posix/regex_internal.h

   1 /* Extended regular expression matching and search library.
   2    Copyright (C) 2002-2013 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    The GNU C Library is free software; you can redistribute it and/or
   7    modify it under the terms of the GNU Lesser General Public
   8    License as published by the Free Software Foundation; either
   9    version 2.1 of the License, or (at your option) any later version.
  10
  11    The GNU C Library 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 GNU
  14    Lesser General Public License for more details.
  15
  16    You should have received a copy of the GNU Lesser General Public
  17    License along with the GNU C Library; if not, see
  18    <http://www.gnu.org/licenses/>.  */
  19
  20 #ifndef _REGEX_INTERNAL_H
  21 #define _REGEX_INTERNAL_H 1
  22
  23 #include <assert.h>
  24 #include <ctype.h>
  25 #include <stdio.h>
  26 #include <stdlib.h>
  27 #include <string.h>
  28
  29 #if defined HAVE_LANGINFO_H || defined HAVE_LANGINFO_CODESET || defined _LIBC
  30 # include <langinfo.h>
  31 #endif
  32 #if defined HAVE_LOCALE_H || defined _LIBC
  33 # include <locale.h>
  34 #endif
  35 #if defined HAVE_WCHAR_H || defined _LIBC
  36 # include <wchar.h>
  37 #endif /* HAVE_WCHAR_H || _LIBC */
  38 #if defined HAVE_WCTYPE_H || defined _LIBC
  39 # include <wctype.h>
  40 #endif /* HAVE_WCTYPE_H || _LIBC */
  41 #if defined HAVE_STDBOOL_H || defined _LIBC
  42 # include <stdbool.h>
  43 #endif /* HAVE_STDBOOL_H || _LIBC */
  44 #if defined HAVE_STDINT_H || defined _LIBC
  45 # include <stdint.h>
  46 #endif /* HAVE_STDINT_H || _LIBC */
  47 #if defined _LIBC
  48 # include <bits/libc-lock.h>
  49 #else
  50 # define __libc_lock_define(CLASS,NAME)
  51 # define __libc_lock_init(NAME) do { } while (0)
  52 # define __libc_lock_lock(NAME) do { } while (0)
  53 # define __libc_lock_unlock(NAME) do { } while (0)
  54 #endif
  55
  56 /* In case that the system doesn't have isblank().  */
  57 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
  58 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
  59 #endif
  60
  61 #ifdef _LIBC
  62 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
  63 #  define _RE_DEFINE_LOCALE_FUNCTIONS 1
  64 #   include <locale/localeinfo.h>
  65 #   include <locale/elem-hash.h>
  66 #   include <locale/coll-lookup.h>
  67 # endif
  68 #endif
  69
  70 /* This is for other GNU distributions with internationalized messages.  */
  71 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
  72 # include <libintl.h>
  73 # ifdef _LIBC
  74 #  undef gettext
  75 #  define gettext(msgid) \
  76   __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES)
  77 # endif
  78 #else
  79 # define gettext(msgid) (msgid)
  80 #endif
  81
  82 #ifndef gettext_noop
  83 /* This define is so xgettext can find the internationalizable
  84    strings.  */
  85 # define gettext_noop(String) String
  86 #endif
  87
  88 /* For loser systems without the definition.  */
  89 #ifndef SIZE_MAX
  90 # define SIZE_MAX ((size_t) -1)
  91 #endif
  92
  93 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
  94 # define RE_ENABLE_I18N
  95 #endif
  96
  97 #if __GNUC__ >= 3
  98 # define BE(expr, val) __builtin_expect (expr, val)
  99 #else
 100 # define BE(expr, val) (expr)
 101 #endif
 102
 103 /* Number of single byte character.  */
 104 #define SBC_MAX 256
 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 __mbrtowc mbrtowc
 118 # define __mempcpy mempcpy
 119 # define __wcrtomb wcrtomb
 120 # define __regfree regfree
 121 # define attribute_hidden
 122 #endif /* not _LIBC */
 123
 124 #ifdef __GNUC__
 125 # define __attribute(arg) __attribute__ (arg)
 126 #else
 127 # define __attribute(arg)
 128 #endif
 129
 130 extern const char __re_error_msgid[] attribute_hidden;
 131 extern const size_t __re_error_msgid_idx[] attribute_hidden;
 132
 133 /* An integer used to represent a set of bits.  It must be unsigned,
 134    and must be at least as wide as unsigned int.  */
 135 typedef unsigned long int bitset_word_t;
 136 /* All bits set in a bitset_word_t.  */
 137 #define BITSET_WORD_MAX ULONG_MAX
 138 /* Number of bits in a bitset_word_t.  */
 139 #define BITSET_WORD_BITS (sizeof (bitset_word_t) * CHAR_BIT)
