gl/regex_internal.h

   1 /* Extended regular expression matching and search library.
   2    Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010 Free
   3    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 #include <langinfo.h>
  32 #ifndef _LIBC
  33 # include "localcharset.h"
  34 #endif
  35 #if defined HAVE_LOCALE_H || defined _LIBC
  36 # include <locale.h>
  37 #endif
  38
  39 #include <wchar.h>
  40 #include <wctype.h>
  41 #include <stdint.h>
  42 #if defined _LIBC
  43 # include <bits/libc-lock.h>
  44 #else
  45 # define __libc_lock_init(NAME) do { } while (0)
  46 # define __libc_lock_lock(NAME) do { } while (0)
  47 # define __libc_lock_unlock(NAME) do { } while (0)
  48 #endif
  49
  50 /* In case that the system doesn't have isblank().  */
  51 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK))
  52 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
  53 #endif
  54
  55 #ifdef _LIBC
  56 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
  57 #  define _RE_DEFINE_LOCALE_FUNCTIONS 1
  58 #   include <locale/localeinfo.h>
  59 #   include <locale/elem-hash.h>
  60 #   include <locale/coll-lookup.h>
  61 # endif
  62 #endif
  63
  64 /* This is for other GNU distributions with internationalized messages.  */
  65 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
  66 # include <libintl.h>
  67 # ifdef _LIBC
  68 #  undef gettext
  69 #  define gettext(msgid) \
  70   INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
  71 # endif
  72 #else
  73 # define gettext(msgid) (msgid)
  74 #endif
  75
  76 #ifndef gettext_noop
  77 /* This define is so xgettext can find the internationalizable
  78    strings.  */
  79 # define gettext_noop(String) String
  80 #endif
  81
  82 /* For loser systems without the definition.  */
  83 #ifndef SIZE_MAX
  84 # define SIZE_MAX ((size_t) -1)
  85 #endif
  86
  87 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCSCOLL) || _LIBC
  88 # define RE_ENABLE_I18N
  89 #endif
  90
  91 #if __GNUC__ >= 3
  92 # define BE(expr, val) __builtin_expect (expr, val)
  93 #else
  94 # define BE(expr, val) (expr)
  95 # ifdef _LIBC
  96 #  define inline
  97 # endif
  98 #endif
  99
 100 /* Number of ASCII characters.  */
 101 #define ASCII_CHARS 0x80
 102
 103 /* Number of single byte characters.  */
 104 #define SBC_MAX (UCHAR_MAX + 1)
 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 __wcrtomb wcrtomb
 118 # define __mbrtowc mbrtowc
 119 # define __regfree regfree
 120 # define attribute_hidden
 121 #endif /* not _LIBC */
 122
 123 #if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)
 124 # define __attribute(arg) __attribute__ (arg)
 125 #else
 126 # define __attribute(arg)
 127 #endif
 128
 129 typedef __re_idx_t Idx;
 130
 131 /* Special return value for failure to match.  */
 132 #define REG_MISSING ((Idx) -1)
 133
 134 /* Special return value for internal error.  */
 135 #define REG_ERROR ((Idx) -2)
 136
 137 /* Test whether N is a valid index, and is not one of the above.  */
 138 #ifdef _REGEX_LARGE_OFFSETS
 139 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR)
 140 #else
 141 # define REG_VALID_INDEX(n) (0 <= (n))
 142 #endif
 143
 144 /* Test whether N is a valid nonzero index.  */
 145 #ifdef _REGEX_LARGE_OFFSETS
 146 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1))
 147 #else
 148 # define REG_VALID_NONZERO_INDEX(n) (0 < (n))
 149 #endif
 150
 151 /* A hash value, suitable for computing hash tables.  */
 152 typedef __re_size_t re_hashval_t;
 153
 154 /* An integer used to represent a set of bits.  It must be unsigned,
 155    and must be at least as wide as unsigned int.  */
 156 typedef unsigned long int bitset_word_t;
 157 /* All bits set in a bitset_word_t.  */
 158 #define BITSET_WORD_MAX ULONG_MAX
 159
 160 /* Number of bits in a bitset_word_t.  For portability to hosts with
