b0f7e657ce37bf7ccfe0bb135aec264c6a311dc8
1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007 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)
9 any later version.
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. */
20 #ifndef _REGEX_INTERNAL_H
21 #define _REGEX_INTERNAL_H 1
23 #include <assert.h>
24 #include <ctype.h>
25 #include <stdbool.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
30 #ifndef _LIBC
31 # include "strcase.h"
32 #endif
34 #if defined HAVE_LANGINFO_H || defined HAVE_LANGINFO_CODESET || defined _LIBC
35 # include <langinfo.h>
36 #endif
37 #if defined HAVE_LOCALE_H || defined _LIBC
38 # include <locale.h>
39 #endif
41 #include <wchar.h>
42 #include <wctype.h>
43 #include <stdint.h>
44 #if defined _LIBC
45 # include <bits/libc-lock.h>
46 #else
47 # define __libc_lock_init(NAME) do { } while (0)
48 # define __libc_lock_lock(NAME) do { } while (0)
49 # define __libc_lock_unlock(NAME) do { } while (0)
50 #endif
52 /* In case that the system doesn't have isblank(). */
53 #if !defined _LIBC && !HAVE_DECL_ISBLANK && !defined isblank
54 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
55 #endif
57 #ifdef _LIBC
58 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
59 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
60 # include <locale/localeinfo.h>
61 # include <locale/elem-hash.h>
62 # include <locale/coll-lookup.h>
63 # endif
64 #endif
66 /* This is for other GNU distributions with internationalized messages. */
67 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
68 # include <libintl.h>
69 # ifdef _LIBC
70 # undef gettext
71 # define gettext(msgid) \
72 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
73 # endif
74 #else
75 # define gettext(msgid) (msgid)
76 #endif
78 #ifndef gettext_noop
79 /* This define is so xgettext can find the internationalizable
80 strings. */
81 # define gettext_noop(String) String
82 #endif
84 /* For loser systems without the definition. */
85 #ifndef SIZE_MAX
86 # define SIZE_MAX ((size_t) -1)
87 #endif
89 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
90 # define RE_ENABLE_I18N
91 #endif
93 #if __GNUC__ >= 3
94 # define BE(expr, val) __builtin_expect (expr, val)
95 #else
96 # define BE(expr, val) (expr)
97 # ifdef _LIBC
98 # define inline
99 # endif
100 #endif
102 /* Number of ASCII characters. */
103 #define ASCII_CHARS 0x80
105 /* Number of single byte characters. */
106 #define SBC_MAX (UCHAR_MAX + 1)
108 #define COLL_ELEM_LEN_MAX 8
110 /* The character which represents newline. */
111 #define NEWLINE_CHAR '\n'
112 #define WIDE_NEWLINE_CHAR L'\n'
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 # ifndef __mempcpy
120 # define __mempcpy mempcpy
121 # endif
122 # define __wcrtomb wcrtomb
123 # define __regfree regfree
124 # define attribute_hidden
125 #endif /* not _LIBC */
127 #if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)
128 # define __attribute(arg) __attribute__ (arg)
129 #else
130 # define __attribute(arg)
131 #endif
133 typedef __re_idx_t Idx;
135 /* Special return value for failure to match. */
136 #define REG_MISSING ((Idx) -1)
138 /* Special return value for internal error. */
139 #define REG_ERROR ((Idx) -2)
141 /* Test whether N is a valid index, and is not one of the above. */
142 #ifdef _REGEX_LARGE_OFFSETS
143 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR)
144 #else
145 # define REG_VALID_INDEX(n) (0 <= (n))
146 #endif
148 /* Test whether N is a valid nonzero index. */
149 #ifdef _REGEX_LARGE_OFFSETS
150 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1))
151 #else
152 # define REG_VALID_NONZERO_INDEX(n) (0 < (n))
153 #endif
155 /* A hash value, suitable for computing hash tables. */
156 typedef __re_size_t re_hashval_t;
158 /* An integer used to represent a set of bits. It must be unsigned,
159 and must be at least as wide as unsigned int. */
160 typedef unsigned long int bitset_word_t;
161 /* All bits set in a bitset_word_t. */
162 #define BITSET_WORD_MAX ULONG_MAX
164 /* Number of bits in a bitset_word_t. For portability to hosts with
165 padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)';
166 instead, deduce it directly from BITSET_WORD_MAX. Avoid
