summaryrefslogtreecommitdiffstats
path: root/lib/regexec.c
diff options
context:
space:
mode:
Diffstat (limited to 'lib/regexec.c')
-rw-r--r--lib/regexec.c4333
1 files changed, 0 insertions, 4333 deletions
diff --git a/lib/regexec.c b/lib/regexec.c
deleted file mode 100644
index a85077c9..00000000
--- a/lib/regexec.c
+++ /dev/null
@@ -1,4333 +0,0 @@
1/* Extended regular expression matching and search library.
2 Copyright (C) 2002, 2003, 2004, 2005 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
5
6 This program is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2, or (at your option)
9 any later version.
10
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
15
16 You should have received a copy of the GNU General Public License along
17 with this program; if not, write to the Free Software Foundation,
18 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
19
20static reg_errcode_t match_ctx_init (re_match_context_t *cache, int eflags,
21 Idx n) internal_function;
22static void match_ctx_clean (re_match_context_t *mctx) internal_function;
23static void match_ctx_free (re_match_context_t *cache) internal_function;
24static reg_errcode_t match_ctx_add_entry (re_match_context_t *cache, Idx node,
25 Idx str_idx, Idx from, Idx to)
26 internal_function;
27static Idx search_cur_bkref_entry (const re_match_context_t *mctx, Idx str_idx)
28 internal_function;
29static reg_errcode_t match_ctx_add_subtop (re_match_context_t *mctx, Idx node,
30 Idx str_idx) internal_function;
31static re_sub_match_last_t * match_ctx_add_sublast (re_sub_match_top_t *subtop,
32 Idx node, Idx str_idx)
33 internal_function;
34static void sift_ctx_init (re_sift_context_t *sctx, re_dfastate_t **sifted_sts,
35 re_dfastate_t **limited_sts, Idx last_node,
36 Idx last_str_idx)
37 internal_function;
38static reg_errcode_t re_search_internal (const regex_t *preg,
39 const char *string, Idx length,
40 Idx start, Idx last_start, Idx stop,
41 size_t nmatch, regmatch_t pmatch[],
42 int eflags) internal_function;
43static regoff_t re_search_2_stub (struct re_pattern_buffer *bufp,
44 const char *string1, Idx length1,
45 const char *string2, Idx length2,
46 Idx start, regoff_t range,
47 struct re_registers *regs,
48 Idx stop, bool ret_len) internal_function;
49static regoff_t re_search_stub (struct re_pattern_buffer *bufp,
50 const char *string, Idx length, Idx start,
51 regoff_t range, Idx stop,
52 struct re_registers *regs,
53 bool ret_len) internal_function;
54static unsigned re_copy_regs (struct re_registers *regs, regmatch_t *pmatch,
55 Idx nregs, int regs_allocated) internal_function;
56static reg_errcode_t prune_impossible_nodes (re_match_context_t *mctx)
57 internal_function;
58static Idx check_matching (re_match_context_t *mctx, bool fl_longest_match,
59 Idx *p_match_first)
60 internal_function;
61static Idx check_halt_state_context (const re_match_context_t *mctx,
62 const re_dfastate_t *state, Idx idx)
63 internal_function;
64static void update_regs (re_dfa_t *dfa, regmatch_t *pmatch,
65 regmatch_t *prev_idx_match, Idx cur_node,
66 Idx cur_idx, Idx nmatch) internal_function;
67static reg_errcode_t push_fail_stack (struct re_fail_stack_t *fs,
68 Idx str_idx, Idx dest_node, Idx nregs,
69 regmatch_t *regs,
70 re_node_set *eps_via_nodes) internal_function;
71static reg_errcode_t set_regs (const regex_t *preg,
72 const re_match_context_t *mctx,
73 size_t nmatch, regmatch_t *pmatch,
74 bool fl_backtrack) internal_function;
75static reg_errcode_t free_fail_stack_return (struct re_fail_stack_t *fs) internal_function;
76
77#ifdef RE_ENABLE_I18N
78static int sift_states_iter_mb (const re_match_context_t *mctx,
79 re_sift_context_t *sctx,
80 Idx node_idx, Idx str_idx, Idx max_str_idx) internal_function;
81#endif /* RE_ENABLE_I18N */
82static reg_errcode_t sift_states_backward (re_match_context_t *mctx,
83 re_sift_context_t *sctx) internal_function;
84static reg_errcode_t build_sifted_states (re_match_context_t *mctx,
85 re_sift_context_t *sctx, Idx str_idx,
86 re_node_set *cur_dest) internal_function;
87static reg_errcode_t update_cur_sifted_state (re_match_context_t *mctx,
88 re_sift_context_t *sctx,
89 Idx str_idx,
90 re_node_set *dest_nodes) internal_function;
91static reg_errcode_t add_epsilon_src_nodes (re_dfa_t *dfa,
92 re_node_set *dest_nodes,
93 const re_node_set *candidates) internal_function;
94static bool check_dst_limits (const re_match_context_t *mctx,
95 const re_node_set *limits,
96 Idx dst_node, Idx dst_idx, Idx src_node,
97 Idx src_idx) internal_function;
98static int check_dst_limits_calc_pos_1 (const re_match_context_t *mctx,
99 int boundaries, Idx subexp_idx,
100 Idx from_node, Idx bkref_idx) internal_function;
101static int check_dst_limits_calc_pos (const re_match_context_t *mctx,
102 Idx limit, Idx subexp_idx,
103 Idx node, Idx str_idx,
104 Idx bkref_idx) internal_function;
105static reg_errcode_t check_subexp_limits (re_dfa_t *dfa,
106 re_node_set *dest_nodes,
107 const re_node_set *candidates,
108 re_node_set *limits,
109 struct re_backref_cache_entry *bkref_ents,
110 Idx str_idx) internal_function;
111static reg_errcode_t sift_states_bkref (re_match_context_t *mctx,
112 re_sift_context_t *sctx,
113 Idx str_idx, const re_node_set *candidates) internal_function;
114static reg_errcode_t merge_state_array (re_dfa_t *dfa, re_dfastate_t **dst,
115 re_dfastate_t **src, Idx num) internal_function;
116static re_dfastate_t *find_recover_state (reg_errcode_t *err,
117 re_match_context_t *mctx) internal_function;
118static re_dfastate_t *transit_state (reg_errcode_t *err,
119 re_match_context_t *mctx,
120 re_dfastate_t *state) internal_function;
121static re_dfastate_t *merge_state_with_log (reg_errcode_t *err,
122 re_match_context_t *mctx,
123 re_dfastate_t *next_state) internal_function;
124static reg_errcode_t check_subexp_matching_top (re_match_context_t *mctx,
125 re_node_set *cur_nodes,
126 Idx str_idx) internal_function;
127#if 0
128static re_dfastate_t *transit_state_sb (reg_errcode_t *err,
129 re_match_context_t *mctx,
130 re_dfastate_t *pstate) internal_function;
131#endif
132#ifdef RE_ENABLE_I18N
133static reg_errcode_t transit_state_mb (re_match_context_t *mctx,
134 re_dfastate_t *pstate) internal_function;
135#endif /* RE_ENABLE_I18N */
136static reg_errcode_t transit_state_bkref (re_match_context_t *mctx,
137 const re_node_set *nodes) internal_function;
138static reg_errcode_t get_subexp (re_match_context_t *mctx,
139 Idx bkref_node, Idx bkref_str_idx) internal_function;
140static reg_errcode_t get_subexp_sub (re_match_context_t *mctx,
141 const re_sub_match_top_t *sub_top,
142 re_sub_match_last_t *sub_last,
143 Idx bkref_node, Idx bkref_str) internal_function;
144static Idx find_subexp_node (const re_dfa_t *dfa, const re_node_set *nodes,
145 Idx subexp_idx, int type) internal_function;
146static reg_errcode_t check_arrival (re_match_context_t *mctx,
147 state_array_t *path, Idx top_node,
148 Idx top_str, Idx last_node, Idx last_str,
149 int type) internal_function;
150static reg_errcode_t check_arrival_add_next_nodes (re_match_context_t *mctx,
151 Idx str_idx,
152 re_node_set *cur_nodes,
153 re_node_set *next_nodes) internal_function;
154static reg_errcode_t check_arrival_expand_ecl (re_dfa_t *dfa,
155 re_node_set *cur_nodes,
156 Idx ex_subexp, int type) internal_function;
157static reg_errcode_t check_arrival_expand_ecl_sub (re_dfa_t *dfa,
158 re_node_set *dst_nodes,
159 Idx target, Idx ex_subexp,
160 int type) internal_function;
161static reg_errcode_t expand_bkref_cache (re_match_context_t *mctx,
162 re_node_set *cur_nodes, Idx cur_str,
163 Idx subexp_num, int type) internal_function;
164static bool build_trtable (re_dfa_t *dfa,
165 re_dfastate_t *state) internal_function;
166#ifdef RE_ENABLE_I18N
167static int check_node_accept_bytes (re_dfa_t *dfa, Idx node_idx,
168 const re_string_t *input, Idx idx) internal_function;
169# ifdef _LIBC
170static unsigned int find_collation_sequence_value (const unsigned char *mbs,
171 size_t name_len) internal_function;
172# endif /* _LIBC */
173#endif /* RE_ENABLE_I18N */
174static Idx group_nodes_into_DFAstates (const re_dfa_t *dfa,
175 const re_dfastate_t *state,
176 re_node_set *states_node,
177 bitset *states_ch) internal_function;
178static bool check_node_accept (const re_match_context_t *mctx,
179 const re_token_t *node, Idx idx)
180 internal_function;
181static reg_errcode_t extend_buffers (re_match_context_t *mctx) internal_function;
182
183/* Entry point for POSIX code. */
184
185/* regexec searches for a given pattern, specified by PREG, in the
186 string STRING.
187
188 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
189 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
190 least NMATCH elements, and we set them to the offsets of the
191 corresponding matched substrings.
192
193 EFLAGS specifies `execution flags' which affect matching: if
194 REG_NOTBOL is set, then ^ does not match at the beginning of the
195 string; if REG_NOTEOL is set, then $ does not match at the end.
196
197 We return 0 if we find a match and REG_NOMATCH if not. */
198
199int
200regexec (const regex_t *__restrict preg, const char *__restrict string,
201 size_t nmatch, regmatch_t pmatch[], int eflags)
202{
203 reg_errcode_t err;
204 Idx start, length;
205#ifdef _LIBC
206 re_dfa_t *dfa = (re_dfa_t *) preg->re_buffer;
207#endif
208
209 if (eflags & ~(REG_NOTBOL | REG_NOTEOL | REG_STARTEND))
210 return REG_BADPAT;
211
212 if (eflags & REG_STARTEND)
213 {
214 start = pmatch[0].rm_so;
215 length = pmatch[0].rm_eo;
216 }
217 else
218 {
219 start = 0;
220 length = strlen (string);
221 }
222
223 __libc_lock_lock (dfa->lock);
224 if (preg->re_no_sub)
225 err = re_search_internal (preg, string, length, start, length,
226 length, 0, NULL, eflags);
227 else
228 err = re_search_internal (preg, string, length, start, length,
229 length, nmatch, pmatch, eflags);
230 __libc_lock_unlock (dfa->lock);
231 return err != REG_NOERROR;
232}
233
234#ifdef _LIBC
235# include <shlib-compat.h>
236versioned_symbol (libc, __regexec, regexec, GLIBC_2_3_4);
237
238# if SHLIB_COMPAT (libc, GLIBC_2_0, GLIBC_2_3_4)
239__typeof__ (__regexec) __compat_regexec;
240
241int
242attribute_compat_text_section
243__compat_regexec (const regex_t *__restrict preg,
244 const char *__restrict string, size_t nmatch,
245 regmatch_t pmatch[], int eflags)
246{
247 return regexec (preg, string, nmatch, pmatch,
248 eflags & (REG_NOTBOL | REG_NOTEOL));
249}
250compat_symbol (libc, __compat_regexec, regexec, GLIBC_2_0);
251# endif
252#endif
253
254/* Entry points for GNU code. */
255
256/* re_match, re_search, re_match_2, re_search_2
257
258 The former two functions operate on STRING with length LENGTH,
259 while the later two operate on concatenation of STRING1 and STRING2
260 with lengths LENGTH1 and LENGTH2, respectively.
261
262 re_match() matches the compiled pattern in BUFP against the string,
263 starting at index START.
264
265 re_search() first tries matching at index START, then it tries to match
266 starting from index START + 1, and so on. The last start position tried
267 is START + RANGE. (Thus RANGE = 0 forces re_search to operate the same
268 way as re_match().)
269
270 The parameter STOP of re_{match,search}_2 specifies that no match exceeding
271 the first STOP characters of the concatenation of the strings should be
272 concerned.
273
274 If REGS is not NULL, and BUFP->re_no_sub is not set, the offsets of the match
275 and all groups is stroed in REGS. (For the "_2" variants, the offsets are
276 computed relative to the concatenation, not relative to the individual
277 strings.)
278
279 On success, re_match* functions return the length of the match, re_search*
280 return the position of the start of the match. Return value -1 means no
281 match was found and -2 indicates an internal error. */
282
283regoff_t
284re_match (struct re_pattern_buffer *bufp, const char *string,
285 Idx length, Idx start, struct re_registers *regs)
286{
287 return re_search_stub (bufp, string, length, start, 0, length, regs, true);
288}
289#ifdef _LIBC
290weak_alias (__re_match, re_match)
291#endif
292
293regoff_t
294re_search (struct re_pattern_buffer *bufp, const char *string,
295 Idx length, Idx start, regoff_t range, struct re_registers *regs)
296{
297 return re_search_stub (bufp, string, length, start, range, length, regs,
298 false);
299}
300#ifdef _LIBC
301weak_alias (__re_search, re_search)
302#endif
303
304regoff_t
305re_match_2 (struct re_pattern_buffer *bufp,
306 const char *string1, Idx length1,
307 const char *string2, Idx length2,
308 Idx start, struct re_registers *regs, Idx stop)
309{
310 return re_search_2_stub (bufp, string1, length1, string2, length2,
311 start, 0, regs, stop, true);
312}
313#ifdef _LIBC
314weak_alias (__re_match_2, re_match_2)
315#endif
316
317regoff_t
318re_search_2 (struct re_pattern_buffer *bufp,
319 const char *string1, Idx length1,
320 const char *string2, Idx length2,
321 Idx start, regoff_t range, struct re_registers *regs, Idx stop)
322{
323 return re_search_2_stub (bufp, string1, length1, string2, length2,
324 start, range, regs, stop, false);
325}
326#ifdef _LIBC
327weak_alias (__re_search_2, re_search_2)
328#endif
329
330static regoff_t
331internal_function
332re_search_2_stub (struct re_pattern_buffer *bufp,
333 const char *string1, Idx length1,
334 const char *string2, Idx length2,
335 Idx start, regoff_t range, struct re_registers *regs,
336 Idx stop, bool ret_len)
337{
338 const char *str;
339 regoff_t rval;
340 Idx len = length1 + length2;
341 char *s = NULL;
342
343 if (BE (length1 < 0 || length2 < 0 || stop < 0 || len < length1, 0))
344 return -2;
345
346 /* Concatenate the strings. */
347 if (length2 > 0)
348 if (length1 > 0)
349 {
350 s = re_malloc (char, len);
351
352 if (BE (s == NULL, 0))
353 return -2;
354 memcpy (s, string1, length1);
355 memcpy (s + length1, string2, length2);
356 str = s;
357 }
358 else
359 str = string2;
360 else
361 str = string1;
362
363 rval = re_search_stub (bufp, str, len, start, range, stop, regs,
364 ret_len);
365 re_free (s);
366 return rval;
367}
368
369/* The parameters have the same meaning as those of re_search.
370 Additional parameters:
371 If RET_LEN is true the length of the match is returned (re_match style);
372 otherwise the position of the match is returned. */
373
374static regoff_t
375internal_function
376re_search_stub (struct re_pattern_buffer *bufp,
377 const char *string, Idx length,
378 Idx start, regoff_t range, Idx stop, struct re_registers *regs,
379 bool ret_len)
380{
381 reg_errcode_t result;
382 regmatch_t *pmatch;
383 Idx nregs;
384 regoff_t rval;
385 int eflags = 0;
386#ifdef _LIBC
387 re_dfa_t *dfa = (re_dfa_t *) bufp->re_buffer;
388#endif
389 Idx last_start = start + range;
390
391 /* Check for out-of-range. */
392 if (BE (start < 0 || start > length, 0))
393 return -1;
394 if (sizeof start < sizeof range)
395 {
396 regoff_t length_offset = length;
397 regoff_t start_offset = start;
398 if (BE (length_offset - start_offset < range, 0))
399 last_start = length;
400 else if (BE (range < - start_offset, 0))
401 last_start = 0;
402 }
403 else
404 {
405 if (BE ((last_start < start) != (range < 0), 0))
406 {
407 /* Overflow occurred when computing last_start; substitute
408 the extreme value. */
409 last_start = range < 0 ? 0 : length;
410 }
411 else
412 {
413 if (BE (length < last_start, 0))
414 last_start = length;
415 else if (BE (last_start < 0, 0))
416 last_start = 0;
417 }
418 }
419
420 __libc_lock_lock (dfa->lock);
421
422 eflags |= (bufp->re_not_bol) ? REG_NOTBOL : 0;
423 eflags |= (bufp->re_not_eol) ? REG_NOTEOL : 0;
424
425 /* Compile fastmap if we haven't yet. */
426 if (start < last_start && bufp->re_fastmap != NULL
427 && !bufp->re_fastmap_accurate)
428 re_compile_fastmap (bufp);
429
430 if (BE (bufp->re_no_sub, 0))
431 regs = NULL;
432
433 /* We need at least 1 register. */
434 if (regs == NULL)
435 nregs = 1;
436 else if (BE (bufp->re_regs_allocated == REG_FIXED
437 && regs->rm_num_regs <= bufp->re_nsub, 0))
438 {
439 nregs = regs->rm_num_regs;
440 if (BE (nregs < 1, 0))
441 {
442 /* Nothing can be copied to regs. */
443 regs = NULL;
444 nregs = 1;
445 }
446 }
447 else
448 nregs = bufp->re_nsub + 1;
449 pmatch = re_xmalloc (regmatch_t, nregs);
450 if (BE (pmatch == NULL, 0))
451 {
452 rval = -2;
453 goto out;
454 }
455
456 result = re_search_internal (bufp, string, length, start, last_start, stop,
457 nregs, pmatch, eflags);
458
459 rval = 0;
460
461 /* I hope we needn't fill ther regs with -1's when no match was found. */
462 if (result != REG_NOERROR)
463 rval = -1;
464 else if (regs != NULL)
465 {
466 /* If caller wants register contents data back, copy them. */
467 bufp->re_regs_allocated = re_copy_regs (regs, pmatch, nregs,
468 bufp->re_regs_allocated);
469 if (BE (bufp->re_regs_allocated == REG_UNALLOCATED, 0))
470 rval = -2;
471 }
472
473 if (BE (rval == 0, 1))
474 {
475 if (ret_len)
476 {
477 assert (pmatch[0].rm_so == start);
478 rval = pmatch[0].rm_eo - start;
479 }
480 else
481 rval = pmatch[0].rm_so;
482 }
483 re_free (pmatch);
484 out:
485 __libc_lock_unlock (dfa->lock);
486 return rval;
487}
488
489static unsigned
490internal_function
491re_copy_regs (struct re_registers *regs, regmatch_t *pmatch, Idx nregs,
492 int regs_allocated)
493{
494 int rval = REG_REALLOCATE;
495 Idx i;
496 Idx need_regs = nregs + 1;
497 /* We need one extra element beyond `rm_num_regs' for the `-1' marker GNU code
498 uses. */
499
500 /* Have the register data arrays been allocated? */
501 if (regs_allocated == REG_UNALLOCATED)
502 { /* No. So allocate them with malloc. */
503 regs->rm_start = re_xmalloc (regoff_t, need_regs);
504 regs->rm_end = re_malloc (regoff_t, need_regs);
505 if (BE (regs->rm_start == NULL, 0) || BE (regs->rm_end == NULL, 0))
506 return REG_UNALLOCATED;
507 regs->rm_num_regs = need_regs;
508 }
509 else if (regs_allocated == REG_REALLOCATE)
510 { /* Yes. If we need more elements than were already
511 allocated, reallocate them. If we need fewer, just
512 leave it alone. */
513 if (BE (need_regs > regs->rm_num_regs, 0))
514 {
515 regoff_t *new_start =
516 re_xrealloc (regs->rm_start, regoff_t, need_regs);
517 regoff_t *new_end = re_realloc (regs->rm_end, regoff_t, need_regs);
518 if (BE (new_start == NULL, 0) || BE (new_end == NULL, 0))
519 return REG_UNALLOCATED;
520 regs->rm_start = new_start;
521 regs->rm_end = new_end;
522 regs->rm_num_regs = need_regs;
523 }
524 }
525 else
526 {
527 assert (regs_allocated == REG_FIXED);
528 /* This function may not be called with REG_FIXED and nregs too big. */
529 assert (regs->rm_num_regs >= nregs);
530 rval = REG_FIXED;
531 }
532
533 /* Copy the regs. */
534 for (i = 0; i < nregs; ++i)
535 {
536 regs->rm_start[i] = pmatch[i].rm_so;
537 regs->rm_end[i] = pmatch[i].rm_eo;
538 }
539 for ( ; i < regs->rm_num_regs; ++i)
540 regs->rm_start[i] = regs->rm_end[i] = -1;
541
542 return rval;
543}
544
545/* Set REGS to hold NUM_REGS registers, storing them in STARTS and
546 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
547 this memory for recording register information. STARTS and ENDS
548 must be allocated using the malloc library routine, and must each
549 be at least NUM_REGS * sizeof (regoff_t) bytes long.
