regex_internal.h revision 251437
1146040Stjr/* Extended regular expression matching and search library. 2250724Sjkim Copyright (C) 2002-2012 Free Software Foundation, Inc. 3146040Stjr This file is part of the GNU C Library. 4146040Stjr Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. 5146040Stjr 6146040Stjr The GNU C Library is free software; you can redistribute it and/or 7146040Stjr modify it under the terms of the GNU Lesser General Public 8146040Stjr License as published by the Free Software Foundation; either 9146040Stjr version 2.1 of the License, or (at your option) any later version. 10146040Stjr 11146040Stjr The GNU C Library is distributed in the hope that it will be useful, 12146040Stjr but WITHOUT ANY WARRANTY; without even the implied warranty of 13146040Stjr MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 14146040Stjr Lesser General Public License for more details. 15146040Stjr 16146040Stjr You should have received a copy of the GNU Lesser General Public 17250724Sjkim License along with the GNU C Library; if not, see 18250724Sjkim <http://www.gnu.org/licenses/>. */ 19146040Stjr 20146040Stjr#ifndef _REGEX_INTERNAL_H 21146040Stjr#define _REGEX_INTERNAL_H 1 22146040Stjr 23146040Stjr#include <assert.h> 24146040Stjr#include <ctype.h> 25146040Stjr#include <stdio.h> 26146040Stjr#include <stdlib.h> 27146040Stjr#include <string.h> 28146040Stjr 29146040Stjr#if defined HAVE_LANGINFO_H || defined HAVE_LANGINFO_CODESET || defined _LIBC 30146040Stjr# include <langinfo.h> 31146040Stjr#endif 32146040Stjr#if defined HAVE_LOCALE_H || defined _LIBC 33146040Stjr# include <locale.h> 34146040Stjr#endif 35146040Stjr#if defined HAVE_WCHAR_H || defined _LIBC 36146040Stjr# include <wchar.h> 37146040Stjr#endif /* HAVE_WCHAR_H || _LIBC */ 38146040Stjr#if defined HAVE_WCTYPE_H || defined _LIBC 39146040Stjr# include <wctype.h> 40146040Stjr#endif /* HAVE_WCTYPE_H || _LIBC */ 41250724Sjkim#if defined HAVE_STDBOOL_H || defined _LIBC 42250724Sjkim# include <stdbool.h> 43250724Sjkim#endif /* HAVE_STDBOOL_H || _LIBC */ 44250724Sjkim#if defined HAVE_STDINT_H || defined _LIBC 45250724Sjkim# include <stdint.h> 46250724Sjkim#endif /* HAVE_STDINT_H || _LIBC */ 47250724Sjkim#if defined _LIBC 48250724Sjkim# include <bits/libc-lock.h> 49250724Sjkim#else 50250724Sjkim# define __libc_lock_define(CLASS,NAME) 51250724Sjkim# define __libc_lock_init(NAME) do { } while (0) 52250724Sjkim# define __libc_lock_lock(NAME) do { } while (0) 53250724Sjkim# define __libc_lock_unlock(NAME) do { } while (0) 54250724Sjkim#endif 55146040Stjr 56146040Stjr/* In case that the system doesn't have isblank(). */ 57146040Stjr#if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank 58146040Stjr# define isblank(ch) ((ch) == ' ' || (ch) == '\t') 59146040Stjr#endif 60146040Stjr 61146040Stjr#ifdef _LIBC 62146040Stjr# ifndef _RE_DEFINE_LOCALE_FUNCTIONS 63146040Stjr# define _RE_DEFINE_LOCALE_FUNCTIONS 1 64146040Stjr# include <locale/localeinfo.h> 65146040Stjr# include <locale/elem-hash.h> 66146040Stjr# include <locale/coll-lookup.h> 67146040Stjr# endif 68146040Stjr#endif 69146040Stjr 70146040Stjr/* This is for other GNU distributions with internationalized messages. */ 71146040Stjr#if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC 72146040Stjr# include <libintl.h> 73146040Stjr# ifdef _LIBC 74146040Stjr# undef gettext 75146040Stjr# define gettext(msgid) \ 76250724Sjkim __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES) 77146040Stjr# endif 78146040Stjr#else 79146040Stjr# define gettext(msgid) (msgid) 80146040Stjr#endif 81146040Stjr 82146040Stjr#ifndef gettext_noop 83146040Stjr/* This define is so xgettext can find the internationalizable 84146040Stjr strings. */ 85146040Stjr# define gettext_noop(String) String 86146040Stjr#endif 87146040Stjr 88250724Sjkim/* For loser systems without the definition. */ 