1 /* -*- buffer-read-only: t -*- vi: set ro: */ 2 /* DO NOT EDIT! GENERATED AUTOMATICALLY! */ 3 /* Extended regular expression matching and search library. 4 Copyright (C) 2002-2011 Free Software Foundation, Inc. 5 This file is part of the GNU C Library. 6 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. 7 8 This program is free software; you can redistribute it and/or modify 9 it under the terms of the GNU General Public License as published by 10 the Free Software Foundation; either version 3, or (at your option) 11 any later version. 12 13 This program is distributed in the hope that it will be useful, 14 but WITHOUT ANY WARRANTY; without even the implied warranty of 15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 16 GNU General Public License for more details. 17 18 You should have received a copy of the GNU General Public License along 19 with this program; if not, write to the Free Software Foundation, 20 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ 21 22 #ifndef _REGEX_INTERNAL_H 23 #define _REGEX_INTERNAL_H 1 24 25 #include <assert.h> 26 #include <ctype.h> 27 #include <stdbool.h> 28 #include <stdio.h> 29 #include <stdlib.h> 30 #include <string.h> 31 32 #include <langinfo.h> 33 #ifndef _LIBC 34 # include "localcharset.h" 35 #endif 36 #include <locale.h> 37 38 #include <wchar.h> 39 #include <wctype.h> 40 #include <stdint.h> 41 #if defined _LIBC 42 # include <bits/libc-lock.h> 43 #else 44 # define __libc_lock_init(NAME) do { } while (0) 45 # define __libc_lock_lock(NAME) do { } while (0) 46 # define __libc_lock_unlock(NAME) do { } while (0) 47 #endif 48 49 /* In case that the system doesn't have isblank(). */ 50 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) 51 # define isblank(ch) ((ch) == ' ' || (ch) == '\t') 52 #endif 53 54 #ifdef _LIBC 55 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS 56 # define _RE_DEFINE_LOCALE_FUNCTIONS 1 57 # include <locale/localeinfo.h> 58 # include <locale/elem-hash.h> 59 # include <locale/coll-lookup.h> 60 # endif 61 #endif 62 63 /* This is for other GNU distributions with internationalized messages. */ 64 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC 65 # include <libintl.h> 66 # ifdef _LIBC 67 # undef gettext 68 # define gettext(msgid) \ 69 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES) 70 # endif 71 #else 72 # define gettext(msgid) (msgid) 73 #endif 74 75 #ifndef gettext_noop 76 /* This define is so xgettext can find the internationalizable 77 strings. */ 78 # define gettext_noop(String) String 79 #endif 80 81 /* For loser systems without the definition. */ 82 #ifndef SIZE_MAX 83 # define SIZE_MAX ((size_t) -1) 84 #endif 85 86 #if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCSCOLL) || _LIBC 87 # define RE_ENABLE_I18N 88 #endif 89 90 #if __GNUC__ >= 3 91 # define BE(expr, val) __builtin_expect (expr, val) 92 #else 93 # define BE(expr, val) (expr) 94 # ifdef _LIBC 95 # define inline 96 # endif 97 #endif 98 99 /* Number of ASCII characters. */ 100 #define ASCII_CHARS 0x80 101 102 /* Number of single byte characters. */ 103 #define SBC_MAX (UCHAR_MAX + 1) 104 105 #define COLL_ELEM_LEN_MAX 8 106 107 /* The character which represents newline. */ 108 #define NEWLINE_CHAR '\n' 109 #define WIDE_NEWLINE_CHAR L'\n' 110 111 /* Rename to standard API for using out of glibc. */ 112 #ifndef _LIBC 113 # define __wctype wctype 114 # define __iswctype iswctype 115 # define __btowc btowc 116 # define __wcrtomb wcrtomb 117 # define __mbrtowc mbrtowc 118 # define __regfree regfree 119 # define attribute_hidden 120 #endif /* not _LIBC */ 121 