compat/regex/regex_internal.h

   1 /* Extended regular expression matching and search library.
   2    Copyright (C) 2002-2005, 2007, 2008, 2010 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(ZOS_USS)
  45 #if defined HAVE_STDINT_H || defined _LIBC
  46 # include <stdint.h>
  47 #endif /* HAVE_STDINT_H || _LIBC */
  48 #endif /* !ZOS_USS */
  49 #if defined _LIBC
  50 # include <bits/libc-lock.h>
  51 #else
  52 # define __libc_lock_define(CLASS,NAME)
  53 # define __libc_lock_init(NAME) do { } while (0)
  54 # define __libc_lock_lock(NAME) do { } while (0)
  55 # define __libc_lock_unlock(NAME) do { } while (0)
  56 #endif
  57
  58 #ifndef GAWK
  59 /* In case that the system doesn't have isblank().  */
  60 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
  61 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
  62 #endif
  63 #else /* GAWK */
  64 /*
  65  * This is a freaking mess. On glibc systems you have to define
  66  * a magic constant to get isblank() out of <ctype.h>, since it's
  67  * a C99 function.  To heck with all that and borrow a page from
  68  * dfa.c's book.
  69  */
  70
  71 static int
  72 is_blank (int c)
  73 {
  74    return (c == ' ' || c == '\t');
  75 }
  76 #endif /* GAWK */
  77
  78 #ifdef _LIBC
  79 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
  80 #  define _RE_DEFINE_LOCALE_FUNCTIONS 1
  81 #   include <locale/localeinfo.h>
  82 #   include <locale/elem-hash.h>
  83 #   include <locale/coll-lookup.h>
  84 # endif
  85 #endif
  86
  87 /* This is for other GNU distributions with internationalized messages.  */
  88 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
  89 # include <libintl.h>
  90 # ifdef _LIBC
  91 #  undef gettext
  92 #  define gettext(msgid) \
  93   INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
  94 # endif
  95 #else
  96 # define gettext(msgid) (msgid)
  97 #endif
  98
  99 #ifndef gettext_noop
 100 /* This define is so xgettext can find the internationalizable
 101    strings.  */
 102 # define gettext_noop(String) String
 103 #endif
 104
 105 /* For loser systems without the definition.  */
 106 #ifndef SIZE_MAX
 107 # define SIZE_MAX ((size_t) -1)
 108 #endif
 109
 110 #ifndef NO_MBSUPPORT
 111 #include "mbsupport.h" /* gawk */
 112 #endif
 113 #ifndef MB_CUR_MAX
 114 #define MB_CUR_MAX 1
 115 #endif
 116
 117 #if (defined MBS_SUPPORT) || _LIBC
 118 # define RE_ENABLE_I18N
 119 #endif
 120
 121 #if __GNUC__ >= 3
 122 # define BE(expr, val) __builtin_expect (expr, val)
 123 #else
 124 # define BE(expr, val) (expr)
 125 # ifdef inline
 126 # undef inline
 127 # endif
 128 # define inline
 129 #endif
 130
 131 /* Number of single byte character.  */
 132 #define SBC_MAX 256
 133
 134 #define COLL_ELEM_LEN_MAX 8
 135
 136 /* The character which represents newline.  */
 137 #define NEWLINE_CHAR '\n'
 138 #define WIDE_NEWLINE_CHAR L'\n'
 139
 140 /* Rename to standard API for using out of glibc.  */
 141 #ifndef _LIBC
 142 # ifdef __wctype
 143 # undef __wctype
 144 # endif
 145 # define __wctype wctype
 146 # ifdef __iswctype
 147 # undef __iswctype
 148 # endif
 149 # define __iswctype iswctype
 150 # define __btowc btowc
 151 # define __mbrtowc mbrtowc
 152 #undef __mempcpy        /* GAWK */
 153 # define __mempcpy mempcpy
 154 # define __wcrtomb wcrtomb
 155 # define __regfree regfree
 156 # define attribute_hidden
 157 #endif /* not _LIBC */
 158
 159 #ifdef __GNUC__
 160 # define __attribute(arg) __attribute__ (arg)
 161 #else
 162 # define __attribute(arg)
 163 #endif
 164
 165 extern const char __re_error_msgid[] attribute_hidden;
 166 extern const size_t __re_error_msgid_idx[] attribute_hidden;
 167
 168 /* An integer used to represent a set of bits.  It must be unsigned,
