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>.
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)
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.
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. */
21 # if defined __GNUC__ && defined __GNUC_MINOR__
22 # define __GNUC_PREREQ(maj, min) \
23 ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
25 # define __GNUC_PREREQ(maj, min) 0
29 #if !__GNUC_PREREQ (3, 1)
30 # define always_inline
33 static reg_errcode_t match_ctx_init (re_match_context_t *cache, int eflags,
34 Idx n) internal_function;
35 static void match_ctx_clean (re_match_context_t *mctx) internal_function;
36 static void match_ctx_free (re_match_context_t *cache) internal_function;
37 static reg_errcode_t match_ctx_add_entry (re_match_context_t *cache, Idx node,
38 Idx str_idx, Idx from, Idx to)
40 static Idx search_cur_bkref_entry (const re_match_context_t *mctx, Idx str_idx)
42 static reg_errcode_t match_ctx_add_subtop (re_match_context_t *mctx, Idx node,
43 Idx str_idx) internal_function;
44 static re_sub_match_last_t * match_ctx_add_sublast (re_sub_match_top_t *subtop,
45 Idx node, Idx str_idx)
47 static void sift_ctx_init (re_sift_context_t *sctx, re_dfastate_t **sifted_sts,
48 re_dfastate_t **limited_sts, Idx last_node,
51 static reg_errcode_t re_search_internal (const regex_t *preg,
52 const char *string, Idx length,
53 Idx start, Idx last_start, Idx stop,
54 size_t nmatch, regmatch_t pmatch[],
55 int eflags) internal_function;
56 static regoff_t re_search_2_stub (struct re_pattern_buffer *bufp,
57 const char *string1, Idx length1,
58 const char *string2, Idx length2,
59 Idx start, regoff_t range,
60 struct re_registers *regs,
61 Idx stop, bool ret_len) internal_function;
62 static regoff_t re_search_stub (struct re_pattern_buffer *bufp,
63 const char *string, Idx length, Idx start,
64 regoff_t range, Idx stop,
65 struct re_registers *regs,
66 bool ret_len) internal_function;
67 static unsigned re_copy_regs (struct re_registers *regs, regmatch_t *pmatch,
68 Idx nregs, int regs_allocated) internal_function;
69 static reg_errcode_t prune_impossible_nodes (re_match_context_t *mctx)
71 static Idx check_matching (re_match_context_t *mctx, bool fl_longest_match,
74 static Idx check_halt_state_context (const re_match_context_t *mctx,
75 const re_dfastate_t *state, Idx idx)
77 static void update_regs (re_dfa_t *dfa, regmatch_t *pmatch,
78 regmatch_t *prev_idx_match, Idx cur_node,
79 Idx cur_idx, Idx nmatch) internal_function;
80 static reg_errcode_t push_fail_stack (struct re_fail_stack_t *fs,
81 Idx str_idx, Idx dest_node, Idx nregs,
83 re_node_set *eps_via_nodes) internal_function;
84 static reg_errcode_t set_regs (const regex_t *preg,
85 const re_match_context_t *mctx,
86 size_t nmatch, regmatch_t *pmatch,
87 bool fl_backtrack) internal_function;
88 static reg_errcode_t free_fail_stack_return (struct re_fail_stack_t *fs) internal_function;
91 static int sift_states_iter_mb (const re_match_context_t *mctx,
92 re_sift_context_t *sctx,
93 Idx node_idx, Idx str_idx, Idx max_str_idx) internal_function;
94 #endif /* RE_ENABLE_I18N */
95 static reg_errcode_t sift_states_backward (re_match_context_t *mctx,
96 re_sift_context_t *sctx) internal_function;
97 static reg_errcode_t build_sifted_states (re_match_context_t *mctx,
98 re_sift_context_t *sctx, Idx str_idx,
99 re_node_set *cur_dest) internal_function;
100 static reg_errcode_t update_cur_sifted_state (re_match_context_t *mctx,
101 re_sift_context_t *sctx,
103 re_node_set *dest_nodes) internal_function;
104 static reg_errcode_t add_epsilon_src_nodes (re_dfa_t *dfa,
105 re_node_set *dest_nodes,
106 const re_node_set *candidates) internal_function;
107 static bool check_dst_limits (const re_match_context_t *mctx,
108 const re_node_set *limits,
109 Idx dst_node, Idx dst_idx, Idx src_node,
110 Idx src_idx) internal_function;
111 static int check_dst_limits_calc_pos_1 (const re_match_context_t *mctx,
112 int boundaries, Idx subexp_idx,
113 Idx from_node, Idx bkref_idx) internal_function;
114 static int check_dst_limits_calc_pos (const re_match_context_t *mctx,
115 Idx limit, Idx subexp_idx,
116 Idx node, Idx str_idx,
117 Idx bkref_idx) internal_function;
118 static reg_errcode_t check_subexp_limits (re_dfa_t *dfa,
119 re_node_set *dest_nodes,
120 const re_node_set *candidates,
122 struct re_backref_cache_entry *bkref_ents,
123 Idx str_idx) internal_function;
124 static reg_errcode_t sift_states_bkref (re_match_context_t *mctx,
125 re_sift_context_t *sctx,
126 Idx str_idx, const re_node_set *candidates) internal_function;
127 static reg_errcode_t merge_state_array (re_dfa_t *dfa, re_dfastate_t **dst,
128 re_dfastate_t **src, Idx num) internal_function;
129 static re_dfastate_t *find_recover_state (reg_errcode_t *err,
130 re_match_context_t *mctx) internal_function;
131 static re_dfastate_t *transit_state (reg_errcode_t *err,
132 re_match_context_t *mctx,
133 re_dfastate_t *state) internal_function;
134 static re_dfastate_t *merge_state_with_log (reg_errcode_t *err,
135 re_match_context_t *mctx,
136 re_dfastate_t *next_state) internal_function;
137 static reg_errcode_t check_subexp_matching_top (re_match_context_t *mctx,
138 re_node_set *cur_nodes,
139 Idx str_idx) internal_function;
141 static re_dfastate_t *transit_state_sb (reg_errcode_t *err,
142 re_match_context_t *mctx,
143 re_dfastate_t *pstate) internal_function;
145 #ifdef RE_ENABLE_I18N
146 static reg_errcode_t transit_state_mb (re_match_context_t *mctx,
147 re_dfastate_t *pstate) internal_function;
148 #endif /* RE_ENABLE_I18N */
149 static reg_errcode_t transit_state_bkref (re_match_context_t *mctx,
150 const re_node_set *nodes) internal_function;
151 static reg_errcode_t get_subexp (re_match_context_t *mctx,
152 Idx bkref_node, Idx bkref_str_idx) internal_function;
153 static reg_errcode_t get_subexp_sub (re_match_context_t *mctx,
154 const re_sub_match_top_t *sub_top,
155 re_sub_match_last_t *sub_last,
156 Idx bkref_node, Idx bkref_str) internal_function;
157 static Idx find_subexp_node (const re_dfa_t *dfa, const re_node_set *nodes,
158 Idx subexp_idx, int type) internal_function;
159 static reg_errcode_t check_arrival (re_match_context_t *mctx,
160 state_array_t *path, Idx top_node,
161 Idx top_str, Idx last_node, Idx last_str,
162 int type) internal_function;
163 static reg_errcode_t check_arrival_add_next_nodes (re_match_context_t *mctx,
165 re_node_set *cur_nodes,
166 re_node_set *next_nodes) internal_function;
167 static reg_errcode_t check_arrival_expand_ecl (re_dfa_t *dfa,
168 re_node_set *cur_nodes,
169 Idx ex_subexp, int type) internal_function;
170 static reg_errcode_t check_arrival_expand_ecl_sub (re_dfa_t *dfa,
171 re_node_set *dst_nodes,
172 Idx target, Idx ex_subexp,
173 int type) internal_function;
174 static reg_errcode_t expand_bkref_cache (re_match_context_t *mctx,
175 re_node_set *cur_nodes, Idx cur_str,
176 Idx subexp_num, int type) internal_function;
177 static bool build_trtable (re_dfa_t *dfa,
178 re_dfastate_t *state) internal_function;
179 #ifdef RE_ENABLE_I18N
180 static int check_node_accept_bytes (re_dfa_t *dfa, Idx node_idx,
181 const re_string_t *input, Idx idx) internal_function;
183 static unsigned int find_collation_sequence_value (const unsigned char *mbs,
184 size_t name_len) internal_function;
186 #endif /* RE_ENABLE_I18N */
187 static Idx group_nodes_into_DFAstates (const re_dfa_t *dfa,
188 const re_dfastate_t *state,
189 re_node_set *states_node,
190 bitset *states_ch) internal_function;
191 static bool check_node_accept (const re_match_context_t *mctx,
192 const re_token_t *node, Idx idx)
194 static reg_errcode_t extend_buffers (re_match_context_t *mctx) internal_function;
196 /* Entry point for POSIX code. */
198 /* regexec searches for a given pattern, specified by PREG, in the
201 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
202 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
203 least NMATCH elements, and we set them to the offsets of the
204 corresponding matched substrings.
206 EFLAGS specifies `execution flags' which affect matching: if
207 REG_NOTBOL is set, then ^ does not match at the beginning of the
208 string; if REG_NOTEOL is set, then $ does not match at the end.
210 We return 0 if we find a match and REG_NOMATCH if not. */
213 regexec (const regex_t *__restrict preg, const char *__restrict string,
214 size_t nmatch, regmatch_t pmatch[], int eflags)
219 re_dfa_t *dfa = (re_dfa_t *) preg->re_buffer;
222 if (eflags & ~(REG_NOTBOL | REG_NOTEOL | REG_STARTEND))
225 if (eflags & REG_STARTEND)
227 start = pmatch[0].rm_so;
228 length = pmatch[0].rm_eo;
233 length = strlen (string);
236 __libc_lock_lock (dfa->lock);
238 err = re_search_internal (preg, string, length, start, length,
239 length, 0, NULL, eflags);
241 err = re_search_internal (preg, string, length, start, length,
242 length, nmatch, pmatch, eflags);
243 __libc_lock_unlock (dfa->lock);
244 return err != REG_NOERROR;
248 # include <shlib-compat.h>
249 versioned_symbol (libc, __regexec, regexec, GLIBC_2_3_4);
251 # if SHLIB_COMPAT (libc, GLIBC_2_0, GLIBC_2_3_4)
252 __typeof__ (__regexec) __compat_regexec;
255 attribute_compat_text_section
256 __compat_regexec (const regex_t *__restrict preg,
257 const char *__restrict string, size_t nmatch,
258 regmatch_t pmatch[], int eflags)
260 return regexec (preg, string, nmatch, pmatch,
261 eflags & (REG_NOTBOL | REG_NOTEOL));
263 compat_symbol (libc, __compat_regexec, regexec, GLIBC_2_0);
267 /* Entry points for GNU code. */
269 /* re_match, re_search, re_match_2, re_search_2
271 The former two functions operate on STRING with length LENGTH,
272 while the later two operate on concatenation of STRING1 and STRING2
273 with lengths LENGTH1 and LENGTH2, respectively.
275 re_match() matches the compiled pattern in BUFP against the string,
276 starting at index START.
278 re_search() first tries matching at index START, then it tries to match
279 starting from index START + 1, and so on. The last start position tried
280 is START + RANGE. (Thus RANGE = 0 forces re_search to operate the same
283 The parameter STOP of re_{match,search}_2 specifies that no match exceeding
284 the first STOP characters of the concatenation of the strings should be
287 If REGS is not NULL, and BUFP->re_no_sub is not set, the offsets of the match
288 and all groups is stroed in REGS. (For the "_2" variants, the offsets are
289 computed relative to the concatenation, not relative to the individual
292 On success, re_match* functions return the length of the match, re_search*
293 return the position of the start of the match. Return value -1 means no
294 match was found and -2 indicates an internal error. */
297 re_match (struct re_pattern_buffer *bufp, const char *string,
298 Idx length, Idx start, struct re_registers *regs)
300 return re_search_stub (bufp, string, length, start, 0, length, regs, true);
303 weak_alias (__re_match, re_match)
307 re_search (struct re_pattern_buffer *bufp, const char *string,
308 Idx length, Idx start, regoff_t range, struct re_registers *regs)
310 return re_search_stub (bufp, string, length, start, range, length, regs,
314 weak_alias (__re_search, re_search)
318 re_match_2 (struct re_pattern_buffer *bufp,
319 const char *string1, Idx length1,
320 const char *string2, Idx length2,
321 Idx start, struct re_registers *regs, Idx stop)
323 return re_search_2_stub (bufp, string1, length1, string2, length2,
324 start, 0, regs, stop, true);
327 weak_alias (__re_match_2, re_match_2)
331 re_search_2 (struct re_pattern_buffer *bufp,
332 const char *string1, Idx length1,
333 const char *string2, Idx length2,
334 Idx start, regoff_t range, struct re_registers *regs, Idx stop)
336 return re_search_2_stub (bufp, string1, length1, string2, length2,
337 start, range, regs, stop, false);
340 weak_alias (__re_search_2, re_search_2)
345 re_search_2_stub (struct re_pattern_buffer *bufp,
346 const char *string1, Idx length1,
347 const char *string2, Idx length2,
348 Idx start, regoff_t range, struct re_registers *regs,
349 Idx stop, bool ret_len)
353 Idx len = length1 + length2;
356 if (BE (length1 < 0 || length2 < 0 || stop < 0 || len < length1, 0))
359 /* Concatenate the strings. */
363 s = re_malloc (char, len);
365 if (BE (s == NULL, 0))
367 memcpy (s, string1, length1);
368 memcpy (s + length1, string2, length2);
376 rval = re_search_stub (bufp, str, len, start, range, stop, regs,
382 /* The parameters have the same meaning as those of re_search.
383 Additional parameters:
384 If RET_LEN is true the length of the match is returned (re_match style);
385 otherwise the position of the match is returned. */
389 re_search_stub (struct re_pattern_buffer *bufp,
390 const char *string, Idx length,
391 Idx start, regoff_t range, Idx stop, struct re_registers *regs,
394 reg_errcode_t result;
400 re_dfa_t *dfa = (re_dfa_t *) bufp->re_buffer;
402 Idx last_start = start + range;
404 /* Check for out-of-range. */
405 if (BE (start < 0 || start > length, 0))
407 if (sizeof start < sizeof range)
409 regoff_t length_offset = length;
410 regoff_t start_offset = start;
411 if (BE (length_offset - start_offset < range, 0))
413 else if (BE (range < - start_offset, 0))
418 if (BE ((last_start < start) != (range < 0), 0))
420 /* Overflow occurred when computing last_start; substitute
421 the extreme value. */
422 last_start = range < 0 ? 0 : length;
426 if (BE (length < last_start, 0))
428 else if (BE (last_start < 0, 0))
433 __libc_lock_lock (dfa->lock);
435 eflags |= (bufp->re_not_bol) ? REG_NOTBOL : 0;
436 eflags |= (bufp->re_not_eol) ? REG_NOTEOL : 0;
438 /* Compile fastmap if we haven't yet. */
439 if (start < last_start && bufp->re_fastmap != NULL
440 && !bufp->re_fastmap_accurate)
441 re_compile_fastmap (bufp);
443 if (BE (bufp->re_no_sub, 0))
446 /* We need at least 1 register. */
449 else if (BE (bufp->re_regs_allocated == REG_FIXED
450 && regs->rm_num_regs <= bufp->re_nsub, 0))
452 nregs = regs->rm_num_regs;
453 if (BE (nregs < 1, 0))
455 /* Nothing can be copied to regs. */
461 nregs = bufp->re_nsub + 1;
462 pmatch = re_xmalloc (regmatch_t, nregs);
463 if (BE (pmatch == NULL, 0))
469 result = re_search_internal (bufp, string, length, start, last_start, stop,
470 nregs, pmatch, eflags);
474 /* I hope we needn't fill ther regs with -1's when no match was found. */
475 if (result != REG_NOERROR)
477 else if (regs != NULL)
479 /* If caller wants register contents data back, copy them. */
480 bufp->re_regs_allocated = re_copy_regs (regs, pmatch, nregs,
481 bufp->re_regs_allocated);
482 if (BE (bufp->re_regs_allocated == REG_UNALLOCATED, 0))
486 if (BE (rval == 0, 1))
490 assert (pmatch[0].rm_so == start);
491 rval = pmatch[0].rm_eo - start;
494 rval = pmatch[0].rm_so;
498 __libc_lock_unlock (dfa->lock);
504 re_copy_regs (struct re_registers *regs, regmatch_t *pmatch, Idx nregs,
507 int rval = REG_REALLOCATE;
509 Idx need_regs = nregs + 1;
510 /* We need one extra element beyond `rm_num_regs' for the `-1' marker GNU code
513 /* Have the register data arrays been allocated? */
514 if (regs_allocated == REG_UNALLOCATED)
515 { /* No. So allocate them with malloc. */
516 regs->rm_start = re_xmalloc (regoff_t, need_regs);
517 regs->rm_end = re_malloc (regoff_t, need_regs);
518 if (BE (regs->rm_start == NULL, 0) || BE (regs->rm_end == NULL, 0))
519 return REG_UNALLOCATED;
520 regs->rm_num_regs = need_regs;
522 else if (regs_allocated == REG_REALLOCATE)
523 { /* Yes. If we need more elements than were already
524 allocated, reallocate them. If we need fewer, just
526 if (BE (need_regs > regs->rm_num_regs, 0))
528 regoff_t *new_start =
529 re_xrealloc (regs->rm_start, regoff_t, need_regs);
530 regoff_t *new_end = re_realloc (regs->rm_end, regoff_t, need_regs);
531 if (BE (new_start == NULL, 0) || BE (new_end == NULL, 0))
532 return REG_UNALLOCATED;
533 regs->rm_start = new_start;
534 regs->rm_end = new_end;
535 regs->rm_num_regs = need_regs;
540 assert (regs_allocated == REG_FIXED);
541 /* This function may not be called with REG_FIXED and nregs too big. */
542 assert (regs->rm_num_regs >= nregs);
547 for (i = 0; i < nregs; ++i)
549 regs->rm_start[i] = pmatch[i].rm_so;
550 regs->rm_end[i] = pmatch[i].rm_eo;
552 for ( ; i < regs->rm_num_regs; ++i)
553 regs->rm_start[i] = regs->rm_end[i] = -1;
558 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
559 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
560 this memory for recording register information. STARTS and ENDS
561 must be allocated using the malloc library routine, and must each
562 be at least NUM_REGS * sizeof (regoff_t) bytes long.
