gl/regex_internal.h

   1 /* Extended regular expression matching and search library.
   2    Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
   3    Free Software Foundation, Inc.
   4    This file is part of the GNU C Library.
   5    Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
   6
   7    This program is free software; you can redistribute it and/or modify
   8    it under the terms of the GNU General Public License as published by
   9    the Free Software Foundation; either version 3, or (at your option)
  10    any later version.
  11
  12    This program is distributed in the hope that it will be useful,
  13    but WITHOUT ANY WARRANTY; without even the implied warranty of
  14    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15    GNU General Public License for more details.
  16
  17    You should have received a copy of the GNU General Public License along
  18    with this program; if not, write to the Free Software Foundation,
  19    Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
  20
  21 #ifndef _REGEX_INTERNAL_H
  22 #define _REGEX_INTERNAL_H 1
  23
  24 #include <assert.h>
  25 #include <ctype.h>
  26 #include <stdbool.h>
  27 #include <stdio.h>
  28 #include <stdlib.h>
  29 #include <string.h>
  30
  31 #ifdef _LIBC
  32 # include <langinfo.h>
  33 #else
  34 # include "localcharset.h"
  35 #endif
  36 #if defined HAVE_LOCALE_H || defined _LIBC
  37 # include <locale.h>
  38 #endif
  39
  40 #include <wchar.h>
  41 #include <wctype.h>
  42 #include <stdint.h>
  43 #if defined _LIBC
  44 # include <bits/libc-lock.h>
  45 #else
  46 # define __libc_lock_init(NAME) do { } while (0)
  47 # define __libc_lock_lock(NAME) do { } while (0)
  48 # define __libc_lock_unlock(NAME) do { } while (0)
  49 #endif
  50
  51 /* In case that the system doesn't have isblank().  */
  52 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK))
  53 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
  54 #endif
  55
  56 #ifdef _LIBC
  57 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
  58 #  define _RE_DEFINE_LOCALE_FUNCTIONS 1
  59 #   include <locale/localeinfo.h>
  60 #   include <locale/elem-hash.h>
  61 #   include <locale/coll-lookup.h>
  62 # endif
  63 #endif
  64
  65 /* This is for other GNU distributions with internationalized messages.  */
  66 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
  67 # include <libintl.h>
  68 # ifdef _LIBC
  69 #  undef gettext
  70 #  define gettext(msgid) \
  71   INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
  72 # endif
  73 #else
  74 # define gettext(msgid) (msgid)
  75 #endif
  76
  77 #ifndef gettext_noop
  78 /* This define is so xgettext can find the internationalizable
  79    strings.  */
  80 # define gettext_noop(String) String
  81 #endif
  82
  83 /* For loser systems without the definition.  */
  84 #ifndef SIZE_MAX
  85 # define SIZE_MAX ((size_t) -1)
  86 #endif
  87
  88 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCSCOLL) || _LIBC
  89 # define RE_ENABLE_I18N
  90 #endif
  91
  92 #if __GNUC__ >= 3
  93 # define BE(expr, val) __builtin_expect (expr, val)
  94 #else
  95 # define BE(expr, val) (expr)
  96 # ifdef _LIBC
  97 #  define inline
  98 # endif
  99 #endif
 100
 101 /* Number of ASCII characters.  */
 102 #define ASCII_CHARS 0x80
 103
 104 /* Number of single byte characters.  */
 105 #define SBC_MAX (UCHAR_MAX + 1)
 106
 107 #define COLL_ELEM_LEN_MAX 8
 108
 109 /* The character which represents newline.  */
 110 #define NEWLINE_CHAR '\n'
 111 #define WIDE_NEWLINE_CHAR L'\n'
 112
 113 /* Rename to standard API for using out of glibc.  */
 114 #ifndef _LIBC
 115 # define __wctype wctype
 116 # define __iswctype iswctype
 117 # define __btowc btowc
 118 # define __wcrtomb wcrtomb
