1 | /* Extended regular expression matching and search library. |
2 | Copyright (C) 2002-2020 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 | #ifdef _LIBC |
81 | # ifndef _RE_DEFINE_LOCALE_FUNCTIONS |
82 | # define _RE_DEFINE_LOCALE_FUNCTIONS 1 |
83 | # include <locale/localeinfo.h> |
84 | # include <locale/coll-lookup.h> |
85 | # endif |
86 | #endif |
87 | |
88 | /* This is for other GNU distributions with internationalized messages. */ |
89 | #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC |
90 | # include <libintl.h> |
91 | # ifdef _LIBC |
92 | # undef gettext |
93 | # define gettext(msgid) \ |
94 | __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES) |
95 | # endif |
96 | #else |
97 | # undef gettext |
98 | # define gettext(msgid) (msgid) |
99 | #endif |
100 | |
101 | #ifndef gettext_noop |
102 | /* This define is so xgettext can find the internationalizable |
103 | strings. */ |
104 | # define gettext_noop(String) String |
105 | #endif |
106 | |
107 | #if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE) || _LIBC |
108 | # define RE_ENABLE_I18N |
109 | #endif |
110 | |
111 | /* Number of ASCII characters. */ |
112 | #define ASCII_CHARS 0x80 |
113 | |
114 | /* Number of single byte characters. */ |
115 | #define SBC_MAX (UCHAR_MAX + 1) |
116 | |
117 | #define COLL_ELEM_LEN_MAX 8 |
118 | |
119 | /* The character which represents newline. */ |
120 | #define NEWLINE_CHAR '\n' |
121 | #define WIDE_NEWLINE_CHAR L'\n' |
122 | |
123 | /* Rename to standard API for using out of glibc. */ |
124 | #ifndef _LIBC |
125 | # undef __wctype |
126 | # undef __iswalnum |
127 | # undef __iswctype |
128 | # undef __towlower |
129 | # undef __towupper |
130 | # define __wctype wctype |
131 | # define __iswalnum iswalnum |
132 | # define __iswctype iswctype |
133 | # define __towlower towlower |
134 | # define __towupper towupper |
135 | # define __btowc btowc |
136 | # define __mbrtowc mbrtowc |
137 | # define __wcrtomb wcrtomb |
138 | # define __regfree regfree |
139 | #endif /* not _LIBC */ |
140 | |
141 | #ifndef SSIZE_MAX |
142 | # define SSIZE_MAX ((ssize_t) (SIZE_MAX / 2)) |
143 | #endif |
144 | |
145 | /* The type of indexes into strings. This is signed, not size_t, |
146 | since the API requires indexes to fit in regoff_t anyway, and using |
147 | signed integers makes the code a bit smaller and presumably faster. |
148 | The traditional GNU regex implementation uses int for indexes. |
149 | The POSIX-compatible implementation uses a possibly-wider type. |
150 | The name 'Idx' is three letters to minimize the hassle of |
151 | reindenting a lot of regex code that formerly used 'int'. */ |
152 | typedef regoff_t Idx; |
153 | #ifdef _REGEX_LARGE_OFFSETS |
154 | # define IDX_MAX SSIZE_MAX |
155 | #else |
156 | # define IDX_MAX INT_MAX |
157 | #endif |
158 | |
159 | /* A hash value, suitable for computing hash tables. */ |
160 | typedef __re_size_t re_hashval_t; |
161 | |
162 | /* An integer used to represent a set of bits. It must be unsigned, |
163 | and must be at least as wide as unsigned int. */ |
164 | typedef unsigned long int bitset_word_t; |
165 | /* All bits set in a bitset_word_t. */ |
166 | #define BITSET_WORD_MAX ULONG_MAX |
167 | |
168 | /* Number of bits in a bitset_word_t. For portability to hosts with |
169 | padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)'; |
170 | instead, deduce it directly from BITSET_WORD_MAX. Avoid |
171 | greater-than-32-bit integers and unconditional shifts by more than |
172 | 31 bits, as they're not portable. */ |
173 | #if BITSET_WORD_MAX == 0xffffffffUL |
174 | # define BITSET_WORD_BITS 32 |
175 | #elif BITSET_WORD_MAX >> 31 >> 4 == 1 |
176 | # define BITSET_WORD_BITS 36 |
