1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
3 Free Software Foundation, Inc.
4 This file is part of the GNU C Library.
5 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 3, or (at your option)
10 any later version.
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License along
18 with this program; if not, write to the Free Software Foundation,
19 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
21 #ifndef _REGEX_INTERNAL_H
22 #define _REGEX_INTERNAL_H 1
24 #include <assert.h>
25 #include <ctype.h>
26 #include <stdbool.h>
27 #include <stdio.h>
28 #include <stdlib.h>
29 #include <string.h>
31 #ifdef _LIBC
32 # include <langinfo.h>
33 #else
34 # include "localcharset.h"
35 #endif
36 #if defined HAVE_LOCALE_H || defined _LIBC
37 # include <locale.h>
38 #endif
40 #include <wchar.h>
41 #include <wctype.h>
42 #include <stdint.h>
43 #if defined _LIBC
44 # include <bits/libc-lock.h>
45 #else
46 # define __libc_lock_init(NAME) do { } while (0)
47 # define __libc_lock_lock(NAME) do { } while (0)
48 # define __libc_lock_unlock(NAME) do { } while (0)
49 #endif
51 /* In case that the system doesn't have isblank(). */
52 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK))
53 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
54 #endif
56 #ifdef _LIBC
57 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
58 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
59 # include <locale/localeinfo.h>
60 # include <locale/elem-hash.h>
61 # include <locale/coll-lookup.h>
62 # endif
63 #endif
65 /* This is for other GNU distributions with internationalized messages. */
66 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
67 # include <libintl.h>
68 # ifdef _LIBC
69 # undef gettext
70 # define gettext(msgid) \
71 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
72 # endif
73 #else
74 # define gettext(msgid) (msgid)
75 #endif
77 #ifndef gettext_noop
78 /* This define is so xgettext can find the internationalizable
79 strings. */
80 # define gettext_noop(String) String
81 #endif
83 /* For loser systems without the definition. */
84 #ifndef SIZE_MAX
85 # define SIZE_MAX ((size_t) -1)
86 #endif
88 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCSCOLL) || _LIBC
89 # define RE_ENABLE_I18N
90 #endif
92 #if __GNUC__ >= 3
93 # define BE(expr, val) __builtin_expect (expr, val)
94 #else
95 # define BE(expr, val) (expr)
96 # ifdef _LIBC
97 # define inline
98 # endif
99 #endif
101 /* Number of ASCII characters. */
102 #define ASCII_CHARS 0x80
104 /* Number of single byte characters. */
105 #define SBC_MAX (UCHAR_MAX + 1)
107 #define COLL_ELEM_LEN_MAX 8
109 /* The character which represents newline. */
110 #define NEWLINE_CHAR '\n'
111 #define WIDE_NEWLINE_CHAR L'\n'
113 /* Rename to standard API for using out of glibc. */
114 #ifndef _LIBC
115 # define __wctype wctype
116 # define __iswctype iswctype
117 # define __btowc btowc
118 # define __wcrtomb wcrtomb
119 # define __mbrtowc mbrtowc
120 # define __regfree regfree
121 # define attribute_hidden
122 #endif /* not _LIBC */
124 #if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)
125 # define __attribute(arg) __attribute__ (arg)
126 #else
127 # define __attribute(arg)
128 #endif
130 typedef __re_idx_t Idx;
132 /* Special return value for failure to match. */
133 #define REG_MISSING ((Idx) -1)
135 /* Special return value for internal error. */
136 #define REG_ERROR ((Idx) -2)
138 /* Test whether N is a valid index, and is not one of the above. */
139 #ifdef _REGEX_LARGE_OFFSETS
140 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR)
141 #else
142 # define REG_VALID_INDEX(n) (0 <= (n))
143 #endif
145 /* Test whether N is a valid nonzero index. */
146 #ifdef _REGEX_LARGE_OFFSETS
147 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1))
148 #else
149 # define REG_VALID_NONZERO_INDEX(n) (0 < (n))
150 #endif
152 /* A hash value, suitable for computing hash tables. */
153 typedef __re_size_t re_hashval_t;
155 /* An integer used to represent a set of bits. It must be unsigned,
156 and must be at least as wide as unsigned int. */
157 typedef unsigned long int bitset_word_t;
158 /* All bits set in a bitset_word_t. */
159 #define BITSET_WORD_MAX ULONG_MAX
161 /* Number of bits in a bitset_word_t. For portability to hosts with