564 If NUM_REGS == 0, then subsequent matches should allocate their own
567 Unless this function is called, the first search or match using
568 PATTERN_BUFFER will allocate its own register data, without
569 freeing the old data. */
572 re_set_registers (struct re_pattern_buffer *bufp, struct re_registers *regs,
573 __re_size_t num_regs, regoff_t *starts, regoff_t *ends)
577 bufp->re_regs_allocated = REG_REALLOCATE;
578 regs->rm_num_regs = num_regs;
579 regs->rm_start = starts;
584 bufp->re_regs_allocated = REG_UNALLOCATED;
585 regs->rm_num_regs = 0;
586 regs->rm_start = regs->rm_end = NULL;
590 weak_alias (__re_set_registers, re_set_registers)
593 /* Entry points compatible with 4.2 BSD regex library. We don't define
594 them unless specifically requested. */
596 #if defined _REGEX_RE_COMP || defined _LIBC
601 re_exec (const char *s)
603 return 0 == regexec (&re_comp_buf, s, 0, NULL, 0);
605 #endif /* _REGEX_RE_COMP */
607 /* Internal entry point. */
609 /* Searches for a compiled pattern PREG in the string STRING, whose
610 length is LENGTH. NMATCH, PMATCH, and EFLAGS have the same
611 meaning as with regexec. LAST_START is START + RANGE, where
612 START and RANGE have the same meaning as with re_search.
613 Return REG_NOERROR if we find a match, and REG_NOMATCH if not,
614 otherwise return the error code.
615 Note: We assume front end functions already check ranges.
616 (0 <= LAST_START && LAST_START <= LENGTH) */
620 re_search_internal (const regex_t *preg,
621 const char *string, Idx length,
622 Idx start, Idx last_start, Idx stop,
623 size_t nmatch, regmatch_t pmatch[],
627 re_dfa_t *dfa = (re_dfa_t *) preg->re_buffer;
628 Idx left_lim, right_lim;
630 bool fl_longest_match;
632 Idx match_first, match_last = REG_MISSING;
636 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
637 re_match_context_t mctx = { .dfa = dfa };
639 re_match_context_t mctx;
641 char *fastmap = ((preg->re_fastmap != NULL && preg->re_fastmap_accurate
642 && start != last_start && !preg->re_can_be_null)
643 ? preg->re_fastmap : NULL);
644 unsigned REG_TRANSLATE_TYPE t =
645 (unsigned REG_TRANSLATE_TYPE) preg->re_translate;
647 #if !(defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L))
648 memset (&mctx, '\0', sizeof (re_match_context_t));
652 extra_nmatch = (nmatch > preg->re_nsub) ? nmatch - (preg->re_nsub + 1) : 0;
653 nmatch -= extra_nmatch;
655 /* Check if the DFA haven't been compiled. */
656 if (BE (preg->re_used == 0 || dfa->init_state == NULL
657 || dfa->init_state_word == NULL || dfa->init_state_nl == NULL
658 || dfa->init_state_begbuf == NULL, 0))
662 /* We assume front-end functions already check them. */
663 assert (0 <= last_start && last_start <= length);
666 /* If initial states with non-begbuf contexts have no elements,
667 the regex must be anchored. If preg->re_newline_anchor is set,
668 we'll never use init_state_nl, so do not check it. */
669 if (dfa->init_state->nodes.nelem == 0
670 && dfa->init_state_word->nodes.nelem == 0
671 && (dfa->init_state_nl->nodes.nelem == 0
672 || !preg->re_newline_anchor))
674 if (start != 0 && last_start != 0)
676 start = last_start = 0;
679 /* We must check the longest matching, if nmatch > 0. */
680 fl_longest_match = (nmatch != 0 || dfa->nbackref);
682 err = re_string_allocate (&mctx.input, string, length, dfa->nodes_len + 1,
684 preg->re_syntax & REG_IGNORE_CASE, dfa);
685 if (BE (err != REG_NOERROR, 0))
687 mctx.input.stop = stop;
688 mctx.input.raw_stop = stop;
689 mctx.input.newline_anchor = preg->re_newline_anchor;
691 err = match_ctx_init (&mctx, eflags, dfa->nbackref * 2);
692 if (BE (err != REG_NOERROR, 0))
695 /* We will log all the DFA states through which the dfa pass,
696 if nmatch > 1, or this dfa has "multibyte node", which is a
697 back-reference or a node which can accept multibyte character or
698 multi character collating element. */
699 if (nmatch > 1 || dfa->has_mb_node)
701 mctx.state_log = re_xmalloc (re_dfastate_t *, mctx.input.bufs_len + 1);
702 if (BE (mctx.state_log == NULL, 0))
709 mctx.state_log = NULL;
712 mctx.input.tip_context = (eflags & REG_NOTBOL) ? CONTEXT_BEGBUF
713 : CONTEXT_NEWLINE | CONTEXT_BEGBUF;
715 /* Check incrementally whether of not the input string match. */
716 incr = (last_start < start) ? -1 : 1;
717 left_lim = (last_start < start) ? last_start : start;
718 right_lim = (last_start < start) ? start : last_start;
719 sb = dfa->mb_cur_max == 1;
722 ? ((sb || !(preg->re_syntax & REG_IGNORE_CASE || t) ? 4 : 0)
723 | (start <= last_start ? 2 : 0)
724 | (t != NULL ? 1 : 0))
727 for (;; match_first += incr)
730 if (match_first < left_lim || right_lim < match_first)
733 /* Advance as rapidly as possible through the string, until we
734 find a plausible place to start matching. This may be done
735 with varying efficiency, so there are various possibilities:
736 only the most common of them are specialized, in order to
737 save on code size. We use a switch statement for speed. */
745 /* Fastmap with single-byte translation, match forward. */
746 while (BE (match_first < right_lim, 1)
747 && !fastmap[t[(unsigned char) string[match_first]]])
749 goto forward_match_found_start_or_reached_end;
752 /* Fastmap without translation, match forward. */
753 while (BE (match_first < right_lim, 1)
754 && !fastmap[(unsigned char) string[match_first]])
757 forward_match_found_start_or_reached_end:
758 if (BE (match_first == right_lim, 0))
760 ch = match_first >= length
761 ? 0 : (unsigned char) string[match_first];
762 if (!fastmap[t ? t[ch] : ch])
769 /* Fastmap without multi-byte translation, match backwards. */
770 while (match_first >= left_lim)
772 ch = match_first >= length
773 ? 0 : (unsigned char) string[match_first];
774 if (fastmap[t ? t[ch] : ch])
778 if (match_first < left_lim)
783 /* In this case, we can't determine easily the current byte,
784 since it might be a component byte of a multibyte
785 character. Then we use the constructed buffer instead. */
788 /* If MATCH_FIRST is out of the valid range, reconstruct the
790 __re_size_t offset = match_first - mctx.input.raw_mbs_idx;
791 if (BE (offset >= (__re_size_t) mctx.input.valid_raw_len, 0))
793 err = re_string_reconstruct (&mctx.input, match_first,
795 if (BE (err != REG_NOERROR, 0))
798 offset = match_first - mctx.input.raw_mbs_idx;
800 /* If MATCH_FIRST is out of the buffer, leave it as '\0'.
801 Note that MATCH_FIRST must not be smaller than 0. */
802 ch = (match_first >= length
803 ? 0 : re_string_byte_at (&mctx.input, offset));
807 if (match_first < left_lim || match_first > right_lim)
816 /* Reconstruct the buffers so that the matcher can assume that
817 the matching starts from the beginning of the buffer. */
818 err = re_string_reconstruct (&mctx.input, match_first, eflags);
819 if (BE (err != REG_NOERROR, 0))
822 #ifdef RE_ENABLE_I18N
823 /* Don't consider this char as a possible match start if it part,
824 yet isn't the head, of a multibyte character. */
825 if (!sb && !re_string_first_byte (&mctx.input, 0))
829 /* It seems to be appropriate one, then use the matcher. */
830 /* We assume that the matching starts from 0. */
831 mctx.state_log_top = mctx.nbkref_ents = mctx.max_mb_elem_len = 0;
832 match_last = check_matching (&mctx, fl_longest_match,
833 start <= last_start ? &match_first : NULL);
834 if (match_last != REG_MISSING)
836 if (BE (match_last == REG_ERROR, 0))
843 mctx.match_last = match_last;
844 if ((!preg->re_no_sub && nmatch > 1) || dfa->nbackref)
846 re_dfastate_t *pstate = mctx.state_log[match_last];
847 mctx.last_node = check_halt_state_context (&mctx, pstate,
850 if ((!preg->re_no_sub && nmatch > 1 && dfa->has_plural_match)
853 err = prune_impossible_nodes (&mctx);
854 if (err == REG_NOERROR)
856 if (BE (err != REG_NOMATCH, 0))
858 match_last = REG_MISSING;
861 break; /* We found a match. */
865 match_ctx_clean (&mctx);
869 assert (match_last != REG_MISSING);
870 assert (err == REG_NOERROR);
873 /* Set pmatch[] if we need. */
878 /* Initialize registers. */
879 for (reg_idx = 1; reg_idx < nmatch; ++reg_idx)
880 pmatch[reg_idx].rm_so = pmatch[reg_idx].rm_eo = -1;
882 /* Set the points where matching start/end. */
884 pmatch[0].rm_eo = mctx.match_last;
885 /* FIXME: This function should fail if mctx.match_last exceeds
886 the maximum possible regoff_t value. We need a new error
887 code REG_OVERFLOW. */
889 if (!preg->re_no_sub && nmatch > 1)
891 err = set_regs (preg, &mctx, nmatch, pmatch,
892 dfa->has_plural_match && dfa->nbackref > 0);
893 if (BE (err != REG_NOERROR, 0))
897 /* At last, add the offset to the each registers, since we slided
898 the buffers so that we could assume that the matching starts
900 for (reg_idx = 0; reg_idx < nmatch; ++reg_idx)
901 if (pmatch[reg_idx].rm_so != -1)
903 #ifdef RE_ENABLE_I18N
904 if (BE (mctx.input.offsets_needed != 0, 0))
906 pmatch[reg_idx].rm_so =
907 (pmatch[reg_idx].rm_so == mctx.input.valid_len
908 ? mctx.input.valid_raw_len
909 : mctx.input.offsets[pmatch[reg_idx].rm_so]);
910 pmatch[reg_idx].rm_eo =
911 (pmatch[reg_idx].rm_eo == mctx.input.valid_len
912 ? mctx.input.valid_raw_len
913 : mctx.input.offsets[pmatch[reg_idx].rm_eo]);
916 assert (mctx.input.offsets_needed == 0);
918 pmatch[reg_idx].rm_so += match_first;
919 pmatch[reg_idx].rm_eo += match_first;
921 for (reg_idx = 0; reg_idx < extra_nmatch; ++reg_idx)
923 pmatch[nmatch + reg_idx].rm_so = -1;
924 pmatch[nmatch + reg_idx].rm_eo = -1;
928 for (reg_idx = 0; reg_idx + 1 < nmatch; reg_idx++)
929 if (dfa->subexp_map[reg_idx] != reg_idx)
931 pmatch[reg_idx + 1].rm_so
932 = pmatch[dfa->subexp_map[reg_idx] + 1].rm_so;
933 pmatch[reg_idx + 1].rm_eo
934 = pmatch[dfa->subexp_map[reg_idx] + 1].rm_eo;
939 re_free (mctx.state_log);
941 match_ctx_free (&mctx);
942 re_string_destruct (&mctx.input);
948 prune_impossible_nodes (re_match_context_t *mctx)
950 re_dfa_t *const dfa = mctx->dfa;
951 Idx halt_node, match_last;
953 re_dfastate_t **sifted_states;
954 re_dfastate_t **lim_states = NULL;
955 re_sift_context_t sctx;
957 assert (mctx->state_log != NULL);
959 match_last = mctx->match_last;
960 halt_node = mctx->last_node;
961 sifted_states = re_xmalloc (re_dfastate_t *, match_last + 1);
962 if (BE (sifted_states == NULL, 0))
969 lim_states = re_xmalloc (re_dfastate_t *, match_last + 1);
970 if (BE (lim_states == NULL, 0))
977 memset (lim_states, '\0',
978 sizeof (re_dfastate_t *) * (match_last + 1));
979 sift_ctx_init (&sctx, sifted_states, lim_states, halt_node,
981 ret = sift_states_backward (mctx, &sctx);
982 re_node_set_free (&sctx.limits);
983 if (BE (ret != REG_NOERROR, 0))
985 if (sifted_states[0] != NULL || lim_states[0] != NULL)
990 if (! REG_VALID_INDEX (match_last))
995 } while (mctx->state_log[match_last] == NULL
996 || !mctx->state_log[match_last]->halt);
997 halt_node = check_halt_state_context (mctx,
998 mctx->state_log[match_last],
1001 ret = merge_state_array (dfa, sifted_states, lim_states,
1003 re_free (lim_states);
1005 if (BE (ret != REG_NOERROR, 0))
1010 sift_ctx_init (&sctx, sifted_states, lim_states, halt_node, match_last);
1011 ret = sift_states_backward (mctx, &sctx);
1012 re_node_set_free (&sctx.limits);
1013 if (BE (ret != REG_NOERROR, 0))
1016 re_free (mctx->state_log);
1017 mctx->state_log = sifted_states;
1018 sifted_states = NULL;
1019 mctx->last_node = halt_node;
1020 mctx->match_last = match_last;
1023 re_free (sifted_states);
1024 re_free (lim_states);
1028 /* Acquire an initial state and return it.
1029 We must select appropriate initial state depending on the context,
1030 since initial states may have constraints like "\<", "^", etc.. */
1032 static inline re_dfastate_t *
1033 __attribute ((always_inline)) internal_function
1034 acquire_init_state_context (reg_errcode_t *err, const re_match_context_t *mctx,
1037 re_dfa_t *const dfa = mctx->dfa;
1038 if (dfa->init_state->has_constraint)
1040 unsigned int context;
1041 context = re_string_context_at (&mctx->input, idx - 1, mctx->eflags);
1042 if (IS_WORD_CONTEXT (context))
1043 return dfa->init_state_word;
1044 else if (IS_ORDINARY_CONTEXT (context))
1045 return dfa->init_state;
1046 else if (IS_BEGBUF_CONTEXT (context) && IS_NEWLINE_CONTEXT (context))
1047 return dfa->init_state_begbuf;
1048 else if (IS_NEWLINE_CONTEXT (context))
1049 return dfa->init_state_nl;
1050 else if (IS_BEGBUF_CONTEXT (context))
1052 /* It is relatively rare case, then calculate on demand. */
1053 return re_acquire_state_context (err, dfa,
1054 dfa->init_state->entrance_nodes,
1058 /* Must not happen? */
1059 return dfa->init_state;
1062 return dfa->init_state;
1065 /* Check whether the regular expression match input string INPUT or not,
1066 and return the index where the matching end. Return REG_MISSING if
1067 there is no match, and return REG_ERROR in case of an error.
1068 FL_LONGEST_MATCH means we want the POSIX longest matching.
1069 If P_MATCH_FIRST is not NULL, and the match fails, it is set to the
1070 next place where we may want to try matching.
