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 _LIBC
  43 # include <bits/libc-lock.h>
  44 #else
  45 # define __libc_lock_define(CLASS,NAME)
  46 # define __libc_lock_init(NAME) do { } while (0)
  47 # define __libc_lock_lock(NAME) do { } while (0)
  48 # define __libc_lock_unlock(NAME) do { } while (0)
  49 #endif
  50
  51 /* In case that the system doesn't have isblank().  */
  52 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
  53 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
  54 #endif
  55
  56 #ifdef _LIBC
  57 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
  58 #  define _RE_DEFINE_LOCALE_FUNCTIONS 1
  59 #   include <locale/localeinfo.h>
  60 #   include <locale/elem-hash.h>
  61 #   include <locale/coll-lookup.h>
  62 # endif
  63 #endif
  64
  65 /* This is for other GNU distributions with internationalized messages.  */
  66 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
  67 # include <libintl.h>
  68 # ifdef _LIBC
  69 #  undef gettext
  70 #  define gettext(msgid) \
  71   INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
  72 # endif
  73 #else
  74 # define gettext(msgid) (msgid)
  75 #endif
  76
  77 #ifndef gettext_noop
  78 /* This define is so xgettext can find the internationalizable
  79    strings.  */
  80 # define gettext_noop(String) String
  81 #endif
  82
  83 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
  84 # define RE_ENABLE_I18N
  85 #endif
  86
  87 #if __GNUC__ >= 3
  88 # define BE(expr, val) __builtin_expect (expr, val)
  89 #else
  90 # define BE(expr, val) (expr)
  91 # define inline
  92 #endif
  93
  94 /* Number of bits in a byte.  */
  95 #define BYTE_BITS 8
  96 /* Number of single byte character.  */
  97 #define SBC_MAX 256
  98
  99 #define COLL_ELEM_LEN_MAX 8
 100
 101 /* The character which represents newline.  */
 102 #define NEWLINE_CHAR '\n'
 103 #define WIDE_NEWLINE_CHAR L'\n'
 104
 105 /* Rename to standard API for using out of glibc.  */
 106 #ifndef _LIBC
 107 # define __wctype wctype
 108 # define __iswctype iswctype
 109 # define __btowc btowc
 110 # define __mempcpy mempcpy
 111 # define __wcrtomb wcrtomb
 112 # define __regfree regfree
 113 # define attribute_hidden
 114 #endif /* not _LIBC */
 115
 116 #ifdef __GNUC__
 117 # define __attribute(arg) __attribute__ (arg)
 118 #else
 119 # define __attribute(arg)
 120 #endif
 121
 122 extern const char __re_error_msgid[] attribute_hidden;
 123 extern const size_t __re_error_msgid_idx[] attribute_hidden;
 124
 125 /* Number of bits in an unsinged int.  */
 126 #define UINT_BITS (sizeof (unsigned int) * BYTE_BITS)
 127 /* Number of unsigned int in an bit_set.  */
 128 #define BITSET_UINTS ((SBC_MAX + UINT_BITS - 1) / UINT_BITS)
 129 typedef unsigned int bitset[BITSET_UINTS];
 130 typedef unsigned int *re_bitset_ptr_t;
 131 typedef const unsigned int *re_const_bitset_ptr_t;
 132
 133 #define bitset_set(set,i) (set[i / UINT_BITS] |= 1 << i % UINT_BITS)
 134 #define bitset_clear(set,i) (set[i / UINT_BITS] &= ~(1 << i % UINT_BITS))
 135 #define bitset_contain(set,i) (set[i / UINT_BITS] & (1 << i % UINT_BITS))
 136 #define bitset_empty(set) memset (set, 0, sizeof (unsigned int) * BITSET_UINTS)
 137 #define bitset_set_all(set) \
 138   memset (set, 255, sizeof (unsigned int) * BITSET_UINTS)
 139 #define bitset_copy(dest,src) \
 140   memcpy (dest, src, sizeof (unsigned int) * BITSET_UINTS)
 141 static inline void bitset_not (bitset set);
 142 static inline void bitset_merge (bitset dest, const bitset src);
 143 static inline void bitset_not_merge (bitset dest, const bitset src);
 144 static inline void bitset_mask (bitset dest, const bitset src);
 145
 146 #define PREV_WORD_CONSTRAINT 0x0001
 147 #define PREV_NOTWORD_CONSTRAINT 0x0002
 148 #define NEXT_WORD_CONSTRAINT 0x0004
