| 1 |
/* Extended regular expression matching and search library. |
| 2 |
Copyright (C) 2002-2024 Free Software Foundation, Inc. |
| 3 |
This file is part of the GNU C Library. |
| 4 |
Contributed by Isamu Hasegawa <[email protected]>. |
| 5 |
|
| 6 |
The GNU C Library is free software; you can redistribute it and/or |
| 7 |
modify it under the terms of the GNU Lesser General Public |
| 8 |
License as published by the Free Software Foundation; either |
| 9 |
version 2.1 of the License, or (at your option) any later version. |
| 10 |
|
| 11 |
The GNU C Library is distributed in the hope that it will be useful, |
| 12 |
but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 |
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 14 |
Lesser General Public License for more details. |
| 15 |
|
| 16 |
You should have received a copy of the GNU Lesser General Public |
| 17 |
License along with the GNU C Library; if not, see |
| 18 |
<https://www.gnu.org/licenses/>. */ |
| 19 |
|
| 20 |
#ifndef _REGEX_INTERNAL_H |
| 21 |
#define _REGEX_INTERNAL_H 1 |
| 22 |
|
| 23 |
#include <ctype.h> |
| 24 |
#include <stdio.h> |
| 25 |
#include <stdlib.h> |
| 26 |
#include <string.h> |
| 27 |
|
| 28 |
#include <langinfo.h> |
| 29 |
#include <locale.h> |
| 30 |
#include <wchar.h> |
| 31 |
#include <wctype.h> |
| 32 |
#include <stdckdint.h> |
| 33 |
#include <stdint.h> |
| 34 |
|
| 35 |
#ifndef _LIBC |
| 36 |
# include <dynarray.h> |
| 37 |
#endif |
| 38 |
|
| 39 |
#include <intprops.h> |
| 40 |
#include <verify.h> |
| 41 |
|
| 42 |
#if defined DEBUG && DEBUG != 0 |
| 43 |
# include <assert.h> |
| 44 |
# define DEBUG_ASSERT(x) assert (x) |
| 45 |
#else |
| 46 |
# define DEBUG_ASSERT(x) assume (x) |
| 47 |
#endif |
| 48 |
|
| 49 |
#ifdef _LIBC |
| 50 |
# include <libc-lock.h> |
| 51 |
# define lock_define(name) __libc_lock_define (, name) |
| 52 |
# define lock_init(lock) (__libc_lock_init (lock), 0) |
| 53 |
# define lock_fini(lock) ((void) 0) |
| 54 |
# define lock_lock(lock) __libc_lock_lock (lock) |
| 55 |
# define lock_unlock(lock) __libc_lock_unlock (lock) |
| 56 |
#elif defined GNULIB_LOCK && !defined GNULIB_REGEX_SINGLE_THREAD |
| 57 |
# include "glthread/lock.h" |
| 58 |
# define lock_define(name) gl_lock_define (, name) |
| 59 |
# define lock_init(lock) glthread_lock_init (&(lock)) |
| 60 |
# define lock_fini(lock) glthread_lock_destroy (&(lock)) |
| 61 |
# define lock_lock(lock) glthread_lock_lock (&(lock)) |
| 62 |
# define lock_unlock(lock) glthread_lock_unlock (&(lock)) |
| 63 |
#elif defined GNULIB_PTHREAD && !defined GNULIB_REGEX_SINGLE_THREAD |
| 64 |
# include <pthread.h> |
| 65 |
# define lock_define(name) pthread_mutex_t name; |
| 66 |
# define lock_init(lock) pthread_mutex_init (&(lock), 0) |
| 67 |
# define lock_fini(lock) pthread_mutex_destroy (&(lock)) |
| 68 |
# define lock_lock(lock) pthread_mutex_lock (&(lock)) |
| 69 |
# define lock_unlock(lock) pthread_mutex_unlock (&(lock)) |
| 70 |
#else |
| 71 |
# define lock_define(name) |
| 72 |
# define lock_init(lock) 0 |
| 73 |
# define lock_fini(lock) ((void) 0) |
| 74 |
/* The 'dfa' avoids an "unused variable 'dfa'" warning from GCC. */ |
| 75 |
# define lock_lock(lock) ((void) dfa) |
| 76 |
# define lock_unlock(lock) ((void) 0) |
| 77 |
#endif |
| 78 |
|
| 79 |
/* In case that the system doesn't have isblank(). */ |
| 80 |
#if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) |
| 81 |
# define isblank(ch) ((ch) == ' ' || (ch) == '\t') |
| 82 |
#endif |
| 83 |
|
| 84 |
/* regex code assumes isascii has its usual numeric meaning, |
| 85 |
even if the portable character set uses EBCDIC encoding, |
| 86 |
and even if wint_t is wider than int. */ |
| 87 |
#ifndef _LIBC |
| 88 |
# undef isascii |
| 89 |
# define isascii(c) (((c) & ~0x7f) == 0) |
| 90 |
