posix/regex_internal.h

   1 /* Extended regular expression matching and search library.
   2    Copyright (C) 2002, 2003, 2004 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
  43 /* In case that the system doesn't have isblank().  */
  44 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
  45 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
  46 #endif
  47
  48 #ifdef _LIBC
  49 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
  50 #  define _RE_DEFINE_LOCALE_FUNCTIONS 1
  51 #   include <locale/localeinfo.h>
  52 #   include <locale/elem-hash.h>
  53 #   include <locale/coll-lookup.h>
  54 # endif
  55 #endif
  56
  57 /* This is for other GNU distributions with internationalized messages.  */
  58 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
  59 # include <libintl.h>
  60 # ifdef _LIBC
  61 #  undef gettext
  62 #  define gettext(msgid) \
  63   INTUSE(__dcgettext) (INTUSE(_libc_intl_domainname), msgid, LC_MESSAGES)
  64 # endif
  65 #else
  66 # define gettext(msgid) (msgid)
  67 #endif
  68
  69 #ifndef gettext_noop
  70 /* This define is so xgettext can find the internationalizable
  71    strings.  */
  72 # define gettext_noop(String) String
  73 #endif
  74
  75 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
  76 # define RE_ENABLE_I18N
  77 #endif
  78
  79 #if __GNUC__ >= 3
  80 # define BE(expr, val) __builtin_expect (expr, val)
  81 #else
  82 # define BE(expr, val) (expr)
  83 # define inline
  84 #endif
  85
  86 /* Number of bits in a byte.  */
  87 #define BYTE_BITS 8
  88 /* Number of single byte character.  */
  89 #define SBC_MAX 256
  90
  91 #define COLL_ELEM_LEN_MAX 8
  92
  93 /* The character which represents newline.  */
  94 #define NEWLINE_CHAR '\n'
  95 #define WIDE_NEWLINE_CHAR L'\n'
  96
  97 /* Rename to standard API for using out of glibc.  */
  98 #ifndef _LIBC
  99 # define __wctype wctype
 100 # define __iswctype iswctype
 101 # define __btowc btowc
 102 # define __mempcpy mempcpy
 103 # define __wcrtomb wcrtomb
 104 # define __regfree regfree
 105 # define attribute_hidden
 106 #endif /* not _LIBC */
 107
 108 #ifdef __GNUC__
 109 # define __attribute(arg) __attribute__ (arg)
 110 #else
 111 # define __attribute(arg)
 112 #endif
 113
 114 extern const char __re_error_msgid[] attribute_hidden;
 115 extern const size_t __re_error_msgid_idx[] attribute_hidden;
 116
 117 /* Number of bits in an unsinged int.  */
 118 #define UINT_BITS (sizeof (unsigned int) * BYTE_BITS)
 119 /* Number of unsigned int in an bit_set.  */
 120 #define BITSET_UINTS ((SBC_MAX + UINT_BITS - 1) / UINT_BITS)
 121 typedef unsigned int bitset[BITSET_UINTS];
 122 typedef unsigned int *re_bitset_ptr_t;
 123 typedef const unsigned int *re_const_bitset_ptr_t;
 124
 125 #define bitset_set(set,i) (set[i / UINT_BITS] |= 1 << i % UINT_BITS)
 126 #define bitset_clear(set,i) (set[i / UINT_BITS] &= ~(1 << i % UINT_BITS))
 127 #define bitset_contain(set,i) (set[i / UINT_BITS] & (1 << i % UINT_BITS))
 128 #define bitset_empty(set) memset (set, 0, sizeof (unsigned int) * BITSET_UINTS)
 129 #define bitset_set_all(set) \
 130   memset (set, 255, sizeof (unsigned int) * BITSET_UINTS)
 131 #define bitset_copy(dest,src) \
 132   memcpy (dest, src, sizeof (unsigned int) * BITSET_UINTS)
 133 static inline void bitset_not (bitset set);
 134 static inline void bitset_merge (bitset dest, const bitset src);
 135 static inline void bitset_not_merge (bitset dest, const bitset src);
 136 static inline void bitset_mask (bitset dest, const bitset src);
 137