167 greater-than-32-bit integers and unconditional shifts by more than
168 31 bits, as they're not portable. */
169 #if BITSET_WORD_MAX == 0xffffffff
170 # define BITSET_WORD_BITS 32
171 #elif BITSET_WORD_MAX >> 31 >> 5 == 1
172 # define BITSET_WORD_BITS 36
173 #elif BITSET_WORD_MAX >> 31 >> 16 == 1
174 # define BITSET_WORD_BITS 48
175 #elif BITSET_WORD_MAX >> 31 >> 28 == 1
176 # define BITSET_WORD_BITS 60
177 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1
178 # define BITSET_WORD_BITS 64
179 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1
180 # define BITSET_WORD_BITS 72
181 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1
182 # define BITSET_WORD_BITS 128
183 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1
184 # define BITSET_WORD_BITS 256
185 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1
186 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */
187 # if BITSET_WORD_BITS <= SBC_MAX
188 # error "Invalid SBC_MAX"
189 # endif
190 #elif BITSET_WORD_MAX == (0xffffffff + 2) * 0xffffffff
191 /* Work around a bug in 64-bit PGC (before version 6.1-2), where the
192 preprocessor mishandles large unsigned values as if they were signed. */
193 # define BITSET_WORD_BITS 64
194 #else
195 # error "Add case for new bitset_word_t size"
196 #endif
198 /* Number of bitset_word_t values in a bitset_t. */
199 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS)
201 typedef bitset_word_t bitset_t[BITSET_WORDS];
202 typedef bitset_word_t *re_bitset_ptr_t;
203 typedef const bitset_word_t *re_const_bitset_ptr_t;
205 #define PREV_WORD_CONSTRAINT 0x0001
206 #define PREV_NOTWORD_CONSTRAINT 0x0002
207 #define NEXT_WORD_CONSTRAINT 0x0004
208 #define NEXT_NOTWORD_CONSTRAINT 0x0008
209 #define PREV_NEWLINE_CONSTRAINT 0x0010
210 #define NEXT_NEWLINE_CONSTRAINT 0x0020
211 #define PREV_BEGBUF_CONSTRAINT 0x0040
212 #define NEXT_ENDBUF_CONSTRAINT 0x0080
213 #define WORD_DELIM_CONSTRAINT 0x0100
214 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
216 typedef enum
217 {
218 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
219 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
220 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
221 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
222 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
223 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
224 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
225 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
226 WORD_DELIM = WORD_DELIM_CONSTRAINT,
227 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
228 } re_context_type;
230 typedef struct
231 {
232 Idx alloc;
233 Idx nelem;
234 Idx *elems;
235 } re_node_set;
237 typedef enum
238 {
239 NON_TYPE = 0,
241 /* Node type, These are used by token, node, tree. */
242 CHARACTER = 1,
243 END_OF_RE = 2,
244 SIMPLE_BRACKET = 3,
245 OP_BACK_REF = 4,
246 OP_PERIOD = 5,
247 #ifdef RE_ENABLE_I18N
248 COMPLEX_BRACKET = 6,
249 OP_UTF8_PERIOD = 7,
250 #endif /* RE_ENABLE_I18N */
252 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
253 when the debugger shows values of this enum type. */
254 #define EPSILON_BIT 8
255 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
256 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
257 OP_ALT = EPSILON_BIT | 2,
258 OP_DUP_ASTERISK = EPSILON_BIT | 3,
259 ANCHOR = EPSILON_BIT | 4,
261 /* Tree type, these are used only by tree. */
262 CONCAT = 16,
263 SUBEXP = 17,
265 /* Token type, these are used only by token. */
266 OP_DUP_PLUS = 18,
267 OP_DUP_QUESTION,
268 OP_OPEN_BRACKET,
269 OP_CLOSE_BRACKET,
270 OP_CHARSET_RANGE,
271 OP_OPEN_DUP_NUM,
272 OP_CLOSE_DUP_NUM,
273 OP_NON_MATCH_LIST,
274 OP_OPEN_COLL_ELEM,
275 OP_CLOSE_COLL_ELEM,
276 OP_OPEN_EQUIV_CLASS,
277 OP_CLOSE_EQUIV_CLASS,
278 OP_OPEN_CHAR_CLASS,
279 OP_CLOSE_CHAR_CLASS,
280 OP_WORD,
281 OP_NOTWORD,
282 OP_SPACE,
283 OP_NOTSPACE,
284 BACK_SLASH
286 } re_token_type_t;
288 #ifdef RE_ENABLE_I18N
289 typedef struct
290 {
291 /* Multibyte characters. */
292 wchar_t *mbchars;
294 /* Collating symbols. */
295 # ifdef _LIBC
296 int32_t *coll_syms;
297 # endif
299 /* Equivalence classes. */
300 # ifdef _LIBC
301 int32_t *equiv_classes;
302 # endif
304 /* Range expressions. */