 161    padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)';
 162    instead, deduce it directly from BITSET_WORD_MAX.  Avoid
 163    greater-than-32-bit integers and unconditional shifts by more than
 164    31 bits, as they're not portable.  */
 165 #if BITSET_WORD_MAX == 0xffffffffUL
 166 # define BITSET_WORD_BITS 32
 167 #elif BITSET_WORD_MAX >> 31 >> 4 == 1
 168 # define BITSET_WORD_BITS 36
 169 #elif BITSET_WORD_MAX >> 31 >> 16 == 1
 170 # define BITSET_WORD_BITS 48
 171 #elif BITSET_WORD_MAX >> 31 >> 28 == 1
 172 # define BITSET_WORD_BITS 60
 173 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1
 174 # define BITSET_WORD_BITS 64
 175 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1
 176 # define BITSET_WORD_BITS 72
 177 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1
 178 # define BITSET_WORD_BITS 128
 179 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1
 180 # define BITSET_WORD_BITS 256
 181 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1
 182 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */
 183 # if BITSET_WORD_BITS <= SBC_MAX
 184 #  error "Invalid SBC_MAX"
 185 # endif
 186 #else
 187 # error "Add case for new bitset_word_t size"
 188 #endif
 189
 190 /* Number of bitset_word_t values in a bitset_t.  */
 191 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS)
 192
 193 typedef bitset_word_t bitset_t[BITSET_WORDS];
 194 typedef bitset_word_t *re_bitset_ptr_t;
 195 typedef const bitset_word_t *re_const_bitset_ptr_t;
 196
 197 #define PREV_WORD_CONSTRAINT 0x0001
 198 #define PREV_NOTWORD_CONSTRAINT 0x0002
 199 #define NEXT_WORD_CONSTRAINT 0x0004
 200 #define NEXT_NOTWORD_CONSTRAINT 0x0008
 201 #define PREV_NEWLINE_CONSTRAINT 0x0010
 202 #define NEXT_NEWLINE_CONSTRAINT 0x0020
 203 #define PREV_BEGBUF_CONSTRAINT 0x0040
 204 #define NEXT_ENDBUF_CONSTRAINT 0x0080
 205 #define WORD_DELIM_CONSTRAINT 0x0100
 206 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
 207
 208 typedef enum
 209 {
 210   INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 211   WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 212   WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 213   INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 214   LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
 215   LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
 216   BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
 217   BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
 218   WORD_DELIM = WORD_DELIM_CONSTRAINT,
 219   NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
 220 } re_context_type;
 221
 222 typedef struct
 223 {
 224   Idx alloc;
 225   Idx nelem;
 226   Idx *elems;
 227 } re_node_set;
 228
 229 typedef enum
 230 {
 231   NON_TYPE = 0,
 232
 233   /* Node type, These are used by token, node, tree.  */
 234   CHARACTER = 1,
 235   END_OF_RE = 2,
 236   SIMPLE_BRACKET = 3,
 237   OP_BACK_REF = 4,
 238   OP_PERIOD = 5,
 239 #ifdef RE_ENABLE_I18N
 240   COMPLEX_BRACKET = 6,
 241   OP_UTF8_PERIOD = 7,
 242 #endif /* RE_ENABLE_I18N */
 243
 244   /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
 245      when the debugger shows values of this enum type.  */
 246 #define EPSILON_BIT 8
 247   OP_OPEN_SUBEXP = EPSILON_BIT | 0,
 248   OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
 249   OP_ALT = EPSILON_BIT | 2,
 250   OP_DUP_ASTERISK = EPSILON_BIT | 3,
 251   ANCHOR = EPSILON_BIT | 4,
 252
 253   /* Tree type, these are used only by tree. */
 254   CONCAT = 16,
 255   SUBEXP = 17,
 256
 257   /* Token type, these are used only by token.  */
 258   OP_DUP_PLUS = 18,
 259   OP_DUP_QUESTION,
 260   OP_OPEN_BRACKET,
 261   OP_CLOSE_BRACKET,
 262   OP_CHARSET_RANGE,
 263   OP_OPEN_DUP_NUM,
 264   OP_CLOSE_DUP_NUM,
 265   OP_NON_MATCH_LIST,
 266   OP_OPEN_COLL_ELEM,
 267   OP_CLOSE_COLL_ELEM,
 268   OP_OPEN_EQUIV_CLASS,