 138 #define PREV_WORD_CONSTRAINT 0x0001
 139 #define PREV_NOTWORD_CONSTRAINT 0x0002
 140 #define NEXT_WORD_CONSTRAINT 0x0004
 141 #define NEXT_NOTWORD_CONSTRAINT 0x0008
 142 #define PREV_NEWLINE_CONSTRAINT 0x0010
 143 #define NEXT_NEWLINE_CONSTRAINT 0x0020
 144 #define PREV_BEGBUF_CONSTRAINT 0x0040
 145 #define NEXT_ENDBUF_CONSTRAINT 0x0080
 146 #define DUMMY_CONSTRAINT 0x0100
 147
 148 typedef enum
 149 {
 150   INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 151   WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 152   WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 153   LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
 154   LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
 155   BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
 156   BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
 157   WORD_DELIM = DUMMY_CONSTRAINT
 158 } re_context_type;
 159
 160 typedef struct
 161 {
 162   int alloc;
 163   int nelem;
 164   int *elems;
 165 } re_node_set;
 166
 167 typedef enum
 168 {
 169   NON_TYPE = 0,
 170
 171   /* Node type, These are used by token, node, tree.  */
 172   CHARACTER = 1,
 173   END_OF_RE = 2,
 174   SIMPLE_BRACKET = 3,
 175   OP_BACK_REF = 4,
 176   OP_PERIOD = 5,
 177 #ifdef RE_ENABLE_I18N
 178   COMPLEX_BRACKET = 6,
 179   OP_UTF8_PERIOD = 7,
 180 #endif /* RE_ENABLE_I18N */
 181
 182   /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
 183      when the debugger shows values of this enum type.  */
 184 #define EPSILON_BIT 8
 185   OP_OPEN_SUBEXP = EPSILON_BIT | 0,
 186   OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
 187   OP_ALT = EPSILON_BIT | 2,
 188   OP_DUP_ASTERISK = EPSILON_BIT | 3,
 189   OP_DUP_PLUS = EPSILON_BIT | 4,
 190   OP_DUP_QUESTION = EPSILON_BIT | 5,
 191   ANCHOR = EPSILON_BIT | 6,
 192   OP_DELETED_SUBEXP = EPSILON_BIT | 7,
 193
 194   /* Tree type, these are used only by tree. */
 195   CONCAT = 16,
 196
 197   /* Token type, these are used only by token.  */
 198   OP_OPEN_BRACKET = 17,
 199   OP_CLOSE_BRACKET,
 200   OP_CHARSET_RANGE,
 201   OP_OPEN_DUP_NUM,
 202   OP_CLOSE_DUP_NUM,
 203   OP_NON_MATCH_LIST,
 204   OP_OPEN_COLL_ELEM,
 205   OP_CLOSE_COLL_ELEM,
 206   OP_OPEN_EQUIV_CLASS,
 207   OP_CLOSE_EQUIV_CLASS,
 208   OP_OPEN_CHAR_CLASS,
 209   OP_CLOSE_CHAR_CLASS,
 210   OP_WORD,
 211   OP_NOTWORD,
 212   OP_SPACE,
 213   OP_NOTSPACE,
 214   BACK_SLASH
 215
 216 } re_token_type_t;
 217
 218 #ifdef RE_ENABLE_I18N
 219 typedef struct
 220 {
 221   /* Multibyte characters.  */
 222   wchar_t *mbchars;
 223
 224   /* Collating symbols.  */
 225 # ifdef _LIBC
 226   int32_t *coll_syms;
 227 # endif
 228
 229   /* Equivalence classes. */
 230 # ifdef _LIBC
 231   int32_t *equiv_classes;
 232 # endif
 233
 234   /* Range expressions. */
 235 # ifdef _LIBC
 236   uint32_t *range_starts;
 237   uint32_t *range_ends;
 238 # else /* not _LIBC */
 239   wchar_t *range_starts;
 240   wchar_t *range_ends;
 241 # endif /* not _LIBC */
 242
 243   /* Character classes. */
 244   wctype_t *char_classes;
 245
 246   /* If this character set is the non-matching list.  */
 247   unsigned int non_match : 1;
 248
 249   /* # of multibyte characters.  */
 250   int nmbchars;
 251
 252   /* # of collating symbols.  */
 253   int ncoll_syms;
 254
 255   /* # of equivalence classes. */
 256   int nequiv_classes;
 257
 258   /* # of range expressions. */
 259   int nranges;
 260
 261   /* # of character classes. */
 262   int nchar_classes;
 263 } re_charset_t;
 264 #endif /* RE_ENABLE_I18N */
 265
 266 typedef struct
 267 {