1071 Note that the matcher assume that the maching starts from the current
1072 index of the buffer. */
1076 check_matching (re_match_context_t *mctx, bool fl_longest_match,
1079 re_dfa_t *const dfa = mctx->dfa;
1082 Idx match_last = REG_MISSING;
1083 Idx cur_str_idx = re_string_cur_idx (&mctx->input);
1084 re_dfastate_t *cur_state;
1085 bool at_init_state = p_match_first != NULL;
1086 Idx next_start_idx = cur_str_idx;
1089 cur_state = acquire_init_state_context (&err, mctx, cur_str_idx);
1090 /* An initial state must not be NULL (invalid). */
1091 if (BE (cur_state == NULL, 0))
1093 assert (err == REG_ESPACE);
1097 if (mctx->state_log != NULL)
1099 mctx->state_log[cur_str_idx] = cur_state;
1101 /* Check OP_OPEN_SUBEXP in the initial state in case that we use them
1102 later. E.g. Processing back references. */
1103 if (BE (dfa->nbackref, 0))
1105 at_init_state = false;
1106 err = check_subexp_matching_top (mctx, &cur_state->nodes, 0);
1107 if (BE (err != REG_NOERROR, 0))
1110 if (cur_state->has_backref)
1112 err = transit_state_bkref (mctx, &cur_state->nodes);
1113 if (BE (err != REG_NOERROR, 0))
1119 /* If the RE accepts NULL string. */
1120 if (BE (cur_state->halt, 0))
1122 if (!cur_state->has_constraint
1123 || check_halt_state_context (mctx, cur_state, cur_str_idx))
1125 if (!fl_longest_match)
1129 match_last = cur_str_idx;
1135 while (!re_string_eoi (&mctx->input))
1137 re_dfastate_t *old_state = cur_state;
1138 Idx next_char_idx = re_string_cur_idx (&mctx->input) + 1;
1140 if (BE (next_char_idx >= mctx->input.bufs_len, 0)
1141 || (BE (next_char_idx >= mctx->input.valid_len, 0)
1142 && mctx->input.valid_len < mctx->input.len))
1144 err = extend_buffers (mctx);
1145 if (BE (err != REG_NOERROR, 0))
1147 assert (err == REG_ESPACE);
1152 cur_state = transit_state (&err, mctx, cur_state);
1153 if (mctx->state_log != NULL)
1154 cur_state = merge_state_with_log (&err, mctx, cur_state);
1156 if (cur_state == NULL)
1158 /* Reached the invalid state or an error. Try to recover a valid
1159 state using the state log, if available and if we have not
1160 already found a valid (even if not the longest) match. */
1161 if (BE (err != REG_NOERROR, 0))
1164 if (mctx->state_log == NULL
1165 || (match && !fl_longest_match)
1166 || (cur_state = find_recover_state (&err, mctx)) == NULL)
1170 if (BE (at_init_state, 0))
1172 if (old_state == cur_state)
1173 next_start_idx = next_char_idx;
1175 at_init_state = false;
1178 if (cur_state->halt)
1180 /* Reached a halt state.
1181 Check the halt state can satisfy the current context. */
1182 if (!cur_state->has_constraint
1183 || check_halt_state_context (mctx, cur_state,
1184 re_string_cur_idx (&mctx->input)))
1186 /* We found an appropriate halt state. */
1187 match_last = re_string_cur_idx (&mctx->input);
1190 /* We found a match, do not modify match_first below. */
1191 p_match_first = NULL;
1192 if (!fl_longest_match)
1199 *p_match_first += next_start_idx;
1204 /* Check NODE match the current context. */
1208 check_halt_node_context (const re_dfa_t *dfa, Idx node, unsigned int context)
1210 re_token_type_t type = dfa->nodes[node].type;
1211 unsigned int constraint = dfa->nodes[node].constraint;
1212 if (type != END_OF_RE)
1216 if (NOT_SATISFY_NEXT_CONSTRAINT (constraint, context))
1221 /* Check the halt state STATE match the current context.
1222 Return 0 if not match, if the node, STATE has, is a halt node and
1223 match the context, return the node. */
1227 check_halt_state_context (const re_match_context_t *mctx,
1228 const re_dfastate_t *state, Idx idx)
1231 unsigned int context;
1233 assert (state->halt);
1235 context = re_string_context_at (&mctx->input, idx, mctx->eflags);
1236 for (i = 0; i < state->nodes.nelem; ++i)
1237 if (check_halt_node_context (mctx->dfa, state->nodes.elems[i], context))
1238 return state->nodes.elems[i];
1242 /* Compute the next node to which "NFA" transit from NODE("NFA" is a NFA
1243 corresponding to the DFA).
1244 Return the destination node, and update EPS_VIA_NODES;
1245 return REG_MISSING in case of errors. */
1249 proceed_next_node (const re_match_context_t *mctx,
1250 Idx nregs, regmatch_t *regs, Idx *pidx, Idx node,
1251 re_node_set *eps_via_nodes, struct re_fail_stack_t *fs)
1253 re_dfa_t *const dfa = mctx->dfa;
1256 if (IS_EPSILON_NODE (dfa->nodes[node].type))
1258 re_node_set *cur_nodes = &mctx->state_log[*pidx]->nodes;
1259 re_node_set *edests = &dfa->edests[node];
1261 ok = re_node_set_insert (eps_via_nodes, node);
1264 /* Pick up a valid destination, or return REG_MISSING if none
1266 for (dest_node = REG_MISSING, i = 0; i < edests->nelem; ++i)
1268 Idx candidate = edests->elems[i];
1269 if (!re_node_set_contains (cur_nodes, candidate))
1271 if (dest_node == REG_MISSING)
1272 dest_node = candidate;
1276 /* In order to avoid infinite loop like "(a*)*", return the second
1277 epsilon-transition if the first was already considered. */
1278 if (re_node_set_contains (eps_via_nodes, dest_node))
1281 /* Otherwise, push the second epsilon-transition on the fail stack. */
1283 && push_fail_stack (fs, *pidx, candidate, nregs, regs,
1287 /* We know we are going to exit. */
1296 re_token_type_t type = dfa->nodes[node].type;
1298 #ifdef RE_ENABLE_I18N
1299 if (dfa->nodes[node].accept_mb)
1300 naccepted = check_node_accept_bytes (dfa, node, &mctx->input, *pidx);
1302 #endif /* RE_ENABLE_I18N */
1303 if (type == OP_BACK_REF)
1305 Idx subexp_idx = dfa->nodes[node].opr.idx + 1;
1306 naccepted = regs[subexp_idx].rm_eo - regs[subexp_idx].rm_so;
1309 if (regs[subexp_idx].rm_so == -1 || regs[subexp_idx].rm_eo == -1)
1313 char *buf = (char *) re_string_get_buffer (&mctx->input);
1314 if (memcmp (buf + regs[subexp_idx].rm_so, buf + *pidx,
1323 ok = re_node_set_insert (eps_via_nodes, node);
1326 dest_node = dfa->edests[node].elems[0];
1327 if (re_node_set_contains (&mctx->state_log[*pidx]->nodes,
1334 || check_node_accept (mctx, dfa->nodes + node, *pidx))
1336 Idx dest_node = dfa->nexts[node];
1337 *pidx = (naccepted == 0) ? *pidx + 1 : *pidx + naccepted;
1338 if (fs && (*pidx > mctx->match_last || mctx->state_log[*pidx] == NULL
1339 || !re_node_set_contains (&mctx->state_log[*pidx]->nodes,
1342 re_node_set_empty (eps_via_nodes);
1349 static reg_errcode_t
1351 push_fail_stack (struct re_fail_stack_t *fs, Idx str_idx, Idx dest_node,
1352 Idx nregs, regmatch_t *regs, re_node_set *eps_via_nodes)
1355 Idx num = fs->num++;
1356 if (fs->num == fs->alloc)
1358 struct re_fail_stack_ent_t *new_array =
1359 re_x2realloc (fs->stack, struct re_fail_stack_ent_t, &fs->alloc);
1360 if (new_array == NULL)
1362 fs->stack = new_array;
1364 fs->stack[num].idx = str_idx;
1365 fs->stack[num].node = dest_node;
1366 fs->stack[num].regs = re_xmalloc (regmatch_t, nregs);
1367 if (fs->stack[num].regs == NULL)
1369 memcpy (fs->stack[num].regs, regs, sizeof (regmatch_t) * nregs);
1370 err = re_node_set_init_copy (&fs->stack[num].eps_via_nodes, eps_via_nodes);
1376 pop_fail_stack (struct re_fail_stack_t *fs, Idx *pidx,
1377 Idx nregs, regmatch_t *regs, re_node_set *eps_via_nodes)
1379 Idx num = --fs->num;
1380 assert (REG_VALID_INDEX (num));
1381 *pidx = fs->stack[num].idx;
1382 memcpy (regs, fs->stack[num].regs, sizeof (regmatch_t) * nregs);
1383 re_node_set_free (eps_via_nodes);
1384 re_free (fs->stack[num].regs);
1385 *eps_via_nodes = fs->stack[num].eps_via_nodes;
1386 return fs->stack[num].node;
1389 /* Set the positions where the subexpressions are starts/ends to registers
1391 Note: We assume that pmatch[0] is already set, and
1392 pmatch[i].rm_so == pmatch[i].rm_eo == -1 for 0 < i < nmatch. */
1394 static reg_errcode_t
1396 set_regs (const regex_t *preg, const re_match_context_t *mctx,
1397 size_t nmatch, regmatch_t *pmatch, bool fl_backtrack)
1399 re_dfa_t *dfa = (re_dfa_t *) preg->re_buffer;
1401 re_node_set eps_via_nodes;
1402 struct re_fail_stack_t *fs;
1403 struct re_fail_stack_t fs_body = { 0, 2, NULL };
1404 regmatch_t *prev_idx_match;
1405 bool prev_idx_match_malloced = false;
1408 assert (nmatch > 1);
1409 assert (mctx->state_log != NULL);
1414 fs->stack = re_xmalloc (struct re_fail_stack_ent_t, fs->alloc);
1415 if (fs->stack == NULL)
1421 cur_node = dfa->init_node;
1422 re_node_set_init_empty (&eps_via_nodes);
1424 if (re_alloc_oversized (nmatch, sizeof (regmatch_t)))
1426 free_fail_stack_return (fs);
1429 if (__libc_use_alloca (nmatch * sizeof (regmatch_t)))
1430 prev_idx_match = (regmatch_t *) alloca (nmatch * sizeof (regmatch_t));
1433 prev_idx_match = re_malloc (regmatch_t, nmatch);
1434 if (prev_idx_match == NULL)
1436 free_fail_stack_return (fs);
1439 prev_idx_match_malloced = true;
1441 memcpy (prev_idx_match, pmatch, sizeof (regmatch_t) * nmatch);
1443 for (idx = pmatch[0].rm_so; idx <= pmatch[0].rm_eo ;)
1445 update_regs (dfa, pmatch, prev_idx_match, cur_node, idx, nmatch);
1447 if (idx == pmatch[0].rm_eo && cur_node == mctx->last_node)
1452 for (reg_idx = 0; reg_idx < nmatch; ++reg_idx)
1453 if (pmatch[reg_idx].rm_so > -1 && pmatch[reg_idx].rm_eo == -1)
1455 if (reg_idx == nmatch)
1457 re_node_set_free (&eps_via_nodes);
1458 if (prev_idx_match_malloced)
1459 re_free (prev_idx_match);
1460 return free_fail_stack_return (fs);
1462 cur_node = pop_fail_stack (fs, &idx, nmatch, pmatch,
1467 re_node_set_free (&eps_via_nodes);
1468 if (prev_idx_match_malloced)
1469 re_free (prev_idx_match);
1474 /* Proceed to next node. */
1475 cur_node = proceed_next_node (mctx, nmatch, pmatch, &idx, cur_node,
1476 &eps_via_nodes, fs);
1478 if (BE (! REG_VALID_INDEX (cur_node), 0))
1480 if (BE (cur_node == REG_ERROR, 0))
1482 re_node_set_free (&eps_via_nodes);
1483 if (prev_idx_match_malloced)
1484 re_free (prev_idx_match);
1485 free_fail_stack_return (fs);
1489 cur_node = pop_fail_stack (fs, &idx, nmatch, pmatch,
1493 re_node_set_free (&eps_via_nodes);
1494 if (prev_idx_match_malloced)
1495 re_free (prev_idx_match);
1500 re_node_set_free (&eps_via_nodes);
1501 if (prev_idx_match_malloced)
1502 re_free (prev_idx_match);
1503 return free_fail_stack_return (fs);
1506 static reg_errcode_t
1508 free_fail_stack_return (struct re_fail_stack_t *fs)
1513 for (fs_idx = 0; fs_idx < fs->num; ++fs_idx)
1515 re_node_set_free (&fs->stack[fs_idx].eps_via_nodes);
1516 re_free (fs->stack[fs_idx].regs);
1518 re_free (fs->stack);
1525 update_regs (re_dfa_t *dfa, regmatch_t *pmatch, regmatch_t *prev_idx_match,
1526 Idx cur_node, Idx cur_idx, Idx nmatch)
1528 int type = dfa->nodes[cur_node].type;
1529 if (type == OP_OPEN_SUBEXP)
1531 Idx reg_num = dfa->nodes[cur_node].opr.idx + 1;
1533 /* We are at the first node of this sub expression. */
1534 if (reg_num < nmatch)
1536 pmatch[reg_num].rm_so = cur_idx;
1537 pmatch[reg_num].rm_eo = -1;
1540 else if (type == OP_CLOSE_SUBEXP)
1542 Idx reg_num = dfa->nodes[cur_node].opr.idx + 1;
1543 if (reg_num < nmatch)
1545 /* We are at the last node of this sub expression. */
1546 if (pmatch[reg_num].rm_so < cur_idx)
1548 pmatch[reg_num].rm_eo = cur_idx;
1549 /* This is a non-empty match or we are not inside an optional
1550 subexpression. Accept this right away. */
1551 memcpy (prev_idx_match, pmatch, sizeof (regmatch_t) * nmatch);
1555 if (dfa->nodes[cur_node].opt_subexp
1556 && prev_idx_match[reg_num].rm_so != -1)
1557 /* We transited through an empty match for an optional
1558 subexpression, like (a?)*, and this is not the subexp's
1559 first match. Copy back the old content of the registers
1560 so that matches of an inner subexpression are undone as
1561 well, like in ((a?))*. */
1562 memcpy (pmatch, prev_idx_match, sizeof (regmatch_t) * nmatch);
1564 /* We completed a subexpression, but it may be part of
1565 an optional one, so do not update PREV_IDX_MATCH. */
1566 pmatch[reg_num].rm_eo = cur_idx;
1572 /* This function checks the STATE_LOG from the SCTX->last_str_idx to 0
1573 and sift the nodes in each states according to the following rules.
1574 Updated state_log will be wrote to STATE_LOG.
1576 Rules: We throw away the Node `a' in the STATE_LOG[STR_IDX] if...
1577 1. When STR_IDX == MATCH_LAST(the last index in the state_log):
1578 If `a' isn't the LAST_NODE and `a' can't epsilon transit to
1579 the LAST_NODE, we throw away the node `a'.
1580 2. When 0 <= STR_IDX < MATCH_LAST and `a' accepts
1581 string `s' and transit to `b':
1582 i. If 'b' isn't in the STATE_LOG[STR_IDX+strlen('s')], we throw
1584 ii. If 'b' is in the STATE_LOG[STR_IDX+strlen('s')] but 'b' is
1585 thrown away, we throw away the node `a'.
1586 3. When 0 <= STR_IDX < MATCH_LAST and 'a' epsilon transit to 'b':
1587 i. If 'b' isn't in the STATE_LOG[STR_IDX], we throw away the
1589 ii. If 'b' is in the STATE_LOG[STR_IDX] but 'b' is thrown away,
1590 we throw away the node `a'. */
1592 #define STATE_NODE_CONTAINS(state,node) \
1593 ((state) != NULL && re_node_set_contains (&(state)->nodes, node))
1595 static reg_errcode_t
1597 sift_states_backward (re_match_context_t *mctx, re_sift_context_t *sctx)
1601 Idx str_idx = sctx->last_str_idx;
1602 re_node_set cur_dest;
1605 assert (mctx->state_log != NULL && mctx->state_log[str_idx] != NULL);
1608 /* Build sifted state_log[str_idx]. It has the nodes which can epsilon
1609 transit to the last_node and the last_node itself. */
1610 err = re_node_set_init_1 (&cur_dest, sctx->last_node);
1611 if (BE (err != REG_NOERROR, 0))
1613 err = update_cur_sifted_state (mctx, sctx, str_idx, &cur_dest);
1614 if (BE (err != REG_NOERROR, 0))
1617 /* Then check each states in the state_log. */
1620 /* Update counters. */
1621 null_cnt = (sctx->sifted_states[str_idx] == NULL) ? null_cnt + 1 : 0;
1622 if (null_cnt > mctx->max_mb_elem_len)
1624 memset (sctx->sifted_states, '\0',
1625 sizeof (re_dfastate_t *) * str_idx);
1626 re_node_set_free (&cur_dest);
1629 re_node_set_empty (&cur_dest);
1632 if (mctx->state_log[str_idx])
1634 err = build_sifted_states (mctx, sctx, str_idx, &cur_dest);
1635 if (BE (err != REG_NOERROR, 0))
1639 /* Add all the nodes which satisfy the following conditions:
1640 - It can epsilon transit to a node in CUR_DEST.
