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, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010 Free 5 Software Foundation, Inc. 6 This file is part of the GNU C Library. 7 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. 8 9 This program is free software; you can redistribute it and/or modify 10 it under the terms of the GNU General Public License as published by 11 the Free Software Foundation; either version 3, or (at your option) 12 any later version. 13 14 This program is distributed in the hope that it will be useful, 15 but WITHOUT ANY WARRANTY; without even the implied warranty of 16 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 17 GNU General Public License for more details. 18 19 You should have received a copy of the GNU General Public License along 20 with this program; if not, write to the Free Software Foundation, 21 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ 22 23 #ifndef _REGEX_INTERNAL_H 24 #define _REGEX_INTERNAL_H 1 25 26 #include <assert.h> 27 #include <ctype.h> 28 #include <stdbool.h> 29 #include <stdio.h> 30 #include <stdlib.h> 31 #include <string.h> 32 33 #include <langinfo.h> 34 #ifndef _LIBC 35 # include "localcharset.h" 36 #endif 37 #if defined HAVE_LOCALE_H || defined _LIBC 38 # include <locale.h> 39 #endif 40 41 #include <wchar.h> 42 #include <wctype.h> 43 #include <stdint.h> 44 #if defined _LIBC 45 # include <bits/libc-lock.h> 46 #else 47 # define __libc_lock_init(NAME) do { } while (0) 48 # define __libc_lock_lock(NAME) do { } while (0) 49 # define __libc_lock_unlock(NAME) do { } while (0) 50 #endif 51 52 /* In case that the system doesn't have isblank(). */ 53 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) 54 # define isblank(ch) ((ch) == ' ' || (ch) == '\t') 55 #endif 56 57 #ifdef _LIBC 58 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS 59 # define _RE_DEFINE_LOCALE_FUNCTIONS 1 60 # include <locale/localeinfo.h> 61 # include <locale/elem-hash.h> 62 # include <locale/coll-lookup.h> 63 # endif 64 #endif 65 66 /* This is for other GNU distributions with internationalized messages. */ 67 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC 68 # include <libintl.h> 69 # ifdef _LIBC 70 # undef gettext 71 # define gettext(msgid) \ 72 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES) 73 # endif 74 #else 75 # define gettext(msgid) (msgid) 76 #endif 77 78 #ifndef gettext_noop 79 /* This define is so xgettext can find the internationalizable 80 strings. */ 81 # define gettext_noop(String) String 82 #endif 83 84 /* For loser systems without the definition. */ 85 #ifndef SIZE_MAX 86 # define SIZE_MAX ((size_t) -1) 87 #endif 88 89 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCSCOLL) || _LIBC 90 # define RE_ENABLE_I18N 91 #endif 92 93 #if __GNUC__ >= 3 94 # define BE(expr, val) __builtin_expect (expr, val) 95 #else 96 # define BE(expr, val) (expr) 97 # ifdef _LIBC 98 # define inline 99 # endif 100 #endif 101 102 /* Number of ASCII characters. */ 103 #define ASCII_CHARS 0x80 104 105 /* Number of single byte characters. */ 106 #define SBC_MAX (UCHAR_MAX + 1) 107 108 #define COLL_ELEM_LEN_MAX 8 109 110 /* The character which represents newline. */ 111 #define NEWLINE_CHAR '\n' 112 #define WIDE_NEWLINE_CHAR L'\n' 113 114 /* Rename to standard API for using out of glibc. */ 115 #ifndef _LIBC 116 # define __wctype wctype 117 # define __iswctype iswctype 118 # define __btowc btowc 119 # define __wcrtomb wcrtomb 120 # define __mbrtowc mbrtowc 121 # define __regfree regfree 122 # define attribute_hidden 123 #endif /* not _LIBC */ 124 125 #if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1) 126 # define __attribute(arg) __attribute__ (arg) 127 #else 128 # define __attribute(arg) 129 #endif 130 131 typedef __re_idx_t Idx; 132 133 /* Special return value for failure to match. */ 134 #define REG_MISSING ((Idx) -1) 135 136 /* Special return value for internal error. */ 137 #define REG_ERROR ((Idx) -2) 138 139 /* Test whether N is a valid index, and is not one of the above. */ 140 #ifdef _REGEX_LARGE_OFFSETS 141 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR) 142 #else 143 # define REG_VALID_INDEX(n) (0 <= (n)) 144 #endif 145 146 /* Test whether N is a valid nonzero index. */ 147 #ifdef _REGEX_LARGE_OFFSETS 148 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1)) 149 #else 150 # define REG_VALID_NONZERO_INDEX(n) (0 < (n)) 151 #endif 152 153 /* A hash value, suitable for computing hash tables. */ 154 typedef __re_size_t re_hashval_t; 155 156 /* An integer used to represent a set of bits. It must be unsigned, 157 and must be at least as wide as unsigned int. */ 158 typedef unsigned long int bitset_word_t; 159 /* All bits set in a bitset_word_t. */ 160 #define BITSET_WORD_MAX ULONG_MAX 161 162 /* Number of bits in a bitset_word_t. For portability to hosts with 163 padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)'; 164 instead, deduce it directly from BITSET_WORD_MAX. Avoid 165 greater-than-32-bit integers and unconditional shifts by more than 166 31 bits, as they're not portable. */ 167 #if BITSET_WORD_MAX == 0xffffffffUL 168 # define BITSET_WORD_BITS 32 169 #elif BITSET_WORD_MAX >> 31 >> 4 == 1 170 # define BITSET_WORD_BITS 36 171 #elif BITSET_WORD_MAX >> 31 >> 16 == 1 172 # define BITSET_WORD_BITS 48 173 #elif BITSET_WORD_MAX >> 31 >> 28 == 1 174 # define BITSET_WORD_BITS 60 175 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1 176 # define BITSET_WORD_BITS 64 177 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1 178 # define BITSET_WORD_BITS 72 179 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1 180 # define BITSET_WORD_BITS 128 181 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1 182 # define BITSET_WORD_BITS 256 183 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1 184 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */ 185 # if BITSET_WORD_BITS <= SBC_MAX 186 # error "Invalid SBC_MAX" 187 # endif 188 #else 189 # error "Add case for new bitset_word_t size" 190 #endif 191 192 /* Number of bitset_word_t values in a bitset_t. */ 193 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) 194 195 typedef bitset_word_t bitset_t[BITSET_WORDS]; 196 typedef bitset_word_t *re_bitset_ptr_t; 197 typedef const bitset_word_t *re_const_bitset_ptr_t; 198 199 #define PREV_WORD_CONSTRAINT 0x0001 200 #define PREV_NOTWORD_CONSTRAINT 0x0002 201 #define NEXT_WORD_CONSTRAINT 0x0004 202 #define NEXT_NOTWORD_CONSTRAINT 0x0008 203 #define PREV_NEWLINE_CONSTRAINT 0x0010 204 #define NEXT_NEWLINE_CONSTRAINT 0x0020 205 #define PREV_BEGBUF_CONSTRAINT 0x0040 206 #define NEXT_ENDBUF_CONSTRAINT 0x0080 207 #define WORD_DELIM_CONSTRAINT 0x0100 208 #define NOT_WORD_DELIM_CONSTRAINT 0x0200 209 210 typedef enum 211 { 212 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, 213 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, 214 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, 215 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, 216 LINE_FIRST = PREV_NEWLINE_CONSTRAINT, 217 LINE_LAST = NEXT_NEWLINE_CONSTRAINT, 218 BUF_FIRST = PREV_BEGBUF_CONSTRAINT, 219 BUF_LAST = NEXT_ENDBUF_CONSTRAINT, 220 WORD_DELIM = WORD_DELIM_CONSTRAINT, 221 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT 222 } re_context_type; 223 224 typedef struct 225 { 226 Idx alloc; 227 Idx