 140 /* Number of bitset_word_t in a bit_set.  */
 141 #define BITSET_WORDS (SBC_MAX / BITSET_WORD_BITS)
 142 typedef bitset_word_t bitset_t[BITSET_WORDS];
 143 typedef bitset_word_t *re_bitset_ptr_t;
 144 typedef const bitset_word_t *re_const_bitset_ptr_t;
 145
 146 #define bitset_set(set,i) \
 147   (set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS)
 148 #define bitset_clear(set,i) \
 149   (set[i / BITSET_WORD_BITS] &= ~((bitset_word_t) 1 << i % BITSET_WORD_BITS))
 150 #define bitset_contain(set,i) \
 151   (set[i / BITSET_WORD_BITS] & ((bitset_word_t) 1 << i % BITSET_WORD_BITS))
 152 #define bitset_empty(set) memset (set, '\0', sizeof (bitset_t))
 153 #define bitset_set_all(set) memset (set, '\xff', sizeof (bitset_t))
 154 #define bitset_copy(dest,src) memcpy (dest, src, sizeof (bitset_t))
 155
 156 #define PREV_WORD_CONSTRAINT 0x0001
 157 #define PREV_NOTWORD_CONSTRAINT 0x0002
 158 #define NEXT_WORD_CONSTRAINT 0x0004
 159 #define NEXT_NOTWORD_CONSTRAINT 0x0008
 160 #define PREV_NEWLINE_CONSTRAINT 0x0010
 161 #define NEXT_NEWLINE_CONSTRAINT 0x0020
 162 #define PREV_BEGBUF_CONSTRAINT 0x0040
 163 #define NEXT_ENDBUF_CONSTRAINT 0x0080
 164 #define WORD_DELIM_CONSTRAINT 0x0100
 165 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
 166
 167 typedef enum
 168 {
 169   INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 170   WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 171   WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 172   INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 173   LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
 174   LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
 175   BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
 176   BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
 177   WORD_DELIM = WORD_DELIM_CONSTRAINT,
 178   NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
 179 } re_context_type;
 180
 181 typedef struct
 182 {
 183   int alloc;
 184   int nelem;
 185   int *elems;
 186 } re_node_set;
 187
 188 typedef enum
 189 {
 190   NON_TYPE = 0,
 191
 192   /* Node type, These are used by token, node, tree.  */
 193   CHARACTER = 1,
 194   END_OF_RE = 2,
 195   SIMPLE_BRACKET = 3,
 196   OP_BACK_REF = 4,
 197   OP_PERIOD = 5,
 198 #ifdef RE_ENABLE_I18N
 199   COMPLEX_BRACKET = 6,
 200   OP_UTF8_PERIOD = 7,
 201 #endif /* RE_ENABLE_I18N */
 202
 203   /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
 204      when the debugger shows values of this enum type.  */
 205 #define EPSILON_BIT 8
 206   OP_OPEN_SUBEXP = EPSILON_BIT | 0,
 207   OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
 208   OP_ALT = EPSILON_BIT | 2,
 209   OP_DUP_ASTERISK = EPSILON_BIT | 3,
 210   ANCHOR = EPSILON_BIT | 4,
 211
 212   /* Tree type, these are used only by tree. */
 213   CONCAT = 16,
 214   SUBEXP = 17,
 215
 216   /* Token type, these are used only by token.  */
 217   OP_DUP_PLUS = 18,
 218   OP_DUP_QUESTION,
 219   OP_OPEN_BRACKET,
 220   OP_CLOSE_BRACKET,
 221   OP_CHARSET_RANGE,
 222   OP_OPEN_DUP_NUM,
 223   OP_CLOSE_DUP_NUM,
 224   OP_NON_MATCH_LIST,
 225   OP_OPEN_COLL_ELEM,
 226   OP_CLOSE_COLL_ELEM,
 227   OP_OPEN_EQUIV_CLASS,
 228   OP_CLOSE_EQUIV_CLASS,
 229   OP_OPEN_CHAR_CLASS,
 230   OP_CLOSE_CHAR_CLASS,
 231   OP_WORD,
 232   OP_NOTWORD,
 233   OP_SPACE,
 234   OP_NOTSPACE,
 235   BACK_SLASH
 236
 237 } re_token_type_t;
 238
 239 #ifdef RE_ENABLE_I18N
 240 typedef struct
 241 {
 242   /* Multibyte characters.  */
 243   wchar_t *mbchars;
 244
 245   /* Collating symbols.  */
 246 # ifdef _LIBC
 247   int32_t *coll_syms;
 248 # endif
 249
 250   /* Equivalence classes. */
 251 # ifdef _LIBC
 252   int32_t *equiv_classes;
 253 # endif
 254
 255   /* Range expressions. */
 256 # ifdef _LIBC
 257   uint32_t *range_starts;
 258   uint32_t *range_ends;
 259 # else /* not _LIBC */
 260   wchar_t *range_starts;
 261   wchar_t *range_ends;
 262 # endif /* not _LIBC */
 263
 264   /* Character classes. */
 265   wctype_t *char_classes;
 266
 267   /* If this character set is the non-matching list.  */
 268   unsigned int non_match : 1;
 269
 270   /* # of multibyte characters.  */
 271   int nmbchars;
 272
 273   /* # of collating symbols.  */
 274   int ncoll_syms;
 275
 276   /* # of equivalence classes. */
 277   int nequiv_classes;
 278