162 padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)';
163 instead, deduce it directly from BITSET_WORD_MAX. Avoid
164 greater-than-32-bit integers and unconditional shifts by more than
165 31 bits, as they're not portable. */
166 #if BITSET_WORD_MAX == 0xffffffffUL
167 # define BITSET_WORD_BITS 32
168 #elif BITSET_WORD_MAX >> 31 >> 4 == 1
169 # define BITSET_WORD_BITS 36
170 #elif BITSET_WORD_MAX >> 31 >> 16 == 1
171 # define BITSET_WORD_BITS 48
172 #elif BITSET_WORD_MAX >> 31 >> 28 == 1
173 # define BITSET_WORD_BITS 60
174 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1
175 # define BITSET_WORD_BITS 64
176 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1
177 # define BITSET_WORD_BITS 72
178 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1
179 # define BITSET_WORD_BITS 128
180 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1
181 # define BITSET_WORD_BITS 256
182 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1
183 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */
184 # if BITSET_WORD_BITS <= SBC_MAX
185 # error "Invalid SBC_MAX"
186 # endif
187 #elif BITSET_WORD_MAX == (0xffffffff + 2) * 0xffffffff
188 /* Work around a bug in 64-bit PGC (before version 6.1-2), where the
189 preprocessor mishandles large unsigned values as if they were signed. */
190 # define BITSET_WORD_BITS 64
191 #else
192 # error "Add case for new bitset_word_t size"
193 #endif
195 /* Number of bitset_word_t values in a bitset_t. */
196 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS)
198 typedef bitset_word_t bitset_t[BITSET_WORDS];
199 typedef bitset_word_t *re_bitset_ptr_t;
200 typedef const bitset_word_t *re_const_bitset_ptr_t;
202 #define PREV_WORD_CONSTRAINT 0x0001
203 #define PREV_NOTWORD_CONSTRAINT 0x0002
204 #define NEXT_WORD_CONSTRAINT 0x0004
205 #define NEXT_NOTWORD_CONSTRAINT 0x0008
206 #define PREV_NEWLINE_CONSTRAINT 0x0010
207 #define NEXT_NEWLINE_CONSTRAINT 0x0020
208 #define PREV_BEGBUF_CONSTRAINT 0x0040
209 #define NEXT_ENDBUF_CONSTRAINT 0x0080
210 #define WORD_DELIM_CONSTRAINT 0x0100
211 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
213 typedef enum
214 {
215 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
216 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
217 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
218 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
219 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
220 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
221 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
222 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
223 WORD_DELIM = WORD_DELIM_CONSTRAINT,
224 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
225 } re_context_type;
227 typedef struct
228 {
229 Idx alloc;
230 Idx nelem;
231 Idx *elems;
232 } re_node_set;
234 typedef enum
235 {
236 NON_TYPE = 0,
238 /* Node type, These are used by token, node, tree. */
239 CHARACTER = 1,
240 END_OF_RE = 2,
241 SIMPLE_BRACKET = 3,
242 OP_BACK_REF = 4,
243 OP_PERIOD = 5,
244 #ifdef RE_ENABLE_I18N
245 COMPLEX_BRACKET = 6,
246 OP_UTF8_PERIOD = 7,
247 #endif /* RE_ENABLE_I18N */
249 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
250 when the debugger shows values of this enum type. */
251 #define EPSILON_BIT 8
252 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
253 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
254 OP_ALT = EPSILON_BIT | 2,
255 OP_DUP_ASTERISK = EPSILON_BIT | 3,
256 ANCHOR = EPSILON_BIT | 4,
258 /* Tree type, these are used only by tree. */
259 CONCAT = 16,
260 SUBEXP = 17,
262 /* Token type, these are used only by token. */
263 OP_DUP_PLUS = 18,
264 OP_DUP_QUESTION,
265 OP_OPEN_BRACKET,
266 OP_CLOSE_BRACKET,
267 OP_CHARSET_RANGE,
268 OP_OPEN_DUP_NUM,
269 OP_CLOSE_DUP_NUM,
270 OP_NON_MATCH_LIST,
271 OP_OPEN_COLL_ELEM,
272 OP_CLOSE_COLL_ELEM,
273 OP_OPEN_EQUIV_CLASS,
274 OP_CLOSE_EQUIV_CLASS,
275 OP_OPEN_CHAR_CLASS,
276 OP_CLOSE_CHAR_CLASS,
277 OP_WORD,
278 OP_NOTWORD,
279 OP_SPACE,
280 OP_NOTSPACE,
281 BACK_SLASH
283 } re_token_type_t;
285 #ifdef RE_ENABLE_I18N
286 typedef struct
287 {
288 /* Multibyte characters. */
289 wchar_t *mbchars;
291 /* Collating symbols. */
292 # ifdef _LIBC
293 int32_t *coll_syms;
294 # endif