550
551 If NUM_REGS == 0, then subsequent matches should allocate their own
552 register data.
553
554 Unless this function is called, the first search or match using
555 PATTERN_BUFFER will allocate its own register data, without
556 freeing the old data. */
557
558void
559re_set_registers (struct re_pattern_buffer *bufp, struct re_registers *regs,
560 __re_size_t num_regs, regoff_t *starts, regoff_t *ends)
561{
562 if (num_regs)
563 {
564 bufp->re_regs_allocated = REG_REALLOCATE;
565 regs->rm_num_regs = num_regs;
566 regs->rm_start = starts;
567 regs->rm_end = ends;
568 }
569 else
570 {
571 bufp->re_regs_allocated = REG_UNALLOCATED;
572 regs->rm_num_regs = 0;
573 regs->rm_start = regs->rm_end = NULL;
574 }
575}
576#ifdef _LIBC
577weak_alias (__re_set_registers, re_set_registers)
578#endif
579
580/* Entry points compatible with 4.2 BSD regex library. We don't define
581 them unless specifically requested. */
582
583#if defined _REGEX_RE_COMP || defined _LIBC
584int
585# ifdef _LIBC
586weak_function
587# endif
588re_exec (const char *s)
589{
590 return 0 == regexec (&re_comp_buf, s, 0, NULL, 0);
591}
592#endif /* _REGEX_RE_COMP */
593
594/* Internal entry point. */
595
596/* Searches for a compiled pattern PREG in the string STRING, whose
597 length is LENGTH. NMATCH, PMATCH, and EFLAGS have the same
598 meaning as with regexec. LAST_START is START + RANGE, where
599 START and RANGE have the same meaning as with re_search.
600 Return REG_NOERROR if we find a match, and REG_NOMATCH if not,
601 otherwise return the error code.
602 Note: We assume front end functions already check ranges.
603 (0 <= LAST_START && LAST_START <= LENGTH) */
604
605static reg_errcode_t
606internal_function
607re_search_internal (const regex_t *preg,
608 const char *string, Idx length,
609 Idx start, Idx last_start, Idx stop,
610 size_t nmatch, regmatch_t pmatch[],
611 int eflags)
612{
613 reg_errcode_t err;
614 re_dfa_t *dfa = (re_dfa_t *) preg->re_buffer;
615 Idx left_lim, right_lim;
616 int incr;
617 bool fl_longest_match;
618 int match_kind;
619 Idx match_first, match_last = REG_MISSING;
620 Idx extra_nmatch;
621 bool sb;
622 int ch;
623#if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
624 re_match_context_t mctx = { .dfa = dfa };
625#else
626 re_match_context_t mctx;
627#endif
628 char *fastmap = ((preg->re_fastmap != NULL && preg->re_fastmap_accurate
629 && start != last_start && !preg->re_can_be_null)
630 ? preg->re_fastmap : NULL);
631 unsigned REG_TRANSLATE_TYPE t =
632 (unsigned REG_TRANSLATE_TYPE) preg->re_translate;
633
634#if !(defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L))
635 memset (&mctx, '\0', sizeof (re_match_context_t));
636 mctx.dfa = dfa;
637#endif
638
639 extra_nmatch = (nmatch > preg->re_nsub) ? nmatch - (preg->re_nsub + 1) : 0;
640 nmatch -= extra_nmatch;
641
642 /* Check if the DFA haven't been compiled. */
643 if (BE (preg->re_used == 0 || dfa->init_state == NULL
644 || dfa->init_state_word == NULL || dfa->init_state_nl == NULL
645 || dfa->init_state_begbuf == NULL, 0))
646 return REG_NOMATCH;
647
648#ifdef DEBUG
649 /* We assume front-end functions already check them. */
650 assert (0 <= last_start && last_start <= length);
651#endif
652
653 /* If initial states with non-begbuf contexts have no elements,
654 the regex must be anchored. If preg->re_newline_anchor is set,
655 we'll never use init_state_nl, so do not check it. */
656 if (dfa->init_state->nodes.nelem == 0
657 && dfa->init_state_word->nodes.nelem == 0
658 && (dfa->init_state_nl->nodes.nelem == 0
659 || !preg->re_newline_anchor))
660 {
661 if (start != 0 && last_start != 0)
662 return REG_NOMATCH;
663 start = last_start = 0;
664 }
665
666 /* We must check the longest matching, if nmatch > 0. */
667 fl_longest_match = (nmatch != 0 || dfa->nbackref);
668
669 err = re_string_allocate (&mctx.input, string, length, dfa->nodes_len + 1,
670 preg->re_translate,
671 preg->re_syntax & REG_IGNORE_CASE, dfa);
672 if (BE (err != REG_NOERROR, 0))
673 goto free_return;
674 mctx.input.stop = stop;
675 mctx.input.raw_stop = stop;
676 mctx.input.newline_anchor = preg->re_newline_anchor;
677
678 err = match_ctx_init (&mctx, eflags, dfa->nbackref * 2);
679 if (BE (err != REG_NOERROR, 0))
680 goto free_return;
681
682 /* We will log all the DFA states through which the dfa pass,
683 if nmatch > 1, or this dfa has "multibyte node", which is a
684 back-reference or a node which can accept multibyte character or
685 multi character collating element. */
686 if (nmatch > 1 || dfa->has_mb_node)
687 {
688 mctx.state_log = re_xmalloc (re_dfastate_t *, mctx.input.bufs_len + 1);
689 if (BE (mctx.state_log == NULL, 0))
690 {
691 err = REG_ESPACE;
692 goto free_return;
693 }
694 }
695 else
696 mctx.state_log = NULL;
697
698 match_first = start;
699 mctx.input.tip_context = (eflags & REG_NOTBOL) ? CONTEXT_BEGBUF
700 : CONTEXT_NEWLINE | CONTEXT_BEGBUF;
701
702 /* Check incrementally whether of not the input string match. */
703 incr = (last_start < start) ? -1 : 1;
704 left_lim = (last_start < start) ? last_start : start;
705 right_lim = (last_start < start) ? start : last_start;
706 sb = dfa->mb_cur_max == 1;
707 match_kind =
708 (fastmap
709 ? ((sb || !(preg->re_syntax & REG_IGNORE_CASE || t) ? 4 : 0)
710 | (start <= last_start ? 2 : 0)
711 | (t != NULL ? 1 : 0))
712 : 8);
713
714 for (;; match_first += incr)
715 {
716 err = REG_NOMATCH;
717 if (match_first < left_lim || right_lim < match_first)
718 goto free_return;
719
720 /* Advance as rapidly as possible through the string, until we
721 find a plausible place to start matching. This may be done
722 with varying efficiency, so there are various possibilities:
723 only the most common of them are specialized, in order to
724 save on code size. We use a switch statement for speed. */
725 switch (match_kind)
726 {
727 case 8:
728 /* No fastmap. */
729 break;
730
731 case 7:
732 /* Fastmap with single-byte translation, match forward. */
733 while (BE (match_first < right_lim, 1)
734 && !fastmap[t[(unsigned char) string[match_first]]])
735 ++match_first;
736 goto forward_match_found_start_or_reached_end;
737
738 case 6:
739 /* Fastmap without translation, match forward. */
740 while (BE (match_first < right_lim, 1)
741 && !fastmap[(unsigned char) string[match_first]])
742 ++match_first;
743
744 forward_match_found_start_or_reached_end:
745 if (BE (match_first == right_lim, 0))
746 {
747 ch = match_first >= length
748 ? 0 : (unsigned char) string[match_first];
749 if (!fastmap[t ? t[ch] : ch])
750 goto free_return;
751 }
752 break;
753
754 case 4:
755 case 5:
756 /* Fastmap without multi-byte translation, match backwards. */
757 while (match_first >= left_lim)
758 {
759 ch = match_first >= length
760 ? 0 : (unsigned char) string[match_first];
761 if (fastmap[t ? t[ch] : ch])
762 break;
763 --match_first;
764 }
765 if (match_first < left_lim)
766 goto free_return;
767 break;
768
769 default:
770 /* In this case, we can't determine easily the current byte,
771 since it might be a component byte of a multibyte
772 character. Then we use the constructed buffer instead. */
773 for (;;)
774 {
775 /* If MATCH_FIRST is out of the valid range, reconstruct the
776 buffers. */
777 __re_size_t offset = match_first - mctx.input.raw_mbs_idx;
778 if (BE (offset >= (__re_size_t) mctx.input.valid_raw_len, 0))
779 {
780 err = re_string_reconstruct (&mctx.input, match_first,
781 eflags);
782 if (BE (err != REG_NOERROR, 0))
783 goto free_return;
784
785 offset = match_first - mctx.input.raw_mbs_idx;
786 }
787 /* If MATCH_FIRST is out of the buffer, leave it as '\0'.
788 Note that MATCH_FIRST must not be smaller than 0. */
789 ch = (match_first >= length
790 ? 0 : re_string_byte_at (&mctx.input, offset));
791 if (fastmap[ch])
792 break;
793 match_first += incr;
794 if (match_first < left_lim || match_first > right_lim)
795 {
796 err = REG_NOMATCH;
797 goto free_return;
798 }
799 }
800 break;
801 }
802
803 /* Reconstruct the buffers so that the matcher can assume that
804 the matching starts from the beginning of the buffer. */
805 err = re_string_reconstruct (&mctx.input, match_first, eflags);
806 if (BE (err != REG_NOERROR, 0))
807 goto free_return;
808
809#ifdef RE_ENABLE_I18N
810 /* Don't consider this char as a possible match start if it part,
811 yet isn't the head, of a multibyte character. */
812 if (!sb && !re_string_first_byte (&mctx.input, 0))
813 continue;
814#endif
815
816 /* It seems to be appropriate one, then use the matcher. */
817 /* We assume that the matching starts from 0. */
818 mctx.state_log_top = mctx.nbkref_ents = mctx.max_mb_elem_len = 0;
819 match_last = check_matching (&mctx, fl_longest_match,
820 start <= last_start ? &match_first : NULL);
821 if (match_last != REG_MISSING)
822 {
823 if (BE (match_last == REG_ERROR, 0))
824 {
825 err = REG_ESPACE;
826 goto free_return;
827 }
828 else
829 {
830 mctx.match_last = match_last;
831 if ((!preg->re_no_sub && nmatch > 1) || dfa->nbackref)
832 {
833 re_dfastate_t *pstate = mctx.state_log[match_last];
834 mctx.last_node = check_halt_state_context (&mctx, pstate,
835 match_last);
836 }
837 if ((!preg->re_no_sub && nmatch > 1 && dfa->has_plural_match)
838 || dfa->nbackref)
839 {
840 err = prune_impossible_nodes (&mctx);
841 if (err == REG_NOERROR)
842 break;
843 if (BE (err != REG_NOMATCH, 0))
844 goto free_return;
845 match_last = REG_MISSING;
846 }
847 else
848 break; /* We found a match. */
849 }
850 }
851
852 match_ctx_clean (&mctx);
853 }
854
855#ifdef DEBUG
856 assert (match_last != REG_MISSING);
857 assert (err == REG_NOERROR);
858#endif
859
860 /* Set pmatch[] if we need. */
861 if (nmatch > 0)
862 {
863 Idx reg_idx;
864
865 /* Initialize registers. */
866 for (reg_idx = 1; reg_idx < nmatch; ++reg_idx)
867 pmatch[reg_idx].rm_so = pmatch[reg_idx].rm_eo = -1;
868
869 /* Set the points where matching start/end. */
870 pmatch[0].rm_so = 0;
871 pmatch[0].rm_eo = mctx.match_last;
872 /* FIXME: This function should fail if mctx.match_last exceeds
873 the maximum possible regoff_t value. We need a new error
874 code REG_OVERFLOW. */
875
876 if (!preg->re_no_sub && nmatch > 1)
877 {
878 err = set_regs (preg, &mctx, nmatch, pmatch,
879 dfa->has_plural_match && dfa->nbackref > 0);
880 if (BE (err != REG_NOERROR, 0))
881 goto free_return;
882 }
883
884 /* At last, add the offset to the each registers, since we slided
885 the buffers so that we could assume that the matching starts
886 from 0. */
887 for (reg_idx = 0; reg_idx < nmatch; ++reg_idx)
888 if (pmatch[reg_idx].rm_so != -1)
889 {
890#ifdef RE_ENABLE_I18N
891 if (BE (mctx.input.offsets_needed != 0, 0))
892 {
893 pmatch[reg_idx].rm_so =
894 (pmatch[reg_idx].rm_so == mctx.input.valid_len
895 ? mctx.input.valid_raw_len
896 : mctx.input.offsets[pmatch[reg_idx].rm_so]);
897 pmatch[reg_idx].rm_eo =
898 (pmatch[reg_idx].rm_eo == mctx.input.valid_len
899 ? mctx.input.valid_raw_len
900 : mctx.input.offsets[pmatch[reg_idx].rm_eo]);
901 }
902#else
903 assert (mctx.input.offsets_needed == 0);
904#endif
905 pmatch[reg_idx].rm_so += match_first;
906 pmatch[reg_idx].rm_eo += match_first;
907 }
908 for (reg_idx = 0; reg_idx < extra_nmatch; ++reg_idx)
909 {
910 pmatch[nmatch + reg_idx].rm_so = -1;
911 pmatch[nmatch + reg_idx].rm_eo = -1;
912 }
913
914 if (dfa->subexp_map)
915 for (reg_idx = 0; reg_idx + 1 < nmatch; reg_idx++)
916 if (dfa->subexp_map[reg_idx] != reg_idx)
917 {
918 pmatch[reg_idx + 1].rm_so
919 = pmatch[dfa->subexp_map[reg_idx] + 1].rm_so;
920 pmatch[reg_idx + 1].rm_eo
921 = pmatch[dfa->subexp_map[reg_idx] + 1].rm_eo;
922 }
923 }
924
925 free_return:
926 re_free (mctx.state_log);
927 if (dfa->nbackref)
928 match_ctx_free (&mctx);
929 re_string_destruct (&mctx.input);
930 return err;
931}
932
933static reg_errcode_t
934internal_function
935prune_impossible_nodes (re_match_context_t *mctx)
936{
937 re_dfa_t *const dfa = mctx->dfa;
938 Idx halt_node, match_last;
939 reg_errcode_t ret;
940 re_dfastate_t **sifted_states;
941 re_dfastate_t **lim_states = NULL;
942 re_sift_context_t sctx;
943#ifdef DEBUG
944 assert (mctx->state_log != NULL);
945#endif
946 match_last = mctx->match_last;
947 halt_node = mctx->last_node;
948 sifted_states = re_xmalloc (re_dfastate_t *, match_last + 1);
949 if (BE (sifted_states == NULL, 0))
950 {
951 ret = REG_ESPACE;
952 goto free_return;
953 }
954 if (dfa->nbackref)
955 {
956 lim_states = re_xmalloc (re_dfastate_t *, match_last + 1);
957 if (BE (lim_states == NULL, 0))
958 {
959 ret = REG_ESPACE;
960 goto free_return;
961 }
962 while (1)
963 {
964 memset (lim_states, '\0',
965 sizeof (re_dfastate_t *) * (match_last + 1));
966 sift_ctx_init (&sctx, sifted_states, lim_states, halt_node,
967 match_last);
968 ret = sift_states_backward (mctx, &sctx);
969 re_node_set_free (&sctx.limits);
970 if (BE (ret != REG_NOERROR, 0))
971 goto free_return;
972 if (sifted_states[0] != NULL || lim_states[0] != NULL)
973 break;
974 do
975 {
976 --match_last;
977 if (! REG_VALID_INDEX (match_last))
978 {
979 ret = REG_NOMATCH;
980 goto free_return;
981 }
982 } while (mctx->state_log[match_last] == NULL
983 || !mctx->state_log[match_last]->halt);
984 halt_node = check_halt_state_context (mctx,
985 mctx->state_log[match_last],
986 match_last);
987 }
988 ret = merge_state_array (dfa, sifted_states, lim_states,
989 match_last + 1);
990 re_free (lim_states);
991 lim_states = NULL;
992 if (BE (ret != REG_NOERROR, 0))
993 goto free_return;
994 }
995 else
996 {
997 sift_ctx_init (&sctx, sifted_states, lim_states, halt_node, match_last);
998 ret = sift_states_backward (mctx, &sctx);
999 re_node_set_free (&sctx.limits);
1000 if (BE (ret != REG_NOERROR, 0))
1001 goto free_return;
1002 }
1003 re_free (mctx->state_log);
1004 mctx->state_log = sifted_states;
1005 sifted_states = NULL;
1006 mctx->last_node = halt_node;
1007 mctx->match_last = match_last;
1008 ret = REG_NOERROR;
1009 free_return:
1010 re_free (sifted_states);
1011 re_free (lim_states);
1012 return ret;
1013}
1014
1015/* Acquire an initial state and return it.
1016 We must select appropriate initial state depending on the context,
1017 since initial states may have constraints like "\<", "^", etc.. */
1018
1019static inline re_dfastate_t *
1020__attribute ((always_inline)) internal_function
1021acquire_init_state_context (reg_errcode_t *err, const re_match_context_t *mctx,
1022 Idx idx)
1023{
1024 re_dfa_t *const dfa = mctx->dfa;
1025 if (dfa->init_state->has_constraint)
1026 {
1027 unsigned int context;
1028 context = re_string_context_at (&mctx->input, idx - 1, mctx->eflags);
1029 if (IS_WORD_CONTEXT (context))
1030 return dfa->init_state_word;
1031 else if (IS_ORDINARY_CONTEXT (context))
1032 return dfa->init_state;
1033 else if (IS_BEGBUF_CONTEXT (context) && IS_NEWLINE_CONTEXT (context))
1034 return dfa->init_state_begbuf;
1035 else if (IS_NEWLINE_CONTEXT (context))
1036 return dfa->init_state_nl;
1037 else if (IS_BEGBUF_CONTEXT (context))
1038 {
1039 /* It is relatively rare case, then calculate on demand. */
1040 return re_acquire_state_context (err, dfa,
1041 dfa->init_state->entrance_nodes,
1042 context);
1043 }
1044 else
1045 /* Must not happen? */
1046 return dfa->init_state;
1047 }
1048 else
1049 return dfa->init_state;
1050}
1051
1052/* Check whether the regular expression match input string INPUT or not,
1053 and return the index where the matching end. Return REG_MISSING if
1054 there is no match, and return REG_ERROR in case of an error.
1055 FL_LONGEST_MATCH means we want the POSIX longest matching.
1056 If P_MATCH_FIRST is not NULL, and the match fails, it is set to the
1057 next place where we may want to try matching.