 169    and must be at least as wide as unsigned int.  */
 170 typedef unsigned long int bitset_word_t;
 171 /* All bits set in a bitset_word_t.  */
 172 #define BITSET_WORD_MAX ULONG_MAX
 173 /* Number of bits in a bitset_word_t.  */
 174 #define BITSET_WORD_BITS (sizeof (bitset_word_t) * CHAR_BIT)
 175 /* Number of bitset_word_t in a bit_set.  */
 176 #define BITSET_WORDS (SBC_MAX / BITSET_WORD_BITS)
 177 typedef bitset_word_t bitset_t[BITSET_WORDS];
 178 typedef bitset_word_t *re_bitset_ptr_t;
 179 typedef const bitset_word_t *re_const_bitset_ptr_t;
 180
 181 #define bitset_set(set,i) \
 182   (set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS)
 183 #define bitset_clear(set,i) \
 184   (set[i / BITSET_WORD_BITS] &= ~((bitset_word_t) 1 << i % BITSET_WORD_BITS))
 185 #define bitset_contain(set,i) \
 186   (set[i / BITSET_WORD_BITS] & ((bitset_word_t) 1 << i % BITSET_WORD_BITS))
 187 #define bitset_empty(set) memset (set, '\0', sizeof (bitset_t))
 188 #define bitset_set_all(set) memset (set, '\xff', sizeof (bitset_t))
 189 #define bitset_copy(dest,src) memcpy (dest, src, sizeof (bitset_t))
 190
 191 #define PREV_WORD_CONSTRAINT 0x0001
 192 #define PREV_NOTWORD_CONSTRAINT 0x0002
 193 #define NEXT_WORD_CONSTRAINT 0x0004
 194 #define NEXT_NOTWORD_CONSTRAINT 0x0008
 195 #define PREV_NEWLINE_CONSTRAINT 0x0010
 196 #define NEXT_NEWLINE_CONSTRAINT 0x0020
 197 #define PREV_BEGBUF_CONSTRAINT 0x0040
 198 #define NEXT_ENDBUF_CONSTRAINT 0x0080
 199 #define WORD_DELIM_CONSTRAINT 0x0100
 200 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
 201
 202 typedef enum
 203 {
 204   INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 205   WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 206   WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 207   INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 208   LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
 209   LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
 210   BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
 211   BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
 212   WORD_DELIM = WORD_DELIM_CONSTRAINT,
 213   NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
 214 } re_context_type;
 215
 216 typedef struct
 217 {
 218   int alloc;
 219   int nelem;
 220   int *elems;
 221 } re_node_set;
 222
 223 typedef enum
 224 {
 225   NON_TYPE = 0,
 226
 227   /* Node type, These are used by token, node, tree.  */
 228   CHARACTER = 1,
 229   END_OF_RE = 2,
 230   SIMPLE_BRACKET = 3,
 231   OP_BACK_REF = 4,
 232   OP_PERIOD = 5,
 233 #ifdef RE_ENABLE_I18N
 234   COMPLEX_BRACKET = 6,
 235   OP_UTF8_PERIOD = 7,
 236 #endif /* RE_ENABLE_I18N */
 237
 238   /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
 239      when the debugger shows values of this enum type.  */
 240 #define EPSILON_BIT 8
 241   OP_OPEN_SUBEXP = EPSILON_BIT | 0,
 242   OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
 243   OP_ALT = EPSILON_BIT | 2,
 244   OP_DUP_ASTERISK = EPSILON_BIT | 3,
 245   ANCHOR = EPSILON_BIT | 4,
 246
 247   /* Tree type, these are used only by tree. */
 248   CONCAT = 16,
 249   SUBEXP = 17,
 250
 251   /* Token type, these are used only by token.  */
 252   OP_DUP_PLUS = 18,
 253   OP_DUP_QUESTION,
 254   OP_OPEN_BRACKET,
 255   OP_CLOSE_BRACKET,
 256   OP_CHARSET_RANGE,
 257   OP_OPEN_DUP_NUM,
 258   OP_CLOSE_DUP_NUM,
 259   OP_NON_MATCH_LIST,
 260   OP_OPEN_COLL_ELEM,
 261   OP_CLOSE_COLL_ELEM,
 262   OP_OPEN_EQUIV_CLASS,
 263   OP_CLOSE_EQUIV_CLASS,
 264   OP_OPEN_CHAR_CLASS,
 265   OP_CLOSE_CHAR_CLASS,
 266   OP_WORD,
 267   OP_NOTWORD,
 268   OP_SPACE,
 269   OP_NOTSPACE,
 270   BACK_SLASH
 271
 272 } re_token_type_t;
 273
 274 #ifdef RE_ENABLE_I18N
 275 typedef struct
 276 {
 277   /* Multibyte characters.  */
