1 /* Extended regular expression matching and search library. 2 Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc. 3 This file is part of the GNU C Library. 4 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. 5 6 This program is free software; you can redistribute it and/or modify 7 it under the terms of the GNU General Public License as published by 8 the Free Software Foundation; either version 2, or (at your option) 9 any later version. 10 11 This program is distributed in the hope that it will be useful, 12 but WITHOUT ANY WARRANTY; without even the implied warranty of 13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 14 GNU General Public License for more details. 15 16 You should have received a copy of the GNU General Public License along 17 with this program; if not, write to the Free Software Foundation, 18 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ 19 20 #ifndef _REGEX_INTERNAL_H 21 #define _REGEX_INTERNAL_H 1 22 23 #include <assert.h> 24 #include <ctype.h> 25 #include <stdbool.h> 26 #include <stdio.h> 27 #include <stdlib.h> 28 #include <string.h> 29 30 #ifdef _LIBC 31 # include <langinfo.h> 32 #else 33 # include "localcharset.h" 34 #endif 35 #if defined HAVE_LOCALE_H || defined _LIBC 36 # include <locale.h> 37 #endif 38 39 #include <wchar.h> 40 #include <wctype.h> 41 #include <stdint.h> 42 #if defined _LIBC 43 # include <bits/libc-lock.h> 44 #else 45 # define __libc_lock_init(NAME) do { } while (0) 46 # define __libc_lock_lock(NAME) do { } while (0) 47 # define __libc_lock_unlock(NAME) do { } while (0) 48 #endif 49 50 /* In case that the system doesn't have isblank(). */ 51 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) 52 # define isblank(ch) ((ch) == ' ' || (ch) == '\t') 53 #endif 54 55 #ifdef _LIBC 56 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS 57 # define _RE_DEFINE_LOCALE_FUNCTIONS 1 58 # include <locale/localeinfo.h> 59 # include <locale/elem-hash.h> 60 # include <locale/coll-lookup.h> 61 # endif 62 #endif 63 64 /* This is for other GNU distributions with internationalized messages. */ 65 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC 66 # include <libintl.h> 67 # ifdef _LIBC 68 # undef gettext 69 # define gettext(msgid) \ 70 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES) 71 # endif 72 #else 73 # define gettext(msgid) (msgid) 74 #endif 75 76 #ifndef gettext_noop 77 /* This define is so xgettext can find the internationalizable 78 strings. */ 79 # define gettext_noop(String) String 80 #endif 81 82 /* For loser systems without the definition. */ 83 #ifndef SIZE_MAX 84 # define SIZE_MAX ((size_t) -1) 85 #endif 86 87 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC 88 # define RE_ENABLE_I18N 89 #endif 90 91 #if __GNUC__ >= 3 92 # define BE(expr, val) __builtin_expect (expr, val) 93 #else 94 # define BE(expr, val) (expr) 95 # ifdef _LIBC 96 # define inline 97 # endif 98 #endif 99 100 /* Number of ASCII characters. */ 101 #define ASCII_CHARS 0x80 102 103 /* Number of single byte characters. */ 104 #define SBC_MAX (UCHAR_MAX + 1) 105 106 #define COLL_ELEM_LEN_MAX 8 107 108 /* The character which represents newline. */ 109 #define NEWLINE_CHAR '\n' 110 #define WIDE_NEWLINE_CHAR L'\n' 111 112 /* Rename to standard API for using out of glibc. */ 113 #ifndef _LIBC 114 # define __wctype wctype 115 # define __iswctype iswctype 116 # define __btowc btowc 117 # define __wcrtomb wcrtomb 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 == 0xffffffff 165 # define BITSET_WORD_BITS 32 166 #elif BITSET_WORD_MAX >> 31 >> 5 == 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 # ifdef __i386__ 421 # define internal_function __attribute ((regparm (3), stdcall)) 422 # else 423 # define internal_function 424 # endif 425 #endif 426 427 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, 428 Idx new_buf_len) 429 internal_function; 430 #ifdef RE_ENABLE_I18N 431 static void build_wcs_buffer (re_string_t *pstr) internal_function; 432 