89250724Sjkim#ifndef SIZE_MAX 90250724Sjkim# define SIZE_MAX ((size_t) -1) 91250724Sjkim#endif 92250724Sjkim 93146040Stjr#if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC 94146040Stjr# define RE_ENABLE_I18N 95146040Stjr#endif 96146040Stjr 97146040Stjr#if __GNUC__ >= 3 98146040Stjr# define BE(expr, val) __builtin_expect (expr, val) 99146040Stjr#else 100146040Stjr# define BE(expr, val) (expr) 101146040Stjr# define inline 102146040Stjr#endif 103146040Stjr 104146040Stjr/* Number of single byte character. */ 105146040Stjr#define SBC_MAX 256 106146040Stjr 107146040Stjr#define COLL_ELEM_LEN_MAX 8 108146040Stjr 109146040Stjr/* The character which represents newline. */ 110146040Stjr#define NEWLINE_CHAR '\n' 111146040Stjr#define WIDE_NEWLINE_CHAR L'\n' 112146040Stjr 113146040Stjr/* Rename to standard API for using out of glibc. */ 114146040Stjr#ifndef _LIBC 115146040Stjr# define __wctype wctype 116146040Stjr# define __iswctype iswctype 117146040Stjr# define __btowc btowc 118250724Sjkim# define __mbrtowc mbrtowc 119146040Stjr# define __mempcpy mempcpy 120146040Stjr# define __wcrtomb wcrtomb 121146040Stjr# define __regfree regfree 122146040Stjr# define attribute_hidden 123146040Stjr#endif /* not _LIBC */ 124146040Stjr 125146040Stjr#ifdef __GNUC__ 126146040Stjr# define __attribute(arg) __attribute__ (arg) 127146040Stjr#else 128146040Stjr# define __attribute(arg) 129146040Stjr#endif 130146040Stjr 131146040Stjrextern const char __re_error_msgid[] attribute_hidden; 132146040Stjrextern const size_t __re_error_msgid_idx[] attribute_hidden; 133146040Stjr 134250724Sjkim/* An integer used to represent a set of bits. It must be unsigned, 135250724Sjkim and must be at least as wide as unsigned int. */ 136250724Sjkimtypedef unsigned long int bitset_word_t; 137250724Sjkim/* All bits set in a bitset_word_t. */ 138250724Sjkim#define BITSET_WORD_MAX ULONG_MAX 139250724Sjkim/* Number of bits in a bitset_word_t. */ 140250724Sjkim#define BITSET_WORD_BITS (sizeof (bitset_word_t) * CHAR_BIT) 141250724Sjkim/* Number of bitset_word_t in a bit_set. */ 142250724Sjkim#define BITSET_WORDS (SBC_MAX / BITSET_WORD_BITS) 143250724Sjkimtypedef bitset_word_t bitset_t[BITSET_WORDS]; 144250724Sjkimtypedef bitset_word_t *re_bitset_ptr_t; 145250724Sjkimtypedef const bitset_word_t *re_const_bitset_ptr_t; 146146040Stjr 147250724Sjkim#define bitset_set(set,i) \ 148250724Sjkim (set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS) 149250724Sjkim#define bitset_clear(set,i) \ 150250724Sjkim (set[i / BITSET_WORD_BITS] &= ~((bitset_word_t) 1 << i % BITSET_WORD_BITS)) 151250724Sjkim#define bitset_contain(set,i) \ 152250724Sjkim (set[i / BITSET_WORD_BITS] & ((bitset_word_t) 1 << i % BITSET_WORD_BITS)) 153250724Sjkim#define bitset_empty(set) memset (set, '\0', sizeof (bitset_t)) 154250724Sjkim#define bitset_set_all(set) memset (set, '\xff', sizeof (bitset_t)) 155250724Sjkim#define bitset_copy(dest,src) memcpy (dest, src, sizeof (bitset_t)) 156146040Stjr 157146040Stjr#define PREV_WORD_CONSTRAINT 0x0001 158146040Stjr#define PREV_NOTWORD_CONSTRAINT 0x0002 159146040Stjr#define NEXT_WORD_CONSTRAINT 0x0004 160146040Stjr#define NEXT_NOTWORD_CONSTRAINT 0x0008 161146040Stjr#define PREV_NEWLINE_CONSTRAINT 0x0010 162146040Stjr#define NEXT_NEWLINE_CONSTRAINT 0x0020 163146040Stjr#define PREV_BEGBUF_CONSTRAINT 0x0040 164146040Stjr#define NEXT_ENDBUF_CONSTRAINT 0x0080 165146040Stjr#define WORD_DELIM_CONSTRAINT 0x0100 166146040Stjr#define NOT_WORD_DELIM_CONSTRAINT 0x0200 167146040Stjr 168146040Stjrtypedef enum 169146040Stjr{ 170146040Stjr INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, 171146040Stjr WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, 172146040Stjr WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, 173146040Stjr INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, 174146040Stjr LINE_FIRST = PREV_NEWLINE_CONSTRAINT, 175146040Stjr LINE_LAST = NEXT_NEWLINE_CONSTRAINT, 176146040Stjr BUF_FIRST = PREV_BEGBUF_CONSTRAINT, 177146040Stjr BUF_LAST = NEXT_ENDBUF_CONSTRAINT, 178146040Stjr WORD_DELIM = WORD_DELIM_CONSTRAINT, 179146040Stjr NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT 180146040Stjr} re_context_type; 181146040Stjr 182146040Stjrtypedef struct 183146040Stjr{ 184146040Stjr int alloc; 185146040Stjr int nelem; 186146040Stjr int *elems; 187146040Stjr} re_node_set; 188146040Stjr 189146040Stjrtypedef enum 190146040Stjr{ 191146040Stjr NON_TYPE = 0, 192146040Stjr 193146040Stjr /* Node type, These are used by token, node, tree. */ 194146040Stjr CHARACTER = 1, 195146040Stjr END_OF_RE = 2, 196146040Stjr SIMPLE_BRACKET = 3, 197146040Stjr OP_BACK_REF = 4, 198146040Stjr OP_PERIOD = 5, 199146040Stjr#ifdef RE_ENABLE_I18N 200146040Stjr COMPLEX_BRACKET = 6, 201146040Stjr OP_UTF8_PERIOD = 7, 202146040Stjr#endif /* RE_ENABLE_I18N */ 203146040Stjr 204146040Stjr /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used 205146040Stjr when the debugger shows values of this enum type. */ 206146040Stjr#define EPSILON_BIT 8 207146040Stjr OP_OPEN_SUBEXP = EPSILON_BIT | 0, 208146040Stjr OP_CLOSE_SUBEXP = EPSILON_BIT | 1, 209146040Stjr OP_ALT = EPSILON_BIT | 2, 210146040Stjr OP_DUP_ASTERISK = EPSILON_BIT | 3, 211146040Stjr ANCHOR = EPSILON_BIT | 4, 212146040Stjr 213146040Stjr /* Tree type, these are used only by tree. */ 214146040Stjr CONCAT = 16, 215146040Stjr SUBEXP = 17, 216146040Stjr 217146040Stjr /* Token type, these are used only by token. */ 218146040Stjr OP_DUP_PLUS = 18, 219146040Stjr OP_DUP_QUESTION, 220146040Stjr OP_OPEN_BRACKET, 221146040Stjr OP_CLOSE_BRACKET, 222146040Stjr OP_CHARSET_RANGE, 223146040Stjr OP_OPEN_DUP_NUM, 224146040Stjr OP_CLOSE_DUP_NUM, 225146040Stjr OP_NON_MATCH_LIST, 226146040Stjr OP_OPEN_COLL_ELEM, 227146040Stjr OP_CLOSE_COLL_ELEM, 228146040Stjr OP_OPEN_EQUIV_CLASS, 229146040Stjr OP_CLOSE_EQUIV_CLASS, 230146040Stjr OP_OPEN_CHAR_CLASS, 231146040Stjr OP_CLOSE_CHAR_CLASS, 232146040Stjr OP_WORD, 233146040Stjr OP_NOTWORD, 234146040Stjr OP_SPACE, 235146040Stjr OP_NOTSPACE, 236146040Stjr BACK_SLASH 237146040Stjr 238146040Stjr} re_token_type_t; 239146040Stjr 240146040Stjr#ifdef RE_ENABLE_I18N 241146040Stjrtypedef struct 242146040Stjr{ 243146040Stjr /* Multibyte characters. */ 244146040Stjr wchar_t *mbchars; 245146040Stjr 246146040Stjr /* Collating symbols. */ 247146040Stjr# ifdef _LIBC 248146040Stjr int32_t *coll_syms; 249146040Stjr# endif 250146040Stjr 251146040Stjr /* Equivalence classes. */ 252146040Stjr# ifdef _LIBC 253146040Stjr int32_t *equiv_classes; 254146040Stjr# endif 255146040Stjr 256146040Stjr /* Range expressions. */ 257146040Stjr# ifdef _LIBC 258146040Stjr uint32_t *range_starts; 259146040Stjr uint32_t *range_ends; 260146040Stjr# else /* not _LIBC */ 261146040Stjr wchar_t *range_starts; 262146040Stjr wchar_t *range_ends; 263146040Stjr# endif /* not _LIBC */ 264146040Stjr 265146040Stjr /* Character classes. */ 266146040Stjr wctype_t *char_classes; 267146040Stjr 268146040Stjr /* If this character set is the non-matching list. */ 269146040Stjr unsigned int non_match : 1; 270146040Stjr 271146040Stjr /* # of multibyte characters. */ 272146040Stjr int nmbchars; 273146040Stjr 274146040Stjr /* # of collating symbols. */ 275146040Stjr int ncoll_syms; 276146040Stjr 277146040Stjr /* # of equivalence classes. */ 278146040Stjr int nequiv_classes; 279146040Stjr 280146040Stjr /* # of range expressions. */ 281146040Stjr int nranges; 282146040Stjr 283146040Stjr /* # of character classes. */ 284146040Stjr int nchar_classes; 285146040Stjr} re_charset_t; 286146040Stjr#endif /* RE_ENABLE_I18N */ 287146040Stjr 288146040Stjrtypedef struct 289146040Stjr{ 290146040Stjr union 291146040Stjr { 292146040Stjr unsigned char c; /* for CHARACTER */ 293146040Stjr