1 | /* Extended regular expression matching and search library. |
2 | Copyright (C) 2002-2019 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 | #include <langinfo.h> |
30 | #include <locale.h> |
31 | #include <wchar.h> |
32 | #include <wctype.h> |
33 | #include <stdbool.h> |
34 | #include <stdint.h> |
35 | |
36 | #include <intprops.h> |
37 | |
38 | #ifdef _LIBC |
39 | # include <libc-lock.h> |
40 | # define lock_define(name) __libc_lock_define (, name) |
41 | # define lock_init(lock) (__libc_lock_init (lock), 0) |
42 | # define lock_fini(lock) ((void) 0) |
43 | # define lock_lock(lock) __libc_lock_lock (lock) |
44 | # define lock_unlock(lock) __libc_lock_unlock (lock) |
45 | #elif defined GNULIB_LOCK && !defined USE_UNLOCKED_IO |
46 | # include "glthread/lock.h" |
47 | /* Use gl_lock_define if empty macro arguments are known to work. |
48 | Otherwise, fall back on less-portable substitutes. */ |
49 | # if ((defined __GNUC__ && !defined __STRICT_ANSI__) \ |
50 | || (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__)) |
51 | # define lock_define(name) gl_lock_define (, name) |
52 | # elif USE_POSIX_THREADS |
53 | # define lock_define(name) pthread_mutex_t name; |
54 | # elif USE_PTH_THREADS |
55 | # define lock_define(name) pth_mutex_t name; |
56 | # elif USE_SOLARIS_THREADS |
57 | # define lock_define(name) mutex_t name; |
58 | # elif USE_WINDOWS_THREADS |
59 | # define lock_define(name) gl_lock_t name; |
60 | # else |
61 | # define lock_define(name) |
62 | # endif |
63 | # define lock_init(lock) glthread_lock_init (&(lock)) |
64 | # define lock_fini(lock) glthread_lock_destroy (&(lock)) |
65 | # define lock_lock(lock) glthread_lock_lock (&(lock)) |
66 | # define lock_unlock(lock) glthread_lock_unlock (&(lock)) |
67 | #elif defined GNULIB_PTHREAD && !defined USE_UNLOCKED_IO |
68 | # include <pthread.h> |
69 | # define lock_define(name) pthread_mutex_t name; |
70 | # define lock_init(lock) pthread_mutex_init (&(lock), 0) |
71 | # define lock_fini(lock) pthread_mutex_destroy (&(lock)) |
72 | # define lock_lock(lock) pthread_mutex_lock (&(lock)) |
73 | # define lock_unlock(lock) pthread_mutex_unlock (&(lock)) |
74 | #else |
75 | # define lock_define(name) |
76 | # define lock_init(lock) 0 |
77 | # define lock_fini(lock) ((void) 0) |
78 | /* The 'dfa' avoids an "unused variable 'dfa'" warning from GCC. */ |
79 | # define lock_lock(lock) ((void) dfa) |
80 | # define lock_unlock(lock) ((void) 0) |
81 | #endif |
82 | |
83 | /* In case that the system doesn't have isblank(). */ |
84 | #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) |
85 | # define isblank(ch) ((ch) == ' ' || (ch) == '\t') |
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 | |
153 | /* The type of indexes into strings. This is signed, not size_t, |
154 | since the API requires indexes to fit in regoff_t anyway, and using |
155 | signed integers makes the code a bit smaller and presumably faster. |
156 | The traditional GNU regex implementation uses int for indexes. |
157 | The POSIX-compatible implementation uses a possibly-wider type. |
158 | The name 'Idx' is three letters to minimize the hassle of |
159 | reindenting a lot of regex code that formerly used 'int'. */ |
160 | typedef regoff_t Idx; |
161 | #ifdef _REGEX_LARGE_OFFSETS |
162 | # define IDX_MAX SSIZE_MAX |
163 | #else |
164 | # define IDX_MAX INT_MAX |