 279   /* # of range expressions. */
 280   int nranges;
 281
 282   /* # of character classes. */
 283   int nchar_classes;
 284 } re_charset_t;
 285 #endif /* RE_ENABLE_I18N */
 286
 287 typedef struct
 288 {
 289   union
 290   {
 291     unsigned char c;            /* for CHARACTER */
 292     re_bitset_ptr_t sbcset;     /* for SIMPLE_BRACKET */
 293 #ifdef RE_ENABLE_I18N
 294     re_charset_t *mbcset;       /* for COMPLEX_BRACKET */
 295 #endif /* RE_ENABLE_I18N */
 296     int idx;                    /* for BACK_REF */
 297     re_context_type ctx_type;   /* for ANCHOR */
 298   } opr;
 299 #if __GNUC__ >= 2
 300   re_token_type_t type : 8;
 301 #else
 302   re_token_type_t type;
 303 #endif
 304   unsigned int constraint : 10; /* context constraint */
 305   unsigned int duplicated : 1;
 306   unsigned int opt_subexp : 1;
 307 #ifdef RE_ENABLE_I18N
 308   unsigned int accept_mb : 1;
 309   /* These 2 bits can be moved into the union if needed (e.g. if running out
 310      of bits; move opr.c to opr.c.c and move the flags to opr.c.flags).  */
 311   unsigned int mb_partial : 1;
 312 #endif
 313   unsigned int word_char : 1;
 314 } re_token_t;
 315
 316 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
 317
 318 struct re_string_t
 319 {
 320   /* Indicate the raw buffer which is the original string passed as an
 321      argument of regexec(), re_search(), etc..  */
 322   const unsigned char *raw_mbs;
 323   /* Store the multibyte string.  In case of "case insensitive mode" like
 324      REG_ICASE, upper cases of the string are stored, otherwise MBS points
 325      the same address that RAW_MBS points.  */
 326   unsigned char *mbs;
 327 #ifdef RE_ENABLE_I18N
 328   /* Store the wide character string which is corresponding to MBS.  */
 329   wint_t *wcs;
 330   int *offsets;
 331   mbstate_t cur_state;
 332 #endif
 333   /* Index in RAW_MBS.  Each character mbs[i] corresponds to
 334      raw_mbs[raw_mbs_idx + i].  */
 335   int raw_mbs_idx;
 336   /* The length of the valid characters in the buffers.  */
 337   int valid_len;
 338   /* The corresponding number of bytes in raw_mbs array.  */
 339   int valid_raw_len;
 340   /* The length of the buffers MBS and WCS.  */
 341   int bufs_len;
 342   /* The index in MBS, which is updated by re_string_fetch_byte.  */
 343   int cur_idx;
 344   /* length of RAW_MBS array.  */
 345   int raw_len;
 346   /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN.  */
 347   int len;
 348   /* End of the buffer may be shorter than its length in the cases such
 349      as re_match_2, re_search_2.  Then, we use STOP for end of the buffer
 350      instead of LEN.  */
 351   int raw_stop;
 352   /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS.  */
 353   int stop;
 354
 355   /* The context of mbs[0].  We store the context independently, since
 356      the context of mbs[0] may be different from raw_mbs[0], which is
 357      the beginning of the input string.  */
 358   unsigned int tip_context;
 359   /* The translation passed as a part of an argument of re_compile_pattern.  */
 360   RE_TRANSLATE_TYPE trans;
 361   /* Copy of re_dfa_t's word_char.  */
 362   re_const_bitset_ptr_t word_char;
 363   /* 1 if REG_ICASE.  */
 364   unsigned char icase;
 365   unsigned char is_utf8;
 366   unsigned char map_notascii;
 367   unsigned char mbs_allocated;
 368   unsigned char offsets_needed;
 369   unsigned char newline_anchor;
 370   unsigned char word_ops_used;
 371   int mb_cur_max;
 372 };
 373 typedef struct re_string_t re_string_t;
 374
 375
 376 struct re_dfa_t;
 377 typedef struct re_dfa_t re_dfa_t;
 378
 379 #ifndef _LIBC
 380 # ifdef __i386__
 381 #  define internal_function   __attribute ((regparm (3), stdcall))
 382 # else
 383 #  define internal_function
 384 # endif
 385 #endif
 386
 387 #ifndef NOT_IN_libc
 388 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
 389                                                 int new_buf_len)
 390      internal_function;
 391 # ifdef RE_ENABLE_I18N
 392 static void build_wcs_buffer (re_string_t *pstr) internal_function;
 393 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
 394   internal_function;
 395 # endif /* RE_ENABLE_I18N */
 396 static void build_upper_buffer (re_string_t *pstr) internal_function;
