]>
Commit | Line | Data |
---|---|---|
d18f76dc ÆAB |
1 | /* Extended regular expression matching and search library. |
2 | Copyright (C) 2002-2005, 2007, 2008, 2010 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 | |
48425792 TZ |
17 | License along with the GNU C Library; if not, see |
18 | <http://www.gnu.org/licenses/>. */ | |
d18f76dc ÆAB |
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> | |
d18f76dc ÆAB |
26 | #include <string.h> |
27 | ||
28 | #if defined HAVE_LANGINFO_H || defined HAVE_LANGINFO_CODESET || defined _LIBC | |
29 | # include <langinfo.h> | |
30 | #endif | |
31 | #if defined HAVE_LOCALE_H || defined _LIBC | |
32 | # include <locale.h> | |
33 | #endif | |
34 | #if defined HAVE_WCHAR_H || defined _LIBC | |
35 | # include <wchar.h> | |
36 | #endif /* HAVE_WCHAR_H || _LIBC */ | |
37 | #if defined HAVE_WCTYPE_H || defined _LIBC | |
38 | # include <wctype.h> | |
39 | #endif /* HAVE_WCTYPE_H || _LIBC */ | |
40 | #if defined HAVE_STDBOOL_H || defined _LIBC | |
41 | # include <stdbool.h> | |
42 | #endif /* HAVE_STDBOOL_H || _LIBC */ | |
43 | #if !defined(ZOS_USS) | |
44 | #if defined HAVE_STDINT_H || defined _LIBC | |
45 | # include <stdint.h> | |
46 | #endif /* HAVE_STDINT_H || _LIBC */ | |
47 | #endif /* !ZOS_USS */ | |
48 | #if defined _LIBC | |
49 | # include <bits/libc-lock.h> | |
50 | #else | |
51 | # define __libc_lock_define(CLASS,NAME) | |
52 | # define __libc_lock_init(NAME) do { } while (0) | |
53 | # define __libc_lock_lock(NAME) do { } while (0) | |
54 | # define __libc_lock_unlock(NAME) do { } while (0) | |
55 | #endif | |
56 | ||
57 | #ifndef GAWK | |
58 | /* In case that the system doesn't have isblank(). */ | |
59 | #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank | |
60 | # define isblank(ch) ((ch) == ' ' || (ch) == '\t') | |
61 | #endif | |
62 | #else /* GAWK */ | |
63 | /* | |
64 | * This is a freaking mess. On glibc systems you have to define | |
65 | * a magic constant to get isblank() out of <ctype.h>, since it's | |
66 | * a C99 function. To heck with all that and borrow a page from | |
67 | * dfa.c's book. | |
68 | */ | |
69 | ||
70 | static int | |
71 | is_blank (int c) | |
72 | { | |
73 | return (c == ' ' || c == '\t'); | |
74 | } | |
75 | #endif /* GAWK */ | |
76 | ||
77 | #ifdef _LIBC | |
78 | # ifndef _RE_DEFINE_LOCALE_FUNCTIONS | |
79 | # define _RE_DEFINE_LOCALE_FUNCTIONS 1 | |
80 | # include <locale/localeinfo.h> | |
81 | # include <locale/elem-hash.h> | |
82 | # include <locale/coll-lookup.h> | |
83 | # endif | |
84 | #endif | |
85 | ||
86 | /* This is for other GNU distributions with internationalized messages. */ | |
87 | #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC | |
88 | # include <libintl.h> | |
89 | # ifdef _LIBC | |
90 | # undef gettext | |
91 | # define gettext(msgid) \ | |
92 | INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES) | |
93 | # endif | |
94 | #else | |
95 | # define gettext(msgid) (msgid) | |
96 | #endif | |
97 | ||
98 | #ifndef gettext_noop | |
99 | /* This define is so xgettext can find the internationalizable | |
100 | strings. */ | |
101 | # define gettext_noop(String) String | |
102 | #endif | |
103 | ||
104 | /* For loser systems without the definition. */ | |
105 | #ifndef SIZE_MAX | |
106 | # define SIZE_MAX ((size_t) -1) | |
107 | #endif | |
108 | ||
109 | #ifndef NO_MBSUPPORT | |
110 | #include "mbsupport.h" /* gawk */ | |
111 | #endif | |
112 | #ifndef MB_CUR_MAX | |
113 | #define MB_CUR_MAX 1 | |
114 | #endif | |
115 | ||