1642 And update state_log. */
1643 err = update_cur_sifted_state (mctx, sctx, str_idx, &cur_dest);
1644 if (BE (err != REG_NOERROR, 0))
1649 re_node_set_free (&cur_dest);
1653 static reg_errcode_t
1655 build_sifted_states (re_match_context_t *mctx, re_sift_context_t *sctx,
1656 Idx str_idx, re_node_set *cur_dest)
1658 re_dfa_t *const dfa = mctx->dfa;
1659 re_node_set *cur_src = &mctx->state_log[str_idx]->non_eps_nodes;
1662 /* Then build the next sifted state.
1663 We build the next sifted state on `cur_dest', and update
1664 `sifted_states[str_idx]' with `cur_dest'.
1666 `cur_dest' is the sifted state from `state_log[str_idx + 1]'.
1667 `cur_src' points the node_set of the old `state_log[str_idx]'
1668 (with the epsilon nodes pre-filtered out). */
1669 for (i = 0; i < cur_src->nelem; i++)
1671 Idx prev_node = cur_src->elems[i];
1676 re_token_type_t type = dfa->nodes[prev_node].type;
1677 assert (!IS_EPSILON_NODE (type));
1679 #ifdef RE_ENABLE_I18N
1680 /* If the node may accept `multi byte'. */
1681 if (dfa->nodes[prev_node].accept_mb)
1682 naccepted = sift_states_iter_mb (mctx, sctx, prev_node,
1683 str_idx, sctx->last_str_idx);
1684 #endif /* RE_ENABLE_I18N */
1686 /* We don't check backreferences here.
1687 See update_cur_sifted_state(). */
1689 && check_node_accept (mctx, dfa->nodes + prev_node, str_idx)
1690 && STATE_NODE_CONTAINS (sctx->sifted_states[str_idx + 1],
1691 dfa->nexts[prev_node]))
1697 if (sctx->limits.nelem)
1699 Idx to_idx = str_idx + naccepted;
1700 if (check_dst_limits (mctx, &sctx->limits,
1701 dfa->nexts[prev_node], to_idx,
1702 prev_node, str_idx))
1705 ok = re_node_set_insert (cur_dest, prev_node);
1713 /* Helper functions. */
1715 static reg_errcode_t
1717 clean_state_log_if_needed (re_match_context_t *mctx, Idx next_state_log_idx)
1719 Idx top = mctx->state_log_top;
1721 if (next_state_log_idx >= mctx->input.bufs_len
1722 || (next_state_log_idx >= mctx->input.valid_len
1723 && mctx->input.valid_len < mctx->input.len))
1726 err = extend_buffers (mctx);
1727 if (BE (err != REG_NOERROR, 0))
1731 if (top < next_state_log_idx)
1733 memset (mctx->state_log + top + 1, '\0',
1734 sizeof (re_dfastate_t *) * (next_state_log_idx - top));
1735 mctx->state_log_top = next_state_log_idx;
1740 static reg_errcode_t
1742 merge_state_array (re_dfa_t *dfa, re_dfastate_t **dst, re_dfastate_t **src,
1747 for (st_idx = 0; st_idx < num; ++st_idx)
1749 if (dst[st_idx] == NULL)
1750 dst[st_idx] = src[st_idx];
1751 else if (src[st_idx] != NULL)
1753 re_node_set merged_set;
1754 err = re_node_set_init_union (&merged_set, &dst[st_idx]->nodes,
1755 &src[st_idx]->nodes);
1756 if (BE (err != REG_NOERROR, 0))
1758 dst[st_idx] = re_acquire_state (&err, dfa, &merged_set);
1759 re_node_set_free (&merged_set);
1760 if (BE (err != REG_NOERROR, 0))
1767 static reg_errcode_t
1769 update_cur_sifted_state (re_match_context_t *mctx, re_sift_context_t *sctx,
1770 Idx str_idx, re_node_set *dest_nodes)
1772 re_dfa_t *const dfa = mctx->dfa;
1774 const re_node_set *candidates;
1775 candidates = ((mctx->state_log[str_idx] == NULL) ? NULL
1776 : &mctx->state_log[str_idx]->nodes);
1778 if (dest_nodes->nelem == 0)
1779 sctx->sifted_states[str_idx] = NULL;
1784 /* At first, add the nodes which can epsilon transit to a node in
1786 err = add_epsilon_src_nodes (dfa, dest_nodes, candidates);
1787 if (BE (err != REG_NOERROR, 0))
1790 /* Then, check the limitations in the current sift_context. */
1791 if (sctx->limits.nelem)
1793 err = check_subexp_limits (dfa, dest_nodes, candidates, &sctx->limits,
1794 mctx->bkref_ents, str_idx);
1795 if (BE (err != REG_NOERROR, 0))
1800 sctx->sifted_states[str_idx] = re_acquire_state (&err, dfa, dest_nodes);
1801 if (BE (err != REG_NOERROR, 0))
1805 if (candidates && mctx->state_log[str_idx]->has_backref)
1807 err = sift_states_bkref (mctx, sctx, str_idx, candidates);
1808 if (BE (err != REG_NOERROR, 0))
1814 static reg_errcode_t
1816 add_epsilon_src_nodes (re_dfa_t *dfa, re_node_set *dest_nodes,
1817 const re_node_set *candidates)
1819 reg_errcode_t err = REG_NOERROR;
1822 re_dfastate_t *state = re_acquire_state (&err, dfa, dest_nodes);
1823 if (BE (err != REG_NOERROR, 0))
1826 if (!state->inveclosure.alloc)
1828 err = re_node_set_alloc (&state->inveclosure, dest_nodes->nelem);
1829 if (BE (err != REG_NOERROR, 0))
1831 for (i = 0; i < dest_nodes->nelem; i++)
1832 re_node_set_merge (&state->inveclosure,
1833 dfa->inveclosures + dest_nodes->elems[i]);
1835 return re_node_set_add_intersect (dest_nodes, candidates,
1836 &state->inveclosure);
1839 static reg_errcode_t
1841 sub_epsilon_src_nodes (re_dfa_t *dfa, Idx node, re_node_set *dest_nodes,
1842 const re_node_set *candidates)
1846 re_node_set *inv_eclosure = dfa->inveclosures + node;
1847 re_node_set except_nodes;
1848 re_node_set_init_empty (&except_nodes);
1849 for (ecl_idx = 0; ecl_idx < inv_eclosure->nelem; ++ecl_idx)
1851 Idx cur_node = inv_eclosure->elems[ecl_idx];
1852 if (cur_node == node)
1854 if (IS_EPSILON_NODE (dfa->nodes[cur_node].type))
1856 Idx edst1 = dfa->edests[cur_node].elems[0];
1857 Idx edst2 = ((dfa->edests[cur_node].nelem > 1)
1858 ? dfa->edests[cur_node].elems[1] : REG_MISSING);
1859 if ((!re_node_set_contains (inv_eclosure, edst1)
1860 && re_node_set_contains (dest_nodes, edst1))
1861 || (REG_VALID_NONZERO_INDEX (edst2)
1862 && !re_node_set_contains (inv_eclosure, edst2)
1863 && re_node_set_contains (dest_nodes, edst2)))
1865 err = re_node_set_add_intersect (&except_nodes, candidates,
1866 dfa->inveclosures + cur_node);
1867 if (BE (err != REG_NOERROR, 0))
1869 re_node_set_free (&except_nodes);
1875 for (ecl_idx = 0; ecl_idx < inv_eclosure->nelem; ++ecl_idx)
1877 Idx cur_node = inv_eclosure->elems[ecl_idx];
1878 if (!re_node_set_contains (&except_nodes, cur_node))
1880 Idx idx = re_node_set_contains (dest_nodes, cur_node) - 1;
1881 re_node_set_remove_at (dest_nodes, idx);
1884 re_node_set_free (&except_nodes);
1890 check_dst_limits (const re_match_context_t *mctx, const re_node_set *limits,
1891 Idx dst_node, Idx dst_idx, Idx src_node, Idx src_idx)
1893 re_dfa_t *const dfa = mctx->dfa;
1894 Idx lim_idx, src_pos, dst_pos;
1896 Idx dst_bkref_idx = search_cur_bkref_entry (mctx, dst_idx);
1897 Idx src_bkref_idx = search_cur_bkref_entry (mctx, src_idx);
1898 for (lim_idx = 0; lim_idx < limits->nelem; ++lim_idx)
1901 struct re_backref_cache_entry *ent;
1902 ent = mctx->bkref_ents + limits->elems[lim_idx];
1903 subexp_idx = dfa->nodes[ent->node].opr.idx;
1905 dst_pos = check_dst_limits_calc_pos (mctx, limits->elems[lim_idx],
1906 subexp_idx, dst_node, dst_idx,
1908 src_pos = check_dst_limits_calc_pos (mctx, limits->elems[lim_idx],
1909 subexp_idx, src_node, src_idx,
1913 <src> <dst> ( <subexp> )
1914 ( <subexp> ) <src> <dst>
1915 ( <subexp1> <src> <subexp2> <dst> <subexp3> ) */
1916 if (src_pos == dst_pos)
1917 continue; /* This is unrelated limitation. */
1926 check_dst_limits_calc_pos_1 (const re_match_context_t *mctx, int boundaries,
1927 Idx subexp_idx, Idx from_node, Idx bkref_idx)
1929 re_dfa_t *const dfa = mctx->dfa;
1930 re_node_set *eclosures = dfa->eclosures + from_node;
1933 /* Else, we are on the boundary: examine the nodes on the epsilon
1935 for (node_idx = 0; node_idx < eclosures->nelem; ++node_idx)
1937 Idx node = eclosures->elems[node_idx];
1938 switch (dfa->nodes[node].type)
1941 if (bkref_idx != REG_MISSING)
1943 struct re_backref_cache_entry *ent = mctx->bkref_ents + bkref_idx;
1949 if (ent->node != node)
1952 if (subexp_idx < BITSET_WORD_BITS
1953 && !(ent->eps_reachable_subexps_map
1954 & ((bitset_word) 1 << subexp_idx)))
1957 /* Recurse trying to reach the OP_OPEN_SUBEXP and
1958 OP_CLOSE_SUBEXP cases below. But, if the
1959 destination node is the same node as the source
1960 node, don't recurse because it would cause an
1961 infinite loop: a regex that exhibits this behavior
1963 dst = dfa->edests[node].elems[0];
1964 if (dst == from_node)
1968 else /* if (boundaries & 2) */
1973 check_dst_limits_calc_pos_1 (mctx, boundaries, subexp_idx,
1975 if (cpos == -1 /* && (boundaries & 1) */)
1977 if (cpos == 0 && (boundaries & 2))
1980 if (subexp_idx < BITSET_WORD_BITS)
1981 ent->eps_reachable_subexps_map &=
1982 ~ ((bitset_word) 1 << subexp_idx);
1984 while (ent++->more);
1988 case OP_OPEN_SUBEXP:
1989 if ((boundaries & 1) && subexp_idx == dfa->nodes[node].opr.idx)
1993 case OP_CLOSE_SUBEXP:
1994 if ((boundaries & 2) && subexp_idx == dfa->nodes[node].opr.idx)
2003 return (boundaries & 2) ? 1 : 0;
2008 check_dst_limits_calc_pos (const re_match_context_t *mctx,
2009 Idx limit, Idx subexp_idx,
2010 Idx from_node, Idx str_idx, Idx bkref_idx)
2012 struct re_backref_cache_entry *lim = mctx->bkref_ents + limit;
2015 /* If we are outside the range of the subexpression, return -1 or 1. */
2016 if (str_idx < lim->subexp_from)
2019 if (lim->subexp_to < str_idx)
2022 /* If we are within the subexpression, return 0. */
2023 boundaries = (str_idx == lim->subexp_from);
2024 boundaries |= (str_idx == lim->subexp_to) << 1;
2025 if (boundaries == 0)
2028 /* Else, examine epsilon closure. */
2029 return check_dst_limits_calc_pos_1 (mctx, boundaries, subexp_idx,
2030 from_node, bkref_idx);
2033 /* Check the limitations of sub expressions LIMITS, and remove the nodes
2034 which are against limitations from DEST_NODES. */
2036 static reg_errcode_t
2038 check_subexp_limits (re_dfa_t *dfa, re_node_set *dest_nodes,
2039 const re_node_set *candidates, re_node_set *limits,
2040 struct re_backref_cache_entry *bkref_ents, Idx str_idx)
2043 Idx node_idx, lim_idx;
2045 for (lim_idx = 0; lim_idx < limits->nelem; ++lim_idx)
2048 struct re_backref_cache_entry *ent;
2049 ent = bkref_ents + limits->elems[lim_idx];
2051 if (str_idx <= ent->subexp_from || ent->str_idx < str_idx)
2052 continue; /* This is unrelated limitation. */
2054 subexp_idx = dfa->nodes[ent->node].opr.idx;
2055 if (ent->subexp_to == str_idx)
2057 Idx ops_node = REG_MISSING;
2058 Idx cls_node = REG_MISSING;
2059 for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx)
2061 Idx node = dest_nodes->elems[node_idx];
2062 re_token_type_t type = dfa->nodes[node].type;
2063 if (type == OP_OPEN_SUBEXP
2064 && subexp_idx == dfa->nodes[node].opr.idx)
2066 else if (type == OP_CLOSE_SUBEXP
2067 && subexp_idx == dfa->nodes[node].opr.idx)
2071 /* Check the limitation of the open subexpression. */
2072 /* Note that (ent->subexp_to = str_idx != ent->subexp_from). */
2073 if (REG_VALID_INDEX (ops_node))
2075 err = sub_epsilon_src_nodes (dfa, ops_node, dest_nodes,
2077 if (BE (err != REG_NOERROR, 0))
2081 /* Check the limitation of the close subexpression. */
2082 if (REG_VALID_INDEX (cls_node))
2083 for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx)
2085 Idx node = dest_nodes->elems[node_idx];
2086 if (!re_node_set_contains (dfa->inveclosures + node,
2088 && !re_node_set_contains (dfa->eclosures + node,
2091 /* It is against this limitation.
2092 Remove it form the current sifted state. */
2093 err = sub_epsilon_src_nodes (dfa, node, dest_nodes,
2095 if (BE (err != REG_NOERROR, 0))
2101 else /* (ent->subexp_to != str_idx) */
2103 for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx)
2105 Idx node = dest_nodes->elems[node_idx];
2106 re_token_type_t type = dfa->nodes[node].type;
2107 if (type == OP_CLOSE_SUBEXP || type == OP_OPEN_SUBEXP)
2109 if (subexp_idx != dfa->nodes[node].opr.idx)
2111 /* It is against this limitation.
