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)
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
40 #if defined HAVE_WCHAR_H || defined _LIBC
41 # include <wchar.h>
42 #endif /* HAVE_WCHAR_H || _LIBC */
43 #if defined HAVE_WCTYPE_H || defined _LIBC
44 # include <wctype.h>
45 #endif /* HAVE_WCTYPE_H || _LIBC */
46 #if defined _LIBC
47 # include <bits/libc-lock.h>
48 #else
49 # define __libc_lock_define(CLASS,NAME)
50 # define __libc_lock_init(NAME) do { } while (0)
51 # define __libc_lock_lock(NAME) do { } while (0)
52 # define __libc_lock_unlock(NAME) do { } while (0)
53 #endif
55 /* In case that the system doesn't have isblank(). */
56 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
57 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
58 #endif
60 #ifdef _LIBC
61 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
62 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
63 # include <locale/localeinfo.h>
64 # include <locale/elem-hash.h>
65 # include <locale/coll-lookup.h>
66 # endif
67 #endif
69 /* This is for other GNU distributions with internationalized messages. */
70 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
71 # include <libintl.h>
72 # ifdef _LIBC
73 # undef gettext
74 # define gettext(msgid) \
75 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
76 # endif
77 #else
78 # define gettext(msgid) (msgid)
79 #endif
81 #ifndef gettext_noop
82 /* This define is so xgettext can find the internationalizable
83 strings. */
84 # define gettext_noop(String) String
85 #endif
87 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
88 # define RE_ENABLE_I18N
89 #endif
91 #if __GNUC__ >= 3
92 # define BE(expr, val) __builtin_expect (expr, val)
93 #else
94 # define BE(expr, val) (expr)
95 #endif
97 /* Number of single byte character. */
98 #define SBC_MAX 256
100 #define COLL_ELEM_LEN_MAX 8
102 /* The character which represents newline. */
103 #define NEWLINE_CHAR '\n'
104 #define WIDE_NEWLINE_CHAR L'\n'
106 /* Rename to standard API for using out of glibc. */
107 #ifndef _LIBC
108 # define __wctype wctype
109 # define __iswctype iswctype
110 # define __btowc btowc
111 # ifndef __mempcpy
112 # define __mempcpy mempcpy
113 # endif
114 # define __wcrtomb wcrtomb
115 # define __regfree regfree
116 # define attribute_hidden
117 #endif /* not _LIBC */
119 #if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)
120 # define __attribute(arg) __attribute__ (arg)
121 #else
122 # define __attribute(arg)
123 #endif
125 extern const char __re_error_msgid[] attribute_hidden;
126 extern const size_t __re_error_msgid_idx[] attribute_hidden;
128 typedef __re_idx_t Idx;
130 /* Special return value for failure to match. */
131 #define REG_MISSING ((Idx) -1)
133 /* Special return value for internal error. */
134 #define REG_ERROR ((Idx) -2)
136 /* Test whether N is a valid index, and is not one of the above. */
137 #ifdef _REGEX_LARGE_OFFSETS
138 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR)
139 #else
140 # define REG_VALID_INDEX(n) (0 <= (n))
141 #endif
143 /* Test whether N is a valid nonzero index. */
144 #ifdef _REGEX_LARGE_OFFSETS
145 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1))
146 #else
147 # define REG_VALID_NONZERO_INDEX(n) (0 < (n))
148 #endif
150 /* A hash value, suitable for computing hash tables. */
151 typedef __re_size_t re_hashval_t;
153 /* An integer used to represent a set of bits. It must be unsigned,
154 and must be at least as wide as unsigned int. */
155 typedef unsigned long int bitset_word;
157 /* Maximum value of a bitset word. It must be useful in preprocessor
158 contexts, and must be consistent with bitset_word. */
159 #define BITSET_WORD_MAX ULONG_MAX
161 /* Number of bits in a bitset word. Avoid greater-than-32-bit
162 integers and unconditional shifts by more than 31 bits, as they're