#endif |
| 91 |
|
| 92 |
#ifdef _LIBC |
| 93 |
# ifndef _RE_DEFINE_LOCALE_FUNCTIONS |
| 94 |
# define _RE_DEFINE_LOCALE_FUNCTIONS 1 |
| 95 |
# include <locale/localeinfo.h> |
| 96 |
# include <locale/coll-lookup.h> |
| 97 |
# endif |
| 98 |
#endif |
| 99 |
|
| 100 |
/* This is for other GNU distributions with internationalized messages. */ |
| 101 |
#if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC |
| 102 |
# include <libintl.h> |
| 103 |
# ifdef _LIBC |
| 104 |
# undef gettext |
| 105 |
# define gettext(msgid) \ |
| 106 |
__dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES) |
| 107 |
# endif |
| 108 |
#else |
| 109 |
# undef gettext |
| 110 |
# define gettext(msgid) (msgid) |
| 111 |
#endif |
| 112 |
|
| 113 |
#ifndef gettext_noop |
| 114 |
/* This define is so xgettext can find the internationalizable |
| 115 |
strings. */ |
| 116 |
# define gettext_noop(String) String |
| 117 |
#endif |
| 118 |
|
| 119 |
/* Number of ASCII characters. */ |
| 120 |
#define ASCII_CHARS 0x80 |
| 121 |
|
| 122 |
/* Number of single byte characters. */ |
| 123 |
#define SBC_MAX (UCHAR_MAX + 1) |
| 124 |
|
| 125 |
#define COLL_ELEM_LEN_MAX 8 |
| 126 |
|
| 127 |
/* The character which represents newline. */ |
| 128 |
#define NEWLINE_CHAR '\n' |
| 129 |
#define WIDE_NEWLINE_CHAR L'\n' |
| 130 |
|
| 131 |
/* Rename to standard API for using out of glibc. */ |
| 132 |
#ifndef _LIBC |
| 133 |
# undef __wctype |
| 134 |
# undef __iswalnum |
| 135 |
# undef __iswctype |
| 136 |
# undef __towlower |
| 137 |
# undef __towupper |
| 138 |
# define __wctype wctype |
| 139 |
# define __iswalnum iswalnum |
| 140 |
# define __iswctype iswctype |
| 141 |
# define __towlower towlower |
| 142 |
# define __towupper towupper |
| 143 |
# define __btowc btowc |
| 144 |
# define __mbrtowc mbrtowc |
| 145 |
# define __wcrtomb wcrtomb |
| 146 |
# define __regfree regfree |
| 147 |
#endif /* not _LIBC */ |
| 148 |
|
| 149 |
/* Types related to integers. Unless protected by #ifdef _LIBC, the |
| 150 |
regex code should avoid exact-width types like int32_t and uint64_t |
| 151 |
as some non-GCC platforms lack them, an issue when this code is |
| 152 |
used in Gnulib. */ |
| 153 |
|
| 154 |
#ifndef ULONG_WIDTH |
| 155 |
# define ULONG_WIDTH REGEX_UINTEGER_WIDTH (ULONG_MAX) |
| 156 |
/* The number of usable bits in an unsigned integer type with maximum |
| 157 |
value MAX, as an int expression suitable in #if. Cover all known |
| 158 |
practical hosts. This implementation exploits the fact that MAX is |
| 159 |
1 less than a power of 2, and merely counts the number of 1 bits in |
| 160 |
MAX; "COBn" means "count the number of 1 bits in the low-order n bits". */ |
| 161 |
# define REGEX_UINTEGER_WIDTH(max) REGEX_COB128 (max) |
| 162 |
# define REGEX_COB128(n) (REGEX_COB64 ((n) >> 31 >> 31 >> 2) + REGEX_COB64 (n)) |
| 163 |
# define REGEX_COB64(n) (REGEX_COB32 ((n) >> 31 >> 1) + REGEX_COB32 (n)) |
| 164 |
# define REGEX_COB32(n) (REGEX_COB16 ((n) >> 16) + REGEX_COB16 (n)) |
| 165 |
# define REGEX_COB16(n) (REGEX_COB8 ((n) >> 8) + REGEX_COB8 (n)) |
| 166 |
# define REGEX_COB8(n) (REGEX_COB4 ((n) >> 4) + REGEX_COB4 (n)) |
| 167 |
# define REGEX_COB4(n) (!!((n) & 8) + !!((n) & 4) + !!((n) & 2) + ((n) & 1)) |
| 168 |
# if ULONG_MAX / 2 + 1 != 1ul << (ULONG_WIDTH - 1) |
| 169 |
# error "ULONG_MAX out of range" |
| 170 |
# endif |
| 171 |
#endif |
| 172 |
|
| 173 |
/* The type of indexes into strings. This is signed, not size_t, |
| 174 |
since the API requires indexes to fit in regoff_t anyway, and using |
| 175 |
signed integers makes the code a bit smaller and presumably faster. |
| 176 |
The traditional GNU regex implementation uses int for indexes. |
| 177 |
The POSIX-compatible implementation uses a possibly-wider type. |