296 /* Equivalence classes. */
297 # ifdef _LIBC
298 int32_t *equiv_classes;
299 # endif
301 /* Range expressions. */
302 # ifdef _LIBC
303 uint32_t *range_starts;
304 uint32_t *range_ends;
305 # else /* not _LIBC */
306 wchar_t *range_starts;
307 wchar_t *range_ends;
308 # endif /* not _LIBC */
310 /* Character classes. */
311 wctype_t *char_classes;
313 /* If this character set is the non-matching list. */
314 unsigned int non_match : 1;
316 /* # of multibyte characters. */
317 Idx nmbchars;
319 /* # of collating symbols. */
320 Idx ncoll_syms;
322 /* # of equivalence classes. */
323 Idx nequiv_classes;
325 /* # of range expressions. */
326 Idx nranges;
328 /* # of character classes. */
329 Idx nchar_classes;
330 } re_charset_t;
331 #endif /* RE_ENABLE_I18N */
333 typedef struct
334 {
335 union
336 {
337 unsigned char c; /* for CHARACTER */
338 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
339 #ifdef RE_ENABLE_I18N
340 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
341 #endif /* RE_ENABLE_I18N */
342 Idx idx; /* for BACK_REF */
343 re_context_type ctx_type; /* for ANCHOR */
344 } opr;
345 #if __GNUC__ >= 2 && !__STRICT_ANSI__
346 re_token_type_t type : 8;
347 #else
348 re_token_type_t type;
349 #endif
350 unsigned int constraint : 10; /* context constraint */
351 unsigned int duplicated : 1;
352 unsigned int opt_subexp : 1;
353 #ifdef RE_ENABLE_I18N
354 unsigned int accept_mb : 1;
355 /* These 2 bits can be moved into the union if needed (e.g. if running out
356 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
357 unsigned int mb_partial : 1;
358 #endif
359 unsigned int word_char : 1;
360 } re_token_t;
362 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
364 struct re_string_t
365 {
366 /* Indicate the raw buffer which is the original string passed as an
367 argument of regexec(), re_search(), etc.. */
368 const unsigned char *raw_mbs;
369 /* Store the multibyte string. In case of "case insensitive mode" like
370 REG_ICASE, upper cases of the string are stored, otherwise MBS points
371 the same address that RAW_MBS points. */
372 unsigned char *mbs;
373 #ifdef RE_ENABLE_I18N
374 /* Store the wide character string which is corresponding to MBS. */
375 wint_t *wcs;
376 Idx *offsets;
377 mbstate_t cur_state;
378 #endif
379 /* Index in RAW_MBS. Each character mbs[i] corresponds to
380 raw_mbs[raw_mbs_idx + i]. */
381 Idx raw_mbs_idx;
382 /* The length of the valid characters in the buffers. */
383 Idx valid_len;
384 /* The corresponding number of bytes in raw_mbs array. */
385 Idx valid_raw_len;
386 /* The length of the buffers MBS and WCS. */
387 Idx bufs_len;
388 /* The index in MBS, which is updated by re_string_fetch_byte. */
389 Idx cur_idx;
390 /* length of RAW_MBS array. */
391 Idx raw_len;
392 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
393 Idx len;
394 /* End of the buffer may be shorter than its length in the cases such
395 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
396 instead of LEN. */
397 Idx raw_stop;
398 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
399 Idx stop;
401 /* The context of mbs[0]. We store the context independently, since
402 the context of mbs[0] may be different from raw_mbs[0], which is
403 the beginning of the input string. */
404 unsigned int tip_context;
405 /* The translation passed as a part of an argument of re_compile_pattern. */
406 RE_TRANSLATE_TYPE trans;
407 /* Copy of re_dfa_t's word_char. */
408 re_const_bitset_ptr_t word_char;
409 /* true if REG_ICASE. */
410 unsigned char icase;
411 unsigned char is_utf8;
412 unsigned char map_notascii;
413 unsigned char mbs_allocated;
414 unsigned char offsets_needed;
415 unsigned char newline_anchor;
416 unsigned char word_ops_used;
417 int mb_cur_max;
418 };
419 typedef struct re_string_t re_string_t;
422 struct re_dfa_t;
423 typedef struct re_dfa_t re_dfa_t;
425 #ifndef _LIBC
426 # if defined __i386__ && !defined __EMX__
427 # define internal_function __attribute ((regparm (3), stdcall))
428 # else
429 # define internal_function
430 # endif
431 #endif
433 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
434 Idx new_buf_len)
435 internal_function;
436 #ifdef RE_ENABLE_I18N
437 static void build_wcs_buffer (re_string_t *pstr) internal_function;