nelem; 228 Idx *elems; 229 } re_node_set; 230 231 typedef enum 232 { 233 NON_TYPE = 0, 234 235 /* Node type, These are used by token, node, tree. */ 236 CHARACTER = 1, 237 END_OF_RE = 2, 238 SIMPLE_BRACKET = 3, 239 OP_BACK_REF = 4, 240 OP_PERIOD = 5, 241 #ifdef RE_ENABLE_I18N 242 COMPLEX_BRACKET = 6, 243 OP_UTF8_PERIOD = 7, 244 #endif /* RE_ENABLE_I18N */ 245 246 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used 247 when the debugger shows values of this enum type. */ 248 #define EPSILON_BIT 8 249 OP_OPEN_SUBEXP = EPSILON_BIT | 0, 250 OP_CLOSE_SUBEXP = EPSILON_BIT | 1, 251 OP_ALT = EPSILON_BIT | 2, 252 OP_DUP_ASTERISK = EPSILON_BIT | 3, 253 ANCHOR = EPSILON_BIT | 4, 254 255 /* Tree type, these are used only by tree. */ 256 CONCAT = 16, 257 SUBEXP = 17, 258 259 /* Token type, these are used only by token. */ 260 OP_DUP_PLUS = 18, 261 OP_DUP_QUESTION, 262 OP_OPEN_BRACKET, 263 OP_CLOSE_BRACKET, 264 OP_CHARSET_RANGE, 265 OP_OPEN_DUP_NUM, 266 OP_CLOSE_DUP_NUM, 267 OP_NON_MATCH_LIST, 268 OP_OPEN_COLL_ELEM, 269 OP_CLOSE_COLL_ELEM, 270 OP_OPEN_EQUIV_CLASS, 271 OP_CLOSE_EQUIV_CLASS, 272 OP_OPEN_CHAR_CLASS, 273 OP_CLOSE_CHAR_CLASS, 274 OP_WORD, 275 OP_NOTWORD, 276 OP_SPACE, 277 OP_NOTSPACE, 278 BACK_SLASH 279 280 } re_token_type_t; 281 282 #ifdef RE_ENABLE_I18N 283 typedef struct 284 { 285 /* Multibyte characters. */ 286 wchar_t *mbchars; 287 288 /* Collating symbols. */ 289 # ifdef _LIBC 290 int32_t *coll_syms; 291 # endif 292 293 /* Equivalence classes. */ 294 # ifdef _LIBC 295 int32_t *equiv_classes; 296 # endif 297 298 /* Range expressions. */ 299 # ifdef _LIBC 300 uint32_t *range_starts; 301 uint32_t *range_ends; 302 # else /* not _LIBC */ 303 wchar_t *range_starts; 304 wchar_t *range_ends; 305 # endif /* not _LIBC */ 306 307 /* Character classes. */ 308 wctype_t *char_classes; 309 310 /* If this character set is the non-matching list. */ 311 unsigned int non_match : 1; 312 313 /* # of multibyte characters. */ 314 Idx nmbchars; 315 316 /* # of collating symbols. */ 317 Idx ncoll_syms; 318 319 /* # of equivalence classes. */ 320 Idx nequiv_classes; 321 322 /* # of range expressions. */ 323 Idx nranges; 324 325 /* # of character classes. */ 326 Idx nchar_classes; 327 } re_charset_t; 328 #endif /* RE_ENABLE_I18N */ 329 330 typedef struct 331 { 332 union 333 { 334 unsigned char c; /* for CHARACTER */ 335 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ 336 #ifdef RE_ENABLE_I18N 337 re_charset_t *mbcset; /* for COMPLEX_BRACKET */ 338 #endif /* RE_ENABLE_I18N */ 339 Idx idx; /* for BACK_REF */ 340 re_context_type ctx_type; /* for ANCHOR */ 341 } opr; 342 #if __GNUC__ >= 2 && !__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 #ifdef RE_ENABLE_I18N 351 unsigned int accept_mb : 1; 352 /* These 2 bits can be moved into the union if needed (e.g. if running out 353 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ 354 unsigned int mb_partial : 1; 355 #endif 356 unsigned int word_char : 1; 357 } re_token_t; 358 359 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) 360 361 struct re_string_t 362 { 363 /* Indicate the raw buffer which is the original string passed as an 364 argument of regexec(), re_search(), etc.. */ 365 const unsigned char *raw_mbs; 366 /* Store the multibyte string. In case of "case insensitive mode" like 367 REG_ICASE, upper cases of the string are stored, otherwise MBS points 368 the same address that RAW_MBS points. */ 369 unsigned char *mbs; 370 #ifdef RE_ENABLE_I18N 371 /* Store the wide character string which is corresponding to MBS. */ 372 wint_t *wcs; 373 Idx *offsets; 374 mbstate_t cur_state; 375 #endif 376 /* Index in RAW_MBS. Each character