Commit | Line | Data |
---|---|---|
eb4a14ed LC |
1 | /* Extended regular expression matching and search library. |
2 | Copyright (C) 2002-2012 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 | This program is free software; you can redistribute it and/or modify | |
7 | it under the terms of the GNU Lesser General Public License as published by | |
8 | the Free Software Foundation; either version 2, or (at your option) | |
9 | any later version. | |
10 | ||
11 | This program 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 | |
14 | GNU Lesser General Public License for more details. | |
15 | ||
16 | You should have received a copy of the GNU Lesser General Public License along | |
17 | with this program; if not, write to the Free Software Foundation, | |
18 | Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ | |
19 | ||
20 | #ifndef _REGEX_INTERNAL_H | |
21 | #define _REGEX_INTERNAL_H 1 | |
22 | ||
23 | #include <assert.h> | |
24 | #include <ctype.h> | |
25 | #include <stdbool.h> | |
26 | #include <stdio.h> | |
27 | #include <stdlib.h> | |
28 | #include <string.h> | |
29 | ||
30 | #include <langinfo.h> | |
31 | #ifndef _LIBC | |
32 | # include "localcharset.h" | |
33 | #endif | |
34 | #include <locale.h> | |
35 | ||
36 | #include <wchar.h> | |
37 | #include <wctype.h> | |
38 | #include <stdint.h> | |
39 | #if defined _LIBC | |
40 | # include <bits/libc-lock.h> | |
41 | #else | |
42 | # define __libc_lock_init(NAME) do { } while (0) | |
43 | # define __libc_lock_lock(NAME) do { } while (0) | |
44 | # define __libc_lock_unlock(NAME) do { } while (0) | |
45 | #endif | |
46 | ||
47 | /* In case that the system doesn't have isblank(). */ | |
48 | #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) | |
49 | # define isblank(ch) ((ch) == ' ' || (ch) == '\t') | |
50 | #endif | |
51 | ||
52 | #ifdef _LIBC | |
53 | # ifndef _RE_DEFINE_LOCALE_FUNCTIONS | |
54 | # define _RE_DEFINE_LOCALE_FUNCTIONS 1 | |
55 | # include <locale/localeinfo.h> | |
56 | # include <locale/elem-hash.h> | |
57 | # include <locale/coll-lookup.h> | |
58 | # endif | |
59 | #endif | |
60 | ||
61 | /* This is for other GNU distributions with internationalized messages. */ | |
62 | #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC | |
63 | # include <libintl.h> | |
64 | # ifdef _LIBC | |
65 | # undef gettext | |
66 | # define gettext(msgid) \ | |
67 | INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES) | |
68 | # endif | |
69 | #else | |
70 | # define gettext(msgid) (msgid) | |
71 | #endif | |
72 | ||
73 | #ifndef gettext_noop | |
74 | /* This define is so xgettext can find the internationalizable | |
75 | strings. */ | |
76 | # define gettext_noop(String) String | |
77 | #endif | |
78 | ||
79 | /* For loser systems without the definition. */ | |
80 | #ifndef SIZE_MAX | |
81 | # define SIZE_MAX ((size_t) -1) | |
82 | #endif | |
83 | ||
84 | #if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCSCOLL) || _LIBC | |
85 | # define RE_ENABLE_I18N | |
86 | #endif | |
87 | ||
88 | #if __GNUC__ >= 3 | |
89 | # define BE(expr, val) __builtin_expect (expr, val) | |
90 | #else | |
91 | # define BE(expr, val) (expr) | |
92 | # ifdef _LIBC | |
93 | # define inline | |
94 | # endif | |
95 | #endif | |
96 | ||
97 | /* Number of ASCII characters. */ | |
98 | #define ASCII_CHARS 0x80 | |
99 | ||
100 | /* Number of single byte characters. */ | |
101 | #define SBC_MAX (UCHAR_MAX + 1) | |
102 | ||
103 | #define COLL_ELEM_LEN_MAX 8 | |
104 | ||
105 | /* The character which represents newline. */ | |
106 | #define NEWLINE_CHAR '\n' | |
107 | #define WIDE_NEWLINE_CHAR L'\n' | |
108 | ||
109 | /* Rename to standard API for using out of glibc. */ | |