438 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
439 internal_function;
440 #endif /* RE_ENABLE_I18N */
441 static void build_upper_buffer (re_string_t *pstr) internal_function;
442 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
443 static unsigned int re_string_context_at (const re_string_t *input, Idx idx,
444 int eflags)
445 internal_function __attribute ((pure));
446 #define re_string_peek_byte(pstr, offset) \
447 ((pstr)->mbs[(pstr)->cur_idx + offset])
448 #define re_string_fetch_byte(pstr) \
449 ((pstr)->mbs[(pstr)->cur_idx++])
450 #define re_string_first_byte(pstr, idx) \
451 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
452 #define re_string_is_single_byte_char(pstr, idx) \
453 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
454 || (pstr)->wcs[(idx) + 1] != WEOF))
455 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
456 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
457 #define re_string_get_buffer(pstr) ((pstr)->mbs)
458 #define re_string_length(pstr) ((pstr)->len)
459 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
460 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
461 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
463 #include <alloca.h>
465 #ifndef _LIBC
466 # if HAVE_ALLOCA
467 /* The OS usually guarantees only one guard page at the bottom of the stack,
468 and a page size can be as small as 4096 bytes. So we cannot safely
469 allocate anything larger than 4096 bytes. Also care for the possibility
470 of a few compiler-allocated temporary stack slots. */
471 # define __libc_use_alloca(n) ((n) < 4032)
472 # else
473 /* alloca is implemented with malloc, so just use malloc. */
474 # define __libc_use_alloca(n) 0
475 # endif
476 #endif
478 #ifndef MAX
479 # define MAX(a,b) ((a) < (b) ? (b) : (a))
480 #endif
482 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
483 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
484 #define re_free(p) free (p)
486 struct bin_tree_t
487 {
488 struct bin_tree_t *parent;
489 struct bin_tree_t *left;
490 struct bin_tree_t *right;
491 struct bin_tree_t *first;
492 struct bin_tree_t *next;
494 re_token_t token;
496 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
497 Otherwise `type' indicate the type of this node. */
498 Idx node_idx;
499 };
500 typedef struct bin_tree_t bin_tree_t;
502 #define BIN_TREE_STORAGE_SIZE \
503 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
505 struct bin_tree_storage_t
506 {
507 struct bin_tree_storage_t *next;
508 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
509 };
510 typedef struct bin_tree_storage_t bin_tree_storage_t;
512 #define CONTEXT_WORD 1
513 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
514 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
515 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
517 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
518 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
519 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
520 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
521 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
523 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
524 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
525 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
526 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
528 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
529 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
530 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
531 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
532 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
534 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
535 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
536 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
537 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
538 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
540 struct re_dfastate_t
541 {
542 re_hashval_t hash;
543 re_node_set nodes;
544 re_node_set non_eps_nodes;
545 re_node_set inveclosure;
546 re_node_set *entrance_nodes;
547 struct re_dfastate_t **trtable, **word_trtable;
548 unsigned int context : 4;
549 unsigned int halt : 1;
550 /* If this state can accept `multi byte'.
