1 | /* Extended regular expression matching and search library. |
2 | Copyright (C) 2002-2023 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 <ctype.h> |
24 | #include <stdio.h> |
25 | #include <stdlib.h> |
26 | #include <string.h> |
27 | |
28 | #include <langinfo.h> |
29 | #include <locale.h> |
30 | #include <wchar.h> |
31 | #include <wctype.h> |
32 | #include <stdbool.h> |
33 | #include <stdint.h> |
34 | |
35 | #ifndef _LIBC |
36 | # include <dynarray.h> |
37 | #endif |
38 | |
39 | #include <intprops.h> |
40 | #include <verify.h> |
41 | |
42 | #if defined DEBUG && DEBUG != 0 |
43 | # include <assert.h> |
44 | # define DEBUG_ASSERT(x) assert (x) |
45 | #else |
46 | # define DEBUG_ASSERT(x) assume (x) |
47 | #endif |
48 | |
49 | #ifdef _LIBC |
50 | # include <libc-lock.h> |
51 | # define lock_define(name) __libc_lock_define (, name) |
52 | # define lock_init(lock) (__libc_lock_init (lock), 0) |
53 | # define lock_fini(lock) ((void) 0) |
54 | # define lock_lock(lock) __libc_lock_lock (lock) |
55 | # define lock_unlock(lock) __libc_lock_unlock (lock) |
56 | #elif defined GNULIB_LOCK && !defined GNULIB_REGEX_SINGLE_THREAD |
57 | # include "glthread/lock.h" |
58 | # define lock_define(name) gl_lock_define (, name) |
59 | # define lock_init(lock) glthread_lock_init (&(lock)) |
60 | # define lock_fini(lock) glthread_lock_destroy (&(lock)) |
61 | # define lock_lock(lock) glthread_lock_lock (&(lock)) |
62 | # define lock_unlock(lock) glthread_lock_unlock (&(lock)) |
63 | #elif defined GNULIB_PTHREAD && !defined GNULIB_REGEX_SINGLE_THREAD |
64 | # include <pthread.h> |
65 | # define lock_define(name) pthread_mutex_t name; |
66 | # define lock_init(lock) pthread_mutex_init (&(lock), 0) |
67 | # define lock_fini(lock) pthread_mutex_destroy (&(lock)) |
68 | # define lock_lock(lock) pthread_mutex_lock (&(lock)) |
69 | # define lock_unlock(lock) pthread_mutex_unlock (&(lock)) |
70 | #else |
71 | # define lock_define(name) |
72 | # define lock_init(lock) 0 |
73 | # define lock_fini(lock) ((void) 0) |
74 | /* The 'dfa' avoids an "unused variable 'dfa'" warning from GCC. */ |
75 | # define lock_lock(lock) ((void) dfa) |
76 | # define lock_unlock(lock) ((void) 0) |
77 | #endif |
78 | |
79 | /* In case that the system doesn't have isblank(). */ |
80 | #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) |
81 | # define isblank(ch) ((ch) == ' ' || (ch) == '\t') |
82 | #endif |
83 | |
84 | /* regex code assumes isascii has its usual numeric meaning, |
85 | even if the portable character set uses EBCDIC encoding, |
86 | and even if wint_t is wider than int. */ |
87 | #ifndef _LIBC |
88 | # undef isascii |
89 | # define isascii(c) (((c) & ~0x7f) == 0) |
90 | #endif |
91 | |
92 | #ifdef _LIBC |
93 | # ifndef _RE_DEFINE_LOCALE_FUNCTIONS |
94 | # define _RE_DEFINE_LOCALE_FUNCTIONS 1 |
95 | # include <locale/localeinfo.h> |
96 | # include <locale/coll-lookup.h> |
97 | # endif |
98 | #endif |
99 | |
100 | /* This is for other GNU distributions with internationalized messages. */ |
101 | #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC |
102 | # include <libintl.h> |
103 | # ifdef _LIBC |
104 | # undef gettext |
105 | # define gettext(msgid) \ |
106 | __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES) |
107 | # endif |
108 | #else |
109 | # undef gettext |
110 | # define gettext(msgid) (msgid) |
111 | #endif |
112 | |
113 | #ifndef gettext_noop |
114 | /* This define is so xgettext can find the internationalizable |