177 | #elif BITSET_WORD_MAX >> 31 >> 16 == 1 |
178 | # define BITSET_WORD_BITS 48 |
179 | #elif BITSET_WORD_MAX >> 31 >> 28 == 1 |
180 | # define BITSET_WORD_BITS 60 |
181 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1 |
182 | # define BITSET_WORD_BITS 64 |
183 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1 |
184 | # define BITSET_WORD_BITS 72 |
185 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1 |
186 | # define BITSET_WORD_BITS 128 |
187 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1 |
188 | # define BITSET_WORD_BITS 256 |
189 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1 |
190 | # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */ |
191 | # if BITSET_WORD_BITS <= SBC_MAX |
192 | # error "Invalid SBC_MAX" |
193 | # endif |
194 | #else |
195 | # error "Add case for new bitset_word_t size" |
196 | #endif |
197 | |
198 | /* Number of bitset_word_t values in a bitset_t. */ |
199 | #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) |
200 | |
201 | typedef bitset_word_t bitset_t[BITSET_WORDS]; |
202 | typedef bitset_word_t *re_bitset_ptr_t; |
203 | typedef const bitset_word_t *re_const_bitset_ptr_t; |
204 | |
205 | #define PREV_WORD_CONSTRAINT 0x0001 |
206 | #define PREV_NOTWORD_CONSTRAINT 0x0002 |
207 | #define NEXT_WORD_CONSTRAINT 0x0004 |
208 | #define NEXT_NOTWORD_CONSTRAINT 0x0008 |
209 | #define PREV_NEWLINE_CONSTRAINT 0x0010 |
210 | #define NEXT_NEWLINE_CONSTRAINT 0x0020 |
211 | #define PREV_BEGBUF_CONSTRAINT 0x0040 |
212 | #define NEXT_ENDBUF_CONSTRAINT 0x0080 |
213 | #define WORD_DELIM_CONSTRAINT 0x0100 |
214 | #define NOT_WORD_DELIM_CONSTRAINT 0x0200 |
215 | |
216 | typedef enum |
217 | { |
218 | INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, |
219 | WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, |
220 | WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, |
221 | INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, |
222 | LINE_FIRST = PREV_NEWLINE_CONSTRAINT, |
223 | LINE_LAST = NEXT_NEWLINE_CONSTRAINT, |
224 | BUF_FIRST = PREV_BEGBUF_CONSTRAINT, |
225 | BUF_LAST = NEXT_ENDBUF_CONSTRAINT, |
226 | WORD_DELIM = WORD_DELIM_CONSTRAINT, |
227 | NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT |
228 | } re_context_type; |
229 | |
230 | typedef struct |
231 | { |
232 | Idx alloc; |
233 | Idx nelem; |
234 | Idx *elems; |
235 | } re_node_set; |
236 | |
237 | typedef enum |
238 | { |
239 | NON_TYPE = 0, |
240 | |
241 | /* Node type, These are used by token, node, tree. */ |
242 | CHARACTER = 1, |
243 | END_OF_RE = 2, |
244 | SIMPLE_BRACKET = 3, |
245 | OP_BACK_REF = 4, |
246 | OP_PERIOD = 5, |
247 | #ifdef RE_ENABLE_I18N |
248 | COMPLEX_BRACKET = 6, |
249 | OP_UTF8_PERIOD = 7, |
250 | #endif /* RE_ENABLE_I18N */ |
251 | |
252 | /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used |
253 | when the debugger shows values of this enum type. */ |
254 | #define EPSILON_BIT 8 |
255 | OP_OPEN_SUBEXP = EPSILON_BIT | 0, |
256 | OP_CLOSE_SUBEXP = EPSILON_BIT | 1, |
257 | OP_ALT = EPSILON_BIT | 2, |
258 | OP_DUP_ASTERISK = EPSILON_BIT | 3, |
259 | ANCHOR = EPSILON_BIT | 4, |
260 | |
261 | /* Tree type, these are used only by tree. */ |
262 | CONCAT = 16, |
263 | SUBEXP = 17, |
264 | |
265 | /* Token type, these are used only by token. */ |
266 | OP_DUP_PLUS = 18, |
267 | OP_DUP_QUESTION, |
268 | OP_OPEN_BRACKET, |
269 | OP_CLOSE_BRACKET, |
270 | OP_CHARSET_RANGE, |
271 | OP_OPEN_DUP_NUM, |
272 | OP_CLOSE_DUP_NUM, |
273 | OP_NON_MATCH_LIST, |
274 | OP_OPEN_COLL_ELEM, |
275 | OP_CLOSE_COLL_ELEM, |
276 | OP_OPEN_EQUIV_CLASS, |
277 | OP_CLOSE_EQUIV_CLASS, |
278 | OP_OPEN_CHAR_CLASS, |
279 | OP_CLOSE_CHAR_CLASS, |
280 | OP_WORD, |
281 | OP_NOTWORD, |
282 | OP_SPACE, |
283 | OP_NOTSPACE, |
284 | BACK_SLASH |