1058 Note that the matcher assume that the maching starts from the current
1059 index of the buffer. */
1060
1061static Idx
1062internal_function
1063check_matching (re_match_context_t *mctx, bool fl_longest_match,
1064 Idx *p_match_first)
1065{
1066 re_dfa_t *const dfa = mctx->dfa;
1067 reg_errcode_t err;
1068 Idx match = 0;
1069 Idx match_last = REG_MISSING;
1070 Idx cur_str_idx = re_string_cur_idx (&mctx->input);
1071 re_dfastate_t *cur_state;
1072 bool at_init_state = p_match_first != NULL;
1073 Idx next_start_idx = cur_str_idx;
1074
1075 err = REG_NOERROR;
1076 cur_state = acquire_init_state_context (&err, mctx, cur_str_idx);
1077 /* An initial state must not be NULL (invalid). */
1078 if (BE (cur_state == NULL, 0))
1079 {
1080 assert (err == REG_ESPACE);
1081 return REG_ERROR;
1082 }
1083
1084 if (mctx->state_log != NULL)
1085 {
1086 mctx->state_log[cur_str_idx] = cur_state;
1087
1088 /* Check OP_OPEN_SUBEXP in the initial state in case that we use them
1089 later. E.g. Processing back references. */
1090 if (BE (dfa->nbackref, 0))
1091 {
1092 at_init_state = false;
1093 err = check_subexp_matching_top (mctx, &cur_state->nodes, 0);
1094 if (BE (err != REG_NOERROR, 0))
1095 return err;
1096
1097 if (cur_state->has_backref)
1098 {
1099 err = transit_state_bkref (mctx, &cur_state->nodes);
1100 if (BE (err != REG_NOERROR, 0))
1101 return err;
1102 }
1103 }
1104 }
1105
1106 /* If the RE accepts NULL string. */
1107 if (BE (cur_state->halt, 0))
1108 {
1109 if (!cur_state->has_constraint
1110 || check_halt_state_context (mctx, cur_state, cur_str_idx))
1111 {
1112 if (!fl_longest_match)
1113 return cur_str_idx;
1114 else
1115 {
1116 match_last = cur_str_idx;
1117 match = 1;
1118 }
1119 }
1120 }
1121
1122 while (!re_string_eoi (&mctx->input))
1123 {
1124 re_dfastate_t *old_state = cur_state;
1125 Idx next_char_idx = re_string_cur_idx (&mctx->input) + 1;
1126
1127 if (BE (next_char_idx >= mctx->input.bufs_len, 0)
1128 || (BE (next_char_idx >= mctx->input.valid_len, 0)
1129 && mctx->input.valid_len < mctx->input.len))
1130 {
1131 err = extend_buffers (mctx);
1132 if (BE (err != REG_NOERROR, 0))
1133 {
1134 assert (err == REG_ESPACE);
1135 return REG_ERROR;
1136 }
1137 }
1138
1139 cur_state = transit_state (&err, mctx, cur_state);
1140 if (mctx->state_log != NULL)
1141 cur_state = merge_state_with_log (&err, mctx, cur_state);
1142
1143 if (cur_state == NULL)
1144 {
1145 /* Reached the invalid state or an error. Try to recover a valid
1146 state using the state log, if available and if we have not
1147 already found a valid (even if not the longest) match. */
1148 if (BE (err != REG_NOERROR, 0))
1149 return REG_ERROR;
1150
1151 if (mctx->state_log == NULL
1152 || (match && !fl_longest_match)
1153 || (cur_state = find_recover_state (&err, mctx)) == NULL)
1154 break;
1155 }
1156
1157 if (BE (at_init_state, 0))
1158 {
1159 if (old_state == cur_state)
1160 next_start_idx = next_char_idx;
1161 else
1162 at_init_state = false;
1163 }
1164
1165 if (cur_state->halt)
1166 {
1167 /* Reached a halt state.
1168 Check the halt state can satisfy the current context. */
1169 if (!cur_state->has_constraint
1170 || check_halt_state_context (mctx, cur_state,
1171 re_string_cur_idx (&mctx->input)))
1172 {
1173 /* We found an appropriate halt state. */
1174 match_last = re_string_cur_idx (&mctx->input);
1175 match = 1;
1176
1177 /* We found a match, do not modify match_first below. */
1178 p_match_first = NULL;
1179 if (!fl_longest_match)
1180 break;
1181 }
1182 }
1183 }
1184
1185 if (p_match_first)
1186 *p_match_first += next_start_idx;
1187
1188 return match_last;
1189}
1190
1191/* Check NODE match the current context. */
1192
1193static bool
1194internal_function
1195check_halt_node_context (const re_dfa_t *dfa, Idx node, unsigned int context)
1196{
1197 re_token_type_t type = dfa->nodes[node].type;
1198 unsigned int constraint = dfa->nodes[node].constraint;
1199 if (type != END_OF_RE)
1200 return false;
1201 if (!constraint)
1202 return true;
1203 if (NOT_SATISFY_NEXT_CONSTRAINT (constraint, context))
1204 return false;
1205 return true;
1206}
1207
1208/* Check the halt state STATE match the current context.
1209 Return 0 if not match, if the node, STATE has, is a halt node and
1210 match the context, return the node. */
1211
1212static Idx
1213internal_function
1214check_halt_state_context (const re_match_context_t *mctx,
1215 const re_dfastate_t *state, Idx idx)
1216{
1217 Idx i;
1218 unsigned int context;
1219#ifdef DEBUG
1220 assert (state->halt);
1221#endif
1222 context = re_string_context_at (&mctx->input, idx, mctx->eflags);
1223 for (i = 0; i < state->nodes.nelem; ++i)
1224 if (check_halt_node_context (mctx->dfa, state->nodes.elems[i], context))
1225 return state->nodes.elems[i];
1226 return 0;
1227}
1228
1229/* Compute the next node to which "NFA" transit from NODE("NFA" is a NFA
1230 corresponding to the DFA).
1231 Return the destination node, and update EPS_VIA_NODES;
1232 return REG_MISSING in case of errors. */
1233
1234static Idx
1235internal_function
1236proceed_next_node (const re_match_context_t *mctx,
1237 Idx nregs, regmatch_t *regs, Idx *pidx, Idx node,
1238 re_node_set *eps_via_nodes, struct re_fail_stack_t *fs)
1239{
1240 re_dfa_t *const dfa = mctx->dfa;
1241 Idx i;
1242 bool ok;
1243 if (IS_EPSILON_NODE (dfa->nodes[node].type))
1244 {
1245 re_node_set *cur_nodes = &mctx->state_log[*pidx]->nodes;
1246 re_node_set *edests = &dfa->edests[node];
1247 Idx dest_node;
1248 ok = re_node_set_insert (eps_via_nodes, node);
1249 if (BE (! ok, 0))
1250 return REG_ERROR;
1251 /* Pick up a valid destination, or return REG_MISSING if none
1252 is found. */
1253 for (dest_node = REG_MISSING, i = 0; i < edests->nelem; ++i)
1254 {
1255 Idx candidate = edests->elems[i];
1256 if (!re_node_set_contains (cur_nodes, candidate))
1257 continue;
1258 if (dest_node == REG_MISSING)
1259 dest_node = candidate;
1260
1261 else
1262 {
1263 /* In order to avoid infinite loop like "(a*)*", return the second
1264 epsilon-transition if the first was already considered. */
1265 if (re_node_set_contains (eps_via_nodes, dest_node))
1266 return candidate;
1267
1268 /* Otherwise, push the second epsilon-transition on the fail stack. */
1269 else if (fs != NULL
1270 && push_fail_stack (fs, *pidx, candidate, nregs, regs,
1271 eps_via_nodes))
1272 return REG_ERROR;
1273
1274 /* We know we are going to exit. */
1275 break;
1276 }
1277 }
1278 return dest_node;
1279 }
1280 else
1281 {
1282 Idx naccepted = 0;
1283 re_token_type_t type = dfa->nodes[node].type;
1284
1285#ifdef RE_ENABLE_I18N
1286 if (dfa->nodes[node].accept_mb)
1287 naccepted = check_node_accept_bytes (dfa, node, &mctx->input, *pidx);
1288 else
1289#endif /* RE_ENABLE_I18N */
1290 if (type == OP_BACK_REF)
1291 {
1292 Idx subexp_idx = dfa->nodes[node].opr.idx + 1;
1293 naccepted = regs[subexp_idx].rm_eo - regs[subexp_idx].rm_so;
1294 if (fs != NULL)
1295 {
1296 if (regs[subexp_idx].rm_so == -1 || regs[subexp_idx].rm_eo == -1)
1297 return REG_MISSING;
1298 else if (naccepted)
1299 {
1300 char *buf = (char *) re_string_get_buffer (&mctx->input);
1301 if (memcmp (buf + regs[subexp_idx].rm_so, buf + *pidx,
1302 naccepted) != 0)
1303 return REG_MISSING;
1304 }
1305 }
1306
1307 if (naccepted == 0)
1308 {
1309 Idx dest_node;
1310 ok = re_node_set_insert (eps_via_nodes, node);
1311 if (BE (! ok, 0))
1312 return REG_ERROR;
1313 dest_node = dfa->edests[node].elems[0];
1314 if (re_node_set_contains (&mctx->state_log[*pidx]->nodes,
1315 dest_node))
1316 return dest_node;
1317 }
1318 }
1319
1320 if (naccepted != 0
1321 || check_node_accept (mctx, dfa->nodes + node, *pidx))
1322 {
1323 Idx dest_node = dfa->nexts[node];
1324 *pidx = (naccepted == 0) ? *pidx + 1 : *pidx + naccepted;
1325 if (fs && (*pidx > mctx->match_last || mctx->state_log[*pidx] == NULL
1326 || !re_node_set_contains (&mctx->state_log[*pidx]->nodes,
1327 dest_node)))
1328 return REG_MISSING;
1329 re_node_set_empty (eps_via_nodes);
1330 return dest_node;
1331 }
1332 }
1333 return REG_MISSING;
1334}
1335
1336static reg_errcode_t
1337internal_function
1338push_fail_stack (struct re_fail_stack_t *fs, Idx str_idx, Idx dest_node,
1339 Idx nregs, regmatch_t *regs, re_node_set *eps_via_nodes)
1340{
1341 reg_errcode_t err;
1342 Idx num = fs->num++;
1343 if (fs->num == fs->alloc)
1344 {
1345 struct re_fail_stack_ent_t *new_array =
1346 re_x2realloc (fs->stack, struct re_fail_stack_ent_t, &fs->alloc);
1347 if (new_array == NULL)
1348 return REG_ESPACE;
1349 fs->stack = new_array;
1350 }
1351 fs->stack[num].idx = str_idx;
1352 fs->stack[num].node = dest_node;
1353 fs->stack[num].regs = re_xmalloc (regmatch_t, nregs);
1354 if (fs->stack[num].regs == NULL)
1355 return REG_ESPACE;
1356 memcpy (fs->stack[num].regs, regs, sizeof (regmatch_t) * nregs);
1357 err = re_node_set_init_copy (&fs->stack[num].eps_via_nodes, eps_via_nodes);
1358 return err;
1359}
1360
1361static Idx
1362internal_function
1363pop_fail_stack (struct re_fail_stack_t *fs, Idx *pidx,
1364 Idx nregs, regmatch_t *regs, re_node_set *eps_via_nodes)
1365{
1366 Idx num = --fs->num;
1367 assert (REG_VALID_INDEX (num));
1368 *pidx = fs->stack[num].idx;
1369 memcpy (regs, fs->stack[num].regs, sizeof (regmatch_t) * nregs);
1370 re_node_set_free (eps_via_nodes);
1371 re_free (fs->stack[num].regs);
1372 *eps_via_nodes = fs->stack[num].eps_via_nodes;
1373 return fs->stack[num].node;
1374}
1375
1376/* Set the positions where the subexpressions are starts/ends to registers
1377 PMATCH.
1378 Note: We assume that pmatch[0] is already set, and
1379 pmatch[i].rm_so == pmatch[i].rm_eo == -1 for 0 < i < nmatch. */
1380
1381static reg_errcode_t
1382internal_function
1383set_regs (const regex_t *preg, const re_match_context_t *mctx,
1384 size_t nmatch, regmatch_t *pmatch, bool fl_backtrack)
1385{
1386 re_dfa_t *dfa = (re_dfa_t *) preg->re_buffer;
1387 Idx idx, cur_node;
1388 re_node_set eps_via_nodes;
1389 struct re_fail_stack_t *fs;
1390 struct re_fail_stack_t fs_body = { 0, 2, NULL };
1391 regmatch_t *prev_idx_match;
1392 bool prev_idx_match_malloced = false;
1393
1394#ifdef DEBUG
1395 assert (nmatch > 1);
1396 assert (mctx->state_log != NULL);
1397#endif
1398 if (fl_backtrack)
1399 {
1400 fs = &fs_body;
1401 fs->stack = re_xmalloc (struct re_fail_stack_ent_t, fs->alloc);
1402 if (fs->stack == NULL)
1403 return REG_ESPACE;
1404 }
1405 else
1406 fs = NULL;
1407
1408 cur_node = dfa->init_node;
1409 re_node_set_init_empty (&eps_via_nodes);
1410
1411 if (re_alloc_oversized (nmatch, sizeof (regmatch_t)))
1412 {
1413 free_fail_stack_return (fs);
1414 return REG_ESPACE;
1415 }
1416 if (__libc_use_alloca (nmatch * sizeof (regmatch_t)))
1417 prev_idx_match = (regmatch_t *) alloca (nmatch * sizeof (regmatch_t));
1418 else
1419 {
1420 prev_idx_match = re_malloc (regmatch_t, nmatch);
1421 if (prev_idx_match == NULL)
1422 {
1423 free_fail_stack_return (fs);
1424 return REG_ESPACE;
1425 }
1426 prev_idx_match_malloced = true;
1427 }
1428 memcpy (prev_idx_match, pmatch, sizeof (regmatch_t) * nmatch);
1429
1430 for (idx = pmatch[0].rm_so; idx <= pmatch[0].rm_eo ;)
1431 {
1432 update_regs (dfa, pmatch, prev_idx_match, cur_node, idx, nmatch);
1433
1434 if (idx == pmatch[0].rm_eo && cur_node == mctx->last_node)
1435 {
1436 Idx reg_idx;
1437 if (fs)
1438 {
1439 for (reg_idx = 0; reg_idx < nmatch; ++reg_idx)
1440 if (pmatch[reg_idx].rm_so > -1 && pmatch[reg_idx].rm_eo == -1)
1441 break;
1442 if (reg_idx == nmatch)
1443 {
1444 re_node_set_free (&eps_via_nodes);
1445 if (prev_idx_match_malloced)
1446 re_free (prev_idx_match);
1447 return free_fail_stack_return (fs);
1448 }
1449 cur_node = pop_fail_stack (fs, &idx, nmatch, pmatch,
1450 &eps_via_nodes);
1451 }
1452 else
1453 {
1454 re_node_set_free (&eps_via_nodes);
1455 if (prev_idx_match_malloced)
1456 re_free (prev_idx_match);
1457 return REG_NOERROR;
1458 }
1459 }
1460
1461 /* Proceed to next node. */
1462 cur_node = proceed_next_node (mctx, nmatch, pmatch, &idx, cur_node,
1463 &eps_via_nodes, fs);
1464
1465 if (BE (! REG_VALID_INDEX (cur_node), 0))
1466 {
1467 if (BE (cur_node == REG_ERROR, 0))
1468 {
1469 re_node_set_free (&eps_via_nodes);
1470 if (prev_idx_match_malloced)
1471 re_free (prev_idx_match);
1472 free_fail_stack_return (fs);
1473 return REG_ESPACE;
1474 }
1475 if (fs)
1476 cur_node = pop_fail_stack (fs, &idx, nmatch, pmatch,
1477 &eps_via_nodes);
1478 else
1479 {
1480 re_node_set_free (&eps_via_nodes);
1481 if (prev_idx_match_malloced)
1482 re_free (prev_idx_match);
1483 return REG_NOMATCH;
1484 }
1485 }
1486 }
1487 re_node_set_free (&eps_via_nodes);
1488 if (prev_idx_match_malloced)
1489 re_free (prev_idx_match);
1490 return free_fail_stack_return (fs);
1491}
1492
1493static reg_errcode_t
1494internal_function
1495free_fail_stack_return (struct re_fail_stack_t *fs)
1496{
1497 if (fs)
1498 {
1499 Idx fs_idx;
1500 for (fs_idx = 0; fs_idx < fs->num; ++fs_idx)
1501 {
1502 re_node_set_free (&fs->stack[fs_idx].eps_via_nodes);
1503 re_free (fs->stack[fs_idx].regs);
1504 }
1505 re_free (fs->stack);
1506 }
1507 return REG_NOERROR;
1508}
1509
1510static void
1511internal_function
1512update_regs (re_dfa_t *dfa, regmatch_t *pmatch, regmatch_t *prev_idx_match,
1513 Idx cur_node, Idx cur_idx, Idx nmatch)
1514{
1515 int type = dfa->nodes[cur_node].type;
1516 if (type == OP_OPEN_SUBEXP)
1517 {
1518 Idx reg_num = dfa->nodes[cur_node].opr.idx + 1;
1519
1520 /* We are at the first node of this sub expression. */
1521 if (reg_num < nmatch)
1522 {
1523 pmatch[reg_num].rm_so = cur_idx;
1524 pmatch[reg_num].rm_eo = -1;
1525 }
1526 }
1527 else if (type == OP_CLOSE_SUBEXP)
1528 {
1529 Idx reg_num = dfa->nodes[cur_node].opr.idx + 1;
1530 if (reg_num < nmatch)
1531 {
1532 /* We are at the last node of this sub expression. */
1533 if (pmatch[reg_num].rm_so < cur_idx)
1534 {
1535 pmatch[reg_num].rm_eo = cur_idx;
1536 /* This is a non-empty match or we are not inside an optional
1537 subexpression. Accept this right away. */
1538 memcpy (prev_idx_match, pmatch, sizeof (regmatch_t) * nmatch);
1539 }
1540 else
1541 {
1542 if (dfa->nodes[cur_node].opt_subexp
1543 && prev_idx_match[reg_num].rm_so != -1)
1544 /* We transited through an empty match for an optional
1545 subexpression, like (a?)*, and this is not the subexp's
1546 first match. Copy back the old content of the registers
1547 so that matches of an inner subexpression are undone as
1548 well, like in ((a?))*. */
1549 memcpy (pmatch, prev_idx_match, sizeof (regmatch_t) * nmatch);
1550 else
1551 /* We completed a subexpression, but it may be part of
1552 an optional one, so do not update PREV_IDX_MATCH. */
1553 pmatch[reg_num].rm_eo = cur_idx;
1554 }
1555 }
1556 }
1557}
1558
1559/* This function checks the STATE_LOG from the SCTX->last_str_idx to 0
1560 and sift the nodes in each states according to the following rules.
1561 Updated state_log will be wrote to STATE_LOG.
1562
1563 Rules: We throw away the Node `a' in the STATE_LOG[STR_IDX] if...
1564 1. When STR_IDX == MATCH_LAST(the last index in the state_log):
1565 If `a' isn't the LAST_NODE and `a' can't epsilon transit to
1566 the LAST_NODE, we throw away the node `a'.
1567 2. When 0 <= STR_IDX < MATCH_LAST and `a' accepts
1568 string `s' and transit to `b':
1569 i. If 'b' isn't in the STATE_LOG[STR_IDX+strlen('s')], we throw
1570 away the node `a'.
1571 ii. If 'b' is in the STATE_LOG[STR_IDX+strlen('s')] but 'b' is
1572 thrown away, we throw away the node `a'.
1573 3. When 0 <= STR_IDX < MATCH_LAST and 'a' epsilon transit to 'b':
1574 i. If 'b' isn't in the STATE_LOG[STR_IDX], we throw away the
1575 node `a'.
1576 ii. If 'b' is in the STATE_LOG[STR_IDX] but 'b' is thrown away,
1577 we throw away the node `a'. */
1578
1579#define STATE_NODE_CONTAINS(state,node) \
1580 ((state) != NULL && re_node_set_contains (&(state)->nodes, node))
1581
1582static reg_errcode_t
1583internal_function
1584sift_states_backward (re_match_context_t *mctx, re_sift_context_t *sctx)
1585{
1586 reg_errcode_t err;
1587 int null_cnt = 0;
1588 Idx str_idx = sctx->last_str_idx;
1589 re_node_set cur_dest;
1590
1591#ifdef DEBUG
1592 assert (mctx->state_log != NULL && mctx->state_log[str_idx] != NULL);
1593#endif
1594
1595 /* Build sifted state_log[str_idx]. It has the nodes which can epsilon
1596 transit to the last_node and the last_node itself. */
1597 err = re_node_set_init_1 (&cur_dest, sctx->last_node);
1598 if (BE (err != REG_NOERROR, 0))
1599 return err;
1600 err = update_cur_sifted_state (mctx, sctx, str_idx, &cur_dest);
1601 if (BE (err != REG_NOERROR, 0))
1602 goto free_return;
1603
1604 /* Then check each states in the state_log. */
1605 while (str_idx > 0)
1606 {
1607 /* Update counters. */
1608 null_cnt = (sctx->sifted_states[str_idx] == NULL) ? null_cnt + 1 : 0;
1609 if (null_cnt > mctx->max_mb_elem_len)
1610 {
1611 memset (sctx->sifted_states, '\0',
1612 sizeof (re_dfastate_t *) * str_idx);
1613 re_node_set_free (&cur_dest);
1614 return REG_NOERROR;
1615 }
1616 re_node_set_empty (&cur_dest);
1617 --str_idx;
1618
1619 if (mctx->state_log[str_idx])
1620 {
1621 err = build_sifted_states (mctx, sctx, str_idx, &cur_dest);
1622 if (BE (err != REG_NOERROR, 0))
1623 goto free_return;
1624 }
1625
1626 /* Add all the nodes which satisfy the following conditions:
1627 - It can epsilon transit to a node in CUR_DEST.