2112 Remove it form the current sifted state. */
2113 err = sub_epsilon_src_nodes (dfa, node, dest_nodes,
2115 if (BE (err != REG_NOERROR, 0))
2124 static reg_errcode_t
2126 sift_states_bkref (re_match_context_t *mctx, re_sift_context_t *sctx,
2127 Idx str_idx, const re_node_set *candidates)
2129 re_dfa_t *const dfa = mctx->dfa;
2132 re_sift_context_t local_sctx;
2133 Idx first_idx = search_cur_bkref_entry (mctx, str_idx);
2135 if (first_idx == REG_MISSING)
2138 local_sctx.sifted_states = NULL; /* Mark that it hasn't been initialized. */
2140 for (node_idx = 0; node_idx < candidates->nelem; ++node_idx)
2143 re_token_type_t type;
2144 struct re_backref_cache_entry *entry;
2145 node = candidates->elems[node_idx];
2146 type = dfa->nodes[node].type;
2147 /* Avoid infinite loop for the REs like "()\1+". */
2148 if (node == sctx->last_node && str_idx == sctx->last_str_idx)
2150 if (type != OP_BACK_REF)
2153 entry = mctx->bkref_ents + first_idx;
2154 enabled_idx = first_idx;
2158 Idx subexp_len, to_idx, dst_node;
2159 re_dfastate_t *cur_state;
2161 if (entry->node != node)
2163 subexp_len = entry->subexp_to - entry->subexp_from;
2164 to_idx = str_idx + subexp_len;
2165 dst_node = (subexp_len ? dfa->nexts[node]
2166 : dfa->edests[node].elems[0]);
2168 if (to_idx > sctx->last_str_idx
2169 || sctx->sifted_states[to_idx] == NULL
2170 || !STATE_NODE_CONTAINS (sctx->sifted_states[to_idx], dst_node)
2171 || check_dst_limits (mctx, &sctx->limits, node,
2172 str_idx, dst_node, to_idx))
2175 if (local_sctx.sifted_states == NULL)
2178 err = re_node_set_init_copy (&local_sctx.limits, &sctx->limits);
2179 if (BE (err != REG_NOERROR, 0))
2182 local_sctx.last_node = node;
2183 local_sctx.last_str_idx = str_idx;
2184 ok = re_node_set_insert (&local_sctx.limits, enabled_idx);
2190 cur_state = local_sctx.sifted_states[str_idx];
2191 err = sift_states_backward (mctx, &local_sctx);
2192 if (BE (err != REG_NOERROR, 0))
2194 if (sctx->limited_states != NULL)
2196 err = merge_state_array (dfa, sctx->limited_states,
2197 local_sctx.sifted_states,
2199 if (BE (err != REG_NOERROR, 0))
2202 local_sctx.sifted_states[str_idx] = cur_state;
2203 re_node_set_remove (&local_sctx.limits, enabled_idx);
2205 /* mctx->bkref_ents may have changed, reload the pointer. */
2206 entry = mctx->bkref_ents + enabled_idx;
2208 while (enabled_idx++, entry++->more);
2212 if (local_sctx.sifted_states != NULL)
2214 re_node_set_free (&local_sctx.limits);
2221 #ifdef RE_ENABLE_I18N
2224 sift_states_iter_mb (const re_match_context_t *mctx, re_sift_context_t *sctx,
2225 Idx node_idx, Idx str_idx, Idx max_str_idx)
2227 re_dfa_t *const dfa = mctx->dfa;
2229 /* Check the node can accept `multi byte'. */
2230 naccepted = check_node_accept_bytes (dfa, node_idx, &mctx->input, str_idx);
2231 if (naccepted > 0 && str_idx + naccepted <= max_str_idx &&
2232 !STATE_NODE_CONTAINS (sctx->sifted_states[str_idx + naccepted],
2233 dfa->nexts[node_idx]))
2234 /* The node can't accept the `multi byte', or the
2235 destination was already thrown away, then the node
2236 could't accept the current input `multi byte'. */
2238 /* Otherwise, it is sure that the node could accept
2239 `naccepted' bytes input. */
2242 #endif /* RE_ENABLE_I18N */
2245 /* Functions for state transition. */
2247 /* Return the next state to which the current state STATE will transit by
2248 accepting the current input byte, and update STATE_LOG if necessary.
2249 If STATE can accept a multibyte char/collating element/back reference
2250 update the destination of STATE_LOG. */
2252 static re_dfastate_t *
2254 transit_state (reg_errcode_t *err, re_match_context_t *mctx,
2255 re_dfastate_t *state)
2257 re_dfastate_t **trtable;
2260 #ifdef RE_ENABLE_I18N
2261 /* If the current state can accept multibyte. */
2262 if (BE (state->accept_mb, 0))
2264 *err = transit_state_mb (mctx, state);
2265 if (BE (*err != REG_NOERROR, 0))
2268 #endif /* RE_ENABLE_I18N */
2270 /* Then decide the next state with the single byte. */
2273 /* don't use transition table */
2274 return transit_state_sb (err, mctx, state);
2277 /* Use transition table */
2278 ch = re_string_fetch_byte (&mctx->input);
2281 trtable = state->trtable;
2282 if (BE (trtable != NULL, 1))
2285 trtable = state->word_trtable;
2286 if (BE (trtable != NULL, 1))
2288 unsigned int context;
2290 = re_string_context_at (&mctx->input,
2291 re_string_cur_idx (&mctx->input) - 1,
2293 if (IS_WORD_CONTEXT (context))
2294 return trtable[ch + SBC_MAX];
2299 if (!build_trtable (mctx->dfa, state))
2305 /* Retry, we now have a transition table. */
2309 /* Update the state_log if we need */
2312 merge_state_with_log (reg_errcode_t *err, re_match_context_t *mctx,
2313 re_dfastate_t *next_state)
2315 re_dfa_t *const dfa = mctx->dfa;
2316 Idx cur_idx = re_string_cur_idx (&mctx->input);
2318 if (cur_idx > mctx->state_log_top)
2320 mctx->state_log[cur_idx] = next_state;
2321 mctx->state_log_top = cur_idx;
2323 else if (mctx->state_log[cur_idx] == 0)
2325 mctx->state_log[cur_idx] = next_state;
2329 re_dfastate_t *pstate;
2330 unsigned int context;
2331 re_node_set next_nodes, *log_nodes, *table_nodes = NULL;
2332 /* If (state_log[cur_idx] != 0), it implies that cur_idx is
2333 the destination of a multibyte char/collating element/
2334 back reference. Then the next state is the union set of
2335 these destinations and the results of the transition table. */
2336 pstate = mctx->state_log[cur_idx];
2337 log_nodes = pstate->entrance_nodes;
2338 if (next_state != NULL)
2340 table_nodes = next_state->entrance_nodes;
2341 *err = re_node_set_init_union (&next_nodes, table_nodes,
2343 if (BE (*err != REG_NOERROR, 0))
2347 next_nodes = *log_nodes;
2348 /* Note: We already add the nodes of the initial state,
2349 then we don't need to add them here. */
2351 context = re_string_context_at (&mctx->input,
2352 re_string_cur_idx (&mctx->input) - 1,
2354 next_state = mctx->state_log[cur_idx]
2355 = re_acquire_state_context (err, dfa, &next_nodes, context);
2356 /* We don't need to check errors here, since the return value of
2357 this function is next_state and ERR is already set. */
2359 if (table_nodes != NULL)
2360 re_node_set_free (&next_nodes);
2363 if (BE (dfa->nbackref, 0) && next_state != NULL)
2365 /* Check OP_OPEN_SUBEXP in the current state in case that we use them
2366 later. We must check them here, since the back references in the
2367 next state might use them. */
2368 *err = check_subexp_matching_top (mctx, &next_state->nodes,
2370 if (BE (*err != REG_NOERROR, 0))
2373 /* If the next state has back references. */
2374 if (next_state->has_backref)
2376 *err = transit_state_bkref (mctx, &next_state->nodes);
2377 if (BE (*err != REG_NOERROR, 0))
2379 next_state = mctx->state_log[cur_idx];
2386 /* Skip bytes in the input that correspond to part of a
2387 multi-byte match, then look in the log for a state
2388 from which to restart matching. */
2389 static re_dfastate_t *
2391 find_recover_state (reg_errcode_t *err, re_match_context_t *mctx)
2393 re_dfastate_t *cur_state = NULL;
2396 Idx max = mctx->state_log_top;
2397 Idx cur_str_idx = re_string_cur_idx (&mctx->input);
2401 if (++cur_str_idx > max)
2403 re_string_skip_bytes (&mctx->input, 1);
2405 while (mctx->state_log[cur_str_idx] == NULL);
2407 cur_state = merge_state_with_log (err, mctx, NULL);
2409 while (*err == REG_NOERROR && cur_state == NULL);
2413 /* Helper functions for transit_state. */
2415 /* From the node set CUR_NODES, pick up the nodes whose types are
2416 OP_OPEN_SUBEXP and which have corresponding back references in the regular
2417 expression. And register them to use them later for evaluating the
2418 correspoding back references. */
2420 static reg_errcode_t
2422 check_subexp_matching_top (re_match_context_t *mctx, re_node_set *cur_nodes,
2425 re_dfa_t *const dfa = mctx->dfa;
2429 /* TODO: This isn't efficient.
2430 Because there might be more than one nodes whose types are
2431 OP_OPEN_SUBEXP and whose index is SUBEXP_IDX, we must check all
2434 for (node_idx = 0; node_idx < cur_nodes->nelem; ++node_idx)
2436 Idx node = cur_nodes->elems[node_idx];
2437 if (dfa->nodes[node].type == OP_OPEN_SUBEXP
2438 && dfa->nodes[node].opr.idx < BITSET_WORD_BITS
2439 && (dfa->used_bkref_map
2440 & ((bitset_word) 1 << dfa->nodes[node].opr.idx)))
2442 err = match_ctx_add_subtop (mctx, node, str_idx);
2443 if (BE (err != REG_NOERROR, 0))
2451 /* Return the next state to which the current state STATE will transit by
2452 accepting the current input byte. */
2454 static re_dfastate_t *
2455 transit_state_sb (reg_errcode_t *err, re_match_context_t *mctx,
2456 re_dfastate_t *state)
2458 re_dfa_t *const dfa = mctx->dfa;
2459 re_node_set next_nodes;
2460 re_dfastate_t *next_state;
2461 Idx node_cnt, cur_str_idx = re_string_cur_idx (&mctx->input);
2462 unsigned int context;
2464 *err = re_node_set_alloc (&next_nodes, state->nodes.nelem + 1);
2465 if (BE (*err != REG_NOERROR, 0))
2467 for (node_cnt = 0; node_cnt < state->nodes.nelem; ++node_cnt)
2469 Idx cur_node = state->nodes.elems[node_cnt];
2470 if (check_node_accept (mctx, dfa->nodes + cur_node, cur_str_idx))
2472 *err = re_node_set_merge (&next_nodes,
2473 dfa->eclosures + dfa->nexts[cur_node]);
2474 if (BE (*err != REG_NOERROR, 0))
2476 re_node_set_free (&next_nodes);
2481 context = re_string_context_at (&mctx->input, cur_str_idx, mctx->eflags);
2482 next_state = re_acquire_state_context (err, dfa, &next_nodes, context);
2483 /* We don't need to check errors here, since the return value of
2484 this function is next_state and ERR is already set. */
2486 re_node_set_free (&next_nodes);
2487 re_string_skip_bytes (&mctx->input, 1);
2492 #ifdef RE_ENABLE_I18N
2493 static reg_errcode_t
2495 transit_state_mb (re_match_context_t *mctx, re_dfastate_t *pstate)
2497 re_dfa_t *const dfa = mctx->dfa;
2501 for (i = 0; i < pstate->nodes.nelem; ++i)
2503 re_node_set dest_nodes, *new_nodes;
2504 Idx cur_node_idx = pstate->nodes.elems[i];
2507 unsigned int context;
2508 re_dfastate_t *dest_state;
2510 if (!dfa->nodes[cur_node_idx].accept_mb)
2513 if (dfa->nodes[cur_node_idx].constraint)
2515 context = re_string_context_at (&mctx->input,
2516 re_string_cur_idx (&mctx->input),
2518 if (NOT_SATISFY_NEXT_CONSTRAINT (dfa->nodes[cur_node_idx].constraint,
2523 /* How many bytes the node can accept? */
2524 naccepted = check_node_accept_bytes (dfa, cur_node_idx, &mctx->input,
2525 re_string_cur_idx (&mctx->input));
2529 /* The node can accepts `naccepted' bytes. */
2530 dest_idx = re_string_cur_idx (&mctx->input) + naccepted;
2531 mctx->max_mb_elem_len = ((mctx->max_mb_elem_len < naccepted) ? naccepted
2532 : mctx->max_mb_elem_len);
2533 err = clean_state_log_if_needed (mctx, dest_idx);
2534 if (BE (err != REG_NOERROR, 0))
2537 assert (dfa->nexts[cur_node_idx] != REG_MISSING);
2539 new_nodes = dfa->eclosures + dfa->nexts[cur_node_idx];
2541 dest_state = mctx->state_log[dest_idx];
2542 if (dest_state == NULL)
2543 dest_nodes = *new_nodes;
2546 err = re_node_set_init_union (&dest_nodes,
2547 dest_state->entrance_nodes, new_nodes);
2548 if (BE (err != REG_NOERROR, 0))
2551 context = re_string_context_at (&mctx->input, dest_idx - 1, mctx->eflags);
2552 mctx->state_log[dest_idx]
2553 = re_acquire_state_context (&err, dfa, &dest_nodes, context);
2554 if (dest_state != NULL)
2555 re_node_set_free (&dest_nodes);
2556 if (BE (mctx->state_log[dest_idx] == NULL && err != REG_NOERROR, 0))
2561 #endif /* RE_ENABLE_I18N */
2563 static reg_errcode_t
2565 transit_state_bkref (re_match_context_t *mctx, const re_node_set *nodes)
2567 re_dfa_t *const dfa = mctx->dfa;
2570 Idx cur_str_idx = re_string_cur_idx (&mctx->input);
2572 for (i = 0; i < nodes->nelem; ++i)
2574 Idx dest_str_idx, prev_nelem, bkc_idx;
2575 Idx node_idx = nodes->elems[i];
2576 unsigned int context;
2577 const re_token_t *node = dfa->nodes + node_idx;
2578 re_node_set *new_dest_nodes;
2580 /* Check whether `node' is a backreference or not. */
2581 if (node->type != OP_BACK_REF)
2584 if (node->constraint)
2586 context = re_string_context_at (&mctx->input, cur_str_idx,
2588 if (NOT_SATISFY_NEXT_CONSTRAINT (node->constraint, context))
2592 /* `node' is a backreference.
2593 Check the substring which the substring matched. */
2594 bkc_idx = mctx->nbkref_ents;
2595 err = get_subexp (mctx, node_idx, cur_str_idx);
2596 if (BE (err != REG_NOERROR, 0))
2599 /* And add the epsilon closures (which is `new_dest_nodes') of
2600 the backreference to appropriate state_log. */
2602 assert (dfa->nexts[node_idx] != REG_MISSING);
2604 for (; bkc_idx < mctx->nbkref_ents; ++bkc_idx)
2607 re_dfastate_t *dest_state;
2608 struct re_backref_cache_entry *bkref_ent;
2609 bkref_ent = mctx->bkref_ents + bkc_idx;
2610 if (bkref_ent->node != node_idx || bkref_ent->str_idx != cur_str_idx)
2612 subexp_len = bkref_ent->subexp_to - bkref_ent->subexp_from;
2613 new_dest_nodes = (subexp_len == 0
2614 ? dfa->eclosures + dfa->edests[node_idx].elems[0]
2615 : dfa->eclosures + dfa->nexts[node_idx]);
2616 dest_str_idx = (cur_str_idx + bkref_ent->subexp_to
2617 - bkref_ent->subexp_from);
2618 context = re_string_context_at (&mctx->input, dest_str_idx - 1,
2620 dest_state = mctx->state_log[dest_str_idx];
2621 prev_nelem = ((mctx->state_log[cur_str_idx] == NULL) ? 0
2622 : mctx->state_log[cur_str_idx]->nodes.nelem);
2623 /* Add `new_dest_node' to state_log. */
2624 if (dest_state == NULL)
2626 mctx->state_log[dest_str_idx]
2627 = re_acquire_state_context (&err, dfa, new_dest_nodes,
2629 if (BE (mctx->state_log[dest_str_idx] == NULL
2630 && err != REG_NOERROR, 0))
2635 re_node_set dest_nodes;
2636 err = re_node_set_init_union (&dest_nodes,
2637 dest_state->entrance_nodes,
2639 if (BE (err != REG_NOERROR, 0))
2641 re_node_set_free (&dest_nodes);
2644 mctx->state_log[dest_str_idx]
2645 = re_acquire_state_context (&err, dfa, &dest_nodes, context);
2646 re_node_set_free (&dest_nodes);
2647 if (BE (mctx->state_log[dest_str_idx] == NULL
2648 && err != REG_NOERROR, 0))
2651 /* We need to check recursively if the backreference can epsilon
2654 && mctx->state_log[cur_str_idx]->nodes.nelem > prev_nelem)
2656 err = check_subexp_matching_top (mctx, new_dest_nodes,
2658 if (BE (err != REG_NOERROR, 0))
2660 err = transit_state_bkref (mctx, new_dest_nodes);
2661 if (BE (err != REG_NOERROR, 0))
2671 /* Enumerate all the candidates which the backreference BKREF_NODE can match
2672 at BKREF_STR_IDX, and register them by match_ctx_add_entry().