115 | strings. */ |
116 | # define gettext_noop(String) String |
117 | #endif |
118 | |
119 | #if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE) || _LIBC |
120 | # define RE_ENABLE_I18N |
121 | #endif |
122 | |
123 | /* Number of ASCII characters. */ |
124 | #define ASCII_CHARS 0x80 |
125 | |
126 | /* Number of single byte characters. */ |
127 | #define SBC_MAX (UCHAR_MAX + 1) |
128 | |
129 | #define COLL_ELEM_LEN_MAX 8 |
130 | |
131 | /* The character which represents newline. */ |
132 | #define NEWLINE_CHAR '\n' |
133 | #define WIDE_NEWLINE_CHAR L'\n' |
134 | |
135 | /* Rename to standard API for using out of glibc. */ |
136 | #ifndef _LIBC |
137 | # undef __wctype |
138 | # undef __iswalnum |
139 | # undef __iswctype |
140 | # undef __towlower |
141 | # undef __towupper |
142 | # define __wctype wctype |
143 | # define __iswalnum iswalnum |
144 | # define __iswctype iswctype |
145 | # define __towlower towlower |
146 | # define __towupper towupper |
147 | # define __btowc btowc |
148 | # define __mbrtowc mbrtowc |
149 | # define __wcrtomb wcrtomb |
150 | # define __regfree regfree |
151 | #endif /* not _LIBC */ |
152 | |
153 | #ifndef SSIZE_MAX |
154 | # define SSIZE_MAX ((ssize_t) (SIZE_MAX / 2)) |
155 | #endif |
156 | #ifndef ULONG_WIDTH |
157 | # define ULONG_WIDTH REGEX_UINTEGER_WIDTH (ULONG_MAX) |
158 | /* The number of usable bits in an unsigned integer type with maximum |
159 | value MAX, as an int expression suitable in #if. Cover all known |
160 | practical hosts. This implementation exploits the fact that MAX is |
161 | 1 less than a power of 2, and merely counts the number of 1 bits in |
162 | MAX; "COBn" means "count the number of 1 bits in the low-order n bits". */ |
163 | # define REGEX_UINTEGER_WIDTH(max) REGEX_COB128 (max) |
164 | # define REGEX_COB128(n) (REGEX_COB64 ((n) >> 31 >> 31 >> 2) + REGEX_COB64 (n)) |
165 | # define REGEX_COB64(n) (REGEX_COB32 ((n) >> 31 >> 1) + REGEX_COB32 (n)) |
166 | # define REGEX_COB32(n) (REGEX_COB16 ((n) >> 16) + REGEX_COB16 (n)) |
167 | # define REGEX_COB16(n) (REGEX_COB8 ((n) >> 8) + REGEX_COB8 (n)) |
168 | # define REGEX_COB8(n) (REGEX_COB4 ((n) >> 4) + REGEX_COB4 (n)) |
169 | # define REGEX_COB4(n) (!!((n) & 8) + !!((n) & 4) + !!((n) & 2) + ((n) & 1)) |
170 | # if ULONG_MAX / 2 + 1 != 1ul << (ULONG_WIDTH - 1) |
171 | # error "ULONG_MAX out of range" |
172 | # endif |
173 | #endif |
174 | |
175 | /* The type of indexes into strings. This is signed, not size_t, |
176 | since the API requires indexes to fit in regoff_t anyway, and using |
177 | signed integers makes the code a bit smaller and presumably faster. |
178 | The traditional GNU regex implementation uses int for indexes. |
179 | The POSIX-compatible implementation uses a possibly-wider type. |
180 | The name 'Idx' is three letters to minimize the hassle of |
181 | reindenting a lot of regex code that formerly used 'int'. */ |
182 | typedef regoff_t Idx; |
183 | #ifdef _REGEX_LARGE_OFFSETS |
184 | # define IDX_MAX SSIZE_MAX |
185 | #else |
186 | # define IDX_MAX INT_MAX |
187 | #endif |
188 | |
189 | /* A hash value, suitable for computing hash tables. */ |
190 | typedef __re_size_t re_hashval_t; |
191 | |
192 | /* An integer used to represent a set of bits. It must be unsigned, |
193 | and must be at least as wide as unsigned int. */ |
194 | typedef unsigned long int bitset_word_t; |
195 | /* All bits set in a bitset_word_t. */ |
196 | #define BITSET_WORD_MAX ULONG_MAX |
197 | /* Number of bits in a bitset_word_t. */ |
