1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002-2005, 2007, 2008 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
5
6 The GNU C Library is free software; you can redistribute it and/or
7 modify it under the terms of the GNU Lesser General Public
8 License as published by the Free Software Foundation; either
9 version 2.1 of the License, or (at your option) any later version.
10
11 The GNU C Library 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 GNU
14 Lesser General Public License for more details.
15
16 You should have received a copy of the GNU Lesser General Public
17 License along with the GNU C Library; if not, write to the Free
18 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
19 02111-1307 USA. */
20
21 #ifndef _REGEX_INTERNAL_H
22 #define _REGEX_INTERNAL_H 1
23
24 #include <assert.h>
25 #include <ctype.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29
30 #if defined HAVE_LANGINFO_H || defined HAVE_LANGINFO_CODESET || defined _LIBC
31 # include <langinfo.h>
32 #endif
33 #if defined HAVE_LOCALE_H || defined _LIBC
34 # include <locale.h>
35 #endif
36 #if defined HAVE_WCHAR_H || defined _LIBC
37 # include <wchar.h>
38 #endif /* HAVE_WCHAR_H || _LIBC */
39 #if defined HAVE_WCTYPE_H || defined _LIBC
40 # include <wctype.h>
41 #endif /* HAVE_WCTYPE_H || _LIBC */
42 #if defined HAVE_STDBOOL_H || defined _LIBC
43 # include <stdbool.h>
44 #endif /* HAVE_STDBOOL_H || _LIBC */
45 #if defined HAVE_STDINT_H || defined _LIBC
46 # include <stdint.h>
47 #endif /* HAVE_STDINT_H || _LIBC */
48 #if defined _LIBC
49 # include <bits/libc-lock.h>
50 #else
51 # define __libc_lock_define(CLASS,NAME)
52 # define __libc_lock_init(NAME) do { } while (0)
53 # define __libc_lock_lock(NAME) do { } while (0)
54 # define __libc_lock_unlock(NAME) do { } while (0)
55 #endif
56
57 /* In case that the system doesn't have isblank(). */
58 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
59 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
60 #endif
61
62 #ifdef _LIBC
63 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
64 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
65 # include <locale/localeinfo.h>
66 # include <locale/elem-hash.h>
67 # include <locale/coll-lookup.h>
68 # endif
69 #endif
70
71 /* This is for other GNU distributions with internationalized messages. */
72 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
73 # include <libintl.h>
74 # ifdef _LIBC
75 # undef gettext
76 # define gettext(msgid) \
77 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
78 # endif
79 #else
80 # define gettext(msgid) (msgid)
81 #endif
82
83 #ifndef gettext_noop
84 /* This define is so xgettext can find the internationalizable
85 strings. */
86 # define gettext_noop(String) String
87 #endif
88
89 /* For loser systems without the definition. */
90 #ifndef SIZE_MAX
91 # define SIZE_MAX ((size_t) -1)
92 #endif
93
94 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
95 # define RE_ENABLE_I18N
96 #endif
97
98 #if __GNUC__ >= 3
99 # define BE(expr, val) __builtin_expect (expr, val)
100 #else
101 # define BE(expr, val) (expr)
102 # define inline
103 #endif
104
105 /* Number of single byte character. */
106 #define SBC_MAX 256
107
108 #define COLL_ELEM_LEN_MAX 8
109
110 /* The character which represents newline. */
111 #define NEWLINE_CHAR '\n'
112 #define WIDE_NEWLINE_CHAR L'\n'
113
114 /* Rename to standard API for using out of glibc. */
115 #ifndef _LIBC
116 # define __wctype wctype
117 # define __iswctype iswctype
118 # define __btowc btowc
119 # define __mbrtowc mbrtowc
120 # define __mempcpy mempcpy
121 # define __wcrtomb wcrtomb
122 # define __regfree regfree
123 # define attribute_hidden
124 #endif /* not _LIBC */
125
126 #ifdef __GNUC__
127 # define __attribute(arg) __attribute__ (arg)
128 #else
129 # define __attribute(arg)
130 #endif
131
132 extern const char __re_error_msgid[] attribute_hidden;
133 extern const size_t __re_error_msgid_idx[] attribute_hidden;
134
135 /* An integer used to represent a set of bits. It must be unsigned,
136 and must be at least as wide as unsigned int. */