| 178 |
The name 'Idx' is three letters to minimize the hassle of |
| 179 |
reindenting a lot of regex code that formerly used 'int'. */ |
| 180 |
typedef regoff_t Idx; |
| 181 |
#ifdef _REGEX_LARGE_OFFSETS |
| 182 |
# define IDX_MAX SSIZE_MAX |
| 183 |
#else |
| 184 |
# define IDX_MAX INT_MAX |
| 185 |
#endif |
| 186 |
|
| 187 |
/* A hash value, suitable for computing hash tables. */ |
| 188 |
typedef __re_size_t re_hashval_t; |
| 189 |
|
| 190 |
/* An integer used to represent a set of bits. It must be unsigned, |
| 191 |
and must be at least as wide as unsigned int. */ |
| 192 |
typedef unsigned long int bitset_word_t; |
| 193 |
/* All bits set in a bitset_word_t. */ |
| 194 |
#define BITSET_WORD_MAX ULONG_MAX |
| 195 |
/* Number of bits in a bitset_word_t. */ |
| 196 |
#define BITSET_WORD_BITS ULONG_WIDTH |
| 197 |
|
| 198 |
/* Number of bitset_word_t values in a bitset_t. */ |
| 199 |
#define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) |
| 200 |
|
| 201 |
typedef bitset_word_t bitset_t[BITSET_WORDS]; |
| 202 |
typedef bitset_word_t *re_bitset_ptr_t; |
| 203 |
typedef const bitset_word_t *re_const_bitset_ptr_t; |
| 204 |
|
| 205 |
#define PREV_WORD_CONSTRAINT 0x0001 |
| 206 |
#define PREV_NOTWORD_CONSTRAINT 0x0002 |
| 207 |
#define NEXT_WORD_CONSTRAINT 0x0004 |
| 208 |
#define NEXT_NOTWORD_CONSTRAINT 0x0008 |
| 209 |
#define PREV_NEWLINE_CONSTRAINT 0x0010 |
| 210 |
#define NEXT_NEWLINE_CONSTRAINT 0x0020 |
| 211 |
#define PREV_BEGBUF_CONSTRAINT 0x0040 |
| 212 |
#define NEXT_ENDBUF_CONSTRAINT 0x0080 |
| 213 |
#define WORD_DELIM_CONSTRAINT 0x0100 |
| 214 |
#define NOT_WORD_DELIM_CONSTRAINT 0x0200 |
| 215 |
|
| 216 |
typedef enum |
| 217 |
{ |
| 218 |
INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, |
| 219 |
WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, |
| 220 |
WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, |
| 221 |
INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, |
| 222 |
LINE_FIRST = PREV_NEWLINE_CONSTRAINT, |
| 223 |
LINE_LAST = NEXT_NEWLINE_CONSTRAINT, |
| 224 |
BUF_FIRST = PREV_BEGBUF_CONSTRAINT, |
| 225 |
BUF_LAST = NEXT_ENDBUF_CONSTRAINT, |
| 226 |
WORD_DELIM = WORD_DELIM_CONSTRAINT, |
| 227 |
NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT |
| 228 |
} re_context_type; |
| 229 |
|
| 230 |
typedef struct |
| 231 |
{ |
| 232 |
Idx alloc; |
| 233 |
Idx nelem; |
| 234 |
Idx *elems; |
| 235 |
} re_node_set; |
| 236 |
|
| 237 |
typedef enum |
| 238 |
{ |
| 239 |
NON_TYPE = 0, |
| 240 |
|
| 241 |
/* Node type, These are used by token, node, tree. */ |
| 242 |
CHARACTER = 1, |
| 243 |
END_OF_RE = 2, |
| 244 |
SIMPLE_BRACKET = 3, |
| 245 |
OP_BACK_REF = 4, |
| 246 |
OP_PERIOD = 5, |
| 247 |
COMPLEX_BRACKET = 6, |
| 248 |
OP_UTF8_PERIOD = 7, |
| 249 |
|
| 250 |
/* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used |
| 251 |
when the debugger shows values of this enum type. */ |
| 252 |
#define EPSILON_BIT 8 |
| 253 |
OP_OPEN_SUBEXP = EPSILON_BIT | 0, |
| 254 |
OP_CLOSE_SUBEXP = EPSILON_BIT | 1, |
| 255 |
OP_ALT = EPSILON_BIT | 2, |
| 256 |
OP_DUP_ASTERISK = EPSILON_BIT | 3, |
| 257 |
ANCHOR = EPSILON_BIT | 4, |
| 258 |
|
| 259 |
/* Tree type, these are used only by tree. */ |
| 260 |
CONCAT = 16, |
| 261 |
SUBEXP = 17, |
| 262 |
|
| 263 |
/* Token type, these are used only by token. */ |
| 264 |
OP_DUP_PLUS = 18, |
| 265 |
OP_DUP_QUESTION, |
| 266 |
OP_OPEN_BRACKET, |
| 267 |
OP_CLOSE_BRACKET, |
| 268 |
OP_CHARSET_RANGE, |
| 269 |
OP_OPEN_DUP_NUM, |
| 270 |
OP_CLOSE_DUP_NUM, |
| 271 |
OP_NON_MATCH_LIST, |
| 272 |
OP_OPEN_COLL_ELEM, |
| 273 |
OP_CLOSE_COLL_ELEM, |
| 274 |
OP_OPEN_EQUIV_CLASS, |
| 275 |
OP_CLOSE_EQUIV_CLASS, |
| 276 |
OP_OPEN_CHAR_CLASS, |
| 277 |
OP_CLOSE_CHAR_CLASS, |
| 278 |
OP_WORD, |
| 279 |
OP_NOTWORD, |