re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ 294146040Stjr#ifdef RE_ENABLE_I18N 295146040Stjr re_charset_t *mbcset; /* for COMPLEX_BRACKET */ 296146040Stjr#endif /* RE_ENABLE_I18N */ 297146040Stjr int idx; /* for BACK_REF */ 298146040Stjr re_context_type ctx_type; /* for ANCHOR */ 299146040Stjr } opr; 300146040Stjr#if __GNUC__ >= 2 301146040Stjr re_token_type_t type : 8; 302146040Stjr#else 303146040Stjr re_token_type_t type; 304146040Stjr#endif 305146040Stjr unsigned int constraint : 10; /* context constraint */ 306146040Stjr unsigned int duplicated : 1; 307146040Stjr unsigned int opt_subexp : 1; 308146040Stjr#ifdef RE_ENABLE_I18N 309146040Stjr unsigned int accept_mb : 1; 310146040Stjr /* These 2 bits can be moved into the union if needed (e.g. if running out 311146040Stjr of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ 312146040Stjr unsigned int mb_partial : 1; 313146040Stjr#endif 314146040Stjr unsigned int word_char : 1; 315146040Stjr} re_token_t; 316146040Stjr 317146040Stjr#define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) 318146040Stjr 319146040Stjrstruct re_string_t 320146040Stjr{ 321146040Stjr /* Indicate the raw buffer which is the original string passed as an 322146040Stjr argument of regexec(), re_search(), etc.. */ 323146040Stjr const unsigned char *raw_mbs; 324146040Stjr /* Store the multibyte string. In case of "case insensitive mode" like 325146040Stjr REG_ICASE, upper cases of the string are stored, otherwise MBS points 326146040Stjr the same address that RAW_MBS points. */ 327146040Stjr unsigned char *mbs; 328146040Stjr#ifdef RE_ENABLE_I18N 329146040Stjr /* Store the wide character string which is corresponding to MBS. */ 330146040Stjr wint_t *wcs; 331146040Stjr int *offsets; 332146040Stjr mbstate_t cur_state; 333146040Stjr#endif 334146040Stjr /* Index in RAW_MBS. Each character mbs[i] corresponds to 335146040Stjr raw_mbs[raw_mbs_idx + i]. */ 336146040Stjr int raw_mbs_idx; 337146040Stjr /* The length of the valid characters in the buffers. */ 338146040Stjr int valid_len; 339146040Stjr /* The corresponding number of bytes in raw_mbs array. */ 340146040Stjr int valid_raw_len; 341146040Stjr /* The length of the buffers MBS and WCS. */ 342146040Stjr int bufs_len; 343146040Stjr /* The index in MBS, which is updated by re_string_fetch_byte. */ 344146040Stjr int cur_idx; 345146040Stjr /* length of RAW_MBS array. */ 346146040Stjr int raw_len; 347146040Stjr /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ 348146040Stjr int len; 349146040Stjr /* End of the buffer may be shorter than its length in the cases such 350146040Stjr as re_match_2, re_search_2. Then, we use STOP for end of the buffer 351146040Stjr instead of LEN. */ 352146040Stjr int raw_stop; 353146040Stjr /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ 354146040Stjr int stop; 355146040Stjr 356146040Stjr /* The context of mbs[0]. We store the context independently, since 357146040Stjr the context of mbs[0] may be different from raw_mbs[0], which is 358146040Stjr the beginning of the input string. */ 359146040Stjr unsigned int tip_context; 360146040Stjr /* The translation passed as a part of an argument of re_compile_pattern. */ 361250724Sjkim RE_TRANSLATE_TYPE trans; 362146040Stjr /* Copy of re_dfa_t's word_char. */ 363146040Stjr re_const_bitset_ptr_t word_char; 364146040Stjr /* 1 if REG_ICASE. */ 365146040Stjr unsigned char icase; 366146040Stjr unsigned char is_utf8; 367146040Stjr unsigned char map_notascii; 368146040Stjr unsigned char mbs_allocated; 369146040Stjr unsigned char offsets_needed; 370146040Stjr unsigned char newline_anchor; 371146040Stjr unsigned char word_ops_used; 372146040Stjr int mb_cur_max; 373146040Stjr}; 374146040Stjrtypedef struct re_string_t re_string_t; 375146040Stjr 376146040Stjr 377146040Stjrstruct re_dfa_t; 378146040Stjrtypedef struct re_dfa_t re_dfa_t; 