165 | #endif |
166 | |
167 | /* A hash value, suitable for computing hash tables. */ |
168 | typedef __re_size_t re_hashval_t; |
169 | |
170 | /* An integer used to represent a set of bits. It must be unsigned, |
171 | and must be at least as wide as unsigned int. */ |
172 | typedef unsigned long int bitset_word_t; |
173 | /* All bits set in a bitset_word_t. */ |
174 | #define BITSET_WORD_MAX ULONG_MAX |
175 | |
176 | /* Number of bits in a bitset_word_t. For portability to hosts with |
177 | padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)'; |
178 | instead, deduce it directly from BITSET_WORD_MAX. Avoid |
179 | greater-than-32-bit integers and unconditional shifts by more than |
180 | 31 bits, as they're not portable. */ |
181 | #if BITSET_WORD_MAX == 0xffffffffUL |
182 | # define BITSET_WORD_BITS 32 |
183 | #elif BITSET_WORD_MAX >> 31 >> 4 == 1 |
184 | # define BITSET_WORD_BITS 36 |
185 | #elif BITSET_WORD_MAX >> 31 >> 16 == 1 |
186 | # define BITSET_WORD_BITS 48 |
187 | #elif BITSET_WORD_MAX >> 31 >> 28 == 1 |
188 | # define BITSET_WORD_BITS 60 |
189 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1 |
190 | # define BITSET_WORD_BITS 64 |
191 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1 |
192 | # define BITSET_WORD_BITS 72 |
193 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1 |
194 | # define BITSET_WORD_BITS 128 |
195 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1 |
196 | # define BITSET_WORD_BITS 256 |
197 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1 |
198 | # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */ |
199 | # if BITSET_WORD_BITS <= SBC_MAX |
200 | # error "Invalid SBC_MAX" |
201 | # endif |
202 | #else |
203 | # error "Add case for new bitset_word_t size" |
204 | #endif |
205 | |
206 | /* Number of bitset_word_t values in a bitset_t. */ |
207 | #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) |
208 | |
209 | typedef bitset_word_t bitset_t[BITSET_WORDS]; |
210 | typedef bitset_word_t *re_bitset_ptr_t; |
211 | typedef const bitset_word_t *re_const_bitset_ptr_t; |
212 | |
213 | #define PREV_WORD_CONSTRAINT 0x0001 |
214 | #define PREV_NOTWORD_CONSTRAINT 0x0002 |
215 | #define NEXT_WORD_CONSTRAINT 0x0004 |
216 | #define NEXT_NOTWORD_CONSTRAINT 0x0008 |
217 | #define PREV_NEWLINE_CONSTRAINT 0x0010 |
218 | #define NEXT_NEWLINE_CONSTRAINT 0x0020 |
219 | #define PREV_BEGBUF_CONSTRAINT 0x0040 |
220 | #define NEXT_ENDBUF_CONSTRAINT 0x0080 |
221 | #define WORD_DELIM_CONSTRAINT 0x0100 |
222 | #define NOT_WORD_DELIM_CONSTRAINT 0x0200 |
223 | |
224 | typedef enum |
225 | { |
226 | INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, |
227 | WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, |
228 | WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, |
229 | INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, |
230 | LINE_FIRST = PREV_NEWLINE_CONSTRAINT, |
231 | LINE_LAST = NEXT_NEWLINE_CONSTRAINT, |
232 | BUF_FIRST = PREV_BEGBUF_CONSTRAINT, |
233 | BUF_LAST = NEXT_ENDBUF_CONSTRAINT, |
234 | WORD_DELIM = WORD_DELIM_CONSTRAINT, |
235 | NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT |
236 | } re_context_type; |
237 | |
238 | typedef struct |
239 | { |
240 | Idx alloc; |
241 | Idx nelem; |
242 | Idx *elems; |
243 | } re_node_set; |
244 | |
245 | typedef enum |
246 | { |
247 | NON_TYPE = 0, |