110 | #ifndef _LIBC | |
111 | # define __wctype wctype | |
112 | # define __iswctype iswctype | |
113 | # define __btowc btowc | |
114 | # define __wcrtomb wcrtomb | |
115 | # define __mbrtowc mbrtowc | |
116 | # define __regfree regfree | |
117 | # define attribute_hidden | |
118 | #endif /* not _LIBC */ | |
119 | ||
120 | #if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1) | |
121 | # define __attribute(arg) __attribute__ (arg) | |
122 | #else | |
123 | # define __attribute(arg) | |
124 | #endif | |
125 | ||
126 | typedef __re_idx_t Idx; | |
127 | ||
128 | /* Special return value for failure to match. */ | |
129 | #define REG_MISSING ((Idx) -1) | |
130 | ||
131 | /* Special return value for internal error. */ | |
132 | #define REG_ERROR ((Idx) -2) | |
133 | ||
134 | /* Test whether N is a valid index, and is not one of the above. */ | |
135 | #ifdef _REGEX_LARGE_OFFSETS | |
136 | # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR) | |
137 | #else | |
138 | # define REG_VALID_INDEX(n) (0 <= (n)) | |
139 | #endif | |
140 | ||
141 | /* Test whether N is a valid nonzero index. */ | |
142 | #ifdef _REGEX_LARGE_OFFSETS | |
143 | # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1)) | |
144 | #else | |
145 | # define REG_VALID_NONZERO_INDEX(n) (0 < (n)) | |
146 | #endif | |
147 | ||
148 | /* A hash value, suitable for computing hash tables. */ | |
149 | typedef __re_size_t re_hashval_t; | |
150 | ||
151 | /* An integer used to represent a set of bits. It must be unsigned, | |
152 | and must be at least as wide as unsigned int. */ | |
153 | typedef unsigned long int bitset_word_t; | |
154 | /* All bits set in a bitset_word_t. */ | |
155 | #define BITSET_WORD_MAX ULONG_MAX | |
156 | ||
157 | /* Number of bits in a bitset_word_t. For portability to hosts with | |
158 | padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)'; | |
159 | instead, deduce it directly from BITSET_WORD_MAX. Avoid | |
160 | greater-than-32-bit integers and unconditional shifts by more than | |
161 | 31 bits, as they're not portable. */ | |
162 | #if BITSET_WORD_MAX == 0xffffffffUL | |
163 | # define BITSET_WORD_BITS 32 | |
164 | #elif BITSET_WORD_MAX >> 31 >> 4 == 1 | |
165 | # define BITSET_WORD_BITS 36 | |
166 | #elif BITSET_WORD_MAX >> 31 >> 16 == 1 | |
167 | # define BITSET_WORD_BITS 48 | |
168 | #elif BITSET_WORD_MAX >> 31 >> 28 == 1 | |
169 | # define BITSET_WORD_BITS 60 | |
170 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1 | |
171 | # define BITSET_WORD_BITS 64 | |
172 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1 | |
173 | # define BITSET_WORD_BITS 72 | |
174 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1 | |
175 | # define BITSET_WORD_BITS 128 | |
176 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1 | |
177 | # define BITSET_WORD_BITS 256 | |
178 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1 | |
179 | # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */ | |
180 | # if BITSET_WORD_BITS <= SBC_MAX | |
181 | # error "Invalid SBC_MAX" | |
182 | # endif | |
183 | #else | |
184 | # error "Add case for new bitset_word_t size" | |
185 | #endif | |
186 | ||
187 | /* Number of bitset_word_t values in a bitset_t. */ | |
188 | #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) | |
189 | ||
190 | typedef bitset_word_t bitset_t[BITSET_WORDS]; | |
191 | typedef bitset_word_t *re_bitset_ptr_t; | |
192 | typedef const bitset_word_t *re_const_bitset_ptr_t; | |
193 | ||
194 | #define PREV_WORD_CONSTRAINT 0x0001 | |
195 | #define PREV_NOTWORD_CONSTRAINT 0x0002 | |
196 | #define NEXT_WORD_CONSTRAINT 0x0004 | |
197 | #define NEXT_NOTWORD_CONSTRAINT 0x0008 | |
198 | #define PREV_NEWLINE_CONSTRAINT 0x0010 | |
