1 | /* Extended regular expression matching and search library. |
2 | Copyright (C) 2002-2021 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 | #include <intprops.h> |
36 | #include <verify.h> |
37 | |
38 | #if defined DEBUG && DEBUG != 0 |
39 | # include <assert.h> |
40 | # define DEBUG_ASSERT(x) assert (x) |
41 | #else |
42 | # define DEBUG_ASSERT(x) assume (x) |
43 | #endif |
44 | |
45 | #ifdef _LIBC |
46 | # include <libc-lock.h> |
47 | # define lock_define(name) __libc_lock_define (, name) |
48 | # define lock_init(lock) (__libc_lock_init (lock), 0) |
49 | # define lock_fini(lock) ((void) 0) |
50 | # define lock_lock(lock) __libc_lock_lock (lock) |
51 | # define lock_unlock(lock) __libc_lock_unlock (lock) |
52 | #elif defined GNULIB_LOCK && !defined USE_UNLOCKED_IO |
53 | # include "glthread/lock.h" |
54 | # define lock_define(name) gl_lock_define (, name) |
55 | # define lock_init(lock) glthread_lock_init (&(lock)) |
56 | # define lock_fini(lock) glthread_lock_destroy (&(lock)) |
57 | # define lock_lock(lock) glthread_lock_lock (&(lock)) |
58 | # define lock_unlock(lock) glthread_lock_unlock (&(lock)) |
59 | #elif defined GNULIB_PTHREAD && !defined USE_UNLOCKED_IO |
60 | # include <pthread.h> |
61 | # define lock_define(name) pthread_mutex_t name; |
62 | # define lock_init(lock) pthread_mutex_init (&(lock), 0) |
63 | # define lock_fini(lock) pthread_mutex_destroy (&(lock)) |
64 | # define lock_lock(lock) pthread_mutex_lock (&(lock)) |
65 | # define lock_unlock(lock) pthread_mutex_unlock (&(lock)) |
66 | #else |
67 | # define lock_define(name) |
68 | # define lock_init(lock) 0 |
69 | # define lock_fini(lock) ((void) 0) |
70 | /* The 'dfa' avoids an "unused variable 'dfa'" warning from GCC. */ |
71 | # define lock_lock(lock) ((void) dfa) |
72 | # define lock_unlock(lock) ((void) 0) |
73 | #endif |
74 | |
75 | /* In case that the system doesn't have isblank(). */ |
76 | #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) |
77 | # define isblank(ch) ((ch) == ' ' || (ch) == '\t') |
78 | #endif |
79 | |
80 | /* regex code assumes isascii has its usual numeric meaning, |
81 | even if the portable character set uses EBCDIC encoding, |
82 | and even if wint_t is wider than int. */ |
83 | #ifndef _LIBC |
84 | # undef isascii |
85 | # define isascii(c) (((c) & ~0x7f) == 0) |
86 | #endif |
87 | |
88 | #ifdef _LIBC |
89 | # ifndef _RE_DEFINE_LOCALE_FUNCTIONS |
90 | # define _RE_DEFINE_LOCALE_FUNCTIONS 1 |
91 | # include <locale/localeinfo.h> |
92 | # include <locale/coll-lookup.h> |
93 | # endif |
94 | #endif |
95 | |
96 | /* This is for other GNU distributions with internationalized messages. */ |
97 | #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC |
98 | # include <libintl.h> |
99 | # ifdef _LIBC |
100 | # undef gettext |
101 | # define gettext(msgid) \ |
102 | __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES) |
103 | # endif |
104 | #else |
105 | # undef gettext |
106 | # define gettext(msgid) (msgid) |
107 | #endif |
108 | |
109 | #ifndef gettext_noop |
110 | /* This define is so xgettext can find the internationalizable |
111 | strings. */ |
112 | # define gettext_noop(String) String |
113 | #endif |
114 | |
115 | #if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE) || _LIBC |
116 | # define RE_ENABLE_I18N |
117 | #endif |
118 | |
119 | /* Number of ASCII characters. */ |
120 | #define ASCII_CHARS 0x80 |
121 | |
122 | /* Number of single byte characters. */ |
123 | #define SBC_MAX (UCHAR_MAX + 1) |
124 | |
125 | #define COLL_ELEM_LEN_MAX 8 |