 119 # define __mbrtowc mbrtowc
 120 # define __regfree regfree
 121 # define attribute_hidden
 122 #endif /* not _LIBC */
 123
 124 #if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)
 125 # define __attribute(arg) __attribute__ (arg)
 126 #else
 127 # define __attribute(arg)
 128 #endif
 129
 130 typedef __re_idx_t Idx;
 131
 132 /* Special return value for failure to match.  */
 133 #define REG_MISSING ((Idx) -1)
 134
 135 /* Special return value for internal error.  */
 136 #define REG_ERROR ((Idx) -2)
 137
 138 /* Test whether N is a valid index, and is not one of the above.  */
 139 #ifdef _REGEX_LARGE_OFFSETS
 140 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR)
 141 #else
 142 # define REG_VALID_INDEX(n) (0 <= (n))
 143 #endif
 144
 145 /* Test whether N is a valid nonzero index.  */
 146 #ifdef _REGEX_LARGE_OFFSETS
 147 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1))
 148 #else
 149 # define REG_VALID_NONZERO_INDEX(n) (0 < (n))
 150 #endif
 151
 152 /* A hash value, suitable for computing hash tables.  */
 153 typedef __re_size_t re_hashval_t;
 154
 155 /* An integer used to represent a set of bits.  It must be unsigned,
 156    and must be at least as wide as unsigned int.  */
 157 typedef unsigned long int bitset_word_t;
 158 /* All bits set in a bitset_word_t.  */
 159 #define BITSET_WORD_MAX ULONG_MAX
 160
 161 /* Number of bits in a bitset_word_t.  For portability to hosts with
 162    padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)';
 163    instead, deduce it directly from BITSET_WORD_MAX.  Avoid
 164    greater-than-32-bit integers and unconditional shifts by more than
 165    31 bits, as they're not portable.  */
 166 #if BITSET_WORD_MAX == 0xffffffffUL
 167 # define BITSET_WORD_BITS 32
 168 #elif BITSET_WORD_MAX >> 31 >> 4 == 1
 169 # define BITSET_WORD_BITS 36
 170 #elif BITSET_WORD_MAX >> 31 >> 16 == 1
 171 # define BITSET_WORD_BITS 48
 172 #elif BITSET_WORD_MAX >> 31 >> 28 == 1
 173 # define BITSET_WORD_BITS 60
 174 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1
 175 # define BITSET_WORD_BITS 64
 176 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1
 177 # define BITSET_WORD_BITS 72
 178 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1
 179 # define BITSET_WORD_BITS 128
 180 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1
 181 # define BITSET_WORD_BITS 256
 182 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1
 183 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */
 184 # if BITSET_WORD_BITS <= SBC_MAX
 185 #  error "Invalid SBC_MAX"
 186 # endif
 187 #else
 188 # error "Add case for new bitset_word_t size"
 189 #endif
 190
 191 /* Number of bitset_word_t values in a bitset_t.  */
 192 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS)
 193
 194 typedef bitset_word_t bitset_t[BITSET_WORDS];
 195 typedef bitset_word_t *re_bitset_ptr_t;
 196 typedef const bitset_word_t *re_const_bitset_ptr_t;
 197
 198 #define PREV_WORD_CONSTRAINT 0x0001
 199 #define PREV_NOTWORD_CONSTRAINT 0x0002
 200 #define NEXT_WORD_CONSTRAINT 0x0004
 201 #define NEXT_NOTWORD_CONSTRAINT 0x0008
 202 #define PREV_NEWLINE_CONSTRAINT 0x0010
 203 #define NEXT_NEWLINE_CONSTRAINT 0x0020
 204 #define PREV_BEGBUF_CONSTRAINT 0x0040
 205 #define NEXT_ENDBUF_CONSTRAINT 0x0080
 206 #define WORD_DELIM_CONSTRAINT 0x0100
 207 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
 208
 209 typedef enum
 210 {
 211   INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 212   WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 213   WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 214   INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 215   LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
 216   LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