1628 - It is in CUR_SRC.
1629 And update state_log. */
1630 err = update_cur_sifted_state (mctx, sctx, str_idx, &cur_dest);
1631 if (BE (err != REG_NOERROR, 0))
1632 goto free_return;
1633 }
1634 err = REG_NOERROR;
1635 free_return:
1636 re_node_set_free (&cur_dest);
1637 return err;
1638}
1639
1640static reg_errcode_t
1641internal_function
1642build_sifted_states (re_match_context_t *mctx, re_sift_context_t *sctx,
1643 Idx str_idx, re_node_set *cur_dest)
1644{
1645 re_dfa_t *const dfa = mctx->dfa;
1646 re_node_set *cur_src = &mctx->state_log[str_idx]->non_eps_nodes;
1647 Idx i;
1648
1649 /* Then build the next sifted state.
1650 We build the next sifted state on `cur_dest', and update
1651 `sifted_states[str_idx]' with `cur_dest'.
1652 Note:
1653 `cur_dest' is the sifted state from `state_log[str_idx + 1]'.
1654 `cur_src' points the node_set of the old `state_log[str_idx]'
1655 (with the epsilon nodes pre-filtered out). */
1656 for (i = 0; i < cur_src->nelem; i++)
1657 {
1658 Idx prev_node = cur_src->elems[i];
1659 int naccepted = 0;
1660 bool ok;
1661
1662#ifdef DEBUG
1663 re_token_type_t type = dfa->nodes[prev_node].type;
1664 assert (!IS_EPSILON_NODE (type));
1665#endif
1666#ifdef RE_ENABLE_I18N
1667 /* If the node may accept `multi byte'. */
1668 if (dfa->nodes[prev_node].accept_mb)
1669 naccepted = sift_states_iter_mb (mctx, sctx, prev_node,
1670 str_idx, sctx->last_str_idx);
1671#endif /* RE_ENABLE_I18N */
1672
1673 /* We don't check backreferences here.
1674 See update_cur_sifted_state(). */
1675 if (!naccepted
1676 && check_node_accept (mctx, dfa->nodes + prev_node, str_idx)
1677 && STATE_NODE_CONTAINS (sctx->sifted_states[str_idx + 1],
1678 dfa->nexts[prev_node]))
1679 naccepted = 1;
1680
1681 if (naccepted == 0)
1682 continue;
1683
1684 if (sctx->limits.nelem)
1685 {
1686 Idx to_idx = str_idx + naccepted;
1687 if (check_dst_limits (mctx, &sctx->limits,
1688 dfa->nexts[prev_node], to_idx,
1689 prev_node, str_idx))
1690 continue;
1691 }
1692 ok = re_node_set_insert (cur_dest, prev_node);
1693 if (BE (! ok, 0))
1694 return REG_ESPACE;
1695 }
1696
1697 return REG_NOERROR;
1698}
1699
1700/* Helper functions. */
1701
1702static reg_errcode_t
1703internal_function
1704clean_state_log_if_needed (re_match_context_t *mctx, Idx next_state_log_idx)
1705{
1706 Idx top = mctx->state_log_top;
1707
1708 if (next_state_log_idx >= mctx->input.bufs_len
1709 || (next_state_log_idx >= mctx->input.valid_len
1710 && mctx->input.valid_len < mctx->input.len))
1711 {
1712 reg_errcode_t err;
1713 err = extend_buffers (mctx);
1714 if (BE (err != REG_NOERROR, 0))
1715 return err;
1716 }
1717
1718 if (top < next_state_log_idx)
1719 {
1720 memset (mctx->state_log + top + 1, '\0',
1721 sizeof (re_dfastate_t *) * (next_state_log_idx - top));
1722 mctx->state_log_top = next_state_log_idx;
1723 }
1724 return REG_NOERROR;
1725}
1726
1727static reg_errcode_t
1728internal_function
1729merge_state_array (re_dfa_t *dfa, re_dfastate_t **dst, re_dfastate_t **src,
1730 Idx num)
1731{
1732 Idx st_idx;
1733 reg_errcode_t err;
1734 for (st_idx = 0; st_idx < num; ++st_idx)
1735 {
1736 if (dst[st_idx] == NULL)
1737 dst[st_idx] = src[st_idx];
1738 else if (src[st_idx] != NULL)
1739 {
1740 re_node_set merged_set;
1741 err = re_node_set_init_union (&merged_set, &dst[st_idx]->nodes,
1742 &src[st_idx]->nodes);
1743 if (BE (err != REG_NOERROR, 0))
1744 return err;
1745 dst[st_idx] = re_acquire_state (&err, dfa, &merged_set);
1746 re_node_set_free (&merged_set);
1747 if (BE (err != REG_NOERROR, 0))
1748 return err;
1749 }
1750 }
1751 return REG_NOERROR;
1752}
1753
1754static reg_errcode_t
1755internal_function
1756update_cur_sifted_state (re_match_context_t *mctx, re_sift_context_t *sctx,
1757 Idx str_idx, re_node_set *dest_nodes)
1758{
1759 re_dfa_t *const dfa = mctx->dfa;
1760 reg_errcode_t err;
1761 const re_node_set *candidates;
1762 candidates = ((mctx->state_log[str_idx] == NULL) ? NULL
1763 : &mctx->state_log[str_idx]->nodes);
1764
1765 if (dest_nodes->nelem == 0)
1766 sctx->sifted_states[str_idx] = NULL;
1767 else
1768 {
1769 if (candidates)
1770 {
1771 /* At first, add the nodes which can epsilon transit to a node in
1772 DEST_NODE. */
1773 err = add_epsilon_src_nodes (dfa, dest_nodes, candidates);
1774 if (BE (err != REG_NOERROR, 0))
1775 return err;
1776
1777 /* Then, check the limitations in the current sift_context. */
1778 if (sctx->limits.nelem)
1779 {
1780 err = check_subexp_limits (dfa, dest_nodes, candidates, &sctx->limits,
1781 mctx->bkref_ents, str_idx);
1782 if (BE (err != REG_NOERROR, 0))
1783 return err;
1784 }
1785 }
1786
1787 sctx->sifted_states[str_idx] = re_acquire_state (&err, dfa, dest_nodes);
1788 if (BE (err != REG_NOERROR, 0))
1789 return err;
1790 }
1791
1792 if (candidates && mctx->state_log[str_idx]->has_backref)
1793 {
1794 err = sift_states_bkref (mctx, sctx, str_idx, candidates);
1795 if (BE (err != REG_NOERROR, 0))
1796 return err;
1797 }
1798 return REG_NOERROR;
1799}
1800
1801static reg_errcode_t
1802internal_function
1803add_epsilon_src_nodes (re_dfa_t *dfa, re_node_set *dest_nodes,
1804 const re_node_set *candidates)
1805{
1806 reg_errcode_t err = REG_NOERROR;
1807 Idx i;
1808
1809 re_dfastate_t *state = re_acquire_state (&err, dfa, dest_nodes);
1810 if (BE (err != REG_NOERROR, 0))
1811 return err;
1812
1813 if (!state->inveclosure.alloc)
1814 {
1815 err = re_node_set_alloc (&state->inveclosure, dest_nodes->nelem);
1816 if (BE (err != REG_NOERROR, 0))
1817 return REG_ESPACE;
1818 for (i = 0; i < dest_nodes->nelem; i++)
1819 re_node_set_merge (&state->inveclosure,
1820 dfa->inveclosures + dest_nodes->elems[i]);
1821 }
1822 return re_node_set_add_intersect (dest_nodes, candidates,
1823 &state->inveclosure);
1824}
1825
1826static reg_errcode_t
1827internal_function
1828sub_epsilon_src_nodes (re_dfa_t *dfa, Idx node, re_node_set *dest_nodes,
1829 const re_node_set *candidates)
1830{
1831 Idx ecl_idx;
1832 reg_errcode_t err;
1833 re_node_set *inv_eclosure = dfa->inveclosures + node;
1834 re_node_set except_nodes;
1835 re_node_set_init_empty (&except_nodes);
1836 for (ecl_idx = 0; ecl_idx < inv_eclosure->nelem; ++ecl_idx)
1837 {
1838 Idx cur_node = inv_eclosure->elems[ecl_idx];
1839 if (cur_node == node)
1840 continue;
1841 if (IS_EPSILON_NODE (dfa->nodes[cur_node].type))
1842 {
1843 Idx edst1 = dfa->edests[cur_node].elems[0];
1844 Idx edst2 = ((dfa->edests[cur_node].nelem > 1)
1845 ? dfa->edests[cur_node].elems[1] : REG_MISSING);
1846 if ((!re_node_set_contains (inv_eclosure, edst1)
1847 && re_node_set_contains (dest_nodes, edst1))
1848 || (REG_VALID_NONZERO_INDEX (edst2)
1849 && !re_node_set_contains (inv_eclosure, edst2)
1850 && re_node_set_contains (dest_nodes, edst2)))
1851 {
1852 err = re_node_set_add_intersect (&except_nodes, candidates,
1853 dfa->inveclosures + cur_node);
1854 if (BE (err != REG_NOERROR, 0))
1855 {
1856 re_node_set_free (&except_nodes);
1857 return err;
1858 }
1859 }
1860 }
1861 }
1862 for (ecl_idx = 0; ecl_idx < inv_eclosure->nelem; ++ecl_idx)
1863 {
1864 Idx cur_node = inv_eclosure->elems[ecl_idx];
1865 if (!re_node_set_contains (&except_nodes, cur_node))
1866 {
1867 Idx idx = re_node_set_contains (dest_nodes, cur_node) - 1;
1868 re_node_set_remove_at (dest_nodes, idx);
1869 }
1870 }
1871 re_node_set_free (&except_nodes);
1872 return REG_NOERROR;
1873}
1874
1875static bool
1876internal_function
1877check_dst_limits (const re_match_context_t *mctx, const re_node_set *limits,
1878 Idx dst_node, Idx dst_idx, Idx src_node, Idx src_idx)
1879{
1880 re_dfa_t *const dfa = mctx->dfa;
1881 Idx lim_idx, src_pos, dst_pos;
1882
1883 Idx dst_bkref_idx = search_cur_bkref_entry (mctx, dst_idx);
1884 Idx src_bkref_idx = search_cur_bkref_entry (mctx, src_idx);
1885 for (lim_idx = 0; lim_idx < limits->nelem; ++lim_idx)
1886 {
1887 Idx subexp_idx;
1888 struct re_backref_cache_entry *ent;
1889 ent = mctx->bkref_ents + limits->elems[lim_idx];
1890 subexp_idx = dfa->nodes[ent->node].opr.idx;
1891
1892 dst_pos = check_dst_limits_calc_pos (mctx, limits->elems[lim_idx],
1893 subexp_idx, dst_node, dst_idx,
1894 dst_bkref_idx);
1895 src_pos = check_dst_limits_calc_pos (mctx, limits->elems[lim_idx],
1896 subexp_idx, src_node, src_idx,
1897 src_bkref_idx);
1898
1899 /* In case of:
1900 <src> <dst> ( <subexp> )
1901 ( <subexp> ) <src> <dst>
1902 ( <subexp1> <src> <subexp2> <dst> <subexp3> ) */
1903 if (src_pos == dst_pos)
1904 continue; /* This is unrelated limitation. */
1905 else
1906 return true;
1907 }
1908 return false;
1909}
1910
1911static int
1912internal_function
1913check_dst_limits_calc_pos_1 (const re_match_context_t *mctx, int boundaries,
1914 Idx subexp_idx, Idx from_node, Idx bkref_idx)
1915{
1916 re_dfa_t *const dfa = mctx->dfa;
1917 re_node_set *eclosures = dfa->eclosures + from_node;
1918 Idx node_idx;
1919
1920 /* Else, we are on the boundary: examine the nodes on the epsilon
1921 closure. */
1922 for (node_idx = 0; node_idx < eclosures->nelem; ++node_idx)
1923 {
1924 Idx node = eclosures->elems[node_idx];
1925 switch (dfa->nodes[node].type)
1926 {
1927 case OP_BACK_REF:
1928 if (bkref_idx != REG_MISSING)
1929 {
1930 struct re_backref_cache_entry *ent = mctx->bkref_ents + bkref_idx;
1931 do
1932 {
1933 Idx dst;
1934 int cpos;
1935
1936 if (ent->node != node)
1937 continue;
1938
1939 if (subexp_idx < BITSET_WORD_BITS
1940 && !(ent->eps_reachable_subexps_map
1941 & ((bitset_word) 1 << subexp_idx)))
1942 continue;
1943
1944 /* Recurse trying to reach the OP_OPEN_SUBEXP and
1945 OP_CLOSE_SUBEXP cases below. But, if the
1946 destination node is the same node as the source
1947 node, don't recurse because it would cause an
1948 infinite loop: a regex that exhibits this behavior
1949 is ()\1*\1* */
1950 dst = dfa->edests[node].elems[0];
1951 if (dst == from_node)
1952 {
1953 if (boundaries & 1)
1954 return -1;
1955 else /* if (boundaries & 2) */
1956 return 0;
1957 }
1958
1959 cpos =
1960 check_dst_limits_calc_pos_1 (mctx, boundaries, subexp_idx,
1961 dst, bkref_idx);
1962 if (cpos == -1 /* && (boundaries & 1) */)
1963 return -1;
1964 if (cpos == 0 && (boundaries & 2))
1965 return 0;
1966
1967 if (subexp_idx < BITSET_WORD_BITS)
1968 ent->eps_reachable_subexps_map &=
1969 ~ ((bitset_word) 1 << subexp_idx);
1970 }
1971 while (ent++->more);
1972 }
1973 break;
1974
1975 case OP_OPEN_SUBEXP:
1976 if ((boundaries & 1) && subexp_idx == dfa->nodes[node].opr.idx)
1977 return -1;
1978 break;
1979
1980 case OP_CLOSE_SUBEXP:
1981 if ((boundaries & 2) && subexp_idx == dfa->nodes[node].opr.idx)
1982 return 0;
1983 break;
1984
1985 default:
1986 break;
1987 }
1988 }
1989
1990 return (boundaries & 2) ? 1 : 0;
1991}
1992
1993static int
1994internal_function
1995check_dst_limits_calc_pos (const re_match_context_t *mctx,
1996 Idx limit, Idx subexp_idx,
1997 Idx from_node, Idx str_idx, Idx bkref_idx)
1998{
1999 struct re_backref_cache_entry *lim = mctx->bkref_ents + limit;
2000 int boundaries;
2001
2002 /* If we are outside the range of the subexpression, return -1 or 1. */
2003 if (str_idx < lim->subexp_from)
2004 return -1;
2005
2006 if (lim->subexp_to < str_idx)
2007 return 1;
2008
2009 /* If we are within the subexpression, return 0. */
2010 boundaries = (str_idx == lim->subexp_from);
2011 boundaries |= (str_idx == lim->subexp_to) << 1;
2012 if (boundaries == 0)
2013 return 0;
2014
2015 /* Else, examine epsilon closure. */
2016 return check_dst_limits_calc_pos_1 (mctx, boundaries, subexp_idx,
2017 from_node, bkref_idx);
2018}
2019
2020/* Check the limitations of sub expressions LIMITS, and remove the nodes
2021 which are against limitations from DEST_NODES. */
2022
2023static reg_errcode_t
2024internal_function
2025check_subexp_limits (re_dfa_t *dfa, re_node_set *dest_nodes,
2026 const re_node_set *candidates, re_node_set *limits,
2027 struct re_backref_cache_entry *bkref_ents, Idx str_idx)
2028{
2029 reg_errcode_t err;
2030 Idx node_idx, lim_idx;
2031
2032 for (lim_idx = 0; lim_idx < limits->nelem; ++lim_idx)
2033 {
2034 Idx subexp_idx;
2035 struct re_backref_cache_entry *ent;
2036 ent = bkref_ents + limits->elems[lim_idx];
2037
2038 if (str_idx <= ent->subexp_from || ent->str_idx < str_idx)
2039 continue; /* This is unrelated limitation. */
2040
2041 subexp_idx = dfa->nodes[ent->node].opr.idx;
2042 if (ent->subexp_to == str_idx)
2043 {
2044 Idx ops_node = REG_MISSING;
2045 Idx cls_node = REG_MISSING;
2046 for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx)
2047 {
2048 Idx node = dest_nodes->elems[node_idx];
2049 re_token_type_t type = dfa->nodes[node].type;
2050 if (type == OP_OPEN_SUBEXP
2051 && subexp_idx == dfa->nodes[node].opr.idx)
2052 ops_node = node;
2053 else if (type == OP_CLOSE_SUBEXP
2054 && subexp_idx == dfa->nodes[node].opr.idx)
2055 cls_node = node;
2056 }
2057
2058 /* Check the limitation of the open subexpression. */
2059 /* Note that (ent->subexp_to = str_idx != ent->subexp_from). */
2060 if (REG_VALID_INDEX (ops_node))
2061 {
2062 err = sub_epsilon_src_nodes (dfa, ops_node, dest_nodes,
2063 candidates);
2064 if (BE (err != REG_NOERROR, 0))
2065 return err;
2066 }
2067
2068 /* Check the limitation of the close subexpression. */
2069 if (REG_VALID_INDEX (cls_node))
2070 for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx)
2071 {
2072 Idx node = dest_nodes->elems[node_idx];
2073 if (!re_node_set_contains (dfa->inveclosures + node,
2074 cls_node)
2075 && !re_node_set_contains (dfa->eclosures + node,
2076 cls_node))
2077 {
2078 /* It is against this limitation.
2079 Remove it form the current sifted state. */
2080 err = sub_epsilon_src_nodes (dfa, node, dest_nodes,
2081 candidates);
2082 if (BE (err != REG_NOERROR, 0))
2083 return err;
2084 --node_idx;
2085 }
2086 }
2087 }
2088 else /* (ent->subexp_to != str_idx) */
2089 {
2090 for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx)
2091 {
2092 Idx node = dest_nodes->elems[node_idx];
2093 re_token_type_t type = dfa->nodes[node].type;
2094 if (type == OP_CLOSE_SUBEXP || type == OP_OPEN_SUBEXP)
2095 {
2096 if (subexp_idx != dfa->nodes[node].opr.idx)
2097 continue;
2098 /* It is against this limitation.