 269   OP_CLOSE_EQUIV_CLASS,
 270   OP_OPEN_CHAR_CLASS,
 271   OP_CLOSE_CHAR_CLASS,
 272   OP_WORD,
 273   OP_NOTWORD,
 274   OP_SPACE,
 275   OP_NOTSPACE,
 276   BACK_SLASH
 277
 278 } re_token_type_t;
 279
 280 #ifdef RE_ENABLE_I18N
 281 typedef struct
 282 {
 283   /* Multibyte characters.  */
 284   wchar_t *mbchars;
 285
 286   /* Collating symbols.  */
 287 # ifdef _LIBC
 288   int32_t *coll_syms;
 289 # endif
 290
 291   /* Equivalence classes. */
 292 # ifdef _LIBC
 293   int32_t *equiv_classes;
 294 # endif
 295
 296   /* Range expressions. */
 297 # ifdef _LIBC
 298   uint32_t *range_starts;
 299   uint32_t *range_ends;
 300 # else /* not _LIBC */
 301   wchar_t *range_starts;
 302   wchar_t *range_ends;
 303 # endif /* not _LIBC */
 304
 305   /* Character classes. */
 306   wctype_t *char_classes;
 307
 308   /* If this character set is the non-matching list.  */
 309   unsigned int non_match : 1;
 310
 311   /* # of multibyte characters.  */
 312   Idx nmbchars;
 313
 314   /* # of collating symbols.  */
 315   Idx ncoll_syms;
 316
 317   /* # of equivalence classes. */
 318   Idx nequiv_classes;
 319
 320   /* # of range expressions. */
 321   Idx nranges;
 322
 323   /* # of character classes. */
 324   Idx nchar_classes;
 325 } re_charset_t;
 326 #endif /* RE_ENABLE_I18N */
 327
 328 typedef struct
 329 {
 330   union
 331   {
 332     unsigned char c;            /* for CHARACTER */
 333     re_bitset_ptr_t sbcset;     /* for SIMPLE_BRACKET */
 334 #ifdef RE_ENABLE_I18N
 335     re_charset_t *mbcset;       /* for COMPLEX_BRACKET */
 336 #endif /* RE_ENABLE_I18N */
 337     Idx idx;                    /* for BACK_REF */
 338     re_context_type ctx_type;   /* for ANCHOR */
 339   } opr;
 340 #if __GNUC__ >= 2 && !__STRICT_ANSI__
 341   re_token_type_t type : 8;
 342 #else
 343   re_token_type_t type;
 344 #endif
 345   unsigned int constraint : 10; /* context constraint */
 346   unsigned int duplicated : 1;
 347   unsigned int opt_subexp : 1;
 348 #ifdef RE_ENABLE_I18N
 349   unsigned int accept_mb : 1;
 350   /* These 2 bits can be moved into the union if needed (e.g. if running out
 351      of bits; move opr.c to opr.c.c and move the flags to opr.c.flags).  */
 352   unsigned int mb_partial : 1;
 353 #endif
 354   unsigned int word_char : 1;
 355 } re_token_t;
 356
 357 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
 358
 359 struct re_string_t
 360 {
 361   /* Indicate the raw buffer which is the original string passed as an
 362      argument of regexec(), re_search(), etc..  */
 363   const unsigned char *raw_mbs;
 364   /* Store the multibyte string.  In case of "case insensitive mode" like
 365      REG_ICASE, upper cases of the string are stored, otherwise MBS points
 366      the same address that RAW_MBS points.  */
 367   unsigned char *mbs;
 368 #ifdef RE_ENABLE_I18N
 369   /* Store the wide character string which is corresponding to MBS.  */
 370   wint_t *wcs;
 371   Idx *offsets;
 372   mbstate_t cur_state;
 373 #endif
 374   /* Index in RAW_MBS.  Each character mbs[i] corresponds to
 375      raw_mbs[raw_mbs_idx + i].  */
 376   Idx raw_mbs_idx;
 377   /* The length of the valid characters in the buffers.  */
 378   Idx valid_len;
 379   /* The corresponding number of bytes in raw_mbs array.  */
 380   Idx valid_raw_len;
 381   /* The length of the buffers MBS and WCS.  */
 382   Idx bufs_len;
 383   /* The index in MBS, which is updated by re_string_fetch_byte.  */
 384   Idx cur_idx;
 385   /* length of RAW_MBS array.  */
 386   Idx raw_len;
 387   /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN.  */
 388   Idx len;
 389   /* End of the buffer may be shorter than its length in the cases such
 390      as re_match_2, re_search_2.  Then, we use STOP for end of the buffer
 391      instead of LEN.  */
 392   Idx raw_stop;
 393   /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS.  */
