]> git.proxmox.com Git - mirror_smartmontools-debian.git/blob - regex/regex_internal.h
import smartmontools 7.0
[mirror_smartmontools-debian.git] / regex / regex_internal.h
1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002-2018 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, see
18 <https://www.gnu.org/licenses/>. */
19
20 #ifndef _REGEX_INTERNAL_H
21 #define _REGEX_INTERNAL_H 1
22
23 #include <assert.h>
24 #include <ctype.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28
29 #ifndef _REGEX_STANDALONE
30 #include <langinfo.h>
31 #endif
32 #include <locale.h>
33 #include <wchar.h>
34 #include <wctype.h>
35 #include <stdbool.h>
36 #include <stdint.h>
37
38 /* Properties of integers. Although Gnulib has intprops.h, glibc does
39 without for now. */
40 #if !defined(_LIBC) && !defined(_REGEX_STANDALONE)
41 # include "intprops.h"
42 #else
43 /* True if the real type T is signed. */
44 # define TYPE_SIGNED(t) (! ((t) 0 < (t) -1))
45
46 /* True if adding the nonnegative Idx values A and B would overflow.
47 If false, set *R to A + B. A, B, and R may be evaluated more than
48 once, or zero times. Although this is not a full implementation of
49 Gnulib INT_ADD_WRAPV, it is good enough for glibc regex code.
50 FIXME: This implementation is a fragile stopgap, and this file would
51 be simpler and more robust if intprops.h were migrated into glibc. */
52 # define INT_ADD_WRAPV(a, b, r) \
53 (IDX_MAX - (a) < (b) ? true : (*(r) = (a) + (b), false))
54 #endif
55
56 #ifdef _LIBC
57 # include <libc-lock.h>
58 # define lock_define(name) __libc_lock_define (, name)
59 # define lock_init(lock) (__libc_lock_init (lock), 0)
60 # define lock_fini(lock) ((void) 0)
61 # define lock_lock(lock) __libc_lock_lock (lock)
62 # define lock_unlock(lock) __libc_lock_unlock (lock)
63 #elif defined GNULIB_LOCK && !defined USE_UNLOCKED_IO
64 # include "glthread/lock.h"
65 /* Use gl_lock_define if empty macro arguments are known to work.
66 Otherwise, fall back on less-portable substitutes. */
67 # if ((defined __GNUC__ && !defined __STRICT_ANSI__) \
68 || (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__))
69 # define lock_define(name) gl_lock_define (, name)
70 # elif USE_POSIX_THREADS
71 # define lock_define(name) pthread_mutex_t name;
72 # elif USE_PTH_THREADS
73 # define lock_define(name) pth_mutex_t name;
74 # elif USE_SOLARIS_THREADS
75 # define lock_define(name) mutex_t name;
76 # elif USE_WINDOWS_THREADS
77 # define lock_define(name) gl_lock_t name;
78 # else
79 # define lock_define(name)
80 # endif
81 # define lock_init(lock) glthread_lock_init (&(lock))
82 # define lock_fini(lock) glthread_lock_destroy (&(lock))
83 # define lock_lock(lock) glthread_lock_lock (&(lock))
84 # define lock_unlock(lock) glthread_lock_unlock (&(lock))
85 #elif defined GNULIB_PTHREAD && !defined USE_UNLOCKED_IO
86 # include <pthread.h>
87 # define lock_define(name) pthread_mutex_t name;
88 # define lock_init(lock) pthread_mutex_init (&(lock), 0)
89 # define lock_fini(lock) pthread_mutex_destroy (&(lock))
90 # define lock_lock(lock) pthread_mutex_lock (&(lock))
91 # define lock_unlock(lock) pthread_mutex_unlock (&(lock))
92 #else
93 # define lock_define(name)
94 # define lock_init(lock) 0
95 # define lock_fini(lock) ((void) 0)
96 /* The 'dfa' avoids an "unused variable 'dfa'" warning from GCC. */
97 # define lock_lock(lock) ((void) dfa)
98 # define lock_unlock(lock) ((void) 0)
99 #endif
100
101 /* In case that the system doesn't have isblank(). */
102 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK))
103 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
104 #endif
105
106 #ifdef _LIBC
107 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
108 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
109 # include <locale/localeinfo.h>