126 | |
127 | /* The character which represents newline. */ |
128 | #define NEWLINE_CHAR '\n' |
129 | #define WIDE_NEWLINE_CHAR L'\n' |
130 | |
131 | /* Rename to standard API for using out of glibc. */ |
132 | #ifndef _LIBC |
133 | # undef __wctype |
134 | # undef __iswalnum |
135 | # undef __iswctype |
136 | # undef __towlower |
137 | # undef __towupper |
138 | # define __wctype wctype |
139 | # define __iswalnum iswalnum |
140 | # define __iswctype iswctype |
141 | # define __towlower towlower |
142 | # define __towupper towupper |
143 | # define __btowc btowc |
144 | # define __mbrtowc mbrtowc |
145 | # define __wcrtomb wcrtomb |
146 | # define __regfree regfree |
147 | #endif /* not _LIBC */ |
148 | |
149 | #ifndef SSIZE_MAX |
150 | # define SSIZE_MAX ((ssize_t) (SIZE_MAX / 2)) |
151 | #endif |
152 | #ifndef ULONG_WIDTH |
153 | # define ULONG_WIDTH REGEX_UINTEGER_WIDTH (ULONG_MAX) |
154 | /* The number of usable bits in an unsigned integer type with maximum |
155 | value MAX, as an int expression suitable in #if. Cover all known |
156 | practical hosts. This implementation exploits the fact that MAX is |
157 | 1 less than a power of 2, and merely counts the number of 1 bits in |
158 | MAX; "COBn" means "count the number of 1 bits in the low-order n bits". */ |
159 | # define REGEX_UINTEGER_WIDTH(max) REGEX_COB128 (max) |
160 | # define REGEX_COB128(n) (REGEX_COB64 ((n) >> 31 >> 31 >> 2) + REGEX_COB64 (n)) |
161 | # define REGEX_COB64(n) (REGEX_COB32 ((n) >> 31 >> 1) + REGEX_COB32 (n)) |
162 | # define REGEX_COB32(n) (REGEX_COB16 ((n) >> 16) + REGEX_COB16 (n)) |
163 | # define REGEX_COB16(n) (REGEX_COB8 ((n) >> 8) + REGEX_COB8 (n)) |
164 | # define REGEX_COB8(n) (REGEX_COB4 ((n) >> 4) + REGEX_COB4 (n)) |
165 | # define REGEX_COB4(n) (!!((n) & 8) + !!((n) & 4) + !!((n) & 2) + ((n) & 1)) |
166 | # if ULONG_MAX / 2 + 1 != 1ul << (ULONG_WIDTH - 1) |
167 | # error "ULONG_MAX out of range" |
168 | # endif |
169 | #endif |
170 | |
171 | /* The type of indexes into strings. This is signed, not size_t, |
172 | since the API requires indexes to fit in regoff_t anyway, and using |
173 | signed integers makes the code a bit smaller and presumably faster. |
174 | The traditional GNU regex implementation uses int for indexes. |
175 | The POSIX-compatible implementation uses a possibly-wider type. |
176 | The name 'Idx' is three letters to minimize the hassle of |
177 | reindenting a lot of regex code that formerly used 'int'. */ |
178 | typedef regoff_t Idx; |
179 | #ifdef _REGEX_LARGE_OFFSETS |
180 | # define IDX_MAX SSIZE_MAX |
181 | #else |
182 | # define IDX_MAX INT_MAX |
183 | #endif |
184 | |
185 | /* A hash value, suitable for computing hash tables. */ |
186 | typedef __re_size_t re_hashval_t; |
187 | |
188 | /* An integer used to represent a set of bits. It must be unsigned, |
189 | and must be at least as wide as unsigned int. */ |
190 | typedef unsigned long int bitset_word_t; |
191 | /* All bits set in a bitset_word_t. */ |
192 | #define BITSET_WORD_MAX ULONG_MAX |
193 | /* Number of bits in a bitset_word_t. */ |
194 | #define BITSET_WORD_BITS ULONG_WIDTH |
195 | |
196 | /* Number of bitset_word_t values in a bitset_t. */ |
197 | #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) |
198 | |
199 | typedef bitset_word_t bitset_t[BITSET_WORDS]; |
200 | typedef bitset_word_t *re_bitset_ptr_t; |
201 | typedef const bitset_word_t *re_const_bitset_ptr_t; |
202 | |
203 | #define PREV_WORD_CONSTRAINT 0x0001 |
204 | #define PREV_NOTWORD_CONSTRAINT 0x0002 |
