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 | |
005de2e8 | 17 | with this program; if not, see <http://www.gnu.org/licenses/>. */ |
eb4a14ed LC |
18 | |
19 | #ifndef _REGEX_INTERNAL_H | |
20 | #define _REGEX_INTERNAL_H 1 | |
21 | ||
22 | #include <assert.h> | |
23 | #include <ctype.h> | |
eb4a14ed LC |
24 | #include <stdio.h> |
25 | #include <stdlib.h> | |
26 | #include <string.h> | |
27 | ||
28 | #include <langinfo.h> | |
29 | #ifndef _LIBC | |
30 | # include "localcharset.h" | |
31 | #endif | |
32 | #include <locale.h> | |
eb4a14ed LC |
33 | #include <wchar.h> |
34 | #include <wctype.h> | |
005de2e8 | 35 | #include <stdbool.h> |
eb4a14ed LC |
36 | #include <stdint.h> |
37 | #if defined _LIBC | |
38 | # include <bits/libc-lock.h> | |
39 | #else | |
005de2e8 | 40 | # define __libc_lock_define(CLASS,NAME) |
eb4a14ed LC |
41 | # define __libc_lock_init(NAME) do { } while (0) |
42 | # define __libc_lock_lock(NAME) do { } while (0) | |
43 | # define __libc_lock_unlock(NAME) do { } while (0) | |
44 | #endif | |
45 | ||
46 | /* In case that the system doesn't have isblank(). */ | |
47 | #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) | |
48 | # define isblank(ch) ((ch) == ' ' || (ch) == '\t') | |
49 | #endif | |
50 | ||
51 | #ifdef _LIBC | |
52 | # ifndef _RE_DEFINE_LOCALE_FUNCTIONS | |
53 | # define _RE_DEFINE_LOCALE_FUNCTIONS 1 | |
54 | # include <locale/localeinfo.h> | |
55 | # include <locale/elem-hash.h> | |
56 | # include <locale/coll-lookup.h> | |
57 | # endif | |
58 | #endif | |
59 | ||
60 | /* This is for other GNU distributions with internationalized messages. */ | |
61 | #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC | |
62 | # include <libintl.h> | |
63 | # ifdef _LIBC | |
64 | # undef gettext | |
65 | # define gettext(msgid) \ | |
66 | INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES) | |
67 | # endif | |
68 | #else | |
69 | # define gettext(msgid) (msgid) | |
70 | #endif | |
71 | ||
72 | #ifndef gettext_noop | |
73 | /* This define is so xgettext can find the internationalizable | |
74 | strings. */ | |
75 | # define gettext_noop(String) String | |
76 | #endif | |
77 | ||
eb4a14ed LC |
78 | #if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCSCOLL) || _LIBC |
79 | # define RE_ENABLE_I18N | |
80 | #endif | |
81 | ||
82 | #if __GNUC__ >= 3 | |
83 | # define BE(expr, val) __builtin_expect (expr, val) | |
84 | #else | |
85 | # define BE(expr, val) (expr) | |
86 | # ifdef _LIBC | |
87 | # define inline | |
88 | # endif | |
89 | #endif | |
90 | ||
91 | /* Number of ASCII characters. */ | |
92 | #define ASCII_CHARS 0x80 | |
93 | ||
94 | /* Number of single byte characters. */ | |
95 | #define SBC_MAX (UCHAR_MAX + 1) | |
96 | ||
97 | #define COLL_ELEM_LEN_MAX 8 | |
98 | ||
99 | /* The character which represents newline. */ | |
100 | #define NEWLINE_CHAR '\n' | |
101 | #define WIDE_NEWLINE_CHAR L'\n' | |
102 | ||
103 | /* Rename to standard API for using out of glibc. */ | |
104 | #ifndef _LIBC | |
105 | # define __wctype wctype | |
106 | # define __iswctype iswctype | |
107 | # define __btowc btowc | |
eb4a14ed | 108 | # define __mbrtowc mbrtowc |
005de2e8 | 109 | # define __wcrtomb wcrtomb |
eb4a14ed LC |
110 | # define __regfree regfree |
111 | # define attribute_hidden | |
112 | #endif /* not _LIBC */ | |
113 | ||
114 | #if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1) | |
115 | # define __attribute(arg) __attribute__ (arg) | |
116 | #else | |
117 | # define __attribute(arg) | |
118 | #endif | |
119 | ||
120 | typedef __re_idx_t Idx; | |
005de2e8 LC |
121 | #ifdef _REGEX_LARGE_OFFSETS |
122 | # define IDX_MAX (SIZE_MAX - 2) | |
123 | #else | |
124 | # define IDX_MAX INT_MAX | |
125 | #endif | |