122 #if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1) 123 # define __attribute(arg) __attribute__ (arg) 124 #else 125 # define __attribute(arg) 126 #endif 127 128 typedef __re_idx_t Idx; 129 130 /* Special return value for failure to match. */ 131 #define REG_MISSING ((Idx) -1) 132 133 /* Special return value for internal error. */ 134 #define REG_ERROR ((Idx) -2) 135 136 /* Test whether N is a valid index, and is not one of the above. */ 137 #ifdef _REGEX_LARGE_OFFSETS 138 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR) 139 #else 140 # define REG_VALID_INDEX(n) (0 <= (n)) 141 #endif 142 143 /* Test whether N is a valid nonzero index. */ 144 #ifdef _REGEX_LARGE_OFFSETS 145 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1)) 146 #else 147 # define REG_VALID_NONZERO_INDEX(n) (0 < (n)) 148 #endif 149 150 /* A hash value, suitable for computing hash tables. */ 151 typedef __re_size_t re_hashval_t; 152 153 /* An integer used to represent a set of bits. It must be unsigned, 154 and must be at least as wide as unsigned int. */ 155 typedef unsigned long int bitset_word_t; 156 /* All bits set in a bitset_word_t. */ 157 #define BITSET_WORD_MAX ULONG_MAX 158 159 /* Number of bits in a bitset_word_t. For portability to hosts with 160 padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)'; 161 instead, deduce it directly from BITSET_WORD_MAX. Avoid 162 greater-than-32-bit integers and unconditional shifts by more than 163 31 bits, as they're not portable. */ 164 #if BITSET_WORD_MAX == 0xffffffffUL 165 # define BITSET_WORD_BITS 32 166 #elif BITSET_WORD_MAX >> 31 >> 4 == 1 167 # define BITSET_WORD_BITS 36 168 #elif BITSET_WORD_MAX >> 31 >> 16 == 1 169 # define BITSET_WORD_BITS 48 170 #elif BITSET_WORD_MAX >> 31 >> 28 == 1 171 # define BITSET_WORD_BITS 60 172 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1 173 # define BITSET_WORD_BITS 64 174 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1 175 # define BITSET_WORD_BITS 72 176 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1 177 # define BITSET_WORD_BITS 128 178 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1 179 # define BITSET_WORD_BITS 256 180 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1 181 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */ 182 # if BITSET_WORD_BITS <= SBC_MAX 183 # error "Invalid SBC_MAX" 184 # endif 185 #else 186 # error "Add case for new bitset_word_t size" 187 #endif 188 189 /* Number of bitset_word_t values in a bitset_t. */ 190 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) 191 192 typedef bitset_word_t bitset_t[BITSET_WORDS]; 193 typedef bitset_word_t *re_bitset_ptr_t; 194 typedef const bitset_word_t *re_const_bitset_ptr_t; 195 196 #define PREV_WORD_CONSTRAINT 0x0001 197 #define PREV_NOTWORD_CONSTRAINT 0x0002 198 #define NEXT_WORD_CONSTRAINT 0x0004 199 #define NEXT_NOTWORD_CONSTRAINT 0x0008 200 #define PREV_NEWLINE_CONSTRAINT 0x0010 201 #define NEXT_NEWLINE_CONSTRAINT 0x0020 202 #define PREV_BEGBUF_CONSTRAINT 0x0040 203 #define NEXT_ENDBUF_CONSTRAINT 0x0080 204 #define WORD_DELIM_CONSTRAINT 0x0100 205 #define NOT_WORD_DELIM_CONSTRAINT 0x0200 206 207 typedef enum 208 { 209 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, 210 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, 211 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, 212 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, 213 LINE_FIRST = PREV_NEWLINE_CONSTRAINT, 214 LINE_LAST = NEXT_NEWLINE_CONSTRAINT, 215 BUF_FIRST = PREV_BEGBUF_CONSTRAINT, 216 BUF_LAST = NEXT_ENDBUF_CONSTRAINT, 217 WORD_DELIM = WORD_DELIM_CONSTRAINT, 218 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT 219 } re_context_type; 220 221 typedef struct 222 { 223 Idx alloc; 224 Idx nelem; 225 Idx *elems; 226 } re_node_set; 227 228 typedef enum 229 { 230 NON_TYPE = 0, 231 232 /* Node type, These are used by token, node, tree. */ 233 CHARACTER = 1, 234 END_OF_RE = 2, 235 SIMPLE_BRACKET = 3, 236 OP_BACK_REF = 4, 237 OP_PERIOD = 5, 238 #ifdef RE_ENABLE_I18N 239 COMPLEX_BRACKET = 6, 240 OP_UTF8_PERIOD = 7, 241 #endif /* RE_ENABLE_I18N */ 242 243 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used 244 when the debugger shows values of this enum type. */ 245 #define EPSILON_BIT 8 246 OP_OPEN_SUBEXP = EPSILON_BIT | 0, 247 OP_CLOSE_SUBEXP = EPSILON_BIT | 1, 248 OP_ALT = EPSILON_BIT | 2, 249 OP_DUP_ASTERISK = EPSILON_BIT | 3, 250 ANCHOR = EPSILON_BIT | 4, 251 252 /* Tree type, these are used only by tree. */ 253 CONCAT = 16, 254 SUBEXP = 17, 255 256 /* Token type, these are used only by token. */ 257 OP_DUP_PLUS = 18, 258 OP_DUP_QUESTION, 259 OP_OPEN_BRACKET, 260 OP_CLOSE_BRACKET, 261 OP_CHARSET_RANGE, 262 OP_OPEN_DUP_NUM, 263 OP_CLOSE_DUP_NUM, 264 OP_NON_MATCH_LIST, 265 OP_OPEN_COLL_ELEM, 266 OP_CLOSE_COLL_ELEM, 267 OP_OPEN_EQUIV_CLASS, 268 OP_CLOSE_EQUIV_CLASS, 269 OP_OPEN_CHAR_CLASS, 270 OP_CLOSE_CHAR_CLASS, 271 OP_WORD, 272 OP_NOTWORD, 273 OP_SPACE, 274 OP_NOTSPACE, 275 BACK_SLASH 276 277 } re_token_type_t; 278 279 #ifdef RE_ENABLE_I18N 280 typedef struct 281 { 282 /* Multibyte characters. */ 283 wchar_t *mbchars; 284 285 /* Collating symbols. */ 286 # ifdef _LIBC 287 int32_t *coll_syms; 288 # endif 289 290 /* Equivalence classes. */ 291 # ifdef _LIBC 292 int32_t *equiv_classes; 293 # endif 294 295 /* Range expressions. */ 296 # ifdef _LIBC 297 uint32_t *range_starts; 298 uint32_t *range_ends; 299 # else /* not _LIBC */ 300 wchar_t *range_starts; 301 wchar_t *range_ends; 302 # endif /* not _LIBC */ 303 304 /* Character classes. */ 305 wctype_t *char_classes; 306 307 /* If this character set is the non-matching list. */ 308 unsigned int non_match : 1; 309 310 /* # of multibyte characters. */ 311 Idx nmbchars; 312 313 /* # of collating symbols. */ 314 Idx ncoll_syms; 315 316 /* # of equivalence classes. */ 317 Idx nequiv_classes; 318 319 /* # of range expressions. */ 320 Idx nranges; 321 322 /* # of character classes. */ 323 Idx nchar_classes; 324 } re_charset_t; 325 #endif /* RE_ENABLE_I18N */ 326 327 typedef struct 328 { 329 union 330 { 331 unsigned char c; /* for CHARACTER */ 332 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ 333 #ifdef RE_ENABLE_I18N 334 re_charset_t *mbcset; /* for COMPLEX_BRACKET */ 335 #endif /* RE_ENABLE_I18N */ 336 Idx idx; /* for BACK_REF */ 337 re_context_type ctx_type; /* for ANCHOR */ 338 } opr; 339 #if __GNUC__ >= 2 && !__STRICT_ANSI__ 340 re_token_type_t type : 8; 341 #else 342 re_token_type_t type; 343 #endif 344 unsigned int constraint : 10; /* context constraint */ 345 unsigned int duplicated : 1; 346 unsigned int opt_subexp : 1; 347 #ifdef RE_ENABLE_I18N 348 unsigned int accept_mb : 1; 349 /* These 2 bits can be moved into the union if needed (e.g. if running out 350 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ 351 unsigned int mb_partial : 1; 352 #endif 353 unsigned int word_char : 1; 354 } re_token_t; 355 356 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) 357 358 struct re_string_t 359 { 360 /* Indicate the raw buffer which is the original string passed as an 361 argument of regexec(), re_search(), etc.. */ 362 const unsigned char *raw_mbs; 363 /* Store the multibyte string. In case of "case