163 not portable. */
164 #if BITSET_WORD_MAX == 0xffffffff
165 # define BITSET_WORD_BITS 32
166 #elif BITSET_WORD_MAX >> 31 >> 5 == 1
167 # define BITSET_WORD_BITS 36
168 #elif BITSET_WORD_MAX >> 31 >> 16 == 1
169 # define BITSET_WORD_BITS 48
170 #elif BITSET_WORD_MAX >> 31 >> 28 == 1
171 # define BITSET_WORD_BITS 60
172 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1
173 # define BITSET_WORD_BITS 64
174 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1
175 # define BITSET_WORD_BITS 72
176 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1
177 # define BITSET_WORD_BITS 128
178 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1
179 # define BITSET_WORD_BITS 256
180 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1
181 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */
182 # if BITSET_WORD_BITS <= SBC_MAX
183 # error "Invalid SBC_MAX"
184 # endif
185 #else
186 # error "Add case for new bitset_word size"
187 #endif
189 /* Number of bitset words in a bitset. */
190 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS)
192 typedef bitset_word bitset[BITSET_WORDS];
193 typedef bitset_word *re_bitset_ptr_t;
194 typedef const bitset_word *re_const_bitset_ptr_t;
196 #define PREV_WORD_CONSTRAINT 0x0001
197 #define PREV_NOTWORD_CONSTRAINT 0x0002
198 #define NEXT_WORD_CONSTRAINT 0x0004
199 #define NEXT_NOTWORD_CONSTRAINT 0x0008
200 #define PREV_NEWLINE_CONSTRAINT 0x0010
201 #define NEXT_NEWLINE_CONSTRAINT 0x0020
202 #define PREV_BEGBUF_CONSTRAINT 0x0040
203 #define NEXT_ENDBUF_CONSTRAINT 0x0080
204 #define WORD_DELIM_CONSTRAINT 0x0100
205 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
207 typedef enum
208 {
209 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
210 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
211 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
212 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
213 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
214 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
215 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
216 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
217 WORD_DELIM = WORD_DELIM_CONSTRAINT,
218 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
219 } re_context_type;
221 typedef struct
222 {
223 Idx alloc;
224 Idx nelem;
225 Idx *elems;
226 } re_node_set;
228 typedef enum
229 {
230 NON_TYPE = 0,
232 /* Node type, These are used by token, node, tree. */
233 CHARACTER = 1,
234 END_OF_RE = 2,
235 SIMPLE_BRACKET = 3,
236 OP_BACK_REF = 4,
237 OP_PERIOD = 5,
238 #ifdef RE_ENABLE_I18N
239 COMPLEX_BRACKET = 6,
240 OP_UTF8_PERIOD = 7,
241 #endif /* RE_ENABLE_I18N */
243 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
244 when the debugger shows values of this enum type. */
245 #define EPSILON_BIT 8
246 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
247 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
248 OP_ALT = EPSILON_BIT | 2,
249 OP_DUP_ASTERISK = EPSILON_BIT | 3,
250 ANCHOR = EPSILON_BIT | 4,
252 /* Tree type, these are used only by tree. */
253 CONCAT = 16,
254 SUBEXP = 17,
256 /* Token type, these are used only by token. */
257 OP_DUP_PLUS = 18,
258 OP_DUP_QUESTION,
259 OP_OPEN_BRACKET,
260 OP_CLOSE_BRACKET,
261 OP_CHARSET_RANGE,
262 OP_OPEN_DUP_NUM,
263 OP_CLOSE_DUP_NUM,
264 OP_NON_MATCH_LIST,
265 OP_OPEN_COLL_ELEM,
266 OP_CLOSE_COLL_ELEM,
267 OP_OPEN_EQUIV_CLASS,
268 OP_CLOSE_EQUIV_CLASS,
269 OP_OPEN_CHAR_CLASS,
270 OP_CLOSE_CHAR_CLASS,
271 OP_WORD,
272 OP_NOTWORD,
273 OP_SPACE,
274 OP_NOTSPACE,
275 BACK_SLASH
277 } re_token_type_t;
279 #ifdef RE_ENABLE_I18N
280 typedef struct
281 {
282 /* Multibyte characters. */
283 wchar_t *mbchars;
285 /* Collating symbols. */
286 # ifdef _LIBC
287 int32_t *coll_syms;
288 # endif
290 /* Equivalence classes. */
291 # ifdef _LIBC