110 # include <locale/coll-lookup.h>
111 # endif
112 #endif
113
114 /* This is for other GNU distributions with internationalized messages. */
115 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
116 # include <libintl.h>
117 # ifdef _LIBC
118 # undef gettext
119 # define gettext(msgid) \
120 __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES)
121 # endif
122 #else
123 # undef gettext
124 # define gettext(msgid) (msgid)
125 #endif
126
127 #ifndef gettext_noop
128 /* This define is so xgettext can find the internationalizable
129 strings. */
130 # define gettext_noop(String) String
131 #endif
132
133 #if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE) || _LIBC
134 # define RE_ENABLE_I18N
135 #endif
136
137 #ifdef __GNUC__
138 #define BE(expr, val) __builtin_expect (expr, val)
139 #else
140 #define BE(expr, val) (expr)
141 #endif
142
143 /* Number of ASCII characters. */
144 #define ASCII_CHARS 0x80
145
146 /* Number of single byte characters. */
147 #define SBC_MAX (UCHAR_MAX + 1)
148
149 #define COLL_ELEM_LEN_MAX 8
150
151 /* The character which represents newline. */
152 #define NEWLINE_CHAR '\n'
153 #define WIDE_NEWLINE_CHAR L'\n'
154
155 /* Rename to standard API for using out of glibc. */
156 #ifndef _LIBC
157 # undef __wctype
158 # undef __iswctype
159 # define __wctype wctype
160 # define __iswalnum iswalnum
161 # define __iswctype iswctype
162 # define __towlower towlower
163 # define __towupper towupper
164 # define __btowc btowc
165 # define __mbrtowc mbrtowc
166 # define __wcrtomb wcrtomb
167 # define __regfree regfree
168 # define attribute_hidden
169 #endif /* not _LIBC */
170
171 #if __GNUC__ < 3 + (__GNUC_MINOR__ < 1)
172 # define __attribute__(arg)
173 #endif
174
175 #ifndef SSIZE_MAX
176 # define SSIZE_MAX ((ssize_t) (SIZE_MAX / 2))
177 #endif
178
179 /* The type of indexes into strings. This is signed, not size_t,
180 since the API requires indexes to fit in regoff_t anyway, and using
181 signed integers makes the code a bit smaller and presumably faster.
182 The traditional GNU regex implementation uses int for indexes.
183 The POSIX-compatible implementation uses a possibly-wider type.
184 The name 'Idx' is three letters to minimize the hassle of
185 reindenting a lot of regex code that formerly used 'int'. */
186 typedef regoff_t Idx;
187 #ifdef _REGEX_LARGE_OFFSETS
188 # define IDX_MAX SSIZE_MAX
189 #else
190 # define IDX_MAX INT_MAX
191 #endif
192
193 /* A hash value, suitable for computing hash tables. */
194 typedef __re_size_t re_hashval_t;
195
196 /* An integer used to represent a set of bits. It must be unsigned,
197 and must be at least as wide as unsigned int. */
198 typedef unsigned long int bitset_word_t;
199 /* All bits set in a bitset_word_t. */
200 #define BITSET_WORD_MAX ULONG_MAX
201
202 /* Number of bits in a bitset_word_t. For portability to hosts with
203 padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)';
204 instead, deduce it directly from BITSET_WORD_MAX. Avoid
205 greater-than-32-bit integers and unconditional shifts by more than
206 31 bits, as they're not portable. */
207 #if BITSET_WORD_MAX == 0xffffffffUL
208 # define BITSET_WORD_BITS 32
209 #elif BITSET_WORD_MAX >> 31 >> 4 == 1
210 # define BITSET_WORD_BITS 36
211 #elif BITSET_WORD_MAX >> 31 >> 16 == 1
212 # define BITSET_WORD_BITS 48
213 #elif BITSET_WORD_MAX >> 31 >> 28 == 1
214 # define BITSET_WORD_BITS 60
215 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1
216 # define BITSET_WORD_BITS 64
217 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1
218 # define BITSET_WORD_BITS 72
219 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1
220 # define BITSET_WORD_BITS 128
221 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1
222 # define BITSET_WORD_BITS 256