137 typedef unsigned long int bitset_word_t;
138 /* All bits set in a bitset_word_t. */
139 #define BITSET_WORD_MAX ULONG_MAX
140 /* Number of bits in a bitset_word_t. */
141 #define BITSET_WORD_BITS (sizeof (bitset_word_t) * CHAR_BIT)
142 /* Number of bitset_word_t in a bit_set. */
143 #define BITSET_WORDS (SBC_MAX / BITSET_WORD_BITS)
144 typedef bitset_word_t bitset_t[BITSET_WORDS];
145 typedef bitset_word_t *re_bitset_ptr_t;
146 typedef const bitset_word_t *re_const_bitset_ptr_t;
147
148 #define bitset_set(set,i) \
149 (set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS)
150 #define bitset_clear(set,i) \
151 (set[i / BITSET_WORD_BITS] &= ~((bitset_word_t) 1 << i % BITSET_WORD_BITS))
152 #define bitset_contain(set,i) \
153 (set[i / BITSET_WORD_BITS] & ((bitset_word_t) 1 << i % BITSET_WORD_BITS))
154 #define bitset_empty(set) memset (set, '\0', sizeof (bitset_t))
155 #define bitset_set_all(set) memset (set, '\xff', sizeof (bitset_t))
156 #define bitset_copy(dest,src) memcpy (dest, src, sizeof (bitset_t))
157
158 #define PREV_WORD_CONSTRAINT 0x0001
159 #define PREV_NOTWORD_CONSTRAINT 0x0002
160 #define NEXT_WORD_CONSTRAINT 0x0004
161 #define NEXT_NOTWORD_CONSTRAINT 0x0008
162 #define PREV_NEWLINE_CONSTRAINT 0x0010
163 #define NEXT_NEWLINE_CONSTRAINT 0x0020
164 #define PREV_BEGBUF_CONSTRAINT 0x0040
165 #define NEXT_ENDBUF_CONSTRAINT 0x0080
166 #define WORD_DELIM_CONSTRAINT 0x0100
167 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
168
169 typedef enum
170 {
171 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
172 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
173 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
174 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
175 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
176 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
177 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
178 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
179 WORD_DELIM = WORD_DELIM_CONSTRAINT,
180 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
181 } re_context_type;
182
183 typedef struct
184 {
185 int alloc;
186 int nelem;
187 int *elems;
188 } re_node_set;
189
190 typedef enum
191 {
192 NON_TYPE = 0,
193
194 /* Node type, These are used by token, node, tree. */
195 CHARACTER = 1,
196 END_OF_RE = 2,
197 SIMPLE_BRACKET = 3,
198 OP_BACK_REF = 4,
199 OP_PERIOD = 5,
200 #ifdef RE_ENABLE_I18N
201 COMPLEX_BRACKET = 6,
202 OP_UTF8_PERIOD = 7,
203 #endif /* RE_ENABLE_I18N */
204
205 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
206 when the debugger shows values of this enum type. */
207 #define EPSILON_BIT 8
208 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
209 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
210 OP_ALT = EPSILON_BIT | 2,
211 OP_DUP_ASTERISK = EPSILON_BIT | 3,
212 ANCHOR = EPSILON_BIT | 4,
213
214 /* Tree type, these are used only by tree. */
215 CONCAT = 16,
216 SUBEXP = 17,
217
218 /* Token type, these are used only by token. */
219 OP_DUP_PLUS = 18,
220 OP_DUP_QUESTION,
221 OP_OPEN_BRACKET,
222 OP_CLOSE_BRACKET,
223 OP_CHARSET_RANGE,
224 OP_OPEN_DUP_NUM,
225 OP_CLOSE_DUP_NUM,
226 OP_NON_MATCH_LIST,
227 OP_OPEN_COLL_ELEM,
228 OP_CLOSE_COLL_ELEM,
229 OP_OPEN_EQUIV_CLASS,
230 OP_CLOSE_EQUIV_CLASS,
231 OP_OPEN_CHAR_CLASS,
232 OP_CLOSE_CHAR_CLASS,
233 OP_WORD,
234 OP_NOTWORD,
235 OP_SPACE,
236 OP_NOTSPACE,
237 BACK_SLASH
238
239 } re_token_type_t;
240
241 #ifdef RE_ENABLE_I18N
242 typedef struct
243 {
244 /* Multibyte characters. */
245 wchar_t *mbchars;
246
247 /* Collating symbols. */
248 # ifdef _LIBC
249 int32_t *coll_syms;
250 # endif
251
252 /* Equivalence classes. */
253 # ifdef _LIBC
254 int32_t *equiv_classes;
255 # endif
256
257 /* Range expressions. */
258 # ifdef _LIBC
259 uint32_t *range_starts;
260 uint32_t *range_ends;
261 # else /* not _LIBC */
262 wchar_t *range_starts;
263 wchar_t *range_ends;
264 # endif /* not _LIBC */
265
266 /* Character classes. */
267 wctype_t *char_classes;
268
269 /* If this character set is the non-matching list. */
270 unsigned int non_match : 1;