eb4a14ed LC |
126 | |
127 | /* Special return value for failure to match. */ | |
128 | #define REG_MISSING ((Idx) -1) | |
129 | ||
130 | /* Special return value for internal error. */ | |
131 | #define REG_ERROR ((Idx) -2) | |
132 | ||
133 | /* Test whether N is a valid index, and is not one of the above. */ | |
134 | #ifdef _REGEX_LARGE_OFFSETS | |
135 | # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR) | |
136 | #else | |
137 | # define REG_VALID_INDEX(n) (0 <= (n)) | |
138 | #endif | |
139 | ||
140 | /* Test whether N is a valid nonzero index. */ | |
141 | #ifdef _REGEX_LARGE_OFFSETS | |
142 | # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1)) | |
143 | #else | |
144 | # define REG_VALID_NONZERO_INDEX(n) (0 < (n)) | |
145 | #endif | |
146 | ||
147 | /* A hash value, suitable for computing hash tables. */ | |
148 | typedef __re_size_t re_hashval_t; | |
149 | ||
150 | /* An integer used to represent a set of bits. It must be unsigned, | |
151 | and must be at least as wide as unsigned int. */ | |
152 | typedef unsigned long int bitset_word_t; | |
153 | /* All bits set in a bitset_word_t. */ | |
154 | #define BITSET_WORD_MAX ULONG_MAX | |
155 | ||
156 | /* Number of bits in a bitset_word_t. For portability to hosts with | |
157 | padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)'; | |
158 | instead, deduce it directly from BITSET_WORD_MAX. Avoid | |
159 | greater-than-32-bit integers and unconditional shifts by more than | |
160 | 31 bits, as they're not portable. */ | |
161 | #if BITSET_WORD_MAX == 0xffffffffUL | |
162 | # define BITSET_WORD_BITS 32 | |
163 | #elif BITSET_WORD_MAX >> 31 >> 4 == 1 | |
164 | # define BITSET_WORD_BITS 36 | |
165 | #elif BITSET_WORD_MAX >> 31 >> 16 == 1 | |
166 | # define BITSET_WORD_BITS 48 | |
167 | #elif BITSET_WORD_MAX >> 31 >> 28 == 1 | |
168 | # define BITSET_WORD_BITS 60 | |
169 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1 | |
170 | # define BITSET_WORD_BITS 64 | |
171 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1 | |
172 | # define BITSET_WORD_BITS 72 | |
173 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1 | |
174 | # define BITSET_WORD_BITS 128 | |
175 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1 | |
176 | # define BITSET_WORD_BITS 256 | |
177 | #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1 | |
178 | # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */ | |
179 | # if BITSET_WORD_BITS <= SBC_MAX | |
180 | # error "Invalid SBC_MAX" | |
181 | # endif | |
182 | #else | |
183 | # error "Add case for new bitset_word_t size" | |
184 | #endif | |
185 | ||
186 | /* Number of bitset_word_t values in a bitset_t. */ | |
187 | #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) | |
188 | ||
189 | typedef bitset_word_t bitset_t[BITSET_WORDS]; | |
190 | typedef bitset_word_t *re_bitset_ptr_t; | |
191 | typedef const bitset_word_t *re_const_bitset_ptr_t; | |
192 | ||
193 | #define PREV_WORD_CONSTRAINT 0x0001 | |
194 | #define PREV_NOTWORD_CONSTRAINT 0x0002 | |
195 | #define NEXT_WORD_CONSTRAINT 0x0004 | |
196 | #define NEXT_NOTWORD_CONSTRAINT 0x0008 | |
197 | #define PREV_NEWLINE_CONSTRAINT 0x0010 | |
198 | #define NEXT_NEWLINE_CONSTRAINT 0x0020 | |
199 | #define PREV_BEGBUF_CONSTRAINT 0x0040 | |
200 | #define NEXT_ENDBUF_CONSTRAINT 0x0080 | |
201 | #define WORD_DELIM_CONSTRAINT 0x0100 | |
202 | #define NOT_WORD_DELIM_CONSTRAINT 0x0200 | |
203 | ||
204 | typedef enum | |
205 | { | |
206 | INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, | |
207 | WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, | |
208 | WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, | |
209 | INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, | |