insensitive mode" like 364 REG_ICASE, upper cases of the string are stored, otherwise MBS points 365 the same address that RAW_MBS points. */ 366 unsigned char *mbs; 367 #ifdef RE_ENABLE_I18N 368 /* Store the wide character string which is corresponding to MBS. */ 369 wint_t *wcs; 370 Idx *offsets; 371 mbstate_t cur_state; 372 #endif 373 /* Index in RAW_MBS. Each character mbs[i] corresponds to 374 raw_mbs[raw_mbs_idx + i]. */ 375 Idx raw_mbs_idx; 376 /* The length of the valid characters in the buffers. */ 377 Idx valid_len; 378 /* The corresponding number of bytes in raw_mbs array. */ 379 Idx valid_raw_len; 380 /* The length of the buffers MBS and WCS. */ 381 Idx bufs_len; 382 /* The index in MBS, which is updated by re_string_fetch_byte. */ 383 Idx cur_idx; 384 /* length of RAW_MBS array. */ 385 Idx raw_len; 386 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ 387 Idx len; 388 /* End of the buffer may be shorter than its length in the cases such 389 as re_match_2, re_search_2. Then, we use STOP for end of the buffer 390 instead of LEN. */ 391 Idx raw_stop; 392 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ 393 Idx stop; 394 395 /* The context of mbs[0]. We store the context independently, since 396 the context of mbs[0] may be different from raw_mbs[0], which is 397 the beginning of the input string. */ 398 unsigned int tip_context; 399 /* The translation passed as a part of an argument of re_compile_pattern. */ 400 RE_TRANSLATE_TYPE trans; 401 /* Copy of re_dfa_t's word_char. */ 402 re_const_bitset_ptr_t word_char; 403 /* true if REG_ICASE. */ 404 unsigned char icase; 405 unsigned char is_utf8; 406 unsigned char map_notascii; 407 unsigned char mbs_allocated; 408 unsigned char offsets_needed; 409 unsigned char newline_anchor; 410 unsigned char word_ops_used; 411 int mb_cur_max; 412 }; 413 typedef struct re_string_t re_string_t; 414 415 416 struct re_dfa_t; 417 typedef struct re_dfa_t re_dfa_t; 418 419 #ifndef _LIBC 420 # define internal_function 421 #endif 422 423 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, 424 Idx new_buf_len) 425 internal_function; 426 #ifdef RE_ENABLE_I18N 427 static void build_wcs_buffer (re_string_t *pstr) internal_function; 428 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr) 429 internal_function; 430 #endif /* RE_ENABLE_I18N */ 431 static void build_upper_buffer (re_string_t *pstr) internal_function; 432 static void re_string_translate_buffer (re_string_t *pstr) internal_function; 433 static unsigned int re_string_context_at (const re_string_t *input, Idx idx, 434 int eflags) 435 internal_function __attribute ((pure)); 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 #include <alloca.h> 454 455 #ifndef _LIBC 456 # if HAVE_ALLOCA 457 /* The OS usually guarantees only one guard page at the bottom of the stack, 458 and a page size can be as small as 4096 bytes. So we cannot safely 459 allocate anything larger than 4096 bytes. Also care for the possibility 460 of a few compiler-allocated temporary stack slots. */ 461 # define __libc_use_alloca(n) ((n) < 4032) 462 # else 463 /* alloca is implemented with malloc, so just use malloc. */ 464 # define __libc_use_alloca(n) 0 465 # undef alloca 466 # define alloca(n) malloc (n) 467 # endif 468 #endif 469 470 #ifndef MAX 471 # define MAX(a,b) ((a) < (b) ? (b) : (a)) 472 #endif 473 474 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) 475 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) 476 #define re_free(p) free (p) 477 478 struct bin_tree_t 479 { 480 struct bin_tree_t *parent; 481 struct bin_tree_t *left; 482 struct bin_tree_t *right; 483 struct bin_tree_t *first; 484 struct bin_tree_t *next; 485 486 re_token_t token; 487 488 /* `node_idx' is the index in dfa->nodes, if `type' == 0. 