116 | #if (defined MBS_SUPPORT) || _LIBC | |
117 | # define RE_ENABLE_I18N | |
118 | #endif | |
119 | ||
120 | #if __GNUC__ >= 3 | |
121 | # define BE(expr, val) __builtin_expect (expr, val) | |
122 | #else | |
123 | # define BE(expr, val) (expr) | |
124 | # ifdef inline | |
125 | # undef inline | |
126 | # endif | |
127 | # define inline | |
128 | #endif | |
129 | ||
130 | /* Number of single byte character. */ | |
131 | #define SBC_MAX 256 | |
132 | ||
133 | #define COLL_ELEM_LEN_MAX 8 | |
134 | ||
135 | /* The character which represents newline. */ | |
136 | #define NEWLINE_CHAR '\n' | |
137 | #define WIDE_NEWLINE_CHAR L'\n' | |
138 | ||
139 | /* Rename to standard API for using out of glibc. */ | |
140 | #ifndef _LIBC | |
141 | # ifdef __wctype | |
142 | # undef __wctype | |
143 | # endif | |
144 | # define __wctype wctype | |
145 | # ifdef __iswctype | |
146 | # undef __iswctype | |
147 | # endif | |
148 | # define __iswctype iswctype | |
149 | # define __btowc btowc | |
150 | # define __mbrtowc mbrtowc | |
151 | #undef __mempcpy /* GAWK */ | |
152 | # define __mempcpy mempcpy | |
153 | # define __wcrtomb wcrtomb | |
154 | # define __regfree regfree | |
155 | # define attribute_hidden | |
156 | #endif /* not _LIBC */ | |
157 | ||
158 | #ifdef __GNUC__ | |
159 | # define __attribute(arg) __attribute__ (arg) | |
160 | #else | |
161 | # define __attribute(arg) | |
162 | #endif | |
163 | ||
164 | extern const char __re_error_msgid[] attribute_hidden; | |
165 | extern const size_t __re_error_msgid_idx[] attribute_hidden; | |
166 | ||
167 | /* An integer used to represent a set of bits. It must be unsigned, | |
168 | and must be at least as wide as unsigned int. */ | |
169 | typedef unsigned long int bitset_word_t; | |
170 | /* All bits set in a bitset_word_t. */ | |
171 | #define BITSET_WORD_MAX ULONG_MAX | |
172 | /* Number of bits in a bitset_word_t. */ | |
173 | #define BITSET_WORD_BITS (sizeof (bitset_word_t) * CHAR_BIT) | |
174 | /* Number of bitset_word_t in a bit_set. */ | |
175 | #define BITSET_WORDS (SBC_MAX / BITSET_WORD_BITS) | |
176 | typedef bitset_word_t bitset_t[BITSET_WORDS]; | |
177 | typedef bitset_word_t *re_bitset_ptr_t; | |
178 | typedef const bitset_word_t *re_const_bitset_ptr_t; | |
179 | ||
180 | #define bitset_set(set,i) \ | |
181 | (set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS) | |
182 | #define bitset_clear(set,i) \ | |
183 | (set[i / BITSET_WORD_BITS] &= ~((bitset_word_t) 1 << i % BITSET_WORD_BITS)) | |
184 | #define bitset_contain(set,i) \ | |
185 | (set[i / BITSET_WORD_BITS] & ((bitset_word_t) 1 << i % BITSET_WORD_BITS)) | |
186 | #define bitset_empty(set) memset (set, '\0', sizeof (bitset_t)) | |
187 | #define bitset_set_all(set) memset (set, '\xff', sizeof (bitset_t)) | |
188 | #define bitset_copy(dest,src) memcpy (dest, src, sizeof (bitset_t)) | |
189 | ||
190 | #define PREV_WORD_CONSTRAINT 0x0001 | |
191 | #define PREV_NOTWORD_CONSTRAINT 0x0002 | |
192 | #define NEXT_WORD_CONSTRAINT 0x0004 | |
193 | #define NEXT_NOTWORD_CONSTRAINT 0x0008 | |
194 | #define PREV_NEWLINE_CONSTRAINT 0x0010 | |
195 | #define NEXT_NEWLINE_CONSTRAINT 0x0020 | |
196 | #define PREV_BEGBUF_CONSTRAINT 0x0040 | |
197 | #define NEXT_ENDBUF_CONSTRAINT 0x0080 | |
198 | #define WORD_DELIM_CONSTRAINT 0x0100 | |
199 | #define NOT_WORD_DELIM_CONSTRAINT 0x0200 | |
200 | ||
201 | typedef enum | |
202 | { | |
203 | INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, | |
204 | WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, | |