2099 Remove it form the current sifted state. */
2100 err = sub_epsilon_src_nodes (dfa, node, dest_nodes,
2101 candidates);
2102 if (BE (err != REG_NOERROR, 0))
2103 return err;
2104 }
2105 }
2106 }
2107 }
2108 return REG_NOERROR;
2109}
2110
2111static reg_errcode_t
2112internal_function
2113sift_states_bkref (re_match_context_t *mctx, re_sift_context_t *sctx,
2114 Idx str_idx, const re_node_set *candidates)
2115{
2116 re_dfa_t *const dfa = mctx->dfa;
2117 reg_errcode_t err;
2118 Idx node_idx, node;
2119 re_sift_context_t local_sctx;
2120 Idx first_idx = search_cur_bkref_entry (mctx, str_idx);
2121
2122 if (first_idx == REG_MISSING)
2123 return REG_NOERROR;
2124
2125 local_sctx.sifted_states = NULL; /* Mark that it hasn't been initialized. */
2126
2127 for (node_idx = 0; node_idx < candidates->nelem; ++node_idx)
2128 {
2129 Idx enabled_idx;
2130 re_token_type_t type;
2131 struct re_backref_cache_entry *entry;
2132 node = candidates->elems[node_idx];
2133 type = dfa->nodes[node].type;
2134 /* Avoid infinite loop for the REs like "()\1+". */
2135 if (node == sctx->last_node && str_idx == sctx->last_str_idx)
2136 continue;
2137 if (type != OP_BACK_REF)
2138 continue;
2139
2140 entry = mctx->bkref_ents + first_idx;
2141 enabled_idx = first_idx;
2142 do
2143 {
2144 bool ok;
2145 Idx subexp_len, to_idx, dst_node;
2146 re_dfastate_t *cur_state;
2147
2148 if (entry->node != node)
2149 continue;
2150 subexp_len = entry->subexp_to - entry->subexp_from;
2151 to_idx = str_idx + subexp_len;
2152 dst_node = (subexp_len ? dfa->nexts[node]
2153 : dfa->edests[node].elems[0]);
2154
2155 if (to_idx > sctx->last_str_idx
2156 || sctx->sifted_states[to_idx] == NULL
2157 || !STATE_NODE_CONTAINS (sctx->sifted_states[to_idx], dst_node)
2158 || check_dst_limits (mctx, &sctx->limits, node,
2159 str_idx, dst_node, to_idx))
2160 continue;
2161
2162 if (local_sctx.sifted_states == NULL)
2163 {
2164 local_sctx = *sctx;
2165 err = re_node_set_init_copy (&local_sctx.limits, &sctx->limits);
2166 if (BE (err != REG_NOERROR, 0))
2167 goto free_return;
2168 }
2169 local_sctx.last_node = node;
2170 local_sctx.last_str_idx = str_idx;
2171 ok = re_node_set_insert (&local_sctx.limits, enabled_idx);
2172 if (BE (! ok, 0))
2173 {
2174 err = REG_ESPACE;
2175 goto free_return;
2176 }
2177 cur_state = local_sctx.sifted_states[str_idx];
2178 err = sift_states_backward (mctx, &local_sctx);
2179 if (BE (err != REG_NOERROR, 0))
2180 goto free_return;
2181 if (sctx->limited_states != NULL)
2182 {
2183 err = merge_state_array (dfa, sctx->limited_states,
2184 local_sctx.sifted_states,
2185 str_idx + 1);
2186 if (BE (err != REG_NOERROR, 0))
2187 goto free_return;
2188 }
2189 local_sctx.sifted_states[str_idx] = cur_state;
2190 re_node_set_remove (&local_sctx.limits, enabled_idx);
2191
2192 /* mctx->bkref_ents may have changed, reload the pointer. */
2193 entry = mctx->bkref_ents + enabled_idx;
2194 }
2195 while (enabled_idx++, entry++->more);
2196 }
2197 err = REG_NOERROR;
2198 free_return:
2199 if (local_sctx.sifted_states != NULL)
2200 {
2201 re_node_set_free (&local_sctx.limits);
2202 }
2203
2204 return err;
2205}
2206
2207
2208#ifdef RE_ENABLE_I18N
2209static int
2210internal_function
2211sift_states_iter_mb (const re_match_context_t *mctx, re_sift_context_t *sctx,
2212 Idx node_idx, Idx str_idx, Idx max_str_idx)
2213{
2214 re_dfa_t *const dfa = mctx->dfa;
2215 int naccepted;
2216 /* Check the node can accept `multi byte'. */
2217 naccepted = check_node_accept_bytes (dfa, node_idx, &mctx->input, str_idx);
2218 if (naccepted > 0 && str_idx + naccepted <= max_str_idx &&
2219 !STATE_NODE_CONTAINS (sctx->sifted_states[str_idx + naccepted],
2220 dfa->nexts[node_idx]))
2221 /* The node can't accept the `multi byte', or the
2222 destination was already thrown away, then the node
2223 could't accept the current input `multi byte'. */
2224 naccepted = 0;
2225 /* Otherwise, it is sure that the node could accept
2226 `naccepted' bytes input. */
2227 return naccepted;
2228}
2229#endif /* RE_ENABLE_I18N */
2230
2231
2232/* Functions for state transition. */
2233
2234/* Return the next state to which the current state STATE will transit by
2235 accepting the current input byte, and update STATE_LOG if necessary.
2236 If STATE can accept a multibyte char/collating element/back reference
2237 update the destination of STATE_LOG. */
2238
2239static re_dfastate_t *
2240internal_function
2241transit_state (reg_errcode_t *err, re_match_context_t *mctx,
2242 re_dfastate_t *state)
2243{
2244 re_dfastate_t **trtable;
2245 unsigned char ch;
2246
2247#ifdef RE_ENABLE_I18N
2248 /* If the current state can accept multibyte. */
2249 if (BE (state->accept_mb, 0))
2250 {
2251 *err = transit_state_mb (mctx, state);
2252 if (BE (*err != REG_NOERROR, 0))
2253 return NULL;
2254 }
2255#endif /* RE_ENABLE_I18N */
2256
2257 /* Then decide the next state with the single byte. */
2258#if 0
2259 if (0)
2260 /* don't use transition table */
2261 return transit_state_sb (err, mctx, state);
2262#endif
2263
2264 /* Use transition table */
2265 ch = re_string_fetch_byte (&mctx->input);
2266 for (;;)
2267 {
2268 trtable = state->trtable;
2269 if (BE (trtable != NULL, 1))
2270 return trtable[ch];
2271
2272 trtable = state->word_trtable;
2273 if (BE (trtable != NULL, 1))
2274 {
2275 unsigned int context;
2276 context
2277 = re_string_context_at (&mctx->input,
2278 re_string_cur_idx (&mctx->input) - 1,
2279 mctx->eflags);
2280 if (IS_WORD_CONTEXT (context))
2281 return trtable[ch + SBC_MAX];
2282 else
2283 return trtable[ch];
2284 }
2285
2286 if (!build_trtable (mctx->dfa, state))
2287 {
2288 *err = REG_ESPACE;
2289 return NULL;
2290 }
2291
2292 /* Retry, we now have a transition table. */
2293 }
2294}
2295
2296/* Update the state_log if we need */
2297re_dfastate_t *
2298internal_function
2299merge_state_with_log (reg_errcode_t *err, re_match_context_t *mctx,
2300 re_dfastate_t *next_state)
2301{
2302 re_dfa_t *const dfa = mctx->dfa;
2303 Idx cur_idx = re_string_cur_idx (&mctx->input);
2304
2305 if (cur_idx > mctx->state_log_top)
2306 {
2307 mctx->state_log[cur_idx] = next_state;
2308 mctx->state_log_top = cur_idx;
2309 }
2310 else if (mctx->state_log[cur_idx] == 0)
2311 {
2312 mctx->state_log[cur_idx] = next_state;
2313 }
2314 else
2315 {
2316 re_dfastate_t *pstate;
2317 unsigned int context;
2318 re_node_set next_nodes, *log_nodes, *table_nodes = NULL;
2319 /* If (state_log[cur_idx] != 0), it implies that cur_idx is
2320 the destination of a multibyte char/collating element/
2321 back reference. Then the next state is the union set of
2322 these destinations and the results of the transition table. */
2323 pstate = mctx->state_log[cur_idx];
2324 log_nodes = pstate->entrance_nodes;
2325 if (next_state != NULL)
2326 {
2327 table_nodes = next_state->entrance_nodes;
2328 *err = re_node_set_init_union (&next_nodes, table_nodes,
2329 log_nodes);
2330 if (BE (*err != REG_NOERROR, 0))
2331 return NULL;
2332 }
2333 else
2334 next_nodes = *log_nodes;
2335 /* Note: We already add the nodes of the initial state,
2336 then we don't need to add them here. */
2337
2338 context = re_string_context_at (&mctx->input,
2339 re_string_cur_idx (&mctx->input) - 1,
2340 mctx->eflags);
2341 next_state = mctx->state_log[cur_idx]
2342 = re_acquire_state_context (err, dfa, &next_nodes, context);
2343 /* We don't need to check errors here, since the return value of
2344 this function is next_state and ERR is already set. */
2345
2346 if (table_nodes != NULL)
2347 re_node_set_free (&next_nodes);
2348 }
2349
2350 if (BE (dfa->nbackref, 0) && next_state != NULL)
2351 {
2352 /* Check OP_OPEN_SUBEXP in the current state in case that we use them
2353 later. We must check them here, since the back references in the
2354 next state might use them. */
2355 *err = check_subexp_matching_top (mctx, &next_state->nodes,
2356 cur_idx);
2357 if (BE (*err != REG_NOERROR, 0))
2358 return NULL;
2359
2360 /* If the next state has back references. */
2361 if (next_state->has_backref)
2362 {
2363 *err = transit_state_bkref (mctx, &next_state->nodes);
2364 if (BE (*err != REG_NOERROR, 0))
2365 return NULL;
2366 next_state = mctx->state_log[cur_idx];
2367 }
2368 }
2369
2370 return next_state;
2371}
2372
2373/* Skip bytes in the input that correspond to part of a
2374 multi-byte match, then look in the log for a state
2375 from which to restart matching. */
2376static re_dfastate_t *
2377internal_function
2378find_recover_state (reg_errcode_t *err, re_match_context_t *mctx)
2379{
2380 re_dfastate_t *cur_state = NULL;
2381 do
2382 {
2383 Idx max = mctx->state_log_top;
2384 Idx cur_str_idx = re_string_cur_idx (&mctx->input);
2385
2386 do
2387 {
2388 if (++cur_str_idx > max)
2389 return NULL;
2390 re_string_skip_bytes (&mctx->input, 1);
2391 }
2392 while (mctx->state_log[cur_str_idx] == NULL);
2393
2394 cur_state = merge_state_with_log (err, mctx, NULL);
2395 }
2396 while (*err == REG_NOERROR && cur_state == NULL);
2397 return cur_state;
2398}
2399
2400/* Helper functions for transit_state. */
2401
2402/* From the node set CUR_NODES, pick up the nodes whose types are
2403 OP_OPEN_SUBEXP and which have corresponding back references in the regular
2404 expression. And register them to use them later for evaluating the
2405 correspoding back references. */
2406
2407static reg_errcode_t
2408internal_function
2409check_subexp_matching_top (re_match_context_t *mctx, re_node_set *cur_nodes,
2410 Idx str_idx)
2411{
2412 re_dfa_t *const dfa = mctx->dfa;
2413 Idx node_idx;
2414 reg_errcode_t err;
2415
2416 /* TODO: This isn't efficient.
2417 Because there might be more than one nodes whose types are
2418 OP_OPEN_SUBEXP and whose index is SUBEXP_IDX, we must check all
2419 nodes.
2420 E.g. RE: (a){2} */
2421 for (node_idx = 0; node_idx < cur_nodes->nelem; ++node_idx)
2422 {
2423 Idx node = cur_nodes->elems[node_idx];
2424 if (dfa->nodes[node].type == OP_OPEN_SUBEXP
2425 && dfa->nodes[node].opr.idx < BITSET_WORD_BITS
2426 && (dfa->used_bkref_map
2427 & ((bitset_word) 1 << dfa->nodes[node].opr.idx)))
2428 {
2429 err = match_ctx_add_subtop (mctx, node, str_idx);
2430 if (BE (err != REG_NOERROR, 0))
2431 return err;
2432 }
2433 }
2434 return REG_NOERROR;
2435}
2436
2437#if 0
2438/* Return the next state to which the current state STATE will transit by
2439 accepting the current input byte. */
2440
2441static re_dfastate_t *
2442transit_state_sb (reg_errcode_t *err, re_match_context_t *mctx,
2443 re_dfastate_t *state)
2444{
2445 re_dfa_t *const dfa = mctx->dfa;
2446 re_node_set next_nodes;
2447 re_dfastate_t *next_state;
2448 Idx node_cnt, cur_str_idx = re_string_cur_idx (&mctx->input);
2449 unsigned int context;
2450
2451 *err = re_node_set_alloc (&next_nodes, state->nodes.nelem + 1);
2452 if (BE (*err != REG_NOERROR, 0))
2453 return NULL;
2454 for (node_cnt = 0; node_cnt < state->nodes.nelem; ++node_cnt)
2455 {
2456 Idx cur_node = state->nodes.elems[node_cnt];
2457 if (check_node_accept (mctx, dfa->nodes + cur_node, cur_str_idx))
2458 {
2459 *err = re_node_set_merge (&next_nodes,
2460 dfa->eclosures + dfa->nexts[cur_node]);
2461 if (BE (*err != REG_NOERROR, 0))
2462 {
2463 re_node_set_free (&next_nodes);
2464 return NULL;
2465 }
2466 }
2467 }
2468 context = re_string_context_at (&mctx->input, cur_str_idx, mctx->eflags);
2469 next_state = re_acquire_state_context (err, dfa, &next_nodes, context);
2470 /* We don't need to check errors here, since the return value of
2471 this function is next_state and ERR is already set. */
2472
2473 re_node_set_free (&next_nodes);
2474 re_string_skip_bytes (&mctx->input, 1);
2475 return next_state;
2476}
2477#endif
2478
2479#ifdef RE_ENABLE_I18N
2480static reg_errcode_t
2481internal_function
2482transit_state_mb (re_match_context_t *mctx, re_dfastate_t *pstate)
2483{
2484 re_dfa_t *const dfa = mctx->dfa;
2485 reg_errcode_t err;
2486 Idx i;
2487
2488 for (i = 0; i < pstate->nodes.nelem; ++i)
2489 {
2490 re_node_set dest_nodes, *new_nodes;
2491 Idx cur_node_idx = pstate->nodes.elems[i];
2492 int naccepted;
2493 Idx dest_idx;
2494 unsigned int context;
2495 re_dfastate_t *dest_state;
2496
2497 if (!dfa->nodes[cur_node_idx].accept_mb)
2498 continue;
2499
2500 if (dfa->nodes[cur_node_idx].constraint)
2501 {
2502 context = re_string_context_at (&mctx->input,
2503 re_string_cur_idx (&mctx->input),
2504 mctx->eflags);
2505 if (NOT_SATISFY_NEXT_CONSTRAINT (dfa->nodes[cur_node_idx].constraint,
2506 context))
2507 continue;
2508 }
2509
2510 /* How many bytes the node can accept? */
2511 naccepted = check_node_accept_bytes (dfa, cur_node_idx, &mctx->input,
2512 re_string_cur_idx (&mctx->input));
2513 if (naccepted == 0)
2514 continue;
2515
2516 /* The node can accepts `naccepted' bytes. */
2517 dest_idx = re_string_cur_idx (&mctx->input) + naccepted;
2518 mctx->max_mb_elem_len = ((mctx->max_mb_elem_len < naccepted) ? naccepted
2519 : mctx->max_mb_elem_len);
2520 err = clean_state_log_if_needed (mctx, dest_idx);
2521 if (BE (err != REG_NOERROR, 0))
2522 return err;
2523#ifdef DEBUG
2524 assert (dfa->nexts[cur_node_idx] != REG_MISSING);
2525#endif
2526 new_nodes = dfa->eclosures + dfa->nexts[cur_node_idx];
2527
2528 dest_state = mctx->state_log[dest_idx];
2529 if (dest_state == NULL)
2530 dest_nodes = *new_nodes;
2531 else
2532 {
2533 err = re_node_set_init_union (&dest_nodes,
2534 dest_state->entrance_nodes, new_nodes);
2535 if (BE (err != REG_NOERROR, 0))
2536 return err;
2537 }
2538 context = re_string_context_at (&mctx->input, dest_idx - 1, mctx->eflags);
2539 mctx->state_log[dest_idx]
2540 = re_acquire_state_context (&err, dfa, &dest_nodes, context);
2541 if (dest_state != NULL)
2542 re_node_set_free (&dest_nodes);
2543 if (BE (mctx->state_log[dest_idx] == NULL && err != REG_NOERROR, 0))
2544 return err;
2545 }
2546 return REG_NOERROR;
2547}
2548#endif /* RE_ENABLE_I18N */
2549
2550static reg_errcode_t
2551internal_function
2552transit_state_bkref (re_match_context_t *mctx, const re_node_set *nodes)
2553{
2554 re_dfa_t *const dfa = mctx->dfa;
2555 reg_errcode_t err;
2556 Idx i;
2557 Idx cur_str_idx = re_string_cur_idx (&mctx->input);
2558
2559 for (i = 0; i < nodes->nelem; ++i)
2560 {
2561 Idx dest_str_idx, prev_nelem, bkc_idx;
2562 Idx node_idx = nodes->elems[i];
2563 unsigned int context;
2564 const re_token_t *node = dfa->nodes + node_idx;
2565 re_node_set *new_dest_nodes;
2566
2567 /* Check whether `node' is a backreference or not. */
2568 if (node->type != OP_BACK_REF)
2569 continue;
2570
2571 if (node->constraint)
2572 {
2573 context = re_string_context_at (&mctx->input, cur_str_idx,
2574 mctx->eflags);
2575 if (NOT_SATISFY_NEXT_CONSTRAINT (node->constraint, context))
2576 continue;
2577 }
2578
2579 /* `node' is a backreference.
2580 Check the substring which the substring matched. */
2581 bkc_idx = mctx->nbkref_ents;
2582 err = get_subexp (mctx, node_idx, cur_str_idx);
2583 if (BE (err != REG_NOERROR, 0))
2584 goto free_return;
2585
2586 /* And add the epsilon closures (which is `new_dest_nodes') of
2587 the backreference to appropriate state_log. */
2588#ifdef DEBUG
2589 assert (dfa->nexts[node_idx] != REG_MISSING);
2590#endif
2591 for (; bkc_idx < mctx->nbkref_ents; ++bkc_idx)
2592 {
2593 Idx subexp_len;
2594 re_dfastate_t *dest_state;
2595 struct re_backref_cache_entry *bkref_ent;
2596 bkref_ent = mctx->bkref_ents + bkc_idx;
2597 if (bkref_ent->node != node_idx || bkref_ent->str_idx != cur_str_idx)
2598 continue;
2599 subexp_len = bkref_ent->subexp_to - bkref_ent->subexp_from;
2600 new_dest_nodes = (subexp_len == 0
2601 ? dfa->eclosures + dfa->edests[node_idx].elems[0]
2602 : dfa->eclosures + dfa->nexts[node_idx]);
2603 dest_str_idx = (cur_str_idx + bkref_ent->subexp_to
2604 - bkref_ent->subexp_from);
2605 context = re_string_context_at (&mctx->input, dest_str_idx - 1,
2606 mctx->eflags);
2607 dest_state = mctx->state_log[dest_str_idx];
2608 prev_nelem = ((mctx->state_log[cur_str_idx] == NULL) ? 0
2609 : mctx->state_log[cur_str_idx]->nodes.nelem);
2610 /* Add `new_dest_node' to state_log. */
2611 if (dest_state == NULL)
2612 {
2613 mctx->state_log[dest_str_idx]
2614 = re_acquire_state_context (&err, dfa, new_dest_nodes,
2615 context);
2616 if (BE (mctx->state_log[dest_str_idx] == NULL
2617 && err != REG_NOERROR, 0))
2618 goto free_return;
2619 }
2620 else
2621 {
2622 re_node_set dest_nodes;
2623 err = re_node_set_init_union (&dest_nodes,
2624 dest_state->entrance_nodes,
2625 new_dest_nodes);
2626 if (BE (err != REG_NOERROR, 0))
2627 {
2628 re_node_set_free (&dest_nodes);
2629 goto free_return;
2630 }
2631 mctx->state_log[dest_str_idx]
2632 = re_acquire_state_context (&err, dfa, &dest_nodes, context);
2633 re_node_set_free (&dest_nodes);
2634 if (BE (mctx->state_log[dest_str_idx] == NULL
2635 && err != REG_NOERROR, 0))
2636 goto free_return;
2637 }
2638 /* We need to check recursively if the backreference can epsilon
2639 transit. */
2640 if (subexp_len == 0
2641 && mctx->state_log[cur_str_idx]->nodes.nelem > prev_nelem)
2642 {
2643 err = check_subexp_matching_top (mctx, new_dest_nodes,
2644 cur_str_idx);
2645 if (BE (err != REG_NOERROR, 0))
2646 goto free_return;
2647 err = transit_state_bkref (mctx, new_dest_nodes);
2648 if (BE (err != REG_NOERROR, 0))
2649 goto free_return;
2650 }
2651 }
2652 }
2653 err = REG_NOERROR;
2654 free_return:
2655 return err;
2656}
2657
2658/* Enumerate all the candidates which the backreference BKREF_NODE can match
2659 at BKREF_STR_IDX, and register them by match_ctx_add_entry().
