHX_ sv, "\\r"); break;
16554 case '\n': Perl_sv_catpvf(aTHX_ sv, "\\n"); break;
16555 case '\t': Perl_sv_catpvf(aTHX_ sv, "\\t"); break;
16556 case '\f': Perl_sv_catpvf(aTHX_ sv, "\\f"); break;
16557 case '\a': Perl_sv_catpvf(aTHX_ sv, "\\a"); break;
16560 Perl_sv_catpvf(aTHX_ sv, "\\x{%x}", c);
16565 const char string = c;
16566 if (c == '-' || c == ']' || c == '\\' || c == '^')
16567 sv_catpvs(sv, "\\");
16568 sv_catpvn(sv, &string, 1);
16573 S_put_range(pTHX_ SV *sv, UV start, UV end)
16576 /* Appends to 'sv' a displayable version of the range of code points from
16577 * 'start' to 'end'. It assumes that only ASCII printables are displayable
16578 * as-is (though some of these will be escaped by put_byte()). For the
16579 * time being, this subroutine only works for latin1 (< 256) code points */
16581 assert(start <= end);
16583 PERL_ARGS_ASSERT_PUT_RANGE;
16585 while (start <= end) {
16586 if (end - start < 3) { /* Individual chars in short ranges */
16587 for (; start <= end; start++) {
16588 put_byte(sv, start);
16593 /* For small ranges that include printable ASCII characters, it's more
16594 * legible to print those characters rather than hex values. For
16595 * larger ranges that include more than printables, it's probably
16596 * clearer to just give the start and end points of the range in hex,
16597 * and that's all we can do if there aren't any printables within the
16600 * On ASCII platforms the range of printables is contiguous. If the
16601 * entire range is printable, we print each character as such. If the
16602 * range is partially printable and partially not, it's less likely
16603 * that the individual printables are meaningful, especially if all or
16604 * almost all of them are in the range. But we err on the side of the
16605 * individual printables being meaningful by using the hex only if the
16606 * range contains all but 2 of the printables.
16608 * On EBCDIC platforms, the printables are scattered around so that the
16609 * maximum range length containing only them is about 10. Anything
16610 * longer we treat as hex; otherwise we examine the range character by
16611 * character to see */
16613 if (start < 256 && (((end < 255) ? end : 255) - start <= 10))
16615 if ((isPRINT_A(start) && isPRINT_A(end))
16616 || (end >= 0x7F && (isPRINT_A(start) && start > 0x21))
16617 || ((end < 0x7D && isPRINT_A(end)) && start < 0x20))
16620 /* If the range beginning isn't an ASCII printable, we find the
16621 * last such in the range, then split the output, so all the
16622 * non-printables are in one subrange; then process the remaining
16623 * portion as usual. If the entire range isn't printables, we
16624 * don't split, but drop down to print as hex */
16625 if (! isPRINT_A(start)) {
16626 UV temp_end = start + 1;
16627 while (temp_end <= end && ! isPRINT_A(temp_end)) {
16630 if (temp_end <= end) {
16631 put_range(sv, start, temp_end - 1);
16637 /* If the range beginning is a digit, output a subrange of just the
16638 * digits, then process the remaining portion as usual */
16639 if (isDIGIT_A(start)) {
16640 put_byte(sv, start);
16641 sv_catpvs(sv, "-");
16642 while (start <= end && isDIGIT_A(start)) start++;
16643 put_byte(sv, start - 1);
16647 /* Similarly for alphabetics. Because in both ASCII and EBCDIC,
16648 * the code points for upper and lower A-Z and a-z aren't
16649 * intermixed, the resulting subrange will consist solely of either
16650 * upper- or lower- alphabetics */
16651 if (isALPHA_A(start)) {