199 | #define NEXT_NEWLINE_CONSTRAINT 0x0020 | |
200 | #define PREV_BEGBUF_CONSTRAINT 0x0040 | |
201 | #define NEXT_ENDBUF_CONSTRAINT 0x0080 | |
202 | #define WORD_DELIM_CONSTRAINT 0x0100 | |
203 | #define NOT_WORD_DELIM_CONSTRAINT 0x0200 | |
204 | ||
205 | typedef enum | |
206 | { | |
207 | INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, | |
208 | WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, | |
209 | WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, | |
210 | INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, | |
211 | LINE_FIRST = PREV_NEWLINE_CONSTRAINT, | |
212 | LINE_LAST = NEXT_NEWLINE_CONSTRAINT, | |
213 | BUF_FIRST = PREV_BEGBUF_CONSTRAINT, | |
214 | BUF_LAST = NEXT_ENDBUF_CONSTRAINT, | |
215 | WORD_DELIM = WORD_DELIM_CONSTRAINT, | |
216 | NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT | |
217 | } re_context_type; | |
218 | ||
219 | typedef struct | |
220 | { | |
221 | Idx alloc; | |
222 | Idx nelem; | |
223 | Idx *elems; | |
224 | } re_node_set; | |
225 | ||
226 | typedef enum | |
227 | { | |
228 | NON_TYPE = 0, | |
229 | ||
230 | /* Node type, These are used by token, node, tree. */ | |
231 | CHARACTER = 1, | |
232 | END_OF_RE = 2, | |
233 | SIMPLE_BRACKET = 3, | |
234 | OP_BACK_REF = 4, | |
235 | OP_PERIOD = 5, | |
236 | #ifdef RE_ENABLE_I18N | |
237 | COMPLEX_BRACKET = 6, | |
238 | OP_UTF8_PERIOD = 7, | |
239 | #endif /* RE_ENABLE_I18N */ | |
240 | ||
241 | /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used | |
242 | when the debugger shows values of this enum type. */ | |
243 | #define EPSILON_BIT 8 | |
244 | OP_OPEN_SUBEXP = EPSILON_BIT | 0, | |
245 | OP_CLOSE_SUBEXP = EPSILON_BIT | 1, | |
246 | OP_ALT = EPSILON_BIT | 2, | |
247 | OP_DUP_ASTERISK = EPSILON_BIT | 3, | |
248 | ANCHOR = EPSILON_BIT | 4, | |
249 | ||
250 | /* Tree type, these are used only by tree. */ | |
251 | CONCAT = 16, | |
252 | SUBEXP = 17, | |
253 | ||
254 | /* Token type, these are used only by token. */ | |
255 | OP_DUP_PLUS = 18, | |
256 | OP_DUP_QUESTION, | |
257 | OP_OPEN_BRACKET, | |
258 | OP_CLOSE_BRACKET, | |
259 | OP_CHARSET_RANGE, | |
260 | OP_OPEN_DUP_NUM, | |
261 | OP_CLOSE_DUP_NUM, | |
262 | OP_NON_MATCH_LIST, | |
263 | OP_OPEN_COLL_ELEM, | |
264 | OP_CLOSE_COLL_ELEM, | |
265 | OP_OPEN_EQUIV_CLASS, | |
266 | OP_CLOSE_EQUIV_CLASS, | |
267 | OP_OPEN_CHAR_CLASS, | |
268 | OP_CLOSE_CHAR_CLASS, | |
269 | OP_WORD, | |
270 | OP_NOTWORD, | |
271 | OP_SPACE, | |
272 | OP_NOTSPACE, | |
273 | BACK_SLASH | |
274 | ||
275 | } re_token_type_t; | |
276 | ||
277 | #ifdef RE_ENABLE_I18N | |
278 | typedef struct | |
279 | { | |
280 | /* Multibyte characters. */ | |
281 | wchar_t *mbchars; | |
282 | ||
283 | /* Collating symbols. */ | |
284 | # ifdef _LIBC | |
285 | int32_t *coll_syms; | |
286 | # endif | |
287 | ||
288 | /* Equivalence classes. */ | |
289 | # ifdef _LIBC | |
290 | int32_t *equiv_classes; | |
291 | # endif | |
292 | ||
293 | /* Range expressions. */ | |
294 | # ifdef _LIBC | |
295 | uint32_t *range_starts; | |
296 | uint32_t *range_ends; | |
297 | # else /* not _LIBC */ | |
298 | wchar_t *range_starts; | |
299 | wchar_t *range_ends; | |
300 | # endif /* not _LIBC */ | |
301 | ||
302 | /* Character classes. */ | |
303 | wctype_t *char_classes; | |
304 | ||
305 | /* If this character set is the non-matching list. */ | |
306 | unsigned int non_match : 1; | |
307 | ||
308 | /* # of multibyte characters. */ | |
309 | Idx nmbchars; | |
310 | ||
311 | /* # of collating symbols. */ | |
312 | Idx ncoll_syms; | |
313 | ||
314 | /* # of equivalence classes. */ | |
315 | Idx nequiv_classes; | |
316 | ||
317 | /* # of range expressions. */ | |
318 | Idx nranges; | |