489 Otherwise `type' indicate the type of this node. */ 490 Idx node_idx; 491 }; 492 typedef struct bin_tree_t bin_tree_t; 493 494 #define BIN_TREE_STORAGE_SIZE \ 495 ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) 496 497 struct bin_tree_storage_t 498 { 499 struct bin_tree_storage_t *next; 500 bin_tree_t data[BIN_TREE_STORAGE_SIZE]; 501 }; 502 typedef struct bin_tree_storage_t bin_tree_storage_t; 503 504 #define CONTEXT_WORD 1 505 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) 506 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) 507 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) 508 509 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) 510 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) 511 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) 512 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) 513 #define IS_ORDINARY_CONTEXT(c) ((c) == 0) 514 515 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') 516 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) 517 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_') 518 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) 519 520 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ 521 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 522 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 523 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ 524 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) 525 526 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ 527 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 528 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 529 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ 530 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) 531 532 struct re_dfastate_t 533 { 534 re_hashval_t hash; 535 re_node_set nodes; 536 re_node_set non_eps_nodes; 537 re_node_set inveclosure; 538 re_node_set *entrance_nodes; 539 struct re_dfastate_t **trtable, **word_trtable; 540 unsigned int context : 4; 541 unsigned int halt : 1; 542 /* If this state can accept `multi byte'. 543 Note that we refer to multibyte characters, and multi character 544 collating elements as `multi byte'. */ 545 unsigned int accept_mb : 1; 546 /* If this state has backreference node(s). */ 547 unsigned int has_backref : 1; 548 unsigned int has_constraint : 1; 549 }; 550 typedef struct re_dfastate_t re_dfastate_t; 551 552 struct re_state_table_entry 553 { 554 Idx num; 555 Idx alloc; 556 re_dfastate_t **array; 557 }; 558 559 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ 560 561 typedef struct 562 { 563 Idx next_idx; 564 Idx alloc; 565 re_dfastate_t **array; 566 } state_array_t; 567 568 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ 569 570 typedef struct 571 { 572 Idx node; 573 Idx str_idx; /* The position NODE match at. */ 574 state_array_t path; 575 } re_sub_match_last_t; 576 577 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. 