248 | |
249 | /* Node type, These are used by token, node, tree. */ |
250 | CHARACTER = 1, |
251 | END_OF_RE = 2, |
252 | SIMPLE_BRACKET = 3, |
253 | OP_BACK_REF = 4, |
254 | OP_PERIOD = 5, |
255 | #ifdef RE_ENABLE_I18N |
256 | COMPLEX_BRACKET = 6, |
257 | OP_UTF8_PERIOD = 7, |
258 | #endif /* RE_ENABLE_I18N */ |
259 | |
260 | /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used |
261 | when the debugger shows values of this enum type. */ |
262 | #define EPSILON_BIT 8 |
263 | OP_OPEN_SUBEXP = EPSILON_BIT | 0, |
264 | OP_CLOSE_SUBEXP = EPSILON_BIT | 1, |
265 | OP_ALT = EPSILON_BIT | 2, |
266 | OP_DUP_ASTERISK = EPSILON_BIT | 3, |
267 | ANCHOR = EPSILON_BIT | 4, |
268 | |
269 | /* Tree type, these are used only by tree. */ |
270 | CONCAT = 16, |
271 | SUBEXP = 17, |
272 | |
273 | /* Token type, these are used only by token. */ |
274 | OP_DUP_PLUS = 18, |
275 | OP_DUP_QUESTION, |
276 | OP_OPEN_BRACKET, |
277 | OP_CLOSE_BRACKET, |
278 | OP_CHARSET_RANGE, |
279 | OP_OPEN_DUP_NUM, |
280 | OP_CLOSE_DUP_NUM, |
281 | OP_NON_MATCH_LIST, |
282 | OP_OPEN_COLL_ELEM, |
283 | OP_CLOSE_COLL_ELEM, |
284 | OP_OPEN_EQUIV_CLASS, |
285 | OP_CLOSE_EQUIV_CLASS, |
286 | OP_OPEN_CHAR_CLASS, |
287 | OP_CLOSE_CHAR_CLASS, |
288 | OP_WORD, |
289 | OP_NOTWORD, |
290 | OP_SPACE, |
291 | OP_NOTSPACE, |
292 | BACK_SLASH |
293 | |
294 | } re_token_type_t; |
295 | |
296 | #ifdef RE_ENABLE_I18N |
297 | typedef struct |
298 | { |
299 | /* Multibyte characters. */ |
300 | wchar_t *mbchars; |
301 | |
302 | /* Collating symbols. */ |
303 | # ifdef _LIBC |
304 | int32_t *coll_syms; |
305 | # endif |
306 | |
307 | /* Equivalence classes. */ |
308 | # ifdef _LIBC |
309 | int32_t *equiv_classes; |
310 | # endif |
311 | |
312 | /* Range expressions. */ |
313 | # ifdef _LIBC |
314 | uint32_t *range_starts; |
315 | uint32_t *range_ends; |
316 | # else /* not _LIBC */ |
317 | wchar_t *range_starts; |
318 | wchar_t *range_ends; |
319 | # endif /* not _LIBC */ |
320 | |
321 | /* Character classes. */ |
322 | wctype_t *char_classes; |
323 | |
324 | /* If this character set is the non-matching list. */ |
325 | unsigned int non_match : 1; |
326 | |
327 | /* # of multibyte characters. */ |
328 | Idx nmbchars; |
329 | |
330 | /* # of collating symbols. */ |
331 | Idx ncoll_syms; |
332 | |
333 | /* # of equivalence classes. */ |
334 | Idx nequiv_classes; |
335 | |
336 | /* # of range expressions. */ |
337 | Idx nranges; |
338 | |
339 | /* # of character classes. */ |
340 | Idx nchar_classes; |
341 | } re_charset_t; |
342 | #endif /* RE_ENABLE_I18N */ |
343 | |
344 | typedef struct |
345 | { |
346 | union |
347 | { |
348 | unsigned char c; /* for CHARACTER */ |
349 | re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ |
350 | #ifdef RE_ENABLE_I18N |
351 | re_charset_t *mbcset; /* for COMPLEX_BRACKET */ |
352 | #endif /* RE_ENABLE_I18N */ |
353 | Idx idx; /* for BACK_REF */ |
354 | re_context_type ctx_type; /* for ANCHOR */ |
355 | } opr; |
356 | #if __GNUC__ >= 2 && !defined __STRICT_ANSI__ |
357 | re_token_type_t type : 8; |
358 | #else |
359 | re_token_type_t type; |
360 | #endif |
361 | unsigned int constraint : 10; /* context constraint */ |
362 | unsigned int duplicated : 1; |
363 | unsigned int opt_subexp : 1; |
364 | #ifdef RE_ENABLE_I18N |
365 | unsigned int accept_mb : 1; |
366 | /* These 2 bits can be moved into the union if needed (e.g. if running out |