271
272 /* # of multibyte characters. */
273 int nmbchars;
274
275 /* # of collating symbols. */
276 int ncoll_syms;
277
278 /* # of equivalence classes. */
279 int nequiv_classes;
280
281 /* # of range expressions. */
282 int nranges;
283
284 /* # of character classes. */
285 int nchar_classes;
286 } re_charset_t;
287 #endif /* RE_ENABLE_I18N */
288
289 typedef struct
290 {
291 union
292 {
293 unsigned char c; /* for CHARACTER */
294 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
295 #ifdef RE_ENABLE_I18N
296 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
297 #endif /* RE_ENABLE_I18N */
298 int idx; /* for BACK_REF */
299 re_context_type ctx_type; /* for ANCHOR */
300 } opr;
301 #if __GNUC__ >= 2
302 re_token_type_t type : 8;
303 #else
304 re_token_type_t type;
305 #endif
306 unsigned int constraint : 10; /* context constraint */
307 unsigned int duplicated : 1;
308 unsigned int opt_subexp : 1;
309 #ifdef RE_ENABLE_I18N
310 unsigned int accept_mb : 1;
311 /* These 2 bits can be moved into the union if needed (e.g. if running out
312 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
313 unsigned int mb_partial : 1;
314 #endif
315 unsigned int word_char : 1;
316 } re_token_t;
317
318 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
319
320 struct re_string_t
321 {
322 /* Indicate the raw buffer which is the original string passed as an
323 argument of regexec(), re_search(), etc.. */
324 const unsigned char *raw_mbs;
325 /* Store the multibyte string. In case of "case insensitive mode" like
326 REG_ICASE, upper cases of the string are stored, otherwise MBS points
327 the same address that RAW_MBS points. */
328 unsigned char *mbs;
329 #ifdef RE_ENABLE_I18N
330 /* Store the wide character string which is corresponding to MBS. */
331 wint_t *wcs;
332 int *offsets;
333 mbstate_t cur_state;
334 #endif
335 /* Index in RAW_MBS. Each character mbs[i] corresponds to
336 raw_mbs[raw_mbs_idx + i]. */
337 int raw_mbs_idx;
338 /* The length of the valid characters in the buffers. */
339 int valid_len;
340 /* The corresponding number of bytes in raw_mbs array. */
341 int valid_raw_len;
342 /* The length of the buffers MBS and WCS. */
343 int bufs_len;
344 /* The index in MBS, which is updated by re_string_fetch_byte. */
345 int cur_idx;
346 /* length of RAW_MBS array. */
347 int raw_len;
348 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
349 int len;
350 /* End of the buffer may be shorter than its length in the cases such
351 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
352 instead of LEN. */
353 int raw_stop;
354 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
355 int stop;
356
357 /* The context of mbs[0]. We store the context independently, since
358 the context of mbs[0] may be different from raw_mbs[0], which is
359 the beginning of the input string. */
360 unsigned int tip_context;
361 /* The translation passed as a part of an argument of re_compile_pattern. */
362 RE_TRANSLATE_TYPE trans;
363 /* Copy of re_dfa_t's word_char. */
364 re_const_bitset_ptr_t word_char;
365 /* 1 if REG_ICASE. */
366 unsigned char icase;
367 unsigned char is_utf8;
368 unsigned char map_notascii;
369 unsigned char mbs_allocated;
370 unsigned char offsets_needed;
371 unsigned char newline_anchor;
372 unsigned char word_ops_used;
373 int mb_cur_max;
374 };
375 typedef struct re_string_t re_string_t;
376
377
378 struct re_dfa_t;
379 typedef struct re_dfa_t re_dfa_t;
380
381 #ifndef _LIBC
382 # ifdef __i386__
383 # define internal_function __attribute ((regparm (3), stdcall))
384 # else
385 # define internal_function
386 # endif
387 #endif
388
389 #ifndef NOT_IN_libc
390 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
391 int new_buf_len)
392 internal_function;
393 # ifdef RE_ENABLE_I18N
394 static void build_wcs_buffer (re_string_t *pstr) internal_function;
395 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
396 internal_function;
397 # endif /* RE_ENABLE_I18N */
398 static void build_upper_buffer (re_string_t *pstr) internal_function;
399 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