 278   wchar_t *mbchars;
 279
 280   /* Collating symbols.  */
 281 # ifdef _LIBC
 282   int32_t *coll_syms;
 283 # endif
 284
 285   /* Equivalence classes. */
 286 # ifdef _LIBC
 287   int32_t *equiv_classes;
 288 # endif
 289
 290   /* Range expressions. */
 291 # ifdef _LIBC
 292   uint32_t *range_starts;
 293   uint32_t *range_ends;
 294 # else /* not _LIBC */
 295   wchar_t *range_starts;
 296   wchar_t *range_ends;
 297 # endif /* not _LIBC */
 298
 299   /* Character classes. */
 300   wctype_t *char_classes;
 301
 302   /* If this character set is the non-matching list.  */
 303   unsigned int non_match : 1;
 304
 305   /* # of multibyte characters.  */
 306   int nmbchars;
 307
 308   /* # of collating symbols.  */
 309   int ncoll_syms;
 310
 311   /* # of equivalence classes. */
 312   int nequiv_classes;
 313
 314   /* # of range expressions. */
 315   int nranges;
 316
 317   /* # of character classes. */
 318   int nchar_classes;
 319 } re_charset_t;
 320 #endif /* RE_ENABLE_I18N */
 321
 322 typedef struct
 323 {
 324   union
 325   {
 326     unsigned char c;            /* for CHARACTER */
 327     re_bitset_ptr_t sbcset;     /* for SIMPLE_BRACKET */
 328 #ifdef RE_ENABLE_I18N
 329     re_charset_t *mbcset;       /* for COMPLEX_BRACKET */
 330 #endif /* RE_ENABLE_I18N */
 331     int idx;                    /* for BACK_REF */
 332     re_context_type ctx_type;   /* for ANCHOR */
 333   } opr;
 334 #if __GNUC__ >= 2
 335   re_token_type_t type : 8;
 336 #else
 337   re_token_type_t type;
 338 #endif
 339   unsigned int constraint : 10; /* context constraint */
 340   unsigned int duplicated : 1;
 341   unsigned int opt_subexp : 1;
 342 #ifdef RE_ENABLE_I18N
 343   unsigned int accept_mb : 1;
 344   /* These 2 bits can be moved into the union if needed (e.g. if running out
 345      of bits; move opr.c to opr.c.c and move the flags to opr.c.flags).  */
 346   unsigned int mb_partial : 1;
 347 #endif
 348   unsigned int word_char : 1;
 349 } re_token_t;
 350
 351 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
 352
 353 struct re_string_t
 354 {
 355   /* Indicate the raw buffer which is the original string passed as an
 356      argument of regexec(), re_search(), etc..  */
 357   const unsigned char *raw_mbs;
 358   /* Store the multibyte string.  In case of "case insensitive mode" like
 359      REG_ICASE, upper cases of the string are stored, otherwise MBS points
 360      the same address that RAW_MBS points.  */
 361   unsigned char *mbs;
 362 #ifdef RE_ENABLE_I18N
 363   /* Store the wide character string which is corresponding to MBS.  */
 364   wint_t *wcs;
 365   int *offsets;
 366   mbstate_t cur_state;
 367 #endif
 368   /* Index in RAW_MBS.  Each character mbs[i] corresponds to
 369      raw_mbs[raw_mbs_idx + i].  */
 370   int raw_mbs_idx;
 371   /* The length of the valid characters in the buffers.  */
 372   int valid_len;
 373   /* The corresponding number of bytes in raw_mbs array.  */
 374   int valid_raw_len;
 375   /* The length of the buffers MBS and WCS.  */
 376   int bufs_len;
 377   /* The index in MBS, which is updated by re_string_fetch_byte.  */
 378   int cur_idx;
 379   /* length of RAW_MBS array.  */
 380   int raw_len;
 381   /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN.  */
 382   int len;
 383   /* End of the buffer may be shorter than its length in the cases such
 384      as re_match_2, re_search_2.  Then, we use STOP for end of the buffer
 385      instead of LEN.  */
 386   int raw_stop;
 387   /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS.  */
 388   int stop;
 389
 390   /* The context of mbs[0].  We store the context independently, since
 391      the context of mbs[0] may be different from raw_mbs[0], which is