210 | LINE_FIRST = PREV_NEWLINE_CONSTRAINT, | |
211 | LINE_LAST = NEXT_NEWLINE_CONSTRAINT, | |
212 | BUF_FIRST = PREV_BEGBUF_CONSTRAINT, | |
213 | BUF_LAST = NEXT_ENDBUF_CONSTRAINT, | |
214 | WORD_DELIM = WORD_DELIM_CONSTRAINT, | |
215 | NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT | |
216 | } re_context_type; | |
217 | ||
218 | typedef struct | |
219 | { | |
220 | Idx alloc; | |
221 | Idx nelem; | |
222 | Idx *elems; | |
223 | } re_node_set; | |
224 | ||
225 | typedef enum | |
226 | { | |
227 | NON_TYPE = 0, | |
228 | ||
229 | /* Node type, These are used by token, node, tree. */ | |
230 | CHARACTER = 1, | |
231 | END_OF_RE = 2, | |
232 | SIMPLE_BRACKET = 3, | |
233 | OP_BACK_REF = 4, | |
234 | OP_PERIOD = 5, | |
235 | #ifdef RE_ENABLE_I18N | |
236 | COMPLEX_BRACKET = 6, | |
237 | OP_UTF8_PERIOD = 7, | |
238 | #endif /* RE_ENABLE_I18N */ | |
239 | ||
240 | /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used | |
241 | when the debugger shows values of this enum type. */ | |
242 | #define EPSILON_BIT 8 | |
243 | OP_OPEN_SUBEXP = EPSILON_BIT | 0, | |
244 | OP_CLOSE_SUBEXP = EPSILON_BIT | 1, | |
245 | OP_ALT = EPSILON_BIT | 2, | |
246 | OP_DUP_ASTERISK = EPSILON_BIT | 3, | |
247 | ANCHOR = EPSILON_BIT | 4, | |
248 | ||
249 | /* Tree type, these are used only by tree. */ | |
250 | CONCAT = 16, | |
251 | SUBEXP = 17, | |
252 | ||
253 | /* Token type, these are used only by token. */ | |
254 | OP_DUP_PLUS = 18, | |
255 | OP_DUP_QUESTION, | |
256 | OP_OPEN_BRACKET, | |
257 | OP_CLOSE_BRACKET, | |
258 | OP_CHARSET_RANGE, | |
259 | OP_OPEN_DUP_NUM, | |
260 | OP_CLOSE_DUP_NUM, | |
261 | OP_NON_MATCH_LIST, | |
262 | OP_OPEN_COLL_ELEM, | |
263 | OP_CLOSE_COLL_ELEM, | |
264 | OP_OPEN_EQUIV_CLASS, | |
265 | OP_CLOSE_EQUIV_CLASS, | |
266 | OP_OPEN_CHAR_CLASS, | |
267 | OP_CLOSE_CHAR_CLASS, | |
268 | OP_WORD, | |
269 | OP_NOTWORD, | |
270 | OP_SPACE, | |
271 | OP_NOTSPACE, | |
272 | BACK_SLASH | |
273 | ||
274 | } re_token_type_t; | |
275 | ||
276 | #ifdef RE_ENABLE_I18N | |
277 | typedef struct | |
278 | { | |
279 | /* Multibyte characters. */ | |
280 | wchar_t *mbchars; | |
281 | ||
282 | /* Collating symbols. */ | |
283 | # ifdef _LIBC | |
284 | int32_t *coll_syms; | |
285 | # endif | |
286 | ||
287 | /* Equivalence classes. */ | |
288 | # ifdef _LIBC | |
289 | int32_t *equiv_classes; | |
290 | # endif | |
291 | ||
292 | /* Range expressions. */ | |
293 | # ifdef _LIBC | |
294 | uint32_t *range_starts; | |
295 | uint32_t *range_ends; | |
296 | # else /* not _LIBC */ | |
297 | wchar_t *range_starts; | |
298 | wchar_t *range_ends; | |
299 | # endif /* not _LIBC */ | |
300 | ||
301 | /* Character classes. */ | |
302 | wctype_t *char_classes; | |
303 | ||
304 | /* If this character set is the non-matching list. */ | |
305 | unsigned int non_match : 1; | |
306 | ||
307 | /* # of multibyte characters. */ | |
308 | Idx nmbchars; | |
309 | ||
310 | /* # of collating symbols. */ | |
311 | Idx ncoll_syms; | |
312 | ||
313 | /* # of equivalence classes. */ | |
314 | Idx nequiv_classes; | |
315 | ||
316 | /* # of range expressions. */ | |
317 | Idx nranges; | |
318 | ||
319 | /* # of character classes. */ | |
320 | Idx nchar_classes; | |
321 | } re_charset_t; | |
322 | #endif /* RE_ENABLE_I18N */ | |
323 | ||
324 | typedef struct | |
325 | { | |
326 | union | |
327 | { | |
328 | unsigned char c; /* for CHARACTER */ | |
329 | re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ | |
330 | #ifdef RE_ENABLE_I18N | |
331 | re_charset_t *mbcset; /* for COMPLEX_BRACKET */ | |
332 | #endif /* RE_ENABLE_I18N */ | |
333 | Idx idx; /* for BACK_REF */ | |
334 | re_context_type ctx_type; /* for ANCHOR */ | |
335 | } opr; | |
336 | #if __GNUC__ >= 2 && !defined __STRICT_ANSI__ | |