285 | |
286 | } re_token_type_t; |
287 | |
288 | #ifdef RE_ENABLE_I18N |
289 | typedef struct |
290 | { |
291 | /* Multibyte characters. */ |
292 | wchar_t *mbchars; |
293 | |
294 | /* Collating symbols. */ |
295 | # ifdef _LIBC |
296 | int32_t *coll_syms; |
297 | # endif |
298 | |
299 | /* Equivalence classes. */ |
300 | # ifdef _LIBC |
301 | int32_t *equiv_classes; |
302 | # endif |
303 | |
304 | /* Range expressions. */ |
305 | # ifdef _LIBC |
306 | uint32_t *range_starts; |
307 | uint32_t *range_ends; |
308 | # else /* not _LIBC */ |
309 | wchar_t *range_starts; |
310 | wchar_t *range_ends; |
311 | # endif /* not _LIBC */ |
312 | |
313 | /* Character classes. */ |
314 | wctype_t *char_classes; |
315 | |
316 | /* If this character set is the non-matching list. */ |
317 | unsigned int non_match : 1; |
318 | |
319 | /* # of multibyte characters. */ |
320 | Idx nmbchars; |
321 | |
322 | /* # of collating symbols. */ |
323 | Idx ncoll_syms; |
324 | |
325 | /* # of equivalence classes. */ |
326 | Idx nequiv_classes; |
327 | |
328 | /* # of range expressions. */ |
329 | Idx nranges; |
330 | |
331 | /* # of character classes. */ |
332 | Idx nchar_classes; |
333 | } re_charset_t; |
334 | #endif /* RE_ENABLE_I18N */ |
335 | |
336 | typedef struct |
337 | { |
338 | union |
339 | { |
340 | unsigned char c; /* for CHARACTER */ |
341 | re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ |
342 | #ifdef RE_ENABLE_I18N |
343 | re_charset_t *mbcset; /* for COMPLEX_BRACKET */ |
344 | #endif /* RE_ENABLE_I18N */ |
345 | Idx idx; /* for BACK_REF */ |
346 | re_context_type ctx_type; /* for ANCHOR */ |
347 | } opr; |
348 | #if __GNUC__ >= 2 && !defined __STRICT_ANSI__ |
349 | re_token_type_t type : 8; |
350 | #else |
351 | re_token_type_t type; |
352 | #endif |
353 | unsigned int constraint : 10; /* context constraint */ |
354 | unsigned int duplicated : 1; |
355 | unsigned int opt_subexp : 1; |
356 | #ifdef RE_ENABLE_I18N |
357 | unsigned int accept_mb : 1; |
358 | /* These 2 bits can be moved into the union if needed (e.g. if running out |
359 | of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ |
360 | unsigned int mb_partial : 1; |
361 | #endif |
362 | unsigned int word_char : 1; |
363 | } re_token_t; |
364 | |
365 | #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) |
366 | |
367 | struct re_string_t |
368 | { |
369 | /* Indicate the raw buffer which is the original string passed as an |
370 | argument of regexec(), re_search(), etc.. */ |
371 | const unsigned char *raw_mbs; |
372 | /* Store the multibyte string. In case of "case insensitive mode" like |
373 | REG_ICASE, upper cases of the string are stored, otherwise MBS points |
374 | the same address that RAW_MBS points. */ |
375 | unsigned char *mbs; |
376 | #ifdef RE_ENABLE_I18N |
377 | /* Store the wide character string which is corresponding to MBS. */ |
378 | wint_t *wcs; |
379 | Idx *offsets; |
380 | mbstate_t cur_state; |
381 | #endif |
382 | /* Index in RAW_MBS. Each character mbs[i] corresponds to |
383 | raw_mbs[raw_mbs_idx + i]. */ |
384 | Idx raw_mbs_idx; |
385 | /* The length of the valid characters in the buffers. */ |
386 | Idx valid_len; |
387 | /* The corresponding number of bytes in raw_mbs array. */ |
388 | Idx valid_raw_len; |
389 | /* The length of the buffers MBS and WCS. */ |
390 | Idx bufs_len; |
391 | /* The index in MBS, which is updated by re_string_fetch_byte. */ |
392 | Idx cur_idx; |
393 | /* length of RAW_MBS array. */ |
394 | Idx raw_len; |
395 | /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ |
396 | Idx len; |
397 | /* End of the buffer may be shorter than its length in the cases such |
398 | as re_match_2, re_search_2. Then, we use STOP for end of the buffer |