 392      the beginning of the input string.  */
 393   unsigned int tip_context;
 394   /* The translation passed as a part of an argument of re_compile_pattern.  */
 395   RE_TRANSLATE_TYPE trans;
 396   /* Copy of re_dfa_t's word_char.  */
 397   re_const_bitset_ptr_t word_char;
 398   /* 1 if REG_ICASE.  */
 399   unsigned char icase;
 400   unsigned char is_utf8;
 401   unsigned char map_notascii;
 402   unsigned char mbs_allocated;
 403   unsigned char offsets_needed;
 404   unsigned char newline_anchor;
 405   unsigned char word_ops_used;
 406   int mb_cur_max;
 407 };
 408 typedef struct re_string_t re_string_t;
 409
 410
 411 struct re_dfa_t;
 412 typedef struct re_dfa_t re_dfa_t;
 413
 414 #ifndef _LIBC
 415 # ifdef __i386__
 416 #  define internal_function   __attribute ((regparm (3), stdcall))
 417 # else
 418 #  define internal_function
 419 # endif
 420 #endif
 421
 422 #ifndef NOT_IN_libc
 423 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
 424                                                 int new_buf_len)
 425      internal_function;
 426 # ifdef RE_ENABLE_I18N
 427 static void build_wcs_buffer (re_string_t *pstr) internal_function;
 428 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
 429   internal_function;
 430 # endif /* RE_ENABLE_I18N */
 431 static void build_upper_buffer (re_string_t *pstr) internal_function;
 432 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
 433 static unsigned int re_string_context_at (const re_string_t *input, int idx,
 434                                           int eflags)
 435      internal_function __attribute ((pure));
 436 #endif
 437 #define re_string_peek_byte(pstr, offset) \
 438   ((pstr)->mbs[(pstr)->cur_idx + offset])
 439 #define re_string_fetch_byte(pstr) \
 440   ((pstr)->mbs[(pstr)->cur_idx++])
 441 #define re_string_first_byte(pstr, idx) \
 442   ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
 443 #define re_string_is_single_byte_char(pstr, idx) \
 444   ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
 445                                 || (pstr)->wcs[(idx) + 1] != WEOF))
 446 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
 447 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
 448 #define re_string_get_buffer(pstr) ((pstr)->mbs)
 449 #define re_string_length(pstr) ((pstr)->len)
 450 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
 451 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
 452 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
 453
 454 #ifndef _LIBC
 455 # if HAVE_ALLOCA
 456 #  if (_MSC_VER)
 457 #   include <malloc.h>
 458 #   define __libc_use_alloca(n) 0
 459 #  else
 460 #   include <alloca.h>
 461 /* The OS usually guarantees only one guard page at the bottom of the stack,
 462    and a page size can be as small as 4096 bytes.  So we cannot safely
 463    allocate anything larger than 4096 bytes.  Also care for the possibility
 464    of a few compiler-allocated temporary stack slots.  */
 465 #  define __libc_use_alloca(n) ((n) < 4032)
 466 #  endif
 467 # else
 468 /* alloca is implemented with malloc, so just use malloc.  */
 469 #  define __libc_use_alloca(n) 0
 470 # endif
 471 #endif
 472
 473 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
 474 /* SunOS 4.1.x realloc doesn't accept null pointers: pre-Standard C. Sigh. */
 475 #define re_realloc(p,t,n) ((p != NULL) ? (t *) realloc (p,(n)*sizeof(t)) : (t *) calloc(n,sizeof(t)))
 476 #define re_free(p) free (p)
 477
 478 struct bin_tree_t
 479 {
 480   struct bin_tree_t *parent;
 481   struct bin_tree_t *left;
 482   struct bin_tree_t *right;
 483   struct bin_tree_t *first;
 484   struct bin_tree_t *next;
 485
 486   re_token_t token;
 487
 488   /* `node_idx' is the index in dfa->nodes, if `type' == 0.