305 # ifdef _LIBC
306 uint32_t *range_starts;
307 uint32_t *range_ends;
308 # else /* not _LIBC */
309 wchar_t *range_starts;
310 wchar_t *range_ends;
311 # endif /* not _LIBC */
313 /* Character classes. */
314 wctype_t *char_classes;
316 /* If this character set is the non-matching list. */
317 unsigned int non_match : 1;
319 /* # of multibyte characters. */
320 Idx nmbchars;
322 /* # of collating symbols. */
323 Idx ncoll_syms;
325 /* # of equivalence classes. */
326 Idx nequiv_classes;
328 /* # of range expressions. */
329 Idx nranges;
331 /* # of character classes. */
332 Idx nchar_classes;
333 } re_charset_t;
334 #endif /* RE_ENABLE_I18N */
336 typedef struct
337 {
338 union
339 {
340 unsigned char c; /* for CHARACTER */
341 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
342 #ifdef RE_ENABLE_I18N
343 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
344 #endif /* RE_ENABLE_I18N */
345 Idx idx; /* for BACK_REF */
346 re_context_type ctx_type; /* for ANCHOR */
347 } opr;
348 #if __GNUC__ >= 2 && !__STRICT_ANSI__
349 re_token_type_t type : 8;
350 #else
351 re_token_type_t type;
352 #endif
353 unsigned int constraint : 10; /* context constraint */
354 unsigned int duplicated : 1;
355 unsigned int opt_subexp : 1;
356 #ifdef RE_ENABLE_I18N
357 unsigned int accept_mb : 1;
358 /* These 2 bits can be moved into the union if needed (e.g. if running out
359 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
360 unsigned int mb_partial : 1;
361 #endif
362 unsigned int word_char : 1;
363 } re_token_t;
365 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
367 struct re_string_t
368 {
369 /* Indicate the raw buffer which is the original string passed as an
370 argument of regexec(), re_search(), etc.. */
371 const unsigned char *raw_mbs;
372 /* Store the multibyte string. In case of "case insensitive mode" like
373 REG_ICASE, upper cases of the string are stored, otherwise MBS points
374 the same address that RAW_MBS points. */
375 unsigned char *mbs;
376 #ifdef RE_ENABLE_I18N
377 /* Store the wide character string which is corresponding to MBS. */
378 wint_t *wcs;
379 Idx *offsets;
380 mbstate_t cur_state;
381 #endif
382 /* Index in RAW_MBS. Each character mbs[i] corresponds to
383 raw_mbs[raw_mbs_idx + i]. */
384 Idx raw_mbs_idx;
385 /* The length of the valid characters in the buffers. */
386 Idx valid_len;
387 /* The corresponding number of bytes in raw_mbs array. */
388 Idx valid_raw_len;
389 /* The length of the buffers MBS and WCS. */
390 Idx bufs_len;
391 /* The index in MBS, which is updated by re_string_fetch_byte. */
392 Idx cur_idx;
393 /* length of RAW_MBS array. */
394 Idx raw_len;
395 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
396 Idx len;
397 /* End of the buffer may be shorter than its length in the cases such
398 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
399 instead of LEN. */
400 Idx raw_stop;
401 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
402 Idx stop;
404 /* The context of mbs[0]. We store the context independently, since
405 the context of mbs[0] may be different from raw_mbs[0], which is
406 the beginning of the input string. */
407 unsigned int tip_context;
408 /* The translation passed as a part of an argument of re_compile_pattern. */
409 RE_TRANSLATE_TYPE trans;
410 /* Copy of re_dfa_t's word_char. */
411 re_const_bitset_ptr_t word_char;
412 /* true if REG_ICASE. */
413 unsigned char icase;
414 unsigned char is_utf8;
415 unsigned char map_notascii;
416 unsigned char mbs_allocated;
417 unsigned char offsets_needed;
418 unsigned char newline_anchor;
419 unsigned char word_ops_used;
420 int mb_cur_max;
421 };
422 typedef struct re_string_t re_string_t;
425 struct re_dfa_t;
426 typedef struct re_dfa_t re_dfa_t;
428 #ifndef _LIBC
429 # ifdef __i386__
430 # define internal_function __attribute ((regparm (3), stdcall))
431 # else
432 # define internal_function
433 # endif
434 #endif
436 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
437 Idx new_buf_len)
438 internal_function;
439 #ifdef RE_ENABLE_I18N
440 static void build_wcs_buffer (re_string_t *pstr) internal_function;
441 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