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr) 433 internal_function; 434 #endif /* RE_ENABLE_I18N */ 435 static void build_upper_buffer (re_string_t *pstr) internal_function; 436 static void re_string_translate_buffer (re_string_t *pstr) internal_function; 437 static unsigned int re_string_context_at (const re_string_t *input, Idx idx, 438 int eflags) 439 internal_function __attribute ((pure)); 440 #define re_string_peek_byte(pstr, offset) \ 441 ((pstr)->mbs[(pstr)->cur_idx + offset]) 442 #define re_string_fetch_byte(pstr) \ 443 ((pstr)->mbs[(pstr)->cur_idx++]) 444 #define re_string_first_byte(pstr, idx) \ 445 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) 446 #define re_string_is_single_byte_char(pstr, idx) \ 447 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ 448 || (pstr)->wcs[(idx) + 1] != WEOF)) 449 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) 450 #define re_string_cur_idx(pstr) ((pstr)->cur_idx) 451 #define re_string_get_buffer(pstr) ((pstr)->mbs) 452 #define re_string_length(pstr) ((pstr)->len) 453 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) 454 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) 455 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) 456 457 #include <alloca.h> 458 459 #ifndef _LIBC 460 # if HAVE_ALLOCA 461 /* The OS usually guarantees only one guard page at the bottom of the stack, 462 and a page size can be as small as 4096 bytes. So we cannot safely 463 allocate anything larger than 4096 bytes. Also care for the possibility 464 of a few compiler-allocated temporary stack slots. */ 465 # define __libc_use_alloca(n) ((n) < 4032) 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 #ifndef MAX 473 # define MAX(a,b) ((a) < (b) ? (b) : (a)) 474 #endif 475 476 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) 477 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) 478 #define re_free(p) free (p) 479 480 struct bin_tree_t 481 { 482 struct bin_tree_t *parent; 483 struct bin_tree_t *left; 484 struct bin_tree_t *right; 485 struct bin_tree_t *first; 486 struct bin_tree_t *next; 487 488 re_token_t token; 489 490 /* `node_idx' is the index in dfa->nodes, if `type' == 0. 491 Otherwise `type' indicate the type of this node. */ 492 Idx node_idx; 493 }; 494 typedef struct bin_tree_t bin_tree_t; 495 496 #define BIN_TREE_STORAGE_SIZE \ 497 ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) 498 499 struct bin_tree_storage_t 500 { 501 struct bin_tree_storage_t *next; 502 bin_tree_t data[BIN_TREE_STORAGE_SIZE]; 503 }; 504 typedef struct bin_tree_storage_t bin_tree_storage_t; 505 506 #define CONTEXT_WORD 1 507 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1) 508 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) 509 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) 510 511 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) 512 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) 513 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) 514 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) 515 #define IS_ORDINARY_CONTEXT(c) ((c) == 0) 516 517 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') 518 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) 519 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_') 520 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) 521 522 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ 523 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 524 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 525 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ 526 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) 527 528 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ 529 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 530 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 531 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ 532 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) 533 534 struct re_dfastate_t 535 { 536 re_hashval_t hash; 537 re_node_set nodes; 538 re_node_set non_eps_nodes; 539 re_node_set inveclosure; 540 re_node_set *entrance_nodes; 541 struct re_dfastate_t **trtable, **word_trtable; 542 unsigned int context : 4; 543 unsigned int halt : 1; 544 /* If this state can accept `multi byte'. 