367 | of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ |
368 | unsigned int mb_partial : 1; |
369 | #endif |
370 | unsigned int word_char : 1; |
371 | } re_token_t; |
372 | |
373 | #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) |
374 | |
375 | struct re_string_t |
376 | { |
377 | /* Indicate the raw buffer which is the original string passed as an |
378 | argument of regexec(), re_search(), etc.. */ |
379 | const unsigned char *raw_mbs; |
380 | /* Store the multibyte string. In case of "case insensitive mode" like |
381 | REG_ICASE, upper cases of the string are stored, otherwise MBS points |
382 | the same address that RAW_MBS points. */ |
383 | unsigned char *mbs; |
384 | #ifdef RE_ENABLE_I18N |
385 | /* Store the wide character string which is corresponding to MBS. */ |
386 | wint_t *wcs; |
387 | Idx *offsets; |
388 | mbstate_t cur_state; |
389 | #endif |
390 | /* Index in RAW_MBS. Each character mbs[i] corresponds to |
391 | raw_mbs[raw_mbs_idx + i]. */ |
392 | Idx raw_mbs_idx; |
393 | /* The length of the valid characters in the buffers. */ |
394 | Idx valid_len; |
395 | /* The corresponding number of bytes in raw_mbs array. */ |
396 | Idx valid_raw_len; |
397 | /* The length of the buffers MBS and WCS. */ |
398 | Idx bufs_len; |
399 | /* The index in MBS, which is updated by re_string_fetch_byte. */ |
400 | Idx cur_idx; |
401 | /* length of RAW_MBS array. */ |
402 | Idx raw_len; |
403 | /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ |
404 | Idx len; |
405 | /* End of the buffer may be shorter than its length in the cases such |
406 | as re_match_2, re_search_2. Then, we use STOP for end of the buffer |
407 | instead of LEN. */ |
408 | Idx raw_stop; |
409 | /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ |
410 | Idx stop; |
411 | |
412 | /* The context of mbs[0]. We store the context independently, since |
413 | the context of mbs[0] may be different from raw_mbs[0], which is |
414 | the beginning of the input string. */ |
415 | unsigned int tip_context; |
416 | /* The translation passed as a part of an argument of re_compile_pattern. */ |
417 | RE_TRANSLATE_TYPE trans; |
418 | /* Copy of re_dfa_t's word_char. */ |
419 | re_const_bitset_ptr_t word_char; |
420 | /* true if REG_ICASE. */ |
421 | unsigned char icase; |
422 | unsigned char is_utf8; |
423 | unsigned char map_notascii; |
424 | unsigned char mbs_allocated; |
425 | unsigned char offsets_needed; |
426 | unsigned char newline_anchor; |
427 | unsigned char word_ops_used; |
428 | int mb_cur_max; |
429 | }; |
430 | typedef struct re_string_t re_string_t; |
431 | |
432 | |
433 | struct re_dfa_t; |
434 | typedef struct re_dfa_t re_dfa_t; |
435 | |
436 | #ifndef _LIBC |
437 | # define IS_IN(libc) false |
438 | #endif |
439 | |
440 | #define re_string_peek_byte(pstr, offset) \ |
441 | ((pstr)->mbs[(pstr)->cur_idx + offset]) |
442 | #define re_string_fetch_byte(pstr) \ |
443 | ((pstr)->mbs[(pstr)->cur_idx++]) |
444 | #define re_string_first_byte(pstr, idx) \ |
445 | ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) |
446 | #define re_string_is_single_byte_char(pstr, idx) \ |
447 | ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ |
448 | || (pstr)->wcs[(idx) + 1] != WEOF)) |
449 | #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) |