337 | re_token_type_t type : 8; | |
338 | #else | |
339 | re_token_type_t type; | |
340 | #endif | |
341 | unsigned int constraint : 10; /* context constraint */ | |
342 | unsigned int duplicated : 1; | |
343 | unsigned int opt_subexp : 1; | |
344 | #ifdef RE_ENABLE_I18N | |
345 | unsigned int accept_mb : 1; | |
346 | /* These 2 bits can be moved into the union if needed (e.g. if running out | |
347 | of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ | |
348 | unsigned int mb_partial : 1; | |
349 | #endif | |
350 | unsigned int word_char : 1; | |
351 | } re_token_t; | |
352 | ||
353 | #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) | |
354 | ||
355 | struct re_string_t | |
356 | { | |
357 | /* Indicate the raw buffer which is the original string passed as an | |
358 | argument of regexec(), re_search(), etc.. */ | |
359 | const unsigned char *raw_mbs; | |
360 | /* Store the multibyte string. In case of "case insensitive mode" like | |
361 | REG_ICASE, upper cases of the string are stored, otherwise MBS points | |
362 | the same address that RAW_MBS points. */ | |
363 | unsigned char *mbs; | |
364 | #ifdef RE_ENABLE_I18N | |
365 | /* Store the wide character string which is corresponding to MBS. */ | |
366 | wint_t *wcs; | |
367 | Idx *offsets; | |
368 | mbstate_t cur_state; | |
369 | #endif | |
370 | /* Index in RAW_MBS. Each character mbs[i] corresponds to | |
371 | raw_mbs[raw_mbs_idx + i]. */ | |
372 | Idx raw_mbs_idx; | |
373 | /* The length of the valid characters in the buffers. */ | |
374 | Idx valid_len; | |
375 | /* The corresponding number of bytes in raw_mbs array. */ | |
376 | Idx valid_raw_len; | |
377 | /* The length of the buffers MBS and WCS. */ | |
378 | Idx bufs_len; | |
379 | /* The index in MBS, which is updated by re_string_fetch_byte. */ | |
380 | Idx cur_idx; | |
381 | /* length of RAW_MBS array. */ | |
382 | Idx raw_len; | |
383 | /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ | |
384 | Idx len; | |
385 | /* End of the buffer may be shorter than its length in the cases such | |
386 | as re_match_2, re_search_2. Then, we use STOP for end of the buffer | |
387 | instead of LEN. */ | |
388 | Idx raw_stop; | |
389 | /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ | |
390 | Idx stop; | |
391 | ||
392 | /* The context of mbs[0]. We store the context independently, since | |
393 | the context of mbs[0] may be different from raw_mbs[0], which is | |
394 | the beginning of the input string. */ | |
395 | unsigned int tip_context; | |
396 | /* The translation passed as a part of an argument of re_compile_pattern. */ | |
397 | RE_TRANSLATE_TYPE trans; | |
398 | /* Copy of re_dfa_t's word_char. */ | |
399 | re_const_bitset_ptr_t word_char; | |
400 | /* true if REG_ICASE. */ | |
401 | unsigned char icase; | |
402 | unsigned char is_utf8; | |
403 | unsigned char map_notascii; | |
404 | unsigned char mbs_allocated; | |
405 | unsigned char offsets_needed; | |
406 | unsigned char newline_anchor; | |
407 | unsigned char word_ops_used; | |
408 | int mb_cur_max; | |
409 | }; | |
410 | typedef struct re_string_t re_string_t; | |
411 | ||
412 | ||
413 | struct re_dfa_t; | |
414 | typedef struct re_dfa_t re_dfa_t; | |
415 | ||
416 | #ifndef _LIBC | |
417 | # define internal_function | |
418 | #endif | |
419 | ||
005de2e8 | 420 | #ifndef NOT_IN_libc |
eb4a14ed LC |
421 | static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, |
422 | Idx new_buf_len) | |
423 | internal_function; | |
005de2e8 | 424 | # ifdef RE_ENABLE_I18N |
eb4a14ed LC |
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) | |
005de2e8 LC |
427 | internal_function; |
428 | # endif /* RE_ENABLE_I18N */ | |
eb4a14ed LC |