205 | #define NEXT_WORD_CONSTRAINT 0x0004 |
206 | #define NEXT_NOTWORD_CONSTRAINT 0x0008 |
207 | #define PREV_NEWLINE_CONSTRAINT 0x0010 |
208 | #define NEXT_NEWLINE_CONSTRAINT 0x0020 |
209 | #define PREV_BEGBUF_CONSTRAINT 0x0040 |
210 | #define NEXT_ENDBUF_CONSTRAINT 0x0080 |
211 | #define WORD_DELIM_CONSTRAINT 0x0100 |
212 | #define NOT_WORD_DELIM_CONSTRAINT 0x0200 |
213 | |
214 | typedef enum |
215 | { |
216 | INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, |
217 | WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, |
218 | WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, |
219 | INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, |
220 | LINE_FIRST = PREV_NEWLINE_CONSTRAINT, |
221 | LINE_LAST = NEXT_NEWLINE_CONSTRAINT, |
222 | BUF_FIRST = PREV_BEGBUF_CONSTRAINT, |
223 | BUF_LAST = NEXT_ENDBUF_CONSTRAINT, |
224 | WORD_DELIM = WORD_DELIM_CONSTRAINT, |
225 | NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT |
226 | } re_context_type; |
227 | |
228 | typedef struct |
229 | { |
230 | Idx alloc; |
231 | Idx nelem; |
232 | Idx *elems; |
233 | } re_node_set; |
234 | |
235 | typedef enum |
236 | { |
237 | NON_TYPE = 0, |
238 | |
239 | /* Node type, These are used by token, node, tree. */ |
240 | CHARACTER = 1, |
241 | END_OF_RE = 2, |
242 | SIMPLE_BRACKET = 3, |
243 | OP_BACK_REF = 4, |
244 | OP_PERIOD = 5, |
245 | #ifdef RE_ENABLE_I18N |
246 | COMPLEX_BRACKET = 6, |
247 | OP_UTF8_PERIOD = 7, |
248 | #endif /* RE_ENABLE_I18N */ |
249 | |
250 | /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used |
251 | when the debugger shows values of this enum type. */ |
252 | #define EPSILON_BIT 8 |
253 | OP_OPEN_SUBEXP = EPSILON_BIT | 0, |
254 | OP_CLOSE_SUBEXP = EPSILON_BIT | 1, |
255 | OP_ALT = EPSILON_BIT | 2, |
256 | OP_DUP_ASTERISK = EPSILON_BIT | 3, |
257 | ANCHOR = EPSILON_BIT | 4, |
258 | |
259 | /* Tree type, these are used only by tree. */ |
260 | CONCAT = 16, |
261 | SUBEXP = 17, |
262 | |
263 | /* Token type, these are used only by token. */ |
264 | OP_DUP_PLUS = 18, |
265 | OP_DUP_QUESTION, |
266 | OP_OPEN_BRACKET, |
267 | OP_CLOSE_BRACKET, |
268 | OP_CHARSET_RANGE, |
269 | OP_OPEN_DUP_NUM, |
270 | OP_CLOSE_DUP_NUM, |
271 | OP_NON_MATCH_LIST, |
272 | OP_OPEN_COLL_ELEM, |
273 | OP_CLOSE_COLL_ELEM, |
274 | OP_OPEN_EQUIV_CLASS, |
275 | OP_CLOSE_EQUIV_CLASS, |
276 | OP_OPEN_CHAR_CLASS, |
277 | OP_CLOSE_CHAR_CLASS, |
278 | OP_WORD, |
279 | OP_NOTWORD, |
280 | OP_SPACE, |
281 | OP_NOTSPACE, |
282 | BACK_SLASH |
283 | |
284 | } re_token_type_t; |
285 | |
286 | #ifdef RE_ENABLE_I18N |
287 | typedef struct |
288 | { |
289 | /* Multibyte characters. */ |
290 | wchar_t *mbchars; |
291 | |
292 | /* Collating symbols. */ |
293 | # ifdef _LIBC |
294 | int32_t *coll_syms; |
295 | # endif |
296 | |
297 | /* Equivalence classes. */ |
298 | # ifdef _LIBC |
299 | int32_t *equiv_classes; |
300 | # endif |
301 | |
302 | /* Range expressions. */ |
303 | # ifdef _LIBC |
304 | uint32_t *range_starts; |
305 | uint32_t *range_ends; |
306 | # else /* not _LIBC */ |
307 | wchar_t *range_starts; |
308 | wchar_t *range_ends; |
309 | # endif /* not _LIBC */ |
310 | |
311 | /* Character classes. */ |
312 | wctype_t *char_classes; |
313 | |
314 | /* If this character set is the non-matching list. */ |
315 | unsigned int non_match : 1; |
316 | |
317 | /* # of multibyte characters. */ |
318 | Idx nmbchars; |
319 | |
320 | /* # of collating symbols. */ |
321 | Idx ncoll_syms; |
322 | |
323 | /* # of equivalence classes. */ |
324 | Idx nequiv_classes; |
325 | |