578 And information about the node, whose type is OP_CLOSE_SUBEXP, 579 corresponding to NODE is stored in LASTS. */ 580 581 typedef struct 582 { 583 Idx str_idx; 584 Idx node; 585 state_array_t *path; 586 Idx alasts; /* Allocation size of LASTS. */ 587 Idx nlasts; /* The number of LASTS. */ 588 re_sub_match_last_t **lasts; 589 } re_sub_match_top_t; 590 591 struct re_backref_cache_entry 592 { 593 Idx node; 594 Idx str_idx; 595 Idx subexp_from; 596 Idx subexp_to; 597 char more; 598 char unused; 599 unsigned short int eps_reachable_subexps_map; 600 }; 601 602 typedef struct 603 { 604 /* The string object corresponding to the input string. */ 605 re_string_t input; 606 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) 607 const re_dfa_t *const dfa; 608 #else 609 const re_dfa_t *dfa; 610 #endif 611 /* EFLAGS of the argument of regexec. */ 612 int eflags; 613 /* Where the matching ends. */ 614 Idx match_last; 615 Idx last_node; 616 /* The state log used by the matcher. */ 617 re_dfastate_t **state_log; 618 Idx state_log_top; 619 /* Back reference cache. */ 620 Idx nbkref_ents; 621 Idx abkref_ents; 622 struct re_backref_cache_entry *bkref_ents; 623 int max_mb_elem_len; 624 Idx nsub_tops; 625 Idx asub_tops; 626 re_sub_match_top_t **sub_tops; 627 } re_match_context_t; 628 629 typedef struct 630 { 631 re_dfastate_t **sifted_states; 632 re_dfastate_t **limited_states; 633 Idx last_node; 634 Idx last_str_idx; 635 re_node_set limits; 636 } re_sift_context_t; 637 638 struct re_fail_stack_ent_t 639 { 640 Idx idx; 641 Idx node; 642 regmatch_t *regs; 643 re_node_set eps_via_nodes; 644 }; 645 646 struct re_fail_stack_t 647 { 648 Idx num; 649 Idx alloc; 650 struct re_fail_stack_ent_t *stack; 651 }; 652 653 struct re_dfa_t 654 { 655 re_token_t *nodes; 656 size_t nodes_alloc; 657 size_t nodes_len; 658 Idx *nexts; 659 Idx *org_indices; 660 re_node_set *edests; 661 re_node_set *eclosures; 662 re_node_set *inveclosures; 663 struct re_state_table_entry *state_table; 664 re_dfastate_t *init_state; 665 re_dfastate_t *init_state_word; 666 re_dfastate_t *init_state_nl; 667 re_dfastate_t *init_state_begbuf; 668 bin_tree_t *str_tree; 669 bin_tree_storage_t *str_tree_storage; 670 re_bitset_ptr_t sb_char; 671 int str_tree_storage_idx; 672 673 /* number of subexpressions `re_nsub' is in regex_t. */ 674 re_hashval_t state_hash_mask; 675 Idx init_node; 676 Idx nbackref; /* The number of backreference in this dfa. */ 677 678 /* Bitmap expressing which backreference is used. */ 679 bitset_word_t used_bkref_map; 680 bitset_word_t completed_bkref_map; 681 682 unsigned int has_plural_match : 1; 683 /* If this dfa has "multibyte node", which is a backreference or 684 a node which can accept multibyte character or multi character 685 collating element. */ 686 unsigned int has_mb_node : 1; 687 unsigned int is_utf8 : 1; 688 unsigned int map_notascii : 1; 689 unsigned int word_ops_used : 1; 690 int mb_cur_max; 691 bitset_t word_char; 692 reg_syntax_t syntax; 693 Idx *subexp_map; 694 #ifdef DEBUG 695 char* re_str; 696 #endif 697 #ifdef _LIBC 698 __libc_lock_define (, lock) 699 #endif 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) 707 708 709 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 } 822 823 static int 824 internal_function __attribute ((pure)) 825 re_string_elem_size_at (const re_string_t *pstr _UNUSED_PARAMETER_, 826 Idx idx _UNUSED_PARAMETER_) 827 { 828 # ifdef _LIBC 829 const unsigned char *p, *extra; 830 const int32_t *table, *indirect; 831 int32_t tmp; 832 # include <locale/weight.h> 833 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); 834 835 if (nrules != 0) 836 { 837 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); 838 extra = (const unsigned char *) 839 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); 840 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, 841 _NL_COLLATE_INDIRECTMB); 842 p = pstr->mbs + idx; 843 tmp = findidx (&p); 844 return p - pstr->mbs - idx; 845 } 846 else 847 # endif /* _LIBC */ 848 return 1; 849 } 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 */ 870