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)); | |
005de2e8 | 434 | #endif |
eb4a14ed LC |
435 | #define re_string_peek_byte(pstr, offset) \ |
436 | ((pstr)->mbs[(pstr)->cur_idx + offset]) | |
437 | #define re_string_fetch_byte(pstr) \ | |
438 | ((pstr)->mbs[(pstr)->cur_idx++]) | |
439 | #define re_string_first_byte(pstr, idx) \ | |
440 | ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) | |
441 | #define re_string_is_single_byte_char(pstr, idx) \ | |
442 | ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ | |
443 | || (pstr)->wcs[(idx) + 1] != WEOF)) | |
444 | #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) | |
445 | #define re_string_cur_idx(pstr) ((pstr)->cur_idx) | |
446 | #define re_string_get_buffer(pstr) ((pstr)->mbs) | |
447 | #define re_string_length(pstr) ((pstr)->len) | |
448 | #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) | |
449 | #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) | |
450 | #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) | |
451 | ||
452 | #include <alloca.h> | |
453 | ||
454 | #ifndef _LIBC | |
455 | # if HAVE_ALLOCA | |
456 | /* The OS usually guarantees only one guard page at the bottom of the stack, | |
457 | and a page size can be as small as 4096 bytes. So we cannot safely | |
458 | allocate anything larger than 4096 bytes. Also care for the possibility | |
459 | of a few compiler-allocated temporary stack slots. */ | |
460 | # define __libc_use_alloca(n) ((n) < 4032) | |
461 | # else | |
462 | /* alloca is implemented with malloc, so just use malloc. */ | |
463 | # define __libc_use_alloca(n) 0 | |
464 | # undef alloca | |
465 | # define alloca(n) malloc (n) | |
466 | # endif | |
467 | #endif | |
468 | ||
469 | #ifndef MAX | |
470 | # define MAX(a,b) ((a) < (b) ? (b) : (a)) | |
471 | #endif | |
005de2e8 LC |
472 | #ifndef MIN |
473 | # define MIN(a,b) ((a) < (b) ? (a) : (b)) | |
474 | #endif | |
eb4a14ed LC |
475 | |
476 | #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) | |
477 | #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) | |
478 | #define re_free(p) free (p) | |
479 | ||
480 | struct bin_tree_t | |
481 | { | |
482 | struct bin_tree_t *parent; | |
483 | struct bin_tree_t *left; | |
484 | struct bin_tree_t *right; | |
485 | struct bin_tree_t *first; | |
486 | struct bin_tree_t *next; | |
487 | ||
488 | re_token_t token; | |
489 | ||
490 | /* 'node_idx' is the index in dfa->nodes, if 'type' == 0. | |
491 | Otherwise 'type' indicate the type of this node. */ | |
492 | Idx node_idx; | |
493 | }; | |
494 | typedef struct bin_tree_t bin_tree_t; | |
495 | ||
496 | #define BIN_TREE_STORAGE_SIZE \ | |
497 | ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) | |
498 | ||
499 | struct bin_tree_storage_t | |
500 | { | |
501 | struct bin_tree_storage_t *next; | |
502 | bin_tree_t data[BIN_TREE_STORAGE_SIZE]; | |
503 | }; | |
504 | typedef struct bin_tree_storage_t bin_tree_storage_t; | |
505 | ||
506 | #define CONTEXT_WORD 1 | |
507 | #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) | |
508 | #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) | |
509 | #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) | |
510 | ||
511 | #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) | |
512 | #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) | |
513 | #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) | |
514 | #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) | |
515 | #define IS_ORDINARY_CONTEXT(c) ((c) == 0) | |
516 | ||
517 | #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') | |
518 | #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) | |
519 | #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_') | |