2673 Note that we might collect inappropriate candidates here.
2674 However, the cost of checking them strictly here is too high, then we
2675 delay these checking for prune_impossible_nodes(). */
2677 static reg_errcode_t
2679 get_subexp (re_match_context_t *mctx, Idx bkref_node, Idx bkref_str_idx)
2681 re_dfa_t *const dfa = mctx->dfa;
2682 Idx subexp_num, sub_top_idx;
2683 const char *buf = (const char *) re_string_get_buffer (&mctx->input);
2684 /* Return if we have already checked BKREF_NODE at BKREF_STR_IDX. */
2685 Idx cache_idx = search_cur_bkref_entry (mctx, bkref_str_idx);
2686 if (cache_idx != REG_MISSING)
2688 const struct re_backref_cache_entry *entry = mctx->bkref_ents + cache_idx;
2690 if (entry->node == bkref_node)
2691 return REG_NOERROR; /* We already checked it. */
2692 while (entry++->more);
2695 subexp_num = dfa->nodes[bkref_node].opr.idx;
2697 /* For each sub expression */
2698 for (sub_top_idx = 0; sub_top_idx < mctx->nsub_tops; ++sub_top_idx)
2701 re_sub_match_top_t *sub_top = mctx->sub_tops[sub_top_idx];
2702 re_sub_match_last_t *sub_last;
2703 Idx sub_last_idx, sl_str, bkref_str_off;
2705 if (dfa->nodes[sub_top->node].opr.idx != subexp_num)
2706 continue; /* It isn't related. */
2708 sl_str = sub_top->str_idx;
2709 bkref_str_off = bkref_str_idx;
2710 /* At first, check the last node of sub expressions we already
2712 for (sub_last_idx = 0; sub_last_idx < sub_top->nlasts; ++sub_last_idx)
2714 regoff_t sl_str_diff;
2715 sub_last = sub_top->lasts[sub_last_idx];
2716 sl_str_diff = sub_last->str_idx - sl_str;
2717 /* The matched string by the sub expression match with the substring
2718 at the back reference? */
2719 if (sl_str_diff > 0)
2721 if (BE (bkref_str_off + sl_str_diff > mctx->input.valid_len, 0))
2723 /* Not enough chars for a successful match. */
2724 if (bkref_str_off + sl_str_diff > mctx->input.len)
2727 err = clean_state_log_if_needed (mctx,
2730 if (BE (err != REG_NOERROR, 0))
2732 buf = (const char *) re_string_get_buffer (&mctx->input);
2734 if (memcmp (buf + bkref_str_off, buf + sl_str, sl_str_diff) != 0)
2735 break; /* We don't need to search this sub expression any more. */
2737 bkref_str_off += sl_str_diff;
2738 sl_str += sl_str_diff;
2739 err = get_subexp_sub (mctx, sub_top, sub_last, bkref_node,
2742 /* Reload buf, since the preceding call might have reallocated
2744 buf = (const char *) re_string_get_buffer (&mctx->input);
2746 if (err == REG_NOMATCH)
2748 if (BE (err != REG_NOERROR, 0))
2752 if (sub_last_idx < sub_top->nlasts)
2754 if (sub_last_idx > 0)
2756 /* Then, search for the other last nodes of the sub expression. */
2757 for (; sl_str <= bkref_str_idx; ++sl_str)
2760 regoff_t sl_str_off;
2761 const re_node_set *nodes;
2762 sl_str_off = sl_str - sub_top->str_idx;
2763 /* The matched string by the sub expression match with the substring
2764 at the back reference? */
2767 if (BE (bkref_str_off >= mctx->input.valid_len, 0))
2769 /* If we are at the end of the input, we cannot match. */
2770 if (bkref_str_off >= mctx->input.len)
2773 err = extend_buffers (mctx);
2774 if (BE (err != REG_NOERROR, 0))
2777 buf = (const char *) re_string_get_buffer (&mctx->input);
2779 if (buf [bkref_str_off++] != buf[sl_str - 1])
2780 break; /* We don't need to search this sub expression
2783 if (mctx->state_log[sl_str] == NULL)
2785 /* Does this state have a ')' of the sub expression? */
2786 nodes = &mctx->state_log[sl_str]->nodes;
2787 cls_node = find_subexp_node (dfa, nodes, subexp_num, OP_CLOSE_SUBEXP);
2788 if (cls_node == REG_MISSING)
2790 if (sub_top->path == NULL)
2792 sub_top->path = re_calloc (state_array_t,
2793 sl_str - sub_top->str_idx + 1);
2794 if (sub_top->path == NULL)
2797 /* Can the OP_OPEN_SUBEXP node arrive the OP_CLOSE_SUBEXP node
2798 in the current context? */
2799 err = check_arrival (mctx, sub_top->path, sub_top->node,
2800 sub_top->str_idx, cls_node, sl_str, OP_CLOSE_SUBEXP);
2801 if (err == REG_NOMATCH)
2803 if (BE (err != REG_NOERROR, 0))
2805 sub_last = match_ctx_add_sublast (sub_top, cls_node, sl_str);
2806 if (BE (sub_last == NULL, 0))
2808 err = get_subexp_sub (mctx, sub_top, sub_last, bkref_node,
2810 if (err == REG_NOMATCH)
2817 /* Helper functions for get_subexp(). */
2819 /* Check SUB_LAST can arrive to the back reference BKREF_NODE at BKREF_STR.
2820 If it can arrive, register the sub expression expressed with SUB_TOP
2823 static reg_errcode_t
2825 get_subexp_sub (re_match_context_t *mctx, const re_sub_match_top_t *sub_top,
2826 re_sub_match_last_t *sub_last, Idx bkref_node, Idx bkref_str)
2830 /* Can the subexpression arrive the back reference? */
2831 err = check_arrival (mctx, &sub_last->path, sub_last->node,
2832 sub_last->str_idx, bkref_node, bkref_str, OP_OPEN_SUBEXP);
2833 if (err != REG_NOERROR)
2835 err = match_ctx_add_entry (mctx, bkref_node, bkref_str, sub_top->str_idx,
2837 if (BE (err != REG_NOERROR, 0))
2839 to_idx = bkref_str + sub_last->str_idx - sub_top->str_idx;
2840 return clean_state_log_if_needed (mctx, to_idx);
2843 /* Find the first node which is '(' or ')' and whose index is SUBEXP_IDX.
2844 Search '(' if FL_OPEN, or search ')' otherwise.
2845 TODO: This function isn't efficient...
2846 Because there might be more than one nodes whose types are
2847 OP_OPEN_SUBEXP and whose index is SUBEXP_IDX, we must check all
2853 find_subexp_node (const re_dfa_t *dfa, const re_node_set *nodes,
2854 Idx subexp_idx, int type)
2857 for (cls_idx = 0; cls_idx < nodes->nelem; ++cls_idx)
2859 Idx cls_node = nodes->elems[cls_idx];
2860 const re_token_t *node = dfa->nodes + cls_node;
2861 if (node->type == type
2862 && node->opr.idx == subexp_idx)
2868 /* Check whether the node TOP_NODE at TOP_STR can arrive to the node
2869 LAST_NODE at LAST_STR. We record the path onto PATH since it will be
2871 Return REG_NOERROR if it can arrive, or REG_NOMATCH otherwise. */
2873 static reg_errcode_t
2875 check_arrival (re_match_context_t *mctx, state_array_t *path,
2876 Idx top_node, Idx top_str, Idx last_node, Idx last_str,
2879 re_dfa_t *const dfa = mctx->dfa;
2881 Idx subexp_num, backup_cur_idx, str_idx, null_cnt;
2882 re_dfastate_t *cur_state = NULL;
2883 re_node_set *cur_nodes, next_nodes;
2884 re_dfastate_t **backup_state_log;
2885 unsigned int context;
2887 subexp_num = dfa->nodes[top_node].opr.idx;
2888 /* Extend the buffer if we need. */
2889 if (BE (path->alloc < last_str + mctx->max_mb_elem_len + 1, 0))
2891 re_dfastate_t **new_array;
2892 Idx old_alloc = path->alloc;
2893 Idx new_alloc = old_alloc + last_str + mctx->max_mb_elem_len + 1;
2894 if (BE (new_alloc < old_alloc, 0))
2896 new_array = re_xrealloc (path->array, re_dfastate_t *, new_alloc);
2897 if (BE (new_array == NULL, 0))
2899 path->array = new_array;
2900 path->alloc = new_alloc;
2901 memset (new_array + old_alloc, '\0',
2902 sizeof (re_dfastate_t *) * (new_alloc - old_alloc));
2905 str_idx = path->next_idx == 0 ? top_str : path->next_idx;
2907 /* Temporary modify MCTX. */
2908 backup_state_log = mctx->state_log;
2909 backup_cur_idx = mctx->input.cur_idx;
2910 mctx->state_log = path->array;
2911 mctx->input.cur_idx = str_idx;
2913 /* Setup initial node set. */
2914 context = re_string_context_at (&mctx->input, str_idx - 1, mctx->eflags);
2915 if (str_idx == top_str)
2917 err = re_node_set_init_1 (&next_nodes, top_node);
2918 if (BE (err != REG_NOERROR, 0))
2920 err = check_arrival_expand_ecl (dfa, &next_nodes, subexp_num, type);
2921 if (BE (err != REG_NOERROR, 0))
2923 re_node_set_free (&next_nodes);
2929 cur_state = mctx->state_log[str_idx];
2930 if (cur_state && cur_state->has_backref)
2932 err = re_node_set_init_copy (&next_nodes, &cur_state->nodes);
2933 if (BE ( err != REG_NOERROR, 0))
2937 re_node_set_init_empty (&next_nodes);
2939 if (str_idx == top_str || (cur_state && cur_state->has_backref))
2941 if (next_nodes.nelem)
2943 err = expand_bkref_cache (mctx, &next_nodes, str_idx,
2945 if (BE ( err != REG_NOERROR, 0))
2947 re_node_set_free (&next_nodes);
2951 cur_state = re_acquire_state_context (&err, dfa, &next_nodes, context);
2952 if (BE (cur_state == NULL && err != REG_NOERROR, 0))
2954 re_node_set_free (&next_nodes);
2957 mctx->state_log[str_idx] = cur_state;
2960 for (null_cnt = 0; str_idx < last_str && null_cnt <= mctx->max_mb_elem_len;)
2962 re_node_set_empty (&next_nodes);
2963 if (mctx->state_log[str_idx + 1])
2965 err = re_node_set_merge (&next_nodes,
2966 &mctx->state_log[str_idx + 1]->nodes);
2967 if (BE (err != REG_NOERROR, 0))
2969 re_node_set_free (&next_nodes);
2975 err = check_arrival_add_next_nodes (mctx, str_idx,
2976 &cur_state->non_eps_nodes, &next_nodes);
2977 if (BE (err != REG_NOERROR, 0))
2979 re_node_set_free (&next_nodes);
2984 if (next_nodes.nelem)
2986 err = check_arrival_expand_ecl (dfa, &next_nodes, subexp_num, type);
2987 if (BE (err != REG_NOERROR, 0))
2989 re_node_set_free (&next_nodes);
2992 err = expand_bkref_cache (mctx, &next_nodes, str_idx,
2994 if (BE ( err != REG_NOERROR, 0))
2996 re_node_set_free (&next_nodes);
3000 context = re_string_context_at (&mctx->input, str_idx - 1, mctx->eflags);
3001 cur_state = re_acquire_state_context (&err, dfa, &next_nodes, context);
3002 if (BE (cur_state == NULL && err != REG_NOERROR, 0))
3004 re_node_set_free (&next_nodes);
3007 mctx->state_log[str_idx] = cur_state;
3008 null_cnt = cur_state == NULL ? null_cnt + 1 : 0;
3010 re_node_set_free (&next_nodes);
3011 cur_nodes = (mctx->state_log[last_str] == NULL ? NULL
3012 : &mctx->state_log[last_str]->nodes);
3013 path->next_idx = str_idx;
3016 mctx->state_log = backup_state_log;
3017 mctx->input.cur_idx = backup_cur_idx;
3019 /* Then check the current node set has the node LAST_NODE. */
3020 if (cur_nodes != NULL && re_node_set_contains (cur_nodes, last_node))
3026 /* Helper functions for check_arrival. */
3028 /* Calculate the destination nodes of CUR_NODES at STR_IDX, and append them
3030 TODO: This function is similar to the functions transit_state*(),
3031 however this function has many additional works.
3032 Can't we unify them? */
3034 static reg_errcode_t
3036 check_arrival_add_next_nodes (re_match_context_t *mctx, Idx str_idx,
3037 re_node_set *cur_nodes,
3038 re_node_set *next_nodes)
3040 re_dfa_t *const dfa = mctx->dfa;
3044 re_node_set union_set;
3045 re_node_set_init_empty (&union_set);
3046 for (cur_idx = 0; cur_idx < cur_nodes->nelem; ++cur_idx)
3049 Idx cur_node = cur_nodes->elems[cur_idx];
3051 re_token_type_t type = dfa->nodes[cur_node].type;
3052 assert (!IS_EPSILON_NODE (type));
3054 #ifdef RE_ENABLE_I18N
3055 /* If the node may accept `multi byte'. */
3056 if (dfa->nodes[cur_node].accept_mb)
3058 naccepted = check_node_accept_bytes (dfa, cur_node, &mctx->input,
3062 re_dfastate_t *dest_state;
3063 Idx next_node = dfa->nexts[cur_node];
3064 Idx next_idx = str_idx + naccepted;
3065 dest_state = mctx->state_log[next_idx];
3066 re_node_set_empty (&union_set);
3069 err = re_node_set_merge (&union_set, &dest_state->nodes);
3070 if (BE (err != REG_NOERROR, 0))
3072 re_node_set_free (&union_set);
3076 ok = re_node_set_insert (&union_set, next_node);
3079 re_node_set_free (&union_set);
3082 mctx->state_log[next_idx] = re_acquire_state (&err, dfa,
3084 if (BE (mctx->state_log[next_idx] == NULL
3085 && err != REG_NOERROR, 0))
3087 re_node_set_free (&union_set);
3092 #endif /* RE_ENABLE_I18N */
3094 || check_node_accept (mctx, dfa->nodes + cur_node, str_idx))
3096 ok = re_node_set_insert (next_nodes, dfa->nexts[cur_node]);
3099 re_node_set_free (&union_set);
3104 re_node_set_free (&union_set);
3108 /* For all the nodes in CUR_NODES, add the epsilon closures of them to
3109 CUR_NODES, however exclude the nodes which are:
3110 - inside the sub expression whose number is EX_SUBEXP, if FL_OPEN.
3111 - out of the sub expression whose number is EX_SUBEXP, if !FL_OPEN.
3114 static reg_errcode_t
3116 check_arrival_expand_ecl (re_dfa_t *dfa, re_node_set *cur_nodes,
3117 Idx ex_subexp, int type)
3120 Idx idx, outside_node;
3121 re_node_set new_nodes;
3123 assert (cur_nodes->nelem);
3125 err = re_node_set_alloc (&new_nodes, cur_nodes->nelem);
3126 if (BE (err != REG_NOERROR, 0))
3128 /* Create a new node set NEW_NODES with the nodes which are epsilon
3129 closures of the node in CUR_NODES. */
3131 for (idx = 0; idx < cur_nodes->nelem; ++idx)
3133 Idx cur_node = cur_nodes->elems[idx];
3134 re_node_set *eclosure = dfa->eclosures + cur_node;
3135 outside_node = find_subexp_node (dfa, eclosure, ex_subexp, type);
3136 if (outside_node == REG_MISSING)
3138 /* There are no problematic nodes, just merge them. */
3139 err = re_node_set_merge (&new_nodes, eclosure);
3140 if (BE (err != REG_NOERROR, 0))
3142 re_node_set_free (&new_nodes);
3148 /* There are problematic nodes, re-calculate incrementally. */
3149 err = check_arrival_expand_ecl_sub (dfa, &new_nodes, cur_node,
3151 if (BE (err != REG_NOERROR, 0))
3153 re_node_set_free (&new_nodes);
3158 re_node_set_free (cur_nodes);
3159 *cur_nodes = new_nodes;
3163 /* Helper function for check_arrival_expand_ecl.