mbs[i] corresponds to 377 raw_mbs[raw_mbs_idx + i]. */ 378 Idx raw_mbs_idx; 379 /* The length of the valid characters in the buffers. */ 380 Idx valid_len; 381 /* The corresponding number of bytes in raw_mbs array. */ 382 Idx valid_raw_len; 383 /* The length of the buffers MBS and WCS. */ 384 Idx bufs_len; 385 /* The index in MBS, which is updated by re_string_fetch_byte. */ 386 Idx cur_idx; 387 /* length of RAW_MBS array. */ 388 Idx raw_len; 389 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ 390 Idx len; 391 /* End of the buffer may be shorter than its length in the cases such 392 as re_match_2, re_search_2. Then, we use STOP for end of the buffer 393 instead of LEN. */ 394 Idx raw_stop; 395 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ 396 Idx stop; 397 398 /* The context of mbs[0]. We store the context independently, since 399 the context of mbs[0] may be different from raw_mbs[0], which is 400 the beginning of the input string. */ 401 unsigned int tip_context; 402 /* The translation passed as a part of an argument of re_compile_pattern. */ 403 RE_TRANSLATE_TYPE trans; 404 /* Copy of re_dfa_t's word_char. */ 405 re_const_bitset_ptr_t word_char; 406 /* true if REG_ICASE. */ 407 unsigned char icase; 408 unsigned char is_utf8; 409 unsigned char map_notascii; 410 unsigned char mbs_allocated; 411 unsigned char offsets_needed; 412 unsigned char newline_anchor; 413 unsigned char word_ops_used; 414 int mb_cur_max; 415 }; 416 typedef struct re_string_t re_string_t; 417 418 419 struct re_dfa_t; 420 typedef struct re_dfa_t re_dfa_t; 421 422 #ifndef _LIBC 423 # if defined __i386__ && !defined __EMX__ 424 # define internal_function __attribute ((regparm (3), stdcall)) 425 # else 426 # define internal_function 427 # endif 428 #endif 429 430 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, 431 Idx new_buf_len) 432 internal_function; 433 #ifdef RE_ENABLE_I18N 434 static void build_wcs_buffer (re_string_t *pstr) internal_function; 435 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr) 436 internal_function; 437 #endif /* RE_ENABLE_I18N */ 438 static void build_upper_buffer (re_string_t *pstr) internal_function; 439 static void re_string_translate_buffer (re_string_t *pstr) internal_function; 440 static unsigned int re_string_context_at (const re_string_t *input, Idx idx, 441 int eflags) 442 internal_function __attribute ((pure)); 443 #define re_string_peek_byte(pstr, offset) \ 444 ((pstr)->mbs[(pstr)->cur_idx + offset]) 445 #define re_string_fetch_byte(pstr) \ 446 ((pstr)->mbs[(pstr)->cur_idx++]) 447 #define re_string_first_byte(pstr, idx) \ 448 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) 449 #define re_string_is_single_byte_char(pstr, idx) \ 450 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ 451 || (pstr)->wcs[(idx) + 1] != WEOF)) 452 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) 453 #define re_string_cur_idx(pstr) ((pstr)->cur_idx) 454 #define re_string_get_buffer(pstr) ((pstr)->mbs) 455 #define re_string_length(pstr) ((pstr)->len) 456 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) 457 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) 458 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) 459 460 #include <alloca.h> 461 462 #ifndef _LIBC 463 # if HAVE_ALLOCA 464 /* The OS usually guarantees only one guard page at the bottom of the stack, 465 and a page size can be as small as 4096 bytes. So we cannot safely 466 allocate anything larger than 4096 bytes. Also care for the possibility 467 of a few compiler-allocated temporary stack slots. */ 468 # define __libc_use_alloca(n) ((n) < 4032) 469 # else 470 /* alloca is implemented with malloc, so