 394   Idx stop;
 395
 396   /* The context of mbs[0].  We store the context independently, since
 397      the context of mbs[0] may be different from raw_mbs[0], which is
 398      the beginning of the input string.  */
 399   unsigned int tip_context;
 400   /* The translation passed as a part of an argument of re_compile_pattern.  */
 401   RE_TRANSLATE_TYPE trans;
 402   /* Copy of re_dfa_t's word_char.  */
 403   re_const_bitset_ptr_t word_char;
 404   /* true if REG_ICASE.  */
 405   unsigned char icase;
 406   unsigned char is_utf8;
 407   unsigned char map_notascii;
 408   unsigned char mbs_allocated;
 409   unsigned char offsets_needed;
 410   unsigned char newline_anchor;
 411   unsigned char word_ops_used;
 412   int mb_cur_max;
 413 };
 414 typedef struct re_string_t re_string_t;
 415
 416
 417 struct re_dfa_t;
 418 typedef struct re_dfa_t re_dfa_t;
 419
 420 #ifndef _LIBC
 421 # if defined __i386__ && !defined __EMX__
 422 #  define internal_function   __attribute ((regparm (3), stdcall))
 423 # else
 424 #  define internal_function
 425 # endif
 426 #endif
 427
 428 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
 429                                                 Idx new_buf_len)
 430      internal_function;
 431 #ifdef RE_ENABLE_I18N
 432 static void build_wcs_buffer (re_string_t *pstr) internal_function;
 433 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
 434      internal_function;
 435 #endif /* RE_ENABLE_I18N */
 436 static void build_upper_buffer (re_string_t *pstr) internal_function;
 437 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
 438 static unsigned int re_string_context_at (const re_string_t *input, Idx idx,
 439                                           int eflags)
 440      internal_function __attribute ((pure));
 441 #define re_string_peek_byte(pstr, offset) \
 442   ((pstr)->mbs[(pstr)->cur_idx + offset])
 443 #define re_string_fetch_byte(pstr) \
 444   ((pstr)->mbs[(pstr)->cur_idx++])
 445 #define re_string_first_byte(pstr, idx) \
 446   ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
 447 #define re_string_is_single_byte_char(pstr, idx) \
 448   ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
 449                                 || (pstr)->wcs[(idx) + 1] != WEOF))
 450 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
 451 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
 452 #define re_string_get_buffer(pstr) ((pstr)->mbs)
 453 #define re_string_length(pstr) ((pstr)->len)
 454 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
 455 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
 456 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
 457
 458 #include <alloca.h>
 459
 460 #ifndef _LIBC
 461 # if HAVE_ALLOCA
 462 /* The OS usually guarantees only one guard page at the bottom of the stack,
 463    and a page size can be as small as 4096 bytes.  So we cannot safely
 464    allocate anything larger than 4096 bytes.  Also care for the possibility
 465    of a few compiler-allocated temporary stack slots.  */
 466 #  define __libc_use_alloca(n) ((n) < 4032)
 467 # else
 468 /* alloca is implemented with malloc, so just use malloc.  */
 469 #  define __libc_use_alloca(n) 0
 470 #  undef alloca
 471 #  define alloca(n) malloc (n)
 472 # endif
 473 #endif
 474
 475 #ifndef MAX
 476 # define MAX(a,b) ((a) < (b) ? (b) : (a))
 477 #endif
 478
 479 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
 480 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
 481 #define re_free(p) free (p)
 482
 483 struct bin_tree_t
 484 {
 485   struct bin_tree_t *parent;
 486   struct bin_tree_t *left;
 487   struct bin_tree_t *right;
 488   struct bin_tree_t *first;
 489   struct bin_tree_t *next;
 490
 491   re_token_t token;
 492
 493   /* `node_idx' is the index in dfa->nodes, if `type' == 0.