399 | instead of LEN. */ |
400 | Idx raw_stop; |
401 | /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ |
402 | Idx stop; |
403 | |
404 | /* The context of mbs[0]. We store the context independently, since |
405 | the context of mbs[0] may be different from raw_mbs[0], which is |
406 | the beginning of the input string. */ |
407 | unsigned int tip_context; |
408 | /* The translation passed as a part of an argument of re_compile_pattern. */ |
409 | RE_TRANSLATE_TYPE trans; |
410 | /* Copy of re_dfa_t's word_char. */ |
411 | re_const_bitset_ptr_t word_char; |
412 | /* true if REG_ICASE. */ |
413 | unsigned char icase; |
414 | unsigned char is_utf8; |
415 | unsigned char map_notascii; |
416 | unsigned char mbs_allocated; |
417 | unsigned char offsets_needed; |
418 | unsigned char newline_anchor; |
419 | unsigned char word_ops_used; |
420 | int mb_cur_max; |
421 | }; |
422 | typedef struct re_string_t re_string_t; |
423 | |
424 | |
425 | struct re_dfa_t; |
426 | typedef struct re_dfa_t re_dfa_t; |
427 | |
428 | #ifndef _LIBC |
429 | # define IS_IN(libc) false |
430 | #endif |
431 | |
432 | #define re_string_peek_byte(pstr, offset) \ |
433 | ((pstr)->mbs[(pstr)->cur_idx + offset]) |
434 | #define re_string_fetch_byte(pstr) \ |
435 | ((pstr)->mbs[(pstr)->cur_idx++]) |
436 | #define re_string_first_byte(pstr, idx) \ |
437 | ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) |
438 | #define re_string_is_single_byte_char(pstr, idx) \ |
439 | ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ |
440 | || (pstr)->wcs[(idx) + 1] != WEOF)) |
441 | #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) |
442 | #define re_string_cur_idx(pstr) ((pstr)->cur_idx) |
443 | #define re_string_get_buffer(pstr) ((pstr)->mbs) |
444 | #define re_string_length(pstr) ((pstr)->len) |
445 | #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) |
446 | #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) |
447 | #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) |
448 | |
449 | #if defined _LIBC || HAVE_ALLOCA |
450 | # include <alloca.h> |
451 | #endif |
452 | |
453 | #ifndef _LIBC |
454 | # if HAVE_ALLOCA |
455 | /* The OS usually guarantees only one guard page at the bottom of the stack, |
456 | and a page size can be as small as 4096 bytes. So we cannot safely |
457 | allocate anything larger than 4096 bytes. Also care for the possibility |
458 | of a few compiler-allocated temporary stack slots. */ |
459 | # define __libc_use_alloca(n) ((n) < 4032) |
460 | # else |
461 | /* alloca is implemented with malloc, so just use malloc. */ |
462 | # define __libc_use_alloca(n) 0 |
463 | # undef alloca |
464 | # define alloca(n) malloc (n) |
465 | # endif |
466 | #endif |
467 | |
468 | #ifdef _LIBC |
469 | # define MALLOC_0_IS_NONNULL 1 |
470 | #elif !defined MALLOC_0_IS_NONNULL |
471 | # define MALLOC_0_IS_NONNULL 0 |
472 | #endif |
473 | |
474 | #ifndef MAX |
475 | # define MAX(a,b) ((a) < (b) ? (b) : (a)) |
476 | #endif |
477 | #ifndef MIN |
478 | # define MIN(a,b) ((a) < (b) ? (a) : (b)) |
479 | #endif |
480 | |
481 | #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) |
482 | #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) |
483 | #define re_free(p) free (p) |
484 | |
485 | struct bin_tree_t |
486 | { |
487 | struct bin_tree_t *parent; |
488 | struct bin_tree_t *left; |
489 | struct bin_tree_t *right; |
490 | struct bin_tree_t *first; |
491 | struct bin_tree_t *next; |
492 | |
493 | re_token_t token; |
494 | |
495 | /* 'node_idx' is the index in dfa->nodes, if 'type' == 0. |
496 | Otherwise 'type' indicate the type of this node. */ |
497 | Idx node_idx; |
498 | }; |
499 | typedef struct bin_tree_t bin_tree_t; |
500 | |
501 | #define BIN_TREE_STORAGE_SIZE \ |