198 | #define BITSET_WORD_BITS ULONG_WIDTH |
199 | |
200 | /* Number of bitset_word_t values in a bitset_t. */ |
201 | #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) |
202 | |
203 | typedef bitset_word_t bitset_t[BITSET_WORDS]; |
204 | typedef bitset_word_t *re_bitset_ptr_t; |
205 | typedef const bitset_word_t *re_const_bitset_ptr_t; |
206 | |
207 | #define PREV_WORD_CONSTRAINT 0x0001 |
208 | #define PREV_NOTWORD_CONSTRAINT 0x0002 |
209 | #define NEXT_WORD_CONSTRAINT 0x0004 |
210 | #define NEXT_NOTWORD_CONSTRAINT 0x0008 |
211 | #define PREV_NEWLINE_CONSTRAINT 0x0010 |
212 | #define NEXT_NEWLINE_CONSTRAINT 0x0020 |
213 | #define PREV_BEGBUF_CONSTRAINT 0x0040 |
214 | #define NEXT_ENDBUF_CONSTRAINT 0x0080 |
215 | #define WORD_DELIM_CONSTRAINT 0x0100 |
216 | #define NOT_WORD_DELIM_CONSTRAINT 0x0200 |
217 | |
218 | typedef enum |
219 | { |
220 | INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, |
221 | WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, |
222 | WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, |
223 | INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, |
224 | LINE_FIRST = PREV_NEWLINE_CONSTRAINT, |
225 | LINE_LAST = NEXT_NEWLINE_CONSTRAINT, |
226 | BUF_FIRST = PREV_BEGBUF_CONSTRAINT, |
227 | BUF_LAST = NEXT_ENDBUF_CONSTRAINT, |
228 | WORD_DELIM = WORD_DELIM_CONSTRAINT, |
229 | NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT |
230 | } re_context_type; |
231 | |
232 | typedef struct |
233 | { |
234 | Idx alloc; |
235 | Idx nelem; |
236 | Idx *elems; |
237 | } re_node_set; |
238 | |
239 | typedef enum |
240 | { |
241 | NON_TYPE = 0, |
242 | |
243 | /* Node type, These are used by token, node, tree. */ |
244 | CHARACTER = 1, |
245 | END_OF_RE = 2, |
246 | SIMPLE_BRACKET = 3, |
247 | OP_BACK_REF = 4, |
248 | OP_PERIOD = 5, |
249 | #ifdef RE_ENABLE_I18N |
250 | COMPLEX_BRACKET = 6, |
251 | OP_UTF8_PERIOD = 7, |
252 | #endif /* RE_ENABLE_I18N */ |
253 | |
254 | /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used |
255 | when the debugger shows values of this enum type. */ |
256 | #define EPSILON_BIT 8 |
257 | OP_OPEN_SUBEXP = EPSILON_BIT | 0, |
258 | OP_CLOSE_SUBEXP = EPSILON_BIT | 1, |
259 | OP_ALT = EPSILON_BIT | 2, |
260 | OP_DUP_ASTERISK = EPSILON_BIT | 3, |
261 | ANCHOR = EPSILON_BIT | 4, |
262 | |
263 | /* Tree type, these are used only by tree. */ |
264 | CONCAT = 16, |
265 | SUBEXP = 17, |
266 | |
267 | /* Token type, these are used only by token. */ |
268 | OP_DUP_PLUS = 18, |
269 | OP_DUP_QUESTION, |
270 | OP_OPEN_BRACKET, |
271 | OP_CLOSE_BRACKET, |
272 | OP_CHARSET_RANGE, |
273 | OP_OPEN_DUP_NUM, |
274 | OP_CLOSE_DUP_NUM, |
275 | OP_NON_MATCH_LIST, |
276 | OP_OPEN_COLL_ELEM, |
277 | OP_CLOSE_COLL_ELEM, |
278 | OP_OPEN_EQUIV_CLASS, |
279 | OP_CLOSE_EQUIV_CLASS, |
280 | OP_OPEN_CHAR_CLASS, |
281 | OP_CLOSE_CHAR_CLASS, |
282 | OP_WORD, |
283 | OP_NOTWORD, |
284 | OP_SPACE, |
285 | OP_NOTSPACE, |
286 | BACK_SLASH |
287 | |
288 | } re_token_type_t; |
289 | |
290 | #ifdef RE_ENABLE_I18N |
291 | typedef struct |
292 | { |
293 | /* Multibyte characters. */ |
294 | wchar_t *mbchars; |
295 | |
296 | /* Collating symbols. */ |
297 | # ifdef _LIBC |
298 | int32_t *coll_syms; |
299 | # endif |
300 | |
301 | /* Equivalence classes. */ |
302 | # ifdef _LIBC |
303 | int32_t *equiv_classes; |
304 | # endif |
305 | |
306 | /* Range expressions. */ |
307 | # ifdef _LIBC |
308 | uint32_t *range_starts; |
309 | uint32_t *range_ends; |
310 | # else /* not _LIBC */ |