 268   union
 269   {
 270     unsigned char c;            /* for CHARACTER */
 271     re_bitset_ptr_t sbcset;     /* for SIMPLE_BRACKET */
 272 #ifdef RE_ENABLE_I18N
 273     re_charset_t *mbcset;       /* for COMPLEX_BRACKET */
 274 #endif /* RE_ENABLE_I18N */
 275     int idx;                    /* for BACK_REF */
 276     re_context_type ctx_type;   /* for ANCHOR */
 277   } opr;
 278 #if __GNUC__ >= 2
 279   re_token_type_t type : 8;
 280 #else
 281   re_token_type_t type;
 282 #endif
 283   unsigned int constraint : 10; /* context constraint */
 284   unsigned int duplicated : 1;
 285   unsigned int opt_subexp : 1;
 286 #ifdef RE_ENABLE_I18N
 287   /* These 2 bits can be moved into the union if needed (e.g. if running out
 288      of bits; move opr.c to opr.c.c and move the flags to opr.c.flags).  */
 289   unsigned int mb_partial : 1;
 290 #endif
 291   unsigned int word_char : 1;
 292 } re_token_t;
 293
 294 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
 295 #define ACCEPT_MB_NODE(type) \
 296   ((type) >= OP_PERIOD && (type) <= OP_UTF8_PERIOD)
 297
 298 struct re_string_t
 299 {
 300   /* Indicate the raw buffer which is the original string passed as an
 301      argument of regexec(), re_search(), etc..  */
 302   const unsigned char *raw_mbs;
 303   /* Store the multibyte string.  In case of "case insensitive mode" like
 304      REG_ICASE, upper cases of the string are stored, otherwise MBS points
 305      the same address that RAW_MBS points.  */
 306   unsigned char *mbs;
 307 #ifdef RE_ENABLE_I18N
 308   /* Store the wide character string which is corresponding to MBS.  */
 309   wint_t *wcs;
 310   int *offsets;
 311   mbstate_t cur_state;
 312 #endif
 313   /* Index in RAW_MBS.  Each character mbs[i] corresponds to
 314      raw_mbs[raw_mbs_idx + i].  */
 315   int raw_mbs_idx;
 316   /* The length of the valid characters in the buffers.  */
 317   int valid_len;
 318   /* The corresponding number of bytes in raw_mbs array.  */
 319   int valid_raw_len;
 320   /* The length of the buffers MBS and WCS.  */
 321   int bufs_len;
 322   /* The index in MBS, which is updated by re_string_fetch_byte.  */
 323   int cur_idx;
 324   /* length of RAW_MBS array.  */
 325   int raw_len;
 326   /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN.  */
 327   int len;
 328   /* End of the buffer may be shorter than its length in the cases such
 329      as re_match_2, re_search_2.  Then, we use STOP for end of the buffer
 330      instead of LEN.  */
 331   int raw_stop;
 332   /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS.  */
 333   int stop;
 334
 335   /* The context of mbs[0].  We store the context independently, since
 336      the context of mbs[0] may be different from raw_mbs[0], which is
 337      the beginning of the input string.  */
 338   unsigned int tip_context;
 339   /* The translation passed as a part of an argument of re_compile_pattern.  */
 340   unsigned RE_TRANSLATE_TYPE trans;
 341   /* Copy of re_dfa_t's word_char.  */
 342   re_const_bitset_ptr_t word_char;
 343   /* 1 if REG_ICASE.  */
 344   unsigned char icase;
 345   unsigned char is_utf8;
 346   unsigned char map_notascii;
 347   unsigned char mbs_allocated;
 348   unsigned char offsets_needed;
 349   unsigned char newline_anchor;
 350   unsigned char word_ops_used;
 351   int mb_cur_max;
 352 };
 353 typedef struct re_string_t re_string_t;
 354
 355
 356 struct re_dfa_t;
 357 typedef struct re_dfa_t re_dfa_t;
 358
 359 #ifndef _LIBC
 360 # ifdef __i386__