just use malloc. */ 471 # define __libc_use_alloca(n) 0 472 # endif 473 #endif 474 475 #ifndef MAX 476 # define MAX(a,b) ((a) < (b) ? (b) : (a)) 477 #endif 478 479 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) 480 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) 481 #define re_free(p) free (p) 482 483 struct bin_tree_t 484 { 485 struct bin_tree_t *parent; 486 struct bin_tree_t *left; 487 struct bin_tree_t *right; 488 struct bin_tree_t *first; 489 struct bin_tree_t *next; 490 491 re_token_t token; 492 493 /* `node_idx' is the index in dfa->nodes, if `type' == 0. 494 Otherwise `type' indicate the type of this node. */ 495 Idx node_idx; 496 }; 497 typedef struct bin_tree_t bin_tree_t; 498 499 #define BIN_TREE_STORAGE_SIZE \ 500 ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) 501 502 struct bin_tree_storage_t 503 { 504 struct bin_tree_storage_t *next; 505 bin_tree_t data[BIN_TREE_STORAGE_SIZE]; 506 }; 507 typedef struct bin_tree_storage_t bin_tree_storage_t; 508 509 #define CONTEXT_WORD 1 510 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) 511 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) 512 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) 513 514 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) 515 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) 516 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) 517 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) 518 #define IS_ORDINARY_CONTEXT(c) ((c) == 0) 519 520 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') 521 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) 522 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_') 523 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) 524 525 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ 526 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 527 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 528 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ 529 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) 530 531 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ 532 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 533 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 534 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ 535 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) 536 537 struct re_dfastate_t 538 { 539 re_hashval_t hash; 540 re_node_set nodes; 541 re_node_set non_eps_nodes; 542 re_node_set inveclosure; 543 re_node_set *entrance_nodes; 544 struct re_dfastate_t **trtable, **word_trtable; 545 unsigned int context : 4; 546 unsigned int halt : 1; 547 /* If this state can accept `multi byte'. 548 Note that we refer to multibyte characters, and multi character 549 collating elements as `multi byte'. */ 550 unsigned int accept_mb : 1; 551 /* If this state has backreference node(s). */ 552 unsigned int has_backref : 1; 553 unsigned int has_constraint : 1; 554 }; 555 typedef struct re_dfastate_t re_dfastate_t; 556 557 struct re_state_table_entry 558 { 559 Idx num; 560 Idx alloc; 561 re_dfastate_t **array; 562 }; 563 564 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ 565 566 typedef struct 567 { 568 Idx next_idx; 569 Idx alloc; 570 re_dfastate_t **array; 571 } state_array_t; 572 573 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ 574 575 typedef struct 576 { 577 Idx node; 578 Idx str_idx; /* The position NODE match at. */ 579 state_array_t path; 580 } re_sub_match_last_t; 581 582 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. 