posix/regex_internal.h

   1 /* Extended regular expression matching and search library.
   2    Copyright (C) 2002, 2003, 2004, 2005 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, write to the Free
  18    Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
  19    02111-1307 USA.  */
  20
  21 #ifndef _REGEX_INTERNAL_H
  22 #define _REGEX_INTERNAL_H 1
  23
  24 #include <assert.h>
  25 #include <ctype.h>
  26 #include <stdio.h>
  27 #include <stdlib.h>
  28 #include <string.h>
  29
  30 #if defined HAVE_LANGINFO_H || defined HAVE_LANGINFO_CODESET || defined _LIBC
  31 # include <langinfo.h>
  32 #endif
  33 #if defined HAVE_LOCALE_H || defined _LIBC
  34 # include <locale.h>
  35 #endif
  36 #if defined HAVE_WCHAR_H || defined _LIBC
  37 # include <wchar.h>
  38 #endif /* HAVE_WCHAR_H || _LIBC */
  39 #if defined HAVE_WCTYPE_H || defined _LIBC
  40 # include <wctype.h>
  41 #endif /* HAVE_WCTYPE_H || _LIBC */
  42 #if defined HAVE_STDBOOL_H || defined _LIBC
  43 # include <stdbool.h>
  44 #endif /* HAVE_STDBOOL_H || _LIBC */
  45 #if defined HAVE_STDINT_H || defined _LIBC
  46 # include <stdint.h>
  47 #endif /* HAVE_STDINT_H || _LIBC */
  48 #if defined _LIBC
  49 # include <bits/libc-lock.h>
  50 #else
  51 # define __libc_lock_define(CLASS,NAME)
  52 # define __libc_lock_init(NAME) do { } while (0)
  53 # define __libc_lock_lock(NAME) do { } while (0)
  54 # define __libc_lock_unlock(NAME) do { } while (0)
  55 #endif
  56
  57 /* In case that the system doesn't have isblank().  */
  58 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
  59 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
  60 #endif
  61
  62 #ifdef _LIBC
  63 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
  64 #  define _RE_DEFINE_LOCALE_FUNCTIONS 1
  65 #   include <locale/localeinfo.h>
  66 #   include <locale/elem-hash.h>
  67 #   include <locale/coll-lookup.h>
  68 # endif
  69 #endif
  70
  71 /* This is for other GNU distributions with internationalized messages.  */
  72 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
  73 # include <libintl.h>
  74 # ifdef _LIBC
  75 #  undef gettext
  76 #  define gettext(msgid) \
  77   INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
  78 # endif
  79 #else
  80 # define gettext(msgid) (msgid)
  81 #endif
  82
  83 #ifndef gettext_noop
  84 /* This define is so xgettext can find the internationalizable
  85    strings.  */
  86 # define gettext_noop(String) String
  87 #endif
  88
  89 /* For loser systems without the definition.  */
  90 #ifndef SIZE_MAX
  91 # define SIZE_MAX ((size_t) -1)
  92 #endif
  93
  94 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
  95 # define RE_ENABLE_I18N
  96 #endif
  97
  98 #if __GNUC__ >= 3
  99 # define BE(expr, val) __builtin_expect (expr, val)
 100 #else
 101 # define BE(expr, val) (expr)
 102 # define inline
 103 #endif
 104
 105 /* Number of single byte character.  */
 106 #define SBC_MAX 256
 107
 108 #define COLL_ELEM_LEN_MAX 8
 109
 110 /* The character which represents newline.  */
 111 #define NEWLINE_CHAR '\n'
 112 #define WIDE_NEWLINE_CHAR L'\n'
 113
 114 /* Rename to standard API for using out of glibc.  */
 115 #ifndef _LIBC
 116 # define __wctype wctype
 117 # define __iswctype iswctype
 118 # define __btowc btowc
 119 # define __mempcpy mempcpy
 120 # define __wcrtomb wcrtomb
 121 # define __regfree regfree
 122 # define attribute_hidden
 123 #endif /* not _LIBC */
 124
 125 #ifdef __GNUC__
 126 # define __attribute(arg) __attribute__ (arg)
 127 #else
 128 # define __attribute(arg)
 129 #endif
 130
 131 extern const char __re_error_msgid[] attribute_hidden;
 132 extern const size_t __re_error_msgid_idx[] attribute_hidden;
 133
 134 /* An integer used to represent a set of bits.  It must be unsigned,
 135    and must be at least as wide as unsigned int.  */
 136 typedef unsigned long int bitset_word_t;
 137 /* All bits set in a bitset_word_t.  */
 138 #define BITSET_WORD_MAX ULONG_MAX
 139 /* Number of bits in a bitset_word_t.  */
 140 #define BITSET_WORD_BITS (sizeof (bitset_word_t) * CHAR_BIT)
 141 /* Number of bitset_word_t in a bit_set.  */
 142 #define BITSET_WORDS (SBC_MAX / BITSET_WORD_BITS)
 143 typedef bitset_word_t bitset_t[BITSET_WORDS];
 144 typedef bitset_word_t *re_bitset_ptr_t;
 145 typedef const bitset_word_t *re_const_bitset_ptr_t;
 146
 147 #define bitset_set(set,i) \
 148   (set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS)
 149 #define bitset_clear(set,i) \
 150   (set[i / BITSET_WORD_BITS] &= ~((bitset_word_t) 1 << i % BITSET_WORD_BITS))
 151 #define bitset_contain(set,i) \
 152   (set[i / BITSET_WORD_BITS] & ((bitset_word_t) 1 << i % BITSET_WORD_BITS))
 153 #define bitset_empty(set) memset (set, '\0', sizeof (bitset_t))
 154 #define bitset_set_all(set) memset (set, '\xff', sizeof (bitset_t))
 155 #define bitset_copy(dest,src) memcpy (dest, src, sizeof (bitset_t))
 156
 157 #define PREV_WORD_CONSTRAINT 0x0001
 158 #define PREV_NOTWORD_CONSTRAINT 0x0002
 159 #define NEXT_WORD_CONSTRAINT 0x0004
 160 #define NEXT_NOTWORD_CONSTRAINT 0x0008
 161 #define PREV_NEWLINE_CONSTRAINT 0x0010
 162 #define NEXT_NEWLINE_CONSTRAINT 0x0020
 163 #define PREV_BEGBUF_CONSTRAINT 0x0040
 164 #define NEXT_ENDBUF_CONSTRAINT 0x0080
 165 #define WORD_DELIM_CONSTRAINT 0x0100
 166 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
 167
 168 typedef enum
 169 {
 170   INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 171   WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 172   WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 173   INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 174   LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
 175   LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
 176   BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
 177   BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
 178   WORD_DELIM = WORD_DELIM_CONSTRAINT,
 179   NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
 180 } re_context_type;
 181
 182 typedef struct
 183 {
 184   int alloc;
 185   int nelem;
 186   int *elems;
 187 } re_node_set;
 188
 189 typedef enum
 190 {
 191   NON_TYPE = 0,
 192
 193   /* Node type, These are used by token, node, tree.  */
 194   CHARACTER = 1,
 195   END_OF_RE = 2,
 196   SIMPLE_BRACKET = 3,
 197   OP_BACK_REF = 4,
 198   OP_PERIOD = 5,
 199 #ifdef RE_ENABLE_I18N
 200   COMPLEX_BRACKET = 6,
 201   OP_UTF8_PERIOD = 7,
 202 #endif /* RE_ENABLE_I18N */
 203
 204   /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
 205      when the debugger shows values of this enum type.  */
 206 #define EPSILON_BIT 8
 207   OP_OPEN_SUBEXP = EPSILON_BIT | 0,
 208   OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
 209   OP_ALT = EPSILON_BIT | 2,
 210   OP_DUP_ASTERISK = EPSILON_BIT | 3,
 211   ANCHOR = EPSILON_BIT | 4,
 212
 213   /* Tree type, these are used only by tree. */
 214   CONCAT = 16,
 215   SUBEXP = 17,
 216
 217   /* Token type, these are used only by token.  */
 218   OP_DUP_PLUS = 18,
 219   OP_DUP_QUESTION,
 220   OP_OPEN_BRACKET,
 221   OP_CLOSE_BRACKET,
 222   OP_CHARSET_RANGE,
 223   OP_OPEN_DUP_NUM,
 224   OP_CLOSE_DUP_NUM,
 225   OP_NON_MATCH_LIST,
 226   OP_OPEN_COLL_ELEM,
 227   OP_CLOSE_COLL_ELEM,
 228   OP_OPEN_EQUIV_CLASS,
 229   OP_CLOSE_EQUIV_CLASS,
 230   OP_OPEN_CHAR_CLASS,
 231   OP_CLOSE_CHAR_CLASS,
 232   OP_WORD,
 233   OP_NOTWORD,
 234   OP_SPACE,
 235   OP_NOTSPACE,
 236   BACK_SLASH
 237
 238 } re_token_type_t;
 239
 240 #ifdef RE_ENABLE_I18N
 241 typedef struct
 242 {
 243   /* Multibyte characters.  */
 244   wchar_t *mbchars;
 245
 246   /* Collating symbols.  */
 247 # ifdef _LIBC
 248   int32_t *coll_syms;
 249 # endif
 250
 251   /* Equivalence classes. */
 252 # ifdef _LIBC
 253   int32_t *equiv_classes;
 254 # endif
 255
 256   /* Range expressions. */
 257 # ifdef _LIBC
 258   uint32_t *range_starts;
 259   uint32_t *range_ends;
 260 # else /* not _LIBC */
 261   wchar_t *range_starts;
 262   wchar_t *range_ends;
 263 # endif /* not _LIBC */
 264
 265   /* Character classes. */
 266   wctype_t *char_classes;
 267
 268   /* If this character set is the non-matching list.  */
 269   unsigned int non_match : 1;
 270
 271   /* # of multibyte characters.  */
 272   int nmbchars;
 273
 274   /* # of collating symbols.  */
 275   int ncoll_syms;
 276
 277   /* # of equivalence classes. */
 278   int nequiv_classes;
 279
 280   /* # of range expressions. */
 281   int nranges;
 282
 283   /* # of character classes. */
 284   int nchar_classes;
 285 } re_charset_t;
 286 #endif /* RE_ENABLE_I18N */
 287
 288 typedef struct
 289 {
 290   union
 291   {
 292     unsigned char c;            /* for CHARACTER */
 293     re_bitset_ptr_t sbcset;     /* for SIMPLE_BRACKET */
 294 #ifdef RE_ENABLE_I18N
 295     re_charset_t *mbcset;       /* for COMPLEX_BRACKET */
 296 #endif /* RE_ENABLE_I18N */
 297     int idx;                    /* for BACK_REF */
 298     re_context_type ctx_type;   /* for ANCHOR */
 299   } opr;
 300 #if __GNUC__ >= 2
 301   re_token_type_t type : 8;
 302 #else
 303   re_token_type_t type;
 304 #endif
 305   unsigned int constraint : 10; /* context constraint */
 306   unsigned int duplicated : 1;
 307   unsigned int opt_subexp : 1;
 308 #ifdef RE_ENABLE_I18N
 309   unsigned int accept_mb : 1;
 310   /* These 2 bits can be moved into the union if needed (e.g. if running out
 311      of bits; move opr.c to opr.c.c and move the flags to opr.c.flags).  */
 312   unsigned int mb_partial : 1;
 313 #endif
 314   unsigned int word_char : 1;
 315 } re_token_t;
 316
 317 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
 318
 319 struct re_string_t
 320 {
 321   /* Indicate the raw buffer which is the original string passed as an
 322      argument of regexec(), re_search(), etc..  */
 323   const unsigned char *raw_mbs;
 324   /* Store the multibyte string.  In case of "case insensitive mode" like
 325      REG_ICASE, upper cases of the string are stored, otherwise MBS points
 326      the same address that RAW_MBS points.  */
 327   unsigned char *mbs;
 328 #ifdef RE_ENABLE_I18N
 329   /* Store the wide character string which is corresponding to MBS.  */
 330   wint_t *wcs;
 331   int *offsets;
 332   mbstate_t cur_state;
 333 #endif
 334   /* Index in RAW_MBS.  Each character mbs[i] corresponds to
 335      raw_mbs[raw_mbs_idx + i].  */
 336   int raw_mbs_idx;
 337   /* The length of the valid characters in the buffers.  */
 338   int valid_len;
 339   /* The corresponding number of bytes in raw_mbs array.  */
 340   int valid_raw_len;
 341   /* The length of the buffers MBS and WCS.  */
 342   int bufs_len;
 343   /* The index in MBS, which is updated by re_string_fetch_byte.  */
 344   int cur_idx;
 345   /* length of RAW_MBS array.  */
 346   int raw_len;
 347   /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN.  */
 348   int len;
 349   /* End of the buffer may be shorter than its length in the cases such
 350      as re_match_2, re_search_2.  Then, we use STOP for end of the buffer
 351      instead of LEN.  */
 352   int raw_stop;
 353   /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS.  */
 354   int stop;
 355
 356   /* The context of mbs[0].  We store the context independently, since
 357      the context of mbs[0] may be different from raw_mbs[0], which is
 358      the beginning of the input string.  */
 359   unsigned int tip_context;
 360   /* The translation passed as a part of an argument of re_compile_pattern.  */
 361   RE_TRANSLATE_TYPE trans;
 362   /* Copy of re_dfa_t's word_char.  */
 363   re_const_bitset_ptr_t word_char;
 364   /* 1 if REG_ICASE.  */
 365   unsigned char icase;
 366   unsigned char is_utf8;
 367   unsigned char map_notascii;
 368   unsigned char mbs_allocated;
 369   unsigned char offsets_needed;
 370   unsigned char newline_anchor;
 371   unsigned char word_ops_used;
 372   int mb_cur_max;
 373 };
 374 typedef struct re_string_t re_string_t;
 375
 376
 377 struct re_dfa_t;
 378 typedef struct re_dfa_t re_dfa_t;
 379
 380 #ifndef _LIBC
 381 # ifdef __i386__
 382 #  define internal_function   __attribute ((regparm (3), stdcall))
 383 # else
 384 #  define internal_function
 385 # endif
 386 #endif
 387
 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 int build_wcs_upper_buffer (re_string_t *pstr) internal_function;
 394 #endif /* RE_ENABLE_I18N */
 395 static void build_upper_buffer (re_string_t *pstr) internal_function;
 396 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
 397 static unsigned int re_string_context_at (const re_string_t *input, int idx,
 398                                           int eflags)
 399      internal_function __attribute ((pure));
 400 #define re_string_peek_byte(pstr, offset) \
 401   ((pstr)->mbs[(pstr)->cur_idx + offset])
 402 #define re_string_fetch_byte(pstr) \
 403   ((pstr)->mbs[(pstr)->cur_idx++])
 404 #define re_string_first_byte(pstr, idx) \
 405   ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
 406 #define re_string_is_single_byte_char(pstr, idx) \
 407   ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
 408                                 || (pstr)->wcs[(idx) + 1] != WEOF))
 409 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
 410 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
 411 #define re_string_get_buffer(pstr) ((pstr)->mbs)
 412 #define re_string_length(pstr) ((pstr)->len)
 413 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
 414 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
 415 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
 416
 417 #include <alloca.h>
 418
 419 #ifndef _LIBC
 420 # if HAVE_ALLOCA
 421 /* The OS usually guarantees only one guard page at the bottom of the stack,
 422    and a page size can be as small as 4096 bytes.  So we cannot safely
 423    allocate anything larger than 4096 bytes.  Also care for the possibility
 424    of a few compiler-allocated temporary stack slots.  */
 425 #  define __libc_use_alloca(n) ((n) < 4032)
 426 # else
 427 /* alloca is implemented with malloc, so just use malloc.  */
 428 #  define __libc_use_alloca(n) 0
 429 # endif
 430 #endif
 431
 432 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
 433 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
 434 #define re_free(p) free (p)
 435
 436 struct bin_tree_t
 437 {
 438   struct bin_tree_t *parent;
 439   struct bin_tree_t *left;
 440   struct bin_tree_t *right;
 441   struct bin_tree_t *first;
 442   struct bin_tree_t *next;
 443
 444   re_token_t token;
 445
 446   /* `node_idx' is the index in dfa->nodes, if `type' == 0.
 447      Otherwise `type' indicate the type of this node.  */
 448   int node_idx;
 449 };
 450 typedef struct bin_tree_t bin_tree_t;
 451
 452 #define BIN_TREE_STORAGE_SIZE \
 453   ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
 454
 455 struct bin_tree_storage_t
 456 {
 457   struct bin_tree_storage_t *next;
 458   bin_tree_t data[BIN_TREE_STORAGE_SIZE];
 459 };
 460 typedef struct bin_tree_storage_t bin_tree_storage_t;
 461
 462 #define CONTEXT_WORD 1
 463 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
 464 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
 465 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
 466
 467 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
 468 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
 469 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
 470 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
 471 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
 472
 473 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
 474 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
 475 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
 476 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
 477
 478 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
 479  ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 480   || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 481   || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
 482   || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
 483
 484 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
 485  ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 486   || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 487   || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
 488   || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
 489
 490 struct re_dfastate_t
 491 {
 492   unsigned int hash;
 493   re_node_set nodes;
 494   re_node_set non_eps_nodes;
 495   re_node_set inveclosure;
 496   re_node_set *entrance_nodes;
 497   struct re_dfastate_t **trtable, **word_trtable;
 498   unsigned int context : 4;
 499   unsigned int halt : 1;
 500   /* If this state can accept `multi byte'.
 501      Note that we refer to multibyte characters, and multi character
 502      collating elements as `multi byte'.  */
 503   unsigned int accept_mb : 1;
 504   /* If this state has backreference node(s).  */
 505   unsigned int has_backref : 1;
 506   unsigned int has_constraint : 1;
 507 };
 508 typedef struct re_dfastate_t re_dfastate_t;
 509
 510 struct re_state_table_entry
 511 {
 512   int num;
 513   int alloc;
 514   re_dfastate_t **array;
 515 };
 516
 517 /* Array type used in re_sub_match_last_t and re_sub_match_top_t.  */
 518
 519 typedef struct
 520 {
 521   int next_idx;
 522   int alloc;
 523   re_dfastate_t **array;
 524 } state_array_t;
 525
 526 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP.  */
 527
 528 typedef struct
 529 {
 530   int node;
 531   int str_idx; /* The position NODE match at.  */
 532   state_array_t path;
 533 } re_sub_match_last_t;
 534
 535 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
 536    And information about the node, whose type is OP_CLOSE_SUBEXP,
 537    corresponding to NODE is stored in LASTS.  */
 538
 539 typedef struct
 540 {
 541   int str_idx;
 542   int node;
 543   state_array_t *path;
 544   int alasts; /* Allocation size of LASTS.  */
 545   int nlasts; /* The number of LASTS.  */
 546   re_sub_match_last_t **lasts;
 547 } re_sub_match_top_t;
 548
 549 struct re_backref_cache_entry
 550 {
 551   int node;
 552   int str_idx;
 553   int subexp_from;
 554   int subexp_to;
 555   char more;
 556   char unused;
 557   unsigned short int eps_reachable_subexps_map;
 558 };
 559
 560 typedef struct
 561 {
 562   /* The string object corresponding to the input string.  */
 563   re_string_t input;
 564 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
 565   const re_dfa_t *const dfa;
 566 #else
 567   const re_dfa_t *dfa;
 568 #endif
 569   /* EFLAGS of the argument of regexec.  */
 570   int eflags;
 571   /* Where the matching ends.  */
 572   int match_last;
 573   int last_node;
 574   /* The state log used by the matcher.  */
 575   re_dfastate_t **state_log;
 576   int state_log_top;
 577   /* Back reference cache.  */
 578   int nbkref_ents;
 579   int abkref_ents;
 580   struct re_backref_cache_entry *bkref_ents;
 581   int max_mb_elem_len;
 582   int nsub_tops;
 583   int asub_tops;
 584   re_sub_match_top_t **sub_tops;
 585 } re_match_context_t;
 586
 587 typedef struct
 588 {
 589   re_dfastate_t **sifted_states;
 590   re_dfastate_t **limited_states;
 591   int last_node;
 592   int last_str_idx;
 593   re_node_set limits;
 594 } re_sift_context_t;
 595
 596 struct re_fail_stack_ent_t
 597 {
 598   int idx;
 599   int node;
 600   regmatch_t *regs;
 601   re_node_set eps_via_nodes;
 602 };
 603
 604 struct re_fail_stack_t
 605 {
 606   int num;
 607   int alloc;
 608   struct re_fail_stack_ent_t *stack;
 609 };
 610
 611 struct re_dfa_t
 612 {
 613   re_token_t *nodes;
 614   size_t nodes_alloc;
 615   size_t nodes_len;
 616   int *nexts;
 617   int *org_indices;
 618   re_node_set *edests;
 619   re_node_set *eclosures;
 620   re_node_set *inveclosures;
 621   struct re_state_table_entry *state_table;
 622   re_dfastate_t *init_state;
 623   re_dfastate_t *init_state_word;
 624   re_dfastate_t *init_state_nl;
 625   re_dfastate_t *init_state_begbuf;
 626   bin_tree_t *str_tree;
 627   bin_tree_storage_t *str_tree_storage;
 628   re_bitset_ptr_t sb_char;
 629   int str_tree_storage_idx;
 630
 631   /* number of subexpressions `re_nsub' is in regex_t.  */
 632   unsigned int state_hash_mask;
 633   int init_node;
 634   int nbackref; /* The number of backreference in this dfa.  */
 635
 636   /* Bitmap expressing which backreference is used.  */
 637   bitset_word_t used_bkref_map;
 638   bitset_word_t completed_bkref_map;
 639
 640   unsigned int has_plural_match : 1;
 641   /* If this dfa has "multibyte node", which is a backreference or
 642      a node which can accept multibyte character or multi character
 643      collating element.  */
 644   unsigned int has_mb_node : 1;
 645   unsigned int is_utf8 : 1;
 646   unsigned int map_notascii : 1;
 647   unsigned int word_ops_used : 1;
 648   int mb_cur_max;
 649   bitset_t word_char;
 650   reg_syntax_t syntax;
 651   int *subexp_map;
 652 #ifdef DEBUG
 653   char* re_str;
 654 #endif
 655   __libc_lock_define (, lock)
 656 };
 657
 658 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
 659 #define re_node_set_remove(set,id) \
 660   (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
 661 #define re_node_set_empty(p) ((p)->nelem = 0)
 662 #define re_node_set_free(set) re_free ((set)->elems)
 663 \f
 664
 665 typedef enum
 666 {
 667   SB_CHAR,
 668   MB_CHAR,
 669   EQUIV_CLASS,
 670   COLL_SYM,
 671   CHAR_CLASS
 672 } bracket_elem_type;
 673
 674 typedef struct
 675 {
 676   bracket_elem_type type;
 677   union
 678   {
 679     unsigned char ch;
 680     unsigned char *name;
 681     wchar_t wch;
 682   } opr;
 683 } bracket_elem_t;
 684
 685
 686 /* Inline functions for bitset operation.  */
 687 static inline void
 688 bitset_not (bitset_t set)
 689 {
 690   int bitset_i;
 691   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 692     set[bitset_i] = ~set[bitset_i];
 693 }
 694
 695 static inline void
 696 bitset_merge (bitset_t dest, const bitset_t src)
 697 {
 698   int bitset_i;
 699   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 700     dest[bitset_i] |= src[bitset_i];
 701 }
 702
 703 static inline void
 704 bitset_mask (bitset_t dest, const bitset_t src)
 705 {
 706   int bitset_i;
 707   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 708     dest[bitset_i] &= src[bitset_i];
 709 }
 710
 711 #ifdef RE_ENABLE_I18N
 712 /* Inline functions for re_string.  */
 713 static inline int
 714 internal_function __attribute ((pure))
 715 re_string_char_size_at (const re_string_t *pstr, int idx)
 716 {
 717   int byte_idx;
 718   if (pstr->mb_cur_max == 1)
 719     return 1;
 720   for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
 721     if (pstr->wcs[idx + byte_idx] != WEOF)
 722       break;
 723   return byte_idx;
 724 }
 725
 726 static inline wint_t
 727 internal_function __attribute ((pure))
 728 re_string_wchar_at (const re_string_t *pstr, int idx)
 729 {
 730   if (pstr->mb_cur_max == 1)
 731     return (wint_t) pstr->mbs[idx];
 732   return (wint_t) pstr->wcs[idx];
 733 }
 734
 735 static int
 736 internal_function __attribute ((pure))
 737 re_string_elem_size_at (const re_string_t *pstr, int idx)
 738 {
 739 # ifdef _LIBC
 740   const unsigned char *p, *extra;
 741   const int32_t *table, *indirect;
 742   int32_t tmp;
 743 #  include <locale/weight.h>
 744   uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
 745
 746   if (nrules != 0)
 747     {
 748       table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
 749       extra = (const unsigned char *)
 750         _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
 751       indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
 752                                                 _NL_COLLATE_INDIRECTMB);
 753       p = pstr->mbs + idx;
 754       tmp = findidx (&p);
 755       return p - pstr->mbs - idx;
 756     }
 757   else
 758 # endif /* _LIBC */
 759     return 1;
 760 }
 761 #endif /* RE_ENABLE_I18N */
 762
 763 #endif /*  _REGEX_INTERNAL_H */