311 | wchar_t *range_starts; |
312 | wchar_t *range_ends; |
313 | # endif /* not _LIBC */ |
314 | |
315 | /* Character classes. */ |
316 | wctype_t *char_classes; |
317 | |
318 | /* If this character set is the non-matching list. */ |
319 | unsigned int non_match : 1; |
320 | |
321 | /* # of multibyte characters. */ |
322 | Idx nmbchars; |
323 | |
324 | /* # of collating symbols. */ |
325 | Idx ncoll_syms; |
326 | |
327 | /* # of equivalence classes. */ |
328 | Idx nequiv_classes; |
329 | |
330 | /* # of range expressions. */ |
331 | Idx nranges; |
332 | |
333 | /* # of character classes. */ |
334 | Idx nchar_classes; |
335 | } re_charset_t; |
336 | #endif /* RE_ENABLE_I18N */ |
337 | |
338 | typedef struct |
339 | { |
340 | union |
341 | { |
342 | unsigned char c; /* for CHARACTER */ |
343 | re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ |
344 | #ifdef RE_ENABLE_I18N |
345 | re_charset_t *mbcset; /* for COMPLEX_BRACKET */ |
346 | #endif /* RE_ENABLE_I18N */ |
347 | Idx idx; /* for BACK_REF */ |
348 | re_context_type ctx_type; /* for ANCHOR */ |
349 | } opr; |
350 | #if (__GNUC__ >= 2 || defined __clang__) && !defined __STRICT_ANSI__ |
351 | re_token_type_t type : 8; |
352 | #else |
353 | re_token_type_t type; |
354 | #endif |
355 | unsigned int constraint : 10; /* context constraint */ |
356 | unsigned int duplicated : 1; |
357 | unsigned int opt_subexp : 1; |
358 | #ifdef RE_ENABLE_I18N |
359 | unsigned int accept_mb : 1; |
360 | /* These 2 bits can be moved into the union if needed (e.g. if running out |
361 | of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ |
362 | unsigned int mb_partial : 1; |
363 | #endif |
364 | unsigned int word_char : 1; |
365 | } re_token_t; |
366 | |
367 | #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) |
368 | |
369 | struct re_string_t |
370 | { |
371 | /* Indicate the raw buffer which is the original string passed as an |
372 | argument of regexec(), re_search(), etc.. */ |
373 | const unsigned char *raw_mbs; |
374 | /* Store the multibyte string. In case of "case insensitive mode" like |
375 | REG_ICASE, upper cases of the string are stored, otherwise MBS points |
376 | the same address that RAW_MBS points. */ |
377 | unsigned char *mbs; |
378 | #ifdef RE_ENABLE_I18N |
379 | /* Store the wide character string which is corresponding to MBS. */ |
380 | wint_t *wcs; |
381 | Idx *offsets; |
382 | mbstate_t cur_state; |
383 | #endif |
384 | /* Index in RAW_MBS. Each character mbs[i] corresponds to |
385 | raw_mbs[raw_mbs_idx + i]. */ |
386 | Idx raw_mbs_idx; |
387 | /* The length of the valid characters in the buffers. */ |
388 | Idx valid_len; |
389 | /* The corresponding number of bytes in raw_mbs array. */ |
390 | Idx valid_raw_len; |
391 | /* The length of the buffers MBS and WCS. */ |
392 | Idx bufs_len; |
393 | /* The index in MBS, which is updated by re_string_fetch_byte. */ |
394 | Idx cur_idx; |
395 | /* length of RAW_MBS array. */ |
396 | Idx raw_len; |
397 | /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ |
398 | Idx len; |
399 | /* End of the buffer may be shorter than its length in the cases such |
400 | as re_match_2, re_search_2. Then, we use STOP for end of the buffer |
401 | instead of LEN. */ |
402 | Idx raw_stop; |
403 | /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ |
404 | Idx stop; |
405 | |
406 | /* The context of mbs[0]. We store the context independently, since |
407 | the context of mbs[0] may be different from raw_mbs[0], which is |