520 | #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) | |
521 | ||
522 | #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ | |
523 | ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ | |
524 | || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ | |
525 | || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ | |
526 | || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) | |
527 | ||
528 | #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ | |
529 | ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ | |
530 | || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ | |
531 | || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ | |
532 | || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) | |
533 | ||
534 | struct re_dfastate_t | |
535 | { | |
536 | re_hashval_t hash; | |
537 | re_node_set nodes; | |
538 | re_node_set non_eps_nodes; | |
539 | re_node_set inveclosure; | |
540 | re_node_set *entrance_nodes; | |
541 | struct re_dfastate_t **trtable, **word_trtable; | |
542 | unsigned int context : 4; | |
543 | unsigned int halt : 1; | |
544 | /* If this state can accept "multi byte". | |
545 | Note that we refer to multibyte characters, and multi character | |
546 | collating elements as "multi byte". */ | |
547 | unsigned int accept_mb : 1; | |
548 | /* If this state has backreference node(s). */ | |
549 | unsigned int has_backref : 1; | |
550 | unsigned int has_constraint : 1; | |
551 | }; | |
552 | typedef struct re_dfastate_t re_dfastate_t; | |
553 | ||
554 | struct re_state_table_entry | |
555 | { | |
556 | Idx num; | |
557 | Idx alloc; | |
558 | re_dfastate_t **array; | |
559 | }; | |
560 | ||
561 | /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ | |
562 | ||
563 | typedef struct | |
564 | { | |
565 | Idx next_idx; | |
566 | Idx alloc; | |
567 | re_dfastate_t **array; | |
568 | } state_array_t; | |
569 | ||
570 | /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ | |
571 | ||
572 | typedef struct | |
573 | { | |
574 | Idx node; | |
575 | Idx str_idx; /* The position NODE match at. */ | |
576 | state_array_t path; | |
577 | } re_sub_match_last_t; | |
578 | ||
579 | /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. | |
580 | And information about the node, whose type is OP_CLOSE_SUBEXP, | |
581 | corresponding to NODE is stored in LASTS. */ | |
582 | ||
583 | typedef struct | |
584 | { | |
585 | Idx str_idx; | |
586 | Idx node; | |
587 | state_array_t *path; | |
588 | Idx alasts; /* Allocation size of LASTS. */ | |
589 | Idx nlasts; /* The number of LASTS. */ | |
590 | re_sub_match_last_t **lasts; | |
591 | } re_sub_match_top_t; | |
592 | ||
593 | struct re_backref_cache_entry | |
594 | { | |
595 | Idx node; | |
596 | Idx str_idx; | |
597 | Idx subexp_from; | |
598 | Idx subexp_to; | |
599 | char more; | |
600 | char unused; | |
601 | unsigned short int eps_reachable_subexps_map; | |
602 | }; | |
603 | ||
604 | typedef struct | |
605 | { | |
606 | /* The string object corresponding to the input string. */ | |
607 | re_string_t input; | |
608 | #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) | |
609 | const re_dfa_t *const dfa; | |
610 | #else | |
611 | const re_dfa_t *dfa; | |
612 | #endif | |
613 | /* EFLAGS of the argument of regexec. */ | |
614 | int eflags; | |
615 | /* Where the matching ends. */ | |
616 | Idx match_last; | |
617 | Idx last_node; | |
618 | /* The state log used by the matcher. */ | |
619 | re_dfastate_t **state_log; | |
620 | Idx state_log_top; | |
621 | /* Back reference cache. */ | |
622 | Idx nbkref_ents; | |
623 | Idx abkref_ents; | |
624 | struct re_backref_cache_entry *bkref_ents; | |
625 | int max_mb_elem_len; | |
626 | Idx nsub_tops; | |