319 | ||
320 | /* # of character classes. */ | |
321 | Idx nchar_classes; | |
322 | } re_charset_t; | |
323 | #endif /* RE_ENABLE_I18N */ | |
324 | ||
325 | typedef struct | |
326 | { | |
327 | union | |
328 | { | |
329 | unsigned char c; /* for CHARACTER */ | |
330 | re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ | |
331 | #ifdef RE_ENABLE_I18N | |
332 | re_charset_t *mbcset; /* for COMPLEX_BRACKET */ | |
333 | #endif /* RE_ENABLE_I18N */ | |
334 | Idx idx; /* for BACK_REF */ | |
335 | re_context_type ctx_type; /* for ANCHOR */ | |
336 | } opr; | |
337 | #if __GNUC__ >= 2 && !defined __STRICT_ANSI__ | |
338 | re_token_type_t type : 8; | |
339 | #else | |
340 | re_token_type_t type; | |
341 | #endif | |
342 | unsigned int constraint : 10; /* context constraint */ | |
343 | unsigned int duplicated : 1; | |
344 | unsigned int opt_subexp : 1; | |
345 | #ifdef RE_ENABLE_I18N | |
346 | unsigned int accept_mb : 1; | |
347 | /* These 2 bits can be moved into the union if needed (e.g. if running out | |
348 | of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ | |
349 | unsigned int mb_partial : 1; | |
350 | #endif | |
351 | unsigned int word_char : 1; | |
352 | } re_token_t; | |
353 | ||
354 | #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) | |
355 | ||
356 | struct re_string_t | |
357 | { | |
358 | /* Indicate the raw buffer which is the original string passed as an | |
359 | argument of regexec(), re_search(), etc.. */ | |
360 | const unsigned char *raw_mbs; | |
361 | /* Store the multibyte string. In case of "case insensitive mode" like | |
362 | REG_ICASE, upper cases of the string are stored, otherwise MBS points | |
363 | the same address that RAW_MBS points. */ | |
364 | unsigned char *mbs; | |
365 | #ifdef RE_ENABLE_I18N | |
366 | /* Store the wide character string which is corresponding to MBS. */ | |
367 | wint_t *wcs; | |
368 | Idx *offsets; | |
369 | mbstate_t cur_state; | |
370 | #endif | |
371 | /* Index in RAW_MBS. Each character mbs[i] corresponds to | |
372 | raw_mbs[raw_mbs_idx + i]. */ | |
373 | Idx raw_mbs_idx; | |
374 | /* The length of the valid characters in the buffers. */ | |
375 | Idx valid_len; | |
376 | /* The corresponding number of bytes in raw_mbs array. */ | |
377 | Idx valid_raw_len; | |
378 | /* The length of the buffers MBS and WCS. */ | |
379 | Idx bufs_len; | |
380 | /* The index in MBS, which is updated by re_string_fetch_byte. */ | |
381 | Idx cur_idx; | |
382 | /* length of RAW_MBS array. */ | |
383 | Idx raw_len; | |
384 | /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ | |
385 | Idx len; | |
386 | /* End of the buffer may be shorter than its length in the cases such | |
387 | as re_match_2, re_search_2. Then, we use STOP for end of the buffer | |
388 | instead of LEN. */ | |
389 | Idx raw_stop; | |
390 | /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ | |
391 | Idx stop; | |
392 | ||
393 | /* The context of mbs[0]. We store the context independently, since | |
394 | the context of mbs[0] may be different from raw_mbs[0], which is | |
395 | the beginning of the input string. */ | |
396 | unsigned int tip_context; | |
397 | /* The translation passed as a part of an argument of re_compile_pattern. */ | |
398 | RE_TRANSLATE_TYPE trans; | |
399 | /* Copy of re_dfa_t's word_char. */ | |
400 | re_const_bitset_ptr_t word_char; | |
401 | /* true if REG_ICASE. */ | |
402 | unsigned char icase; | |
403 | unsigned char is_utf8; | |
404 | unsigned char map_notascii; | |
405 | unsigned char mbs_allocated; | |
406 | unsigned char offsets_needed; | |
407 | unsigned char newline_anchor; | |
408 | unsigned char word_ops_used; | |
409 | int mb_cur_max; | |
410 | }; | |