379146040Stjr 380146040Stjr#ifndef _LIBC 381146040Stjr# ifdef __i386__ 382146040Stjr# define internal_function __attribute ((regparm (3), stdcall)) 383146040Stjr# else 384146040Stjr# define internal_function 385146040Stjr# endif 386146040Stjr#endif 387146040Stjr 388250724Sjkim#ifndef NOT_IN_libc 389146040Stjrstatic reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, 390146040Stjr int new_buf_len) 391146040Stjr internal_function; 392146040Stjr# ifdef RE_ENABLE_I18N 393146040Stjrstatic void build_wcs_buffer (re_string_t *pstr) internal_function; 394250724Sjkimstatic reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr) 395250724Sjkim internal_function; 396146040Stjr# endif /* RE_ENABLE_I18N */ 397146040Stjrstatic void build_upper_buffer (re_string_t *pstr) internal_function; 398146040Stjrstatic void re_string_translate_buffer (re_string_t *pstr) internal_function; 399146040Stjrstatic unsigned int re_string_context_at (const re_string_t *input, int idx, 400146040Stjr int eflags) 401146040Stjr internal_function __attribute ((pure)); 402146040Stjr#endif 403146040Stjr#define re_string_peek_byte(pstr, offset) \ 404146040Stjr ((pstr)->mbs[(pstr)->cur_idx + offset]) 405146040Stjr#define re_string_fetch_byte(pstr) \ 406146040Stjr ((pstr)->mbs[(pstr)->cur_idx++]) 407146040Stjr#define re_string_first_byte(pstr, idx) \ 408146040Stjr ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) 409146040Stjr#define re_string_is_single_byte_char(pstr, idx) \ 410146040Stjr ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ 411146040Stjr || (pstr)->wcs[(idx) + 1] != WEOF)) 412146040Stjr#define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) 413146040Stjr#define re_string_cur_idx(pstr) ((pstr)->cur_idx) 414146040Stjr#define re_string_get_buffer(pstr) ((pstr)->mbs) 415146040Stjr#define re_string_length(pstr) ((pstr)->len) 416146040Stjr#define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) 417146040Stjr#define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) 418146040Stjr#define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) 419146040Stjr 420251437Sjkim#ifndef __FreeBSD__ 421250724Sjkim#include <alloca.h> 422251437Sjkim#endif 423250724Sjkim 424250724Sjkim#ifndef _LIBC 425250724Sjkim# if HAVE_ALLOCA 426250724Sjkim/* The OS usually guarantees only one guard page at the bottom of the stack, 427250724Sjkim and a page size can be as small as 4096 bytes. So we cannot safely 428250724Sjkim allocate anything larger than 4096 bytes. Also care for the possibility 429250724Sjkim of a few compiler-allocated temporary stack slots. */ 430250724Sjkim# define __libc_use_alloca(n) ((n) < 4032) 431250724Sjkim# else 432250724Sjkim/* alloca is implemented with malloc, so just use malloc. */ 433250724Sjkim# define __libc_use_alloca(n) 0 434250724Sjkim# endif 435250724Sjkim#endif 436250724Sjkim 437146040Stjr#define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) 438146040Stjr#define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) 439146040Stjr#define re_free(p) free (p) 440146040Stjr 441146040Stjrstruct bin_tree_t 442146040Stjr{ 443146040Stjr struct bin_tree_t *parent; 444146040Stjr struct bin_tree_t *left; 445146040Stjr struct bin_tree_t *right; 446146040Stjr struct bin_tree_t *first; 447146040Stjr struct bin_tree_t *next; 448146040Stjr 449146040Stjr re_token_t token; 450146040Stjr 451146040Stjr /* `node_idx' is the index in dfa->nodes, if `type' == 0. 