223 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1
224 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */
225 # if BITSET_WORD_BITS <= SBC_MAX
226 # error "Invalid SBC_MAX"
227 # endif
228 #else
229 # error "Add case for new bitset_word_t size"
230 #endif
231
232 /* Number of bitset_word_t values in a bitset_t. */
233 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS)
234
235 typedef bitset_word_t bitset_t[BITSET_WORDS];
236 typedef bitset_word_t *re_bitset_ptr_t;
237 typedef const bitset_word_t *re_const_bitset_ptr_t;
238
239 #define PREV_WORD_CONSTRAINT 0x0001
240 #define PREV_NOTWORD_CONSTRAINT 0x0002
241 #define NEXT_WORD_CONSTRAINT 0x0004
242 #define NEXT_NOTWORD_CONSTRAINT 0x0008
243 #define PREV_NEWLINE_CONSTRAINT 0x0010
244 #define NEXT_NEWLINE_CONSTRAINT 0x0020
245 #define PREV_BEGBUF_CONSTRAINT 0x0040
246 #define NEXT_ENDBUF_CONSTRAINT 0x0080
247 #define WORD_DELIM_CONSTRAINT 0x0100
248 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
249
250 typedef enum
251 {
252 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
253 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
254 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
255 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
256 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
257 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
258 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
259 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
260 WORD_DELIM = WORD_DELIM_CONSTRAINT,
261 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
262 } re_context_type;
263
264 typedef struct
265 {
266 Idx alloc;
267 Idx nelem;
268 Idx *elems;
269 } re_node_set;
270
271 typedef enum
272 {
273 NON_TYPE = 0,
274
275 /* Node type, These are used by token, node, tree. */
276 CHARACTER = 1,
277 END_OF_RE = 2,
278 SIMPLE_BRACKET = 3,
279 OP_BACK_REF = 4,
280 OP_PERIOD = 5,
281 #ifdef RE_ENABLE_I18N
282 COMPLEX_BRACKET = 6,
283 OP_UTF8_PERIOD = 7,
284 #endif /* RE_ENABLE_I18N */
285
286 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
287 when the debugger shows values of this enum type. */
288 #define EPSILON_BIT 8
289 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
290 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
291 OP_ALT = EPSILON_BIT | 2,
292 OP_DUP_ASTERISK = EPSILON_BIT | 3,
293 ANCHOR = EPSILON_BIT | 4,
294
295 /* Tree type, these are used only by tree. */
296 CONCAT = 16,
297 SUBEXP = 17,
298
299 /* Token type, these are used only by token. */
300 OP_DUP_PLUS = 18,
301 OP_DUP_QUESTION,
302 OP_OPEN_BRACKET,
303 OP_CLOSE_BRACKET,
304 OP_CHARSET_RANGE,
305 OP_OPEN_DUP_NUM,
306 OP_CLOSE_DUP_NUM,
307 OP_NON_MATCH_LIST,
308 OP_OPEN_COLL_ELEM,
309 OP_CLOSE_COLL_ELEM,
310 OP_OPEN_EQUIV_CLASS,
311 OP_CLOSE_EQUIV_CLASS,
312 OP_OPEN_CHAR_CLASS,
313 OP_CLOSE_CHAR_CLASS,
314 OP_WORD,
315 OP_NOTWORD,
316 OP_SPACE,
317 OP_NOTSPACE,
318 BACK_SLASH
319
320 } re_token_type_t;
321
322 #ifdef RE_ENABLE_I18N
323 typedef struct
324 {
325 /* Multibyte characters. */
326 wchar_t *mbchars;
327
328 /* Collating symbols. */
329 # ifdef _LIBC
330 int32_t *coll_syms;
331 # endif
332
333 /* Equivalence classes. */
334 # ifdef _LIBC
335 int32_t *equiv_classes;
336 # endif
337
338 /* Range expressions. */
339 # ifdef _LIBC
340 uint32_t *range_starts;
341 uint32_t *range_ends;
342 # else /* not _LIBC */
343 wchar_t *range_starts;
344 wchar_t *range_ends;
345 # endif /* not _LIBC */
346
347 /* Character classes. */
348 wctype_t *char_classes;
349
350 /* If this character set is the non-matching list. */
351 unsigned int non_match : 1;
352
353 /* # of multibyte characters. */
354 Idx nmbchars;
355
356 /* # of collating symbols. */
357 Idx ncoll_syms;
358
359 /* # of equivalence classes. */
360 Idx nequiv_classes;
361
362 /* # of range expressions. */