 217   BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
 218   BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
 219   WORD_DELIM = WORD_DELIM_CONSTRAINT,
 220   NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
 221 } re_context_type;
 222
 223 typedef struct
 224 {
 225   Idx alloc;
 226   Idx nelem;
 227   Idx *elems;
 228 } re_node_set;
 229
 230 typedef enum
 231 {
 232   NON_TYPE = 0,
 233
 234   /* Node type, These are used by token, node, tree.  */
 235   CHARACTER = 1,
 236   END_OF_RE = 2,
 237   SIMPLE_BRACKET = 3,
 238   OP_BACK_REF = 4,
 239   OP_PERIOD = 5,
 240 #ifdef RE_ENABLE_I18N
 241   COMPLEX_BRACKET = 6,
 242   OP_UTF8_PERIOD = 7,
 243 #endif /* RE_ENABLE_I18N */
 244
 245   /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
 246      when the debugger shows values of this enum type.  */
 247 #define EPSILON_BIT 8
 248   OP_OPEN_SUBEXP = EPSILON_BIT | 0,
 249   OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
 250   OP_ALT = EPSILON_BIT | 2,
 251   OP_DUP_ASTERISK = EPSILON_BIT | 3,
 252   ANCHOR = EPSILON_BIT | 4,
 253
 254   /* Tree type, these are used only by tree. */
 255   CONCAT = 16,
 256   SUBEXP = 17,
 257
 258   /* Token type, these are used only by token.  */
 259   OP_DUP_PLUS = 18,
 260   OP_DUP_QUESTION,
 261   OP_OPEN_BRACKET,
 262   OP_CLOSE_BRACKET,
 263   OP_CHARSET_RANGE,
 264   OP_OPEN_DUP_NUM,
 265   OP_CLOSE_DUP_NUM,
 266   OP_NON_MATCH_LIST,
 267   OP_OPEN_COLL_ELEM,
 268   OP_CLOSE_COLL_ELEM,
 269   OP_OPEN_EQUIV_CLASS,
 270   OP_CLOSE_EQUIV_CLASS,
 271   OP_OPEN_CHAR_CLASS,
 272   OP_CLOSE_CHAR_CLASS,
 273   OP_WORD,
 274   OP_NOTWORD,
 275   OP_SPACE,
 276   OP_NOTSPACE,
 277   BACK_SLASH
 278
 279 } re_token_type_t;
 280
 281 #ifdef RE_ENABLE_I18N
 282 typedef struct
 283 {
 284   /* Multibyte characters.  */
 285   wchar_t *mbchars;
 286
 287   /* Collating symbols.  */
 288 # ifdef _LIBC
 289   int32_t *coll_syms;
 290 # endif
 291
 292   /* Equivalence classes. */
 293 # ifdef _LIBC
 294   int32_t *equiv_classes;
 295 # endif
 296
 297   /* Range expressions. */
 298 # ifdef _LIBC
 299   uint32_t *range_starts;
 300   uint32_t *range_ends;
 301 # else /* not _LIBC */
 302   wchar_t *range_starts;
 303   wchar_t *range_ends;
 304 # endif /* not _LIBC */
 305
 306   /* Character classes. */
 307   wctype_t *char_classes;
 308
 309   /* If this character set is the non-matching list.  */
 310   unsigned int non_match : 1;
 311
 312   /* # of multibyte characters.  */
 313   Idx nmbchars;
 314
 315   /* # of collating symbols.  */
 316   Idx ncoll_syms;
 317
 318   /* # of equivalence classes. */
 319   Idx nequiv_classes;
 320
 321   /* # of range expressions. */
 322   Idx nranges;
 323
 324   /* # of character classes. */
 325   Idx nchar_classes;
 326 } re_charset_t;
 327 #endif /* RE_ENABLE_I18N */
 328
 329 typedef struct
 330 {
 331   union
 332   {
 333     unsigned char c;            /* for CHARACTER */
 334     re_bitset_ptr_t sbcset;     /* for SIMPLE_BRACKET */
 335 #ifdef RE_ENABLE_I18N
 336     re_charset_t *mbcset;       /* for COMPLEX_BRACKET */
 337 #endif /* RE_ENABLE_I18N */
 338     Idx idx;                    /* for BACK_REF */
 339     re_context_type ctx_type;   /* for ANCHOR */
 340   } opr;
 341 #if __GNUC__ >= 2 && !__STRICT_ANSI__
 342   re_token_type_t type : 8;
 343 #else
 344   re_token_type_t type;
 345 #endif
 346   unsigned int constraint : 10; /* context constraint */
 347   unsigned int duplicated : 1;
 348   unsigned int opt_subexp : 1;
 349 #ifdef RE_ENABLE_I18N
 350   unsigned int accept_mb : 1;
 351   /* These 2 bits can be moved into the union if needed (e.g. if running out
 352      of bits; move opr.c to opr.c.c and move the flags to opr.c.flags).  */
 353   unsigned int mb_partial : 1;
 354 #endif