 397 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
 398 static unsigned int re_string_context_at (const re_string_t *input, int idx,
 399                                           int eflags)
 400      internal_function __attribute ((pure));
 401 #endif
 402 #define re_string_peek_byte(pstr, offset) \
 403   ((pstr)->mbs[(pstr)->cur_idx + offset])
 404 #define re_string_fetch_byte(pstr) \
 405   ((pstr)->mbs[(pstr)->cur_idx++])
 406 #define re_string_first_byte(pstr, idx) \
 407   ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
 408 #define re_string_is_single_byte_char(pstr, idx) \
 409   ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
 410                                 || (pstr)->wcs[(idx) + 1] != WEOF))
 411 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
 412 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
 413 #define re_string_get_buffer(pstr) ((pstr)->mbs)
 414 #define re_string_length(pstr) ((pstr)->len)
 415 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
 416 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
 417 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
 418
 419 #include <alloca.h>
 420
 421 #ifndef _LIBC
 422 # if HAVE_ALLOCA
 423 /* The OS usually guarantees only one guard page at the bottom of the stack,
 424    and a page size can be as small as 4096 bytes.  So we cannot safely
 425    allocate anything larger than 4096 bytes.  Also care for the possibility
 426    of a few compiler-allocated temporary stack slots.  */
 427 #  define __libc_use_alloca(n) ((n) < 4032)
 428 # else
 429 /* alloca is implemented with malloc, so just use malloc.  */
 430 #  define __libc_use_alloca(n) 0
 431 # endif
 432 #endif
 433
 434 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
 435 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
 436 #define re_free(p) free (p)
 437
 438 struct bin_tree_t
 439 {
 440   struct bin_tree_t *parent;
 441   struct bin_tree_t *left;
 442   struct bin_tree_t *right;
 443   struct bin_tree_t *first;
 444   struct bin_tree_t *next;
 445
 446   re_token_t token;
 447
 448   /* `node_idx' is the index in dfa->nodes, if `type' == 0.
 449      Otherwise `type' indicate the type of this node.  */
 450   int node_idx;
 451 };
 452 typedef struct bin_tree_t bin_tree_t;
 453
 454 #define BIN_TREE_STORAGE_SIZE \
 455   ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
 456
 457 struct bin_tree_storage_t
 458 {
 459   struct bin_tree_storage_t *next;
 460   bin_tree_t data[BIN_TREE_STORAGE_SIZE];
 461 };
 462 typedef struct bin_tree_storage_t bin_tree_storage_t;
 463
 464 #define CONTEXT_WORD 1
 465 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
 466 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
 467 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
 468
 469 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
 470 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
 471 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
 472 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
 473 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
 474
 475 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
 476 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
 477 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
 478 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
 479
 480 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
 481  ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 482   || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 483   || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
 484   || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
 485
 486 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
 487  ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 488   || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 489   || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
 490   || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
 491
 492 struct re_dfastate_t
 493 {
 494   unsigned int hash;
 495   re_node_set nodes;
 496   re_node_set non_eps_nodes;
 497   re_node_set inveclosure;
 498   re_node_set *entrance_nodes;
 499   struct re_dfastate_t **trtable, **word_trtable;
 500   unsigned int context : 4;
 501   unsigned int halt : 1;
 502   /* If this state can accept `multi byte'.
 503      Note that we refer to multibyte characters, and multi character
 504      collating elements as `multi byte'.  */
 505   unsigned int accept_mb : 1;
 506   /* If this state has backreference node(s).  */
 507   unsigned int has_backref : 1;
 508   unsigned int has_constraint : 1;
 509 };
 510 typedef struct re_dfastate_t re_dfastate_t;
 511
 512 struct re_state_table_entry
 513 {
 514   int num;
 515   int alloc;
 516   re_dfastate_t **array;
 517 };
 518
 519 /* Array type used in re_sub_match_last_t and re_sub_match_top_t.  */
 520
 521 typedef struct
 522 {
 523   int next_idx;
 524   int alloc;
 525   re_dfastate_t **array;
 526 } state_array_t;
 527
 528 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP.  */
 529
 530 typedef struct
 531 {
 532   int node;
 533   int str_idx; /* The position NODE match at.  */
 534   state_array_t path;
 535 } re_sub_match_last_t;
 536
 537 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
 538    And information about the node, whose type is OP_CLOSE_SUBEXP,
 539    corresponding to NODE is stored in LASTS.  */
 540
 541 typedef struct
 542 {
 543   int str_idx;
 544   int node;
 545   state_array_t *path;
 546   int alasts; /* Allocation size of LASTS.  */
 547   int nlasts; /* The number of LASTS.  */
 548   re_sub_match_last_t **lasts;
 549 } re_sub_match_top_t;
 550
 551 struct re_backref_cache_entry
 552 {
 553   int node;
 554   int str_idx;
 555   int subexp_from;
 556   int subexp_to;
 557   char more;
 558   char unused;
 559   unsigned short int eps_reachable_subexps_map;
 560 };
 561
 562 typedef struct
 563 {
 564   /* The string object corresponding to the input string.  */
 565   re_string_t input;
 566 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
 567   const re_dfa_t *const dfa;
 568 #else
 569   const re_dfa_t *dfa;
 570 #endif
 571   /* EFLAGS of the argument of regexec.  */
 572   int eflags;
 573   /* Where the matching ends.  */
 574   int match_last;
 575   int last_node;
 576   /* The state log used by the matcher.  */
 577   re_dfastate_t **state_log;
 578   int state_log_top;
 579   /* Back reference cache.  */
 580   int nbkref_ents;
 581   int abkref_ents;
 582   struct re_backref_cache_entry *bkref_ents;
 583   int max_mb_elem_len;
 584   int nsub_tops;
 585   int asub_tops;
 586   re_sub_match_top_t **sub_tops;
 587 } re_match_context_t;
 588
 589 typedef struct
 590 {
 591   re_dfastate_t **sifted_states;
 592   re_dfastate_t **limited_states;
 593   int last_node;
 594   int last_str_idx;
 595   re_node_set limits;
 596 } re_sift_context_t;
 597
 598 struct re_fail_stack_ent_t
 599 {
 600   int idx;
 601   int node;
 602   regmatch_t *regs;
 603   re_node_set eps_via_nodes;
 604 };
 605
 606 struct re_fail_stack_t
 607 {
 608   int num;
 609   int alloc;
 610   struct re_fail_stack_ent_t *stack;
 611 };
 612
 613 struct re_dfa_t
 614 {
 615   re_token_t *nodes;
 616   size_t nodes_alloc;
 617   size_t nodes_len;
 618   int *nexts;
 619   int *org_indices;
 620   re_node_set *edests;
 621   re_node_set *eclosures;
 622   re_node_set *inveclosures;
 623   struct re_state_table_entry *state_table;
 624   re_dfastate_t *init_state;
 625   re_dfastate_t *init_state_word;
 626   re_dfastate_t *init_state_nl;
 627   re_dfastate_t *init_state_begbuf;
 628   bin_tree_t *str_tree;
 629   bin_tree_storage_t *str_tree_storage;
 630   re_bitset_ptr_t sb_char;
 631   int str_tree_storage_idx;
 632
 633   /* number of subexpressions `re_nsub' is in regex_t.  */
 634   unsigned int state_hash_mask;
 635   int init_node;
 636   int nbackref; /* The number of backreference in this dfa.  */
 637
