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 && !HAVE_DECL_ISBLANK && !defined 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 128typedef __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. */ 151typedef __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. */ 155typedef 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#elif BITSET_WORD_MAX == (0xffffffff + 2) * 0xffffffff 186/* Work around a bug in 64-bit PGC (before version 6.1-2), where the 187 preprocessor mishandles large unsigned values as if they were signed. */ 188# define BITSET_WORD_BITS 64 189#else 190# error "Add case for new bitset_word_t size" 191#endif 192 193/* Number of bitset_word_t values in a bitset_t. */ 194#define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) 195 196typedef bitset_word_t bitset_t[BITSET_WORDS]; 197typedef bitset_word_t *re_bitset_ptr_t; 198typedef const bitset_word_t *re_const_bitset_ptr_t; 199 200#define PREV_WORD_CONSTRAINT 0x0001 201#define PREV_NOTWORD_CONSTRAINT 0x0002 202#define NEXT_WORD_CONSTRAINT 0x0004 203#define NEXT_NOTWORD_CONSTRAINT 0x0008 204#define PREV_NEWLINE_CONSTRAINT 0x0010 205#define NEXT_NEWLINE_CONSTRAINT 0x0020 206#define PREV_BEGBUF_CONSTRAINT 0x0040 207#define NEXT_ENDBUF_CONSTRAINT 0x0080 208#define WORD_DELIM_CONSTRAINT 0x0100 209#define NOT_WORD_DELIM_CONSTRAINT 0x0200 210 211typedef enum 212{ 213 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, 214 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, 215 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, 216 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, 217 LINE_FIRST = PREV_NEWLINE_CONSTRAINT, 218 LINE_LAST = NEXT_NEWLINE_CONSTRAINT, 219 BUF_FIRST = PREV_BEGBUF_CONSTRAINT, 220 BUF_LAST = NEXT_ENDBUF_CONSTRAINT, 221 WORD_DELIM = WORD_DELIM_CONSTRAINT, 222 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT 223} re_context_type; 224 225typedef struct 226{ 227 Idx alloc; 228 Idx nelem; 229 Idx *elems; 230} re_node_set; 231 232typedef enum 233{ 234 NON_TYPE = 0, 235 236 /* Node type, These are used by token, node, tree. */ 237 CHARACTER = 1, 238 END_OF_RE = 2, 239 SIMPLE_BRACKET = 3, 240 OP_BACK_REF = 4, 241 OP_PERIOD = 5, 242#ifdef RE_ENABLE_I18N 243 COMPLEX_BRACKET = 6, 244 OP_UTF8_PERIOD = 7, 245#endif /* RE_ENABLE_I18N */ 246 247 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used 248 when the debugger shows values of this enum type. */ 249#define EPSILON_BIT 8 250 OP_OPEN_SUBEXP = EPSILON_BIT | 0, 251 OP_CLOSE_SUBEXP = EPSILON_BIT | 1, 252 OP_ALT = EPSILON_BIT | 2, 253 OP_DUP_ASTERISK = EPSILON_BIT | 3, 254 ANCHOR = EPSILON_BIT | 4, 255 256 /* Tree type, these are used only by tree. */ 257 CONCAT = 16, 258 SUBEXP = 17, 259 260 /* Token type, these are used only by token. */ 261 OP_DUP_PLUS = 18, 262 OP_DUP_QUESTION, 263 OP_OPEN_BRACKET, 264 OP_CLOSE_BRACKET, 265 OP_CHARSET_RANGE, 266 OP_OPEN_DUP_NUM, 267 OP_CLOSE_DUP_NUM, 268 OP_NON_MATCH_LIST, 269 OP_OPEN_COLL_ELEM, 270 OP_CLOSE_COLL_ELEM, 271 OP_OPEN_EQUIV_CLASS, 272 OP_CLOSE_EQUIV_CLASS, 273 OP_OPEN_CHAR_CLASS, 274 OP_CLOSE_CHAR_CLASS, 275 OP_WORD, 276 OP_NOTWORD, 277 OP_SPACE, 278 OP_NOTSPACE, 279 BACK_SLASH 280 281} re_token_type_t; 282 283#ifdef RE_ENABLE_I18N 284typedef struct 285{ 286 /* Multibyte characters. */ 287 wchar_t *mbchars; 288 289 /* Collating symbols. */ 290# ifdef _LIBC 291 int32_t *coll_syms; 292# endif 293 294 /* Equivalence classes. */ 295# ifdef _LIBC 296 int32_t *equiv_classes; 297# endif 298 299 /* Range expressions. */ 300# ifdef _LIBC 301 uint32_t *range_starts; 302 uint32_t *range_ends; 303# else /* not _LIBC */ 304 wchar_t *range_starts; 305 wchar_t *range_ends; 306# endif /* not _LIBC */ 307 308 /* Character classes. */ 309 wctype_t *char_classes; 310 311 /* If this character set is the non-matching list. */ 312 unsigned int non_match : 1; 313 