205 | WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, | |
206 | INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, | |
207 | LINE_FIRST = PREV_NEWLINE_CONSTRAINT, | |
208 | LINE_LAST = NEXT_NEWLINE_CONSTRAINT, | |
209 | BUF_FIRST = PREV_BEGBUF_CONSTRAINT, | |
210 | BUF_LAST = NEXT_ENDBUF_CONSTRAINT, | |
211 | WORD_DELIM = WORD_DELIM_CONSTRAINT, | |
212 | NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT | |
213 | } re_context_type; | |
214 | ||
215 | typedef struct | |
216 | { | |
217 | int alloc; | |
218 | int nelem; | |
219 | int *elems; | |
220 | } re_node_set; | |
221 | ||
222 | typedef enum | |
223 | { | |
224 | NON_TYPE = 0, | |
225 | ||
226 | /* Node type, These are used by token, node, tree. */ | |
227 | CHARACTER = 1, | |
228 | END_OF_RE = 2, | |
229 | SIMPLE_BRACKET = 3, | |
230 | OP_BACK_REF = 4, | |
231 | OP_PERIOD = 5, | |
232 | #ifdef RE_ENABLE_I18N | |
233 | COMPLEX_BRACKET = 6, | |
234 | OP_UTF8_PERIOD = 7, | |
235 | #endif /* RE_ENABLE_I18N */ | |
236 | ||
237 | /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used | |
238 | when the debugger shows values of this enum type. */ | |
239 | #define EPSILON_BIT 8 | |
240 | OP_OPEN_SUBEXP = EPSILON_BIT | 0, | |
241 | OP_CLOSE_SUBEXP = EPSILON_BIT | 1, | |
242 | OP_ALT = EPSILON_BIT | 2, | |
243 | OP_DUP_ASTERISK = EPSILON_BIT | 3, | |
244 | ANCHOR = EPSILON_BIT | 4, | |
245 | ||
246 | /* Tree type, these are used only by tree. */ | |
247 | CONCAT = 16, | |
248 | SUBEXP = 17, | |
249 | ||
250 | /* Token type, these are used only by token. */ | |
251 | OP_DUP_PLUS = 18, | |
252 | OP_DUP_QUESTION, | |
253 | OP_OPEN_BRACKET, | |
254 | OP_CLOSE_BRACKET, | |
255 | OP_CHARSET_RANGE, | |
256 | OP_OPEN_DUP_NUM, | |
257 | OP_CLOSE_DUP_NUM, | |
258 | OP_NON_MATCH_LIST, | |
259 | OP_OPEN_COLL_ELEM, | |
260 | OP_CLOSE_COLL_ELEM, | |
261 | OP_OPEN_EQUIV_CLASS, | |
262 | OP_CLOSE_EQUIV_CLASS, | |
263 | OP_OPEN_CHAR_CLASS, | |
264 | OP_CLOSE_CHAR_CLASS, | |
265 | OP_WORD, | |
266 | OP_NOTWORD, | |
267 | OP_SPACE, | |
268 | OP_NOTSPACE, | |
269 | BACK_SLASH | |
270 | ||
271 | } re_token_type_t; | |
272 | ||
273 | #ifdef RE_ENABLE_I18N | |
274 | typedef struct | |
275 | { | |
276 | /* Multibyte characters. */ | |
277 | wchar_t *mbchars; | |
278 | ||
279 | /* Collating symbols. */ | |
280 | # ifdef _LIBC | |
281 | int32_t *coll_syms; | |
282 | # endif | |
283 | ||
284 | /* Equivalence classes. */ | |
285 | # ifdef _LIBC | |
286 | int32_t *equiv_classes; | |
287 | # endif | |
288 | ||
289 | /* Range expressions. */ | |
290 | # ifdef _LIBC | |
291 | uint32_t *range_starts; | |
292 | uint32_t *range_ends; | |
293 | # else /* not _LIBC */ | |
294 | wchar_t *range_starts; | |
295 | wchar_t *range_ends; | |
296 | # endif /* not _LIBC */ | |
297 | ||
298 | /* Character classes. */ | |
299 | wctype_t *char_classes; | |
300 | ||
301 | /* If this character set is the non-matching list. */ | |
302 | unsigned int non_match : 1; | |
303 | ||
304 | /* # of multibyte characters. */ | |
305 | int nmbchars; | |
306 | ||
307 | /* # of collating symbols. */ | |
308 | int ncoll_syms; | |
309 | ||
310 | /* # of equivalence classes. */ | |
311 | int nequiv_classes; | |
312 | ||
313 | /* # of range expressions. */ | |
314 | int nranges; | |
315 | ||
316 | /* # of character classes. */ | |
317 | int nchar_classes; | |
318 | } re_charset_t; | |
319 | #endif /* RE_ENABLE_I18N */ | |
320 | ||
321 | typedef struct | |
322 | { | |
323 | union | |
324 | { | |
325 | unsigned char c; /* for CHARACTER */ | |