545 Note that we refer to multibyte characters, and multi character 546 collating elements as `multi byte'. */ 547 unsigned int accept_mb : 1; 548 /* If this state has backreference node(s). */ 549 unsigned int has_backref : 1; 550 unsigned int has_constraint : 1; 551 }; 552 typedef struct re_dfastate_t re_dfastate_t; 553 554 struct re_state_table_entry 555 { 556 Idx num; 557 Idx alloc; 558 re_dfastate_t **array; 559 }; 560 561 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ 562 563 typedef struct 564 { 565 Idx next_idx; 566 Idx alloc; 567 re_dfastate_t **array; 568 } state_array_t; 569 570 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ 571 572 typedef struct 573 { 574 Idx node; 575 Idx str_idx; /* The position NODE match at. */ 576 state_array_t path; 577 } re_sub_match_last_t; 578 579 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP. 580 And information about the node, whose type is OP_CLOSE_SUBEXP, 581 corresponding to NODE is stored in LASTS. */ 582 583 typedef struct 584 { 585 Idx str_idx; 586 Idx node; 587 state_array_t *path; 588 Idx alasts; /* Allocation size of LASTS. */ 589 Idx nlasts; /* The number of LASTS. */ 590 re_sub_match_last_t **lasts; 591 } re_sub_match_top_t; 592 593 struct re_backref_cache_entry 594 { 595 Idx node; 596 Idx str_idx; 597 Idx subexp_from; 598 Idx subexp_to; 599 char more; 600 char unused; 601 unsigned short int eps_reachable_subexps_map; 602 }; 603 604 typedef struct 605 { 606 /* The string object corresponding to the input string. */ 607 re_string_t input; 608 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) 609 const re_dfa_t *const dfa; 610 #else 611 const re_dfa_t *dfa; 612 #endif 613 /* EFLAGS of the argument of regexec. */ 614 int eflags; 615 /* Where the matching ends. */ 616 Idx match_last; 617 Idx last_node; 618 /* The state log used by the matcher. */ 619 re_dfastate_t **state_log; 620 Idx state_log_top; 621 /* Back reference cache. */ 622 Idx nbkref_ents; 623 Idx abkref_ents; 624 struct re_backref_cache_entry *bkref_ents; 625 int max_mb_elem_len; 626 Idx nsub_tops; 627 Idx asub_tops; 628 re_sub_match_top_t **sub_tops; 629 } re_match_context_t; 630 631 typedef struct 632 { 633 re_dfastate_t **sifted_states; 634 re_dfastate_t **limited_states; 635 Idx last_node; 636 Idx last_str_idx; 637 re_node_set limits; 638 } re_sift_context_t; 639 640 struct re_fail_stack_ent_t 641 { 642 Idx idx; 643 Idx node; 644 regmatch_t *regs; 645 re_node_set eps_via_nodes; 646 }; 647 648 struct re_fail_stack_t 649 { 650 Idx num; 651 Idx alloc; 652 struct re_fail_stack_ent_t *stack; 653 }; 654 655 struct re_dfa_t 656 { 657 re_token_t *nodes; 658 size_t nodes_alloc; 659 size_t nodes_len; 660 Idx *nexts; 661 Idx *org_indices; 662 re_node_set *edests; 663 re_node_set *eclosures; 664 re_node_set *inveclosures; 665 struct re_state_table_entry *state_table; 666 re_dfastate_t *init_state; 667 re_dfastate_t *init_state_word; 668 re_dfastate_t *init_state_nl; 669 re_dfastate_t *init_state_begbuf; 670 bin_tree_t *str_tree; 671 bin_tree_storage_t *str_tree_storage; 672 re_bitset_ptr_t sb_char; 673 int str_tree_storage_idx; 674 675 /* number of subexpressions `re_nsub' is in regex_t. */ 676 re_hashval_t state_hash_mask; 677 Idx init_node; 678 Idx nbackref; /* The number of backreference in this dfa. */ 679 680 /* Bitmap expressing which backreference is used. */ 681 bitset_word_t used_bkref_map; 682 bitset_word_t completed_bkref_map; 683 684 unsigned int has_plural_match : 1; 685 /* If this dfa has "multibyte node", which is a backreference or 686 a node which can accept multibyte character or multi character 687 collating element. */ 688 unsigned int has_mb_node : 1; 689 unsigned int is_utf8 : 1; 690 unsigned int map_notascii : 1; 691 unsigned int word_ops_used : 