 494      Otherwise `type' indicate the type of this node.  */
 495   Idx node_idx;
 496 };
 497 typedef struct bin_tree_t bin_tree_t;
 498
 499 #define BIN_TREE_STORAGE_SIZE \
 500   ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
 501
 502 struct bin_tree_storage_t
 503 {
 504   struct bin_tree_storage_t *next;
 505   bin_tree_t data[BIN_TREE_STORAGE_SIZE];
 506 };
 507 typedef struct bin_tree_storage_t bin_tree_storage_t;
 508
 509 #define CONTEXT_WORD 1
 510 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
 511 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
 512 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
 513
 514 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
 515 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
 516 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
 517 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
 518 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
 519
 520 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
 521 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
 522 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
 523 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
 524
 525 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
 526  ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 527   || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 528   || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
 529   || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
 530
 531 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
 532  ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 533   || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 534   || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
 535   || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
 536
 537 struct re_dfastate_t
 538 {
 539   re_hashval_t hash;
 540   re_node_set nodes;
 541   re_node_set non_eps_nodes;
 542   re_node_set inveclosure;
 543   re_node_set *entrance_nodes;
 544   struct re_dfastate_t **trtable, **word_trtable;
 545   unsigned int context : 4;
 546   unsigned int halt : 1;
 547   /* If this state can accept `multi byte'.
 548      Note that we refer to multibyte characters, and multi character
 549      collating elements as `multi byte'.  */
 550   unsigned int accept_mb : 1;
 551   /* If this state has backreference node(s).  */
 552   unsigned int has_backref : 1;
 553   unsigned int has_constraint : 1;
 554 };
 555 typedef struct re_dfastate_t re_dfastate_t;
 556
 557 struct re_state_table_entry
 558 {
 559   Idx num;
 560   Idx alloc;
 561   re_dfastate_t **array;
 562 };
 563
 564 /* Array type used in re_sub_match_last_t and re_sub_match_top_t.  */
 565
 566 typedef struct
 567 {
 568   Idx next_idx;
 569   Idx alloc;
 570   re_dfastate_t **array;
 571 } state_array_t;
 572
 573 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP.  */
 574
 575 typedef struct
 576 {
 577   Idx node;
 578   Idx str_idx; /* The position NODE match at.  */
 579   state_array_t path;
 580 } re_sub_match_last_t;
 581
 582 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
 583    And information about the node, whose type is OP_CLOSE_SUBEXP,
 584    corresponding to NODE is stored in LASTS.  */
 585
 586 typedef struct
 587 {
 588   Idx str_idx;
 589   Idx node;
 590   state_array_t *path;
 591   Idx alasts; /* Allocation size of LASTS.  */
 592   Idx nlasts; /* The number of LASTS.  */
 593   re_sub_match_last_t **lasts;
 594 } re_sub_match_top_t;
 595
 596 struct re_backref_cache_entry
 597 {
 598   Idx node;
 599   Idx str_idx;
 600   Idx subexp_from;
 601   Idx subexp_to;
 602   char more;
 603   char unused;
 604   unsigned short int eps_reachable_subexps_map;
 605 };
 606
 607 typedef struct