583 And information about the node, whose type is OP_CLOSE_SUBEXP, 584 corresponding to NODE is stored in LASTS. */ 585 586 typedef struct 587 { 588 Idx str_idx; 589 Idx node; 590 state_array_t *path; 591 Idx alasts; /* Allocation size of LASTS. */ 592 Idx nlasts; /* The number of LASTS. */ 593 re_sub_match_last_t **lasts; 594 } re_sub_match_top_t; 595 596 struct re_backref_cache_entry 597 { 598 Idx node; 599 Idx str_idx; 600 Idx subexp_from; 601 Idx subexp_to; 602 char more; 603 char unused; 604 unsigned short int eps_reachable_subexps_map; 605 }; 606 607 typedef struct 608 { 609 /* The string object corresponding to the input string. */ 610 re_string_t input; 611 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) 612 const re_dfa_t *const dfa; 613 #else 614 const re_dfa_t *dfa; 615 #endif 616 /* EFLAGS of the argument of regexec. */ 617 int eflags; 618 /* Where the matching ends. */ 619 Idx match_last; 620 Idx last_node; 621 /* The state log used by the matcher. */ 622 re_dfastate_t **state_log; 623 Idx state_log_top; 624 /* Back reference cache. */ 625 Idx nbkref_ents; 626 Idx abkref_ents; 627 struct re_backref_cache_entry *bkref_ents; 628 int max_mb_elem_len; 629 Idx nsub_tops; 630 Idx asub_tops; 631 re_sub_match_top_t **sub_tops; 632 } re_match_context_t; 633 634 typedef struct 635 { 636 re_dfastate_t **sifted_states; 637 re_dfastate_t **limited_states; 638 Idx last_node; 639 Idx last_str_idx; 640 re_node_set limits; 641 } re_sift_context_t; 642 643 struct re_fail_stack_ent_t 644 { 645 Idx idx; 646 Idx node; 647 regmatch_t *regs; 648 re_node_set eps_via_nodes; 649 }; 650 651 struct re_fail_stack_t 652 { 653 Idx num; 654 Idx alloc; 655 struct re_fail_stack_ent_t *stack; 656 }; 657 658 struct re_dfa_t 659 { 660 re_token_t *nodes; 661 size_t nodes_alloc; 662 size_t nodes_len; 663 Idx *nexts; 664 Idx *org_indices; 665 re_node_set *edests; 666 re_node_set *eclosures; 667 re_node_set *inveclosures; 668 struct re_state_table_entry *state_table; 669 re_dfastate_t *init_state; 670 re_dfastate_t *init_state_word; 671 re_dfastate_t *init_state_nl; 672 re_dfastate_t *init_state_begbuf; 673 bin_tree_t *str_tree; 674 bin_tree_storage_t *str_tree_storage; 675 re_bitset_ptr_t sb_char; 676 int str_tree_storage_idx; 677 678 /* number of subexpressions `re_nsub' is in regex_t. */ 679 re_hashval_t state_hash_mask; 680 Idx init_node; 681 Idx nbackref; /* The number of backreference in this dfa. */ 682 683 /* Bitmap expressing which backreference is used. */ 684 bitset_word_t used_bkref_map; 685 bitset_word_t completed_bkref_map; 686 687 unsigned int has_plural_match : 1; 688 /* If this dfa has "multibyte node", which is a backreference or 689 a node which can accept multibyte character or multi character 690 collating element. */ 691 unsigned int has_mb_node : 1; 692 unsigned int is_utf8 : 1; 693 unsigned int map_notascii : 1; 694 unsigned int word_ops_used : 1; 695 int mb_cur_max; 696 bitset_t word_char; 697 reg_syntax_t syntax; 698 Idx *subexp_map; 699 #ifdef DEBUG 700 char* re_str; 701 #endif 702 #ifdef _LIBC 703 __libc_lock_define (, lock) 704 #endif 705 }; 706 707 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) 708 #define re_node_set_remove(set,id) \ 709 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) 710 #define re_node_set_empty(p) ((p)->nelem = 0) 711 #define re_node_set_free(set) re_free ((set)->elems) 712 713 714 typedef enum 715 { 716 SB_CHAR, 717 MB_CHAR, 718 EQUIV_CLASS, 719 COLL_SYM, 720 CHAR_CLASS 721 } bracket_elem_type; 722 723 typedef struct 724 { 725 bracket_elem_type type; 726 union 727 { 728 unsigned char ch; 729 unsigned char *name; 730 wchar_t wch; 731 } opr; 732 } bracket_elem_t; 733 734 735 /* Inline functions for bitset_t operation. */ 736 737 static inline void 738 bitset_set (bitset_t set, Idx