3164 Check incrementally the epsilon closure of TARGET, and if it isn't
3165 problematic append it to DST_NODES. */
3167 static reg_errcode_t
3169 check_arrival_expand_ecl_sub (re_dfa_t *dfa, re_node_set *dst_nodes,
3170 Idx target, Idx ex_subexp, int type)
3173 for (cur_node = target; !re_node_set_contains (dst_nodes, cur_node);)
3177 if (dfa->nodes[cur_node].type == type
3178 && dfa->nodes[cur_node].opr.idx == ex_subexp)
3180 if (type == OP_CLOSE_SUBEXP)
3182 ok = re_node_set_insert (dst_nodes, cur_node);
3188 ok = re_node_set_insert (dst_nodes, cur_node);
3191 if (dfa->edests[cur_node].nelem == 0)
3193 if (dfa->edests[cur_node].nelem == 2)
3196 check_arrival_expand_ecl_sub (dfa, dst_nodes,
3197 dfa->edests[cur_node].elems[1],
3199 if (BE (ret != REG_NOERROR, 0))
3202 cur_node = dfa->edests[cur_node].elems[0];
3208 /* For all the back references in the current state, calculate the
3209 destination of the back references by the appropriate entry
3210 in MCTX->BKREF_ENTS. */
3212 static reg_errcode_t
3214 expand_bkref_cache (re_match_context_t *mctx, re_node_set *cur_nodes,
3215 Idx cur_str, Idx subexp_num, int type)
3217 re_dfa_t *const dfa = mctx->dfa;
3219 Idx cache_idx_start = search_cur_bkref_entry (mctx, cur_str);
3220 struct re_backref_cache_entry *ent;
3222 if (cache_idx_start == REG_MISSING)
3226 ent = mctx->bkref_ents + cache_idx_start;
3229 Idx to_idx, next_node;
3231 /* Is this entry ENT is appropriate? */
3232 if (!re_node_set_contains (cur_nodes, ent->node))
3235 to_idx = cur_str + ent->subexp_to - ent->subexp_from;
3236 /* Calculate the destination of the back reference, and append it
3237 to MCTX->STATE_LOG. */
3238 if (to_idx == cur_str)
3240 /* The backreference did epsilon transit, we must re-check all the
3241 node in the current state. */
3242 re_node_set new_dests;
3243 reg_errcode_t err2, err3;
3244 next_node = dfa->edests[ent->node].elems[0];
3245 if (re_node_set_contains (cur_nodes, next_node))
3247 err = re_node_set_init_1 (&new_dests, next_node);
3248 err2 = check_arrival_expand_ecl (dfa, &new_dests, subexp_num, type);
3249 err3 = re_node_set_merge (cur_nodes, &new_dests);
3250 re_node_set_free (&new_dests);
3251 if (BE (err != REG_NOERROR || err2 != REG_NOERROR
3252 || err3 != REG_NOERROR, 0))
3254 err = (err != REG_NOERROR ? err
3255 : (err2 != REG_NOERROR ? err2 : err3));
3258 /* TODO: It is still inefficient... */
3263 re_node_set union_set;
3264 next_node = dfa->nexts[ent->node];
3265 if (mctx->state_log[to_idx])
3268 if (re_node_set_contains (&mctx->state_log[to_idx]->nodes,
3271 err = re_node_set_init_copy (&union_set,
3272 &mctx->state_log[to_idx]->nodes);
3273 ok = re_node_set_insert (&union_set, next_node);
3274 if (BE (err != REG_NOERROR || ! ok, 0))
3276 re_node_set_free (&union_set);
3277 err = err != REG_NOERROR ? err : REG_ESPACE;
3283 err = re_node_set_init_1 (&union_set, next_node);
3284 if (BE (err != REG_NOERROR, 0))
3287 mctx->state_log[to_idx] = re_acquire_state (&err, dfa, &union_set);
3288 re_node_set_free (&union_set);
3289 if (BE (mctx->state_log[to_idx] == NULL
3290 && err != REG_NOERROR, 0))
3294 while (ent++->more);
3298 /* Build transition table for the state.
3299 Return true if successful. */
3303 build_trtable (re_dfa_t *dfa, re_dfastate_t *state)
3308 bool need_word_trtable = false;
3309 bitset_word elem, mask;
3310 bool dests_node_malloced = false, dest_states_malloced = false;
3311 Idx ndests; /* Number of the destination states from `state'. */
3312 re_dfastate_t **trtable;
3313 re_dfastate_t **dest_states = NULL, **dest_states_word, **dest_states_nl;
3314 re_node_set follows, *dests_node;
3320 re_node_set dests_node[SBC_MAX];
3321 bitset dests_ch[SBC_MAX];
3324 /* We build DFA states which corresponds to the destination nodes
3325 from `state'. `dests_node[i]' represents the nodes which i-th
3326 destination state contains, and `dests_ch[i]' represents the
3327 characters which i-th destination state accepts. */
3328 if (__libc_use_alloca (sizeof (struct dests_alloc)))
3329 dests_alloc = (struct dests_alloc *) alloca (sizeof dests_alloc[0]);
3332 dests_alloc = re_malloc (struct dests_alloc, 1);
3333 if (BE (dests_alloc == NULL, 0))
3335 dests_node_malloced = true;
3337 dests_node = dests_alloc->dests_node;
3338 dests_ch = dests_alloc->dests_ch;
3340 /* Initialize transiton table. */
3341 state->word_trtable = state->trtable = NULL;
3343 /* At first, group all nodes belonging to `state' into several
3345 ndests = group_nodes_into_DFAstates (dfa, state, dests_node, dests_ch);
3346 if (BE (! REG_VALID_NONZERO_INDEX (ndests), 0))
3348 if (dests_node_malloced)
3352 state->trtable = re_calloc (re_dfastate_t *, SBC_MAX);
3358 err = re_node_set_alloc (&follows, ndests + 1);
3359 if (BE (err != REG_NOERROR, 0))
3362 /* Avoid arithmetic overflow in size calculation. */
3363 if (BE (((SIZE_MAX - (sizeof (re_node_set) + sizeof (bitset)) * SBC_MAX)
3364 / (3 * sizeof (re_dfastate_t *)))
3368 if (__libc_use_alloca ((sizeof (re_node_set) + sizeof (bitset)) * SBC_MAX
3369 + ndests * 3 * sizeof (re_dfastate_t *)))
3370 dest_states = (re_dfastate_t **)
3371 alloca (ndests * 3 * sizeof (re_dfastate_t *));
3374 dest_states = (re_dfastate_t **)
3375 malloc (ndests * 3 * sizeof (re_dfastate_t *));
3376 if (BE (dest_states == NULL, 0))
3379 if (dest_states_malloced)
3381 re_node_set_free (&follows);
3382 for (i = 0; i < ndests; ++i)
3383 re_node_set_free (dests_node + i);
3384 if (dests_node_malloced)
3388 dest_states_malloced = true;
3390 dest_states_word = dest_states + ndests;
3391 dest_states_nl = dest_states_word + ndests;
3392 bitset_empty (acceptable);
3394 /* Then build the states for all destinations. */
3395 for (i = 0; i < ndests; ++i)
3398 re_node_set_empty (&follows);
3399 /* Merge the follows of this destination states. */
3400 for (j = 0; j < dests_node[i].nelem; ++j)
3402 next_node = dfa->nexts[dests_node[i].elems[j]];
3403 if (next_node != REG_MISSING)
3405 err = re_node_set_merge (&follows, dfa->eclosures + next_node);
3406 if (BE (err != REG_NOERROR, 0))
3410 dest_states[i] = re_acquire_state_context (&err, dfa, &follows, 0);
3411 if (BE (dest_states[i] == NULL && err != REG_NOERROR, 0))
3413 /* If the new state has context constraint,
3414 build appropriate states for these contexts. */
3415 if (dest_states[i]->has_constraint)
3417 dest_states_word[i] = re_acquire_state_context (&err, dfa, &follows,
3419 if (BE (dest_states_word[i] == NULL && err != REG_NOERROR, 0))
3422 if (dest_states[i] != dest_states_word[i] && dfa->mb_cur_max > 1)
3423 need_word_trtable = true;
3425 dest_states_nl[i] = re_acquire_state_context (&err, dfa, &follows,
3427 if (BE (dest_states_nl[i] == NULL && err != REG_NOERROR, 0))
3432 dest_states_word[i] = dest_states[i];
3433 dest_states_nl[i] = dest_states[i];
3435 bitset_merge (acceptable, dests_ch[i]);
3438 if (!BE (need_word_trtable, 0))
3440 /* We don't care about whether the following character is a word
3441 character, or we are in a single-byte character set so we can
3442 discern by looking at the character code: allocate a
3443 256-entry transition table. */
3444 trtable = state->trtable = re_calloc (re_dfastate_t *, SBC_MAX);
3445 if (BE (trtable == NULL, 0))
3448 /* For all characters ch...: */
3449 for (i = 0; i < BITSET_WORDS; ++i)
3450 for (ch = i * BITSET_WORD_BITS, elem = acceptable[i], mask = 1;
3452 mask <<= 1, elem >>= 1, ++ch)
3453 if (BE (elem & 1, 0))
3455 /* There must be exactly one destination which accepts
3456 character ch. See group_nodes_into_DFAstates. */
3457 for (j = 0; (dests_ch[j][i] & mask) == 0; ++j)
3460 /* j-th destination accepts the word character ch. */
3461 if (dfa->word_char[i] & mask)
3462 trtable[ch] = dest_states_word[j];
3464 trtable[ch] = dest_states[j];
3469 /* We care about whether the following character is a word
3470 character, and we are in a multi-byte character set: discern
3471 by looking at the character code: build two 256-entry
3472 transition tables, one starting at trtable[0] and one
3473 starting at trtable[SBC_MAX]. */
3474 trtable = state->word_trtable = re_calloc (re_dfastate_t *, 2 * SBC_MAX);
3475 if (BE (trtable == NULL, 0))
3478 /* For all characters ch...: */
3479 for (i = 0; i < BITSET_WORDS; ++i)
3480 for (ch = i * BITSET_WORD_BITS, elem = acceptable[i], mask = 1;
3482 mask <<= 1, elem >>= 1, ++ch)
3483 if (BE (elem & 1, 0))
3485 /* There must be exactly one destination which accepts
3486 character ch. See group_nodes_into_DFAstates. */
3487 for (j = 0; (dests_ch[j][i] & mask) == 0; ++j)
3490 /* j-th destination accepts the word character ch. */
3491 trtable[ch] = dest_states[j];
3492 trtable[ch + SBC_MAX] = dest_states_word[j];
3497 if (bitset_contain (acceptable, NEWLINE_CHAR))
3499 /* The current state accepts newline character. */
3500 for (j = 0; j < ndests; ++j)
3501 if (bitset_contain (dests_ch[j], NEWLINE_CHAR))
3503 /* k-th destination accepts newline character. */
3504 trtable[NEWLINE_CHAR] = dest_states_nl[j];
3505 if (need_word_trtable)
3506 trtable[NEWLINE_CHAR + SBC_MAX] = dest_states_nl[j];
3507 /* There must be only one destination which accepts
3508 newline. See group_nodes_into_DFAstates. */
3513 if (dest_states_malloced)
3516 re_node_set_free (&follows);
3517 for (i = 0; i < ndests; ++i)
3518 re_node_set_free (dests_node + i);
3520 if (dests_node_malloced)
3526 /* Group all nodes belonging to STATE into several destinations.
3527 Then for all destinations, set the nodes belonging to the destination
3528 to DESTS_NODE[i] and set the characters accepted by the destination
3529 to DEST_CH[i]. This function return the number of destinations. */
3533 group_nodes_into_DFAstates (const re_dfa_t *dfa, const re_dfastate_t *state,
3534 re_node_set *dests_node, bitset *dests_ch)
3539 Idx ndests; /* Number of the destinations from `state'. */
3540 bitset accepts; /* Characters a node can accept. */
3541 const re_node_set *cur_nodes = &state->nodes;
3542 bitset_empty (accepts);
3545 /* For all the nodes belonging to `state', */
3546 for (i = 0; i < cur_nodes->nelem; ++i)
3548 re_token_t *node = &dfa->nodes[cur_nodes->elems[i]];
3549 re_token_type_t type = node->type;
3550 unsigned int constraint = node->constraint;
3552 /* Enumerate all single byte character this node can accept. */
3553 if (type == CHARACTER)
3554 bitset_set (accepts, node->opr.c);
3555 else if (type == SIMPLE_BRACKET)
3557 bitset_merge (accepts, node->opr.sbcset);
3559 else if (type == OP_PERIOD)
3561 #ifdef RE_ENABLE_I18N
3562 if (dfa->mb_cur_max > 1)
3563 bitset_merge (accepts, dfa->sb_char);
3566 bitset_set_all (accepts);
3567 if (!(dfa->syntax & REG_DOT_NEWLINE))
3568 bitset_clear (accepts, '\n');
3569 if (dfa->syntax & REG_DOT_NOT_NULL)
3570 bitset_clear (accepts, '\0');
3572 #ifdef RE_ENABLE_I18N
3573 else if (type == OP_UTF8_PERIOD)
3575 if (SBC_MAX / 2 % BITSET_WORD_BITS == 0)
3576 memset (accepts, -1, sizeof accepts / 2);
3578 bitset_merge (accepts, utf8_sb_map);
3579 if (!(dfa->syntax & REG_DOT_NEWLINE))
3580 bitset_clear (accepts, '\n');
3581 if (dfa->syntax & REG_DOT_NOT_NULL)
3582 bitset_clear (accepts, '\0');
3588 /* Check the `accepts' and sift the characters which are not
3589 match it the context. */
3592 if (constraint & NEXT_NEWLINE_CONSTRAINT)
3594 bool accepts_newline = bitset_contain (accepts, NEWLINE_CHAR);
3595 bitset_empty (accepts);
3596 if (accepts_newline)
3597 bitset_set (accepts, NEWLINE_CHAR);
3601 if (constraint & NEXT_ENDBUF_CONSTRAINT)
3603 bitset_empty (accepts);
3607 if (constraint & NEXT_WORD_CONSTRAINT)
3609 bitset_word any_set = 0;
3610 if (type == CHARACTER && !node->word_char)
3612 bitset_empty (accepts);
3615 #ifdef RE_ENABLE_I18N
3616 if (dfa->mb_cur_max > 1)
3617 for (j = 0; j < BITSET_WORDS; ++j)
3618 any_set |= (accepts[j] &= (dfa->word_char[j] | ~dfa->sb_char[j]));
3621 for (j = 0; j < BITSET_WORDS; ++j)
3622 any_set |= (accepts[j] &= dfa->word_char[j]);
3626 if (constraint & NEXT_NOTWORD_CONSTRAINT)
3628 bitset_word any_set = 0;
3629 if (type == CHARACTER && node->word_char)
3631 bitset_empty (accepts);
3634 #ifdef RE_ENABLE_I18N
3635 if (dfa->mb_cur_max > 1)
3636 for (j = 0; j < BITSET_WORDS; ++j)
3637 any_set |= (accepts[j] &= ~(dfa->word_char[j] & dfa->sb_char[j]));
3640 for (j = 0; j < BITSET_WORDS; ++j)
3641 any_set |= (accepts[j] &= ~dfa->word_char[j]);
3647 /* Then divide `accepts' into DFA states, or create a new
3648 state. Above, we make sure that accepts is not empty. */
3649 for (j = 0; j < ndests; ++j)
3651 bitset intersec; /* Intersection sets, see below. */
3653 /* Flags, see below. */
3654 bitset_word has_intersec, not_subset, not_consumed;
3656 /* Optimization, skip if this state doesn't accept the character. */
3657 if (type == CHARACTER && !bitset_contain (dests_ch[j], node->opr.c))
3660 /* Enumerate the intersection set of this state and `accepts'. */
3662 for (k = 0; k < BITSET_WORDS; ++k)
3663 has_intersec |= intersec[k] = accepts[k] & dests_ch[j][k];
3664 /* And skip if the intersection set is empty. */
3668 /* Then check if this state is a subset of `accepts'. */
3669 not_subset = not_consumed = 0;
3670 for (k = 0; k < BITSET_WORDS; ++k)
3672 not_subset |= remains[k] = ~accepts[k] & dests_ch[j][k];
3673 not_consumed |= accepts[k] = accepts[k] & ~dests_ch[j][k];
3676 /* If this state isn't a subset of `accepts', create a
3677 new group state, which has the `remains'. */
3680 bitset_copy (dests_ch[ndests], remains);
3681 bitset_copy (dests_ch[j], intersec);
3682 err = re_node_set_init_copy (dests_node + ndests, &dests_node[j]);
3683 if (BE (err != REG_NOERROR, 0))
3688 /* Put the position in the current group. */
3689 ok = re_node_set_insert (&dests_node[j], cur_nodes->elems[i]);
3693 /* If all characters are consumed, go to next node. */
3697 /* Some characters remain, create a new group. */
3700 bitset_copy (dests_ch[ndests], accepts);
3701 err = re_node_set_init_1 (dests_node + ndests, cur_nodes->elems[i]);
3702 if (BE (err != REG_NOERROR, 0))
3705 bitset_empty (accepts);
3710 for (j = 0; j < ndests; ++j)
3711 re_node_set_free (dests_node + j);
3715 #ifdef RE_ENABLE_I18N
3716 /* Check how many bytes the node `dfa->nodes[node_idx]' accepts.