326 | re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ | |
327 | #ifdef RE_ENABLE_I18N | |
328 | re_charset_t *mbcset; /* for COMPLEX_BRACKET */ | |
329 | #endif /* RE_ENABLE_I18N */ | |
330 | int idx; /* for BACK_REF */ | |
331 | re_context_type ctx_type; /* for ANCHOR */ | |
332 | } opr; | |
333 | #if __GNUC__ >= 2 | |
334 | re_token_type_t type : 8; | |
335 | #else | |
336 | re_token_type_t type; | |
337 | #endif | |
338 | unsigned int constraint : 10; /* context constraint */ | |
339 | unsigned int duplicated : 1; | |
340 | unsigned int opt_subexp : 1; | |
341 | #ifdef RE_ENABLE_I18N | |
342 | unsigned int accept_mb : 1; | |
343 | /* These 2 bits can be moved into the union if needed (e.g. if running out | |
344 | of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ | |
345 | unsigned int mb_partial : 1; | |
346 | #endif | |
347 | unsigned int word_char : 1; | |
348 | } re_token_t; | |
349 | ||
350 | #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) | |
351 | ||
352 | struct re_string_t | |
353 | { | |
354 | /* Indicate the raw buffer which is the original string passed as an | |
355 | argument of regexec(), re_search(), etc.. */ | |
356 | const unsigned char *raw_mbs; | |
357 | /* Store the multibyte string. In case of "case insensitive mode" like | |
358 | REG_ICASE, upper cases of the string are stored, otherwise MBS points | |
359 | the same address that RAW_MBS points. */ | |
360 | unsigned char *mbs; | |
361 | #ifdef RE_ENABLE_I18N | |
362 | /* Store the wide character string which is corresponding to MBS. */ | |
363 | wint_t *wcs; | |
364 | int *offsets; | |
365 | mbstate_t cur_state; | |
366 | #endif | |
367 | /* Index in RAW_MBS. Each character mbs[i] corresponds to | |
368 | raw_mbs[raw_mbs_idx + i]. */ | |
369 | int raw_mbs_idx; | |
370 | /* The length of the valid characters in the buffers. */ | |
371 | int valid_len; | |
372 | /* The corresponding number of bytes in raw_mbs array. */ | |
373 | int valid_raw_len; | |
374 | /* The length of the buffers MBS and WCS. */ | |
375 | int bufs_len; | |
376 | /* The index in MBS, which is updated by re_string_fetch_byte. */ | |
377 | int cur_idx; | |
378 | /* length of RAW_MBS array. */ | |
379 | int raw_len; | |
380 | /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ | |
381 | int len; | |
382 | /* End of the buffer may be shorter than its length in the cases such | |
383 | as re_match_2, re_search_2. Then, we use STOP for end of the buffer | |
384 | instead of LEN. */ | |
385 | int raw_stop; | |
386 | /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ | |
387 | int stop; | |
388 | ||
389 | /* The context of mbs[0]. We store the context independently, since | |
390 | the context of mbs[0] may be different from raw_mbs[0], which is | |
391 | the beginning of the input string. */ | |
392 | unsigned int tip_context; | |
393 | /* The translation passed as a part of an argument of re_compile_pattern. */ | |
394 | RE_TRANSLATE_TYPE trans; | |
395 | /* Copy of re_dfa_t's word_char. */ | |
396 | re_const_bitset_ptr_t word_char; | |
397 | /* 1 if REG_ICASE. */ | |
398 | unsigned char icase; | |
399 | unsigned char is_utf8; | |
400 | unsigned char map_notascii; | |
401 | unsigned char mbs_allocated; | |
402 | unsigned char offsets_needed; | |
403 | unsigned char newline_anchor; | |
404 | unsigned char word_ops_used; | |
405 | int mb_cur_max; | |
406 | }; | |
407 | typedef struct re_string_t re_string_t; | |
408 | ||
409 | ||
410 | struct re_dfa_t; | |