551 Note that we refer to multibyte characters, and multi character
552 collating elements as `multi byte'. */
553 unsigned int accept_mb : 1;
554 /* If this state has backreference node(s). */
555 unsigned int has_backref : 1;
556 unsigned int has_constraint : 1;
557 };
558 typedef struct re_dfastate_t re_dfastate_t;
560 struct re_state_table_entry
561 {
562 Idx num;
563 Idx alloc;
564 re_dfastate_t **array;
565 };
567 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
569 typedef struct
570 {
571 Idx next_idx;
572 Idx alloc;
573 re_dfastate_t **array;
574 } state_array_t;
576 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
578 typedef struct
579 {
580 Idx node;
581 Idx str_idx; /* The position NODE match at. */
582 state_array_t path;
583 } re_sub_match_last_t;
585 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
586 And information about the node, whose type is OP_CLOSE_SUBEXP,
587 corresponding to NODE is stored in LASTS. */
589 typedef struct
590 {
591 Idx str_idx;
592 Idx node;
593 state_array_t *path;
594 Idx alasts; /* Allocation size of LASTS. */
595 Idx nlasts; /* The number of LASTS. */
596 re_sub_match_last_t **lasts;
597 } re_sub_match_top_t;
599 struct re_backref_cache_entry
600 {
601 Idx node;
602 Idx str_idx;
603 Idx subexp_from;
604 Idx subexp_to;
605 char more;
606 char unused;
607 unsigned short int eps_reachable_subexps_map;
608 };
610 typedef struct
611 {
612 /* The string object corresponding to the input string. */
613 re_string_t input;
614 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
615 const re_dfa_t *const dfa;
616 #else
617 const re_dfa_t *dfa;
618 #endif
619 /* EFLAGS of the argument of regexec. */
620 int eflags;
621 /* Where the matching ends. */
622 Idx match_last;
623 Idx last_node;
624 /* The state log used by the matcher. */
625 re_dfastate_t **state_log;
626 Idx state_log_top;
627 /* Back reference cache. */
628 Idx nbkref_ents;
629 Idx abkref_ents;
630 struct re_backref_cache_entry *bkref_ents;
631 int max_mb_elem_len;
632 Idx nsub_tops;
633 Idx asub_tops;
634 re_sub_match_top_t **sub_tops;
635 } re_match_context_t;
637 typedef struct
638 {
639 re_dfastate_t **sifted_states;
640 re_dfastate_t **limited_states;
641 Idx last_node;
642 Idx last_str_idx;
643 re_node_set limits;
644 } re_sift_context_t;
646 struct re_fail_stack_ent_t
647 {
648 Idx idx;
649 Idx node;
650 regmatch_t *regs;
651 re_node_set eps_via_nodes;
652 };
654 struct re_fail_stack_t
655 {
656 Idx num;
657 Idx alloc;
658 struct re_fail_stack_ent_t *stack;
659 };
661 struct re_dfa_t
662 {
663 re_token_t *nodes;
664 size_t nodes_alloc;
665 size_t nodes_len;
666 Idx *nexts;
667 Idx *org_indices;
668 re_node_set *edests;
669 re_node_set *eclosures;
670 re_node_set *inveclosures;
671 struct re_state_table_entry *state_table;
672 re_dfastate_t *init_state;
673 re_dfastate_t *init_state_word;
674 re_dfastate_t *init_state_nl;
675 re_dfastate_t *init_state_begbuf;
676 bin_tree_t *str_tree;
677 bin_tree_storage_t *str_tree_storage;
678 re_bitset_ptr_t sb_char;
679 int str_tree_storage_idx;
681 /* number of subexpressions `re_nsub' is in regex_t. */
682 re_hashval_t state_hash_mask;
683 Idx init_node;
684 Idx nbackref; /* The number of backreference in this dfa. */
686 /* Bitmap expressing which backreference is used. */
687 bitset_word_t used_bkref_map;
688 bitset_word_t completed_bkref_map;
690 unsigned int has_plural_match : 1;
691 /* If this dfa has "multibyte node", which is a backreference or
692 a node which can accept multibyte character or multi character
693 collating element. */
694 unsigned int has_mb_node : 1;
695 unsigned int is_utf8 : 1;
696 unsigned int map_notascii : 1;
697 unsigned int word_ops_used : 1;
698 int mb_cur_max;
699 bitset_t word_char;
700 reg_syntax_t syntax;
701 Idx *subexp_map;
702 #ifdef DEBUG