 355   unsigned int word_char : 1;
 356 } re_token_t;
 357
 358 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
 359
 360 struct re_string_t
 361 {
 362   /* Indicate the raw buffer which is the original string passed as an
 363      argument of regexec(), re_search(), etc..  */
 364   const unsigned char *raw_mbs;
 365   /* Store the multibyte string.  In case of "case insensitive mode" like
 366      REG_ICASE, upper cases of the string are stored, otherwise MBS points
 367      the same address that RAW_MBS points.  */
 368   unsigned char *mbs;
 369 #ifdef RE_ENABLE_I18N
 370   /* Store the wide character string which is corresponding to MBS.  */
 371   wint_t *wcs;
 372   Idx *offsets;
 373   mbstate_t cur_state;
 374 #endif
 375   /* Index in RAW_MBS.  Each character mbs[i] corresponds to
 376      raw_mbs[raw_mbs_idx + i].  */
 377   Idx raw_mbs_idx;
 378   /* The length of the valid characters in the buffers.  */
 379   Idx valid_len;
 380   /* The corresponding number of bytes in raw_mbs array.  */
 381   Idx valid_raw_len;
 382   /* The length of the buffers MBS and WCS.  */
 383   Idx bufs_len;
 384   /* The index in MBS, which is updated by re_string_fetch_byte.  */
 385   Idx cur_idx;
 386   /* length of RAW_MBS array.  */
 387   Idx raw_len;
 388   /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN.  */
 389   Idx len;
 390   /* End of the buffer may be shorter than its length in the cases such
 391      as re_match_2, re_search_2.  Then, we use STOP for end of the buffer
 392      instead of LEN.  */
 393   Idx raw_stop;
 394   /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS.  */
 395   Idx stop;
 396
 397   /* The context of mbs[0].  We store the context independently, since
 398      the context of mbs[0] may be different from raw_mbs[0], which is
 399      the beginning of the input string.  */
 400   unsigned int tip_context;
 401   /* The translation passed as a part of an argument of re_compile_pattern.  */
 402   RE_TRANSLATE_TYPE trans;
 403   /* Copy of re_dfa_t's word_char.  */
 404   re_const_bitset_ptr_t word_char;
 405   /* true if REG_ICASE.  */
 406   unsigned char icase;
 407   unsigned char is_utf8;
 408   unsigned char map_notascii;
 409   unsigned char mbs_allocated;
 410   unsigned char offsets_needed;
 411   unsigned char newline_anchor;
 412   unsigned char word_ops_used;
 413   int mb_cur_max;
 414 };
 415 typedef struct re_string_t re_string_t;
 416
 417
 418 struct re_dfa_t;
 419 typedef struct re_dfa_t re_dfa_t;
 420
 421 #ifndef _LIBC
 422 # if defined __i386__ && !defined __EMX__
 423 #  define internal_function   __attribute ((regparm (3), stdcall))
 424 # else
 425 #  define internal_function
 426 # endif
 427 #endif
 428
 429 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
 430                                                 Idx new_buf_len)
 431      internal_function;
 432 #ifdef RE_ENABLE_I18N
 433 static void build_wcs_buffer (re_string_t *pstr) internal_function;
 434 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
 435      internal_function;
 436 #endif /* RE_ENABLE_I18N */
 437 static void build_upper_buffer (re_string_t *pstr) internal_function;
 438 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
 439 static unsigned int re_string_context_at (const re_string_t *input, Idx idx,
 440                                           int eflags)
 441      internal_function __attribute ((pure));
 442 #define re_string_peek_byte(pstr, offset) \
 443   ((pstr)->mbs[(pstr)->cur_idx + offset])
 444 #define re_string_fetch_byte(pstr) \
 445   ((pstr)->mbs[(pstr)->cur_idx++])
 446 #define re_string_first_byte(pstr, idx) \
 447   ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
 448 #define re_string_is_single_byte_char(pstr, idx) \