16652 put_byte(sv, start);
16653 sv_catpvs(sv, "-");
16654 while (start <= end && isALPHA_A(start)) start++;
16655 put_byte(sv, start - 1);
16659 /* We output any remaining printables as individual characters */
16660 if (isPUNCT_A(start) || isSPACE_A(start)) {
16661 while (start <= end && (isPUNCT_A(start) || isSPACE_A(start))) {
16662 put_byte(sv, start);
16669 /* Here is a control or non-ascii. Output the range or subrange as
16671 Perl_sv_catpvf(aTHX_ sv, "\\x{%02" UVXf "}-\\x{%02" UVXf "}",
16673 (end < 256) ? end : 255);
16679 S_put_latin1_charclass_innards(pTHX_ SV *sv, char *bitmap)
16681 /* Appends to 'sv' a displayable version of the innards of the bracketed
16682 * character class whose bitmap is 'bitmap'; Returns 'TRUE' if it actually
16683 * output anything */
16686 bool has_output_anything = FALSE;
16688 PERL_ARGS_ASSERT_PUT_LATIN1_CHARCLASS_INNARDS;
16690 for (i = 0; i < 256; i++) {
16691 if (BITMAP_TEST((U8 *) bitmap,i)) {
16693 /* The character at index i should be output. Find the next
16694 * character that should NOT be output */
16696 for (j = i + 1; j < 256; j++) {
16697 if (! BITMAP_TEST((U8 *) bitmap, j)) {
16702 /* Everything between them is a single range that should be output
16704 put_range(sv, i, j - 1);
16705 has_output_anything = TRUE;
16710 return has_output_anything;
16713 #define CLEAR_OPTSTART \
16714 if (optstart) STMT_START { \
16715 DEBUG_OPTIMISE_r(PerlIO_printf(Perl_debug_log, \
16716 " (%"IVdf" nodes)\n", (IV)(node - optstart))); \
16720 #define DUMPUNTIL(b,e) \
16722 node=dumpuntil(r,start,(b),(e),last,sv,indent+1,depth+1);
16724 STATIC const regnode *
16725 S_dumpuntil(pTHX_ const regexp *r, const regnode *start, const regnode *node,
16726 const regnode *last, const regnode *plast,
16727 SV* sv, I32 indent, U32 depth)
16729 U8 op = PSEUDO; /* Arbitrary non-END op. */
16730 const regnode *next;
16731 const regnode *optstart= NULL;
16733 RXi_GET_DECL(r,ri);
16734 GET_RE_DEBUG_FLAGS_DECL;
16736 PERL_ARGS_ASSERT_DUMPUNTIL;
16738 #ifdef DEBUG_DUMPUNTIL
16739 PerlIO_printf(Perl_debug_log, "--- %d : %d - %d - %d\n",indent,node-start,
16740 last ? last-start : 0,plast ? plast-start : 0);
16743 if (plast && plast < last)
16746 while (PL_regkind[op] != END && (!last || node < last)) {
16748 /* While that wasn't END last time... */
16751 if (op == CLOSE || op == WHILEM)
16753 next = regnext((regnode *)node);
16756 if (OP(node) == OPTIMIZED) {
16757 if (!optstart && RE_DEBUG_FLAG(RE_DEBUG_COMPILE_OPTIMISE))
16764 regprop(r, sv, node, NULL);
16765 PerlIO_printf(Perl_debug_log, "%4"IVdf":%*s%s", (IV)(node - start),
16766 (int)(2*indent + 1), "", SvPVX_const(sv));
16768 if (OP(node) != OPTIMIZED) {
16769 if (next == NULL) /* Next ptr. */
16770 PerlIO_printf(Perl_debug_log, " (0)");
16771 else if (PL_regkind[(U8)op] == BRANCH
16772 && PL_regkind[OP(next)] != BRANCH )
16773 PerlIO_printf(Perl_debug_log, " (FAIL)");
16775 PerlIO_printf(Perl_debug_log, " (%"IVdf")", (IV)(next - start));
16776 (void)PerlIO_putc(Perl_debug_log, '\n');
16780 if (PL_regkind[(U8)op] == BRANCHJ) {
16783 const regnode *nnode = (OP(next) == LONGJMP
16784 ? regnext((regnode *)next)
16786 if (last && nnode > last)
16788 DUMPUNTIL(NEXTOPER(NEXTOPER(node)), nnode);
16791 else if (PL_regkind[(U8)op] == BRANCH) {
16793 DUMPUNTIL(NEXTOPER(node), next);
16795 else if ( PL_regkind[(U8)op] == TRIE ) {
16796 const regnode *this_trie = node;
16797 const char op = OP(node);
16798 const U32 n = ARG(node);
16799 const reg_ac_data * const ac = op>=AHOCORASICK ?