502 | ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) |
503 | |
504 | struct bin_tree_storage_t |
505 | { |
506 | struct bin_tree_storage_t *next; |
507 | bin_tree_t data[BIN_TREE_STORAGE_SIZE]; |
508 | }; |
509 | typedef struct bin_tree_storage_t bin_tree_storage_t; |
510 | |
511 | #define CONTEXT_WORD 1 |
512 | #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) |
513 | #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) |
514 | #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) |
515 | |
516 | #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) |
517 | #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) |
518 | #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) |
519 | #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) |
520 | #define IS_ORDINARY_CONTEXT(c) ((c) == 0) |
521 | |
522 | #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') |
523 | #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) |
524 | #define IS_WIDE_WORD_CHAR(ch) (__iswalnum (ch) || (ch) == L'_') |
525 | #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) |
526 | |
527 | #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ |
528 | ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ |
529 | || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ |
530 | || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ |
531 | || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) |
532 | |
533 | #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ |
534 | ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ |
535 | || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ |
536 | || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ |
537 | || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) |
538 | |
539 | struct re_dfastate_t |
540 | { |
541 | re_hashval_t hash; |
542 | re_node_set nodes; |
543 | re_node_set non_eps_nodes; |
544 | re_node_set inveclosure; |
545 | re_node_set *entrance_nodes; |
546 | struct re_dfastate_t **trtable, **word_trtable; |
547 | unsigned int context : 4; |
548 | unsigned int halt : 1; |
549 | /* If this state can accept "multi byte". |
550 | Note that we refer to multibyte characters, and multi character |
551 | collating elements as "multi byte". */ |
552 | unsigned int accept_mb : 1; |
553 | /* If this state has backreference node(s). */ |
554 | unsigned int has_backref : 1; |
555 | unsigned int has_constraint : 1; |
556 | }; |
557 | typedef struct re_dfastate_t re_dfastate_t; |
558 | |
559 | struct re_state_table_entry |
560 | { |
561 | Idx num; |
562 | Idx alloc; |
563 | re_dfastate_t **array; |
564 | }; |
565 | |
566 | /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ |
567 | |
568 | typedef struct |
569 | { |
570 | Idx next_idx; |
571 | Idx alloc; |
572 | re_dfastate_t **array; |
573 | } state_array_t; |
574 | |
575 | /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ |
576 | |
577 | typedef struct |
578 | { |
579 | Idx node; |
580 | Idx str_idx; /* The position NODE match at. */ |
581 | state_array_t path; |
582 | } re_sub_match_last_t; |
583 | |
584 | /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. |
585 | And information about the node, whose type is OP_CLOSE_SUBEXP, |
586 | corresponding to NODE is stored in LASTS. */ |
587 | |
588 | typedef struct |
589 | { |
590 | Idx str_idx; |
591 | Idx node; |
592 | state_array_t *path; |
593 | Idx alasts; /* Allocation size of LASTS. */ |
594 | Idx nlasts; /* The number of LASTS. */ |
595 | re_sub_match_last_t **lasts; |
596 | } re_sub_match_top_t; |
597 | |
598 | struct re_backref_cache_entry |
599 | { |
600 | Idx node; |
601 | Idx str_idx; |
602 | Idx subexp_from; |
603 | Idx subexp_to; |
604 | char more; |
605 | char unused; |