627 | Idx asub_tops; | |
628 | re_sub_match_top_t **sub_tops; | |
629 | } re_match_context_t; | |
630 | ||
631 | typedef struct | |
632 | { | |
633 | re_dfastate_t **sifted_states; | |
634 | re_dfastate_t **limited_states; | |
635 | Idx last_node; | |
636 | Idx last_str_idx; | |
637 | re_node_set limits; | |
638 | } re_sift_context_t; | |
639 | ||
640 | struct re_fail_stack_ent_t | |
641 | { | |
642 | Idx idx; | |
643 | Idx node; | |
644 | regmatch_t *regs; | |
645 | re_node_set eps_via_nodes; | |
646 | }; | |
647 | ||
648 | struct re_fail_stack_t | |
649 | { | |
650 | Idx num; | |
651 | Idx alloc; | |
652 | struct re_fail_stack_ent_t *stack; | |
653 | }; | |
654 | ||
655 | struct re_dfa_t | |
656 | { | |
657 | re_token_t *nodes; | |
658 | size_t nodes_alloc; | |
659 | size_t nodes_len; | |
660 | Idx *nexts; | |
661 | Idx *org_indices; | |
662 | re_node_set *edests; | |
663 | re_node_set *eclosures; | |
664 | re_node_set *inveclosures; | |
665 | struct re_state_table_entry *state_table; | |
666 | re_dfastate_t *init_state; | |
667 | re_dfastate_t *init_state_word; | |
668 | re_dfastate_t *init_state_nl; | |
669 | re_dfastate_t *init_state_begbuf; | |
670 | bin_tree_t *str_tree; | |
671 | bin_tree_storage_t *str_tree_storage; | |
672 | re_bitset_ptr_t sb_char; | |
673 | int str_tree_storage_idx; | |
674 | ||
675 | /* number of subexpressions 're_nsub' is in regex_t. */ | |
676 | re_hashval_t state_hash_mask; | |
677 | Idx init_node; | |
678 | Idx nbackref; /* The number of backreference in this dfa. */ | |
679 | ||
680 | /* Bitmap expressing which backreference is used. */ | |
681 | bitset_word_t used_bkref_map; | |
682 | bitset_word_t completed_bkref_map; | |
683 | ||
684 | unsigned int has_plural_match : 1; | |
685 | /* If this dfa has "multibyte node", which is a backreference or | |
686 | a node which can accept multibyte character or multi character | |
687 | collating element. */ | |
688 | unsigned int has_mb_node : 1; | |
689 | unsigned int is_utf8 : 1; | |
690 | unsigned int map_notascii : 1; | |
691 | unsigned int word_ops_used : 1; | |
692 | int mb_cur_max; | |
693 | bitset_t word_char; | |
694 | reg_syntax_t syntax; | |
695 | Idx *subexp_map; | |
696 | #ifdef DEBUG | |
697 | char* re_str; | |
698 | #endif | |
eb4a14ed | 699 | __libc_lock_define (, lock) |
eb4a14ed LC |
700 | }; |
701 | ||
702 | #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) | |
703 | #define re_node_set_remove(set,id) \ | |
704 | (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) | |
705 | #define re_node_set_empty(p) ((p)->nelem = 0) | |
706 | #define re_node_set_free(set) re_free ((set)->elems) | |
707 | \f | |
708 | ||
709 | typedef enum | |
710 | { | |
711 | SB_CHAR, | |
712 | MB_CHAR, | |
713 | EQUIV_CLASS, | |
714 | COLL_SYM, | |
715 | CHAR_CLASS | |
716 | } bracket_elem_type; | |
717 | ||
718 | typedef struct | |
719 | { | |
720 | bracket_elem_type type; | |
721 | union | |
722 | { | |
723 | unsigned char ch; | |
724 | unsigned char *name; | |
725 | wchar_t wch; | |
726 | } opr; | |
727 | } bracket_elem_t; | |
728 | ||
729 | ||
730 | /* Inline functions for bitset_t operation. */ | |
731 | ||
732 | static inline void | |
733 | bitset_set (bitset_t set, Idx i) | |
734 | { | |
735 | set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; | |
736 | } | |
737 | ||
738 | static inline void | |
739 | bitset_clear (bitset_t set, Idx i) | |
740 | { | |
741 | set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); | |
742 | } | |
743 | ||
744 | static inline bool | |
745 | bitset_contain (const bitset_t set, Idx i) | |
746 | { | |
747 | return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; | |
748 | } | |
749 | ||
750 | static inline void | |