450 | #define re_string_cur_idx(pstr) ((pstr)->cur_idx) |
451 | #define re_string_get_buffer(pstr) ((pstr)->mbs) |
452 | #define re_string_length(pstr) ((pstr)->len) |
453 | #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) |
454 | #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) |
455 | #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) |
456 | |
457 | #if defined _LIBC || HAVE_ALLOCA |
458 | # include <alloca.h> |
459 | #endif |
460 | |
461 | #ifndef _LIBC |
462 | # if HAVE_ALLOCA |
463 | /* The OS usually guarantees only one guard page at the bottom of the stack, |
464 | and a page size can be as small as 4096 bytes. So we cannot safely |
465 | allocate anything larger than 4096 bytes. Also care for the possibility |
466 | of a few compiler-allocated temporary stack slots. */ |
467 | # define __libc_use_alloca(n) ((n) < 4032) |
468 | # else |
469 | /* alloca is implemented with malloc, so just use malloc. */ |
470 | # define __libc_use_alloca(n) 0 |
471 | # undef alloca |
472 | # define alloca(n) malloc (n) |
473 | # endif |
474 | #endif |
475 | |
476 | #ifdef _LIBC |
477 | # define MALLOC_0_IS_NONNULL 1 |
478 | #elif !defined MALLOC_0_IS_NONNULL |
479 | # define MALLOC_0_IS_NONNULL 0 |
480 | #endif |
481 | |
482 | #ifndef MAX |
483 | # define MAX(a,b) ((a) < (b) ? (b) : (a)) |
484 | #endif |
485 | #ifndef MIN |
486 | # define MIN(a,b) ((a) < (b) ? (a) : (b)) |
487 | #endif |
488 | |
489 | #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) |
490 | #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) |
491 | #define re_free(p) free (p) |
492 | |
493 | struct bin_tree_t |
494 | { |
495 | struct bin_tree_t *parent; |
496 | struct bin_tree_t *left; |
497 | struct bin_tree_t *right; |
498 | struct bin_tree_t *first; |
499 | struct bin_tree_t *next; |
500 | |
501 | re_token_t token; |
502 | |
503 | /* 'node_idx' is the index in dfa->nodes, if 'type' == 0. |
504 | Otherwise 'type' indicate the type of this node. */ |
505 | Idx node_idx; |
506 | }; |
507 | typedef struct bin_tree_t bin_tree_t; |
508 | |
509 | #define BIN_TREE_STORAGE_SIZE \ |
510 | ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) |
511 | |
512 | struct bin_tree_storage_t |
513 | { |
514 | struct bin_tree_storage_t *next; |
515 | bin_tree_t data[BIN_TREE_STORAGE_SIZE]; |
516 | }; |
517 | typedef struct bin_tree_storage_t bin_tree_storage_t; |
518 | |
519 | #define CONTEXT_WORD 1 |
520 | #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) |
521 | #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) |
522 | #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) |
523 | |
524 | #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) |
525 | #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) |
526 | #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) |
527 | #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) |
528 | #define IS_ORDINARY_CONTEXT(c) ((c) == 0) |
529 | |
530 | #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') |
531 | #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) |
532 | #define IS_WIDE_WORD_CHAR(ch) (__iswalnum (ch) || (ch) == L'_') |
533 | #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) |
534 | |
535 | #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ |
536 | ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ |
537 | || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ |
538 | || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ |
539 | || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) |
540 | |
541 | #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ |
542 | ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ |
543 | || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ |
544 | || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ |
545 | || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) |
546 | |
547 | struct re_dfastate_t |
548 | { |
549 | re_hashval_t hash; |
550 | re_node_set nodes; |
551 | re_node_set non_eps_nodes; |
552 | re_node_set inveclosure; |
553 | re_node_set *entrance_nodes; |
554 | struct re_dfastate_t **trtable, **word_trtable; |
555 | unsigned int context : 4; |
556 | unsigned int halt : 1; |
557 | /* If this state can accept "multi byte". |
558 | Note that we refer to multibyte characters, and multi character |
559 | collating elements as "multi byte". */ |
560 | unsigned int accept_mb : 1; |
561 | /* If this state has backreference node(s). */ |
562 | unsigned int has_backref : 1; |
563 | unsigned int has_constraint : 1; |
564 | }; |
565 | typedef struct re_dfastate_t re_dfastate_t; |
566 | |
567 | struct re_state_table_entry |
568 | { |
569 | Idx num; |
570 | Idx alloc; |
571 | re_dfastate_t **array; |
572 | }; |
573 | |
574 | /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ |
575 | |
576 | typedef struct |
577 | { |
578 | Idx next_idx; |
579 | Idx alloc; |
580 | re_dfastate_t **array; |
581 | } state_array_t; |
582 | |
583 | /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ |
584 | |
585 | typedef struct |
586 | { |
587 | Idx node; |
588 | Idx str_idx; /* The position NODE match at. */ |
589 | state_array_t path; |
590 | } re_sub_match_last_t; |
591 | |
592 | /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. |
593 | And information about the node, whose type is OP_CLOSE_SUBEXP, |
594 | corresponding to NODE is stored in LASTS. */ |
595 | |
596 | typedef struct |
597 | { |
598 | Idx str_idx; |
599 | Idx node; |
600 | state_array_t *path; |
601 | Idx alasts; /* Allocation size of LASTS. */ |
602 | Idx nlasts; /* The number of LASTS. */ |
603 | re_sub_match_last_t **lasts; |
604 | } re_sub_match_top_t; |
605 | |
606 | struct re_backref_cache_entry |
607 | { |
608 | Idx node; |
609 | Idx str_idx; |
610 | Idx subexp_from; |
611 | Idx subexp_to; |
612 | char more; |
613 | char unused; |
614 | unsigned short int eps_reachable_subexps_map; |
615 | }; |
616 | |
617 | typedef struct |
618 | { |
619 | /* The string object corresponding to the input string. */ |
620 | re_string_t input; |
621 | #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) |
622 | const re_dfa_t *const dfa; |
623 | #else |
624 | const re_dfa_t *dfa; |
625 | #endif |
626 | /* EFLAGS of the argument of regexec. */ |
627 | int eflags; |
628 | /* Where the matching ends. */ |
629 | Idx match_last; |
630 | Idx last_node; |
631 | /* The state log used by the matcher. */ |
632 | re_dfastate_t **state_log; |
633 | Idx state_log_top; |
634 | /* Back reference cache. */ |
635 | Idx nbkref_ents; |
636 | Idx abkref_ents; |
637 | struct re_backref_cache_entry *bkref_ents; |
638 | int max_mb_elem_len; |
639 | Idx nsub_tops; |
640 | Idx asub_tops; |
641 | re_sub_match_top_t **sub_tops; |
642 | } re_match_context_t; |