442 internal_function;
443 #endif /* RE_ENABLE_I18N */
444 static void build_upper_buffer (re_string_t *pstr) internal_function;
445 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
446 static unsigned int re_string_context_at (const re_string_t *input, Idx idx,
447 int eflags)
448 internal_function __attribute ((pure));
449 #define re_string_peek_byte(pstr, offset) \
450 ((pstr)->mbs[(pstr)->cur_idx + offset])
451 #define re_string_fetch_byte(pstr) \
452 ((pstr)->mbs[(pstr)->cur_idx++])
453 #define re_string_first_byte(pstr, idx) \
454 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
455 #define re_string_is_single_byte_char(pstr, idx) \
456 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
457 || (pstr)->wcs[(idx) + 1] != WEOF))
458 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
459 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
460 #define re_string_get_buffer(pstr) ((pstr)->mbs)
461 #define re_string_length(pstr) ((pstr)->len)
462 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
463 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
464 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
466 #include <alloca.h>
468 #ifndef _LIBC
469 # if HAVE_ALLOCA
470 /* The OS usually guarantees only one guard page at the bottom of the stack,
471 and a page size can be as small as 4096 bytes. So we cannot safely
472 allocate anything larger than 4096 bytes. Also care for the possibility
473 of a few compiler-allocated temporary stack slots. */
474 # define __libc_use_alloca(n) ((n) < 4032)
475 # else
476 /* alloca is implemented with malloc, so just use malloc. */
477 # define __libc_use_alloca(n) 0
478 # endif
479 #endif
481 #ifndef MAX
482 # define MAX(a,b) ((a) < (b) ? (b) : (a))
483 #endif
485 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
486 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
487 #define re_free(p) free (p)
489 struct bin_tree_t
490 {
491 struct bin_tree_t *parent;
492 struct bin_tree_t *left;
493 struct bin_tree_t *right;
494 struct bin_tree_t *first;
495 struct bin_tree_t *next;
497 re_token_t token;
499 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
500 Otherwise `type' indicate the type of this node. */
501 Idx node_idx;
502 };
503 typedef struct bin_tree_t bin_tree_t;
505 #define BIN_TREE_STORAGE_SIZE \
506 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
508 struct bin_tree_storage_t
509 {
510 struct bin_tree_storage_t *next;
511 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
512 };
513 typedef struct bin_tree_storage_t bin_tree_storage_t;
515 #define CONTEXT_WORD 1
516 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
517 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
518 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
520 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
521 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
522 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
523 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
524 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
526 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
527 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
528 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
529 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
531 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
532 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
533 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
534 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
535 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
537 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
538 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
539 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
540 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
541 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
543 struct re_dfastate_t
544 {
545 re_hashval_t hash;
546 re_node_set nodes;
547 re_node_set non_eps_nodes;
548 re_node_set inveclosure;
549 re_node_set *entrance_nodes;
550 struct re_dfastate_t **trtable, **word_trtable;
551 unsigned int context : 4;
552 unsigned int halt : 1;
553 /* If this state can accept `multi byte'.