400 static unsigned int re_string_context_at (const re_string_t *input, int idx,
401 int eflags)
402 internal_function __attribute ((pure));
403 #endif
404 #define re_string_peek_byte(pstr, offset) \
405 ((pstr)->mbs[(pstr)->cur_idx + offset])
406 #define re_string_fetch_byte(pstr) \
407 ((pstr)->mbs[(pstr)->cur_idx++])
408 #define re_string_first_byte(pstr, idx) \
409 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
410 #define re_string_is_single_byte_char(pstr, idx) \
411 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
412 || (pstr)->wcs[(idx) + 1] != WEOF))
413 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
414 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
415 #define re_string_get_buffer(pstr) ((pstr)->mbs)
416 #define re_string_length(pstr) ((pstr)->len)
417 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
418 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
419 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
420
421 #ifdef WIN32
422 # include <malloc.h>
423 #else
424 # include <alloca.h>
425 #endif
426
427 #ifndef _LIBC
428 # if HAVE_ALLOCA
429 /* The OS usually guarantees only one guard page at the bottom of the stack,
430 and a page size can be as small as 4096 bytes. So we cannot safely
431 allocate anything larger than 4096 bytes. Also care for the possibility
432 of a few compiler-allocated temporary stack slots. */
433 # define __libc_use_alloca(n) ((n) < 4032)
434 # else
435 /* alloca is implemented with malloc, so just use malloc. */
436 # define __libc_use_alloca(n) 0
437 # endif
438 #endif
439
440 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
441 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
442 #define re_free(p) free (p)
443
444 struct bin_tree_t
445 {
446 struct bin_tree_t *parent;
447 struct bin_tree_t *left;
448 struct bin_tree_t *right;
449 struct bin_tree_t *first;
450 struct bin_tree_t *next;
451
452 re_token_t token;
453
454 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
455 Otherwise `type' indicate the type of this node. */
456 int node_idx;
457 };
458 typedef struct bin_tree_t bin_tree_t;
459
460 #define BIN_TREE_STORAGE_SIZE \
461 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
462
463 struct bin_tree_storage_t
464 {
465 struct bin_tree_storage_t *next;
466 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
467 };
468 typedef struct bin_tree_storage_t bin_tree_storage_t;
469
470 #define CONTEXT_WORD 1
471 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
472 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
473 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
474
475 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
476 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
477 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
478 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
479 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
480
481 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
482 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
483 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
484 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
485
486 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
487 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
488 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
489 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
490 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
491
492 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
493 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
494 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
495 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
496 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
497
498 struct re_dfastate_t
499 {
500 unsigned int hash;
501 re_node_set nodes;
502 re_node_set non_eps_nodes;
503 re_node_set inveclosure;
504 re_node_set *entrance_nodes;
505 struct re_dfastate_t **trtable, **word_trtable;
506 unsigned int context : 4;
507 unsigned int halt : 1;
508 /* If this state can accept `multi byte'.