363 Idx nranges;
364
365 /* # of character classes. */
366 Idx nchar_classes;
367 } re_charset_t;
368 #endif /* RE_ENABLE_I18N */
369
370 typedef struct
371 {
372 union
373 {
374 unsigned char c; /* for CHARACTER */
375 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
376 #ifdef RE_ENABLE_I18N
377 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
378 #endif /* RE_ENABLE_I18N */
379 Idx idx; /* for BACK_REF */
380 re_context_type ctx_type; /* for ANCHOR */
381 } opr;
382 #if __GNUC__ >= 2 && !defined __STRICT_ANSI__
383 re_token_type_t type : 8;
384 #else
385 re_token_type_t type;
386 #endif
387 unsigned int constraint : 10; /* context constraint */
388 unsigned int duplicated : 1;
389 unsigned int opt_subexp : 1;
390 #ifdef RE_ENABLE_I18N
391 unsigned int accept_mb : 1;
392 /* These 2 bits can be moved into the union if needed (e.g. if running out
393 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
394 unsigned int mb_partial : 1;
395 #endif
396 unsigned int word_char : 1;
397 } re_token_t;
398
399 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
400
401 struct re_string_t
402 {
403 /* Indicate the raw buffer which is the original string passed as an
404 argument of regexec(), re_search(), etc.. */
405 const unsigned char *raw_mbs;
406 /* Store the multibyte string. In case of "case insensitive mode" like
407 REG_ICASE, upper cases of the string are stored, otherwise MBS points
408 the same address that RAW_MBS points. */
409 unsigned char *mbs;
410 #ifdef RE_ENABLE_I18N
411 /* Store the wide character string which is corresponding to MBS. */
412 wint_t *wcs;
413 Idx *offsets;
414 mbstate_t cur_state;
415 #endif
416 /* Index in RAW_MBS. Each character mbs[i] corresponds to
417 raw_mbs[raw_mbs_idx + i]. */
418 Idx raw_mbs_idx;
419 /* The length of the valid characters in the buffers. */
420 Idx valid_len;
421 /* The corresponding number of bytes in raw_mbs array. */
422 Idx valid_raw_len;
423 /* The length of the buffers MBS and WCS. */
424 Idx bufs_len;
425 /* The index in MBS, which is updated by re_string_fetch_byte. */
426 Idx cur_idx;
427 /* length of RAW_MBS array. */
428 Idx raw_len;
429 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
430 Idx len;
431 /* End of the buffer may be shorter than its length in the cases such
432 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
433 instead of LEN. */
434 Idx raw_stop;
435 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
436 Idx stop;
437
438 /* The context of mbs[0]. We store the context independently, since
439 the context of mbs[0] may be different from raw_mbs[0], which is
440 the beginning of the input string. */
441 unsigned int tip_context;
442 /* The translation passed as a part of an argument of re_compile_pattern. */
443 RE_TRANSLATE_TYPE trans;
444 /* Copy of re_dfa_t's word_char. */
445 re_const_bitset_ptr_t word_char;
446 /* true if REG_ICASE. */
447 unsigned char icase;
448 unsigned char is_utf8;
449 unsigned char map_notascii;
450 unsigned char mbs_allocated;
451 unsigned char offsets_needed;
452 unsigned char newline_anchor;
453 unsigned char word_ops_used;
454 int mb_cur_max;
455 };
456 typedef struct re_string_t re_string_t;
457
458
459 struct re_dfa_t;
460 typedef struct re_dfa_t re_dfa_t;
461
462 #ifndef _LIBC
463 # define IS_IN(libc) false
464 #endif
465
466 #define re_string_peek_byte(pstr, offset) \
467 ((pstr)->mbs[(pstr)->cur_idx + offset])
468 #define re_string_fetch_byte(pstr) \
469 ((pstr)->mbs[(pstr)->cur_idx++])
470 #define re_string_first_byte(pstr, idx) \
471 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
472 #define re_string_is_single_byte_char(pstr, idx) \
473 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
474 || (pstr)->wcs[(idx) + 1] != WEOF))
475 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