703 char* re_str;
704 #endif
705 #ifdef _LIBC
706 __libc_lock_define (, lock)
707 #endif
708 };
710 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
711 #define re_node_set_remove(set,id) \
712 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
713 #define re_node_set_empty(p) ((p)->nelem = 0)
714 #define re_node_set_free(set) re_free ((set)->elems)
715 \f
717 typedef enum
718 {
719 SB_CHAR,
720 MB_CHAR,
721 EQUIV_CLASS,
722 COLL_SYM,
723 CHAR_CLASS
724 } bracket_elem_type;
726 typedef struct
727 {
728 bracket_elem_type type;
729 union
730 {
731 unsigned char ch;
732 unsigned char *name;
733 wchar_t wch;
734 } opr;
735 } bracket_elem_t;
738 /* Inline functions for bitset_t operation. */
740 static inline void
741 bitset_set (bitset_t set, Idx i)
742 {
743 set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS;
744 }
746 static inline void
747 bitset_clear (bitset_t set, Idx i)
748 {
749 set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS);
750 }
752 static inline bool
753 bitset_contain (const bitset_t set, Idx i)
754 {
755 return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1;
756 }
758 static inline void
759 bitset_empty (bitset_t set)
760 {
761 memset (set, '\0', sizeof (bitset_t));
762 }
764 static inline void
765 bitset_set_all (bitset_t set)
766 {
767 memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS));
768 if (SBC_MAX % BITSET_WORD_BITS != 0)
769 set[BITSET_WORDS - 1] =
770 ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1;
771 }
773 static inline void
774 bitset_copy (bitset_t dest, const bitset_t src)
775 {
776 memcpy (dest, src, sizeof (bitset_t));
777 }
779 static inline void
780 bitset_not (bitset_t set)
781 {
782 int bitset_i;
783 for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i)
784 set[bitset_i] = ~set[bitset_i];
785 if (SBC_MAX % BITSET_WORD_BITS != 0)
786 set[BITSET_WORDS - 1] =
787 ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1)
788 & ~set[BITSET_WORDS - 1]);
789 }
791 static inline void
792 bitset_merge (bitset_t dest, const bitset_t src)
793 {
794 int bitset_i;
795 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
796 dest[bitset_i] |= src[bitset_i];
797 }
799 static inline void
800 bitset_mask (bitset_t dest, const bitset_t src)
801 {
802 int bitset_i;
803 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
804 dest[bitset_i] &= src[bitset_i];
805 }
807 #ifdef RE_ENABLE_I18N
808 /* Inline functions for re_string. */
809 static inline int
810 internal_function __attribute ((pure))
811 re_string_char_size_at (const re_string_t *pstr, Idx idx)
812 {
813 int byte_idx;
814 if (pstr->mb_cur_max == 1)
815 return 1;
816 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
817 if (pstr->wcs[idx + byte_idx] != WEOF)
818 break;
819 return byte_idx;
820 }
822 static inline wint_t
823 internal_function __attribute ((pure))
824 re_string_wchar_at (const re_string_t *pstr, Idx idx)
825 {
826 if (pstr->mb_cur_max == 1)
827 return (wint_t) pstr->mbs[idx];
828 return (wint_t) pstr->wcs[idx];
829 }
831 static int
832 internal_function __attribute ((pure))
833 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
834 {
835 # ifdef _LIBC
836 const unsigned char *p, *extra;
837 const int32_t *table, *indirect;
838 int32_t tmp;
839 # include <locale/weight.h>
840 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
842 if (nrules != 0)
843 {
844 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
845 extra = (const unsigned char *)
846 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
847 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
848 _NL_COLLATE_INDIRECTMB);
849 p = pstr->mbs + idx;
850 tmp = findidx (&p);
851 return p - pstr->mbs - idx;
852 }
853 else
854 # endif /* _LIBC */
855 return 1;
856 }
857 #endif /* RE_ENABLE_I18N */
859 #endif /* _REGEX_INTERNAL_H */