 638   /* Bitmap expressing which backreference is used.  */
 639   bitset_word_t used_bkref_map;
 640   bitset_word_t completed_bkref_map;
 641
 642   unsigned int has_plural_match : 1;
 643   /* If this dfa has "multibyte node", which is a backreference or
 644      a node which can accept multibyte character or multi character
 645      collating element.  */
 646   unsigned int has_mb_node : 1;
 647   unsigned int is_utf8 : 1;
 648   unsigned int map_notascii : 1;
 649   unsigned int word_ops_used : 1;
 650   int mb_cur_max;
 651   bitset_t word_char;
 652   reg_syntax_t syntax;
 653   int *subexp_map;
 654 #ifdef DEBUG
 655   char* re_str;
 656 #endif
 657   __libc_lock_define (, lock)
 658 };
 659
 660 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
 661 #define re_node_set_remove(set,id) \
 662   (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
 663 #define re_node_set_empty(p) ((p)->nelem = 0)
 664 #define re_node_set_free(set) re_free ((set)->elems)
 665 \f
 666
 667 typedef enum
 668 {
 669   SB_CHAR,
 670   MB_CHAR,
 671   EQUIV_CLASS,
 672   COLL_SYM,
 673   CHAR_CLASS
 674 } bracket_elem_type;
 675
 676 typedef struct
 677 {
 678   bracket_elem_type type;
 679   union
 680   {
 681     unsigned char ch;
 682     unsigned char *name;
 683     wchar_t wch;
 684   } opr;
 685 } bracket_elem_t;
 686
 687
 688 /* Inline functions for bitset operation.  */
 689 static void
 690 bitset_not (bitset_t set)
 691 {
 692   int bitset_i;
 693   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 694     set[bitset_i] = ~set[bitset_i];
 695 }
 696
 697 static void
 698 bitset_merge (bitset_t dest, const bitset_t src)
 699 {
 700   int bitset_i;
 701   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 702     dest[bitset_i] |= src[bitset_i];
 703 }
 704
 705 static void
 706 bitset_mask (bitset_t dest, const bitset_t src)
 707 {
 708   int bitset_i;
 709   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 710     dest[bitset_i] &= src[bitset_i];
 711 }
 712
 713 #ifdef RE_ENABLE_I18N
 714 /* Inline functions for re_string.  */
 715 static int
 716 internal_function __attribute ((pure))
 717 re_string_char_size_at (const re_string_t *pstr, int idx)
 718 {
 719   int byte_idx;
 720   if (pstr->mb_cur_max == 1)
 721     return 1;
 722   for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
 723     if (pstr->wcs[idx + byte_idx] != WEOF)
 724       break;
 725   return byte_idx;
 726 }
 727
 728 static wint_t
 729 internal_function __attribute ((pure))
 730 re_string_wchar_at (const re_string_t *pstr, int idx)
 731 {
 732   if (pstr->mb_cur_max == 1)
 733     return (wint_t) pstr->mbs[idx];
 734   return (wint_t) pstr->wcs[idx];
 735 }
 736
 737 # ifndef NOT_IN_libc
 738 static int
 739 internal_function __attribute ((pure))
 740 re_string_elem_size_at (const re_string_t *pstr, int idx)
 741 {
 742 #  ifdef _LIBC
 743   const unsigned char *p, *extra;
 744   const int32_t *table, *indirect;
 745 #   include <locale/weight.h>
 746   uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
 747
 748   if (nrules != 0)
 749     {
 750       table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
 751       extra = (const unsigned char *)
 752         _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
 753       indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
 754                                                 _NL_COLLATE_INDIRECTMB);
 755       p = pstr->mbs + idx;
 756       findidx (&p, pstr->len - idx);
 757       return p - pstr->mbs - idx;
 758     }
 759   else
 760 #  endif /* _LIBC */
 761     return 1;
 762 }
 763 # endif
 764 #endif /* RE_ENABLE_I18N */
 765
 766 #endif /*  _REGEX_INTERNAL_H */