411 | typedef struct re_dfa_t re_dfa_t; | |
412 | ||
413 | #ifndef _LIBC | |
414 | # ifdef __i386__ | |
415 | # define internal_function __attribute ((regparm (3), stdcall)) | |
416 | # else | |
417 | # define internal_function | |
418 | # endif | |
419 | #endif | |
420 | ||
421 | #ifndef NOT_IN_libc | |
422 | static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, | |
423 | int new_buf_len) | |
424 | internal_function; | |
425 | # ifdef RE_ENABLE_I18N | |
426 | static void build_wcs_buffer (re_string_t *pstr) internal_function; | |
427 | static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr) | |
428 | internal_function; | |
429 | # endif /* RE_ENABLE_I18N */ | |
430 | static void build_upper_buffer (re_string_t *pstr) internal_function; | |
431 | static void re_string_translate_buffer (re_string_t *pstr) internal_function; | |
432 | static unsigned int re_string_context_at (const re_string_t *input, int idx, | |
433 | int eflags) | |
434 | internal_function __attribute ((pure)); | |
435 | #endif | |
436 | #define re_string_peek_byte(pstr, offset) \ | |
437 | ((pstr)->mbs[(pstr)->cur_idx + offset]) | |
438 | #define re_string_fetch_byte(pstr) \ | |
439 | ((pstr)->mbs[(pstr)->cur_idx++]) | |
440 | #define re_string_first_byte(pstr, idx) \ | |
441 | ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) | |
442 | #define re_string_is_single_byte_char(pstr, idx) \ | |
443 | ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ | |
444 | || (pstr)->wcs[(idx) + 1] != WEOF)) | |
445 | #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) | |
446 | #define re_string_cur_idx(pstr) ((pstr)->cur_idx) | |
447 | #define re_string_get_buffer(pstr) ((pstr)->mbs) | |
448 | #define re_string_length(pstr) ((pstr)->len) | |
449 | #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) | |
450 | #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) | |
451 | #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) | |
452 | ||
453 | #ifndef _LIBC | |
454 | # if HAVE_ALLOCA | |
455 | # if (_MSC_VER) | |
456 | # include <malloc.h> | |
457 | # define __libc_use_alloca(n) 0 | |
458 | # else | |
459 | # include <alloca.h> | |
460 | /* The OS usually guarantees only one guard page at the bottom of the stack, | |
461 | and a page size can be as small as 4096 bytes. So we cannot safely | |
462 | allocate anything larger than 4096 bytes. Also care for the possibility | |
463 | of a few compiler-allocated temporary stack slots. */ | |
464 | # define __libc_use_alloca(n) ((n) < 4032) | |
465 | # endif | |
466 | # else | |
467 | /* alloca is implemented with malloc, so just use malloc. */ | |
468 | # define __libc_use_alloca(n) 0 | |
469 | # endif | |
470 | #endif | |
471 | ||
472 | #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) | |
473 | /* SunOS 4.1.x realloc doesn't accept null pointers: pre-Standard C. Sigh. */ | |
474 | #define re_realloc(p,t,n) ((p != NULL) ? (t *) realloc (p,(n)*sizeof(t)) : (t *) calloc(n,sizeof(t))) | |
475 | #define re_free(p) free (p) | |
476 | ||
477 | struct bin_tree_t | |
478 | { | |
479 | struct bin_tree_t *parent; | |
480 | struct bin_tree_t *left; | |
481 | struct bin_tree_t *right; | |
482 | struct bin_tree_t *first; | |
483 | struct bin_tree_t *next; | |
484 | ||
485 | re_token_t token; | |
486 | ||
487 | /* `node_idx' is the index in dfa->nodes, if `type' == 0. | |
488 | Otherwise `type' indicate the type of this node. */ | |
489 | int node_idx; | |
490 | }; | |
491 | typedef struct bin_tree_t bin_tree_t; | |
492 | ||