326 | /* # of range expressions. */ |
327 | Idx nranges; |
328 | |
329 | /* # of character classes. */ |
330 | Idx nchar_classes; |
331 | } re_charset_t; |
332 | #endif /* RE_ENABLE_I18N */ |
333 | |
334 | typedef struct |
335 | { |
336 | union |
337 | { |
338 | unsigned char c; /* for CHARACTER */ |
339 | re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ |
340 | #ifdef RE_ENABLE_I18N |
341 | re_charset_t *mbcset; /* for COMPLEX_BRACKET */ |
342 | #endif /* RE_ENABLE_I18N */ |
343 | Idx idx; /* for BACK_REF */ |
344 | re_context_type ctx_type; /* for ANCHOR */ |
345 | } opr; |
346 | #if (__GNUC__ >= 2 || defined __clang__) && !defined __STRICT_ANSI__ |
347 | re_token_type_t type : 8; |
348 | #else |
349 | re_token_type_t type; |
350 | #endif |
351 | unsigned int constraint : 10; /* context constraint */ |
352 | unsigned int duplicated : 1; |
353 | unsigned int opt_subexp : 1; |
354 | #ifdef RE_ENABLE_I18N |
355 | unsigned int accept_mb : 1; |
356 | /* These 2 bits can be moved into the union if needed (e.g. if running out |
357 | of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ |
358 | unsigned int mb_partial : 1; |
359 | #endif |
360 | unsigned int word_char : 1; |
361 | } re_token_t; |
362 | |
363 | #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) |
364 | |
365 | struct re_string_t |
366 | { |
367 | /* Indicate the raw buffer which is the original string passed as an |
368 | argument of regexec(), re_search(), etc.. */ |
369 | const unsigned char *raw_mbs; |
370 | /* Store the multibyte string. In case of "case insensitive mode" like |
371 | REG_ICASE, upper cases of the string are stored, otherwise MBS points |
372 | the same address that RAW_MBS points. */ |
373 | unsigned char *mbs; |
374 | #ifdef RE_ENABLE_I18N |
375 | /* Store the wide character string which is corresponding to MBS. */ |
376 | wint_t *wcs; |
377 | Idx *offsets; |
378 | mbstate_t cur_state; |
379 | #endif |
380 | /* Index in RAW_MBS. Each character mbs[i] corresponds to |
381 | raw_mbs[raw_mbs_idx + i]. */ |
382 | Idx raw_mbs_idx; |
383 | /* The length of the valid characters in the buffers. */ |
384 | Idx valid_len; |
385 | /* The corresponding number of bytes in raw_mbs array. */ |
386 | Idx valid_raw_len; |
387 | /* The length of the buffers MBS and WCS. */ |
388 | Idx bufs_len; |
389 | /* The index in MBS, which is updated by re_string_fetch_byte. */ |
390 | Idx cur_idx; |
391 | /* length of RAW_MBS array. */ |
392 | Idx raw_len; |
393 | /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ |
394 | Idx len; |
395 | /* End of the buffer may be shorter than its length in the cases such |
396 | as re_match_2, re_search_2. Then, we use STOP for end of the buffer |
397 | instead of LEN. */ |
398 | Idx raw_stop; |
399 | /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ |
400 | Idx stop; |
401 | |
402 | /* The context of mbs[0]. We store the context independently, since |
403 | the context of mbs[0] may be different from raw_mbs[0], which is |
404 | the beginning of the input string. */ |
405 | unsigned int tip_context; |
406 | /* The translation passed as a part of an argument of re_compile_pattern. */ |
407 | RE_TRANSLATE_TYPE trans; |
408 | /* Copy of re_dfa_t's word_char. */ |
409 | re_const_bitset_ptr_t word_char; |
410 | /* true if REG_ICASE. */ |
411 | unsigned char icase; |
412 | unsigned char is_utf8; |
413 | unsigned char map_notascii; |
414 | unsigned char mbs_allocated; |
415 | unsigned char offsets_needed; |
416 | unsigned char newline_anchor; |