476 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
477 #define re_string_get_buffer(pstr) ((pstr)->mbs)
478 #define re_string_length(pstr) ((pstr)->len)
479 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
480 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
481 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
482
483 #if defined _LIBC || HAVE_ALLOCA
484 # include <alloca.h>
485 #endif
486
487 #ifndef _LIBC
488 # if HAVE_ALLOCA
489 /* The OS usually guarantees only one guard page at the bottom of the stack,
490 and a page size can be as small as 4096 bytes. So we cannot safely
491 allocate anything larger than 4096 bytes. Also care for the possibility
492 of a few compiler-allocated temporary stack slots. */
493 # define __libc_use_alloca(n) ((n) < 4032)
494 # else
495 /* alloca is implemented with malloc, so just use malloc. */
496 # define __libc_use_alloca(n) 0
497 # undef alloca
498 # define alloca(n) malloc (n)
499 # endif
500 #endif
501
502 #ifdef _LIBC
503 # define MALLOC_0_IS_NONNULL 1
504 #elif !defined MALLOC_0_IS_NONNULL
505 # define MALLOC_0_IS_NONNULL 0
506 #endif
507
508 #ifndef MAX
509 # define MAX(a,b) ((a) < (b) ? (b) : (a))
510 #endif
511 #ifndef MIN
512 # define MIN(a,b) ((a) < (b) ? (a) : (b))
513 #endif
514
515 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
516 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
517 #define re_free(p) free (p)
518
519 struct bin_tree_t
520 {
521 struct bin_tree_t *parent;
522 struct bin_tree_t *left;
523 struct bin_tree_t *right;
524 struct bin_tree_t *first;
525 struct bin_tree_t *next;
526
527 re_token_t token;
528
529 /* 'node_idx' is the index in dfa->nodes, if 'type' == 0.
530 Otherwise 'type' indicate the type of this node. */
531 Idx node_idx;
532 };
533 typedef struct bin_tree_t bin_tree_t;
534
535 #define BIN_TREE_STORAGE_SIZE \
536 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
537
538 struct bin_tree_storage_t
539 {
540 struct bin_tree_storage_t *next;
541 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
542 };
543 typedef struct bin_tree_storage_t bin_tree_storage_t;
544
545 #define CONTEXT_WORD 1
546 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
547 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
548 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
549
550 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
551 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
552 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
553 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
554 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
555
556 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
557 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
558 #define IS_WIDE_WORD_CHAR(ch) (__iswalnum (ch) || (ch) == L'_')
559 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
560
561 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
562 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
563 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
564 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
565 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
566
567 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
568 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
569 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
570 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
571 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
572
573 struct re_dfastate_t
574 {
575 re_hashval_t hash;
576 re_node_set nodes;
577 re_node_set non_eps_nodes;
578 re_node_set inveclosure;
579 re_node_set *entrance_nodes;
580 struct re_dfastate_t **trtable, **word_trtable;
581 unsigned int context : 4;
582 unsigned int halt : 1;
583 /* If this state can accept "multi byte".