411 | typedef struct re_string_t re_string_t; | |
412 | ||
413 | ||
414 | struct re_dfa_t; | |
415 | typedef struct re_dfa_t re_dfa_t; | |
416 | ||
417 | #ifndef _LIBC | |
418 | # define internal_function | |
419 | #endif | |
420 | ||
421 | static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, | |
422 | Idx new_buf_len) | |
423 | internal_function; | |
424 | #ifdef RE_ENABLE_I18N | |
425 | static void build_wcs_buffer (re_string_t *pstr) internal_function; | |
426 | static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr) | |
427 | internal_function; | |
428 | #endif /* RE_ENABLE_I18N */ | |
429 | static void build_upper_buffer (re_string_t *pstr) internal_function; | |
430 | static void re_string_translate_buffer (re_string_t *pstr) internal_function; | |
431 | static unsigned int re_string_context_at (const re_string_t *input, Idx idx, | |
432 | int eflags) | |
433 | internal_function __attribute ((pure)); | |
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 | #include <alloca.h> | |
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 | #ifndef MAX | |
469 | # define MAX(a,b) ((a) < (b) ? (b) : (a)) | |
470 | #endif | |
471 | ||
472 | #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) | |
473 | #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) | |
474 | #define re_free(p) free (p) | |
475 | ||
476 | struct bin_tree_t | |
477 | { | |
478 | struct bin_tree_t *parent; | |
479 | struct bin_tree_t *left; | |
480 | struct bin_tree_t *right; | |
481 | struct bin_tree_t *first; | |
482 | struct bin_tree_t *next; | |
483 | ||
484 | re_token_t token; | |
485 | ||
486 | /* 'node_idx' is the index in dfa->nodes, if 'type' == 0. | |
487 | Otherwise 'type' indicate the type of this node. */ | |
488 | Idx node_idx; | |
489 | }; | |
490 | typedef struct bin_tree_t bin_tree_t; | |
491 | ||
492 | #define BIN_TREE_STORAGE_SIZE \ | |
493 | ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) | |
494 | ||
495 | struct bin_tree_storage_t | |
496 | { | |
497 | struct bin_tree_storage_t *next; | |
498 | bin_tree_t data[BIN_TREE_STORAGE_SIZE]; | |
499 | }; | |
500 | typedef struct bin_tree_storage_t bin_tree_storage_t; | |
501 | ||
502 | #define CONTEXT_WORD 1 | |
503 | #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) | |
504 | #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) | |
505 | #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) | |
506 | ||
507 | #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) | |
508 | #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) | |
509 | #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) | |
510 | #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) | |
511 | #define IS_ORDINARY_CONTEXT(c) ((c) == 0) | |
512 | ||
513 | #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') | |
514 | #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) | |
515 | #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_') | |
516 | #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) | |
517 | ||
518 | #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ | |
519 | ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ | |
520 | || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ | |
521 | || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ | |
522 | || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) | |
523 | ||
524 | #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ | |
525 | ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ | |
526 | || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ | |
527 | || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ | |
528 | || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) | |