 449   ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
 450                                 || (pstr)->wcs[(idx) + 1] != WEOF))
 451 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
 452 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
 453 #define re_string_get_buffer(pstr) ((pstr)->mbs)
 454 #define re_string_length(pstr) ((pstr)->len)
 455 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
 456 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
 457 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
 458
 459 #include <alloca.h>
 460
 461 #ifndef _LIBC
 462 # if HAVE_ALLOCA
 463 /* The OS usually guarantees only one guard page at the bottom of the stack,
 464    and a page size can be as small as 4096 bytes.  So we cannot safely
 465    allocate anything larger than 4096 bytes.  Also care for the possibility
 466    of a few compiler-allocated temporary stack slots.  */
 467 #  define __libc_use_alloca(n) ((n) < 4032)
 468 # else
 469 /* alloca is implemented with malloc, so just use malloc.  */
 470 #  define __libc_use_alloca(n) 0
 471 # endif
 472 #endif
 473
 474 #ifndef MAX
 475 # define MAX(a,b) ((a) < (b) ? (b) : (a))
 476 #endif
 477
 478 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
 479 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
 480 #define re_free(p) free (p)
 481
 482 struct bin_tree_t
 483 {
 484   struct bin_tree_t *parent;
 485   struct bin_tree_t *left;
 486   struct bin_tree_t *right;
 487   struct bin_tree_t *first;
 488   struct bin_tree_t *next;
 489
 490   re_token_t token;
 491
 492   /* `node_idx' is the index in dfa->nodes, if `type' == 0.
 493      Otherwise `type' indicate the type of this node.  */
 494   Idx node_idx;
 495 };
 496 typedef struct bin_tree_t bin_tree_t;
 497
 498 #define BIN_TREE_STORAGE_SIZE \
 499   ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
 500
 501 struct bin_tree_storage_t
 502 {
 503   struct bin_tree_storage_t *next;
 504   bin_tree_t data[BIN_TREE_STORAGE_SIZE];
 505 };
 506 typedef struct bin_tree_storage_t bin_tree_storage_t;
 507
 508 #define CONTEXT_WORD 1
 509 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
 510 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
 511 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
 512
 513 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
 514 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
 515 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
 516 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
 517 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
 518
 519 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
 520 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
 521 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
 522 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
 523
 524 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
 525  ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 526   || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 527   || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
 528   || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
 529
 530 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
 531  ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 532   || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 533   || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
 534   || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
 535
 536 struct re_dfastate_t
 537 {
 538   re_hashval_t hash;
 539   re_node_set nodes;
 540   re_node_set non_eps_nodes;
 541   re_node_set inveclosure;
 542   re_node_set *entrance_nodes;
 543   struct re_dfastate_t **trtable, **word_trtable;
 544   unsigned int context : 4;
 545   unsigned int halt : 1;
 546   /* If this state can accept `multi byte'.