417 | unsigned char word_ops_used; |
418 | int mb_cur_max; |
419 | }; |
420 | typedef struct re_string_t re_string_t; |
421 | |
422 | |
423 | struct re_dfa_t; |
424 | typedef struct re_dfa_t re_dfa_t; |
425 | |
426 | #ifndef _LIBC |
427 | # define IS_IN(libc) false |
428 | #endif |
429 | |
430 | #define re_string_peek_byte(pstr, offset) \ |
431 | ((pstr)->mbs[(pstr)->cur_idx + offset]) |
432 | #define re_string_fetch_byte(pstr) \ |
433 | ((pstr)->mbs[(pstr)->cur_idx++]) |
434 | #define re_string_first_byte(pstr, idx) \ |
435 | ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) |
436 | #define re_string_is_single_byte_char(pstr, idx) \ |
437 | ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ |
438 | || (pstr)->wcs[(idx) + 1] != WEOF)) |
439 | #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) |
440 | #define re_string_cur_idx(pstr) ((pstr)->cur_idx) |
441 | #define re_string_get_buffer(pstr) ((pstr)->mbs) |
442 | #define re_string_length(pstr) ((pstr)->len) |
443 | #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) |
444 | #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) |
445 | #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) |
446 | |
447 | #if defined _LIBC || HAVE_ALLOCA |
448 | # include <alloca.h> |
449 | #endif |
450 | |
451 | #ifndef _LIBC |
452 | # if HAVE_ALLOCA |
453 | /* The OS usually guarantees only one guard page at the bottom of the stack, |
454 | and a page size can be as small as 4096 bytes. So we cannot safely |
455 | allocate anything larger than 4096 bytes. Also care for the possibility |
456 | of a few compiler-allocated temporary stack slots. */ |
457 | # define __libc_use_alloca(n) ((n) < 4032) |
458 | # else |
459 | /* alloca is implemented with malloc, so just use malloc. */ |
460 | # define __libc_use_alloca(n) 0 |
461 | # undef alloca |
462 | # define alloca(n) malloc (n) |
463 | # endif |
464 | #endif |
465 | |
466 | #ifdef _LIBC |
467 | # define MALLOC_0_IS_NONNULL 1 |
468 | #elif !defined MALLOC_0_IS_NONNULL |
469 | # define MALLOC_0_IS_NONNULL 0 |
470 | #endif |
471 | |
472 | #ifndef MAX |
473 | # define MAX(a,b) ((a) < (b) ? (b) : (a)) |
474 | #endif |
475 | #ifndef MIN |
476 | # define MIN(a,b) ((a) < (b) ? (a) : (b)) |
477 | #endif |
478 | |
479 | #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) |
480 | #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) |
481 | #define re_free(p) free (p) |
482 | |
483 | struct bin_tree_t |
484 | { |
485 | struct bin_tree_t *parent; |
486 | struct bin_tree_t *left; |
487 | struct bin_tree_t *right; |
488 | struct bin_tree_t *first; |
489 | struct bin_tree_t *next; |
490 | |
491 | re_token_t token; |
492 | |
493 | /* 'node_idx' is the index in dfa->nodes, if 'type' == 0. |
494 | Otherwise 'type' indicate the type of this node. */ |
495 | Idx node_idx; |
496 | }; |
497 | typedef struct bin_tree_t bin_tree_t; |
498 | |
499 | #define BIN_TREE_STORAGE_SIZE \ |
500 | ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) |
501 | |
502 | struct bin_tree_storage_t |
503 | { |
504 | struct bin_tree_storage_t *next; |
505 | bin_tree_t data[BIN_TREE_STORAGE_SIZE]; |
506 | }; |
507 | typedef struct bin_tree_storage_t bin_tree_storage_t; |
508 | |
509 | #define CONTEXT_WORD 1 |
510 | #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) |
511 | #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) |
512 | #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) |
513 | |
514 | #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) |
515 | #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) |