529 | ||
530 | struct re_dfastate_t | |
531 | { | |
532 | re_hashval_t hash; | |
533 | re_node_set nodes; | |
534 | re_node_set non_eps_nodes; | |
535 | re_node_set inveclosure; | |
536 | re_node_set *entrance_nodes; | |
537 | struct re_dfastate_t **trtable, **word_trtable; | |
538 | unsigned int context : 4; | |
539 | unsigned int halt : 1; | |
540 | /* If this state can accept "multi byte". | |
541 | Note that we refer to multibyte characters, and multi character | |
542 | collating elements as "multi byte". */ | |
543 | unsigned int accept_mb : 1; | |
544 | /* If this state has backreference node(s). */ | |
545 | unsigned int has_backref : 1; | |
546 | unsigned int has_constraint : 1; | |
547 | }; | |
548 | typedef struct re_dfastate_t re_dfastate_t; | |
549 | ||
550 | struct re_state_table_entry | |
551 | { | |
552 | Idx num; | |
553 | Idx alloc; | |
554 | re_dfastate_t **array; | |
555 | }; | |
556 | ||
557 | /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ | |
558 | ||
559 | typedef struct | |
560 | { | |
561 | Idx next_idx; | |
562 | Idx alloc; | |
563 | re_dfastate_t **array; | |
564 | } state_array_t; | |
565 | ||
566 | /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ | |
567 | ||
568 | typedef struct | |
569 | { | |
570 | Idx node; | |
571 | Idx str_idx; /* The position NODE match at. */ | |
572 | state_array_t path; | |
573 | } re_sub_match_last_t; | |
574 | ||
575 | /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. | |
576 | And information about the node, whose type is OP_CLOSE_SUBEXP, | |
577 | corresponding to NODE is stored in LASTS. */ | |
578 | ||
579 | typedef struct | |
580 | { | |
581 | Idx str_idx; | |
582 | Idx node; | |
583 | state_array_t *path; | |
584 | Idx alasts; /* Allocation size of LASTS. */ | |
585 | Idx nlasts; /* The number of LASTS. */ | |
586 | re_sub_match_last_t **lasts; | |
587 | } re_sub_match_top_t; | |
588 | ||
589 | struct re_backref_cache_entry | |
590 | { | |
591 | Idx node; | |
592 | Idx str_idx; | |
593 | Idx subexp_from; | |
594 | Idx subexp_to; | |
595 | char more; | |
596 | char unused; | |
597 | unsigned short int eps_reachable_subexps_map; | |
598 | }; | |
599 | ||
600 | typedef struct | |
601 | { | |
602 | /* The string object corresponding to the input string. */ | |
603 | re_string_t input; | |
604 | #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) | |
605 | const re_dfa_t *const dfa; | |
606 | #else | |
607 | const re_dfa_t *dfa; | |
608 | #endif | |
609 | /* EFLAGS of the argument of regexec. */ | |
610 | int eflags; | |
611 | /* Where the matching ends. */ | |
612 | Idx match_last; | |
613 | Idx last_node; | |
614 | /* The state log used by the matcher. */ | |
615 | re_dfastate_t **state_log; | |
616 | Idx state_log_top; | |
617 | /* Back reference cache. */ | |
618 | Idx nbkref_ents; | |
619 | Idx abkref_ents; | |
620 | struct re_backref_cache_entry *bkref_ents; | |
621 | int max_mb_elem_len; | |
622 | Idx nsub_tops; | |
623 | Idx asub_tops; | |
624 | re_sub_match_top_t **sub_tops; | |
625 | } re_match_context_t; | |
626 | ||
627 | typedef struct | |
628 | { | |
629 | re_dfastate_t **sifted_states; | |
630 | re_dfastate_t **limited_states; | |
631 | Idx last_node; | |
632 | Idx last_str_idx; | |
633 | re_node_set limits; | |
634 | } re_sift_context_t; | |
635 | ||
636 | struct re_fail_stack_ent_t | |
637 | { | |
638 | Idx idx; | |
639 | Idx node; | |
640 | regmatch_t *regs; | |
641 | re_node_set eps_via_nodes; | |
642 | }; | |
643 | ||
644 | struct re_fail_stack_t | |
645 | { | |
646 | Idx num; | |
647 | Idx alloc; | |
648 | struct re_fail_stack_ent_t *stack; | |
649 | }; | |
650 | ||
651 | struct re_dfa_t | |
652 | { | |