584 Note that we refer to multibyte characters, and multi character
585 collating elements as "multi byte". */
586 unsigned int accept_mb : 1;
587 /* If this state has backreference node(s). */
588 unsigned int has_backref : 1;
589 unsigned int has_constraint : 1;
590 };
591 typedef struct re_dfastate_t re_dfastate_t;
592
593 struct re_state_table_entry
594 {
595 Idx num;
596 Idx alloc;
597 re_dfastate_t **array;
598 };
599
600 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
601
602 typedef struct
603 {
604 Idx next_idx;
605 Idx alloc;
606 re_dfastate_t **array;
607 } state_array_t;
608
609 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
610
611 typedef struct
612 {
613 Idx node;
614 Idx str_idx; /* The position NODE match at. */
615 state_array_t path;
616 } re_sub_match_last_t;
617
618 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
619 And information about the node, whose type is OP_CLOSE_SUBEXP,
620 corresponding to NODE is stored in LASTS. */
621
622 typedef struct
623 {
624 Idx str_idx;
625 Idx node;
626 state_array_t *path;
627 Idx alasts; /* Allocation size of LASTS. */
628 Idx nlasts; /* The number of LASTS. */
629 re_sub_match_last_t **lasts;
630 } re_sub_match_top_t;
631
632 struct re_backref_cache_entry
633 {
634 Idx node;
635 Idx str_idx;
636 Idx subexp_from;
637 Idx subexp_to;
638 char more;
639 char unused;
640 unsigned short int eps_reachable_subexps_map;
641 };
642
643 typedef struct
644 {
645 /* The string object corresponding to the input string. */
646 re_string_t input;
647 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
648 const re_dfa_t *const dfa;
649 #else
650 const re_dfa_t *dfa;
651 #endif
652 /* EFLAGS of the argument of regexec. */
653 int eflags;
654 /* Where the matching ends. */
655 Idx match_last;
656 Idx last_node;
657 /* The state log used by the matcher. */
658 re_dfastate_t **state_log;
659 Idx state_log_top;
660 /* Back reference cache. */
661 Idx nbkref_ents;
662 Idx abkref_ents;
663 struct re_backref_cache_entry *bkref_ents;
664 int max_mb_elem_len;
665 Idx nsub_tops;
666 Idx asub_tops;
667 re_sub_match_top_t **sub_tops;
668 } re_match_context_t;
669
670 typedef struct
671 {
672 re_dfastate_t **sifted_states;
673 re_dfastate_t **limited_states;
674 Idx last_node;
675 Idx last_str_idx;
676 re_node_set limits;
677 } re_sift_context_t;
678
679 struct re_fail_stack_ent_t
680 {
681 Idx idx;
682 Idx node;
683 regmatch_t *regs;
684 re_node_set eps_via_nodes;
685 };
686
687 struct re_fail_stack_t
688 {
689 Idx num;
690 Idx alloc;
691 struct re_fail_stack_ent_t *stack;
692 };
693
694 struct re_dfa_t
695 {
696 re_token_t *nodes;
697 size_t nodes_alloc;
698 size_t nodes_len;
699 Idx *nexts;
700 Idx *org_indices;
701 re_node_set *edests;
702 re_node_set *eclosures;
703 re_node_set *inveclosures;
704 struct re_state_table_entry *state_table;
705 re_dfastate_t *init_state;
706 re_dfastate_t *init_state_word;
707 re_dfastate_t *init_state_nl;
708 re_dfastate_t *init_state_begbuf;
709 bin_tree_t *str_tree;
710 bin_tree_storage_t *str_tree_storage;
711 re_bitset_ptr_t sb_char;
712 int str_tree_storage_idx;
713
714 /* number of subexpressions 're_nsub' is in regex_t. */
715 re_hashval_t state_hash_mask;
716 Idx init_node;
717 Idx nbackref; /* The number of backreference in this dfa. */
718
719 /* Bitmap expressing which backreference is used. */
720 bitset_word_t used_bkref_map;
721 bitset_word_t completed_bkref_map;
722
723 unsigned int has_plural_match : 1;
724 /* If this dfa has "multibyte node", which is a backreference or
725 a node which can accept multibyte character or multi character
726 collating element. */
727 unsigned int has_mb_node : 1;
728 unsigned int is_utf8 : 1;
729 unsigned int map_notascii : 1;
730 unsigned int word_ops_used : 1;
731 int mb_cur_max;
732 bitset_t word_char;
733 reg_syntax_t syntax;
734 Idx *subexp_map;
735 #ifdef DEBUG
736 char* re_str;
737 #endif
738 lock_define (lock)
739 };
740
741 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
742 #define re_node_set_remove(set,id) \
743 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
744 #define re_node_set_empty(p) ((p)->nelem = 0)
745 #define re_node_set_free(set) re_free ((set)->elems)