i) 739 { 740 set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; 741 } 742 743 static inline void 744 bitset_clear (bitset_t set, Idx i) 745 { 746 set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); 747 } 748 749 static inline bool 750 bitset_contain (const bitset_t set, Idx i) 751 { 752 return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; 753 } 754 755 static inline void 756 bitset_empty (bitset_t set) 757 { 758 memset (set, '\0', sizeof (bitset_t)); 759 } 760 761 static inline void 762 bitset_set_all (bitset_t set) 763 { 764 memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); 765 if (SBC_MAX % BITSET_WORD_BITS != 0) 766 set[BITSET_WORDS - 1] = 767 ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; 768 } 769 770 static inline void 771 bitset_copy (bitset_t dest, const bitset_t src) 772 { 773 memcpy (dest, src, sizeof (bitset_t)); 774 } 775 776 static inline void 777 bitset_not (bitset_t set) 778 { 779 int bitset_i; 780 for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) 781 set[bitset_i] = ~set[bitset_i]; 782 if (SBC_MAX % BITSET_WORD_BITS != 0) 783 set[BITSET_WORDS - 1] = 784 ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) 785 & ~set[BITSET_WORDS - 1]); 786 } 787 788 static inline void 789 bitset_merge (bitset_t dest, const bitset_t src) 790 { 791 int bitset_i; 792 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 793 dest[bitset_i] |= src[bitset_i]; 794 } 795 796 static inline void 797 bitset_mask (bitset_t dest, const bitset_t src) 798 { 799 int bitset_i; 800 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 801 dest[bitset_i] &= src[bitset_i]; 802 } 803 804 #ifdef RE_ENABLE_I18N 805 /* Inline functions for re_string. */ 806 static inline int 807 internal_function __attribute ((pure)) 808 re_string_char_size_at (const re_string_t *pstr, Idx idx) 809 { 810 int byte_idx; 811 if (pstr->mb_cur_max == 1) 812 return 1; 813 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) 814 if (pstr->wcs[idx + byte_idx] != WEOF) 815 break; 816 return byte_idx; 817 } 818 819 static inline wint_t 820 internal_function __attribute ((pure)) 821 re_string_wchar_at (const re_string_t *pstr, Idx idx) 822 { 823 if (pstr->mb_cur_max == 1) 824 return (wint_t) pstr->mbs[idx]; 825 return (wint_t) pstr->wcs[idx]; 826 } 827 828 static int 829 internal_function __attribute ((pure)) 830 re_string_elem_size_at (const re_string_t *pstr _UNUSED_PARAMETER_, 831 Idx idx _UNUSED_PARAMETER_) 832 { 833 # ifdef _LIBC 834 const unsigned char *p, *extra; 835 const int32_t *table, *indirect; 836 int32_t tmp; 837 # include <locale/weight.h> 838 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); 839 840 if (nrules != 0) 841 { 842 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); 843 extra = (const unsigned char *) 844 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); 845 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, 846 _NL_COLLATE_INDIRECTMB); 847 p = pstr->mbs + idx; 848 tmp = findidx (&p); 849 return p - pstr->mbs - idx; 850 } 851 else 852 # endif /* _LIBC */ 853 return 1; 854 } 855 #endif /* RE_ENABLE_I18N */ 856 857 #ifndef __GNUC_PREREQ 858 # if defined __GNUC__ && defined __GNUC_MINOR__ 859 # define __GNUC_PREREQ(maj, min) \ 860 ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min)) 861 # else 862 # define __GNUC_PREREQ(maj, min) 0 863 # endif 864 #endif 865 866 #if __GNUC_PREREQ (3,4) 867 # undef __attribute_warn_unused_result__ 868 # define __attribute_warn_unused_result__ \ 869 __attribute__ ((__warn_unused_result__)) 870 #else 871 # define __attribute_warn_unused_result__ /* empty */ 872 #endif 873 874 #endif /* _REGEX_INTERNAL_H */ 875