643 | |
644 | typedef struct |
645 | { |
646 | re_dfastate_t **sifted_states; |
647 | re_dfastate_t **limited_states; |
648 | Idx last_node; |
649 | Idx last_str_idx; |
650 | re_node_set limits; |
651 | } re_sift_context_t; |
652 | |
653 | struct re_fail_stack_ent_t |
654 | { |
655 | Idx idx; |
656 | Idx node; |
657 | regmatch_t *regs; |
658 | re_node_set eps_via_nodes; |
659 | }; |
660 | |
661 | struct re_fail_stack_t |
662 | { |
663 | Idx num; |
664 | Idx alloc; |
665 | struct re_fail_stack_ent_t *stack; |
666 | }; |
667 | |
668 | struct re_dfa_t |
669 | { |
670 | re_token_t *nodes; |
671 | size_t nodes_alloc; |
672 | size_t nodes_len; |
673 | Idx *nexts; |
674 | Idx *org_indices; |
675 | re_node_set *edests; |
676 | re_node_set *eclosures; |
677 | re_node_set *inveclosures; |
678 | struct re_state_table_entry *state_table; |
679 | re_dfastate_t *init_state; |
680 | re_dfastate_t *init_state_word; |
681 | re_dfastate_t *init_state_nl; |
682 | re_dfastate_t *init_state_begbuf; |
683 | bin_tree_t *str_tree; |
684 | bin_tree_storage_t *str_tree_storage; |
685 | re_bitset_ptr_t sb_char; |
686 | int str_tree_storage_idx; |
687 | |
688 | /* number of subexpressions 're_nsub' is in regex_t. */ |
689 | re_hashval_t state_hash_mask; |
690 | Idx init_node; |
691 | Idx nbackref; /* The number of backreference in this dfa. */ |
692 | |
693 | /* Bitmap expressing which backreference is used. */ |
694 | bitset_word_t used_bkref_map; |
695 | bitset_word_t completed_bkref_map; |
696 | |
697 | unsigned int has_plural_match : 1; |
698 | /* If this dfa has "multibyte node", which is a backreference or |
699 | a node which can accept multibyte character or multi character |
700 | collating element. */ |
701 | unsigned int has_mb_node : 1; |
702 | unsigned int is_utf8 : 1; |
703 | unsigned int map_notascii : 1; |
704 | unsigned int word_ops_used : 1; |
705 | int mb_cur_max; |
706 | bitset_t word_char; |
707 | reg_syntax_t syntax; |
708 | Idx *subexp_map; |
709 | #ifdef DEBUG |
710 | char* re_str; |
711 | #endif |
712 | lock_define (lock) |
713 | }; |
714 | |
715 | #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) |
716 | #define re_node_set_remove(set,id) \ |
717 | (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) |
718 | #define re_node_set_empty(p) ((p)->nelem = 0) |
719 | #define re_node_set_free(set) re_free ((set)->elems) |
720 | |
721 | |
722 | typedef enum |
723 | { |
724 | SB_CHAR, |
725 | MB_CHAR, |
726 | EQUIV_CLASS, |
727 | COLL_SYM, |
728 | CHAR_CLASS |
729 | } bracket_elem_type; |
730 | |
731 | typedef struct |
732 | { |
733 | bracket_elem_type type; |
734 | union |
735 | { |
736 | unsigned char ch; |
737 | unsigned char *name; |
738 | wchar_t wch; |
739 | } opr; |
740 | } bracket_elem_t; |
741 | |
742 | |
743 | /* Functions for bitset_t operation. */ |
744 | |
745 | static inline void |
746 | bitset_set (bitset_t set, Idx i) |
747 | { |
748 | set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; |
749 | } |
750 | |
751 | static inline void |
752 | bitset_clear (bitset_t set, Idx i) |
753 | { |
754 | set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); |
755 | } |
756 | |
757 | static inline bool |
758 | bitset_contain (const bitset_t set, Idx i) |
759 | { |
760 | return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; |
761 | } |
762 | |
763 | static inline void |