746 \f
747
748 typedef enum
749 {
750 SB_CHAR,
751 MB_CHAR,
752 EQUIV_CLASS,
753 COLL_SYM,
754 CHAR_CLASS
755 } bracket_elem_type;
756
757 typedef struct
758 {
759 bracket_elem_type type;
760 union
761 {
762 unsigned char ch;
763 unsigned char *name;
764 wchar_t wch;
765 } opr;
766 } bracket_elem_t;
767
768
769 /* Functions for bitset_t operation. */
770
771 static inline void
772 bitset_set (bitset_t set, Idx i)
773 {
774 set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS;
775 }
776
777 static inline void
778 bitset_clear (bitset_t set, Idx i)
779 {
780 set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS);
781 }
782
783 static inline bool
784 bitset_contain (const bitset_t set, Idx i)
785 {
786 return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1;
787 }
788
789 static inline void
790 bitset_empty (bitset_t set)
791 {
792 memset (set, '\0', sizeof (bitset_t));
793 }
794
795 static inline void
796 bitset_set_all (bitset_t set)
797 {
798 memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS));
799 if (SBC_MAX % BITSET_WORD_BITS != 0)
800 set[BITSET_WORDS - 1] =
801 ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1;
802 }
803
804 static inline void
805 bitset_copy (bitset_t dest, const bitset_t src)
806 {
807 memcpy (dest, src, sizeof (bitset_t));
808 }
809
810 static inline void
811 bitset_not (bitset_t set)
812 {
813 int bitset_i;
814 for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i)
815 set[bitset_i] = ~set[bitset_i];
816 if (SBC_MAX % BITSET_WORD_BITS != 0)
817 set[BITSET_WORDS - 1] =
818 ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1)
819 & ~set[BITSET_WORDS - 1]);
820 }
821
822 static inline void
823 bitset_merge (bitset_t dest, const bitset_t src)
824 {
825 int bitset_i;
826 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
827 dest[bitset_i] |= src[bitset_i];
828 }
829
830 static inline void
831 bitset_mask (bitset_t dest, const bitset_t src)
832 {
833 int bitset_i;
834 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
835 dest[bitset_i] &= src[bitset_i];
836 }
837
838 #ifdef RE_ENABLE_I18N
839 /* Functions for re_string. */
840 static int
841 __attribute__ ((pure, unused))
842 re_string_char_size_at (const re_string_t *pstr, Idx idx)
843 {
844 int byte_idx;
845 if (pstr->mb_cur_max == 1)
846 return 1;
847 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
848 if (pstr->wcs[idx + byte_idx] != WEOF)
849 break;
850 return byte_idx;
851 }
852
853 static wint_t
854 __attribute__ ((pure, unused))
855 re_string_wchar_at (const re_string_t *pstr, Idx idx)
856 {
857 if (pstr->mb_cur_max == 1)
858 return (wint_t) pstr->mbs[idx];
859 return (wint_t) pstr->wcs[idx];
860 }
861
862 # ifdef _LIBC
863 # include <locale/weight.h>
864 # endif
865
866 static int
867 __attribute__ ((pure, unused))
868 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
869 {
870 # ifdef _LIBC
871 const unsigned char *p, *extra;
872 const int32_t *table, *indirect;
873 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
874
875 if (nrules != 0)
876 {
877 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
878 extra = (const unsigned char *)
879 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
880 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
881 _NL_COLLATE_INDIRECTMB);
882 p = pstr->mbs + idx;
883 findidx (table, indirect, extra, &p, pstr->len - idx);
884 return p - pstr->mbs - idx;
885 }
886 else
887 # endif /* _LIBC */
888 return 1;
889 }
890 #endif /* RE_ENABLE_I18N */
891
892 #ifndef __GNUC_PREREQ
893 # if defined __GNUC__ && defined __GNUC_MINOR__
894 # define __GNUC_PREREQ(maj, min) \
895 ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
896 # else
897 # define __GNUC_PREREQ(maj, min) 0
898 # endif
899 #endif
900
901 #if __GNUC_PREREQ (3,4)
902 # undef __attribute_warn_unused_result__
903 # define __attribute_warn_unused_result__ \
904 __attribute__ ((__warn_unused_result__))
905 #else
906 # define __attribute_warn_unused_result__ /* empty */
907 #endif
908
909 #ifndef FALLTHROUGH
910 # if __GNUC__ < 7
911 # define FALLTHROUGH ((void) 0)
912 # else
913 # define FALLTHROUGH __attribute__ ((__fallthrough__))
914 # endif
915 #endif
916
917 #endif /* _REGEX_INTERNAL_H */