| 280 |
OP_SPACE, |
| 281 |
OP_NOTSPACE, |
| 282 |
BACK_SLASH |
| 283 |
|
| 284 |
} re_token_type_t; |
| 285 |
|
| 286 |
typedef struct |
| 287 |
{ |
| 288 |
/* Multibyte characters. */ |
| 289 |
wchar_t *mbchars; |
| 290 |
|
| 291 |
#ifdef _LIBC |
| 292 |
/* Collating symbols. */ |
| 293 |
int32_t *coll_syms; |
| 294 |
#endif |
| 295 |
|
| 296 |
#ifdef _LIBC |
| 297 |
/* Equivalence classes. */ |
| 298 |
int32_t *equiv_classes; |
| 299 |
#endif |
| 300 |
|
| 301 |
/* Range expressions. */ |
| 302 |
#ifdef _LIBC |
| 303 |
uint32_t *range_starts; |
| 304 |
uint32_t *range_ends; |
| 305 |
#else |
| 306 |
wchar_t *range_starts; |
| 307 |
wchar_t *range_ends; |
| 308 |
#endif |
| 309 |
|
| 310 |
/* Character classes. */ |
| 311 |
wctype_t *char_classes; |
| 312 |
|
| 313 |
/* If this character set is the non-matching list. */ |
| 314 |
unsigned int non_match : 1; |
| 315 |
|
| 316 |
/* # of multibyte characters. */ |
| 317 |
Idx nmbchars; |
| 318 |
|
| 319 |
/* # of collating symbols. */ |
| 320 |
Idx ncoll_syms; |
| 321 |
|
| 322 |
/* # of equivalence classes. */ |
| 323 |
Idx nequiv_classes; |
| 324 |
|
| 325 |
/* # of range expressions. */ |
| 326 |
Idx nranges; |
| 327 |
|
| 328 |
/* # of character classes. */ |
| 329 |
Idx nchar_classes; |
| 330 |
} re_charset_t; |
| 331 |
|
| 332 |
typedef struct |
| 333 |
{ |
| 334 |
union |
| 335 |
{ |
| 336 |
unsigned char c; /* for CHARACTER */ |
| 337 |
re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ |
| 338 |
re_charset_t *mbcset; /* for COMPLEX_BRACKET */ |
| 339 |
Idx idx; /* for BACK_REF */ |
| 340 |
re_context_type ctx_type; /* for ANCHOR */ |
| 341 |
} opr; |
| 342 |
#if (__GNUC__ >= 2 || defined __clang__) && !defined __STRICT_ANSI__ |
| 343 |
re_token_type_t type : 8; |
| 344 |
#else |
| 345 |
re_token_type_t type; |
| 346 |
#endif |
| 347 |
unsigned int constraint : 10; /* context constraint */ |
| 348 |
unsigned int duplicated : 1; |
| 349 |
unsigned int opt_subexp : 1; |
| 350 |
unsigned int accept_mb : 1; |
| 351 |
/* These 2 bits can be moved into the union if needed (e.g. if running out |
| 352 |
of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ |
| 353 |
unsigned int mb_partial : 1; |
| 354 |
unsigned int word_char : 1; |
| 355 |
} re_token_t; |
| 356 |
|
| 357 |
#define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) |
| 358 |
|
| 359 |
struct re_string_t |
| 360 |
{ |
| 361 |
/* Indicate the raw buffer which is the original string passed as an |
| 362 |
argument of regexec(), re_search(), etc.. */ |
| 363 |
const unsigned char *raw_mbs; |
| 364 |
/* Store the multibyte string. In case of "case insensitive mode" like |
| 365 |
REG_ICASE, upper cases of the string are stored, otherwise MBS points |
| 366 |
the same address that RAW_MBS points. */ |
| 367 |
unsigned char *mbs; |
| 368 |
/* Store the wide character string which is corresponding to MBS. */ |
| 369 |
wint_t *wcs; |
| 370 |
Idx *offsets; |
| 371 |
mbstate_t cur_state; |
| 372 |
/* Index in RAW_MBS. Each character mbs[i] corresponds to |
| 373 |
raw_mbs[raw_mbs_idx + i]. */ |
| 374 |
Idx raw_mbs_idx; |
| 375 |
/* The length of the valid characters in the buffers. */ |
| 376 |
Idx valid_len; |
| 377 |
/* The corresponding number of bytes in raw_mbs array. */ |
| 378 |
Idx valid_raw_len; |
| 379 |
/* The length of the buffers MBS and WCS. */ |
| 380 |
Idx bufs_len; |
| 381 |
/* The index in MBS, which is updated by re_string_fetch_byte. */ |
| 382 |
Idx cur_idx; |
| 383 |
/* length of RAW_MBS array. */ |
| 384 |
Idx raw_len; |
| 385 |
/* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ |
| 386 |
Idx len; |
| 387 |
/* End of the buffer may be shorter than its length in the cases such |
| 388 |
as re_match_2, re_search_2. Then, we use STOP for end of the buffer |