 489      Otherwise `type' indicate the type of this node.  */
 490   int node_idx;
 491 };
 492 typedef struct bin_tree_t bin_tree_t;
 493
 494 #define BIN_TREE_STORAGE_SIZE \
 495   ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
 496
 497 struct bin_tree_storage_t
 498 {
 499   struct bin_tree_storage_t *next;
 500   bin_tree_t data[BIN_TREE_STORAGE_SIZE];
 501 };
 502 typedef struct bin_tree_storage_t bin_tree_storage_t;
 503
 504 #define CONTEXT_WORD 1
 505 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
 506 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
 507 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
 508
 509 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
 510 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
 511 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
 512 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
 513 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
 514
 515 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
 516 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
 517 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
 518 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
 519
 520 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
 521  ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 522   || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 523   || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
 524   || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
 525
 526 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
 527  ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 528   || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 529   || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
 530   || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
 531
 532 struct re_dfastate_t
 533 {
 534   unsigned int hash;
 535   re_node_set nodes;
 536   re_node_set non_eps_nodes;
 537   re_node_set inveclosure;
 538   re_node_set *entrance_nodes;
 539   struct re_dfastate_t **trtable, **word_trtable;
 540   unsigned int context : 4;
 541   unsigned int halt : 1;
 542   /* If this state can accept `multi byte'.
 543      Note that we refer to multibyte characters, and multi character
 544      collating elements as `multi byte'.  */
 545   unsigned int accept_mb : 1;
 546   /* If this state has backreference node(s).  */
 547   unsigned int has_backref : 1;
 548   unsigned int has_constraint : 1;
 549 };
 550 typedef struct re_dfastate_t re_dfastate_t;
 551
 552 struct re_state_table_entry
 553 {
 554   int num;
 555   int alloc;
 556   re_dfastate_t **array;
 557 };
 558
 559 /* Array type used in re_sub_match_last_t and re_sub_match_top_t.  */
 560
 561 typedef struct
 562 {
 563   int next_idx;
 564   int alloc;
 565   re_dfastate_t **array;
 566 } state_array_t;
 567
 568 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP.  */
 569
 570 typedef struct
 571 {
 572   int node;
 573   int str_idx; /* The position NODE match at.  */
 574   state_array_t path;
 575 } re_sub_match_last_t;
 576
 577 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
 578    And information about the node, whose type is OP_CLOSE_SUBEXP,
 579    corresponding to NODE is stored in LASTS.  */
 580
 581 typedef struct
 582 {
 583   int str_idx;
 584   int node;
 585   state_array_t *path;
 586   int alasts; /* Allocation size of LASTS.  */
 587   int nlasts; /* The number of LASTS.  */
 588   re_sub_match_last_t **lasts;
 589 } re_sub_match_top_t;
 590
 591 struct re_backref_cache_entry
 592 {
 593   int node;
 594   int str_idx;
 595   int subexp_from;
 596   int subexp_to;
 597   char more;
 598   char unused;
 599   unsigned short int eps_reachable_subexps_map;
 600 };
 601
 602 typedef struct
 603 {
 604   /* The string object corresponding to the input string.  */
 605   re_string_t input;
 606 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
 607   const re_dfa_t *const dfa;
 608 #else
 609   const re_dfa_t *dfa;
 610 #endif
 611   /* EFLAGS of the argument of regexec.  */
 612   int eflags;
 613   /* Where the matching ends.  */
 614   int match_last;
 615   int last_node;
 616   /* The state log used by the matcher.  */
 617   re_dfastate_t **state_log;
 618   int state_log_top;
 619   /* Back reference cache.  */
 620   int nbkref_ents;
 621   int abkref_ents;
 622   struct re_backref_cache_entry *bkref_ents;
 623   int max_mb_elem_len;
 624   int nsub_tops;
 625   int asub_tops;
 626   re_sub_match_top_t **sub_tops;
 627 } re_match_context_t;
 628
 629 typedef struct
 630 {
 631   re_dfastate_t **sifted_states;
 632   re_dfastate_t **limited_states;
 633   int last_node;
 634   int last_str_idx;
 635   re_node_set limits;
 636 } re_sift_context_t;
 637
 638 struct re_fail_stack_ent_t
 639 {