 608 {
 609   /* The string object corresponding to the input string.  */
 610   re_string_t input;
 611 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
 612   const re_dfa_t *const dfa;
 613 #else
 614   const re_dfa_t *dfa;
 615 #endif
 616   /* EFLAGS of the argument of regexec.  */
 617   int eflags;
 618   /* Where the matching ends.  */
 619   Idx match_last;
 620   Idx last_node;
 621   /* The state log used by the matcher.  */
 622   re_dfastate_t **state_log;
 623   Idx state_log_top;
 624   /* Back reference cache.  */
 625   Idx nbkref_ents;
 626   Idx abkref_ents;
 627   struct re_backref_cache_entry *bkref_ents;
 628   int max_mb_elem_len;
 629   Idx nsub_tops;
 630   Idx asub_tops;
 631   re_sub_match_top_t **sub_tops;
 632 } re_match_context_t;
 633
 634 typedef struct
 635 {
 636   re_dfastate_t **sifted_states;
 637   re_dfastate_t **limited_states;
 638   Idx last_node;
 639   Idx last_str_idx;
 640   re_node_set limits;
 641 } re_sift_context_t;
 642
 643 struct re_fail_stack_ent_t
 644 {
 645   Idx idx;
 646   Idx node;
 647   regmatch_t *regs;
 648   re_node_set eps_via_nodes;
 649 };
 650
 651 struct re_fail_stack_t
 652 {
 653   Idx num;
 654   Idx alloc;
 655   struct re_fail_stack_ent_t *stack;
 656 };
 657
 658 struct re_dfa_t
 659 {
 660   re_token_t *nodes;
 661   size_t nodes_alloc;
 662   size_t nodes_len;
 663   Idx *nexts;
 664   Idx *org_indices;
 665   re_node_set *edests;
 666   re_node_set *eclosures;
 667   re_node_set *inveclosures;
 668   struct re_state_table_entry *state_table;
 669   re_dfastate_t *init_state;
 670   re_dfastate_t *init_state_word;
 671   re_dfastate_t *init_state_nl;
 672   re_dfastate_t *init_state_begbuf;
 673   bin_tree_t *str_tree;
 674   bin_tree_storage_t *str_tree_storage;
 675   re_bitset_ptr_t sb_char;
 676   int str_tree_storage_idx;
 677
 678   /* number of subexpressions `re_nsub' is in regex_t.  */
 679   re_hashval_t state_hash_mask;
 680   Idx init_node;
 681   Idx nbackref; /* The number of backreference in this dfa.  */
 682
 683   /* Bitmap expressing which backreference is used.  */
 684   bitset_word_t used_bkref_map;
 685   bitset_word_t completed_bkref_map;
 686
 687   unsigned int has_plural_match : 1;
 688   /* If this dfa has "multibyte node", which is a backreference or
 689      a node which can accept multibyte character or multi character
 690      collating element.  */
 691   unsigned int has_mb_node : 1;
 692   unsigned int is_utf8 : 1;
 693   unsigned int map_notascii : 1;
 694   unsigned int word_ops_used : 1;
 695   int mb_cur_max;
 696   bitset_t word_char;
 697   reg_syntax_t syntax;
 698   Idx *subexp_map;
 699 #ifdef DEBUG
 700   char* re_str;
 701 #endif
 702 #ifdef _LIBC
 703   __libc_lock_define (, lock)
 704 #endif
 705 };
 706
 707 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
 708 #define re_node_set_remove(set,id) \
 709   (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
 710 #define re_node_set_empty(p) ((p)->nelem = 0)
 711 #define re_node_set_free(set) re_free ((set)->elems)
 712 \f
 713
 714 typedef enum
 715 {
 716   SB_CHAR,
 717   MB_CHAR,
 718   EQUIV_CLASS,
 719   COLL_SYM,
 720   CHAR_CLASS
 721 } bracket_elem_type;
 722
 723 typedef struct
 724 {
 725   bracket_elem_type type;
 726   union
 727   {
 728     unsigned char ch;
 729     unsigned char *name;
 730     wchar_t wch;
 731   } opr;
 732 } bracket_elem_t;
 733
 734
 735 /* Inline functions for bitset_t operation.  */
 736
 737 static inline void
 738 bitset_set (bitset_t set, Idx i)
 739 {
 740   set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS;
 741 }
 742
 743 static inline void
 744 bitset_clear (bitset_t set, Idx i)
 745 {
 746   set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS);