 149 #define NEXT_NOTWORD_CONSTRAINT 0x0008
 150 #define PREV_NEWLINE_CONSTRAINT 0x0010
 151 #define NEXT_NEWLINE_CONSTRAINT 0x0020
 152 #define PREV_BEGBUF_CONSTRAINT 0x0040
 153 #define NEXT_ENDBUF_CONSTRAINT 0x0080
 154 #define WORD_DELIM_CONSTRAINT 0x0100
 155 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
 156
 157 typedef enum
 158 {
 159   INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 160   WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 161   WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 162   INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 163   LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
 164   LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
 165   BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
 166   BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
 167   WORD_DELIM = WORD_DELIM_CONSTRAINT,
 168   NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
 169 } re_context_type;
 170
 171 typedef struct
 172 {
 173   int alloc;
 174   int nelem;
 175   int *elems;
 176 } re_node_set;
 177
 178 typedef enum
 179 {
 180   NON_TYPE = 0,
 181
 182   /* Node type, These are used by token, node, tree.  */
 183   CHARACTER = 1,
 184   END_OF_RE = 2,
 185   SIMPLE_BRACKET = 3,
 186   OP_BACK_REF = 4,
 187   OP_PERIOD = 5,
 188 #ifdef RE_ENABLE_I18N
 189   COMPLEX_BRACKET = 6,
 190   OP_UTF8_PERIOD = 7,
 191 #endif /* RE_ENABLE_I18N */
 192
 193   /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
 194      when the debugger shows values of this enum type.  */
 195 #define EPSILON_BIT 8
 196   OP_OPEN_SUBEXP = EPSILON_BIT | 0,
 197   OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
 198   OP_ALT = EPSILON_BIT | 2,
 199   OP_DUP_ASTERISK = EPSILON_BIT | 3,
 200   ANCHOR = EPSILON_BIT | 4,
 201
 202   /* Tree type, these are used only by tree. */
 203   CONCAT = 16,
 204   SUBEXP = 17,
 205
 206   /* Token type, these are used only by token.  */
 207   OP_DUP_PLUS = 18,
 208   OP_DUP_QUESTION,
 209   OP_OPEN_BRACKET,
 210   OP_CLOSE_BRACKET,
 211   OP_CHARSET_RANGE,
 212   OP_OPEN_DUP_NUM,
 213   OP_CLOSE_DUP_NUM,
 214   OP_NON_MATCH_LIST,
 215   OP_OPEN_COLL_ELEM,
 216   OP_CLOSE_COLL_ELEM,
 217   OP_OPEN_EQUIV_CLASS,
 218   OP_CLOSE_EQUIV_CLASS,
 219   OP_OPEN_CHAR_CLASS,
 220   OP_CLOSE_CHAR_CLASS,
 221   OP_WORD,
 222   OP_NOTWORD,
 223   OP_SPACE,
 224   OP_NOTSPACE,
 225   BACK_SLASH
 226
 227 } re_token_type_t;
 228
 229 #ifdef RE_ENABLE_I18N
 230 typedef struct
 231 {
 232   /* Multibyte characters.  */
 233   wchar_t *mbchars;
 234
 235   /* Collating symbols.  */
 236 # ifdef _LIBC
 237   int32_t *coll_syms;
 238 # endif
 239
 240   /* Equivalence classes. */
 241 # ifdef _LIBC
 242   int32_t *equiv_classes;
 243 # endif
 244
 245   /* Range expressions. */
 246 # ifdef _LIBC
 247   uint32_t *range_starts;
 248   uint32_t *range_ends;
 249 # else /* not _LIBC */
 250   wchar_t *range_starts;
 251   wchar_t *range_ends;
 252 # endif /* not _LIBC */
 253
 254   /* Character classes. */
 255   wctype_t *char_classes;
 256
 257   /* If this character set is the non-matching list.  */
 258   unsigned int non_match : 1;
 259
 260   /* # of multibyte characters.  */
 261   int nmbchars;
 262
 263   /* # of collating symbols.  */
 264   int ncoll_syms;
 265
 266   /* # of equivalence classes. */
 267   int nequiv_classes;
 268
 269   /* # of range expressions. */
 270   int nranges;
 271
 272   /* # of character classes. */
 273   int nchar_classes;
 274 } re_charset_t;
 275 #endif /* RE_ENABLE_I18N */
 276
 277 typedef struct
 278 {
 279   union
 280   {
 281     unsigned char c;            /* for CHARACTER */
 282     re_bitset_ptr_t sbcset;     /* for SIMPLE_BRACKET */
 283 #ifdef RE_ENABLE_I18N
 284     re_charset_t *mbcset;       /* for COMPLEX_BRACKET */