292 int32_t *equiv_classes;
293 # endif
295 /* Range expressions. */
296 # ifdef _LIBC
297 uint32_t *range_starts;
298 uint32_t *range_ends;
299 # else /* not _LIBC */
300 wchar_t *range_starts;
301 wchar_t *range_ends;
302 # endif /* not _LIBC */
304 /* Character classes. */
305 wctype_t *char_classes;
307 /* If this character set is the non-matching list. */
308 unsigned int non_match : 1;
310 /* # of multibyte characters. */
311 Idx nmbchars;
313 /* # of collating symbols. */
314 Idx ncoll_syms;
316 /* # of equivalence classes. */
317 Idx nequiv_classes;
319 /* # of range expressions. */
320 Idx nranges;
322 /* # of character classes. */
323 Idx nchar_classes;
324 } re_charset_t;
325 #endif /* RE_ENABLE_I18N */
327 typedef struct
328 {
329 union
330 {
331 unsigned char c; /* for CHARACTER */
332 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
333 #ifdef RE_ENABLE_I18N
334 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
335 #endif /* RE_ENABLE_I18N */
336 Idx idx; /* for BACK_REF */
337 re_context_type ctx_type; /* for ANCHOR */
338 } opr;
339 #if __GNUC__ >= 2
340 re_token_type_t type : 8;
341 #else
342 re_token_type_t type;
343 #endif
344 unsigned int constraint : 10; /* context constraint */
345 unsigned int duplicated : 1;
346 unsigned int opt_subexp : 1;
347 #ifdef RE_ENABLE_I18N
348 unsigned int accept_mb : 1;
349 /* These 2 bits can be moved into the union if needed (e.g. if running out
350 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
351 unsigned int mb_partial : 1;
352 #endif
353 unsigned int word_char : 1;
354 } re_token_t;
356 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
358 struct re_string_t
359 {
360 /* Indicate the raw buffer which is the original string passed as an
361 argument of regexec(), re_search(), etc.. */
362 const unsigned char *raw_mbs;
363 /* Store the multibyte string. In case of "case insensitive mode" like
364 REG_ICASE, upper cases of the string are stored, otherwise MBS points
365 the same address that RAW_MBS points. */
366 unsigned char *mbs;
367 #ifdef RE_ENABLE_I18N
368 /* Store the wide character string which is corresponding to MBS. */
369 wint_t *wcs;
370 Idx *offsets;
371 mbstate_t cur_state;
372 #endif
373 /* Index in RAW_MBS. Each character mbs[i] corresponds to
374 raw_mbs[raw_mbs_idx + i]. */
375 Idx raw_mbs_idx;
376 /* The length of the valid characters in the buffers. */
377 Idx valid_len;
378 /* The corresponding number of bytes in raw_mbs array. */
379 Idx valid_raw_len;
380 /* The length of the buffers MBS and WCS. */
381 Idx bufs_len;
382 /* The index in MBS, which is updated by re_string_fetch_byte. */
383 Idx cur_idx;
384 /* length of RAW_MBS array. */
385 Idx raw_len;
386 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
387 Idx len;
388 /* End of the buffer may be shorter than its length in the cases such
389 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
390 instead of LEN. */
391 Idx raw_stop;
392 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
393 Idx stop;
395 /* The context of mbs[0]. We store the context independently, since
396 the context of mbs[0] may be different from raw_mbs[0], which is
397 the beginning of the input string. */
398 unsigned int tip_context;
399 /* The translation passed as a part of an argument of re_compile_pattern. */
400 unsigned REG_TRANSLATE_TYPE trans;
401 /* Copy of re_dfa_t's word_char. */
402 re_const_bitset_ptr_t word_char;
403 /* true if REG_ICASE. */
404 unsigned char icase;
405 unsigned char is_utf8;
406 unsigned char map_notascii;
407 unsigned char mbs_allocated;
408 unsigned char offsets_needed;
409 unsigned char newline_anchor;
410 unsigned char word_ops_used;
411 int mb_cur_max;
412 };
413 typedef struct re_string_t re_string_t;
416 struct re_dfa_t;
417 typedef struct re_dfa_t re_dfa_t;