 361 #  define internal_function   __attribute ((regparm (3), stdcall))
 362 # else
 363 #  define internal_function
 364 # endif
 365 #endif
 366
 367 #ifndef RE_NO_INTERNAL_PROTOTYPES
 368 static reg_errcode_t re_string_allocate (re_string_t *pstr, const char *str,
 369                                          int len, int init_len,
 370                                          RE_TRANSLATE_TYPE trans, int icase,
 371                                          const re_dfa_t *dfa)
 372      internal_function;
 373 static reg_errcode_t re_string_construct (re_string_t *pstr, const char *str,
 374                                           int len, RE_TRANSLATE_TYPE trans,
 375                                           int icase, const re_dfa_t *dfa)
 376      internal_function;
 377 static reg_errcode_t re_string_reconstruct (re_string_t *pstr, int idx,
 378                                             int eflags) internal_function;
 379 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
 380                                                 int new_buf_len)
 381      internal_function;
 382 # ifdef RE_ENABLE_I18N
 383 static void build_wcs_buffer (re_string_t *pstr) internal_function;
 384 static int build_wcs_upper_buffer (re_string_t *pstr) internal_function;
 385 # endif /* RE_ENABLE_I18N */
 386 static void build_upper_buffer (re_string_t *pstr) internal_function;
 387 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
 388 static void re_string_destruct (re_string_t *pstr) internal_function;
 389 # ifdef RE_ENABLE_I18N
 390 static int re_string_elem_size_at (const re_string_t *pstr, int idx)
 391      internal_function;
 392 static inline int re_string_char_size_at (const re_string_t *pstr, int idx)
 393      internal_function;
 394 static inline wint_t re_string_wchar_at (const re_string_t *pstr, int idx)
 395      internal_function;
 396 # endif /* RE_ENABLE_I18N */
 397 static unsigned int re_string_context_at (const re_string_t *input, int idx,
 398                                           int eflags) internal_function;
 399 static unsigned char re_string_peek_byte_case (const re_string_t *pstr,
 400                                                int idx) internal_function;
 401 static unsigned char re_string_fetch_byte_case (re_string_t *pstr)
 402      internal_function;
 403 #endif
 404 #define re_string_peek_byte(pstr, offset) \
 405   ((pstr)->mbs[(pstr)->cur_idx + offset])
 406 #define re_string_fetch_byte(pstr) \
 407   ((pstr)->mbs[(pstr)->cur_idx++])
 408 #define re_string_first_byte(pstr, idx) \
 409   ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
 410 #define re_string_is_single_byte_char(pstr, idx) \
 411   ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
 412                                 || (pstr)->wcs[(idx) + 1] != WEOF))
 413 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
 414 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
 415 #define re_string_get_buffer(pstr) ((pstr)->mbs)
 416 #define re_string_length(pstr) ((pstr)->len)
 417 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
 418 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
 419 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
 420
 421 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
 422 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
 423 #define re_free(p) free (p)
 424
 425 struct bin_tree_t
 426 {
 427   struct bin_tree_t *parent;
 428   struct bin_tree_t *left;
 429   struct bin_tree_t *right;
 430
 431   /* `node_idx' is the index in dfa->nodes, if `type' == 0.