606 | unsigned short int eps_reachable_subexps_map; |
607 | }; |
608 | |
609 | typedef struct |
610 | { |
611 | /* The string object corresponding to the input string. */ |
612 | re_string_t input; |
613 | const re_dfa_t *const dfa; |
614 | /* EFLAGS of the argument of regexec. */ |
615 | int eflags; |
616 | /* Where the matching ends. */ |
617 | Idx match_last; |
618 | Idx last_node; |
619 | /* The state log used by the matcher. */ |
620 | re_dfastate_t **state_log; |
621 | Idx state_log_top; |
622 | /* Back reference cache. */ |
623 | Idx nbkref_ents; |
624 | Idx abkref_ents; |
625 | struct re_backref_cache_entry *bkref_ents; |
626 | int max_mb_elem_len; |
627 | Idx nsub_tops; |
628 | Idx asub_tops; |
629 | re_sub_match_top_t **sub_tops; |
630 | } re_match_context_t; |
631 | |
632 | typedef struct |
633 | { |
634 | re_dfastate_t **sifted_states; |
635 | re_dfastate_t **limited_states; |
636 | Idx last_node; |
637 | Idx last_str_idx; |
638 | re_node_set limits; |
639 | } re_sift_context_t; |
640 | |
641 | struct re_fail_stack_ent_t |
642 | { |
643 | Idx idx; |
644 | Idx node; |
645 | regmatch_t *regs; |
646 | re_node_set eps_via_nodes; |
647 | }; |
648 | |
649 | struct re_fail_stack_t |
650 | { |
651 | Idx num; |
652 | Idx alloc; |
653 | struct re_fail_stack_ent_t *stack; |
654 | }; |
655 | |
656 | struct re_dfa_t |
657 | { |
658 | re_token_t *nodes; |
659 | size_t nodes_alloc; |
660 | size_t nodes_len; |
661 | Idx *nexts; |
662 | Idx *org_indices; |
663 | re_node_set *edests; |
664 | re_node_set *eclosures; |
665 | re_node_set *inveclosures; |
666 | struct re_state_table_entry *state_table; |
667 | re_dfastate_t *init_state; |
668 | re_dfastate_t *init_state_word; |
669 | re_dfastate_t *init_state_nl; |
670 | re_dfastate_t *init_state_begbuf; |
671 | bin_tree_t *str_tree; |
672 | bin_tree_storage_t *str_tree_storage; |
673 | re_bitset_ptr_t sb_char; |
674 | int str_tree_storage_idx; |
675 | |
676 | /* number of subexpressions 're_nsub' is in regex_t. */ |
677 | re_hashval_t state_hash_mask; |
678 | Idx init_node; |
679 | Idx nbackref; /* The number of backreference in this dfa. */ |
680 | |
681 | /* Bitmap expressing which backreference is used. */ |
682 | bitset_word_t used_bkref_map; |
683 | bitset_word_t completed_bkref_map; |
684 | |
685 | unsigned int has_plural_match : 1; |
686 | /* If this dfa has "multibyte node", which is a backreference or |
687 | a node which can accept multibyte character or multi character |
688 | collating element. */ |
689 | unsigned int has_mb_node : 1; |
690 | unsigned int is_utf8 : 1; |
691 | unsigned int map_notascii : 1; |
692 | unsigned int word_ops_used : 1; |
693 | int mb_cur_max; |
694 | bitset_t word_char; |
695 | reg_syntax_t syntax; |
696 | Idx *subexp_map; |
697 | #ifdef DEBUG |
698 | char* re_str; |
699 | #endif |
700 | lock_define (lock) |
701 | }; |
702 | |
703 | #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) |
704 | #define re_node_set_remove(set,id) \ |
705 | (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) |
706 | #define re_node_set_empty(p) ((p)->nelem = 0) |
707 | #define re_node_set_free(set) re_free ((set)->elems) |
708 | |
709 | |
710 | typedef enum |
711 | { |
712 | SB_CHAR, |
713 | MB_CHAR, |
714 | EQUIV_CLASS, |
715 | COLL_SYM, |
716 | CHAR_CLASS |
717 | } bracket_elem_type; |
718 | |
719 | typedef struct |
720 | { |
721 | bracket_elem_type type; |
722 | union |
723 | { |
724 | unsigned char ch; |
725 | unsigned char *name; |
726 | wchar_t wch; |
727 | } opr; |
728 | } bracket_elem_t; |
729 | |
730 | |
731 | /* Functions for bitset_t operation. */ |
732 | |
733 | static inline void |
734 | bitset_set (bitset_t set, Idx i) |
735 | { |
736 | set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; |
737 | } |