| 389 |
instead of LEN. */ |
| 390 |
Idx raw_stop; |
| 391 |
/* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ |
| 392 |
Idx stop; |
| 393 |
|
| 394 |
/* The context of mbs[0]. We store the context independently, since |
| 395 |
the context of mbs[0] may be different from raw_mbs[0], which is |
| 396 |
the beginning of the input string. */ |
| 397 |
unsigned int tip_context; |
| 398 |
/* The translation passed as a part of an argument of re_compile_pattern. */ |
| 399 |
RE_TRANSLATE_TYPE trans; |
| 400 |
/* Copy of re_dfa_t's word_char. */ |
| 401 |
re_const_bitset_ptr_t word_char; |
| 402 |
/* true if REG_ICASE. */ |
| 403 |
unsigned char icase; |
| 404 |
unsigned char is_utf8; |
| 405 |
unsigned char map_notascii; |
| 406 |
unsigned char mbs_allocated; |
| 407 |
unsigned char offsets_needed; |
| 408 |
unsigned char newline_anchor; |
| 409 |
unsigned char word_ops_used; |
| 410 |
int mb_cur_max; |
| 411 |
}; |
| 412 |
typedef struct re_string_t re_string_t; |
| 413 |
|
| 414 |
|
| 415 |
struct re_dfa_t; |
| 416 |
typedef struct re_dfa_t re_dfa_t; |
| 417 |
|
| 418 |
#ifndef _LIBC |
| 419 |
# define IS_IN(libc) false |
| 420 |
#endif |
| 421 |
|
| 422 |
#define re_string_peek_byte(pstr, offset) \ |
| 423 |
((pstr)->mbs[(pstr)->cur_idx + offset]) |
| 424 |
#define re_string_fetch_byte(pstr) \ |
| 425 |
((pstr)->mbs[(pstr)->cur_idx++]) |
| 426 |
#define re_string_first_byte(pstr, idx) \ |
| 427 |
((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) |
| 428 |
#define re_string_is_single_byte_char(pstr, idx) \ |
| 429 |
((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ |
| 430 |
|| (pstr)->wcs[(idx) + 1] != WEOF)) |
| 431 |
#define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) |
| 432 |
#define re_string_cur_idx(pstr) ((pstr)->cur_idx) |
| 433 |
#define re_string_get_buffer(pstr) ((pstr)->mbs) |
| 434 |
#define re_string_length(pstr) ((pstr)->len) |
| 435 |
#define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) |
| 436 |
#define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) |
| 437 |
#define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) |
| 438 |
|
| 439 |
#ifndef MAX |
| 440 |
# define MAX(a,b) ((a) < (b) ? (b) : (a)) |
| 441 |
#endif |
| 442 |
#ifndef MIN |
| 443 |
# define MIN(a,b) ((a) < (b) ? (a) : (b)) |
| 444 |
#endif |
| 445 |
|
| 446 |
#define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) |
| 447 |
#define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) |
| 448 |
#define re_free(p) free (p) |
| 449 |
|
| 450 |
struct bin_tree_t |
| 451 |
{ |
| 452 |
struct bin_tree_t *parent; |
| 453 |
struct bin_tree_t *left; |
| 454 |
struct bin_tree_t *right; |
| 455 |
struct bin_tree_t *first; |
| 456 |
struct bin_tree_t *next; |
| 457 |
|
| 458 |
re_token_t token; |
| 459 |
|
| 460 |
/* 'node_idx' is the index in dfa->nodes, if 'type' == 0. |
| 461 |
Otherwise 'type' indicate the type of this node. */ |
| 462 |
Idx node_idx; |
| 463 |
}; |
| 464 |
typedef struct bin_tree_t bin_tree_t; |
| 465 |
|
| 466 |
#define BIN_TREE_STORAGE_SIZE \ |
| 467 |
((1024 - sizeof (void *)) / sizeof (bin_tree_t)) |
| 468 |
|
| 469 |
struct bin_tree_storage_t |
| 470 |
{ |
| 471 |
struct bin_tree_storage_t *next; |
| 472 |
bin_tree_t data[BIN_TREE_STORAGE_SIZE]; |
| 473 |
}; |
| 474 |
typedef struct bin_tree_storage_t bin_tree_storage_t; |
| 475 |
|
| 476 |
#define CONTEXT_WORD 1 |
| 477 |
#define CONTEXT_NEWLINE (CONTEXT_WORD << 1) |
| 478 |
#define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) |
| 479 |
#define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) |
| 480 |
|
| 481 |
#define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) |
| 482 |
#define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) |
| 483 |