452146040Stjr Otherwise `type' indicate the type of this node. */ 453146040Stjr int node_idx; 454146040Stjr}; 455146040Stjrtypedef struct bin_tree_t bin_tree_t; 456146040Stjr 457146040Stjr#define BIN_TREE_STORAGE_SIZE \ 458146040Stjr ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) 459146040Stjr 460146040Stjrstruct bin_tree_storage_t 461146040Stjr{ 462146040Stjr struct bin_tree_storage_t *next; 463146040Stjr bin_tree_t data[BIN_TREE_STORAGE_SIZE]; 464146040Stjr}; 465146040Stjrtypedef struct bin_tree_storage_t bin_tree_storage_t; 466146040Stjr 467146040Stjr#define CONTEXT_WORD 1 468146040Stjr#define CONTEXT_NEWLINE (CONTEXT_WORD << 1) 469146040Stjr#define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) 470146040Stjr#define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) 471146040Stjr 472146040Stjr#define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) 473146040Stjr#define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) 474146040Stjr#define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) 475146040Stjr#define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) 476146040Stjr#define IS_ORDINARY_CONTEXT(c) ((c) == 0) 477146040Stjr 478146040Stjr#define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') 479146040Stjr#define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) 480146040Stjr#define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_') 481146040Stjr#define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) 482146040Stjr 483146040Stjr#define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ 484146040Stjr ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 485146040Stjr || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 486146040Stjr || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ 487146040Stjr || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) 488146040Stjr 489146040Stjr#define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ 490146040Stjr ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ 491146040Stjr || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ 492146040Stjr || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ 493146040Stjr || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) 494146040Stjr 495146040Stjrstruct re_dfastate_t 496146040Stjr{ 497146040Stjr unsigned int hash; 498146040Stjr re_node_set nodes; 499146040Stjr re_node_set non_eps_nodes; 500146040Stjr re_node_set inveclosure; 501146040Stjr re_node_set *entrance_nodes; 502146040Stjr struct re_dfastate_t **trtable, **word_trtable; 503146040Stjr unsigned int context : 4; 504146040Stjr unsigned int halt : 1; 505146040Stjr /* If this state can accept `multi byte'. 506146040Stjr Note that we refer to multibyte characters, and multi character 507146040Stjr collating elements as `multi byte'. */ 508146040Stjr unsigned int accept_mb : 1; 509146040Stjr /* If this state has backreference node(s). */ 510146040Stjr unsigned int has_backref : 1; 511146040Stjr unsigned int has_constraint : 1; 512146040Stjr}; 513146040Stjrtypedef struct re_dfastate_t re_dfastate_t; 514146040Stjr 515146040Stjrstruct re_state_table_entry 516146040Stjr{ 517146040Stjr int num; 518146040Stjr int alloc; 519146040Stjr re_dfastate_t **array; 520146040Stjr}; 521146040Stjr 522146040Stjr/* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ 523146040Stjr 524146040Stjrtypedef struct 525146040Stjr{ 526146040Stjr int next_idx; 527146040Stjr int alloc; 528146040Stjr re_dfastate_t **array; 529146040Stjr} state_array_t; 530146040Stjr 531146040Stjr/* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ 532146040Stjr 533146040Stjrtypedef struct 534146040Stjr{ 535146040Stjr int node; 536146040Stjr int str_idx; /* The position NODE match at. */ 537146040Stjr state_array_t path; 538146040Stjr} re_sub_match_last_t; 539146040Stjr 540146040Stjr/* Store information about the node NODE whose type is OP_OPEN_SUBEXP. 