419 #ifndef _LIBC
420 # ifdef __i386__
421 # define internal_function __attribute ((regparm (3), stdcall))
422 # else
423 # define internal_function
424 # endif
425 #endif
427 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
428 Idx new_buf_len)
429 internal_function;
430 #ifdef RE_ENABLE_I18N
431 static void build_wcs_buffer (re_string_t *pstr) internal_function;
432 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
433 internal_function;
434 #endif /* RE_ENABLE_I18N */
435 static void build_upper_buffer (re_string_t *pstr) internal_function;
436 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
437 static unsigned int re_string_context_at (const re_string_t *input,
438 Idx idx, int eflags)
439 internal_function __attribute ((pure));
441 #define re_string_peek_byte(pstr, offset) \
442 ((pstr)->mbs[(pstr)->cur_idx + offset])
443 #define re_string_fetch_byte(pstr) \
444 ((pstr)->mbs[(pstr)->cur_idx++])
445 #define re_string_first_byte(pstr, idx) \
446 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
447 #define re_string_is_single_byte_char(pstr, idx) \
448 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
449 || (pstr)->wcs[(idx) + 1] != WEOF))
450 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
451 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
452 #define re_string_get_buffer(pstr) ((pstr)->mbs)
453 #define re_string_length(pstr) ((pstr)->len)
454 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
455 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
456 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
458 #include <alloca.h>
460 #ifndef _LIBC
461 # if HAVE_ALLOCA
462 /* The OS usually guarantees only one guard page at the bottom of the stack,
463 and a page size can be as small as 4096 bytes. So we cannot safely
464 allocate anything larger than 4096 bytes. Also care for the possibility
465 of a few compiler-allocated temporary stack slots. */
466 # define __libc_use_alloca(n) ((n) < 4032)
467 # else
468 /* alloca is implemented with malloc, so just use malloc. */
469 # define __libc_use_alloca(n) 0
470 # endif
471 #endif
473 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
474 #define re_xmalloc(t,n) ((t *) re_xnmalloc (n, sizeof (t)))
475 #define re_calloc(t,n) ((t *) calloc (n, sizeof (t)))
476 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
477 #define re_xrealloc(p,t,n) ((t *) re_xnrealloc (p, n, sizeof (t)))
478 #define re_x2realloc(p,t,pn) ((t *) re_x2nrealloc (p, pn, sizeof (t)))
479 #define re_free(p) free (p)
481 #ifndef SIZE_MAX
482 # define SIZE_MAX ((size_t) -1)
483 #endif
485 /* Return true if an array of N objects, each of size S, cannot exist
486 due to size arithmetic overflow. S must be nonzero. */
487 static inline bool
488 re_alloc_oversized (size_t n, size_t s)
489 {
490 return BE (SIZE_MAX / s < n, 0);
491 }
493 /* Return true if an array of (2 * N + 1) objects, each of size S,
494 cannot exist due to size arithmetic overflow. S must be nonzero. */
495 static inline bool
496 re_x2alloc_oversized (size_t n, size_t s)
497 {
498 return BE ((SIZE_MAX / s - 1) / 2 < n, 0);
499 }
501 /* Allocate an array of N objects, each with S bytes of memory,
502 dynamically, with error checking. S must be nonzero. */
503 static inline void *
504 re_xnmalloc (size_t n, size_t s)
505 {
506 return re_alloc_oversized (n, s) ? NULL : malloc (n * s);
507 }
509 /* Change the size of an allocated block of memory P to an array of N
510 objects each of S bytes, with error checking. S must be nonzero. */
511 static inline void *
512 re_xnrealloc (void *p, size_t n, size_t s)
513 {
514 return re_alloc_oversized (n, s) ? NULL : realloc (p, n * s);
515 }
517 /* Reallocate a block of memory P to an array of (2 * (*PN) + 1)
518 objects each of S bytes, with error checking. S must be nonzero.