 285 #endif /* RE_ENABLE_I18N */
 286     int idx;                    /* for BACK_REF */
 287     re_context_type ctx_type;   /* for ANCHOR */
 288   } opr;
 289 #if __GNUC__ >= 2
 290   re_token_type_t type : 8;
 291 #else
 292   re_token_type_t type;
 293 #endif
 294   unsigned int constraint : 10; /* context constraint */
 295   unsigned int duplicated : 1;
 296   unsigned int opt_subexp : 1;
 297 #ifdef RE_ENABLE_I18N
 298   unsigned int accept_mb : 1;
 299   /* These 2 bits can be moved into the union if needed (e.g. if running out
 300      of bits; move opr.c to opr.c.c and move the flags to opr.c.flags).  */
 301   unsigned int mb_partial : 1;
 302 #endif
 303   unsigned int word_char : 1;
 304 } re_token_t;
 305
 306 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
 307
 308 struct re_string_t
 309 {
 310   /* Indicate the raw buffer which is the original string passed as an
 311      argument of regexec(), re_search(), etc..  */
 312   const unsigned char *raw_mbs;
 313   /* Store the multibyte string.  In case of "case insensitive mode" like
 314      REG_ICASE, upper cases of the string are stored, otherwise MBS points
 315      the same address that RAW_MBS points.  */
 316   unsigned char *mbs;
 317 #ifdef RE_ENABLE_I18N
 318   /* Store the wide character string which is corresponding to MBS.  */
 319   wint_t *wcs;
 320   int *offsets;
 321   mbstate_t cur_state;
 322 #endif
 323   /* Index in RAW_MBS.  Each character mbs[i] corresponds to
 324      raw_mbs[raw_mbs_idx + i].  */
 325   int raw_mbs_idx;
 326   /* The length of the valid characters in the buffers.  */
 327   int valid_len;
 328   /* The corresponding number of bytes in raw_mbs array.  */
 329   int valid_raw_len;
 330   /* The length of the buffers MBS and WCS.  */
 331   int bufs_len;
 332   /* The index in MBS, which is updated by re_string_fetch_byte.  */
 333   int cur_idx;
 334   /* length of RAW_MBS array.  */
 335   int raw_len;
 336   /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN.  */
 337   int len;
 338   /* End of the buffer may be shorter than its length in the cases such
 339      as re_match_2, re_search_2.  Then, we use STOP for end of the buffer
 340      instead of LEN.  */
 341   int raw_stop;
 342   /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS.  */
 343   int stop;
 344
 345   /* The context of mbs[0].  We store the context independently, since
 346      the context of mbs[0] may be different from raw_mbs[0], which is
 347      the beginning of the input string.  */
 348   unsigned int tip_context;
 349   /* The translation passed as a part of an argument of re_compile_pattern.  */
 350   unsigned RE_TRANSLATE_TYPE trans;
 351   /* Copy of re_dfa_t's word_char.  */
 352   re_const_bitset_ptr_t word_char;
 353   /* 1 if REG_ICASE.  */
 354   unsigned char icase;
 355   unsigned char is_utf8;
 356   unsigned char map_notascii;
 357   unsigned char mbs_allocated;
 358   unsigned char offsets_needed;
 359   unsigned char newline_anchor;
 360   unsigned char word_ops_used;
 361   int mb_cur_max;
 362 };
 363 typedef struct re_string_t re_string_t;
 364
 365
 366 struct re_dfa_t;
 367 typedef struct re_dfa_t re_dfa_t;
 368
 369 #ifndef _LIBC
 370 # ifdef __i386__
 371 #  define internal_function   __attribute ((regparm (3), stdcall))
 372 # else
 373 #  define internal_function
 374 # endif
 375 #endif
 376
 377 #ifndef RE_NO_INTERNAL_PROTOTYPES
 378 static reg_errcode_t re_string_allocate (re_string_t *pstr, const char *str,
 379                                          int len, int init_len,
 380                                          RE_TRANSLATE_TYPE trans, int icase,
 381                                          const re_dfa_t *dfa)
 382      internal_function;
 383 static reg_errcode_t re_string_construct (re_string_t *pstr, const char *str,