2660 Note that we might collect inappropriate candidates here.
2661 However, the cost of checking them strictly here is too high, then we
2662 delay these checking for prune_impossible_nodes(). */
2663
2664static reg_errcode_t
2665internal_function
2666get_subexp (re_match_context_t *mctx, Idx bkref_node, Idx bkref_str_idx)
2667{
2668 re_dfa_t *const dfa = mctx->dfa;
2669 Idx subexp_num, sub_top_idx;
2670 const char *buf = (const char *) re_string_get_buffer (&mctx->input);
2671 /* Return if we have already checked BKREF_NODE at BKREF_STR_IDX. */
2672 Idx cache_idx = search_cur_bkref_entry (mctx, bkref_str_idx);
2673 if (cache_idx != REG_MISSING)
2674 {
2675 const struct re_backref_cache_entry *entry = mctx->bkref_ents + cache_idx;
2676 do
2677 if (entry->node == bkref_node)
2678 return REG_NOERROR; /* We already checked it. */
2679 while (entry++->more);
2680 }
2681
2682 subexp_num = dfa->nodes[bkref_node].opr.idx;
2683
2684 /* For each sub expression */
2685 for (sub_top_idx = 0; sub_top_idx < mctx->nsub_tops; ++sub_top_idx)
2686 {
2687 reg_errcode_t err;
2688 re_sub_match_top_t *sub_top = mctx->sub_tops[sub_top_idx];
2689 re_sub_match_last_t *sub_last;
2690 Idx sub_last_idx, sl_str, bkref_str_off;
2691
2692 if (dfa->nodes[sub_top->node].opr.idx != subexp_num)
2693 continue; /* It isn't related. */
2694
2695 sl_str = sub_top->str_idx;
2696 bkref_str_off = bkref_str_idx;
2697 /* At first, check the last node of sub expressions we already
2698 evaluated. */
2699 for (sub_last_idx = 0; sub_last_idx < sub_top->nlasts; ++sub_last_idx)
2700 {
2701 regoff_t sl_str_diff;
2702 sub_last = sub_top->lasts[sub_last_idx];
2703 sl_str_diff = sub_last->str_idx - sl_str;
2704 /* The matched string by the sub expression match with the substring
2705 at the back reference? */
2706 if (sl_str_diff > 0)
2707 {
2708 if (BE (bkref_str_off + sl_str_diff > mctx->input.valid_len, 0))
2709 {
2710 /* Not enough chars for a successful match. */
2711 if (bkref_str_off + sl_str_diff > mctx->input.len)
2712 break;
2713
2714 err = clean_state_log_if_needed (mctx,
2715 bkref_str_off
2716 + sl_str_diff);
2717 if (BE (err != REG_NOERROR, 0))
2718 return err;
2719 buf = (const char *) re_string_get_buffer (&mctx->input);
2720 }
2721 if (memcmp (buf + bkref_str_off, buf + sl_str, sl_str_diff) != 0)
2722 break; /* We don't need to search this sub expression any more. */
2723 }
2724 bkref_str_off += sl_str_diff;
2725 sl_str += sl_str_diff;
2726 err = get_subexp_sub (mctx, sub_top, sub_last, bkref_node,
2727 bkref_str_idx);
2728
2729 /* Reload buf, since the preceding call might have reallocated
2730 the buffer. */
2731 buf = (const char *) re_string_get_buffer (&mctx->input);
2732
2733 if (err == REG_NOMATCH)
2734 continue;
2735 if (BE (err != REG_NOERROR, 0))
2736 return err;
2737 }
2738
2739 if (sub_last_idx < sub_top->nlasts)
2740 continue;
2741 if (sub_last_idx > 0)
2742 ++sl_str;
2743 /* Then, search for the other last nodes of the sub expression. */
2744 for (; sl_str <= bkref_str_idx; ++sl_str)
2745 {
2746 Idx cls_node;
2747 regoff_t sl_str_off;
2748 const re_node_set *nodes;
2749 sl_str_off = sl_str - sub_top->str_idx;
2750 /* The matched string by the sub expression match with the substring
2751 at the back reference? */
2752 if (sl_str_off > 0)
2753 {
2754 if (BE (bkref_str_off >= mctx->input.valid_len, 0))
2755 {
2756 /* If we are at the end of the input, we cannot match. */
2757 if (bkref_str_off >= mctx->input.len)
2758 break;
2759
2760 err = extend_buffers (mctx);
2761 if (BE (err != REG_NOERROR, 0))
2762 return err;
2763
2764 buf = (const char *) re_string_get_buffer (&mctx->input);
2765 }
2766 if (buf [bkref_str_off++] != buf[sl_str - 1])
2767 break; /* We don't need to search this sub expression
2768 any more. */
2769 }
2770 if (mctx->state_log[sl_str] == NULL)
2771 continue;
2772 /* Does this state have a ')' of the sub expression? */
2773 nodes = &mctx->state_log[sl_str]->nodes;
2774 cls_node = find_subexp_node (dfa, nodes, subexp_num, OP_CLOSE_SUBEXP);
2775 if (cls_node == REG_MISSING)
2776 continue; /* No. */
2777 if (sub_top->path == NULL)
2778 {
2779 sub_top->path = re_calloc (state_array_t,
2780 sl_str - sub_top->str_idx + 1);
2781 if (sub_top->path == NULL)
2782 return REG_ESPACE;
2783 }
2784 /* Can the OP_OPEN_SUBEXP node arrive the OP_CLOSE_SUBEXP node
2785 in the current context? */
2786 err = check_arrival (mctx, sub_top->path, sub_top->node,
2787 sub_top->str_idx, cls_node, sl_str, OP_CLOSE_SUBEXP);
2788 if (err == REG_NOMATCH)
2789 continue;
2790 if (BE (err != REG_NOERROR, 0))
2791 return err;
2792 sub_last = match_ctx_add_sublast (sub_top, cls_node, sl_str);
2793 if (BE (sub_last == NULL, 0))
2794 return REG_ESPACE;
2795 err = get_subexp_sub (mctx, sub_top, sub_last, bkref_node,
2796 bkref_str_idx);
2797 if (err == REG_NOMATCH)
2798 continue;
2799 }
2800 }
2801 return REG_NOERROR;
2802}
2803
2804/* Helper functions for get_subexp(). */
2805
2806/* Check SUB_LAST can arrive to the back reference BKREF_NODE at BKREF_STR.
2807 If it can arrive, register the sub expression expressed with SUB_TOP
2808 and SUB_LAST. */
2809
2810static reg_errcode_t
2811internal_function
2812get_subexp_sub (re_match_context_t *mctx, const re_sub_match_top_t *sub_top,
2813 re_sub_match_last_t *sub_last, Idx bkref_node, Idx bkref_str)
2814{
2815 reg_errcode_t err;
2816 Idx to_idx;
2817 /* Can the subexpression arrive the back reference? */
2818 err = check_arrival (mctx, &sub_last->path, sub_last->node,
2819 sub_last->str_idx, bkref_node, bkref_str, OP_OPEN_SUBEXP);
2820 if (err != REG_NOERROR)
2821 return err;
2822 err = match_ctx_add_entry (mctx, bkref_node, bkref_str, sub_top->str_idx,
2823 sub_last->str_idx);
2824 if (BE (err != REG_NOERROR, 0))
2825 return err;
2826 to_idx = bkref_str + sub_last->str_idx - sub_top->str_idx;
2827 return clean_state_log_if_needed (mctx, to_idx);
2828}
2829
2830/* Find the first node which is '(' or ')' and whose index is SUBEXP_IDX.
2831 Search '(' if FL_OPEN, or search ')' otherwise.
2832 TODO: This function isn't efficient...
2833 Because there might be more than one nodes whose types are
2834 OP_OPEN_SUBEXP and whose index is SUBEXP_IDX, we must check all
2835 nodes.
2836 E.g. RE: (a){2} */
2837
2838static Idx
2839internal_function
2840find_subexp_node (const re_dfa_t *dfa, const re_node_set *nodes,
2841 Idx subexp_idx, int type)
2842{
2843 Idx cls_idx;
2844 for (cls_idx = 0; cls_idx < nodes->nelem; ++cls_idx)
2845 {
2846 Idx cls_node = nodes->elems[cls_idx];
2847 const re_token_t *node = dfa->nodes + cls_node;
2848 if (node->type == type
2849 && node->opr.idx == subexp_idx)
2850 return cls_node;
2851 }
2852 return REG_MISSING;
2853}
2854
2855/* Check whether the node TOP_NODE at TOP_STR can arrive to the node
2856 LAST_NODE at LAST_STR. We record the path onto PATH since it will be
2857 heavily reused.
2858 Return REG_NOERROR if it can arrive, or REG_NOMATCH otherwise. */
2859
2860static reg_errcode_t
2861internal_function
2862check_arrival (re_match_context_t *mctx, state_array_t *path,
2863 Idx top_node, Idx top_str, Idx last_node, Idx last_str,
2864 int type)
2865{
2866 re_dfa_t *const dfa = mctx->dfa;
2867 reg_errcode_t err;
2868 Idx subexp_num, backup_cur_idx, str_idx, null_cnt;
2869 re_dfastate_t *cur_state = NULL;
2870 re_node_set *cur_nodes, next_nodes;
2871 re_dfastate_t **backup_state_log;
2872 unsigned int context;
2873
2874 subexp_num = dfa->nodes[top_node].opr.idx;
2875 /* Extend the buffer if we need. */
2876 if (BE (path->alloc < last_str + mctx->max_mb_elem_len + 1, 0))
2877 {
2878 re_dfastate_t **new_array;
2879 Idx old_alloc = path->alloc;
2880 Idx new_alloc = old_alloc + last_str + mctx->max_mb_elem_len + 1;
2881 if (BE (new_alloc < old_alloc, 0))
2882 return REG_ESPACE;
2883 new_array = re_xrealloc (path->array, re_dfastate_t *, new_alloc);
2884 if (BE (new_array == NULL, 0))
2885 return REG_ESPACE;
2886 path->array = new_array;
2887 path->alloc = new_alloc;
2888 memset (new_array + old_alloc, '\0',
2889 sizeof (re_dfastate_t *) * (new_alloc - old_alloc));
2890 }
2891
2892 str_idx = path->next_idx == 0 ? top_str : path->next_idx;
2893
2894 /* Temporary modify MCTX. */
2895 backup_state_log = mctx->state_log;
2896 backup_cur_idx = mctx->input.cur_idx;
2897 mctx->state_log = path->array;
2898 mctx->input.cur_idx = str_idx;
2899
2900 /* Setup initial node set. */
2901 context = re_string_context_at (&mctx->input, str_idx - 1, mctx->eflags);
2902 if (str_idx == top_str)
2903 {
2904 err = re_node_set_init_1 (&next_nodes, top_node);
2905 if (BE (err != REG_NOERROR, 0))
2906 return err;
2907 err = check_arrival_expand_ecl (dfa, &next_nodes, subexp_num, type);
2908 if (BE (err != REG_NOERROR, 0))
2909 {
2910 re_node_set_free (&next_nodes);
2911 return err;
2912 }
2913 }
2914 else
2915 {
2916 cur_state = mctx->state_log[str_idx];
2917 if (cur_state && cur_state->has_backref)
2918 {
2919 err = re_node_set_init_copy (&next_nodes, &cur_state->nodes);
2920 if (BE ( err != REG_NOERROR, 0))
2921 return err;
2922 }
2923 else
2924 re_node_set_init_empty (&next_nodes);
2925 }
2926 if (str_idx == top_str || (cur_state && cur_state->has_backref))
2927 {
2928 if (next_nodes.nelem)
2929 {
2930 err = expand_bkref_cache (mctx, &next_nodes, str_idx,
2931 subexp_num, type);
2932 if (BE ( err != REG_NOERROR, 0))
2933 {
2934 re_node_set_free (&next_nodes);
2935 return err;
2936 }
2937 }
2938 cur_state = re_acquire_state_context (&err, dfa, &next_nodes, context);
2939 if (BE (cur_state == NULL && err != REG_NOERROR, 0))
2940 {
2941 re_node_set_free (&next_nodes);
2942 return err;
2943 }
2944 mctx->state_log[str_idx] = cur_state;
2945 }
2946
2947 for (null_cnt = 0; str_idx < last_str && null_cnt <= mctx->max_mb_elem_len;)
2948 {
2949 re_node_set_empty (&next_nodes);
2950 if (mctx->state_log[str_idx + 1])
2951 {
2952 err = re_node_set_merge (&next_nodes,
2953 &mctx->state_log[str_idx + 1]->nodes);
2954 if (BE (err != REG_NOERROR, 0))
2955 {
2956 re_node_set_free (&next_nodes);
2957 return err;
2958 }
2959 }
2960 if (cur_state)
2961 {
2962 err = check_arrival_add_next_nodes (mctx, str_idx,
2963 &cur_state->non_eps_nodes, &next_nodes);
2964 if (BE (err != REG_NOERROR, 0))
2965 {
2966 re_node_set_free (&next_nodes);
2967 return err;
2968 }
2969 }
2970 ++str_idx;
2971 if (next_nodes.nelem)
2972 {
2973 err = check_arrival_expand_ecl (dfa, &next_nodes, subexp_num, type);
2974 if (BE (err != REG_NOERROR, 0))
2975 {
2976 re_node_set_free (&next_nodes);
2977 return err;
2978 }
2979 err = expand_bkref_cache (mctx, &next_nodes, str_idx,
2980 subexp_num, type);
2981 if (BE ( err != REG_NOERROR, 0))
2982 {
2983 re_node_set_free (&next_nodes);
2984 return err;
2985 }
2986 }
2987 context = re_string_context_at (&mctx->input, str_idx - 1, mctx->eflags);
2988 cur_state = re_acquire_state_context (&err, dfa, &next_nodes, context);
2989 if (BE (cur_state == NULL && err != REG_NOERROR, 0))
2990 {
2991 re_node_set_free (&next_nodes);
2992 return err;
2993 }
2994 mctx->state_log[str_idx] = cur_state;
2995 null_cnt = cur_state == NULL ? null_cnt + 1 : 0;
2996 }
2997 re_node_set_free (&next_nodes);
2998 cur_nodes = (mctx->state_log[last_str] == NULL ? NULL
2999 : &mctx->state_log[last_str]->nodes);
3000 path->next_idx = str_idx;
3001
3002 /* Fix MCTX. */
3003 mctx->state_log = backup_state_log;
3004 mctx->input.cur_idx = backup_cur_idx;
3005
3006 /* Then check the current node set has the node LAST_NODE. */
3007 if (cur_nodes != NULL && re_node_set_contains (cur_nodes, last_node))
3008 return REG_NOERROR;
3009
3010 return REG_NOMATCH;
3011}
3012
3013/* Helper functions for check_arrival. */
3014
3015/* Calculate the destination nodes of CUR_NODES at STR_IDX, and append them
3016 to NEXT_NODES.
3017 TODO: This function is similar to the functions transit_state*(),
3018 however this function has many additional works.
3019 Can't we unify them? */
3020
3021static reg_errcode_t
3022internal_function
3023check_arrival_add_next_nodes (re_match_context_t *mctx, Idx str_idx,
3024 re_node_set *cur_nodes,
3025 re_node_set *next_nodes)
3026{
3027 re_dfa_t *const dfa = mctx->dfa;
3028 bool ok;
3029 Idx cur_idx;
3030 reg_errcode_t err;
3031 re_node_set union_set;
3032 re_node_set_init_empty (&union_set);
3033 for (cur_idx = 0; cur_idx < cur_nodes->nelem; ++cur_idx)
3034 {
3035 int naccepted = 0;
3036 Idx cur_node = cur_nodes->elems[cur_idx];
3037#ifdef DEBUG
3038 re_token_type_t type = dfa->nodes[cur_node].type;
3039 assert (!IS_EPSILON_NODE (type));
3040#endif
3041#ifdef RE_ENABLE_I18N
3042 /* If the node may accept `multi byte'. */
3043 if (dfa->nodes[cur_node].accept_mb)
3044 {
3045 naccepted = check_node_accept_bytes (dfa, cur_node, &mctx->input,
3046 str_idx);
3047 if (naccepted > 1)
3048 {
3049 re_dfastate_t *dest_state;
3050 Idx next_node = dfa->nexts[cur_node];
3051 Idx next_idx = str_idx + naccepted;
3052 dest_state = mctx->state_log[next_idx];
3053 re_node_set_empty (&union_set);
3054 if (dest_state)
3055 {
3056 err = re_node_set_merge (&union_set, &dest_state->nodes);
3057 if (BE (err != REG_NOERROR, 0))
3058 {
3059 re_node_set_free (&union_set);
3060 return err;
3061 }
3062 }
3063 ok = re_node_set_insert (&union_set, next_node);
3064 if (BE (! ok, 0))
3065 {
3066 re_node_set_free (&union_set);
3067 return REG_ESPACE;
3068 }
3069 mctx->state_log[next_idx] = re_acquire_state (&err, dfa,
3070 &union_set);
3071 if (BE (mctx->state_log[next_idx] == NULL
3072 && err != REG_NOERROR, 0))
3073 {
3074 re_node_set_free (&union_set);
3075 return err;
3076 }
3077 }
3078 }
3079#endif /* RE_ENABLE_I18N */
3080 if (naccepted
3081 || check_node_accept (mctx, dfa->nodes + cur_node, str_idx))
3082 {
3083 ok = re_node_set_insert (next_nodes, dfa->nexts[cur_node]);
3084 if (BE (! ok, 0))
3085 {
3086 re_node_set_free (&union_set);
3087 return REG_ESPACE;
3088 }
3089 }
3090 }
3091 re_node_set_free (&union_set);
3092 return REG_NOERROR;
3093}
3094
3095/* For all the nodes in CUR_NODES, add the epsilon closures of them to
3096 CUR_NODES, however exclude the nodes which are:
3097 - inside the sub expression whose number is EX_SUBEXP, if FL_OPEN.
3098 - out of the sub expression whose number is EX_SUBEXP, if !FL_OPEN.
3099*/
3100
3101static reg_errcode_t
3102internal_function
3103check_arrival_expand_ecl (re_dfa_t *dfa, re_node_set *cur_nodes,
3104 Idx ex_subexp, int type)
3105{
3106 reg_errcode_t err;
3107 Idx idx, outside_node;
3108 re_node_set new_nodes;
3109#ifdef DEBUG
3110 assert (cur_nodes->nelem);
3111#endif
3112 err = re_node_set_alloc (&new_nodes, cur_nodes->nelem);
3113 if (BE (err != REG_NOERROR, 0))
3114 return err;
3115 /* Create a new node set NEW_NODES with the nodes which are epsilon
3116 closures of the node in CUR_NODES. */
3117
3118 for (idx = 0; idx < cur_nodes->nelem; ++idx)
3119 {
3120 Idx cur_node = cur_nodes->elems[idx];
3121 re_node_set *eclosure = dfa->eclosures + cur_node;
3122 outside_node = find_subexp_node (dfa, eclosure, ex_subexp, type);
3123 if (outside_node == REG_MISSING)
3124 {
3125 /* There are no problematic nodes, just merge them. */
3126 err = re_node_set_merge (&new_nodes, eclosure);
3127 if (BE (err != REG_NOERROR, 0))
3128 {
3129 re_node_set_free (&new_nodes);
3130 return err;
3131 }
3132 }
3133 else
3134 {
3135 /* There are problematic nodes, re-calculate incrementally. */
3136 err = check_arrival_expand_ecl_sub (dfa, &new_nodes, cur_node,
3137 ex_subexp, type);
3138 if (BE (err != REG_NOERROR, 0))
3139 {
3140 re_node_set_free (&new_nodes);
3141 return err;
3142 }
3143 }
3144 }
3145 re_node_set_free (cur_nodes);
3146 *cur_nodes = new_nodes;
3147 return REG_NOERROR;
3148}
3149
3150/* Helper function for check_arrival_expand_ecl.