408 | the beginning of the input string. */ |
409 | unsigned int tip_context; |
410 | /* The translation passed as a part of an argument of re_compile_pattern. */ |
411 | RE_TRANSLATE_TYPE trans; |
412 | /* Copy of re_dfa_t's word_char. */ |
413 | re_const_bitset_ptr_t word_char; |
414 | /* true if REG_ICASE. */ |
415 | unsigned char icase; |
416 | unsigned char is_utf8; |
417 | unsigned char map_notascii; |
418 | unsigned char mbs_allocated; |
419 | unsigned char offsets_needed; |
420 | unsigned char newline_anchor; |
421 | unsigned char word_ops_used; |
422 | int mb_cur_max; |
423 | }; |
424 | typedef struct re_string_t re_string_t; |
425 | |
426 | |
427 | struct re_dfa_t; |
428 | typedef struct re_dfa_t re_dfa_t; |
429 | |
430 | #ifndef _LIBC |
431 | # define IS_IN(libc) false |
432 | #endif |
433 | |
434 | #define re_string_peek_byte(pstr, offset) \ |
435 | ((pstr)->mbs[(pstr)->cur_idx + offset]) |
436 | #define re_string_fetch_byte(pstr) \ |
437 | ((pstr)->mbs[(pstr)->cur_idx++]) |
438 | #define re_string_first_byte(pstr, idx) \ |
439 | ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) |
440 | #define re_string_is_single_byte_char(pstr, idx) \ |
441 | ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ |
442 | || (pstr)->wcs[(idx) + 1] != WEOF)) |
443 | #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) |
444 | #define re_string_cur_idx(pstr) ((pstr)->cur_idx) |
445 | #define re_string_get_buffer(pstr) ((pstr)->mbs) |
446 | #define re_string_length(pstr) ((pstr)->len) |
447 | #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) |
448 | #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) |
449 | #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) |
450 | |
451 | #ifdef _LIBC |
452 | # define MALLOC_0_IS_NONNULL 1 |
453 | #elif !defined MALLOC_0_IS_NONNULL |
454 | # define MALLOC_0_IS_NONNULL 0 |
455 | #endif |
456 | |
457 | #ifndef MAX |
458 | # define MAX(a,b) ((a) < (b) ? (b) : (a)) |
459 | #endif |
460 | #ifndef MIN |
461 | # define MIN(a,b) ((a) < (b) ? (a) : (b)) |
462 | #endif |
463 | |
464 | #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) |
465 | #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) |
466 | #define re_free(p) free (p) |
467 | |
468 | struct bin_tree_t |
469 | { |
470 | struct bin_tree_t *parent; |
471 | struct bin_tree_t *left; |
472 | struct bin_tree_t *right; |
473 | struct bin_tree_t *first; |
474 | struct bin_tree_t *next; |
475 | |
476 | re_token_t token; |
477 | |
478 | /* 'node_idx' is the index in dfa->nodes, if 'type' == 0. |
479 | Otherwise 'type' indicate the type of this node. */ |
480 | Idx node_idx; |
481 | }; |
482 | typedef struct bin_tree_t bin_tree_t; |
483 | |
484 | #define BIN_TREE_STORAGE_SIZE \ |
485 | ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) |
486 | |
487 | struct bin_tree_storage_t |
488 | { |
489 | struct bin_tree_storage_t *next; |
490 | bin_tree_t data[BIN_TREE_STORAGE_SIZE]; |
491 | }; |
492 | typedef struct bin_tree_storage_t bin_tree_storage_t; |
493 | |
494 | #define CONTEXT_WORD 1 |
495 | #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) |
496 | #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) |
497 | #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) |
498 | |
499 | #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) |
500 | #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) |
501 | #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) |
502 | #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) |