 384                                           int len, RE_TRANSLATE_TYPE trans,
 385                                           int icase, const re_dfa_t *dfa)
 386      internal_function;
 387 static reg_errcode_t re_string_reconstruct (re_string_t *pstr, int idx,
 388                                             int eflags) internal_function;
 389 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
 390                                                 int new_buf_len)
 391      internal_function;
 392 # ifdef RE_ENABLE_I18N
 393 static void build_wcs_buffer (re_string_t *pstr) internal_function;
 394 static int build_wcs_upper_buffer (re_string_t *pstr) internal_function;
 395 # endif /* RE_ENABLE_I18N */
 396 static void build_upper_buffer (re_string_t *pstr) internal_function;
 397 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
 398 static void re_string_destruct (re_string_t *pstr) internal_function;
 399 # ifdef RE_ENABLE_I18N
 400 static int re_string_elem_size_at (const re_string_t *pstr, int idx)
 401      internal_function __attribute ((pure));
 402 static inline int re_string_char_size_at (const re_string_t *pstr, int idx)
 403      internal_function __attribute ((pure));
 404 static inline wint_t re_string_wchar_at (const re_string_t *pstr, int idx)
 405      internal_function __attribute ((pure));
 406 # endif /* RE_ENABLE_I18N */
 407 static unsigned int re_string_context_at (const re_string_t *input, int idx,
 408                                           int eflags)
 409      internal_function __attribute ((pure));
 410 static unsigned char re_string_peek_byte_case (const re_string_t *pstr,
 411                                                int idx)
 412      internal_function __attribute ((pure));
 413 static unsigned char re_string_fetch_byte_case (re_string_t *pstr)
 414      internal_function __attribute ((pure));
 415 #endif
 416 #define re_string_peek_byte(pstr, offset) \
 417   ((pstr)->mbs[(pstr)->cur_idx + offset])
 418 #define re_string_fetch_byte(pstr) \
 419   ((pstr)->mbs[(pstr)->cur_idx++])
 420 #define re_string_first_byte(pstr, idx) \
 421   ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
 422 #define re_string_is_single_byte_char(pstr, idx) \
 423   ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
 424                                 || (pstr)->wcs[(idx) + 1] != WEOF))
 425 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
 426 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
 427 #define re_string_get_buffer(pstr) ((pstr)->mbs)
 428 #define re_string_length(pstr) ((pstr)->len)
 429 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
 430 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
 431 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
 432
 433 #include <alloca.h>
 434
 435 #ifndef _LIBC
 436 # if HAVE_ALLOCA
 437 /* The OS usually guarantees only one guard page at the bottom of the stack,
 438    and a page size can be as small as 4096 bytes.  So we cannot safely
 439    allocate anything larger than 4096 bytes.  Also care for the possibility
 440    of a few compiler-allocated temporary stack slots.  */
 441 #  define __libc_use_alloca(n) ((n) < 4032)
 442 # else
 443 /* alloca is implemented with malloc, so just use malloc.  */
 444 #  define __libc_use_alloca(n) 0
 445 # endif
 446 #endif
 447
 448 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
 449 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
 450 #define re_free(p) free (p)
 451
 452 struct bin_tree_t
 453 {
 454   struct bin_tree_t *parent;
 455   struct bin_tree_t *left;
 456   struct bin_tree_t *right;
 457   struct bin_tree_t *first;
 458   struct bin_tree_t *next;
 459
 460   re_token_t token;
 461
 462   /* `node_idx' is the index in dfa->nodes, if `type' == 0.