314 /* # of multibyte characters. */ 315 Idx nmbchars; 316 317 /* # of collating symbols. */ 318 Idx ncoll_syms; 319 320 /* # of equivalence classes. */ 321 Idx nequiv_classes; 322 323 /* # of range expressions. */ 324 Idx nranges; 325 326 /* # of character classes. */ 327 Idx nchar_classes; 328} re_charset_t; 329#endif /* RE_ENABLE_I18N */ 330 331typedef struct 332{ 333 union 334 { 335 unsigned char c; /* for CHARACTER */ 336 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ 337#ifdef RE_ENABLE_I18N 338 re_charset_t *mbcset; /* for COMPLEX_BRACKET */ 339#endif /* RE_ENABLE_I18N */ 340 Idx idx; /* for BACK_REF */ 341 re_context_type ctx_type; /* for ANCHOR */ 342 } opr; 343#if __GNUC__ >= 2 && !__STRICT_ANSI__ 344 re_token_type_t type : 8; 345#else 346 re_token_type_t type; 347#endif 348 unsigned int constraint : 10; /* context constraint */ 349 unsigned int duplicated : 1; 350 unsigned int opt_subexp : 1; 351#ifdef RE_ENABLE_I18N 352 unsigned int accept_mb : 1; 353 /* These 2 bits can be moved into the union if needed (e.g. if running out 354 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ 355 unsigned int mb_partial : 1; 356#endif 357 unsigned int word_char : 1; 358} re_token_t; 359 360#define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) 361 362struct re_string_t 363{ 364 /* Indicate the raw buffer which is the original string passed as an 365 argument of regexec(), re_search(), etc.. */ 366 const unsigned char *raw_mbs; 367 /* Store the multibyte string. In case of "case insensitive mode" like 368 REG_ICASE, upper cases of the string are stored, otherwise MBS points 369 the same address that RAW_MBS points. */ 370 unsigned char *mbs; 371#ifdef RE_ENABLE_I18N 372 /* Store the wide character string which is corresponding to MBS. */ 373 wint_t *wcs; 374 Idx *offsets; 375 mbstate_t cur_state; 376#endif 377 /* Index in RAW_MBS. Each character mbs[i] corresponds to 378 raw_mbs[raw_mbs_idx + i]. */ 379 Idx raw_mbs_idx; 380 /* The length of the valid characters in the buffers. */ 381 Idx valid_len; 382 /* The corresponding number of bytes in raw_mbs array. */ 383 Idx valid_raw_len; 384 /* The length of the buffers MBS and WCS. */ 385 Idx bufs_len; 386 /* The index in MBS, which is updated by re_string_fetch_byte. */ 387 Idx cur_idx; 388 /* length of RAW_MBS array. */ 389 Idx raw_len; 390 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ 391 Idx len; 392 /* End of the buffer may be shorter than its length in the cases such 393 as re_match_2, re_search_2. Then, we use STOP for end of the buffer 394 instead of LEN. */ 395 Idx raw_stop; 396 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ 397 Idx stop; 398 399 /* The context of mbs[0]. We store the context independently, since 400 the context of mbs[0] may be different from raw_mbs[0], which is 401 the beginning of the input string. */ 402 unsigned int tip_context; 403 /* The translation passed as a part of an argument of re_compile_pattern. */ 404 RE_TRANSLATE_TYPE trans; 405 /* Copy of re_dfa_t's word_char. */ 406 re_const_bitset_ptr_t word_char; 407 /* true if REG_ICASE. */ 408 unsigned char icase; 409 unsigned char is_utf8; 410 unsigned char map_notascii; 411 unsigned char mbs_allocated; 412 unsigned char offsets_needed; 413 unsigned char newline_anchor; 414 unsigned char word_ops_used; 415 int mb_cur_max; 416}; 417typedef struct re_string_t re_string_t; 418 419 420struct re_dfa_t; 421typedef struct re_dfa_t re_dfa_t; 422 423#ifndef _LIBC 424# ifdef __i386__ 425# define internal_function __attribute ((regparm (3), stdcall)) 426# else 427# define internal_function 428# endif 429#endif 430 431static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, 432 Idx new_buf_len) 433 internal_function; 434#ifdef RE_ENABLE_I18N 435static void build_wcs_buffer (re_string_t *pstr) internal_function; 