 547      Note that we refer to multibyte characters, and multi character
 548      collating elements as `multi byte'.  */
 549   unsigned int accept_mb : 1;
 550   /* If this state has backreference node(s).  */
 551   unsigned int has_backref : 1;
 552   unsigned int has_constraint : 1;
 553 };
 554 typedef struct re_dfastate_t re_dfastate_t;
 555
 556 struct re_state_table_entry
 557 {
 558   Idx num;
 559   Idx alloc;
 560   re_dfastate_t **array;
 561 };
 562
 563 /* Array type used in re_sub_match_last_t and re_sub_match_top_t.  */
 564
 565 typedef struct
 566 {
 567   Idx next_idx;
 568   Idx alloc;
 569   re_dfastate_t **array;
 570 } state_array_t;
 571
 572 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP.  */
 573
 574 typedef struct
 575 {
 576   Idx node;
 577   Idx str_idx; /* The position NODE match at.  */
 578   state_array_t path;
 579 } re_sub_match_last_t;
 580
 581 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
 582    And information about the node, whose type is OP_CLOSE_SUBEXP,
 583    corresponding to NODE is stored in LASTS.  */
 584
 585 typedef struct
 586 {
 587   Idx str_idx;
 588   Idx node;
 589   state_array_t *path;
 590   Idx alasts; /* Allocation size of LASTS.  */
 591   Idx nlasts; /* The number of LASTS.  */
 592   re_sub_match_last_t **lasts;
 593 } re_sub_match_top_t;
 594
 595 struct re_backref_cache_entry
 596 {
 597   Idx node;
 598   Idx str_idx;
 599   Idx subexp_from;
 600   Idx subexp_to;
 601   char more;
 602   char unused;
 603   unsigned short int eps_reachable_subexps_map;
 604 };
 605
 606 typedef struct
 607 {
 608   /* The string object corresponding to the input string.  */
 609   re_string_t input;
 610 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
 611   const re_dfa_t *const dfa;
 612 #else
 613   const re_dfa_t *dfa;
 614 #endif
 615   /* EFLAGS of the argument of regexec.  */
 616   int eflags;
 617   /* Where the matching ends.  */
 618   Idx match_last;
 619   Idx last_node;
 620   /* The state log used by the matcher.  */
 621   re_dfastate_t **state_log;
 622   Idx state_log_top;
 623   /* Back reference cache.  */
 624   Idx nbkref_ents;
 625   Idx abkref_ents;
 626   struct re_backref_cache_entry *bkref_ents;
 627   int max_mb_elem_len;
 628   Idx nsub_tops;
 629   Idx asub_tops;
 630   re_sub_match_top_t **sub_tops;
 631 } re_match_context_t;
 632
 633 typedef struct
 634 {
 635   re_dfastate_t **sifted_states;
 636   re_dfastate_t **limited_states;
 637   Idx last_node;
 638   Idx last_str_idx;
 639   re_node_set limits;
 640 } re_sift_context_t;
 641
 642 struct re_fail_stack_ent_t
 643 {
 644   Idx idx;
 645   Idx node;
 646   regmatch_t *regs;
 647   re_node_set eps_via_nodes;
 648 };
 649
 650 struct re_fail_stack_t
 651 {
 652   Idx num;
 653   Idx alloc;
 654   struct re_fail_stack_ent_t *stack;
 655 };
 656
 657 struct re_dfa_t
 658 {
 659   re_token_t *nodes;
 660   size_t nodes_alloc;
 661   size_t nodes_len;
 662   Idx *nexts;
 663   Idx *org_indices;
 664   re_node_set *edests;
 665   re_node_set *eclosures;
 666   re_node_set *inveclosures;
 667   struct re_state_table_entry *state_table;
 668   re_dfastate_t *init_state;
 669   re_dfastate_t *init_state_word;
 670   re_dfastate_t *init_state_nl;
 671   re_dfastate_t *init_state_begbuf;
 672   bin_tree_t *str_tree;
 673   bin_tree_storage_t *str_tree_storage;
 674   re_bitset_ptr_t sb_char;
 675   int str_tree_storage_idx;
 676
 677   /* number of subexpressions `re_nsub' is in regex_t.  */
 678   re_hashval_t state_hash_mask;
 679   Idx init_node;
 680   Idx nbackref; /* The number of backreference in this dfa.  */
 681
 682   /* Bitmap expressing which backreference is used.  */
 683   bitset_word_t used_bkref_map;
 684   bitset_word_t completed_bkref_map;
 685
 686   unsigned int has_plural_match : 1;
 687   /* If this dfa has "multibyte node", which is a backreference or
 688      a node which can accept multibyte character or multi character
 689      collating element.  */
 690   unsigned int has_mb_node : 1;
 691   unsigned int is_utf8 : 1;
 692   unsigned int map_notascii : 1;
 693   unsigned int word_ops_used : 1;
 694   int mb_cur_max;
 695   bitset_t word_char;
 696   reg_syntax_t syntax;
 697   Idx *subexp_map;
 698 #ifdef DEBUG
 699   char* re_str;