738 | |
739 | static inline void |
740 | bitset_clear (bitset_t set, Idx i) |
741 | { |
742 | set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); |
743 | } |
744 | |
745 | static inline bool |
746 | bitset_contain (const bitset_t set, Idx i) |
747 | { |
748 | return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; |
749 | } |
750 | |
751 | static inline void |
752 | bitset_empty (bitset_t set) |
753 | { |
754 | memset (set, '\0', sizeof (bitset_t)); |
755 | } |
756 | |
757 | static inline void |
758 | bitset_set_all (bitset_t set) |
759 | { |
760 | memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); |
761 | if (SBC_MAX % BITSET_WORD_BITS != 0) |
762 | set[BITSET_WORDS - 1] = |
763 | ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; |
764 | } |
765 | |
766 | static inline void |
767 | bitset_copy (bitset_t dest, const bitset_t src) |
768 | { |
769 | memcpy (dest, src, sizeof (bitset_t)); |
770 | } |
771 | |
772 | static inline void |
773 | bitset_not (bitset_t set) |
774 | { |
775 | int bitset_i; |
776 | for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) |
777 | set[bitset_i] = ~set[bitset_i]; |
778 | if (SBC_MAX % BITSET_WORD_BITS != 0) |
779 | set[BITSET_WORDS - 1] = |
780 | ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) |
781 | & ~set[BITSET_WORDS - 1]); |
782 | } |
783 | |
784 | static inline void |
785 | bitset_merge (bitset_t dest, const bitset_t src) |
786 | { |
787 | int bitset_i; |
788 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) |
789 | dest[bitset_i] |= src[bitset_i]; |
790 | } |
791 | |
792 | static inline void |
793 | bitset_mask (bitset_t dest, const bitset_t src) |
794 | { |
795 | int bitset_i; |
796 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) |
797 | dest[bitset_i] &= src[bitset_i]; |
798 | } |
799 | |
800 | #ifdef RE_ENABLE_I18N |
801 | /* Functions for re_string. */ |
802 | static int |
803 | __attribute__ ((pure, unused)) |
804 | re_string_char_size_at (const re_string_t *pstr, Idx idx) |
805 | { |
806 | int byte_idx; |
807 | if (pstr->mb_cur_max == 1) |
808 | return 1; |
809 | for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) |
810 | if (pstr->wcs[idx + byte_idx] != WEOF) |
811 | break; |
812 | return byte_idx; |
813 | } |
814 | |
815 | static wint_t |
816 | __attribute__ ((pure, unused)) |
817 | re_string_wchar_at (const re_string_t *pstr, Idx idx) |
818 | { |
819 | if (pstr->mb_cur_max == 1) |
820 | return (wint_t) pstr->mbs[idx]; |
821 | return (wint_t) pstr->wcs[idx]; |
822 | } |
823 | |
824 | # ifdef _LIBC |
825 | # include <locale/weight.h> |
826 | # endif |
827 | |
828 | static int |
829 | __attribute__ ((pure, unused)) |
830 | re_string_elem_size_at (const re_string_t *pstr, Idx idx) |
831 | { |
832 | # ifdef _LIBC |
833 | const unsigned char *p, *; |
834 | const int32_t *table, *indirect; |
835 | uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); |
836 | |
837 | if (nrules != 0) |
838 | { |
839 | table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); |
840 | extra = (const unsigned char *) |
841 | _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); |
842 | indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, |
843 | _NL_COLLATE_INDIRECTMB); |
844 | p = pstr->mbs + idx; |
845 | findidx (table, indirect, extra, &p, pstr->len - idx); |
846 | return p - pstr->mbs - idx; |
847 | } |
848 | else |
849 | # endif /* _LIBC */ |
850 | return 1; |
851 | } |
852 | #endif /* RE_ENABLE_I18N */ |
853 | |
854 | #ifndef FALLTHROUGH |
855 | # if __GNUC__ < 7 |
856 | # define FALLTHROUGH ((void) 0) |
857 | # else |
858 | # define FALLTHROUGH __attribute__ ((__fallthrough__)) |
859 | # endif |
860 | #endif |
861 | |
862 | #endif /* _REGEX_INTERNAL_H */ |
863 | |