436static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr) 437 internal_function; 438#endif /* RE_ENABLE_I18N */ 439static void build_upper_buffer (re_string_t *pstr) internal_function; 440static void re_string_translate_buffer (re_string_t *pstr) internal_function; 441static unsigned int re_string_context_at (const re_string_t *input, Idx idx, 442 int eflags) 443 internal_function __attribute ((pure)); 444#define re_string_peek_byte(pstr, offset) \ 445 ((pstr)->mbs[(pstr)->cur_idx + offset]) 446#define re_string_fetch_byte(pstr) \ 447 ((pstr)->mbs[(pstr)->cur_idx++]) 448#define re_string_first_byte(pstr, idx) \ 449 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) 450#define re_string_is_single_byte_char(pstr, idx) \ 451 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ 452 || (pstr)->wcs[(idx) + 1] != WEOF)) 453#define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) 454#define re_string_cur_idx(pstr) ((pstr)->cur_idx) 455#define re_string_get_buffer(pstr) ((pstr)->mbs) 456#define re_string_length(pstr) ((pstr)->len) 457#define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) 458#define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) 459#define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) 460 461#include <alloca.h> 462 463#ifndef _LIBC 464# if HAVE_ALLOCA 465/* The OS usually guarantees only one guard page at the bottom of the stack, 466 and a page size can be as small as 4096 bytes. So we cannot safely 467 allocate anything larger than 4096 bytes. Also care for the possibility 468 of a few compiler-allocated temporary stack slots. */ 469# define __libc_use_alloca(n) ((n) < 4032) 470# else 471/* alloca is implemented with malloc, so just use malloc. */ 472# define __libc_use_alloca(n) 0 473# endif 474#endif 475 476#ifndef MAX 477# define MAX(a,b) ((a) < (b) ? (b) : (a)) 478#endif 479 480#define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) 481#define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) 482#define re_free(p) free (p) 483 484struct bin_tree_t 485{ 486 struct bin_tree_t *parent; 487 struct bin_tree_t *left; 488 struct bin_tree_t *right; 489 struct bin_tree_t *first; 490 struct bin_tree_t *next; 491 492 re_token_t token; 493 494 /* `node_idx' is the index in dfa->nodes, if `type' == 0. 495 Otherwise `type' indicate the type of this node. */ 496 Idx node_idx; 497}; 498typedef struct bin_tree_t bin_tree_t; 499 500#define BIN_TREE_STORAGE_SIZE \ 501 ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) 502 503struct bin_tree_storage_t 504{ 505 struct bin_tree_storage_t *next; 506 bin_tree_t data[BIN_TREE_STORAGE_SIZE]; 507}; 508typedef struct bin_tree_storage_t bin_tree_storage_t; 509 510#define CONTEXT_WORD 1 511#define CONTEXT_NEWLINE (CONTEXT_WORD << 1) 512#define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) 513#define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) 514 515#define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) 516#define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) 517#define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) 518#define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) 519#define IS_ORDINARY_CONTEXT(c) ((c) == 0) 520 521#define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') 522#define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) 523#define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_') 524#define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) 525 526#define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ 527 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 528 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 529 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ 530 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) 531 532#define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ 533 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 534 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 535 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ 536 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) 537 538struct re_dfastate_t 539{ 540 re_hashval_t hash; 541 re_node_set nodes; 542 re_node_set non_eps_nodes; 543 re_node_set inveclosure; 544 re_node_set *entrance_nodes; 545 struct re_dfastate_t **trtable, **word_trtable; 546 unsigned int context : 4; 547 unsigned int halt : 1; 548 /* If this state can accept `multi byte'. 549 Note that we refer to multibyte characters, and multi character 550 collating elements as `multi byte'. */ 551 unsigned int accept_mb : 1; 552 /* If this state has backreference node(s). */ 553 unsigned int has_backref : 1; 554 unsigned int has_constraint : 1; 555}; 556typedef struct re_dfastate_t re_dfastate_t; 557 558struct re_state_table_entry 559{ 560 Idx num; 561 Idx alloc; 562 re_dfastate_t **array; 563}; 564 565/* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ 566 567typedef struct 568{ 569 Idx next_idx; 570 Idx alloc; 571 re_dfastate_t **array; 572} state_array_t; 573 574/* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ 575 576typedef struct 577{ 578 Idx node; 579 Idx str_idx; /* The position NODE match at. */ 580 state_array_t path; 581} re_sub_match_last_t; 582 583/* Store information about the node NODE whose type is OP_OPEN_SUBEXP. 584 And information about the node, whose type is OP_CLOSE_SUBEXP, 585 corresponding to NODE is stored in LASTS. */ 586 587typedef struct 588{ 589 Idx str_idx; 590 Idx node; 591 state_array_t *path; 592 Idx alasts; /* Allocation size of LASTS. */ 593 Idx nlasts; /* The number of LASTS. */ 594 re_sub_match_last_t **lasts; 595} re_sub_match_top_t; 596 597struct re_backref_cache_entry 598{ 599 Idx node; 600 Idx str_idx; 601 Idx subexp_from; 602 Idx subexp_to; 603 char more; 604 char unused; 605 unsigned short int eps_reachable_subexps_map; 606}; 607 608typedef struct 609{ 610 /* The string object corresponding to the input string. */ 611 re_string_t input; 612#if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) 613 const re_dfa_t *const dfa; 614#else 615 const re_dfa_t *dfa; 616#endif 617 /* EFLAGS of the argument of regexec. */ 618 int eflags; 619 /* Where the matching ends. */ 620 Idx match_last; 621 Idx last_node; 622 /* The state log used by the matcher. */ 623 re_dfastate_t **state_log; 624 Idx state_log_top; 625 /* Back reference cache. */ 626 Idx nbkref_ents; 627 Idx abkref_ents; 628 struct re_backref_cache_entry *bkref_ents; 629 int max_mb_elem_len; 630 Idx nsub_tops; 631 Idx asub_tops; 632 re_sub_match_top_t **sub_tops; 633} re_match_context_t; 634 635typedef struct 636{ 637 re_dfastate_t **sifted_states; 638 re_dfastate_t **limited_states; 639 Idx last_node; 640 Idx last_str_idx; 641 re_node_set limits; 642} re_sift_context_t; 643 644struct re_fail_stack_ent_t 645{ 646 Idx idx; 647 Idx node; 648 regmatch_t *regs; 649 re_node_set eps_via_nodes; 650}; 651 652struct re_fail_stack_t 653{ 654 Idx num; 655 Idx alloc; 656 struct re_fail_stack_ent_t *stack; 657}; 658 659struct re_dfa_t 660{ 661 re_token_t *nodes; 662 size_t nodes_alloc; 663 size_t nodes_len; 664 Idx *nexts; 665 Idx *org_indices; 666 re_node_set *edests; 667 re_node_set *eclosures; 668 re_node_set *inveclosures; 669 struct re_state_table_entry *state_table; 670 re_dfastate_t *init_state; 671 re_dfastate_t *init_state_word; 672 re_dfastate_t *init_state_nl; 673 re_dfastate_t *init_state_begbuf; 674 bin_tree_t *str_tree; 675 bin_tree_storage_t *str_tree_storage; 676 re_bitset_ptr_t sb_char; 677 int str_tree_storage_idx; 678 679 /* number of subexpressions `re_nsub' is in regex_t. */ 680 re_hashval_t state_hash_mask; 681 Idx init_node; 682 Idx nbackref; /* The number of