751 | bitset_empty (bitset_t set) | |
752 | { | |
753 | memset (set, '\0', sizeof (bitset_t)); | |
754 | } | |
755 | ||
756 | static inline void | |
757 | bitset_set_all (bitset_t set) | |
758 | { | |
759 | memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); | |
760 | if (SBC_MAX % BITSET_WORD_BITS != 0) | |
761 | set[BITSET_WORDS - 1] = | |
762 | ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; | |
763 | } | |
764 | ||
765 | static inline void | |
766 | bitset_copy (bitset_t dest, const bitset_t src) | |
767 | { | |
768 | memcpy (dest, src, sizeof (bitset_t)); | |
769 | } | |
770 | ||
771 | static inline void | |
772 | bitset_not (bitset_t set) | |
773 | { | |
774 | int bitset_i; | |
775 | for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) | |
776 | set[bitset_i] = ~set[bitset_i]; | |
777 | if (SBC_MAX % BITSET_WORD_BITS != 0) | |
778 | set[BITSET_WORDS - 1] = | |
779 | ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) | |
780 | & ~set[BITSET_WORDS - 1]); | |
781 | } | |
782 | ||
783 | static inline void | |
784 | bitset_merge (bitset_t dest, const bitset_t src) | |
785 | { | |
786 | int bitset_i; | |
787 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) | |
788 | dest[bitset_i] |= src[bitset_i]; | |
789 | } | |
790 | ||
791 | static inline void | |
792 | bitset_mask (bitset_t dest, const bitset_t src) | |
793 | { | |
794 | int bitset_i; | |
795 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) | |
796 | dest[bitset_i] &= src[bitset_i]; | |
797 | } | |
798 | ||
799 | #ifdef RE_ENABLE_I18N | |
800 | /* Inline functions for re_string. */ | |
801 | static inline int | |
802 | internal_function __attribute ((pure)) | |
803 | re_string_char_size_at (const re_string_t *pstr, Idx idx) | |
804 | { | |
805 | int byte_idx; | |
806 | if (pstr->mb_cur_max == 1) | |
807 | return 1; | |
808 | for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) | |
809 | if (pstr->wcs[idx + byte_idx] != WEOF) | |
810 | break; | |
811 | return byte_idx; | |
812 | } | |
813 | ||
814 | static inline wint_t | |
815 | internal_function __attribute ((pure)) | |
816 | re_string_wchar_at (const re_string_t *pstr, Idx idx) | |
817 | { | |
818 | if (pstr->mb_cur_max == 1) | |
819 | return (wint_t) pstr->mbs[idx]; | |
820 | return (wint_t) pstr->wcs[idx]; | |
821 | } | |
822 | ||
005de2e8 | 823 | # ifndef NOT_IN_libc |
eb4a14ed LC |
824 | static int |
825 | internal_function __attribute ((pure)) | |
826 | re_string_elem_size_at (const re_string_t *pstr, Idx idx) | |
827 | { | |
005de2e8 | 828 | # ifdef _LIBC |
eb4a14ed LC |
829 | const unsigned char *p, *extra; |
830 | const int32_t *table, *indirect; | |
005de2e8 | 831 | # include <locale/weight.h> |
eb4a14ed LC |
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; | |
005de2e8 | 842 | findidx (&p, pstr->len - idx); |
eb4a14ed LC |
843 | return p - pstr->mbs - idx; |
844 | } | |
845 | else | |
005de2e8 | 846 | # endif /* _LIBC */ |
eb4a14ed LC |
847 | return 1; |
848 | } | |
005de2e8 | 849 | # endif |
eb4a14ed LC |
850 | #endif /* RE_ENABLE_I18N */ |
851 | ||
852 | #ifndef __GNUC_PREREQ | |
853 | # if defined __GNUC__ && defined __GNUC_MINOR__ | |
854 | # define __GNUC_PREREQ(maj, min) \ | |
855 | ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min)) | |
856 | # else | |
857 | # define __GNUC_PREREQ(maj, min) 0 | |
858 | # endif | |
859 | #endif | |
860 | ||
861 | #if __GNUC_PREREQ (3,4) | |
862 | # undef __attribute_warn_unused_result__ | |
863 | # define __attribute_warn_unused_result__ \ | |
864 | __attribute__ ((__warn_unused_result__)) | |
865 | #else | |
866 | # define __attribute_warn_unused_result__ /* empty */ | |
867 | #endif | |
868 | ||
869 | #endif /* _REGEX_INTERNAL_H */ |