3151 Check incrementally the epsilon closure of TARGET, and if it isn't
3152 problematic append it to DST_NODES. */
3153
3154static reg_errcode_t
3155internal_function
3156check_arrival_expand_ecl_sub (re_dfa_t *dfa, re_node_set *dst_nodes,
3157 Idx target, Idx ex_subexp, int type)
3158{
3159 Idx cur_node;
3160 for (cur_node = target; !re_node_set_contains (dst_nodes, cur_node);)
3161 {
3162 bool ok;
3163
3164 if (dfa->nodes[cur_node].type == type
3165 && dfa->nodes[cur_node].opr.idx == ex_subexp)
3166 {
3167 if (type == OP_CLOSE_SUBEXP)
3168 {
3169 ok = re_node_set_insert (dst_nodes, cur_node);
3170 if (BE (! ok, 0))
3171 return REG_ESPACE;
3172 }
3173 break;
3174 }
3175 ok = re_node_set_insert (dst_nodes, cur_node);
3176 if (BE (! ok, 0))
3177 return REG_ESPACE;
3178 if (dfa->edests[cur_node].nelem == 0)
3179 break;
3180 if (dfa->edests[cur_node].nelem == 2)
3181 {
3182 reg_errcode_t ret =
3183 check_arrival_expand_ecl_sub (dfa, dst_nodes,
3184 dfa->edests[cur_node].elems[1],
3185 ex_subexp, type);
3186 if (BE (ret != REG_NOERROR, 0))
3187 return ret;
3188 }
3189 cur_node = dfa->edests[cur_node].elems[0];
3190 }
3191 return REG_NOERROR;
3192}
3193
3194
3195/* For all the back references in the current state, calculate the
3196 destination of the back references by the appropriate entry
3197 in MCTX->BKREF_ENTS. */
3198
3199static reg_errcode_t
3200internal_function
3201expand_bkref_cache (re_match_context_t *mctx, re_node_set *cur_nodes,
3202 Idx cur_str, Idx subexp_num, int type)
3203{
3204 re_dfa_t *const dfa = mctx->dfa;
3205 reg_errcode_t err;
3206 Idx cache_idx_start = search_cur_bkref_entry (mctx, cur_str);
3207 struct re_backref_cache_entry *ent;
3208
3209 if (cache_idx_start == REG_MISSING)
3210 return REG_NOERROR;
3211
3212 restart:
3213 ent = mctx->bkref_ents + cache_idx_start;
3214 do
3215 {
3216 Idx to_idx, next_node;
3217
3218 /* Is this entry ENT is appropriate? */
3219 if (!re_node_set_contains (cur_nodes, ent->node))
3220 continue; /* No. */
3221
3222 to_idx = cur_str + ent->subexp_to - ent->subexp_from;
3223 /* Calculate the destination of the back reference, and append it
3224 to MCTX->STATE_LOG. */
3225 if (to_idx == cur_str)
3226 {
3227 /* The backreference did epsilon transit, we must re-check all the
3228 node in the current state. */
3229 re_node_set new_dests;
3230 reg_errcode_t err2, err3;
3231 next_node = dfa->edests[ent->node].elems[0];
3232 if (re_node_set_contains (cur_nodes, next_node))
3233 continue;
3234 err = re_node_set_init_1 (&new_dests, next_node);
3235 err2 = check_arrival_expand_ecl (dfa, &new_dests, subexp_num, type);
3236 err3 = re_node_set_merge (cur_nodes, &new_dests);
3237 re_node_set_free (&new_dests);
3238 if (BE (err != REG_NOERROR || err2 != REG_NOERROR
3239 || err3 != REG_NOERROR, 0))
3240 {
3241 err = (err != REG_NOERROR ? err
3242 : (err2 != REG_NOERROR ? err2 : err3));
3243 return err;
3244 }
3245 /* TODO: It is still inefficient... */
3246 goto restart;
3247 }
3248 else
3249 {
3250 re_node_set union_set;
3251 next_node = dfa->nexts[ent->node];
3252 if (mctx->state_log[to_idx])
3253 {
3254 bool ok;
3255 if (re_node_set_contains (&mctx->state_log[to_idx]->nodes,
3256 next_node))
3257 continue;
3258 err = re_node_set_init_copy (&union_set,
3259 &mctx->state_log[to_idx]->nodes);
3260 ok = re_node_set_insert (&union_set, next_node);
3261 if (BE (err != REG_NOERROR || ! ok, 0))
3262 {
3263 re_node_set_free (&union_set);
3264 err = err != REG_NOERROR ? err : REG_ESPACE;
3265 return err;
3266 }
3267 }
3268 else
3269 {
3270 err = re_node_set_init_1 (&union_set, next_node);
3271 if (BE (err != REG_NOERROR, 0))
3272 return err;
3273 }
3274 mctx->state_log[to_idx] = re_acquire_state (&err, dfa, &union_set);
3275 re_node_set_free (&union_set);
3276 if (BE (mctx->state_log[to_idx] == NULL
3277 && err != REG_NOERROR, 0))
3278 return err;
3279 }
3280 }
3281 while (ent++->more);
3282 return REG_NOERROR;
3283}
3284
3285/* Build transition table for the state.
3286 Return true if successful. */
3287
3288static bool
3289internal_function
3290build_trtable (re_dfa_t *dfa, re_dfastate_t *state)
3291{
3292 reg_errcode_t err;
3293 Idx i, j;
3294 int ch;
3295 bool need_word_trtable = false;
3296 bitset_word elem, mask;
3297 bool dests_node_malloced = false, dest_states_malloced = false;
3298 Idx ndests; /* Number of the destination states from `state'. */
3299 re_dfastate_t **trtable;
3300 re_dfastate_t **dest_states = NULL, **dest_states_word, **dest_states_nl;
3301 re_node_set follows, *dests_node;
3302 bitset *dests_ch;
3303 bitset acceptable;
3304
3305 struct dests_alloc
3306 {
3307 re_node_set dests_node[SBC_MAX];
3308 bitset dests_ch[SBC_MAX];
3309 } *dests_alloc;
3310
3311 /* We build DFA states which corresponds to the destination nodes
3312 from `state'. `dests_node[i]' represents the nodes which i-th
3313 destination state contains, and `dests_ch[i]' represents the
3314 characters which i-th destination state accepts. */
3315 if (__libc_use_alloca (sizeof (struct dests_alloc)))
3316 dests_alloc = (struct dests_alloc *) alloca (sizeof dests_alloc[0]);
3317 else
3318 {
3319 dests_alloc = re_malloc (struct dests_alloc, 1);
3320 if (BE (dests_alloc == NULL, 0))
3321 return false;
3322 dests_node_malloced = true;
3323 }
3324 dests_node = dests_alloc->dests_node;
3325 dests_ch = dests_alloc->dests_ch;
3326
3327 /* Initialize transiton table. */
3328 state->word_trtable = state->trtable = NULL;
3329
3330 /* At first, group all nodes belonging to `state' into several
3331 destinations. */
3332 ndests = group_nodes_into_DFAstates (dfa, state, dests_node, dests_ch);
3333 if (BE (! REG_VALID_NONZERO_INDEX (ndests), 0))
3334 {
3335 if (dests_node_malloced)
3336 free (dests_alloc);
3337 if (ndests == 0)
3338 {
3339 state->trtable = re_calloc (re_dfastate_t *, SBC_MAX);
3340 return true;
3341 }
3342 return false;
3343 }
3344
3345 err = re_node_set_alloc (&follows, ndests + 1);
3346 if (BE (err != REG_NOERROR, 0))
3347 goto out_free;
3348
3349 /* Avoid arithmetic overflow in size calculation. */
3350 if (BE (((SIZE_MAX - (sizeof (re_node_set) + sizeof (bitset)) * SBC_MAX)
3351 / (3 * sizeof (re_dfastate_t *)))
3352 < ndests, 0))
3353 goto out_free;
3354
3355 if (__libc_use_alloca ((sizeof (re_node_set) + sizeof (bitset)) * SBC_MAX
3356 + ndests * 3 * sizeof (re_dfastate_t *)))
3357 dest_states = (re_dfastate_t **)
3358 alloca (ndests * 3 * sizeof (re_dfastate_t *));
3359 else
3360 {
3361 dest_states = (re_dfastate_t **)
3362 malloc (ndests * 3 * sizeof (re_dfastate_t *));
3363 if (BE (dest_states == NULL, 0))
3364 {
3365out_free:
3366 if (dest_states_malloced)
3367 free (dest_states);
3368 re_node_set_free (&follows);
3369 for (i = 0; i < ndests; ++i)
3370 re_node_set_free (dests_node + i);
3371 if (dests_node_malloced)
3372 free (dests_alloc);
3373 return false;
3374 }
3375 dest_states_malloced = true;
3376 }
3377 dest_states_word = dest_states + ndests;
3378 dest_states_nl = dest_states_word + ndests;
3379 bitset_empty (acceptable);
3380
3381 /* Then build the states for all destinations. */
3382 for (i = 0; i < ndests; ++i)
3383 {
3384 Idx next_node;
3385 re_node_set_empty (&follows);
3386 /* Merge the follows of this destination states. */
3387 for (j = 0; j < dests_node[i].nelem; ++j)
3388 {
3389 next_node = dfa->nexts[dests_node[i].elems[j]];
3390 if (next_node != REG_MISSING)
3391 {
3392 err = re_node_set_merge (&follows, dfa->eclosures + next_node);
3393 if (BE (err != REG_NOERROR, 0))
3394 goto out_free;
3395 }
3396 }
3397 dest_states[i] = re_acquire_state_context (&err, dfa, &follows, 0);
3398 if (BE (dest_states[i] == NULL && err != REG_NOERROR, 0))
3399 goto out_free;
3400 /* If the new state has context constraint,
3401 build appropriate states for these contexts. */
3402 if (dest_states[i]->has_constraint)
3403 {
3404 dest_states_word[i] = re_acquire_state_context (&err, dfa, &follows,
3405 CONTEXT_WORD);
3406 if (BE (dest_states_word[i] == NULL && err != REG_NOERROR, 0))
3407 goto out_free;
3408
3409 if (dest_states[i] != dest_states_word[i] && dfa->mb_cur_max > 1)
3410 need_word_trtable = true;
3411
3412 dest_states_nl[i] = re_acquire_state_context (&err, dfa, &follows,
3413 CONTEXT_NEWLINE);
3414 if (BE (dest_states_nl[i] == NULL && err != REG_NOERROR, 0))
3415 goto out_free;
3416 }
3417 else
3418 {
3419 dest_states_word[i] = dest_states[i];
3420 dest_states_nl[i] = dest_states[i];
3421 }
3422 bitset_merge (acceptable, dests_ch[i]);
3423 }
3424
3425 if (!BE (need_word_trtable, 0))
3426 {
3427 /* We don't care about whether the following character is a word
3428 character, or we are in a single-byte character set so we can
3429 discern by looking at the character code: allocate a
3430 256-entry transition table. */
3431 trtable = state->trtable = re_calloc (re_dfastate_t *, SBC_MAX);
3432 if (BE (trtable == NULL, 0))
3433 goto out_free;
3434
3435 /* For all characters ch...: */
3436 for (i = 0; i < BITSET_WORDS; ++i)
3437 for (ch = i * BITSET_WORD_BITS, elem = acceptable[i], mask = 1;
3438 elem;
3439 mask <<= 1, elem >>= 1, ++ch)
3440 if (BE (elem & 1, 0))
3441 {
3442 /* There must be exactly one destination which accepts
3443 character ch. See group_nodes_into_DFAstates. */
3444 for (j = 0; (dests_ch[j][i] & mask) == 0; ++j)
3445 ;
3446
3447 /* j-th destination accepts the word character ch. */
3448 if (dfa->word_char[i] & mask)
3449 trtable[ch] = dest_states_word[j];
3450 else
3451 trtable[ch] = dest_states[j];
3452 }
3453 }
3454 else
3455 {
3456 /* We care about whether the following character is a word
3457 character, and we are in a multi-byte character set: discern
3458 by looking at the character code: build two 256-entry
3459 transition tables, one starting at trtable[0] and one
3460 starting at trtable[SBC_MAX]. */
3461 trtable = state->word_trtable = re_calloc (re_dfastate_t *, 2 * SBC_MAX);
3462 if (BE (trtable == NULL, 0))
3463 goto out_free;
3464
3465 /* For all characters ch...: */
3466 for (i = 0; i < BITSET_WORDS; ++i)
3467 for (ch = i * BITSET_WORD_BITS, elem = acceptable[i], mask = 1;
3468 elem;
3469 mask <<= 1, elem >>= 1, ++ch)
3470 if (BE (elem & 1, 0))
3471 {
3472 /* There must be exactly one destination which accepts
3473 character ch. See group_nodes_into_DFAstates. */
3474 for (j = 0; (dests_ch[j][i] & mask) == 0; ++j)
3475 ;
3476
3477 /* j-th destination accepts the word character ch. */
3478 trtable[ch] = dest_states[j];
3479 trtable[ch + SBC_MAX] = dest_states_word[j];
3480 }
3481 }
3482
3483 /* new line */
3484 if (bitset_contain (acceptable, NEWLINE_CHAR))
3485 {
3486 /* The current state accepts newline character. */
3487 for (j = 0; j < ndests; ++j)
3488 if (bitset_contain (dests_ch[j], NEWLINE_CHAR))
3489 {
3490 /* k-th destination accepts newline character. */
3491 trtable[NEWLINE_CHAR] = dest_states_nl[j];
3492 if (need_word_trtable)
3493 trtable[NEWLINE_CHAR + SBC_MAX] = dest_states_nl[j];
3494 /* There must be only one destination which accepts
3495 newline. See group_nodes_into_DFAstates. */
3496 break;
3497 }
3498 }
3499
3500 if (dest_states_malloced)
3501 free (dest_states);
3502
3503 re_node_set_free (&follows);
3504 for (i = 0; i < ndests; ++i)
3505 re_node_set_free (dests_node + i);
3506
3507 if (dests_node_malloced)
3508 free (dests_alloc);
3509
3510 return true;
3511}
3512
3513/* Group all nodes belonging to STATE into several destinations.
3514 Then for all destinations, set the nodes belonging to the destination
3515 to DESTS_NODE[i] and set the characters accepted by the destination
3516 to DEST_CH[i]. This function return the number of destinations. */
3517
3518static Idx
3519internal_function
3520group_nodes_into_DFAstates (const re_dfa_t *dfa, const re_dfastate_t *state,
3521 re_node_set *dests_node, bitset *dests_ch)
3522{
3523 reg_errcode_t err;
3524 bool ok;
3525 Idx i, j, k;
3526 Idx ndests; /* Number of the destinations from `state'. */
3527 bitset accepts; /* Characters a node can accept. */
3528 const re_node_set *cur_nodes = &state->nodes;
3529 bitset_empty (accepts);
3530 ndests = 0;
3531
3532 /* For all the nodes belonging to `state', */
3533 for (i = 0; i < cur_nodes->nelem; ++i)
3534 {
3535 re_token_t *node = &dfa->nodes[cur_nodes->elems[i]];
3536 re_token_type_t type = node->type;
3537 unsigned int constraint = node->constraint;
3538
3539 /* Enumerate all single byte character this node can accept. */
3540 if (type == CHARACTER)
3541 bitset_set (accepts, node->opr.c);
3542 else if (type == SIMPLE_BRACKET)
3543 {
3544 bitset_merge (accepts, node->opr.sbcset);
3545 }
3546 else if (type == OP_PERIOD)
3547 {
3548#ifdef RE_ENABLE_I18N
3549 if (dfa->mb_cur_max > 1)
3550 bitset_merge (accepts, dfa->sb_char);
3551 else
3552#endif
3553 bitset_set_all (accepts);
3554 if (!(dfa->syntax & REG_DOT_NEWLINE))
3555 bitset_clear (accepts, '\n');
3556 if (dfa->syntax & REG_DOT_NOT_NULL)
3557 bitset_clear (accepts, '\0');
3558 }
3559#ifdef RE_ENABLE_I18N
3560 else if (type == OP_UTF8_PERIOD)
3561 {
3562 if (SBC_MAX / 2 % BITSET_WORD_BITS == 0)
3563 memset (accepts, -1, sizeof accepts / 2);
3564 else
3565 bitset_merge (accepts, utf8_sb_map);
3566 if (!(dfa->syntax & REG_DOT_NEWLINE))
3567 bitset_clear (accepts, '\n');
3568 if (dfa->syntax & REG_DOT_NOT_NULL)
3569 bitset_clear (accepts, '\0');
3570 }
3571#endif
3572 else
3573 continue;
3574
3575 /* Check the `accepts' and sift the characters which are not
3576 match it the context. */
3577 if (constraint)
3578 {
3579 if (constraint & NEXT_NEWLINE_CONSTRAINT)
3580 {
3581 bool accepts_newline = bitset_contain (accepts, NEWLINE_CHAR);
3582 bitset_empty (accepts);
3583 if (accepts_newline)
3584 bitset_set (accepts, NEWLINE_CHAR);
3585 else
3586 continue;
3587 }
3588 if (constraint & NEXT_ENDBUF_CONSTRAINT)
3589 {
3590 bitset_empty (accepts);
3591 continue;
3592 }
3593
3594 if (constraint & NEXT_WORD_CONSTRAINT)
3595 {
3596 bitset_word any_set = 0;
3597 if (type == CHARACTER && !node->word_char)
3598 {
3599 bitset_empty (accepts);
3600 continue;
3601 }
3602#ifdef RE_ENABLE_I18N
3603 if (dfa->mb_cur_max > 1)
3604 for (j = 0; j < BITSET_WORDS; ++j)
3605 any_set |= (accepts[j] &= (dfa->word_char[j] | ~dfa->sb_char[j]));
3606 else
3607#endif
3608 for (j = 0; j < BITSET_WORDS; ++j)
3609 any_set |= (accepts[j] &= dfa->word_char[j]);
3610 if (!any_set)
3611 continue;
3612 }
3613 if (constraint & NEXT_NOTWORD_CONSTRAINT)
3614 {
3615 bitset_word any_set = 0;
3616 if (type == CHARACTER && node->word_char)
3617 {
3618 bitset_empty (accepts);
3619 continue;
3620 }
3621#ifdef RE_ENABLE_I18N
3622 if (dfa->mb_cur_max > 1)
3623 for (j = 0; j < BITSET_WORDS; ++j)
3624 any_set |= (accepts[j] &= ~(dfa->word_char[j] & dfa->sb_char[j]));
3625 else
3626#endif
3627 for (j = 0; j < BITSET_WORDS; ++j)
3628 any_set |= (accepts[j] &= ~dfa->word_char[j]);
3629 if (!any_set)
3630 continue;
3631 }
3632 }
3633
3634 /* Then divide `accepts' into DFA states, or create a new
3635 state. Above, we make sure that accepts is not empty. */
3636 for (j = 0; j < ndests; ++j)
3637 {
3638 bitset intersec; /* Intersection sets, see below. */
3639 bitset remains;
3640 /* Flags, see below. */
3641 bitset_word has_intersec, not_subset, not_consumed;
3642
3643 /* Optimization, skip if this state doesn't accept the character. */
3644 if (type == CHARACTER && !bitset_contain (dests_ch[j], node->opr.c))
3645 continue;
3646
3647 /* Enumerate the intersection set of this state and `accepts'. */
3648 has_intersec = 0;
3649 for (k = 0; k < BITSET_WORDS; ++k)
3650 has_intersec |= intersec[k] = accepts[k] & dests_ch[j][k];
3651 /* And skip if the intersection set is empty. */
3652 if (!has_intersec)
3653 continue;
3654
3655 /* Then check if this state is a subset of `accepts'. */
3656 not_subset = not_consumed = 0;
3657 for (k = 0; k < BITSET_WORDS; ++k)
3658 {
3659 not_subset |= remains[k] = ~accepts[k] & dests_ch[j][k];
3660 not_consumed |= accepts[k] = accepts[k] & ~dests_ch[j][k];
3661 }
3662
3663 /* If this state isn't a subset of `accepts', create a
3664 new group state, which has the `remains'. */
3665 if (not_subset)
3666 {
3667 bitset_copy (dests_ch[ndests], remains);
3668 bitset_copy (dests_ch[j], intersec);
3669 err = re_node_set_init_copy (dests_node + ndests, &dests_node[j]);
3670 if (BE (err != REG_NOERROR, 0))
3671 goto error_return;
3672 ++ndests;
3673 }
3674
3675 /* Put the position in the current group. */
3676 ok = re_node_set_insert (&dests_node[j], cur_nodes->elems[i]);
3677 if (BE (! ok, 0))
3678 goto error_return;
3679
3680 /* If all characters are consumed, go to next node. */
3681 if (!not_consumed)
3682 break;
3683 }
3684 /* Some characters remain, create a new group. */
3685 if (j == ndests)
3686 {
3687 bitset_copy (dests_ch[ndests], accepts);
3688 err = re_node_set_init_1 (dests_node + ndests, cur_nodes->elems[i]);
3689 if (BE (err != REG_NOERROR, 0))
3690 goto error_return;
3691 ++ndests;
3692 bitset_empty (accepts);
3693 }
3694 }
3695 return ndests;
3696 error_return:
3697 for (j = 0; j < ndests; ++j)
3698 re_node_set_free (dests_node + j);
3699 return REG_MISSING;
3700}
3701
3702#ifdef RE_ENABLE_I18N
3703/* Check how many bytes the node `dfa->nodes[node_idx]' accepts.