1; 692 int mb_cur_max; 693 bitset_t word_char; 694 reg_syntax_t syntax; 695 Idx *subexp_map; 696 #ifdef DEBUG 697 char* re_str; 698 #endif 699 #ifdef _LIBC 700 __libc_lock_define (, lock) 701 #endif 702 }; 703 704 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) 705 #define re_node_set_remove(set,id) \ 706 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) 707 #define re_node_set_empty(p) ((p)->nelem = 0) 708 #define re_node_set_free(set) re_free ((set)->elems) 709 710 711 typedef enum 712 { 713 SB_CHAR, 714 MB_CHAR, 715 EQUIV_CLASS, 716 COLL_SYM, 717 CHAR_CLASS 718 } bracket_elem_type; 719 720 typedef struct 721 { 722 bracket_elem_type type; 723 union 724 { 725 unsigned char ch; 726 unsigned char *name; 727 wchar_t wch; 728 } opr; 729 } bracket_elem_t; 730 731 732 /* Inline functions for bitset_t operation. */ 733 734 static inline void 735 bitset_set (bitset_t set, Idx i) 736 { 737 set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; 738 } 739 740 static inline void 741 bitset_clear (bitset_t set, Idx i) 742 { 743 set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); 744 } 745 746 static inline bool 747 bitset_contain (const bitset_t set, Idx i) 748 { 749 return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; 750 } 751 752 static inline void 753 bitset_empty (bitset_t set) 754 { 755 memset (set, '\0', sizeof (bitset_t)); 756 } 757 758 static inline void 759 bitset_set_all (bitset_t set) 760 { 761 memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); 762 if (SBC_MAX % BITSET_WORD_BITS != 0) 763 set[BITSET_WORDS - 1] = 764 ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; 765 } 766 767 static inline void 768 bitset_copy (bitset_t dest, const bitset_t src) 769 { 770 memcpy (dest, src, sizeof (bitset_t)); 771 } 772 773 static inline void 774 bitset_not (bitset_t set) 775 { 776 int bitset_i; 777 for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) 778 set[bitset_i] = ~set[bitset_i]; 779 if (SBC_MAX % BITSET_WORD_BITS != 0) 780 set[BITSET_WORDS - 1] = 781 ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) 782 & ~set[BITSET_WORDS - 1]); 783 } 784 785 static inline void 786 bitset_merge (bitset_t dest, const bitset_t src) 787 { 788 int bitset_i; 789 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 790 dest[bitset_i] |= src[bitset_i]; 791 } 792 793 static inline void 794 bitset_mask (bitset_t dest, const bitset_t src) 795 { 796 int bitset_i; 797 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 798 dest[bitset_i] &= src[bitset_i]; 799 } 800 801 #ifdef RE_ENABLE_I18N 802 /* Inline functions for re_string. */ 803 static inline int 804 internal_function __attribute ((pure)) 805 re_string_char_size_at (const re_string_t *pstr, Idx idx) 806 { 807 int byte_idx; 808 if (pstr->mb_cur_max == 1) 809 return 1; 810 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) 811 if (pstr->wcs[idx + byte_idx] != WEOF) 812 break; 813 return byte_idx; 814 } 815 816 static inline wint_t 817 internal_function __attribute ((pure)) 818 re_string_wchar_at (const re_string_t *pstr, Idx idx) 819 { 820 if (pstr->mb_cur_max == 1) 821 return (wint_t) pstr->mbs[idx]; 822 return (wint_t) pstr->wcs[idx]; 823 } 824 825 static int 826 internal_function __attribute ((pure)) 827 re_string_elem_size_at (const re_string_t *pstr, Idx idx) 828 { 829 # ifdef _LIBC 830 const unsigned char *p, *extra; 831 const int32_t *table, *indirect; 832 int32_t tmp; 833 # include <locale/weight.h> 834 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); 835 836 if (nrules != 0) 837 { 838 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); 839 extra = (const unsigned char *) 840 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); 841 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, 842 _NL_COLLATE_INDIRECTMB); 843 p = pstr->mbs + idx; 844 tmp = findidx (&p); 845 return p - pstr->mbs - idx; 846 } 847 else 848 # endif /* _LIBC */ 849 return 1; 850 } 851 #endif /* RE_ENABLE_I18N */ 852 853 #endif /* _REGEX_INTERNAL_H */