 700 #endif
 701 #ifdef _LIBC
 702   __libc_lock_define (, lock)
 703 #endif
 704 };
 705
 706 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
 707 #define re_node_set_remove(set,id) \
 708   (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
 709 #define re_node_set_empty(p) ((p)->nelem = 0)
 710 #define re_node_set_free(set) re_free ((set)->elems)
 711 \f
 712
 713 typedef enum
 714 {
 715   SB_CHAR,
 716   MB_CHAR,
 717   EQUIV_CLASS,
 718   COLL_SYM,
 719   CHAR_CLASS
 720 } bracket_elem_type;
 721
 722 typedef struct
 723 {
 724   bracket_elem_type type;
 725   union
 726   {
 727     unsigned char ch;
 728     unsigned char *name;
 729     wchar_t wch;
 730   } opr;
 731 } bracket_elem_t;
 732
 733
 734 /* Inline functions for bitset_t operation.  */
 735
 736 static inline void
 737 bitset_set (bitset_t set, Idx i)
 738 {
 739   set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS;
 740 }
 741
 742 static inline void
 743 bitset_clear (bitset_t set, Idx i)
 744 {
 745   set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS);
 746 }
 747
 748 static inline bool
 749 bitset_contain (const bitset_t set, Idx i)
 750 {
 751   return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1;
 752 }
 753
 754 static inline void
 755 bitset_empty (bitset_t set)
 756 {
 757   memset (set, '\0', sizeof (bitset_t));
 758 }
 759
 760 static inline void
 761 bitset_set_all (bitset_t set)
 762 {
 763   memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS));
 764   if (SBC_MAX % BITSET_WORD_BITS != 0)
 765     set[BITSET_WORDS - 1] =
 766       ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1;
 767 }
 768
 769 static inline void
 770 bitset_copy (bitset_t dest, const bitset_t src)
 771 {
 772   memcpy (dest, src, sizeof (bitset_t));
 773 }
 774
 775 static inline void
 776 bitset_not (bitset_t set)
 777 {
 778   int bitset_i;
 779   for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i)
 780     set[bitset_i] = ~set[bitset_i];
 781   if (SBC_MAX % BITSET_WORD_BITS != 0)
 782     set[BITSET_WORDS - 1] =
 783       ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1)
 784        & ~set[BITSET_WORDS - 1]);
 785 }
 786
 787 static inline void
 788 bitset_merge (bitset_t dest, const bitset_t src)
 789 {
 790   int bitset_i;
 791   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 792     dest[bitset_i] |= src[bitset_i];
 793 }
 794
 795 static inline void
 796 bitset_mask (bitset_t dest, const bitset_t src)
 797 {
 798   int bitset_i;
 799   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 800     dest[bitset_i] &= src[bitset_i];
 801 }
 802
 803 #ifdef RE_ENABLE_I18N
 804 /* Inline functions for re_string.  */
 805 static inline int
 806 internal_function __attribute ((pure))
 807 re_string_char_size_at (const re_string_t *pstr, Idx idx)
 808 {
 809   int byte_idx;
 810   if (pstr->mb_cur_max == 1)
 811     return 1;
 812   for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
 813     if (pstr->wcs[idx + byte_idx] != WEOF)
 814       break;
 815   return byte_idx;
 816 }
 817
 818 static inline wint_t
 819 internal_function __attribute ((pure))
 820 re_string_wchar_at (const re_string_t *pstr, Idx idx)
 821 {
 822   if (pstr->mb_cur_max == 1)
 823     return (wint_t) pstr->mbs[idx];
 824   return (wint_t) pstr->wcs[idx];
 825 }
 826
 827 static int
 828 internal_function __attribute ((pure))
 829 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
 830 {
 831 # ifdef _LIBC
 832   const unsigned char *p, *extra;
 833   const int32_t *table, *indirect;
 834   int32_t tmp;
 835 #  include <locale/weight.h>
 836   uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
 837
 838   if (nrules != 0)
 839     {
 840       table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
 841       extra = (const unsigned char *)
 842         _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
 843       indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
 844                                                 _NL_COLLATE_INDIRECTMB);
 845       p = pstr->mbs + idx;
 846       tmp = findidx (&p);
 847       return p - pstr->mbs - idx;
 848     }
 849   else
 850 # endif /* _LIBC */
 851     return 1;
 852 }
 853 #endif /* RE_ENABLE_I18N */
 854
 855 #endif /*  _REGEX_INTERNAL_H */