 463      Otherwise `type' indicate the type of this node.  */
 464   int node_idx;
 465 };
 466 typedef struct bin_tree_t bin_tree_t;
 467
 468 #define BIN_TREE_STORAGE_SIZE \
 469   ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
 470
 471 struct bin_tree_storage_t
 472 {
 473   struct bin_tree_storage_t *next;
 474   bin_tree_t data[BIN_TREE_STORAGE_SIZE];
 475 };
 476 typedef struct bin_tree_storage_t bin_tree_storage_t;
 477
 478 #define CONTEXT_WORD 1
 479 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
 480 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
 481 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
 482
 483 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
 484 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
 485 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
 486 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
 487 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
 488
 489 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
 490 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
 491 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
 492 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
 493
 494 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
 495  ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 496   || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 497   || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
 498   || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
 499
 500 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
 501  ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 502   || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 503   || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
 504   || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
 505
 506 struct re_dfastate_t
 507 {
 508   unsigned int hash;
 509   re_node_set nodes;
 510   re_node_set non_eps_nodes;
 511   re_node_set inveclosure;
 512   re_node_set *entrance_nodes;
 513   struct re_dfastate_t **trtable, **word_trtable;
 514   unsigned int context : 4;
 515   unsigned int halt : 1;
 516   /* If this state can accept `multi byte'.
 517      Note that we refer to multibyte characters, and multi character
 518      collating elements as `multi byte'.  */
 519   unsigned int accept_mb : 1;
 520   /* If this state has backreference node(s).  */
 521   unsigned int has_backref : 1;
 522   unsigned int has_constraint : 1;
 523 };
 524 typedef struct re_dfastate_t re_dfastate_t;
 525
 526 struct re_state_table_entry
 527 {
 528   int num;
 529   int alloc;
 530   re_dfastate_t **array;
 531 };
 532
 533 /* Array type used in re_sub_match_last_t and re_sub_match_top_t.  */
 534
 535 typedef struct
 536 {
 537   int next_idx;
 538   int alloc;
 539   re_dfastate_t **array;
 540 } state_array_t;
 541
 542 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP.  */
 543
 544 typedef struct
 545 {
 546   int node;
 547   int str_idx; /* The position NODE match at.  */
 548   state_array_t path;
 549 } re_sub_match_last_t;
 550
 551 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
 552    And information about the node, whose type is OP_CLOSE_SUBEXP,
 553    corresponding to NODE is stored in LASTS.  */
 554
 555 typedef struct
 556 {
 557   int str_idx;
 558   int node;
 559   state_array_t *path;
 560   int alasts; /* Allocation size of LASTS.  */
 561   int nlasts; /* The number of LASTS.  */
 562   re_sub_match_last_t **lasts;
 563 } re_sub_match_top_t;
 564
 565 struct re_backref_cache_entry
 566 {
 567   int node;
 568   int str_idx;
 569   int subexp_from;
 570   int subexp_to;
 571   char more;
 572   char unused;
 573   unsigned short int eps_reachable_subexps_map;
 574 };
 575
 576 typedef struct
 577 {
 578   /* The string object corresponding to the input string.  */
 579   re_string_t input;
 580 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