519 If the allocation is successful, set *PN to the new allocation
520 count and return the resulting pointer. Otherwise, return
521 NULL. */
522 static inline void *
523 re_x2nrealloc (void *p, size_t *pn, size_t s)
524 {
525 if (re_x2alloc_oversized (*pn, s))
526 return NULL;
527 else
528 {
529 /* Add 1 in case *PN is zero. */
530 size_t n1 = 2 * *pn + 1;
531 p = realloc (p, n1 * s);
532 if (BE (p != NULL, 1))
533 *pn = n1;
534 return p;
535 }
536 }
538 struct bin_tree_t
539 {
540 struct bin_tree_t *parent;
541 struct bin_tree_t *left;
542 struct bin_tree_t *right;
543 struct bin_tree_t *first;
544 struct bin_tree_t *next;
546 re_token_t token;
548 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
549 Otherwise `type' indicate the type of this node. */
550 Idx node_idx;
551 };
552 typedef struct bin_tree_t bin_tree_t;
554 #define BIN_TREE_STORAGE_SIZE \
555 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
557 struct bin_tree_storage_t
558 {
559 struct bin_tree_storage_t *next;
560 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
561 };
562 typedef struct bin_tree_storage_t bin_tree_storage_t;
564 #define CONTEXT_WORD 1
565 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
566 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
567 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
569 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
570 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
571 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
572 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
573 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
575 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
576 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
577 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
578 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
580 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
581 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
582 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
583 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
584 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
586 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
587 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
588 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
589 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
590 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
592 struct re_dfastate_t
593 {
594 re_hashval_t hash;
595 re_node_set nodes;
596 re_node_set non_eps_nodes;
597 re_node_set inveclosure;
598 re_node_set *entrance_nodes;
599 struct re_dfastate_t **trtable, **word_trtable;
600 unsigned int context : 4;
601 unsigned int halt : 1;
602 /* If this state can accept `multi byte'.