516 | #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) |
517 | #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) |
518 | #define IS_ORDINARY_CONTEXT(c) ((c) == 0) |
519 | |
520 | #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') |
521 | #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) |
522 | #define IS_WIDE_WORD_CHAR(ch) (__iswalnum (ch) || (ch) == L'_') |
523 | #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) |
524 | |
525 | #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ |
526 | ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ |
527 | || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ |
528 | || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ |
529 | || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) |
530 | |
531 | #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ |
532 | ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ |
533 | || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ |
534 | || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ |
535 | || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) |
536 | |
537 | struct re_dfastate_t |
538 | { |
539 | re_hashval_t hash; |
540 | re_node_set nodes; |
541 | re_node_set non_eps_nodes; |
542 | re_node_set inveclosure; |
543 | re_node_set *entrance_nodes; |
544 | struct re_dfastate_t **trtable, **word_trtable; |
545 | unsigned int context : 4; |
546 | unsigned int halt : 1; |
547 | /* If this state can accept "multi byte". |
548 | Note that we refer to multibyte characters, and multi character |
549 | collating elements as "multi byte". */ |
550 | unsigned int accept_mb : 1; |
551 | /* If this state has backreference node(s). */ |
552 | unsigned int has_backref : 1; |
553 | unsigned int has_constraint : 1; |
554 | }; |
555 | typedef struct re_dfastate_t re_dfastate_t; |
556 | |
557 | struct re_state_table_entry |
558 | { |
559 | Idx num; |
560 | Idx alloc; |
561 | re_dfastate_t **array; |
562 | }; |
563 | |
564 | /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ |
565 | |
566 | typedef struct |
567 | { |
568 | Idx next_idx; |
569 | Idx alloc; |
570 | re_dfastate_t **array; |
571 | } state_array_t; |
572 | |
573 | /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ |
574 | |
575 | typedef struct |
576 | { |
577 | Idx node; |
578 | Idx str_idx; /* The position NODE match at. */ |
579 | state_array_t path; |
580 | } re_sub_match_last_t; |
581 | |
582 | /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. |
583 | And information about the node, whose type is OP_CLOSE_SUBEXP, |
584 | corresponding to NODE is stored in LASTS. */ |
585 | |
586 | typedef struct |
587 | { |
588 | Idx str_idx; |
589 | Idx node; |
590 | state_array_t *path; |
591 | Idx alasts; /* Allocation size of LASTS. */ |
592 | Idx nlasts; /* The number of LASTS. */ |
593 | re_sub_match_last_t **lasts; |
594 | } re_sub_match_top_t; |
595 | |
596 | struct re_backref_cache_entry |
597 | { |
598 | Idx node; |
599 | Idx str_idx; |
600 | Idx subexp_from; |
601 | Idx subexp_to; |
602 | bitset_word_t eps_reachable_subexps_map; |
603 | char more; |
604 | }; |
605 | |
606 | typedef struct |
607 | { |
608 | /* The string object corresponding to the input string. */ |
609 | re_string_t input; |
610 | const re_dfa_t *const dfa; |
611 | /* EFLAGS of the argument of regexec. */ |
612 | int eflags; |
613 | /* Where the matching ends. */ |
614 | Idx match_last; |
615 | Idx last_node; |
616 | /* The state log used by the matcher. */ |
617 | re_dfastate_t **state_log; |