backreference in this dfa. */ 683 684 /* Bitmap expressing which backreference is used. */ 685 bitset_word_t used_bkref_map; 686 bitset_word_t completed_bkref_map; 687 688 unsigned int has_plural_match : 1; 689 /* If this dfa has "multibyte node", which is a backreference or 690 a node which can accept multibyte character or multi character 691 collating element. */ 692 unsigned int has_mb_node : 1; 693 unsigned int is_utf8 : 1; 694 unsigned int map_notascii : 1; 695 unsigned int word_ops_used : 1; 696 int mb_cur_max; 697 bitset_t word_char; 698 reg_syntax_t syntax; 699 Idx *subexp_map; 700#ifdef DEBUG 701 char* re_str; 702#endif 703#ifdef _LIBC 704 __libc_lock_define (, lock) 705#endif 706}; 707 708#define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) 709#define re_node_set_remove(set,id) \ 710 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) 711#define re_node_set_empty(p) ((p)->nelem = 0) 712#define re_node_set_free(set) re_free ((set)->elems) 713 714 715typedef enum 716{ 717 SB_CHAR, 718 MB_CHAR, 719 EQUIV_CLASS, 720 COLL_SYM, 721 CHAR_CLASS 722} bracket_elem_type; 723 724typedef struct 725{ 726 bracket_elem_type type; 727 union 728 { 729 unsigned char ch; 730 unsigned char *name; 731 wchar_t wch; 732 } opr; 733} bracket_elem_t; 734 735 736/* Inline functions for bitset_t operation. */ 737 738static inline void 739bitset_set (bitset_t set, Idx i) 740{ 741 set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; 742} 743 744static inline void 745bitset_clear (bitset_t set, Idx i) 746{ 747 set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); 748} 749 750static inline bool 751bitset_contain (const bitset_t set, Idx i) 752{ 753 return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; 754} 755 756static inline void 757bitset_empty (bitset_t set) 758{ 759 memset (set, '\0', sizeof (bitset_t)); 760} 761 762static inline void 763bitset_set_all (bitset_t set) 764{ 765 memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); 766 if (SBC_MAX % BITSET_WORD_BITS != 0) 767 set[BITSET_WORDS - 1] = 768 ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; 769} 770 771static inline void 772bitset_copy (bitset_t dest, const bitset_t src) 773{ 774 memcpy (dest, src, sizeof (bitset_t)); 775} 776 777static inline void 778bitset_not (bitset_t set) 779{ 780 int bitset_i; 781 for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) 782 set[bitset_i] = ~set[bitset_i]; 783 if (SBC_MAX % BITSET_WORD_BITS != 0) 784 set[BITSET_WORDS - 1] = 785 ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) 786 & ~set[BITSET_WORDS - 1]); 787} 788 789static inline void 790bitset_merge (bitset_t dest, const bitset_t src) 791{ 792 int bitset_i; 793 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 794 dest[bitset_i] |= src[bitset_i]; 795} 796 797static inline void 798bitset_mask (bitset_t dest, const bitset_t src) 799{ 800 int bitset_i; 801 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 802 dest[bitset_i] &= src[bitset_i]; 803} 804 805#ifdef RE_ENABLE_I18N 806/* Inline functions for re_string. */ 807static inline int 808internal_function __attribute ((pure)) 809re_string_char_size_at (const re_string_t *pstr, Idx idx) 810{ 811 int byte_idx; 812 if (pstr->mb_cur_max == 1) 813 return 1; 814 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) 815 if (pstr->wcs[idx + byte_idx] != WEOF) 816 break; 817 return byte_idx; 818} 819 820static inline wint_t 821internal_function __attribute ((pure)) 822re_string_wchar_at (const re_string_t *pstr, Idx idx) 823{ 824 if (pstr->mb_cur_max == 1) 825 return (wint_t) pstr->mbs[idx]; 826 return (wint_t) pstr->wcs[idx]; 827} 828 829static int 830internal_function __attribute ((pure)) 831re_string_elem_size_at (const re_string_t *pstr, Idx idx) 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#endif /* _REGEX_INTERNAL_H */ 858