554 Note that we refer to multibyte characters, and multi character
555 collating elements as `multi byte'. */
556 unsigned int accept_mb : 1;
557 /* If this state has backreference node(s). */
558 unsigned int has_backref : 1;
559 unsigned int has_constraint : 1;
560 };
561 typedef struct re_dfastate_t re_dfastate_t;
563 struct re_state_table_entry
564 {
565 Idx num;
566 Idx alloc;
567 re_dfastate_t **array;
568 };
570 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
572 typedef struct
573 {
574 Idx next_idx;
575 Idx alloc;
576 re_dfastate_t **array;
577 } state_array_t;
579 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
581 typedef struct
582 {
583 Idx node;
584 Idx str_idx; /* The position NODE match at. */
585 state_array_t path;
586 } re_sub_match_last_t;
588 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
589 And information about the node, whose type is OP_CLOSE_SUBEXP,
590 corresponding to NODE is stored in LASTS. */
592 typedef struct
593 {
594 Idx str_idx;
595 Idx node;
596 state_array_t *path;
597 Idx alasts; /* Allocation size of LASTS. */
598 Idx nlasts; /* The number of LASTS. */
599 re_sub_match_last_t **lasts;
600 } re_sub_match_top_t;
602 struct re_backref_cache_entry
603 {
604 Idx node;
605 Idx str_idx;
606 Idx subexp_from;
607 Idx subexp_to;
608 char more;
609 char unused;
610 unsigned short int eps_reachable_subexps_map;
611 };
613 typedef struct
614 {
615 /* The string object corresponding to the input string. */
616 re_string_t input;
617 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
618 const re_dfa_t *const dfa;
619 #else
620 const re_dfa_t *dfa;
621 #endif
622 /* EFLAGS of the argument of regexec. */
623 int eflags;
624 /* Where the matching ends. */
625 Idx match_last;
626 Idx last_node;
627 /* The state log used by the matcher. */
628 re_dfastate_t **state_log;
629 Idx state_log_top;
630 /* Back reference cache. */
631 Idx nbkref_ents;
632 Idx abkref_ents;
633 struct re_backref_cache_entry *bkref_ents;
634 int max_mb_elem_len;
635 Idx nsub_tops;
636 Idx asub_tops;
637 re_sub_match_top_t **sub_tops;
638 } re_match_context_t;
640 typedef struct
641 {
642 re_dfastate_t **sifted_states;
643 re_dfastate_t **limited_states;
644 Idx last_node;
645 Idx last_str_idx;
646 re_node_set limits;
647 } re_sift_context_t;
649 struct re_fail_stack_ent_t
650 {
651 Idx idx;
652 Idx node;
653 regmatch_t *regs;
654 re_node_set eps_via_nodes;
655 };
657 struct re_fail_stack_t
658 {
659 Idx num;
660 Idx alloc;
661 struct re_fail_stack_ent_t *stack;
662 };
664 struct re_dfa_t
665 {
666 re_token_t *nodes;
667 size_t nodes_alloc;
668 size_t nodes_len;
669 Idx *nexts;
670 Idx *org_indices;
671 re_node_set *edests;
672 re_node_set *eclosures;
673 re_node_set *inveclosures;
674 struct re_state_table_entry *state_table;
675 re_dfastate_t *init_state;
676 re_dfastate_t *init_state_word;
677 re_dfastate_t *init_state_nl;
678 re_dfastate_t *init_state_begbuf;
679 bin_tree_t *str_tree;
680 bin_tree_storage_t *str_tree_storage;
681 re_bitset_ptr_t sb_char;
682 int str_tree_storage_idx;
684 /* number of subexpressions `re_nsub' is in regex_t. */
685 re_hashval_t state_hash_mask;
686 Idx init_node;
687 Idx nbackref; /* The number of backreference in this dfa. */
689 /* Bitmap expressing which backreference is used. */
690 bitset_word_t used_bkref_map;
691 bitset_word_t completed_bkref_map;
693 unsigned int has_plural_match : 1;
694 /* If this dfa has "multibyte node", which is a backreference or
695 a node which can accept multibyte character or multi character
696 collating element. */
697 unsigned int has_mb_node : 1;
698 unsigned int is_utf8 : 1;
699 unsigned int map_notascii : 1;
700 unsigned int word_ops_used : 1;
701 int mb_cur_max;
702 bitset_t word_char;
703 reg_syntax_t syntax;
704 Idx *subexp_map;
705 #ifdef DEBUG