#define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) |
| 484 |
#define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) |
| 485 |
#define IS_ORDINARY_CONTEXT(c) ((c) == 0) |
| 486 |
|
| 487 |
#define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') |
| 488 |
#define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) |
| 489 |
#define IS_WIDE_WORD_CHAR(ch) (__iswalnum (ch) || (ch) == L'_') |
| 490 |
#define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) |
| 491 |
|
| 492 |
#define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ |
| 493 |
((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ |
| 494 |
|| ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ |
| 495 |
|| ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ |
| 496 |
|| ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) |
| 497 |
|
| 498 |
#define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ |
| 499 |
((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ |
| 500 |
|| (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ |
| 501 |
|| (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ |
| 502 |
|| (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) |
| 503 |
|
| 504 |
struct re_dfastate_t |
| 505 |
{ |
| 506 |
re_hashval_t hash; |
| 507 |
re_node_set nodes; |
| 508 |
re_node_set non_eps_nodes; |
| 509 |
re_node_set inveclosure; |
| 510 |
re_node_set *entrance_nodes; |
| 511 |
struct re_dfastate_t **trtable, **word_trtable; |
| 512 |
unsigned int context : 4; |
| 513 |
unsigned int halt : 1; |
| 514 |
/* If this state can accept "multi byte". |
| 515 |
Note that we refer to multibyte characters, and multi character |
| 516 |
collating elements as "multi byte". */ |
| 517 |
unsigned int accept_mb : 1; |
| 518 |
/* If this state has backreference node(s). */ |
| 519 |
unsigned int has_backref : 1; |
| 520 |
unsigned int has_constraint : 1; |
| 521 |
}; |
| 522 |
typedef struct re_dfastate_t re_dfastate_t; |
| 523 |
|
| 524 |
struct re_state_table_entry |
| 525 |
{ |
| 526 |
Idx num; |
| 527 |
Idx alloc; |
| 528 |
re_dfastate_t **array; |
| 529 |
}; |
| 530 |
|
| 531 |
/* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ |
| 532 |
|
| 533 |
typedef struct |
| 534 |
{ |
| 535 |
Idx next_idx; |
| 536 |
Idx alloc; |
| 537 |
re_dfastate_t **array; |
| 538 |
} state_array_t; |
| 539 |
|
| 540 |
/* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ |
| 541 |
|
| 542 |
typedef struct |
| 543 |
{ |
| 544 |
Idx node; |
| 545 |
Idx str_idx; /* The position NODE match at. */ |
| 546 |
state_array_t path; |
| 547 |
} re_sub_match_last_t; |
| 548 |
|
| 549 |
/* Store information about the node NODE whose type is OP_OPEN_SUBEXP. |
| 550 |
And information about the node, whose type is OP_CLOSE_SUBEXP, |
| 551 |
corresponding to NODE is stored in LASTS. */ |
| 552 |
|
| 553 |
typedef struct |
| 554 |
{ |
| 555 |
Idx str_idx; |
| 556 |
Idx node; |
| 557 |
state_array_t *path; |
| 558 |
Idx alasts; /* Allocation size of LASTS. */ |
| 559 |
Idx nlasts; /* The number of LASTS. */ |
| 560 |
re_sub_match_last_t **lasts; |
| 561 |
} re_sub_match_top_t; |
| 562 |
|
| 563 |
struct re_backref_cache_entry |
| 564 |
{ |
| 565 |
Idx node; |
| 566 |
Idx str_idx; |
| 567 |
Idx subexp_from; |
| 568 |
Idx subexp_to; |
| 569 |
bitset_word_t eps_reachable_subexps_map; |
| 570 |
char more; |
| 571 |
}; |
| 572 |
|
| 573 |
typedef struct |
| 574 |
{ |
| 575 |
/* The string object corresponding to the input string. */ |
| 576 |
re_string_t input; |
| 577 |
const re_dfa_t *const dfa; |
| 578 |
/* EFLAGS of the argument of regexec. */ |
| 579 |
int eflags; |
| 580 |
/* Where the matching ends. */ |
| 581 |
Idx match_last; |
| 582 |
Idx last_node; |
| 583 |
/* The state log used by the matcher. */ |
| 584 |