16800 (reg_ac_data *)ri->data->data[n] :
16802 const reg_trie_data * const trie =
16803 (reg_trie_data*)ri->data->data[op<AHOCORASICK ? n : ac->trie];
16805 AV *const trie_words
16806 = MUTABLE_AV(ri->data->data[n + TRIE_WORDS_OFFSET]);
16808 const regnode *nextbranch= NULL;
16811 for (word_idx= 0; word_idx < (I32)trie->wordcount; word_idx++) {
16812 SV ** const elem_ptr = av_fetch(trie_words,word_idx,0);
16814 PerlIO_printf(Perl_debug_log, "%*s%s ",
16815 (int)(2*(indent+3)), "",
16817 ? pv_pretty(sv, SvPV_nolen_const(*elem_ptr),
16818 SvCUR(*elem_ptr), 60,
16819 PL_colors[0], PL_colors[1],
16821 ? PERL_PV_ESCAPE_UNI
16823 | PERL_PV_PRETTY_ELLIPSES
16824 | PERL_PV_PRETTY_LTGT
16829 U16 dist= trie->jump[word_idx+1];
16830 PerlIO_printf(Perl_debug_log, "(%"UVuf")\n",
16831 (UV)((dist ? this_trie + dist : next) - start));
16834 nextbranch= this_trie + trie->jump[0];
16835 DUMPUNTIL(this_trie + dist, nextbranch);
16837 if (nextbranch && PL_regkind[OP(nextbranch)]==BRANCH)
16838 nextbranch= regnext((regnode *)nextbranch);
16840 PerlIO_printf(Perl_debug_log, "\n");
16843 if (last && next > last)
16848 else if ( op == CURLY ) { /* "next" might be very big: optimizer */
16849 DUMPUNTIL(NEXTOPER(node) + EXTRA_STEP_2ARGS,
16850 NEXTOPER(node) + EXTRA_STEP_2ARGS + 1);
16852 else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
16854 DUMPUNTIL(NEXTOPER(node) + EXTRA_STEP_2ARGS, next);
16856 else if ( op == PLUS || op == STAR) {
16857 DUMPUNTIL(NEXTOPER(node), NEXTOPER(node) + 1);
16859 else if (PL_regkind[(U8)op] == ANYOF) {
16860 /* arglen 1 + class block */
16861 node += 1 + ((ANYOF_FLAGS(node) & ANYOF_POSIXL)
16862 ? ANYOF_POSIXL_SKIP
16864 node = NEXTOPER(node);
16866 else if (PL_regkind[(U8)op] == EXACT) {
16867 /* Literal string, where present. */
16868 node += NODE_SZ_STR(node) - 1;
16869 node = NEXTOPER(node);
16872 node = NEXTOPER(node);
16873 node += regarglen[(U8)op];
16875 if (op == CURLYX || op == OPEN)
16879 #ifdef DEBUG_DUMPUNTIL
16880 PerlIO_printf(Perl_debug_log, "--- %d\n", (int)indent);
16885 #endif /* DEBUGGING */
16889 * c-indentation-style: bsd
16890 * c-basic-offset: 4
16891 * indent-tabs-mode: nil
16894 * ex: set ts=8 sts=4 sw=4 et: