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