re_dfastate_t **state_log; |
| 585 |
Idx state_log_top; |
| 586 |
/* Back reference cache. */ |
| 587 |
Idx nbkref_ents; |
| 588 |
Idx abkref_ents; |
| 589 |
struct re_backref_cache_entry *bkref_ents; |
| 590 |
int max_mb_elem_len; |
| 591 |
Idx nsub_tops; |
| 592 |
Idx asub_tops; |
| 593 |
re_sub_match_top_t **sub_tops; |
| 594 |
} re_match_context_t; |
| 595 |
|
| 596 |
typedef struct |
| 597 |
{ |
| 598 |
re_dfastate_t **sifted_states; |
| 599 |
re_dfastate_t **limited_states; |
| 600 |
Idx last_node; |
| 601 |
Idx last_str_idx; |
| 602 |
re_node_set limits; |
| 603 |
} re_sift_context_t; |
| 604 |
|
| 605 |
struct re_fail_stack_ent_t |
| 606 |
{ |
| 607 |
Idx idx; |
| 608 |
Idx node; |
| 609 |
regmatch_t *regs; |
| 610 |
re_node_set eps_via_nodes; |
| 611 |
}; |
| 612 |
|
| 613 |
struct re_fail_stack_t |
| 614 |
{ |
| 615 |
Idx num; |
| 616 |
Idx alloc; |
| 617 |
struct re_fail_stack_ent_t *stack; |
| 618 |
}; |
| 619 |
|
| 620 |
struct re_dfa_t |
| 621 |
{ |
| 622 |
re_token_t *nodes; |
| 623 |
size_t nodes_alloc; |
| 624 |
size_t nodes_len; |
| 625 |
Idx *nexts; |
| 626 |
Idx *org_indices; |
| 627 |
re_node_set *edests; |
| 628 |
re_node_set *eclosures; |
| 629 |
re_node_set *inveclosures; |
| 630 |
struct re_state_table_entry *state_table; |
| 631 |
re_dfastate_t *init_state; |
| 632 |
re_dfastate_t *init_state_word; |
| 633 |
re_dfastate_t *init_state_nl; |
| 634 |
re_dfastate_t *init_state_begbuf; |
| 635 |
bin_tree_t *str_tree; |
| 636 |
bin_tree_storage_t *str_tree_storage; |
| 637 |
re_bitset_ptr_t sb_char; |
| 638 |
int str_tree_storage_idx; |
| 639 |
|
| 640 |
/* number of subexpressions 're_nsub' is in regex_t. */ |
| 641 |
re_hashval_t state_hash_mask; |
| 642 |
Idx init_node; |
| 643 |
Idx nbackref; /* The number of backreference in this dfa. */ |
| 644 |
|
| 645 |
/* Bitmap expressing which backreference is used. */ |
| 646 |
bitset_word_t used_bkref_map; |
| 647 |
bitset_word_t completed_bkref_map; |
| 648 |
|
| 649 |
unsigned int has_plural_match : 1; |
| 650 |
/* If this dfa has "multibyte node", which is a backreference or |
| 651 |
a node which can accept multibyte character or multi character |
| 652 |
collating element. */ |
| 653 |
unsigned int has_mb_node : 1; |
| 654 |
unsigned int is_utf8 : 1; |
| 655 |
unsigned int map_notascii : 1; |
| 656 |
unsigned int word_ops_used : 1; |
| 657 |
int mb_cur_max; |
| 658 |
bitset_t word_char; |
| 659 |
reg_syntax_t syntax; |
| 660 |
Idx *subexp_map; |
| 661 |
#ifdef DEBUG |
| 662 |
char* re_str; |
| 663 |
#endif |
| 664 |
lock_define (lock) |
| 665 |
}; |
| 666 |
|
| 667 |
#define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) |
| 668 |
#define re_node_set_remove(set,id) \ |
| 669 |
(re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) |
| 670 |
#define re_node_set_empty(p) ((p)->nelem = 0) |
| 671 |
#define re_node_set_free(set) re_free ((set)->elems) |
| 672 |
|
| 673 |
|
| 674 |
typedef enum |
| 675 |
{ |
| 676 |
SB_CHAR, |
| 677 |
MB_CHAR, |
| 678 |
EQUIV_CLASS, |
| 679 |
COLL_SYM, |
| 680 |
CHAR_CLASS |
| 681 |
} bracket_elem_type; |
| 682 |
|
| 683 |
typedef struct |
| 684 |
{ |
| 685 |
bracket_elem_type type; |
| 686 |
union |
| 687 |
{ |
| 688 |
unsigned char ch; |
| 689 |
unsigned char *name; |
| 690 |
wchar_t wch; |
| 691 |
} opr; |
| 692 |
} bracket_elem_t; |
| 693 |
|
| 694 |
|
| 695 |
/* Functions for bitset_t operation. */ |
| 696 |
|
| 697 |
static inline void |
| 698 |
bitset_set (bitset_t set, Idx i) |
| 699 |
{ |
| 700 |
set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; |
| 701 |
} |
| 702 |
|
| 703 |
static inline void |
| 704 |
bitset_clear (bitset_t set, Idx i) |
| 705 |
{ |
| 706 |
set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); |
| 707 |
} |
| 708 |
|
| 709 |
static inline bool |