 432      Otherwise `type' indicate the type of this node.  */
 433   re_token_type_t type;
 434   int node_idx;
 435
 436   int first;
 437   int next;
 438   re_node_set eclosure;
 439 };
 440 typedef struct bin_tree_t bin_tree_t;
 441
 442 #define BIN_TREE_STORAGE_SIZE \
 443   ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
 444
 445 struct bin_tree_storage_t
 446 {
 447   struct bin_tree_storage_t *next;
 448   bin_tree_t data[BIN_TREE_STORAGE_SIZE];
 449 };
 450 typedef struct bin_tree_storage_t bin_tree_storage_t;
 451
 452 #define CONTEXT_WORD 1
 453 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
 454 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
 455 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
 456
 457 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
 458 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
 459 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
 460 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
 461 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
 462
 463 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
 464 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
 465 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
 466 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
 467
 468 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
 469  ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 470   || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 471   || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
 472   || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
 473
 474 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
 475  ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 476   || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 477   || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
 478   || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
 479
 480 struct re_dfastate_t
 481 {
 482   unsigned int hash;
 483   re_node_set nodes;
 484   re_node_set *entrance_nodes;
 485   struct re_dfastate_t **trtable;
 486   unsigned int context : 4;
 487   unsigned int halt : 1;
 488   /* If this state can accept `multi byte'.
 489      Note that we refer to multibyte characters, and multi character
 490      collating elements as `multi byte'.  */
 491   unsigned int accept_mb : 1;
 492   /* If this state has backreference node(s).  */
 493   unsigned int has_backref : 1;
 494   unsigned int has_constraint : 1;
 495   unsigned int word_trtable : 1;
 496 };
 497 typedef struct re_dfastate_t re_dfastate_t;
 498
 499 typedef struct
 500 {
 501   /* start <= node < end  */
 502   int start;
 503   int end;
 504 } re_subexp_t;
 505
 506 struct re_state_table_entry
 507 {
 508   int num;
 509   int alloc;
 510   re_dfastate_t **array;
 511 };
 512
 513 /* Array type used in re_sub_match_last_t and re_sub_match_top_t.  */
 514
 515 typedef struct
 516 {
 517   int next_idx;
 518   int alloc;
 519   re_dfastate_t **array;
 520 } state_array_t;
 521
 522 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP.  */
 523
 524 typedef struct
 525 {
 526   int node;
 527   int str_idx; /* The position NODE match at.  */
 528   state_array_t path;
 529 } re_sub_match_last_t;
 530
 531 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
 532    And information about the node, whose type is OP_CLOSE_SUBEXP,
 533    corresponding to NODE is stored in LASTS.  */
 534
 535 typedef struct
 536 {
 537   int str_idx;
 538   int node;
 539   int next_last_offset;
 540   state_array_t *path;
 541   int alasts; /* Allocation size of LASTS.  */
 542   int nlasts; /* The number of LASTS.  */
 543   re_sub_match_last_t **lasts;
 544 } re_sub_match_top_t;
 545
 546 struct re_backref_cache_entry
 547 {
 548   int node;
 549   int str_idx;
 550   int subexp_from;
 551   int subexp_to;
 552   char more;
 553   char unused;
 554   unsigned short int eps_reachable_subexps_map;
 555 };
 556
 557 typedef struct
 558 {
 559   /* The string object corresponding to the input string.  */
 560   re_string_t input;