509 Note that we refer to multibyte characters, and multi character
510 collating elements as `multi byte'. */
511 unsigned int accept_mb : 1;
512 /* If this state has backreference node(s). */
513 unsigned int has_backref : 1;
514 unsigned int has_constraint : 1;
515 };
516 typedef struct re_dfastate_t re_dfastate_t;
517
518 struct re_state_table_entry
519 {
520 int num;
521 int alloc;
522 re_dfastate_t **array;
523 };
524
525 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
526
527 typedef struct
528 {
529 int next_idx;
530 int alloc;
531 re_dfastate_t **array;
532 } state_array_t;
533
534 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
535
536 typedef struct
537 {
538 int node;
539 int str_idx; /* The position NODE match at. */
540 state_array_t path;
541 } re_sub_match_last_t;
542
543 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
544 And information about the node, whose type is OP_CLOSE_SUBEXP,
545 corresponding to NODE is stored in LASTS. */
546
547 typedef struct
548 {
549 int str_idx;
550 int node;
551 state_array_t *path;
552 int alasts; /* Allocation size of LASTS. */
553 int nlasts; /* The number of LASTS. */
554 re_sub_match_last_t **lasts;
555 } re_sub_match_top_t;
556
557 struct re_backref_cache_entry
558 {
559 int node;
560 int str_idx;
561 int subexp_from;
562 int subexp_to;
563 char more;
564 char unused;
565 unsigned short int eps_reachable_subexps_map;
566 };
567
568 typedef struct
569 {
570 /* The string object corresponding to the input string. */
571 re_string_t input;
572 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
573 const re_dfa_t *const dfa;
574 #else
575 const re_dfa_t *dfa;
576 #endif
577 /* EFLAGS of the argument of regexec. */
578 int eflags;
579 /* Where the matching ends. */
580 int match_last;
581 int last_node;
582 /* The state log used by the matcher. */
583 re_dfastate_t **state_log;
584 int state_log_top;
585 /* Back reference cache. */
586 int nbkref_ents;
587 int abkref_ents;
588 struct re_backref_cache_entry *bkref_ents;
589 int max_mb_elem_len;
590 int nsub_tops;
591 int asub_tops;
592 re_sub_match_top_t **sub_tops;
593 } re_match_context_t;
594
595 typedef struct
596 {
597 re_dfastate_t **sifted_states;
598 re_dfastate_t **limited_states;
599 int last_node;
600 int last_str_idx;
601 re_node_set limits;
602 } re_sift_context_t;
603
604 struct re_fail_stack_ent_t
605 {
606 int idx;
607 int node;
608 regmatch_t *regs;
609 re_node_set eps_via_nodes;
610 };
611
612 struct re_fail_stack_t
613 {
614 int num;
615 int alloc;
616 struct re_fail_stack_ent_t *stack;
617 };
618
619 struct re_dfa_t
620 {
621 re_token_t *nodes;
622 size_t nodes_alloc;
623 size_t nodes_len;
624 int *nexts;
625 int *org_indices;
626 re_node_set *edests;
627 re_node_set *eclosures;
628 re_node_set *inveclosures;
629 struct re_state_table_entry *state_table;
630 re_dfastate_t *init_state;
631 re_dfastate_t *init_state_word;
632 re_dfastate_t *init_state_nl;
633 re_dfastate_t *init_state_begbuf;
634 bin_tree_t *str_tree;
635 bin_tree_storage_t *str_tree_storage;
636 re_bitset_ptr_t sb_char;
637 int str_tree_storage_idx;
638
639 /* number of subexpressions `re_nsub' is in regex_t. */
640 unsigned int state_hash_mask;
641 int init_node;
642 int nbackref; /* The number of backreference in this dfa. */
643
644 /* Bitmap expressing which backreference is used. */
645 bitset_word_t used_bkref_map;
646 bitset_word_t completed_bkref_map;