| 710 |
bitset_contain (const bitset_t set, Idx i) |
| 711 |
{ |
| 712 |
return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; |
| 713 |
} |
| 714 |
|
| 715 |
static inline void |
| 716 |
bitset_empty (bitset_t set) |
| 717 |
{ |
| 718 |
memset (set, '\0', sizeof (bitset_t)); |
| 719 |
} |
| 720 |
|
| 721 |
static inline void |
| 722 |
bitset_set_all (bitset_t set) |
| 723 |
{ |
| 724 |
memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); |
| 725 |
if (SBC_MAX % BITSET_WORD_BITS != 0) |
| 726 |
set[BITSET_WORDS - 1] = |
| 727 |
((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; |
| 728 |
} |
| 729 |
|
| 730 |
static inline void |
| 731 |
bitset_copy (bitset_t dest, const bitset_t src) |
| 732 |
{ |
| 733 |
memcpy (dest, src, sizeof (bitset_t)); |
| 734 |
} |
| 735 |
|
| 736 |
static inline void |
| 737 |
bitset_not (bitset_t set) |
| 738 |
{ |
| 739 |
int bitset_i; |
| 740 |
for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) |
| 741 |
set[bitset_i] = ~set[bitset_i]; |
| 742 |
if (SBC_MAX % BITSET_WORD_BITS != 0) |
| 743 |
set[BITSET_WORDS - 1] = |
| 744 |
((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) |
| 745 |
& ~set[BITSET_WORDS - 1]); |
| 746 |
} |
| 747 |
|
| 748 |
static inline void |
| 749 |
bitset_merge (bitset_t dest, const bitset_t src) |
| 750 |
{ |
| 751 |
int bitset_i; |
| 752 |
for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) |
| 753 |
dest[bitset_i] |= src[bitset_i]; |
| 754 |
} |
| 755 |
|
| 756 |
static inline void |
| 757 |
bitset_mask (bitset_t dest, const bitset_t src) |
| 758 |
{ |
| 759 |
int bitset_i; |
| 760 |
for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) |
| 761 |
dest[bitset_i] &= src[bitset_i]; |
| 762 |
} |
| 763 |
|
| 764 |
/* Functions for re_string. */ |
| 765 |
static int |
| 766 |
__attribute__ ((pure, unused)) |
| 767 |
re_string_char_size_at (const re_string_t *pstr, Idx idx) |
| 768 |
{ |
| 769 |
int byte_idx; |
| 770 |
if (pstr->mb_cur_max == 1) |
| 771 |
return 1; |
| 772 |
for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) |
| 773 |
if (pstr->wcs[idx + byte_idx] != WEOF) |
| 774 |
break; |
| 775 |
return byte_idx; |
| 776 |
} |
| 777 |
|
| 778 |
static wint_t |
| 779 |
__attribute__ ((pure, unused)) |
| 780 |
re_string_wchar_at (const re_string_t *pstr, Idx idx) |
| 781 |
{ |
| 782 |
if (pstr->mb_cur_max == 1) |
| 783 |
return (wint_t) pstr->mbs[idx]; |
| 784 |
return (wint_t) pstr->wcs[idx]; |
| 785 |
} |
| 786 |
|
| 787 |
#ifdef _LIBC |
| 788 |
# include <locale/weight.h> |
| 789 |
#endif |
| 790 |
|
| 791 |
static int |
| 792 |
__attribute__ ((pure, unused)) |
| 793 |
re_string_elem_size_at (const re_string_t *pstr, Idx idx) |
| 794 |
{ |
| 795 |
#ifdef _LIBC |
| 796 |
const unsigned char *p, *extra; |
| 797 |
const int32_t *table, *indirect; |
| 798 |
uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); |
| 799 |
|
| 800 |
if (nrules != 0) |
| 801 |
{ |
| 802 |
table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); |
| 803 |
extra = (const unsigned char *) |
| 804 |
_NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); |
| 805 |
indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, |
| 806 |
_NL_COLLATE_INDIRECTMB); |
| 807 |
p = pstr->mbs + idx; |
| 808 |
findidx (table, indirect, extra, &p, pstr->len - idx); |
| 809 |
return p - pstr->mbs - idx; |
| 810 |
} |
| 811 |
#endif /* _LIBC */ |
| 812 |
|
| 813 |
return 1; |
| 814 |
} |
| 815 |
|
| 816 |
#ifdef _LIBC |
| 817 |
# if __glibc_has_attribute (__fallthrough__) |
| 818 |
# define FALLTHROUGH __attribute__ ((__fallthrough__)) |
| 819 |
# else |
| 820 |
# define FALLTHROUGH ((void) 0) |
| 821 |
# endif |
| 822 |
#else |
| 823 |
# include "attribute.h" |
| 824 |
#endif |
| 825 |
|
| 826 |
#endif /* _REGEX_INTERNAL_H */ |