 581   re_dfa_t *const dfa;
 582 #else
 583   re_dfa_t *dfa;
 584 #endif
 585   /* EFLAGS of the argument of regexec.  */
 586   int eflags;
 587   /* Where the matching ends.  */
 588   int match_last;
 589   int last_node;
 590   /* The state log used by the matcher.  */
 591   re_dfastate_t **state_log;
 592   int state_log_top;
 593   /* Back reference cache.  */
 594   int nbkref_ents;
 595   int abkref_ents;
 596   struct re_backref_cache_entry *bkref_ents;
 597   int max_mb_elem_len;
 598   int nsub_tops;
 599   int asub_tops;
 600   re_sub_match_top_t **sub_tops;
 601 } re_match_context_t;
 602
 603 typedef struct
 604 {
 605   re_dfastate_t **sifted_states;
 606   re_dfastate_t **limited_states;
 607   int last_node;
 608   int last_str_idx;
 609   re_node_set limits;
 610 } re_sift_context_t;
 611
 612 struct re_fail_stack_ent_t
 613 {
 614   int idx;
 615   int node;
 616   regmatch_t *regs;
 617   re_node_set eps_via_nodes;
 618 };
 619
 620 struct re_fail_stack_t
 621 {
 622   int num;
 623   int alloc;
 624   struct re_fail_stack_ent_t *stack;
 625 };
 626
 627 struct re_dfa_t
 628 {
 629   re_token_t *nodes;
 630   int nodes_alloc;
 631   int nodes_len;
 632   int *nexts;
 633   int *org_indices;
 634   re_node_set *edests;
 635   re_node_set *eclosures;
 636   re_node_set *inveclosures;
 637   struct re_state_table_entry *state_table;
 638   re_dfastate_t *init_state;
 639   re_dfastate_t *init_state_word;
 640   re_dfastate_t *init_state_nl;
 641   re_dfastate_t *init_state_begbuf;
 642   bin_tree_t *str_tree;
 643   bin_tree_storage_t *str_tree_storage;
 644   re_bitset_ptr_t sb_char;
 645   int str_tree_storage_idx;
 646
 647   /* number of subexpressions `re_nsub' is in regex_t.  */
 648   unsigned int state_hash_mask;
 649   int init_node;
 650   int nbackref; /* The number of backreference in this dfa.  */
 651
 652   /* Bitmap expressing which backreference is used.  */
 653   unsigned int used_bkref_map;
 654   unsigned int completed_bkref_map;
 655
 656   unsigned int has_plural_match : 1;
 657   /* If this dfa has "multibyte node", which is a backreference or
 658      a node which can accept multibyte character or multi character
 659      collating element.  */
 660   unsigned int has_mb_node : 1;
 661   unsigned int is_utf8 : 1;
 662   unsigned int map_notascii : 1;
 663   unsigned int word_ops_used : 1;
 664   int mb_cur_max;
 665   bitset word_char;
 666   reg_syntax_t syntax;
 667   int *subexp_map;
 668 #ifdef DEBUG
 669   char* re_str;
 670 #endif
 671   __libc_lock_define (, lock)
 672 };
 673
 674 #ifndef RE_NO_INTERNAL_PROTOTYPES
 675 static reg_errcode_t re_node_set_alloc (re_node_set *set, int size) internal_function;
 676 static reg_errcode_t re_node_set_init_1 (re_node_set *set, int elem) internal_function;
 677 static reg_errcode_t re_node_set_init_2 (re_node_set *set, int elem1,
 678                                          int elem2) internal_function;
 679 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
 680 static reg_errcode_t re_node_set_init_copy (re_node_set *dest,
 681                                             const re_node_set *src) internal_function;
 682 static reg_errcode_t re_node_set_add_intersect (re_node_set *dest,
 683                                                 const re_node_set *src1,
 684                                                 const re_node_set *src2) internal_function;
 685 static reg_errcode_t re_node_set_init_union (re_node_set *dest,
 686                                              const re_node_set *src1,
 687                                              const re_node_set *src2) internal_function;
 688 static reg_errcode_t re_node_set_merge (re_node_set *dest,
 689                                         const re_node_set *src) internal_function;
 690 static int re_node_set_insert (re_node_set *set, int elem) internal_function;