647
648 unsigned int has_plural_match : 1;
649 /* If this dfa has "multibyte node", which is a backreference or
650 a node which can accept multibyte character or multi character
651 collating element. */
652 unsigned int has_mb_node : 1;
653 unsigned int is_utf8 : 1;
654 unsigned int map_notascii : 1;
655 unsigned int word_ops_used : 1;
656 int mb_cur_max;
657 bitset_t word_char;
658 reg_syntax_t syntax;
659 int *subexp_map;
660 #ifdef DEBUG
661 char* re_str;
662 #endif
663 __libc_lock_define (, lock)
664 };
665
666 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
667 #define re_node_set_remove(set,id) \
668 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
669 #define re_node_set_empty(p) ((p)->nelem = 0)
670 #define re_node_set_free(set) re_free ((set)->elems)
671
672
673 typedef enum
674 {
675 SB_CHAR,
676 MB_CHAR,
677 EQUIV_CLASS,
678 COLL_SYM,
679 CHAR_CLASS
680 } bracket_elem_type;
681
682 typedef struct
683 {
684 bracket_elem_type type;
685 union
686 {
687 unsigned char ch;
688 unsigned char *name;
689 wchar_t wch;
690 } opr;
691 } bracket_elem_t;
692
693
694 /* Inline functions for bitset operation. */
695 static inline void
bitset_not(bitset_t set)696 bitset_not (bitset_t set)
697 {
698 int bitset_i;
699 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
700 set[bitset_i] = ~set[bitset_i];
701 }
702
703 static inline void
bitset_merge(bitset_t dest,const bitset_t src)704 bitset_merge (bitset_t dest, const bitset_t src)
705 {
706 int bitset_i;
707 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
708 dest[bitset_i] |= src[bitset_i];
709 }
710
711 static inline void
bitset_mask(bitset_t dest,const bitset_t src)712 bitset_mask (bitset_t dest, const bitset_t src)
713 {
714 int bitset_i;
715 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
716 dest[bitset_i] &= src[bitset_i];
717 }
718
719 #ifdef RE_ENABLE_I18N
720 /* Inline functions for re_string. */
721 static inline int
internal_function(pure)722 internal_function __attribute ((pure))
723 re_string_char_size_at (const re_string_t *pstr, int idx)
724 {
725 int byte_idx;
726 if (pstr->mb_cur_max == 1)
727 return 1;
728 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
729 if (pstr->wcs[idx + byte_idx] != WEOF)
730 break;
731 return byte_idx;
732 }
733
734 static inline wint_t
internal_function(pure)735 internal_function __attribute ((pure))
736 re_string_wchar_at (const re_string_t *pstr, int idx)
737 {
738 if (pstr->mb_cur_max == 1)
739 return (wint_t) pstr->mbs[idx];
740 return (wint_t) pstr->wcs[idx];
741 }
742
743 # ifndef NOT_IN_libc
744 static int
internal_function(pure)745 internal_function __attribute ((pure))
746 re_string_elem_size_at (const re_string_t *pstr, int idx)
747 {
748 # ifdef _LIBC
749 const unsigned char *p, *extra;
750 const int32_t *table, *indirect;
751 int32_t tmp;
752 # include <locale/weight.h>
753 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
754
755 if (nrules != 0)
756 {
757 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
758 extra = (const unsigned char *)
759 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
760 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
761 _NL_COLLATE_INDIRECTMB);
762 p = pstr->mbs + idx;
763 tmp = findidx (&p);
764 return p - pstr->mbs - idx;
765 }
766 else
767 # endif /* _LIBC */
768 return 1;
769 }
770 # endif
771 #endif /* RE_ENABLE_I18N */
772
773 #endif /* _REGEX_INTERNAL_H */
774