603 Note that we refer to multibyte characters, and multi character
604 collating elements as `multi byte'. */
605 unsigned int accept_mb : 1;
606 /* If this state has backreference node(s). */
607 unsigned int has_backref : 1;
608 unsigned int has_constraint : 1;
609 };
610 typedef struct re_dfastate_t re_dfastate_t;
612 struct re_state_table_entry
613 {
614 Idx num;
615 Idx alloc;
616 re_dfastate_t **array;
617 };
619 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
621 typedef struct
622 {
623 Idx next_idx;
624 Idx alloc;
625 re_dfastate_t **array;
626 } state_array_t;
628 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
630 typedef struct
631 {
632 Idx node;
633 Idx str_idx; /* The position NODE match at. */
634 state_array_t path;
635 } re_sub_match_last_t;
637 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
638 And information about the node, whose type is OP_CLOSE_SUBEXP,
639 corresponding to NODE is stored in LASTS. */
641 typedef struct
642 {
643 Idx str_idx;
644 Idx node;
645 state_array_t *path;
646 Idx alasts; /* Allocation size of LASTS. */
647 Idx nlasts; /* The number of LASTS. */
648 re_sub_match_last_t **lasts;
649 } re_sub_match_top_t;
651 struct re_backref_cache_entry
652 {
653 Idx node;
654 Idx str_idx;
655 Idx subexp_from;
656 Idx subexp_to;
657 char more;
658 char unused;
659 unsigned short int eps_reachable_subexps_map;
660 };
662 typedef struct
663 {
664 /* The string object corresponding to the input string. */
665 re_string_t input;
666 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
667 re_dfa_t *const dfa;
668 #else
669 re_dfa_t *dfa;
670 #endif
671 /* EFLAGS of the argument of regexec. */
672 int eflags;
673 /* Where the matching ends. */
674 Idx match_last;
675 Idx last_node;
676 /* The state log used by the matcher. */
677 re_dfastate_t **state_log;
678 Idx state_log_top;
679 /* Back reference cache. */
680 Idx nbkref_ents;
681 Idx abkref_ents;
682 struct re_backref_cache_entry *bkref_ents;
683 int max_mb_elem_len;
684 Idx nsub_tops;
685 Idx asub_tops;
686 re_sub_match_top_t **sub_tops;
687 } re_match_context_t;
689 typedef struct
690 {
691 re_dfastate_t **sifted_states;
692 re_dfastate_t **limited_states;
693 Idx last_node;
694 Idx last_str_idx;
695 re_node_set limits;
696 } re_sift_context_t;
698 struct re_fail_stack_ent_t
699 {
700 Idx idx;
701 Idx node;
702 regmatch_t *regs;
703 re_node_set eps_via_nodes;
704 };
706 struct re_fail_stack_t
707 {
708 Idx num;
709 Idx alloc;
710 struct re_fail_stack_ent_t *stack;
711 };
713 struct re_dfa_t
714 {
715 re_token_t *nodes;
716 Idx nodes_alloc;
717 Idx nodes_len;
718 Idx *nexts;
719 Idx *org_indices;
720 re_node_set *edests;
721 re_node_set *eclosures;
722 re_node_set *inveclosures;
723 struct re_state_table_entry *state_table;
724 re_dfastate_t *init_state;
725 re_dfastate_t *init_state_word;
726 re_dfastate_t *init_state_nl;
727 re_dfastate_t *init_state_begbuf;
728 bin_tree_t *str_tree;
729 bin_tree_storage_t *str_tree_storage;
730 re_bitset_ptr_t sb_char;
731 int str_tree_storage_idx;
733 /* number of subexpressions `re_nsub' is in regex_t. */
734 re_hashval_t state_hash_mask;
735 Idx init_node;
736 Idx nbackref; /* The number of backreference in this dfa. */
738 /* Bitmap expressing which backreference is used. */
739 bitset_word used_bkref_map;
740 bitset_word completed_bkref_map;
742 unsigned int has_plural_match : 1;
743 /* If this dfa has "multibyte node", which is a backreference or
744 a node which can accept multibyte character or multi character
745 collating element. */
746 unsigned int has_mb_node : 1;
747 unsigned int is_utf8 : 1;
748 unsigned int map_notascii : 1;
749 unsigned int word_ops_used : 1;
750 int mb_cur_max;
751 bitset word_char;
752 reg_syntax_t syntax;
753 Idx *subexp_map;