 691 static int re_node_set_insert_last (re_node_set *set,
 692                                     int elem) internal_function;
 693 static int re_node_set_compare (const re_node_set *set1,
 694                                 const re_node_set *set2)
 695      internal_function __attribute ((pure));
 696 static int re_node_set_contains (const re_node_set *set, int elem)
 697      internal_function __attribute ((pure));
 698 static void re_node_set_remove_at (re_node_set *set, int idx) internal_function;
 699 #define re_node_set_remove(set,id) \
 700   (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
 701 #define re_node_set_empty(p) ((p)->nelem = 0)
 702 #define re_node_set_free(set) re_free ((set)->elems)
 703 static int re_dfa_add_node (re_dfa_t *dfa, re_token_t token) internal_function;
 704 static re_dfastate_t *re_acquire_state (reg_errcode_t *err, re_dfa_t *dfa,
 705                                         const re_node_set *nodes) internal_function;
 706 static re_dfastate_t *re_acquire_state_context (reg_errcode_t *err,
 707                                                 re_dfa_t *dfa,
 708                                                 const re_node_set *nodes,
 709                                                 unsigned int context) internal_function;
 710 static void free_state (re_dfastate_t *state) internal_function;
 711 #endif
 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 operation.  */
 736 static inline void
 737 bitset_not (bitset set)
 738 {
 739   int bitset_i;
 740   for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
 741     set[bitset_i] = ~set[bitset_i];
 742 }
 743
 744 static inline void
 745 bitset_merge (bitset dest, const bitset src)
 746 {
 747   int bitset_i;
 748   for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
 749     dest[bitset_i] |= src[bitset_i];
 750 }
 751
 752 static inline void
 753 bitset_not_merge (bitset dest, const bitset src)
 754 {
 755   int i;
 756   for (i = 0; i < BITSET_UINTS; ++i)
 757     dest[i] |= ~src[i];
 758 }
 759
 760 static inline void
 761 bitset_mask (bitset dest, const bitset src)
 762 {
 763   int bitset_i;
 764   for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
 765     dest[bitset_i] &= src[bitset_i];
 766 }
 767
 768 #if defined RE_ENABLE_I18N && !defined RE_NO_INTERNAL_PROTOTYPES
 769 /* Inline functions for re_string.  */
 770 static inline int
 771 internal_function
 772 re_string_char_size_at (const re_string_t *pstr, int idx)
 773 {
 774   int byte_idx;
 775   if (pstr->mb_cur_max == 1)
 776     return 1;
 777   for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
 778     if (pstr->wcs[idx + byte_idx] != WEOF)
 779       break;
 780   return byte_idx;
 781 }
 782
 783 static inline wint_t
 784 internal_function
 785 re_string_wchar_at (const re_string_t *pstr, int idx)
 786 {
 787   if (pstr->mb_cur_max == 1)
 788     return (wint_t) pstr->mbs[idx];
 789   return (wint_t) pstr->wcs[idx];
 790 }
 791
 792 static int
 793 internal_function
 794 re_string_elem_size_at (const re_string_t *pstr, int idx)
 795 {
 796 #ifdef _LIBC
 797   const unsigned char *p, *extra;
 798   const int32_t *table, *indirect;
 799   int32_t tmp;
 800 # include <locale/weight.h>
 801   uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
 802
 803   if (nrules != 0)
 804     {
 805       table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
 806       extra = (const unsigned char *)
 807         _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
 808       indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
 809                                                 _NL_COLLATE_INDIRECTMB);
 810       p = pstr->mbs + idx;
 811       tmp = findidx (&p);
 812       return p - pstr->mbs - idx;
 813     }
 814   else
 815 #endif /* _LIBC */
 816     return 1;
 817 }
 818 #endif /* RE_ENABLE_I18N */
 819
 820 #endif /*  _REGEX_INTERNAL_H */