 747 }
 748
 749 static inline bool
 750 bitset_contain (const bitset_t set, Idx i)
 751 {
 752   return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1;
 753 }
 754
 755 static inline void
 756 bitset_empty (bitset_t set)
 757 {
 758   memset (set, '\0', sizeof (bitset_t));
 759 }
 760
 761 static inline void
 762 bitset_set_all (bitset_t set)
 763 {
 764   memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS));
 765   if (SBC_MAX % BITSET_WORD_BITS != 0)
 766     set[BITSET_WORDS - 1] =
 767       ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1;
 768 }
 769
 770 static inline void
 771 bitset_copy (bitset_t dest, const bitset_t src)
 772 {
 773   memcpy (dest, src, sizeof (bitset_t));
 774 }
 775
 776 static inline void
 777 bitset_not (bitset_t set)
 778 {
 779   int bitset_i;
 780   for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i)
 781     set[bitset_i] = ~set[bitset_i];
 782   if (SBC_MAX % BITSET_WORD_BITS != 0)
 783     set[BITSET_WORDS - 1] =
 784       ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1)
 785        & ~set[BITSET_WORDS - 1]);
 786 }
 787
 788 static inline void
 789 bitset_merge (bitset_t dest, const bitset_t src)
 790 {
 791   int bitset_i;
 792   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 793     dest[bitset_i] |= src[bitset_i];
 794 }
 795
 796 static inline void
 797 bitset_mask (bitset_t dest, const bitset_t src)
 798 {
 799   int bitset_i;
 800   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 801     dest[bitset_i] &= src[bitset_i];
 802 }
 803
 804 #ifdef RE_ENABLE_I18N
 805 /* Inline functions for re_string.  */
 806 static inline int
 807 internal_function __attribute ((pure))
 808 re_string_char_size_at (const re_string_t *pstr, Idx idx)
 809 {
 810   int byte_idx;
 811   if (pstr->mb_cur_max == 1)
 812     return 1;
 813   for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
 814     if (pstr->wcs[idx + byte_idx] != WEOF)
 815       break;
 816   return byte_idx;
 817 }
 818
 819 static inline wint_t
 820 internal_function __attribute ((pure))
 821 re_string_wchar_at (const re_string_t *pstr, Idx idx)
 822 {
 823   if (pstr->mb_cur_max == 1)
 824     return (wint_t) pstr->mbs[idx];
 825   return (wint_t) pstr->wcs[idx];
 826 }
 827
 828 static int
 829 internal_function __attribute ((pure))
 830 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
 831 {
 832 # ifdef _LIBC
 833   const unsigned char *p, *extra;
 834   const int32_t *table, *indirect;
 835   int32_t tmp;
 836 #  include <locale/weight.h>
 837   uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
 838
 839   if (nrules != 0)
 840     {
 841       table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
 842       extra = (const unsigned char *)
 843         _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
 844       indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
 845                                                 _NL_COLLATE_INDIRECTMB);
 846       p = pstr->mbs + idx;
 847       tmp = findidx (&p);
 848       return p - pstr->mbs - idx;
 849     }
 850   else
 851 # endif /* _LIBC */
 852     return 1;
 853 }
 854 #endif /* RE_ENABLE_I18N */
 855
 856 #ifndef __GNUC_PREREQ
 857 # if defined __GNUC__ && defined __GNUC_MINOR__
 858 #  define __GNUC_PREREQ(maj, min) \
 859          ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
 860 # else
 861 #  define __GNUC_PREREQ(maj, min) 0
 862 # endif
 863 #endif
 864
 865 #if __GNUC_PREREQ (3,4)
 866 # undef __attribute_warn_unused_result__
 867 # define __attribute_warn_unused_result__ \
 868    __attribute__ ((__warn_unused_result__))
 869 #else
 870 # define __attribute_warn_unused_result__ /* empty */
 871 #endif
 872
 873 #endif /*  _REGEX_INTERNAL_H */