754 #ifdef DEBUG
755 char* re_str;
756 #endif
757 __libc_lock_define (, lock)
758 };
760 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
761 #define re_node_set_remove(set,id) \
762 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
763 #define re_node_set_empty(p) ((p)->nelem = 0)
764 #define re_node_set_free(set) re_free ((set)->elems)
766 static void free_state (re_dfastate_t *state) internal_function;
767 \f
769 typedef enum
770 {
771 SB_CHAR,
772 MB_CHAR,
773 EQUIV_CLASS,
774 COLL_SYM,
775 CHAR_CLASS
776 } bracket_elem_type;
778 typedef struct
779 {
780 bracket_elem_type type;
781 union
782 {
783 unsigned char ch;
784 unsigned char *name;
785 wchar_t wch;
786 } opr;
787 } bracket_elem_t;
790 /* Inline functions for bitset operation. */
792 static inline void
793 bitset_set (bitset set, Idx i)
794 {
795 set[i / BITSET_WORD_BITS] |= (bitset_word) 1 << i % BITSET_WORD_BITS;
796 }
798 static inline void
799 bitset_clear (bitset set, Idx i)
800 {
801 set[i / BITSET_WORD_BITS] &= ~ ((bitset_word) 1 << i % BITSET_WORD_BITS);
802 }
804 static inline bool
805 bitset_contain (const bitset set, Idx i)
806 {
807 return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1;
808 }
810 static inline void
811 bitset_empty (bitset set)
812 {
813 memset (set, 0, sizeof (bitset));
814 }
816 static inline void
817 bitset_set_all (bitset set)
818 {
819 memset (set, -1, sizeof (bitset_word) * (SBC_MAX / BITSET_WORD_BITS));
820 if (SBC_MAX % BITSET_WORD_BITS != 0)
821 set[BITSET_WORDS - 1] =
822 ((bitset_word) 1 << SBC_MAX % BITSET_WORD_BITS) - 1;
823 }
825 static inline void
826 bitset_copy (bitset dest, const bitset src)
827 {
828 memcpy (dest, src, sizeof (bitset));
829 }
831 static inline void
832 bitset_not (bitset set)
833 {
834 int i;
835 for (i = 0; i < SBC_MAX / BITSET_WORD_BITS; ++i)
836 set[i] = ~set[i];
837 if (SBC_MAX % BITSET_WORD_BITS != 0)
838 set[BITSET_WORDS - 1] =
839 ((((bitset_word) 1 << SBC_MAX % BITSET_WORD_BITS) - 1)
840 & ~set[BITSET_WORDS - 1]);
841 }
843 static inline void
844 bitset_merge (bitset dest, const bitset src)
845 {
846 int i;
847 for (i = 0; i < BITSET_WORDS; ++i)
848 dest[i] |= src[i];
849 }
851 static inline void
852 bitset_mask (bitset dest, const bitset src)
853 {
854 int i;
855 for (i = 0; i < BITSET_WORDS; ++i)
856 dest[i] &= src[i];
857 }
859 #if defined RE_ENABLE_I18N
860 /* Inline functions for re_string. */
861 static inline int
862 internal_function __attribute ((pure))
863 re_string_char_size_at (const re_string_t *pstr, Idx idx)
864 {
865 int byte_idx;
866 if (pstr->mb_cur_max == 1)
867 return 1;
868 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
869 if (pstr->wcs[idx + byte_idx] != WEOF)
870 break;
871 return byte_idx;
872 }
874 static inline wint_t
875 internal_function __attribute ((pure))
876 re_string_wchar_at (const re_string_t *pstr, Idx idx)
877 {
878 if (pstr->mb_cur_max == 1)
879 return (wint_t) pstr->mbs[idx];
880 return (wint_t) pstr->wcs[idx];
881 }
883 static int
884 internal_function __attribute ((pure))
885 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
886 {
887 #ifdef _LIBC
888 const unsigned char *p, *extra;
889 const int32_t *table, *indirect;
890 int32_t tmp;
891 # include <locale/weight.h>
892 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
894 if (nrules != 0)
895 {
896 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
897 extra = (const unsigned char *)
898 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
899 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
900 _NL_COLLATE_INDIRECTMB);
901 p = pstr->mbs + idx;
902 tmp = findidx (&p);
903 return p - pstr->mbs - idx;
904 }
905 else
906 #endif /* _LIBC */
907 return 1;
908 }
909 #endif /* RE_ENABLE_I18N */
911 #endif /* _REGEX_INTERNAL_H */