 640   int idx;
 641   int node;
 642   regmatch_t *regs;
 643   re_node_set eps_via_nodes;
 644 };
 645
 646 struct re_fail_stack_t
 647 {
 648   int num;
 649   int alloc;
 650   struct re_fail_stack_ent_t *stack;
 651 };
 652
 653 struct re_dfa_t
 654 {
 655   re_token_t *nodes;
 656   size_t nodes_alloc;
 657   size_t nodes_len;
 658   int *nexts;
 659   int *org_indices;
 660   re_node_set *edests;
 661   re_node_set *eclosures;
 662   re_node_set *inveclosures;
 663   struct re_state_table_entry *state_table;
 664   re_dfastate_t *init_state;
 665   re_dfastate_t *init_state_word;
 666   re_dfastate_t *init_state_nl;
 667   re_dfastate_t *init_state_begbuf;
 668   bin_tree_t *str_tree;
 669   bin_tree_storage_t *str_tree_storage;
 670   re_bitset_ptr_t sb_char;
 671   int str_tree_storage_idx;
 672
 673   /* number of subexpressions `re_nsub' is in regex_t.  */
 674   unsigned int state_hash_mask;
 675   int init_node;
 676   int nbackref; /* The number of backreference in this dfa.  */
 677
 678   /* Bitmap expressing which backreference is used.  */
 679   bitset_word_t used_bkref_map;
 680   bitset_word_t completed_bkref_map;
 681
 682   unsigned int has_plural_match : 1;
 683   /* If this dfa has "multibyte node", which is a backreference or
 684      a node which can accept multibyte character or multi character
 685      collating element.  */
 686   unsigned int has_mb_node : 1;
 687   unsigned int is_utf8 : 1;
 688   unsigned int map_notascii : 1;
 689   unsigned int word_ops_used : 1;
 690   int mb_cur_max;
 691   bitset_t word_char;
 692   reg_syntax_t syntax;
 693   int *subexp_map;
 694 #ifdef DEBUG
 695   char* re_str;
 696 #endif
 697 #if defined _LIBC
 698   __libc_lock_define (, lock)
 699 #endif
 700 };
 701
 702 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
 703 #define re_node_set_remove(set,id) \
 704   (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
 705 #define re_node_set_empty(p) ((p)->nelem = 0)
 706 #define re_node_set_free(set) re_free ((set)->elems)
 707 \f
 708
 709 typedef enum
 710 {
 711   SB_CHAR,
 712   MB_CHAR,
 713   EQUIV_CLASS,
 714   COLL_SYM,
 715   CHAR_CLASS
 716 } bracket_elem_type;
 717
 718 typedef struct
 719 {
 720   bracket_elem_type type;
 721   union
 722   {
 723     unsigned char ch;
 724     unsigned char *name;
 725     wchar_t wch;
 726   } opr;
 727 } bracket_elem_t;
 728
 729
 730 /* Inline functions for bitset operation.  */
 731 static inline void
 732 bitset_not (bitset_t set)
 733 {
 734   int bitset_i;
 735   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 736     set[bitset_i] = ~set[bitset_i];
 737 }
 738
 739 static inline void
 740 bitset_merge (bitset_t dest, const bitset_t src)
 741 {
 742   int bitset_i;
 743   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 744     dest[bitset_i] |= src[bitset_i];
 745 }
 746
 747 static inline void
 748 bitset_mask (bitset_t dest, const bitset_t src)
 749 {
 750   int bitset_i;
 751   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 752     dest[bitset_i] &= src[bitset_i];
 753 }
 754
 755 #ifdef RE_ENABLE_I18N
 756 /* Inline functions for re_string.  */
 757 static inline int
 758 internal_function __attribute ((pure))
 759 re_string_char_size_at (const re_string_t *pstr, int idx)
 760 {
 761   int byte_idx;
 762   if (pstr->mb_cur_max == 1)
 763     return 1;
 764   for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
 765     if (pstr->wcs[idx + byte_idx] != WEOF)
 766       break;
 767   return byte_idx;
 768 }
 769
 770 static inline wint_t
 771 internal_function __attribute ((pure))
 772 re_string_wchar_at (const re_string_t *pstr, int idx)
 773 {
 774   if (pstr->mb_cur_max == 1)
 775     return (wint_t) pstr->mbs[idx];
 776   return (wint_t) pstr->wcs[idx];
 777 }
 778
 779 # ifndef NOT_IN_libc
 780 static int
 781 internal_function __attribute ((pure))
 782 re_string_elem_size_at (const re_string_t *pstr, int idx)
 783 {
 784 #  ifdef _LIBC
 785   const unsigned char *p, *extra;
 786   const int32_t *table, *indirect;
 787   int32_t tmp;
 788 #   include <locale/weight.h>
 789   uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
 790
 791   if (nrules != 0)
 792     {
 793       table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
 794       extra = (const unsigned char *)
 795         _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
 796       indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
 797                                                 _NL_COLLATE_INDIRECTMB);
 798       p = pstr->mbs + idx;
 799       tmp = findidx (&p);
 800       return p - pstr->mbs - idx;
 801     }
 802   else
 803 #  endif /* _LIBC */
 804     return 1;
 805 }
 806 # endif
 807 #endif /* RE_ENABLE_I18N */
 808
 809 #endif /*  _REGEX_INTERNAL_H */