503 | #define IS_ORDINARY_CONTEXT(c) ((c) == 0) |
504 | |
505 | #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') |
506 | #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) |
507 | #define IS_WIDE_WORD_CHAR(ch) (__iswalnum (ch) || (ch) == L'_') |
508 | #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) |
509 | |
510 | #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ |
511 | ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ |
512 | || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ |
513 | || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ |
514 | || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) |
515 | |
516 | #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ |
517 | ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ |
518 | || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ |
519 | || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ |
520 | || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) |
521 | |
522 | struct re_dfastate_t |
523 | { |
524 | re_hashval_t hash; |
525 | re_node_set nodes; |
526 | re_node_set non_eps_nodes; |
527 | re_node_set inveclosure; |
528 | re_node_set *entrance_nodes; |
529 | struct re_dfastate_t **trtable, **word_trtable; |
530 | unsigned int context : 4; |
531 | unsigned int halt : 1; |
532 | /* If this state can accept "multi byte". |
533 | Note that we refer to multibyte characters, and multi character |
534 | collating elements as "multi byte". */ |
535 | unsigned int accept_mb : 1; |
536 | /* If this state has backreference node(s). */ |
537 | unsigned int has_backref : 1; |
538 | unsigned int has_constraint : 1; |
539 | }; |
540 | typedef struct re_dfastate_t re_dfastate_t; |
541 | |
542 | struct re_state_table_entry |
543 | { |
544 | Idx num; |
545 | Idx alloc; |
546 | re_dfastate_t **array; |
547 | }; |
548 | |
549 | /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ |
550 | |
551 | typedef struct |
552 | { |
553 | Idx next_idx; |
554 | Idx alloc; |
555 | re_dfastate_t **array; |
556 | } state_array_t; |
557 | |
558 | /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ |
559 | |
560 | typedef struct |
561 | { |
562 | Idx node; |
563 | Idx str_idx; /* The position NODE match at. */ |
564 | state_array_t path; |
565 | } re_sub_match_last_t; |
566 | |
567 | /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. |
568 | And information about the node, whose type is OP_CLOSE_SUBEXP, |
569 | corresponding to NODE is stored in LASTS. */ |
570 | |
571 | typedef struct |
572 | { |
573 | Idx str_idx; |
574 | Idx node; |
575 | state_array_t *path; |
576 | Idx alasts; /* Allocation size of LASTS. */ |
577 | Idx nlasts; /* The number of LASTS. */ |
578 | re_sub_match_last_t **lasts; |
579 | } re_sub_match_top_t; |
580 | |
581 | struct re_backref_cache_entry |
582 | { |
583 | Idx node; |
584 | Idx str_idx; |
585 | Idx subexp_from; |
586 | Idx subexp_to; |
587 | bitset_word_t eps_reachable_subexps_map; |
588 | char more; |
589 | }; |
590 | |
591 | typedef struct |
592 | { |
593 | /* The string object corresponding to the input string. */ |
594 | re_string_t input; |
595 | const re_dfa_t *const dfa; |
596 | /* EFLAGS of the argument of regexec. */ |
597 | int eflags; |
598 | /* Where the matching ends. */ |
599 | Idx match_last; |
600 | Idx last_node; |
601 | /* The state log used by the matcher. */ |
602 | re_dfastate_t **state_log; |
603 | Idx state_log_top; |
604 | /* Back reference cache. */ |
605 | Idx nbkref_ents; |