653 | re_token_t *nodes; | |
654 | size_t nodes_alloc; | |
655 | size_t nodes_len; | |
656 | Idx *nexts; | |
657 | Idx *org_indices; | |
658 | re_node_set *edests; | |
659 | re_node_set *eclosures; | |
660 | re_node_set *inveclosures; | |
661 | struct re_state_table_entry *state_table; | |
662 | re_dfastate_t *init_state; | |
663 | re_dfastate_t *init_state_word; | |
664 | re_dfastate_t *init_state_nl; | |
665 | re_dfastate_t *init_state_begbuf; | |
666 | bin_tree_t *str_tree; | |
667 | bin_tree_storage_t *str_tree_storage; | |
668 | re_bitset_ptr_t sb_char; | |
669 | int str_tree_storage_idx; | |
670 | ||
671 | /* number of subexpressions 're_nsub' is in regex_t. */ | |
672 | re_hashval_t state_hash_mask; | |
673 | Idx init_node; | |
674 | Idx nbackref; /* The number of backreference in this dfa. */ | |
675 | ||
676 | /* Bitmap expressing which backreference is used. */ | |
677 | bitset_word_t used_bkref_map; | |
678 | bitset_word_t completed_bkref_map; | |
679 | ||
680 | unsigned int has_plural_match : 1; | |
681 | /* If this dfa has "multibyte node", which is a backreference or | |
682 | a node which can accept multibyte character or multi character | |
683 | collating element. */ | |
684 | unsigned int has_mb_node : 1; | |
685 | unsigned int is_utf8 : 1; | |
686 | unsigned int map_notascii : 1; | |
687 | unsigned int word_ops_used : 1; | |
688 | int mb_cur_max; | |
689 | bitset_t word_char; | |
690 | reg_syntax_t syntax; | |
691 | Idx *subexp_map; | |
692 | #ifdef DEBUG | |
693 | char* re_str; | |
694 | #endif | |
695 | #ifdef _LIBC | |
696 | __libc_lock_define (, lock) | |
697 | #endif | |
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 | \f | |
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 | /* Inline 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 | /* Inline functions for re_string. */ | |
799 | static inline int | |
800 | internal_function __attribute ((pure)) | |
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 inline wint_t | |
813 | internal_function __attribute ((pure)) | |
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 | static int | |
822 | internal_function __attribute ((pure)) | |
823 | re_string_elem_size_at (const re_string_t *pstr, Idx idx) | |
824 | { | |
825 | # ifdef _LIBC | |
826 | const unsigned char *p, *extra; | |
827 | const int32_t *table, *indirect; | |
828 | int32_t tmp; | |
829 | # include <locale/weight.h> | |
830 | uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); | |
831 | ||
832 | if (nrules != 0) | |
833 | { | |
834 | table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); | |
835 | extra = (const unsigned char *) | |
836 | _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); | |
837 | indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, | |
838 | _NL_COLLATE_INDIRECTMB); | |
839 | p = pstr->mbs + idx; | |
840 | tmp = findidx (&p); | |
841 | return p - pstr->mbs - idx; | |
842 | } | |
843 | else | |
844 | # endif /* _LIBC */ | |
845 | return 1; | |
846 | } | |
847 | #endif /* RE_ENABLE_I18N */ | |
848 | ||
849 | #ifndef __GNUC_PREREQ | |
850 | # if defined __GNUC__ && defined __GNUC_MINOR__ | |
851 | # define __GNUC_PREREQ(maj, min) \ | |
852 | ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min)) | |
853 | # else | |
854 | # define __GNUC_PREREQ(maj, min) 0 | |
855 | # endif | |
856 | #endif | |
857 | ||
858 | #if __GNUC_PREREQ (3,4) | |
859 | # undef __attribute_warn_unused_result__ | |
860 | # define __attribute_warn_unused_result__ \ | |
861 | __attribute__ ((__warn_unused_result__)) | |
862 | #else | |
863 | # define __attribute_warn_unused_result__ /* empty */ | |
864 | #endif | |
865 | ||
866 | #endif /* _REGEX_INTERNAL_H */ |