3704 Return the number of the bytes the node accepts.
3705 STR_IDX is the current index of the input string.
3706
3707 This function handles the nodes which can accept one character, or
3708 one collating element like '.', '[a-z]', opposite to the other nodes
3709 can only accept one byte. */
3710
3711static int
3712internal_function
3713check_node_accept_bytes (re_dfa_t *dfa, Idx node_idx,
3714 const re_string_t *input, Idx str_idx)
3715{
3716 const re_token_t *node = dfa->nodes + node_idx;
3717 int char_len, elem_len;
3718 Idx i;
3719
3720 if (BE (node->type == OP_UTF8_PERIOD, 0))
3721 {
3722 unsigned char c = re_string_byte_at (input, str_idx), d;
3723 if (BE (c < 0xc2, 1))
3724 return 0;
3725
3726 if (str_idx + 2 > input->len)
3727 return 0;
3728
3729 d = re_string_byte_at (input, str_idx + 1);
3730 if (c < 0xe0)
3731 return (d < 0x80 || d > 0xbf) ? 0 : 2;
3732 else if (c < 0xf0)
3733 {
3734 char_len = 3;
3735 if (c == 0xe0 && d < 0xa0)
3736 return 0;
3737 }
3738 else if (c < 0xf8)
3739 {
3740 char_len = 4;
3741 if (c == 0xf0 && d < 0x90)
3742 return 0;
3743 }
3744 else if (c < 0xfc)
3745 {
3746 char_len = 5;
3747 if (c == 0xf8 && d < 0x88)
3748 return 0;
3749 }
3750 else if (c < 0xfe)
3751 {
3752 char_len = 6;
3753 if (c == 0xfc && d < 0x84)
3754 return 0;
3755 }
3756 else
3757 return 0;
3758
3759 if (str_idx + char_len > input->len)
3760 return 0;
3761
3762 for (i = 1; i < char_len; ++i)
3763 {
3764 d = re_string_byte_at (input, str_idx + i);
3765 if (d < 0x80 || d > 0xbf)
3766 return 0;
3767 }
3768 return char_len;
3769 }
3770
3771 char_len = re_string_char_size_at (input, str_idx);
3772 if (node->type == OP_PERIOD)
3773 {
3774 if (char_len <= 1)
3775 return 0;
3776 /* FIXME: I don't think this if is needed, as both '\n'
3777 and '\0' are char_len == 1. */
3778 /* '.' accepts any one character except the following two cases. */
3779 if ((!(dfa->syntax & REG_DOT_NEWLINE) &&
3780 re_string_byte_at (input, str_idx) == '\n') ||
3781 ((dfa->syntax & REG_DOT_NOT_NULL) &&
3782 re_string_byte_at (input, str_idx) == '\0'))
3783 return 0;
3784 return char_len;
3785 }
3786
3787 elem_len = re_string_elem_size_at (input, str_idx);
3788 if ((elem_len <= 1 && char_len <= 1) || char_len == 0)
3789 return 0;
3790
3791 if (node->type == COMPLEX_BRACKET)
3792 {
3793 const re_charset_t *cset = node->opr.mbcset;
3794# ifdef _LIBC
3795 const unsigned char *pin
3796 = ((const unsigned char *) re_string_get_buffer (input) + str_idx);
3797 Idx j;
3798 uint32_t nrules;
3799# endif /* _LIBC */
3800 int match_len = 0;
3801 wchar_t wc = ((cset->nranges || cset->nchar_classes || cset->nmbchars)
3802 ? re_string_wchar_at (input, str_idx) : 0);
3803
3804 /* match with multibyte character? */
3805 for (i = 0; i < cset->nmbchars; ++i)
3806 if (wc == cset->mbchars[i])
3807 {
3808 match_len = char_len;
3809 goto check_node_accept_bytes_match;
3810 }
3811 /* match with character_class? */
3812 for (i = 0; i < cset->nchar_classes; ++i)
3813 {
3814 wctype_t wt = cset->char_classes[i];
3815 if (__iswctype (wc, wt))
3816 {
3817 match_len = char_len;
3818 goto check_node_accept_bytes_match;
3819 }
3820 }
3821
3822# ifdef _LIBC
3823 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3824 if (nrules != 0)
3825 {
3826 unsigned int in_collseq = 0;
3827 const int32_t *table, *indirect;
3828 const unsigned char *weights, *extra;
3829 const char *collseqwc;
3830 int32_t idx;
3831 /* This #include defines a local function! */
3832# include <locale/weight.h>
3833
3834 /* match with collating_symbol? */
3835 if (cset->ncoll_syms)
3836 extra = (const unsigned char *)
3837 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
3838 for (i = 0; i < cset->ncoll_syms; ++i)
3839 {
3840 const unsigned char *coll_sym = extra + cset->coll_syms[i];
3841 /* Compare the length of input collating element and
3842 the length of current collating element. */
3843 if (*coll_sym != elem_len)
3844 continue;
3845 /* Compare each bytes. */
3846 for (j = 0; j < *coll_sym; j++)
3847 if (pin[j] != coll_sym[1 + j])
3848 break;
3849 if (j == *coll_sym)
3850 {
3851 /* Match if every bytes is equal. */
3852 match_len = j;
3853 goto check_node_accept_bytes_match;
3854 }
3855 }
3856
3857 if (cset->nranges)
3858 {
3859 if (elem_len <= char_len)
3860 {
3861 collseqwc = _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQWC);
3862 in_collseq = __collseq_table_lookup (collseqwc, wc);
3863 }
3864 else
3865 in_collseq = find_collation_sequence_value (pin, elem_len);
3866 }
3867 /* match with range expression? */
3868 for (i = 0; i < cset->nranges; ++i)
3869 if (cset->range_starts[i] <= in_collseq
3870 && in_collseq <= cset->range_ends[i])
3871 {
3872 match_len = elem_len;
3873 goto check_node_accept_bytes_match;
3874 }
3875
3876 /* match with equivalence_class? */
3877 if (cset->nequiv_classes)
3878 {
3879 const unsigned char *cp = pin;
3880 table = (const int32_t *)
3881 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
3882 weights = (const unsigned char *)
3883 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB);
3884 extra = (const unsigned char *)
3885 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
3886 indirect = (const int32_t *)
3887 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB);
3888 idx = findidx (&cp);
3889 if (idx > 0)
3890 for (i = 0; i < cset->nequiv_classes; ++i)
3891 {
3892 int32_t equiv_class_idx = cset->equiv_classes[i];
3893 size_t weight_len = weights[idx];
3894 if (weight_len == weights[equiv_class_idx])
3895 {
3896 Idx cnt = 0;
3897 while (cnt <= weight_len
3898 && (weights[equiv_class_idx + 1 + cnt]
3899 == weights[idx + 1 + cnt]))
3900 ++cnt;
3901 if (cnt > weight_len)
3902 {
3903 match_len = elem_len;
3904 goto check_node_accept_bytes_match;
3905 }
3906 }
3907 }
3908 }
3909 }
3910 else
3911# endif /* _LIBC */
3912 {
3913 /* match with range expression? */
3914#if __GNUC__ >= 2
3915 wchar_t cmp_buf[] = {L'\0', L'\0', wc, L'\0', L'\0', L'\0'};
3916#else
3917 wchar_t cmp_buf[] = {L'\0', L'\0', L'\0', L'\0', L'\0', L'\0'};
3918 cmp_buf[2] = wc;
3919#endif
3920 for (i = 0; i < cset->nranges; ++i)
3921 {
3922 cmp_buf[0] = cset->range_starts[i];
3923 cmp_buf[4] = cset->range_ends[i];
3924 if (wcscoll (cmp_buf, cmp_buf + 2) <= 0
3925 && wcscoll (cmp_buf + 2, cmp_buf + 4) <= 0)
3926 {
3927 match_len = char_len;
3928 goto check_node_accept_bytes_match;
3929 }
3930 }
3931 }
3932 check_node_accept_bytes_match:
3933 if (!cset->non_match)
3934 return match_len;
3935 else
3936 {
3937 if (match_len > 0)
3938 return 0;
3939 else
3940 return (elem_len > char_len) ? elem_len : char_len;
3941 }
3942 }
3943 return 0;
3944}
3945
3946# ifdef _LIBC
3947static unsigned int
3948find_collation_sequence_value (const unsigned char *mbs, size_t mbs_len)
3949{
3950 uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3951 if (nrules == 0)
3952 {
3953 if (mbs_len == 1)
3954 {
3955 /* No valid character. Match it as a single byte character. */
3956 const unsigned char *collseq = (const unsigned char *)
3957 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQMB);
3958 return collseq[mbs[0]];
3959 }
3960 return UINT_MAX;
3961 }
3962 else
3963 {
3964 int32_t idx;
3965 const unsigned char *extra = (const unsigned char *)
3966 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
3967 int32_t extrasize = (const unsigned char *)
3968 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB + 1) - extra;
3969
3970 for (idx = 0; idx < extrasize;)
3971 {
3972 int mbs_cnt;
3973 bool found = false;
3974 int32_t elem_mbs_len;
3975 /* Skip the name of collating element name. */
3976 idx = idx + extra[idx] + 1;
3977 elem_mbs_len = extra[idx++];
3978 if (mbs_len == elem_mbs_len)
3979 {
3980 for (mbs_cnt = 0; mbs_cnt < elem_mbs_len; ++mbs_cnt)
3981 if (extra[idx + mbs_cnt] != mbs[mbs_cnt])
3982 break;
3983 if (mbs_cnt == elem_mbs_len)
3984 /* Found the entry. */
3985 found = true;
3986 }
3987 /* Skip the byte sequence of the collating element. */
3988 idx += elem_mbs_len;
3989 /* Adjust for the alignment. */
3990 idx = (idx + 3) & ~3;
3991 /* Skip the collation sequence value. */
3992 idx += sizeof (uint32_t);
3993 /* Skip the wide char sequence of the collating element. */
3994 idx = idx + sizeof (uint32_t) * (extra[idx] + 1);
3995 /* If we found the entry, return the sequence value. */
3996 if (found)
3997 return *(uint32_t *) (extra + idx);
3998 /* Skip the collation sequence value. */
3999 idx += sizeof (uint32_t);
4000 }
4001 return UINT_MAX;
4002 }
4003}
4004# endif /* _LIBC */
4005#endif /* RE_ENABLE_I18N */
4006
4007/* Check whether the node accepts the byte which is IDX-th
4008 byte of the INPUT. */
4009
4010static bool
4011internal_function
4012check_node_accept (const re_match_context_t *mctx, const re_token_t *node,
4013 Idx idx)
4014{
4015 unsigned char ch;
4016 ch = re_string_byte_at (&mctx->input, idx);
4017 switch (node->type)
4018 {
4019 case CHARACTER:
4020 if (node->opr.c != ch)
4021 return false;
4022 break;
4023
4024 case SIMPLE_BRACKET:
4025 if (!bitset_contain (node->opr.sbcset, ch))
4026 return false;
4027 break;
4028
4029#ifdef RE_ENABLE_I18N
4030 case OP_UTF8_PERIOD:
4031 if (ch >= 0x80)
4032 return false;
4033 /* FALLTHROUGH */
4034#endif
4035 case OP_PERIOD:
4036 if ((ch == '\n' && !(mctx->dfa->syntax & REG_DOT_NEWLINE))
4037 || (ch == '\0' && (mctx->dfa->syntax & REG_DOT_NOT_NULL)))
4038 return false;
4039 break;
4040
4041 default:
4042 return false;
4043 }
4044
4045 if (node->constraint)
4046 {
4047 /* The node has constraints. Check whether the current context
4048 satisfies the constraints. */
4049 unsigned int context = re_string_context_at (&mctx->input, idx,
4050 mctx->eflags);
4051 if (NOT_SATISFY_NEXT_CONSTRAINT (node->constraint, context))
4052 return false;
4053 }
4054
4055 return true;
4056}
4057
4058/* Extend the buffers, if the buffers have run out. */
4059
4060static reg_errcode_t
4061internal_function
4062extend_buffers (re_match_context_t *mctx)
4063{
4064 reg_errcode_t ret;
4065 re_string_t *pstr = &mctx->input;
4066
4067 /* Double the lengthes of the buffers. */
4068 ret = re_string_realloc_buffers (pstr, pstr->bufs_len * 2);
4069 if (BE (ret != REG_NOERROR, 0))
4070 return ret;
4071
4072 if (mctx->state_log != NULL)
4073 {
4074 /* And double the length of state_log. */
4075 /* XXX We have no indication of the size of this buffer. If this
4076 allocation fail we have no indication that the state_log array
4077 does not have the right size. */
4078 re_dfastate_t **new_array = re_xrealloc (mctx->state_log, re_dfastate_t *,
4079 pstr->bufs_len + 1);
4080 if (BE (new_array == NULL, 0))
4081 return REG_ESPACE;
4082 mctx->state_log = new_array;
4083 }
4084
4085 /* Then reconstruct the buffers. */
4086 if (pstr->icase)
4087 {
4088#ifdef RE_ENABLE_I18N
4089 if (pstr->mb_cur_max > 1)
4090 {
4091 ret = build_wcs_upper_buffer (pstr);
4092 if (BE (ret != REG_NOERROR, 0))
4093 return ret;
4094 }
4095 else
4096#endif /* RE_ENABLE_I18N */
4097 build_upper_buffer (pstr);
4098 }
4099 else
4100 {
4101#ifdef RE_ENABLE_I18N
4102 if (pstr->mb_cur_max > 1)
4103 build_wcs_buffer (pstr);
4104 else
4105#endif /* RE_ENABLE_I18N */
4106 {
4107 if (pstr->trans != NULL)
4108 re_string_translate_buffer (pstr);
4109 }
4110 }
4111 return REG_NOERROR;
4112}
4113
4114
4115/* Functions for matching context. */
4116
4117/* Initialize MCTX. */
4118
4119static reg_errcode_t
4120internal_function
4121match_ctx_init (re_match_context_t *mctx, int eflags, Idx n)
4122{
4123 mctx->eflags = eflags;
4124 mctx->match_last = REG_MISSING;
4125 if (n > 0)
4126 {
4127 mctx->bkref_ents = re_xmalloc (struct re_backref_cache_entry, n);
4128 mctx->sub_tops = re_xmalloc (re_sub_match_top_t *, n);
4129 if (BE (mctx->bkref_ents == NULL || mctx->sub_tops == NULL, 0))
4130 return REG_ESPACE;
4131 }
4132 /* Already zero-ed by the caller.
4133 else
4134 mctx->bkref_ents = NULL;
4135 mctx->nbkref_ents = 0;
4136 mctx->nsub_tops = 0; */
4137 mctx->abkref_ents = n;
4138 mctx->max_mb_elem_len = 1;
4139 mctx->asub_tops = n;
4140 return REG_NOERROR;
4141}
4142
4143/* Clean the entries which depend on the current input in MCTX.
4144 This function must be invoked when the matcher changes the start index
4145 of the input, or changes the input string. */
4146
4147static void
4148internal_function
4149match_ctx_clean (re_match_context_t *mctx)
4150{
4151 Idx st_idx;
4152 for (st_idx = 0; st_idx < mctx->nsub_tops; ++st_idx)
4153 {
4154 Idx sl_idx;
4155 re_sub_match_top_t *top = mctx->sub_tops[st_idx];
4156 for (sl_idx = 0; sl_idx < top->nlasts; ++sl_idx)
4157 {
4158 re_sub_match_last_t *last = top->lasts[sl_idx];
4159 re_free (last->path.array);
4160 re_free (last);
4161 }
4162 re_free (top->lasts);
4163 if (top->path)
4164 {
4165 re_free (top->path->array);
4166 re_free (top->path);
4167 }
4168 free (top);
4169 }
4170
4171 mctx->nsub_tops = 0;
4172 mctx->nbkref_ents = 0;
4173}
4174
4175/* Free all the memory associated with MCTX. */
4176
4177static void
4178internal_function
4179match_ctx_free (re_match_context_t *mctx)
4180{
4181 /* First, free all the memory associated with MCTX->SUB_TOPS. */
4182 match_ctx_clean (mctx);
4183 re_free (mctx->sub_tops);
4184 re_free (mctx->bkref_ents);
4185}
4186
4187/* Add a new backreference entry to MCTX.
4188 Note that we assume that caller never call this function with duplicate
4189 entry, and call with STR_IDX which isn't smaller than any existing entry.
4190*/
4191
4192static reg_errcode_t
4193internal_function
4194match_ctx_add_entry (re_match_context_t *mctx, Idx node, Idx str_idx,
4195 Idx from, Idx to)
4196{
4197 if (mctx->nbkref_ents >= mctx->abkref_ents)
4198 {
4199 struct re_backref_cache_entry* new_entry;
4200 new_entry = re_x2realloc (mctx->bkref_ents, struct re_backref_cache_entry,
4201 &mctx->abkref_ents);
4202 if (BE (new_entry == NULL, 0))
4203 {
4204 re_free (mctx->bkref_ents);
4205 return REG_ESPACE;
4206 }
4207 mctx->bkref_ents = new_entry;
4208 memset (mctx->bkref_ents + mctx->nbkref_ents, '\0',
4209 (sizeof (struct re_backref_cache_entry)
4210 * (mctx->abkref_ents - mctx->nbkref_ents)));
4211 }
4212 if (mctx->nbkref_ents > 0
4213 && mctx->bkref_ents[mctx->nbkref_ents - 1].str_idx == str_idx)
4214 mctx->bkref_ents[mctx->nbkref_ents - 1].more = 1;
4215
4216 mctx->bkref_ents[mctx->nbkref_ents].node = node;
4217 mctx->bkref_ents[mctx->nbkref_ents].str_idx = str_idx;
4218 mctx->bkref_ents[mctx->nbkref_ents].subexp_from = from;
4219 mctx->bkref_ents[mctx->nbkref_ents].subexp_to = to;
4220
4221 /* This is a cache that saves negative results of check_dst_limits_calc_pos.
4222 If bit N is clear, means that this entry won't epsilon-transition to
4223 an OP_OPEN_SUBEXP or OP_CLOSE_SUBEXP for the N+1-th subexpression. If
4224 it is set, check_dst_limits_calc_pos_1 will recurse and try to find one
4225 such node.
4226
4227 A backreference does not epsilon-transition unless it is empty, so set
4228 to all zeros if FROM != TO. */
4229 mctx->bkref_ents[mctx->nbkref_ents].eps_reachable_subexps_map
4230 = (from == to ? -1 : 0);
4231
4232 mctx->bkref_ents[mctx->nbkref_ents++].more = 0;
4233 if (mctx->max_mb_elem_len < to - from)
4234 mctx->max_mb_elem_len = to - from;
4235 return REG_NOERROR;
4236}
4237
4238/* Return the first entry with the same str_idx, or REG_MISSING if none is
4239 found. Note that MCTX->BKREF_ENTS is already sorted by MCTX->STR_IDX. */
4240
4241static Idx
4242internal_function
4243search_cur_bkref_entry (const re_match_context_t *mctx, Idx str_idx)
4244{
4245 Idx left, right, mid, last;
4246 last = right = mctx->nbkref_ents;
4247 for (left = 0; left < right;)
4248 {
4249 mid = (left + right) / 2;
4250 if (mctx->bkref_ents[mid].str_idx < str_idx)
4251 left = mid + 1;
4252 else
4253 right = mid;
4254 }
4255 if (left < last && mctx->bkref_ents[left].str_idx == str_idx)
4256 return left;
4257 else
4258 return REG_MISSING;
4259}
4260
4261/* Register the node NODE, whose type is OP_OPEN_SUBEXP, and which matches
4262 at STR_IDX. */
4263
4264static reg_errcode_t
4265internal_function
4266match_ctx_add_subtop (re_match_context_t *mctx, Idx node, Idx str_idx)
4267{
4268#ifdef DEBUG
4269 assert (mctx->sub_tops != NULL);
4270 assert (mctx->asub_tops > 0);
4271#endif
4272 if (BE (mctx->nsub_tops == mctx->asub_tops, 0))
4273 {
4274 Idx new_asub_tops = mctx->asub_tops;
4275 re_sub_match_top_t **new_array = re_x2realloc (mctx->sub_tops,
4276 re_sub_match_top_t *,
4277 &new_asub_tops);
4278 if (BE (new_array == NULL, 0))
4279 return REG_ESPACE;
4280 mctx->sub_tops = new_array;
4281 mctx->asub_tops = new_asub_tops;
4282 }
4283 mctx->sub_tops[mctx->nsub_tops] = re_calloc (re_sub_match_top_t, 1);
4284 if (BE (mctx->sub_tops[mctx->nsub_tops] == NULL, 0))
4285 return REG_ESPACE;
4286 mctx->sub_tops[mctx->nsub_tops]->node = node;
4287 mctx->sub_tops[mctx->nsub_tops++]->str_idx = str_idx;
4288 return REG_NOERROR;
4289}
4290
4291/* Register the node NODE, whose type is OP_CLOSE_SUBEXP, and which matches
4292 at STR_IDX, whose corresponding OP_OPEN_SUBEXP is SUB_TOP. */
4293
4294static re_sub_match_last_t *
4295internal_function
4296match_ctx_add_sublast (re_sub_match_top_t *subtop, Idx node, Idx str_idx)
4297{
4298 re_sub_match_last_t *new_entry;
4299 if (BE (subtop->nlasts == subtop->alasts, 0))
4300 {
4301 Idx new_alasts = subtop->alasts;
4302 re_sub_match_last_t **new_array = re_x2realloc (subtop->lasts,
4303 re_sub_match_last_t *,
4304 &new_alasts);
4305 if (BE (new_array == NULL, 0))
4306 return NULL;
4307 subtop->lasts = new_array;
4308 subtop->alasts = new_alasts;
4309 }
4310 new_entry = re_calloc (re_sub_match_last_t, 1);
4311 if (BE (new_entry != NULL, 1))
4312 {
4313 subtop->lasts[subtop->nlasts] = new_entry;
4314 new_entry->node = node;
4315 new_entry->str_idx = str_idx;
4316 ++subtop->nlasts;
4317 }
4318 return new_entry;
4319}
4320
4321static void
4322internal_function
4323sift_ctx_init (re_sift_context_t *sctx,
4324 re_dfastate_t **sifted_sts,
4325 re_dfastate_t **limited_sts,
4326 Idx last_node, Idx last_str_idx)
4327{
4328 sctx->sifted_states = sifted_sts;
4329 sctx->limited_states = limited_sts;
4330 sctx->last_node = last_node;
4331 sctx->last_str_idx = last_str_idx;
4332 re_node_set_init_empty (&sctx->limits);
4333}