606 | Idx abkref_ents; |
607 | struct re_backref_cache_entry *bkref_ents; |
608 | int max_mb_elem_len; |
609 | Idx nsub_tops; |
610 | Idx asub_tops; |
611 | re_sub_match_top_t **sub_tops; |
612 | } re_match_context_t; |
613 | |
614 | typedef struct |
615 | { |
616 | re_dfastate_t **sifted_states; |
617 | re_dfastate_t **limited_states; |
618 | Idx last_node; |
619 | Idx last_str_idx; |
620 | re_node_set limits; |
621 | } re_sift_context_t; |
622 | |
623 | struct re_fail_stack_ent_t |
624 | { |
625 | Idx idx; |
626 | Idx node; |
627 | regmatch_t *regs; |
628 | re_node_set eps_via_nodes; |
629 | }; |
630 | |
631 | struct re_fail_stack_t |
632 | { |
633 | Idx num; |
634 | Idx alloc; |
635 | struct re_fail_stack_ent_t *stack; |
636 | }; |
637 | |
638 | struct re_dfa_t |
639 | { |
640 | re_token_t *nodes; |
641 | size_t nodes_alloc; |
642 | size_t nodes_len; |
643 | Idx *nexts; |
644 | Idx *org_indices; |
645 | re_node_set *edests; |
646 | re_node_set *eclosures; |
647 | re_node_set *inveclosures; |
648 | struct re_state_table_entry *state_table; |
649 | re_dfastate_t *init_state; |
650 | re_dfastate_t *init_state_word; |
651 | re_dfastate_t *init_state_nl; |
652 | re_dfastate_t *init_state_begbuf; |
653 | bin_tree_t *str_tree; |
654 | bin_tree_storage_t *str_tree_storage; |
655 | re_bitset_ptr_t sb_char; |
656 | int str_tree_storage_idx; |
657 | |
658 | /* number of subexpressions 're_nsub' is in regex_t. */ |
659 | re_hashval_t state_hash_mask; |
660 | Idx init_node; |
661 | Idx nbackref; /* The number of backreference in this dfa. */ |
662 | |
663 | /* Bitmap expressing which backreference is used. */ |
664 | bitset_word_t used_bkref_map; |
665 | bitset_word_t completed_bkref_map; |
666 | |
667 | unsigned int has_plural_match : 1; |
668 | /* If this dfa has "multibyte node", which is a backreference or |
669 | a node which can accept multibyte character or multi character |
670 | collating element. */ |
671 | unsigned int has_mb_node : 1; |
672 | unsigned int is_utf8 : 1; |
673 | unsigned int map_notascii : 1; |
674 | unsigned int word_ops_used : 1; |
675 | int mb_cur_max; |
676 | bitset_t word_char; |
677 | reg_syntax_t syntax; |
678 | Idx *subexp_map; |
679 | #ifdef DEBUG |
680 | char* re_str; |
681 | #endif |
682 | lock_define (lock) |
683 | }; |
684 | |
685 | #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) |
686 | #define re_node_set_remove(set,id) \ |
687 | (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) |
688 | #define re_node_set_empty(p) ((p)->nelem = 0) |
689 | #define re_node_set_free(set) re_free ((set)->elems) |
690 | |
691 | |
692 | typedef enum |
693 | { |
694 | SB_CHAR, |
695 | MB_CHAR, |
696 | EQUIV_CLASS, |
697 | COLL_SYM, |
698 | CHAR_CLASS |
699 | } bracket_elem_type; |
700 | |
701 | typedef struct |
702 | { |
703 | bracket_elem_type type; |
704 | union |
705 | { |
706 | unsigned char ch; |
707 | unsigned char *name; |
708 | wchar_t wch; |
709 | } opr; |
710 | } bracket_elem_t; |
711 | |
712 | |
713 | /* Functions for bitset_t operation. */ |
714 | |
715 | static inline void |
716 | bitset_set (bitset_t set, Idx i) |
717 | { |
718 | set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; |
719 | } |
720 | |
721 | static inline void |
722 | bitset_clear (bitset_t set, Idx i) |
723 | { |
724 | set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); |
725 | } |
726 | |
727 | static inline bool |
728 | bitset_contain (const bitset_t set, Idx i) |
729 | { |