618 | Idx state_log_top; |
619 | /* Back reference cache. */ |
620 | Idx nbkref_ents; |
621 | Idx abkref_ents; |
622 | struct re_backref_cache_entry *bkref_ents; |
623 | int max_mb_elem_len; |
624 | Idx nsub_tops; |
625 | Idx asub_tops; |
626 | re_sub_match_top_t **sub_tops; |
627 | } re_match_context_t; |
628 | |
629 | typedef struct |
630 | { |
631 | re_dfastate_t **sifted_states; |
632 | re_dfastate_t **limited_states; |
633 | Idx last_node; |
634 | Idx last_str_idx; |
635 | re_node_set limits; |
636 | } re_sift_context_t; |
637 | |
638 | struct re_fail_stack_ent_t |
639 | { |
640 | Idx idx; |
641 | Idx node; |
642 | regmatch_t *regs; |
643 | re_node_set eps_via_nodes; |
644 | }; |
645 | |
646 | struct re_fail_stack_t |
647 | { |
648 | Idx num; |
649 | Idx alloc; |
650 | struct re_fail_stack_ent_t *stack; |
651 | }; |
652 | |
653 | struct re_dfa_t |
654 | { |
655 | re_token_t *nodes; |
656 | size_t nodes_alloc; |
657 | size_t nodes_len; |
658 | Idx *nexts; |
659 | Idx *org_indices; |
660 | re_node_set *edests; |
661 | re_node_set *eclosures; |
662 | re_node_set *inveclosures; |
663 | struct re_state_table_entry *state_table; |
664 | re_dfastate_t *init_state; |
665 | re_dfastate_t *init_state_word; |
666 | re_dfastate_t *init_state_nl; |
667 | re_dfastate_t *init_state_begbuf; |
668 | bin_tree_t *str_tree; |
669 | bin_tree_storage_t *str_tree_storage; |
670 | re_bitset_ptr_t sb_char; |
671 | int str_tree_storage_idx; |
672 | |
673 | /* number of subexpressions 're_nsub' is in regex_t. */ |
674 | re_hashval_t state_hash_mask; |
675 | Idx init_node; |
676 | Idx nbackref; /* The number of backreference in this dfa. */ |
677 | |
678 | /* Bitmap expressing which backreference is used. */ |
679 | bitset_word_t used_bkref_map; |
680 | bitset_word_t completed_bkref_map; |
681 | |
682 | unsigned int has_plural_match : 1; |
683 | /* If this dfa has "multibyte node", which is a backreference or |
684 | a node which can accept multibyte character or multi character |
685 | collating element. */ |
686 | unsigned int has_mb_node : 1; |
687 | unsigned int is_utf8 : 1; |
688 | unsigned int map_notascii : 1; |
689 | unsigned int word_ops_used : 1; |
690 | int mb_cur_max; |
691 | bitset_t word_char; |
692 | reg_syntax_t syntax; |
693 | Idx *subexp_map; |
694 | #ifdef DEBUG |
695 | char* re_str; |
696 | #endif |
697 | lock_define (lock) |
698 | }; |
699 | |
700 | #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) |
701 | #define re_node_set_remove(set,id) \ |
702 | (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) |
703 | #define re_node_set_empty(p) ((p)->nelem = 0) |
704 | #define re_node_set_free(set) re_free ((set)->elems) |
705 | |
706 | |
707 | typedef enum |
708 | { |
709 | SB_CHAR, |
710 | MB_CHAR, |
711 | EQUIV_CLASS, |
712 | COLL_SYM, |
713 | CHAR_CLASS |
714 | } bracket_elem_type; |
715 | |
716 | typedef struct |
717 | { |
718 | bracket_elem_type type; |
719 | union |
720 | { |
721 | unsigned char ch; |
722 | unsigned char *name; |
723 | wchar_t wch; |
724 | } opr; |
725 | } bracket_elem_t; |
726 | |
727 | |
728 | /* Functions for bitset_t operation. */ |
729 | |
730 | static inline void |
731 | bitset_set (bitset_t set, Idx i) |
732 | { |
733 | set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; |
734 | } |
735 | |
736 | static inline void |
737 | bitset_clear (bitset_t set, Idx i) |
738 | { |
739 | set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); |
740 | } |
741 | |
742 | static inline bool |
743 | bitset_contain (const bitset_t set, Idx i) |