3717 Return the number of the bytes the node accepts.
3718 STR_IDX is the current index of the input string.
3720 This function handles the nodes which can accept one character, or
3721 one collating element like '.', '[a-z]', opposite to the other nodes
3722 can only accept one byte. */
3726 check_node_accept_bytes (re_dfa_t *dfa, Idx node_idx,
3727 const re_string_t *input, Idx str_idx)
3729 const re_token_t *node = dfa->nodes + node_idx;
3730 int char_len, elem_len;
3733 if (BE (node->type == OP_UTF8_PERIOD, 0))
3735 unsigned char c = re_string_byte_at (input, str_idx), d;
3736 if (BE (c < 0xc2, 1))
3739 if (str_idx + 2 > input->len)
3742 d = re_string_byte_at (input, str_idx + 1);
3744 return (d < 0x80 || d > 0xbf) ? 0 : 2;
3748 if (c == 0xe0 && d < 0xa0)
3754 if (c == 0xf0 && d < 0x90)
3760 if (c == 0xf8 && d < 0x88)
3766 if (c == 0xfc && d < 0x84)
3772 if (str_idx + char_len > input->len)
3775 for (i = 1; i < char_len; ++i)
3777 d = re_string_byte_at (input, str_idx + i);
3778 if (d < 0x80 || d > 0xbf)
3784 char_len = re_string_char_size_at (input, str_idx);
3785 if (node->type == OP_PERIOD)
3789 /* FIXME: I don't think this if is needed, as both '\n'
3790 and '\0' are char_len == 1. */
3791 /* '.' accepts any one character except the following two cases. */
3792 if ((!(dfa->syntax & REG_DOT_NEWLINE) &&
3793 re_string_byte_at (input, str_idx) == '\n') ||
3794 ((dfa->syntax & REG_DOT_NOT_NULL) &&
3795 re_string_byte_at (input, str_idx) == '\0'))
3800 elem_len = re_string_elem_size_at (input, str_idx);
3801 if ((elem_len <= 1 && char_len <= 1) || char_len == 0)
3804 if (node->type == COMPLEX_BRACKET)
3806 const re_charset_t *cset = node->opr.mbcset;
3808 const unsigned char *pin
3809 = ((const unsigned char *) re_string_get_buffer (input) + str_idx);
3814 wchar_t wc = ((cset->nranges || cset->nchar_classes || cset->nmbchars)
3815 ? re_string_wchar_at (input, str_idx) : 0);
3817 /* match with multibyte character? */
3818 for (i = 0; i < cset->nmbchars; ++i)
3819 if (wc == cset->mbchars[i])
3821 match_len = char_len;
3822 goto check_node_accept_bytes_match;
3824 /* match with character_class? */
3825 for (i = 0; i < cset->nchar_classes; ++i)
3827 wctype_t wt = cset->char_classes[i];
3828 if (__iswctype (wc, wt))
3830 match_len = char_len;
3831 goto check_node_accept_bytes_match;
3836 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3839 unsigned int in_collseq = 0;
3840 const int32_t *table, *indirect;
3841 const unsigned char *weights, *extra;
3842 const char *collseqwc;
3844 /* This #include defines a local function! */
3845 # include <locale/weight.h>
3847 /* match with collating_symbol? */
3848 if (cset->ncoll_syms)
3849 extra = (const unsigned char *)
3850 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
3851 for (i = 0; i < cset->ncoll_syms; ++i)
3853 const unsigned char *coll_sym = extra + cset->coll_syms[i];
3854 /* Compare the length of input collating element and
3855 the length of current collating element. */
3856 if (*coll_sym != elem_len)
3858 /* Compare each bytes. */
3859 for (j = 0; j < *coll_sym; j++)
3860 if (pin[j] != coll_sym[1 + j])
3864 /* Match if every bytes is equal. */
3866 goto check_node_accept_bytes_match;
3872 if (elem_len <= char_len)
3874 collseqwc = _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQWC);
3875 in_collseq = __collseq_table_lookup (collseqwc, wc);
3878 in_collseq = find_collation_sequence_value (pin, elem_len);
3880 /* match with range expression? */
3881 for (i = 0; i < cset->nranges; ++i)
3882 if (cset->range_starts[i] <= in_collseq
3883 && in_collseq <= cset->range_ends[i])
3885 match_len = elem_len;
3886 goto check_node_accept_bytes_match;
3889 /* match with equivalence_class? */
3890 if (cset->nequiv_classes)
3892 const unsigned char *cp = pin;
3893 table = (const int32_t *)
3894 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
3895 weights = (const unsigned char *)
3896 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB);
3897 extra = (const unsigned char *)
3898 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
3899 indirect = (const int32_t *)
3900 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB);
3901 idx = findidx (&cp);
3903 for (i = 0; i < cset->nequiv_classes; ++i)
3905 int32_t equiv_class_idx = cset->equiv_classes[i];
3906 size_t weight_len = weights[idx];
3907 if (weight_len == weights[equiv_class_idx])
3910 while (cnt <= weight_len
3911 && (weights[equiv_class_idx + 1 + cnt]
3912 == weights[idx + 1 + cnt]))
3914 if (cnt > weight_len)
3916 match_len = elem_len;
3917 goto check_node_accept_bytes_match;
3926 /* match with range expression? */
3928 wchar_t cmp_buf[] = {L'\0', L'\0', wc, L'\0', L'\0', L'\0'};
3930 wchar_t cmp_buf[] = {L'\0', L'\0', L'\0', L'\0', L'\0', L'\0'};
3933 for (i = 0; i < cset->nranges; ++i)
3935 cmp_buf[0] = cset->range_starts[i];
3936 cmp_buf[4] = cset->range_ends[i];
3937 if (wcscoll (cmp_buf, cmp_buf + 2) <= 0
3938 && wcscoll (cmp_buf + 2, cmp_buf + 4) <= 0)
3940 match_len = char_len;
3941 goto check_node_accept_bytes_match;
3945 check_node_accept_bytes_match:
3946 if (!cset->non_match)
3953 return (elem_len > char_len) ? elem_len : char_len;
3961 find_collation_sequence_value (const unsigned char *mbs, size_t mbs_len)
3963 uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3968 /* No valid character. Match it as a single byte character. */
3969 const unsigned char *collseq = (const unsigned char *)
3970 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQMB);
3971 return collseq[mbs[0]];
3978 const unsigned char *extra = (const unsigned char *)
3979 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
3980 int32_t extrasize = (const unsigned char *)
3981 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB + 1) - extra;
3983 for (idx = 0; idx < extrasize;)
3987 int32_t elem_mbs_len;
3988 /* Skip the name of collating element name. */
3989 idx = idx + extra[idx] + 1;
3990 elem_mbs_len = extra[idx++];
3991 if (mbs_len == elem_mbs_len)
3993 for (mbs_cnt = 0; mbs_cnt < elem_mbs_len; ++mbs_cnt)
3994 if (extra[idx + mbs_cnt] != mbs[mbs_cnt])
3996 if (mbs_cnt == elem_mbs_len)
3997 /* Found the entry. */
4000 /* Skip the byte sequence of the collating element. */
4001 idx += elem_mbs_len;
4002 /* Adjust for the alignment. */
4003 idx = (idx + 3) & ~3;
4004 /* Skip the collation sequence value. */
4005 idx += sizeof (uint32_t);
4006 /* Skip the wide char sequence of the collating element. */
4007 idx = idx + sizeof (uint32_t) * (extra[idx] + 1);
4008 /* If we found the entry, return the sequence value. */
4010 return *(uint32_t *) (extra + idx);
4011 /* Skip the collation sequence value. */
4012 idx += sizeof (uint32_t);
4018 #endif /* RE_ENABLE_I18N */
4020 /* Check whether the node accepts the byte which is IDX-th
4021 byte of the INPUT. */
4025 check_node_accept (const re_match_context_t *mctx, const re_token_t *node,
4029 ch = re_string_byte_at (&mctx->input, idx);
4033 if (node->opr.c != ch)
4037 case SIMPLE_BRACKET:
4038 if (!bitset_contain (node->opr.sbcset, ch))
4042 #ifdef RE_ENABLE_I18N
4043 case OP_UTF8_PERIOD:
4049 if ((ch == '\n' && !(mctx->dfa->syntax & REG_DOT_NEWLINE))
4050 || (ch == '\0' && (mctx->dfa->syntax & REG_DOT_NOT_NULL)))
4058 if (node->constraint)
4060 /* The node has constraints. Check whether the current context
4061 satisfies the constraints. */
4062 unsigned int context = re_string_context_at (&mctx->input, idx,
4064 if (NOT_SATISFY_NEXT_CONSTRAINT (node->constraint, context))
4071 /* Extend the buffers, if the buffers have run out. */
4073 static reg_errcode_t
4075 extend_buffers (re_match_context_t *mctx)
4078 re_string_t *pstr = &mctx->input;
4080 /* Double the lengthes of the buffers. */
4081 ret = re_string_realloc_buffers (pstr, pstr->bufs_len * 2);
4082 if (BE (ret != REG_NOERROR, 0))
4085 if (mctx->state_log != NULL)
4087 /* And double the length of state_log. */
4088 /* XXX We have no indication of the size of this buffer. If this
4089 allocation fail we have no indication that the state_log array
4090 does not have the right size. */
4091 re_dfastate_t **new_array = re_xrealloc (mctx->state_log, re_dfastate_t *,
4092 pstr->bufs_len + 1);
4093 if (BE (new_array == NULL, 0))
4095 mctx->state_log = new_array;
4098 /* Then reconstruct the buffers. */
4101 #ifdef RE_ENABLE_I18N
4102 if (pstr->mb_cur_max > 1)
4104 ret = build_wcs_upper_buffer (pstr);
4105 if (BE (ret != REG_NOERROR, 0))
4109 #endif /* RE_ENABLE_I18N */
4110 build_upper_buffer (pstr);
4114 #ifdef RE_ENABLE_I18N
4115 if (pstr->mb_cur_max > 1)
4116 build_wcs_buffer (pstr);
4118 #endif /* RE_ENABLE_I18N */
4120 if (pstr->trans != NULL)
4121 re_string_translate_buffer (pstr);
4128 /* Functions for matching context. */
4130 /* Initialize MCTX. */
4132 static reg_errcode_t
4134 match_ctx_init (re_match_context_t *mctx, int eflags, Idx n)
4136 mctx->eflags = eflags;
4137 mctx->match_last = REG_MISSING;
4140 mctx->bkref_ents = re_xmalloc (struct re_backref_cache_entry, n);
4141 mctx->sub_tops = re_xmalloc (re_sub_match_top_t *, n);
4142 if (BE (mctx->bkref_ents == NULL || mctx->sub_tops == NULL, 0))
4145 /* Already zero-ed by the caller.
4147 mctx->bkref_ents = NULL;
4148 mctx->nbkref_ents = 0;
4149 mctx->nsub_tops = 0; */
4150 mctx->abkref_ents = n;
4151 mctx->max_mb_elem_len = 1;
4152 mctx->asub_tops = n;
4156 /* Clean the entries which depend on the current input in MCTX.
4157 This function must be invoked when the matcher changes the start index
4158 of the input, or changes the input string. */
4162 match_ctx_clean (re_match_context_t *mctx)
4165 for (st_idx = 0; st_idx < mctx->nsub_tops; ++st_idx)
4168 re_sub_match_top_t *top = mctx->sub_tops[st_idx];
4169 for (sl_idx = 0; sl_idx < top->nlasts; ++sl_idx)
4171 re_sub_match_last_t *last = top->lasts[sl_idx];
4172 re_free (last->path.array);
4175 re_free (top->lasts);
4178 re_free (top->path->array);
4179 re_free (top->path);
4184 mctx->nsub_tops = 0;
4185 mctx->nbkref_ents = 0;
4188 /* Free all the memory associated with MCTX. */
4192 match_ctx_free (re_match_context_t *mctx)
4194 /* First, free all the memory associated with MCTX->SUB_TOPS. */
4195 match_ctx_clean (mctx);
4196 re_free (mctx->sub_tops);
4197 re_free (mctx->bkref_ents);
4200 /* Add a new backreference entry to MCTX.
4201 Note that we assume that caller never call this function with duplicate
4202 entry, and call with STR_IDX which isn't smaller than any existing entry.
4205 static reg_errcode_t
4207 match_ctx_add_entry (re_match_context_t *mctx, Idx node, Idx str_idx,
4210 if (mctx->nbkref_ents >= mctx->abkref_ents)
4212 struct re_backref_cache_entry* new_entry;
4213 new_entry = re_x2realloc (mctx->bkref_ents, struct re_backref_cache_entry,
4214 &mctx->abkref_ents);
4215 if (BE (new_entry == NULL, 0))
4217 re_free (mctx->bkref_ents);
4220 mctx->bkref_ents = new_entry;
4221 memset (mctx->bkref_ents + mctx->nbkref_ents, '\0',
4222 (sizeof (struct re_backref_cache_entry)
4223 * (mctx->abkref_ents - mctx->nbkref_ents)));
4225 if (mctx->nbkref_ents > 0
4226 && mctx->bkref_ents[mctx->nbkref_ents - 1].str_idx == str_idx)
4227 mctx->bkref_ents[mctx->nbkref_ents - 1].more = 1;
4229 mctx->bkref_ents[mctx->nbkref_ents].node = node;
4230 mctx->bkref_ents[mctx->nbkref_ents].str_idx = str_idx;
4231 mctx->bkref_ents[mctx->nbkref_ents].subexp_from = from;
4232 mctx->bkref_ents[mctx->nbkref_ents].subexp_to = to;
4234 /* This is a cache that saves negative results of check_dst_limits_calc_pos.
4235 If bit N is clear, means that this entry won't epsilon-transition to
4236 an OP_OPEN_SUBEXP or OP_CLOSE_SUBEXP for the N+1-th subexpression. If
4237 it is set, check_dst_limits_calc_pos_1 will recurse and try to find one
4240 A backreference does not epsilon-transition unless it is empty, so set
4241 to all zeros if FROM != TO. */
4242 mctx->bkref_ents[mctx->nbkref_ents].eps_reachable_subexps_map
4243 = (from == to ? -1 : 0);
4245 mctx->bkref_ents[mctx->nbkref_ents++].more = 0;
4246 if (mctx->max_mb_elem_len < to - from)
4247 mctx->max_mb_elem_len = to - from;
4251 /* Return the first entry with the same str_idx, or REG_MISSING if none is
4252 found. Note that MCTX->BKREF_ENTS is already sorted by MCTX->STR_IDX. */
4256 search_cur_bkref_entry (const re_match_context_t *mctx, Idx str_idx)
4258 Idx left, right, mid, last;
4259 last = right = mctx->nbkref_ents;
4260 for (left = 0; left < right;)
4262 mid = (left + right) / 2;
4263 if (mctx->bkref_ents[mid].str_idx < str_idx)
4268 if (left < last && mctx->bkref_ents[left].str_idx == str_idx)
4274 /* Register the node NODE, whose type is OP_OPEN_SUBEXP, and which matches
4277 static reg_errcode_t
4279 match_ctx_add_subtop (re_match_context_t *mctx, Idx node, Idx str_idx)
4282 assert (mctx->sub_tops != NULL);
4283 assert (mctx->asub_tops > 0);
4285 if (BE (mctx->nsub_tops == mctx->asub_tops, 0))
4287 Idx new_asub_tops = mctx->asub_tops;
4288 re_sub_match_top_t **new_array = re_x2realloc (mctx->sub_tops,
4289 re_sub_match_top_t *,
4291 if (BE (new_array == NULL, 0))
4293 mctx->sub_tops = new_array;
4294 mctx->asub_tops = new_asub_tops;
4296 mctx->sub_tops[mctx->nsub_tops] = re_calloc (re_sub_match_top_t, 1);
4297 if (BE (mctx->sub_tops[mctx->nsub_tops] == NULL, 0))
4299 mctx->sub_tops[mctx->nsub_tops]->node = node;
4300 mctx->sub_tops[mctx->nsub_tops++]->str_idx = str_idx;
4304 /* Register the node NODE, whose type is OP_CLOSE_SUBEXP, and which matches
4305 at STR_IDX, whose corresponding OP_OPEN_SUBEXP is SUB_TOP. */
4307 static re_sub_match_last_t *
4309 match_ctx_add_sublast (re_sub_match_top_t *subtop, Idx node, Idx str_idx)
4311 re_sub_match_last_t *new_entry;
4312 if (BE (subtop->nlasts == subtop->alasts, 0))
4314 Idx new_alasts = subtop->alasts;
4315 re_sub_match_last_t **new_array = re_x2realloc (subtop->lasts,
4316 re_sub_match_last_t *,
4318 if (BE (new_array == NULL, 0))
4320 subtop->lasts = new_array;
4321 subtop->alasts = new_alasts;
4323 new_entry = re_calloc (re_sub_match_last_t, 1);
4324 if (BE (new_entry != NULL, 1))
4326 subtop->lasts[subtop->nlasts] = new_entry;
4327 new_entry->node = node;
4328 new_entry->str_idx = str_idx;
4336 sift_ctx_init (re_sift_context_t *sctx,
4337 re_dfastate_t **sifted_sts,
4338 re_dfastate_t **limited_sts,
4339 Idx last_node, Idx last_str_idx)
4341 sctx->sifted_states = sifted_sts;
4342 sctx->limited_states = limited_sts;
4343 sctx->last_node = last_node;
4344 sctx->last_str_idx = last_str_idx;
4345 re_node_set_init_empty (&sctx->limits);