541146040Stjr And information about the node, whose type is OP_CLOSE_SUBEXP, 542146040Stjr corresponding to NODE is stored in LASTS. */ 543146040Stjr 544146040Stjrtypedef struct 545146040Stjr{ 546146040Stjr int str_idx; 547146040Stjr int node; 548146040Stjr state_array_t *path; 549146040Stjr int alasts; /* Allocation size of LASTS. */ 550146040Stjr int nlasts; /* The number of LASTS. */ 551146040Stjr re_sub_match_last_t **lasts; 552146040Stjr} re_sub_match_top_t; 553146040Stjr 554146040Stjrstruct re_backref_cache_entry 555146040Stjr{ 556146040Stjr int node; 557146040Stjr int str_idx; 558146040Stjr int subexp_from; 559146040Stjr int subexp_to; 560146040Stjr char more; 561146040Stjr char unused; 562146040Stjr unsigned short int eps_reachable_subexps_map; 563146040Stjr}; 564146040Stjr 565146040Stjrtypedef struct 566146040Stjr{ 567146040Stjr /* The string object corresponding to the input string. */ 568146040Stjr re_string_t input; 569146040Stjr#if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) 570250724Sjkim const re_dfa_t *const dfa; 571146040Stjr#else 572250724Sjkim const re_dfa_t *dfa; 573146040Stjr#endif 574146040Stjr /* EFLAGS of the argument of regexec. */ 575146040Stjr int eflags; 576146040Stjr /* Where the matching ends. */ 577146040Stjr int match_last; 578146040Stjr int last_node; 579146040Stjr /* The state log used by the matcher. */ 580146040Stjr re_dfastate_t **state_log; 581146040Stjr int state_log_top; 582146040Stjr /* Back reference cache. */ 583146040Stjr int nbkref_ents; 584146040Stjr int abkref_ents; 585146040Stjr struct re_backref_cache_entry *bkref_ents; 586146040Stjr int max_mb_elem_len; 587146040Stjr int nsub_tops; 588146040Stjr int asub_tops; 589146040Stjr re_sub_match_top_t **sub_tops; 590146040Stjr} re_match_context_t; 591146040Stjr 592146040Stjrtypedef struct 593146040Stjr{ 594146040Stjr re_dfastate_t **sifted_states; 595146040Stjr re_dfastate_t **limited_states; 596146040Stjr int last_node; 597146040Stjr int last_str_idx; 598146040Stjr re_node_set limits; 599146040Stjr} re_sift_context_t; 600146040Stjr 601146040Stjrstruct re_fail_stack_ent_t 602146040Stjr{ 603146040Stjr int idx; 604146040Stjr int node; 605146040Stjr regmatch_t *regs; 606146040Stjr re_node_set eps_via_nodes; 607146040Stjr}; 608146040Stjr 609146040Stjrstruct re_fail_stack_t 610146040Stjr{ 611146040Stjr int num; 612146040Stjr int alloc; 613146040Stjr struct re_fail_stack_ent_t *stack; 614146040Stjr}; 615146040Stjr 616146040Stjrstruct re_dfa_t 617146040Stjr{ 618146040Stjr re_token_t *nodes; 619250724Sjkim size_t nodes_alloc; 620250724Sjkim size_t nodes_len; 621146040Stjr int *nexts; 622146040Stjr int *org_indices; 623146040Stjr re_node_set *edests; 624146040Stjr re_node_set *eclosures; 625146040Stjr re_node_set *inveclosures; 626146040Stjr struct re_state_table_entry *state_table; 627146040Stjr re_dfastate_t *init_state; 628146040Stjr re_dfastate_t *init_state_word; 629146040Stjr re_dfastate_t *init_state_nl; 630146040Stjr re_dfastate_t *init_state_begbuf; 631146040Stjr bin_tree_t *str_tree; 632146040Stjr bin_tree_storage_t *str_tree_storage; 633146040Stjr re_bitset_ptr_t sb_char; 634146040Stjr int str_tree_storage_idx; 635146040Stjr 636146040Stjr /* number of subexpressions `re_nsub' is in regex_t. */ 637146040Stjr unsigned int state_hash_mask; 638146040Stjr int init_node; 639146040Stjr int nbackref; /* The number of backreference in this dfa. */ 640146040Stjr 641146040Stjr /* Bitmap expressing which backreference is used. */ 642250724Sjkim bitset_word_t used_bkref_map; 643250724Sjkim bitset_word_t completed_bkref_map; 644146040Stjr 645146040Stjr unsigned int has_plural_match : 1; 646146040Stjr /* If this dfa has "multibyte node", which is a backreference or 647146040Stjr a node which can accept multibyte character or multi character 648146040Stjr collating element. */ 649146040Stjr unsigned int has_mb_node : 1; 650146040Stjr unsigned int is_utf8 : 1; 651146040Stjr unsigned int map_notascii : 1; 652146040Stjr unsigned int word_ops_used : 1; 653146040Stjr int mb_cur_max; 654250724Sjkim bitset_t word_char; 655146040Stjr reg_syntax_t syntax; 656146040Stjr int *subexp_map; 657146040Stjr#ifdef DEBUG 658146040Stjr char* re_str; 659146040Stjr#endif 660250724Sjkim __libc_lock_define (, lock) 661146040Stjr}; 662146040Stjr 663146040Stjr#define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) 664146040Stjr#define re_node_set_remove(set,id) \ 665146040Stjr (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) 666146040Stjr#define re_node_set_empty(p) ((p)->nelem = 0) 667146040Stjr#define re_node_set_free(set) re_free ((set)->elems) 668146040Stjr 669146040Stjr 670146040Stjrtypedef enum 671146040Stjr{ 672146040Stjr SB_CHAR, 673146040Stjr MB_CHAR, 674146040Stjr EQUIV_CLASS, 675146040Stjr COLL_SYM, 676146040Stjr CHAR_CLASS 677146040Stjr} bracket_elem_type; 678146040Stjr 679146040Stjrtypedef struct 680146040Stjr{ 681146040Stjr bracket_elem_type type; 682146040Stjr union 683146040Stjr { 684146040Stjr unsigned char ch; 685146040Stjr unsigned char *name; 686146040Stjr wchar_t wch; 687146040Stjr } opr; 688146040Stjr} bracket_elem_t; 689146040Stjr 690146040Stjr 691146040Stjr/* Inline functions for bitset operation. */ 692146040Stjrstatic inline void 693250724Sjkimbitset_not (bitset_t set) 694146040Stjr{ 695146040Stjr int bitset_i; 696250724Sjkim for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 697146040Stjr set[bitset_i] = ~set[bitset_i]; 698146040Stjr} 699146040Stjr 700146040Stjrstatic inline void 701250724Sjkimbitset_merge (bitset_t dest, const bitset_t src) 702146040Stjr{ 703146040Stjr int bitset_i; 704250724Sjkim for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 705146040Stjr dest[bitset_i] |= src[bitset_i]; 706146040Stjr} 707146040Stjr 708146040Stjrstatic inline void 709250724Sjkimbitset_mask (bitset_t dest, const bitset_t src) 710146040Stjr{ 711146040Stjr int bitset_i; 712250724Sjkim for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) 713146040Stjr dest[bitset_i] &= src[bitset_i]; 714146040Stjr} 715146040Stjr 716250724Sjkim#ifdef RE_ENABLE_I18N 717146040Stjr/* Inline functions for re_string. */ 718146040Stjrstatic inline int 719250724Sjkiminternal_function __attribute ((pure)) 720146040Stjrre_string_char_size_at (const re_string_t *pstr, int idx) 721146040Stjr{ 722146040Stjr int byte_idx; 723146040Stjr if (pstr->mb_cur_max == 1) 724146040Stjr return 1; 725146040Stjr for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) 726146040Stjr if (pstr->wcs[idx + byte_idx] != WEOF) 727146040Stjr break; 728146040Stjr return byte_idx; 729146040Stjr} 730146040Stjr 731146040Stjrstatic inline wint_t 732250724Sjkiminternal_function __attribute ((pure)) 733146040Stjrre_string_wchar_at (const re_string_t *pstr, int idx) 734146040Stjr{ 735146040Stjr if (pstr->mb_cur_max == 1) 736146040Stjr return (wint_t) pstr->mbs[idx]; 737146040Stjr return (wint_t) pstr->wcs[idx]; 738146040Stjr} 739146040Stjr 740250724Sjkim# ifndef NOT_IN_libc 741146040Stjrstatic int 742250724Sjkiminternal_function __attribute ((pure)) 743146040Stjrre_string_elem_size_at (const re_string_t *pstr, int idx) 744146040Stjr{ 745250724Sjkim# ifdef _LIBC 746146040Stjr const unsigned char *p, *extra; 747146040Stjr const int32_t *table, *indirect; 748250724Sjkim# include <locale/weight.h> 749146040Stjr uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); 750146040Stjr 751146040Stjr if (nrules != 0) 752146040Stjr { 753146040Stjr table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); 754146040Stjr extra = (const unsigned char *) 755146040Stjr _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); 756146040Stjr indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, 757146040Stjr _NL_COLLATE_INDIRECTMB); 758146040Stjr p = pstr->mbs + idx; 759250724Sjkim findidx (&p, pstr->len - idx); 760146040Stjr return p - pstr->mbs - idx; 761146040Stjr } 762146040Stjr else 763250724Sjkim# endif /* _LIBC */ 764146040Stjr return 1; 765146040Stjr} 766250724Sjkim# endif 767146040Stjr#endif /* RE_ENABLE_I18N */ 768146040Stjr 769146040Stjr#endif /* _REGEX_INTERNAL_H */ 770