730 | return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; |
731 | } |
732 | |
733 | static inline void |
734 | bitset_empty (bitset_t set) |
735 | { |
736 | memset (set, '\0', sizeof (bitset_t)); |
737 | } |
738 | |
739 | static inline void |
740 | bitset_set_all (bitset_t set) |
741 | { |
742 | memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); |
743 | if (SBC_MAX % BITSET_WORD_BITS != 0) |
744 | set[BITSET_WORDS - 1] = |
745 | ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; |
746 | } |
747 | |
748 | static inline void |
749 | bitset_copy (bitset_t dest, const bitset_t src) |
750 | { |
751 | memcpy (dest, src, sizeof (bitset_t)); |
752 | } |
753 | |
754 | static inline void |
755 | bitset_not (bitset_t set) |
756 | { |
757 | int bitset_i; |
758 | for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) |
759 | set[bitset_i] = ~set[bitset_i]; |
760 | if (SBC_MAX % BITSET_WORD_BITS != 0) |
761 | set[BITSET_WORDS - 1] = |
762 | ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) |
763 | & ~set[BITSET_WORDS - 1]); |
764 | } |
765 | |
766 | static inline void |
767 | bitset_merge (bitset_t dest, const bitset_t src) |
768 | { |
769 | int bitset_i; |
770 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) |
771 | dest[bitset_i] |= src[bitset_i]; |
772 | } |
773 | |
774 | static inline void |
775 | bitset_mask (bitset_t dest, const bitset_t src) |
776 | { |
777 | int bitset_i; |
778 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) |
779 | dest[bitset_i] &= src[bitset_i]; |
780 | } |
781 | |
782 | #ifdef RE_ENABLE_I18N |
783 | /* Functions for re_string. */ |
784 | static int |
785 | __attribute__ ((pure, unused)) |
786 | re_string_char_size_at (const re_string_t *pstr, Idx idx) |
787 | { |
788 | int byte_idx; |
789 | if (pstr->mb_cur_max == 1) |
790 | return 1; |
791 | for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) |
792 | if (pstr->wcs[idx + byte_idx] != WEOF) |
793 | break; |
794 | return byte_idx; |
795 | } |
796 | |
797 | static wint_t |
798 | __attribute__ ((pure, unused)) |
799 | re_string_wchar_at (const re_string_t *pstr, Idx idx) |
800 | { |
801 | if (pstr->mb_cur_max == 1) |
802 | return (wint_t) pstr->mbs[idx]; |
803 | return (wint_t) pstr->wcs[idx]; |
804 | } |
805 | |
806 | # ifdef _LIBC |
807 | # include <locale/weight.h> |
808 | # endif |
809 | |
810 | static int |
811 | __attribute__ ((pure, unused)) |
812 | re_string_elem_size_at (const re_string_t *pstr, Idx idx) |
813 | { |
814 | # ifdef _LIBC |
815 | const unsigned char *p, *; |
816 | const int32_t *table, *indirect; |
817 | uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); |
818 | |
819 | if (nrules != 0) |
820 | { |
821 | table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); |
822 | extra = (const unsigned char *) |
823 | _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); |
824 | indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, |
825 | _NL_COLLATE_INDIRECTMB); |
826 | p = pstr->mbs + idx; |
827 | findidx (table, indirect, extra, &p, pstr->len - idx); |
828 | return p - pstr->mbs - idx; |
829 | } |
830 | else |
831 | # endif /* _LIBC */ |
832 | return 1; |
833 | } |
834 | #endif /* RE_ENABLE_I18N */ |
835 | |
836 | #ifdef _LIBC |
837 | # if __GNUC__ >= 7 |
838 | # define FALLTHROUGH __attribute__ ((__fallthrough__)) |
839 | # else |
840 | # define FALLTHROUGH ((void) 0) |
841 | # endif |
842 | #else |
843 | # include "attribute.h" |
844 | #endif |
845 | |
846 | #endif /* _REGEX_INTERNAL_H */ |
847 | |