493 | #define BIN_TREE_STORAGE_SIZE \ | |
494 | ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) | |
495 | ||
496 | struct bin_tree_storage_t | |
497 | { | |
498 | struct bin_tree_storage_t *next; | |
499 | bin_tree_t data[BIN_TREE_STORAGE_SIZE]; | |
500 | }; | |
501 | typedef struct bin_tree_storage_t bin_tree_storage_t; | |
502 | ||
503 | #define CONTEXT_WORD 1 | |
504 | #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) | |
505 | #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) | |
506 | #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) | |
507 | ||
508 | #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) | |
509 | #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) | |
510 | #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) | |
511 | #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) | |
512 | #define IS_ORDINARY_CONTEXT(c) ((c) == 0) | |
513 | ||
514 | #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') | |
515 | #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) | |
516 | #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_') | |
517 | #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) | |
518 | ||
519 | #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ | |
520 | ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ | |
521 | || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ | |
522 | || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ | |
523 | || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) | |
524 | ||
525 | #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ | |
526 | ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ | |
527 | || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ | |
528 | || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ | |
529 | || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) | |
530 | ||
531 | struct re_dfastate_t | |
532 | { | |
533 | unsigned int hash; | |
534 | re_node_set nodes; | |
535 | re_node_set non_eps_nodes; | |
536 | re_node_set inveclosure; | |
537 | re_node_set *entrance_nodes; | |
538 | struct re_dfastate_t **trtable, **word_trtable; | |
539 | unsigned int context : 4; | |
540 | unsigned int halt : 1; | |
541 | /* If this state can accept `multi byte'. | |
542 | Note that we refer to multibyte characters, and multi character | |
543 | collating elements as `multi byte'. */ | |
544 | unsigned int accept_mb : 1; | |
545 | /* If this state has backreference node(s). */ | |
546 | unsigned int has_backref : 1; | |
547 | unsigned int has_constraint : 1; | |
548 | }; | |
549 | typedef struct re_dfastate_t re_dfastate_t; | |
550 | ||
551 | struct re_state_table_entry | |
552 | { | |
553 | int num; | |
554 | int alloc; | |
555 | re_dfastate_t **array; | |
556 | }; | |
557 | ||
558 | /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ | |
559 | ||
560 | typedef struct | |
561 | { | |
562 | int next_idx; | |
563 | int alloc; | |
564 | re_dfastate_t **array; | |
565 | } state_array_t; | |
566 | ||
567 | /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ | |
568 | ||
569 | typedef struct | |
570 | { | |
571 | int node; | |
572 | int str_idx; /* The position NODE match at. */ | |
573 | state_array_t path; | |
574 | } re_sub_match_last_t; | |
575 | ||
576 | /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. | |
577 | And information about the node, whose type is OP_CLOSE_SUBEXP, | |
578 | corresponding to NODE is stored in LASTS. */ | |
579 | ||
580 | typedef struct | |
581 | { | |
582 | int str_idx; | |