 561 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
 562   re_dfa_t *const dfa;
 563 #else
 564   re_dfa_t *dfa;
 565 #endif
 566   /* EFLAGS of the argument of regexec.  */
 567   int eflags;
 568   /* Where the matching ends.  */
 569   int match_last;
 570   int last_node;
 571   /* The state log used by the matcher.  */
 572   re_dfastate_t **state_log;
 573   int state_log_top;
 574   /* Back reference cache.  */
 575   int nbkref_ents;
 576   int abkref_ents;
 577   struct re_backref_cache_entry *bkref_ents;
 578   int max_mb_elem_len;
 579   int nsub_tops;
 580   int asub_tops;
 581   re_sub_match_top_t **sub_tops;
 582 } re_match_context_t;
 583
 584 typedef struct
 585 {
 586   re_dfastate_t **sifted_states;
 587   re_dfastate_t **limited_states;
 588   int last_node;
 589   int last_str_idx;
 590   re_node_set limits;
 591 } re_sift_context_t;
 592
 593 struct re_fail_stack_ent_t
 594 {
 595   int idx;
 596   int node;
 597   regmatch_t *regs;
 598   re_node_set eps_via_nodes;
 599 };
 600
 601 struct re_fail_stack_t
 602 {
 603   int num;
 604   int alloc;
 605   struct re_fail_stack_ent_t *stack;
 606 };
 607
 608 struct re_dfa_t
 609 {
 610   re_subexp_t *subexps;
 611   re_token_t *nodes;
 612   int nodes_alloc;
 613   int nodes_len;
 614   int *nexts;
 615   int *org_indices;
 616   re_node_set *edests;
 617   re_node_set *eclosures;
 618   re_node_set *inveclosures;
 619   struct re_state_table_entry *state_table;
 620   re_dfastate_t *init_state;
 621   re_dfastate_t *init_state_word;
 622   re_dfastate_t *init_state_nl;
 623   re_dfastate_t *init_state_begbuf;
 624   bin_tree_t *str_tree;
 625   bin_tree_storage_t *str_tree_storage;
 626   re_bitset_ptr_t sb_char;
 627   int str_tree_storage_idx;
 628
 629   /* number of subexpressions `re_nsub' is in regex_t.  */
 630   int subexps_alloc;
 631   unsigned int state_hash_mask;
 632   int states_alloc;
 633   int init_node;
 634   int nbackref; /* The number of backreference in this dfa.  */
 635   /* Bitmap expressing which backreference is used.  */
 636   unsigned int used_bkref_map;
 637   unsigned int has_plural_match : 1;
 638   /* If this dfa has "multibyte node", which is a backreference or
 639      a node which can accept multibyte character or multi character
 640      collating element.  */
 641   unsigned int has_mb_node : 1;
 642   unsigned int is_utf8 : 1;
 643   unsigned int map_notascii : 1;
 644   unsigned int word_ops_used : 1;
 645   int mb_cur_max;
 646   bitset word_char;
 647   reg_syntax_t syntax;
 648   int *subexp_map;
 649 #ifdef DEBUG
 650   char* re_str;
 651 #endif
 652 };
 653
 654 #ifndef RE_NO_INTERNAL_PROTOTYPES
 655 static reg_errcode_t re_node_set_alloc (re_node_set *set, int size) internal_function;
 656 static reg_errcode_t re_node_set_init_1 (re_node_set *set, int elem) internal_function;
 657 static reg_errcode_t re_node_set_init_2 (re_node_set *set, int elem1,
 658                                          int elem2) internal_function;
 659 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
 660 static reg_errcode_t re_node_set_init_copy (re_node_set *dest,
 661                                             const re_node_set *src) internal_function;
 662 static reg_errcode_t re_node_set_add_intersect (re_node_set *dest,
 663                                                 const re_node_set *src1,
 664                                                 const re_node_set *src2) internal_function;
 665 static reg_errcode_t re_node_set_init_union (re_node_set *dest,
 666                                              const re_node_set *src1,
 667                                              const re_node_set *src2) internal_function;
 668 static reg_errcode_t re_node_set_merge (re_node_set *dest,
 669                                         const re_node_set *src) internal_function;
 670 static int re_node_set_insert (re_node_set *set, int elem) internal_function;
 671 static int re_node_set_insert_last (re_node_set *set,
 672                                     int elem) internal_function;