764 | bitset_empty (bitset_t set) |
765 | { |
766 | memset (set, '\0', sizeof (bitset_t)); |
767 | } |
768 | |
769 | static inline void |
770 | bitset_set_all (bitset_t set) |
771 | { |
772 | memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); |
773 | if (SBC_MAX % BITSET_WORD_BITS != 0) |
774 | set[BITSET_WORDS - 1] = |
775 | ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; |
776 | } |
777 | |
778 | static inline void |
779 | bitset_copy (bitset_t dest, const bitset_t src) |
780 | { |
781 | memcpy (dest, src, sizeof (bitset_t)); |
782 | } |
783 | |
784 | static inline void |
785 | bitset_not (bitset_t set) |
786 | { |
787 | int bitset_i; |
788 | for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) |
789 | set[bitset_i] = ~set[bitset_i]; |
790 | if (SBC_MAX % BITSET_WORD_BITS != 0) |
791 | set[BITSET_WORDS - 1] = |
792 | ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) |
793 | & ~set[BITSET_WORDS - 1]); |
794 | } |
795 | |
796 | static inline void |
797 | bitset_merge (bitset_t dest, const bitset_t src) |
798 | { |
799 | int bitset_i; |
800 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) |
801 | dest[bitset_i] |= src[bitset_i]; |
802 | } |
803 | |
804 | static inline void |
805 | bitset_mask (bitset_t dest, const bitset_t src) |
806 | { |
807 | int bitset_i; |
808 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) |
809 | dest[bitset_i] &= src[bitset_i]; |
810 | } |
811 | |
812 | #ifdef RE_ENABLE_I18N |
813 | /* Functions for re_string. */ |
814 | static int |
815 | __attribute__ ((pure, unused)) |
816 | re_string_char_size_at (const re_string_t *pstr, Idx idx) |
817 | { |
818 | int byte_idx; |
819 | if (pstr->mb_cur_max == 1) |
820 | return 1; |
821 | for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) |
822 | if (pstr->wcs[idx + byte_idx] != WEOF) |
823 | break; |
824 | return byte_idx; |
825 | } |
826 | |
827 | static wint_t |
828 | __attribute__ ((pure, unused)) |
829 | re_string_wchar_at (const re_string_t *pstr, Idx idx) |
830 | { |
831 | if (pstr->mb_cur_max == 1) |
832 | return (wint_t) pstr->mbs[idx]; |
833 | return (wint_t) pstr->wcs[idx]; |
834 | } |
835 | |
836 | # ifdef _LIBC |
837 | # include <locale/weight.h> |
838 | # endif |
839 | |
840 | static int |
841 | __attribute__ ((pure, unused)) |
842 | re_string_elem_size_at (const re_string_t *pstr, Idx idx) |
843 | { |
844 | # ifdef _LIBC |
845 | const unsigned char *p, *; |
846 | const int32_t *table, *indirect; |
847 | uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); |
848 | |
849 | if (nrules != 0) |
850 | { |
851 | table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); |
852 | extra = (const unsigned char *) |
853 | _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); |
854 | indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, |
855 | _NL_COLLATE_INDIRECTMB); |
856 | p = pstr->mbs + idx; |
857 | findidx (table, indirect, extra, &p, pstr->len - idx); |
858 | return p - pstr->mbs - idx; |
859 | } |
860 | else |
861 | # endif /* _LIBC */ |
862 | return 1; |
863 | } |
864 | #endif /* RE_ENABLE_I18N */ |
865 | |
866 | #ifndef FALLTHROUGH |
867 | # if __GNUC__ < 7 |
868 | # define FALLTHROUGH ((void) 0) |
869 | # else |
870 | # define FALLTHROUGH __attribute__ ((__fallthrough__)) |
871 | # endif |
872 | #endif |
873 | |
874 | #endif /* _REGEX_INTERNAL_H */ |
875 | |