583 | int node; | |
584 | state_array_t *path; | |
585 | int alasts; /* Allocation size of LASTS. */ | |
586 | int nlasts; /* The number of LASTS. */ | |
587 | re_sub_match_last_t **lasts; | |
588 | } re_sub_match_top_t; | |
589 | ||
590 | struct re_backref_cache_entry | |
591 | { | |
592 | int node; | |
593 | int str_idx; | |
594 | int subexp_from; | |
595 | int subexp_to; | |
596 | char more; | |
597 | char unused; | |
598 | unsigned short int eps_reachable_subexps_map; | |
599 | }; | |
600 | ||
601 | typedef struct | |
602 | { | |
603 | /* The string object corresponding to the input string. */ | |
604 | re_string_t input; | |
605 | #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) | |
606 | const re_dfa_t *const dfa; | |
607 | #else | |
608 | const re_dfa_t *dfa; | |
609 | #endif | |
610 | /* EFLAGS of the argument of regexec. */ | |
611 | int eflags; | |
612 | /* Where the matching ends. */ | |
613 | int match_last; | |
614 | int last_node; | |
615 | /* The state log used by the matcher. */ | |
616 | re_dfastate_t **state_log; | |
617 | int state_log_top; | |
618 | /* Back reference cache. */ | |
619 | int nbkref_ents; | |
620 | int abkref_ents; | |
621 | struct re_backref_cache_entry *bkref_ents; | |
622 | int max_mb_elem_len; | |
623 | int nsub_tops; | |
624 | int asub_tops; | |
625 | re_sub_match_top_t **sub_tops; | |
626 | } re_match_context_t; | |
627 | ||
628 | typedef struct | |
629 | { | |
630 | re_dfastate_t **sifted_states; | |
631 | re_dfastate_t **limited_states; | |
632 | int last_node; | |
633 | int last_str_idx; | |
634 | re_node_set limits; | |
635 | } re_sift_context_t; | |
636 | ||
637 | struct re_fail_stack_ent_t | |
638 | { | |
639 | int idx; | |
640 | int node; | |
641 | regmatch_t *regs; | |
642 | re_node_set eps_via_nodes; | |
643 | }; | |
644 | ||
645 | struct re_fail_stack_t | |
646 | { | |
647 | int num; | |
648 | int alloc; | |
649 | struct re_fail_stack_ent_t *stack; | |
650 | }; | |
651 | ||
652 | struct re_dfa_t | |
653 | { | |
654 | re_token_t *nodes; | |
655 | size_t nodes_alloc; | |
656 | size_t nodes_len; | |
657 | int *nexts; | |
658 | int *org_indices; | |
659 | re_node_set *edests; | |
660 | re_node_set *eclosures; | |
661 | re_node_set *inveclosures; | |
662 | struct re_state_table_entry *state_table; | |
663 | re_dfastate_t *init_state; | |
664 | re_dfastate_t *init_state_word; | |
665 | re_dfastate_t *init_state_nl; | |
666 | re_dfastate_t *init_state_begbuf; | |
667 | bin_tree_t *str_tree; | |
668 | bin_tree_storage_t *str_tree_storage; | |
669 | re_bitset_ptr_t sb_char; | |
670 | int str_tree_storage_idx; | |
671 | ||
672 | /* number of subexpressions `re_nsub' is in regex_t. */ | |
673 | unsigned int state_hash_mask; | |
674 | int init_node; | |
675 | int nbackref; /* The number of backreference in this dfa. */ | |
676 | ||
677 | /* Bitmap expressing which backreference is used. */ | |
678 | bitset_word_t used_bkref_map; | |
679 | bitset_word_t completed_bkref_map; | |
680 | ||
681 | unsigned int has_plural_match : 1; | |
682 | /* If this dfa has "multibyte node", which is a backreference or | |
683 | a node which can accept multibyte character or multi character | |
684 | collating element. */ | |
685 | unsigned int has_mb_node : 1; | |
686 | unsigned int is_utf8 : 1; | |
687 | unsigned int map_notascii : 1; | |
688 | unsigned int word_ops_used : 1; | |
689 | int mb_cur_max; | |
690 | bitset_t word_char; | |
691 | reg_syntax_t syntax; | |
692 | int *subexp_map; | |
693 | #ifdef DEBUG | |