 673 static int re_node_set_compare (const re_node_set *set1,
 674                                 const re_node_set *set2) internal_function;
 675 static int re_node_set_contains (const re_node_set *set, int elem) internal_function;
 676 static void re_node_set_remove_at (re_node_set *set, int idx) internal_function;
 677 #define re_node_set_remove(set,id) \
 678   (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
 679 #define re_node_set_empty(p) ((p)->nelem = 0)
 680 #define re_node_set_free(set) re_free ((set)->elems)
 681 static int re_dfa_add_node (re_dfa_t *dfa, re_token_t token, int mode) internal_function;
 682 static re_dfastate_t *re_acquire_state (reg_errcode_t *err, re_dfa_t *dfa,
 683                                         const re_node_set *nodes) internal_function;
 684 static re_dfastate_t *re_acquire_state_context (reg_errcode_t *err,
 685                                                 re_dfa_t *dfa,
 686                                                 const re_node_set *nodes,
 687                                                 unsigned int context) internal_function;
 688 static void free_state (re_dfastate_t *state) internal_function;
 689 #endif
 690 \f
 691
 692 typedef enum
 693 {
 694   SB_CHAR,
 695   MB_CHAR,
 696   EQUIV_CLASS,
 697   COLL_SYM,
 698   CHAR_CLASS
 699 } bracket_elem_type;
 700
 701 typedef struct
 702 {
 703   bracket_elem_type type;
 704   union
 705   {
 706     unsigned char ch;
 707     unsigned char *name;
 708     wchar_t wch;
 709   } opr;
 710 } bracket_elem_t;
 711
 712
 713 /* Inline functions for bitset operation.  */
 714 static inline void
 715 bitset_not (bitset set)
 716 {
 717   int bitset_i;
 718   for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
 719     set[bitset_i] = ~set[bitset_i];
 720 }
 721
 722 static inline void
 723 bitset_merge (bitset dest, const bitset src)
 724 {
 725   int bitset_i;
 726   for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
 727     dest[bitset_i] |= src[bitset_i];
 728 }
 729
 730 static inline void
 731 bitset_not_merge (bitset dest, const bitset src)
 732 {
 733   int i;
 734   for (i = 0; i < BITSET_UINTS; ++i)
 735     dest[i] |= ~src[i];
 736 }
 737
 738 static inline void
 739 bitset_mask (bitset dest, const bitset src)
 740 {
 741   int bitset_i;
 742   for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
 743     dest[bitset_i] &= src[bitset_i];
 744 }
 745
 746 #if defined RE_ENABLE_I18N && !defined RE_NO_INTERNAL_PROTOTYPES
 747 /* Inline functions for re_string.  */
 748 static inline int
 749 internal_function
 750 re_string_char_size_at (const re_string_t *pstr, int idx)
 751 {
 752   int byte_idx;
 753   if (pstr->mb_cur_max == 1)
 754     return 1;
 755   for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
 756     if (pstr->wcs[idx + byte_idx] != WEOF)
 757       break;
 758   return byte_idx;
 759 }
 760
 761 static inline wint_t
 762 internal_function
 763 re_string_wchar_at (const re_string_t *pstr, int idx)
 764 {
 765   if (pstr->mb_cur_max == 1)
 766     return (wint_t) pstr->mbs[idx];
 767   return (wint_t) pstr->wcs[idx];
 768 }
 769
 770 static int
 771 internal_function
 772 re_string_elem_size_at (const re_string_t *pstr, int idx)
 773 {
 774 #ifdef _LIBC
 775   const unsigned char *p, *extra;
 776   const int32_t *table, *indirect;
 777   int32_t tmp;
 778 # include <locale/weight.h>
 779   uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
 780
 781   if (nrules != 0)
 782     {
 783       table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
 784       extra = (const unsigned char *)
 785         _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
 786       indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
 787                                                 _NL_COLLATE_INDIRECTMB);
 788       p = pstr->mbs + idx;
 789       tmp = findidx (&p);
 790       return p - pstr->mbs - idx;
 791     }
 792   else
 793 #endif /* _LIBC */
 794     return 1;
 795 }
 796 #endif /* RE_ENABLE_I18N */
 797
 798 #endif /*  _REGEX_INTERNAL_H */