744 | { |
745 | return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; |
746 | } |
747 | |
748 | static inline void |
749 | bitset_empty (bitset_t set) |
750 | { |
751 | memset (set, '\0', sizeof (bitset_t)); |
752 | } |
753 | |
754 | static inline void |
755 | bitset_set_all (bitset_t set) |
756 | { |
757 | memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); |
758 | if (SBC_MAX % BITSET_WORD_BITS != 0) |
759 | set[BITSET_WORDS - 1] = |
760 | ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; |
761 | } |
762 | |
763 | static inline void |
764 | bitset_copy (bitset_t dest, const bitset_t src) |
765 | { |
766 | memcpy (dest, src, sizeof (bitset_t)); |
767 | } |
768 | |
769 | static inline void |
770 | bitset_not (bitset_t set) |
771 | { |
772 | int bitset_i; |
773 | for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) |
774 | set[bitset_i] = ~set[bitset_i]; |
775 | if (SBC_MAX % BITSET_WORD_BITS != 0) |
776 | set[BITSET_WORDS - 1] = |
777 | ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) |
778 | & ~set[BITSET_WORDS - 1]); |
779 | } |
780 | |
781 | static inline void |
782 | bitset_merge (bitset_t dest, const bitset_t src) |
783 | { |
784 | int bitset_i; |
785 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) |
786 | dest[bitset_i] |= src[bitset_i]; |
787 | } |
788 | |
789 | static inline void |
790 | bitset_mask (bitset_t dest, const bitset_t src) |
791 | { |
792 | int bitset_i; |
793 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) |
794 | dest[bitset_i] &= src[bitset_i]; |
795 | } |
796 | |
797 | #ifdef RE_ENABLE_I18N |
798 | /* Functions for re_string. */ |
799 | static int |
800 | __attribute__ ((pure, unused)) |
801 | re_string_char_size_at (const re_string_t *pstr, Idx idx) |
802 | { |
803 | int byte_idx; |
804 | if (pstr->mb_cur_max == 1) |
805 | return 1; |
806 | for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) |
807 | if (pstr->wcs[idx + byte_idx] != WEOF) |
808 | break; |
809 | return byte_idx; |
810 | } |
811 | |
812 | static wint_t |
813 | __attribute__ ((pure, unused)) |
814 | re_string_wchar_at (const re_string_t *pstr, Idx idx) |
815 | { |
816 | if (pstr->mb_cur_max == 1) |
817 | return (wint_t) pstr->mbs[idx]; |
818 | return (wint_t) pstr->wcs[idx]; |
819 | } |
820 | |
821 | # ifdef _LIBC |
822 | # include <locale/weight.h> |
823 | # endif |
824 | |
825 | static int |
826 | __attribute__ ((pure, unused)) |
827 | re_string_elem_size_at (const re_string_t *pstr, Idx idx) |
828 | { |
829 | # ifdef _LIBC |
830 | const unsigned char *p, *; |
831 | const int32_t *table, *indirect; |
832 | uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); |
833 | |
834 | if (nrules != 0) |
835 | { |
836 | table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); |
837 | extra = (const unsigned char *) |
838 | _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); |
839 | indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, |
840 | _NL_COLLATE_INDIRECTMB); |
841 | p = pstr->mbs + idx; |
842 | findidx (table, indirect, extra, &p, pstr->len - idx); |
843 | return p - pstr->mbs - idx; |
844 | } |
845 | else |
846 | # endif /* _LIBC */ |
847 | return 1; |
848 | } |
849 | #endif /* RE_ENABLE_I18N */ |
850 | |
851 | #ifdef _LIBC |
852 | # if __GNUC__ >= 7 |
853 | # define FALLTHROUGH __attribute__ ((__fallthrough__)) |
854 | # else |
855 | # define FALLTHROUGH ((void) 0) |
856 | # endif |
857 | #else |
858 | # include "attribute.h" |
859 | #endif |
860 | |
861 | #endif /* _REGEX_INTERNAL_H */ |
862 | |