706 char* re_str;
707 #endif
708 #ifdef _LIBC
709 __libc_lock_define (, lock)
710 #endif
711 };
713 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
714 #define re_node_set_remove(set,id) \
715 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
716 #define re_node_set_empty(p) ((p)->nelem = 0)
717 #define re_node_set_free(set) re_free ((set)->elems)
718 \f
720 typedef enum
721 {
722 SB_CHAR,
723 MB_CHAR,
724 EQUIV_CLASS,
725 COLL_SYM,
726 CHAR_CLASS
727 } bracket_elem_type;
729 typedef struct
730 {
731 bracket_elem_type type;
732 union
733 {
734 unsigned char ch;
735 unsigned char *name;
736 wchar_t wch;
737 } opr;
738 } bracket_elem_t;
741 /* Inline functions for bitset_t operation. */
743 static inline void
744 bitset_set (bitset_t set, Idx i)
745 {
746 set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS;
747 }
749 static inline void
750 bitset_clear (bitset_t set, Idx i)
751 {
752 set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS);
753 }
755 static inline bool
756 bitset_contain (const bitset_t set, Idx i)
757 {
758 return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1;
759 }
761 static inline void
762 bitset_empty (bitset_t set)
763 {
764 memset (set, '\0', sizeof (bitset_t));
765 }
767 static inline void
768 bitset_set_all (bitset_t set)
769 {
770 memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS));
771 if (SBC_MAX % BITSET_WORD_BITS != 0)
772 set[BITSET_WORDS - 1] =
773 ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1;
774 }
776 static inline void
777 bitset_copy (bitset_t dest, const bitset_t src)
778 {
779 memcpy (dest, src, sizeof (bitset_t));
780 }
782 static inline void
783 bitset_not (bitset_t set)
784 {
785 int bitset_i;
786 for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i)
787 set[bitset_i] = ~set[bitset_i];
788 if (SBC_MAX % BITSET_WORD_BITS != 0)
789 set[BITSET_WORDS - 1] =
790 ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1)
791 & ~set[BITSET_WORDS - 1]);
792 }
794 static inline void
795 bitset_merge (bitset_t dest, const bitset_t src)
796 {
797 int bitset_i;
798 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
799 dest[bitset_i] |= src[bitset_i];
800 }
802 static inline void
803 bitset_mask (bitset_t dest, const bitset_t src)
804 {
805 int bitset_i;
806 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
807 dest[bitset_i] &= src[bitset_i];
808 }
810 #ifdef RE_ENABLE_I18N
811 /* Inline functions for re_string. */
812 static inline int
813 internal_function __attribute ((pure))
814 re_string_char_size_at (const re_string_t *pstr, Idx idx)
815 {
816 int byte_idx;
817 if (pstr->mb_cur_max == 1)
818 return 1;
819 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
820 if (pstr->wcs[idx + byte_idx] != WEOF)
821 break;
822 return byte_idx;
823 }
825 static inline wint_t
826 internal_function __attribute ((pure))
827 re_string_wchar_at (const re_string_t *pstr, Idx idx)
828 {
829 if (pstr->mb_cur_max == 1)
830 return (wint_t) pstr->mbs[idx];
831 return (wint_t) pstr->wcs[idx];
832 }
834 static int
835 internal_function __attribute ((pure))
836 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
837 {
838 # ifdef _LIBC
839 const unsigned char *p, *extra;
840 const int32_t *table, *indirect;
841 int32_t tmp;
842 # include <locale/weight.h>
843 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
845 if (nrules != 0)
846 {
847 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
848 extra = (const unsigned char *)
849 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
850 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
851 _NL_COLLATE_INDIRECTMB);
852 p = pstr->mbs + idx;
853 tmp = findidx (&p);
854 return p - pstr->mbs - idx;
855 }
856 else
857 # endif /* _LIBC */
858 return 1;
859 }
860 #endif /* RE_ENABLE_I18N */
862 #endif /* _REGEX_INTERNAL_H */