694 | char* re_str; | |
695 | #endif | |
696 | #if defined _LIBC | |
697 | __libc_lock_define (, lock) | |
698 | #endif | |
699 | }; | |
700 | ||
701 | #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) | |
702 | #define re_node_set_remove(set,id) \ | |
703 | (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) | |
704 | #define re_node_set_empty(p) ((p)->nelem = 0) | |
705 | #define re_node_set_free(set) re_free ((set)->elems) | |
706 | \f | |
707 | ||
708 | typedef enum | |
709 | { | |
710 | SB_CHAR, | |
711 | MB_CHAR, | |
712 | EQUIV_CLASS, | |
713 | COLL_SYM, | |
714 | CHAR_CLASS | |
715 | } bracket_elem_type; | |
716 | ||
717 | typedef struct | |
718 | { | |
719 | bracket_elem_type type; | |
720 | union | |
721 | { | |
722 | unsigned char ch; | |
723 | unsigned char *name; | |
724 | wchar_t wch; | |
725 | } opr; | |
726 | } bracket_elem_t; | |
727 | ||
728 | ||
729 | /* Inline functions for bitset operation. */ | |
730 | static inline void | |
731 | bitset_not (bitset_t set) | |
732 | { | |
733 | int bitset_i; | |
734 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) | |
735 | set[bitset_i] = ~set[bitset_i]; | |
736 | } | |
737 | ||
738 | static inline void | |
739 | bitset_merge (bitset_t dest, const bitset_t src) | |
740 | { | |
741 | int bitset_i; | |
742 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) | |
743 | dest[bitset_i] |= src[bitset_i]; | |
744 | } | |
745 | ||
746 | static inline void | |
747 | bitset_mask (bitset_t dest, const bitset_t src) | |
748 | { | |
749 | int bitset_i; | |
750 | for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) | |
751 | dest[bitset_i] &= src[bitset_i]; | |
752 | } | |
753 | ||
754 | #ifdef RE_ENABLE_I18N | |
755 | /* Inline functions for re_string. */ | |
756 | static inline int | |
757 | internal_function __attribute ((pure)) | |
758 | re_string_char_size_at (const re_string_t *pstr, int idx) | |
759 | { | |
760 | int byte_idx; | |
761 | if (pstr->mb_cur_max == 1) | |
762 | return 1; | |
763 | for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) | |
764 | if (pstr->wcs[idx + byte_idx] != WEOF) | |
765 | break; | |
766 | return byte_idx; | |
767 | } | |
768 | ||
769 | static inline wint_t | |
770 | internal_function __attribute ((pure)) | |
771 | re_string_wchar_at (const re_string_t *pstr, int idx) | |
772 | { | |
773 | if (pstr->mb_cur_max == 1) | |
774 | return (wint_t) pstr->mbs[idx]; | |
775 | return (wint_t) pstr->wcs[idx]; | |
776 | } | |
777 | ||
778 | # ifndef NOT_IN_libc | |
779 | static int | |
780 | internal_function __attribute ((pure)) | |
781 | re_string_elem_size_at (const re_string_t *pstr, int idx) | |
782 | { | |
783 | # ifdef _LIBC | |
784 | const unsigned char *p, *extra; | |
785 | const int32_t *table, *indirect; | |
786 | int32_t tmp; | |
787 | # include <locale/weight.h> | |
788 | uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); | |
789 | ||
790 | if (nrules != 0) | |
791 | { | |
792 | table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); | |
793 | extra = (const unsigned char *) | |
794 | _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); | |
795 | indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, | |
796 | _NL_COLLATE_INDIRECTMB); | |
797 | p = pstr->mbs + idx; | |
798 | tmp = findidx (&p); | |
799 | return p - pstr->mbs - idx; | |
800 | } | |
801 | else | |
802 | # endif /* _LIBC */ | |
803 | return 1; | |
804 | } | |
805 | # endif | |
806 | #endif /* RE_ENABLE_I18N */ | |
807 | ||
808 | #endif /* _REGEX_INTERNAL_H */ |