lib/regex_internal.h

   1 /* Extended regular expression matching and search library.
   2    Copyright (C) 2002-2018 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    <https://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 #include <langinfo.h>
  30 #include <locale.h>
  31 #include <wchar.h>
  32 #include <wctype.h>
  33 #include <stdbool.h>
  34 #include <stdint.h>
  35
  36 /* Properties of integers.  Although Gnulib has intprops.h, glibc does
  37    without for now.  */
  38 #ifndef _LIBC
  39 # include "intprops.h"
  40 #else
  41 /* True if the real type T is signed.  */
  42 # define TYPE_SIGNED(t) (! ((t) 0 < (t) -1))
  43
  44 /* True if adding the nonnegative Idx values A and B would overflow.
  45    If false, set *R to A + B.  A, B, and R may be evaluated more than
  46    once, or zero times.  Although this is not a full implementation of
  47    Gnulib INT_ADD_WRAPV, it is good enough for glibc regex code.
  48    FIXME: This implementation is a fragile stopgap, and this file would
  49    be simpler and more robust if intprops.h were migrated into glibc.  */
  50 # define INT_ADD_WRAPV(a, b, r) \
  51    (IDX_MAX - (a) < (b) ? true : (*(r) = (a) + (b), false))
  52 #endif
  53
  54 #ifdef _LIBC
  55 # include <libc-lock.h>
  56 # define lock_define(name) __libc_lock_define (, name)
  57 # define lock_init(lock) (__libc_lock_init (lock), 0)
  58 # define lock_fini(lock) ((void) 0)
  59 # define lock_lock(lock) __libc_lock_lock (lock)
  60 # define lock_unlock(lock) __libc_lock_unlock (lock)
  61 #elif defined GNULIB_LOCK && !defined USE_UNLOCKED_IO
  62 # include "glthread/lock.h"
  63   /* Use gl_lock_define if empty macro arguments are known to work.
  64      Otherwise, fall back on less-portable substitutes.  */
  65 # if ((defined __GNUC__ && !defined __STRICT_ANSI__) \
  66       || (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__))
  67 #  define lock_define(name) gl_lock_define (, name)
  68 # elif USE_POSIX_THREADS
  69 #  define lock_define(name) pthread_mutex_t name;
  70 # elif USE_PTH_THREADS
  71 #  define lock_define(name) pth_mutex_t name;
  72 # elif USE_SOLARIS_THREADS
  73 #  define lock_define(name) mutex_t name;
  74 # elif USE_WINDOWS_THREADS
  75 #  define lock_define(name) gl_lock_t name;
  76 # else
  77 #  define lock_define(name)
  78 # endif
  79 # define lock_init(lock) glthread_lock_init (&(lock))
  80 # define lock_fini(lock) glthread_lock_destroy (&(lock))
  81 # define lock_lock(lock) glthread_lock_lock (&(lock))
  82 # define lock_unlock(lock) glthread_lock_unlock (&(lock))
  83 #elif defined GNULIB_PTHREAD && !defined USE_UNLOCKED_IO
  84 # include <pthread.h>
  85 # define lock_define(name) pthread_mutex_t name;
  86 # define lock_init(lock) pthread_mutex_init (&(lock), 0)
  87 # define lock_fini(lock) pthread_mutex_destroy (&(lock))
  88 # define lock_lock(lock) pthread_mutex_lock (&(lock))
  89 # define lock_unlock(lock) pthread_mutex_unlock (&(lock))
  90 #else
  91 # define lock_define(name)
  92 # define lock_init(lock) 0
  93 # define lock_fini(lock) ((void) 0)
  94   /* The 'dfa' avoids an "unused variable 'dfa'" warning from GCC.  */
  95 # define lock_lock(lock) ((void) dfa)
  96 # define lock_unlock(lock) ((void) 0)
  97 #endif
  98
  99 /* In case that the system doesn't have isblank().  */
 100 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK))
 101 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
 102 #endif
 103
 104 #ifdef _LIBC
 105 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
 106 #  define _RE_DEFINE_LOCALE_FUNCTIONS 1
 107 #   include <locale/localeinfo.h>
 108 #   include <locale/coll-lookup.h>
 109 # endif
 110 #endif
 111
 112 /* This is for other GNU distributions with internationalized messages.  */
 113 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
 114 # include <libintl.h>
 115 # ifdef _LIBC
 116 #  undef gettext
 117 #  define gettext(msgid) \
 118   __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES)
 119 # endif
 120 #else
 121 # undef gettext
 122 # define gettext(msgid) (msgid)
 123 #endif
 124
 125 #ifndef gettext_noop
 126 /* This define is so xgettext can find the internationalizable
 127    strings.  */
 128 # define gettext_noop(String) String
 129 #endif
 130
 131 #if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE) || _LIBC
 132 # define RE_ENABLE_I18N
 133 #endif
 134
 135 #define BE(expr, val) __builtin_expect (expr, val)
 136
 137 /* Number of ASCII characters.  */
 138 #define ASCII_CHARS 0x80
 139
 140 /* Number of single byte characters.  */
 141 #define SBC_MAX (UCHAR_MAX + 1)
 142
 143 #define COLL_ELEM_LEN_MAX 8
 144
 145 /* The character which represents newline.  */
 146 #define NEWLINE_CHAR '\n'
 147 #define WIDE_NEWLINE_CHAR L'\n'
 148
 149 /* Rename to standard API for using out of glibc.  */
 150 #ifndef _LIBC
 151 # undef __wctype
 152 # undef __iswalnum
 153 # undef __iswctype
 154 # undef __towlower
 155 # undef __towupper
 156 # define __wctype wctype
 157 # define __iswalnum iswalnum
 158 # define __iswctype iswctype
 159 # define __towlower towlower
 160 # define __towupper towupper
 161 # define __btowc btowc
 162 # define __mbrtowc mbrtowc
 163 # define __wcrtomb wcrtomb
 164 # define __regfree regfree
 165 # define attribute_hidden
 166 #endif /* not _LIBC */
 167
 168 #if __GNUC__ < 3 + (__GNUC_MINOR__ < 1)
 169 # define __attribute__(arg)
 170 #endif
 171
 172 #ifndef SSIZE_MAX
 173 # define SSIZE_MAX ((ssize_t) (SIZE_MAX / 2))
 174 #endif
 175
 176 /* The type of indexes into strings.  This is signed, not size_t,
 177    since the API requires indexes to fit in regoff_t anyway, and using
 178    signed integers makes the code a bit smaller and presumably faster.
 179    The traditional GNU regex implementation uses int for indexes.
 180    The POSIX-compatible implementation uses a possibly-wider type.
 181    The name 'Idx' is three letters to minimize the hassle of
 182    reindenting a lot of regex code that formerly used 'int'.  */
 183 typedef regoff_t Idx;
 184 #ifdef _REGEX_LARGE_OFFSETS
 185 # define IDX_MAX SSIZE_MAX
 186 #else
 187 # define IDX_MAX INT_MAX
 188 #endif
 189
 190 /* A hash value, suitable for computing hash tables.  */
 191 typedef __re_size_t re_hashval_t;
 192
 193 /* An integer used to represent a set of bits.  It must be unsigned,
 194    and must be at least as wide as unsigned int.  */
 195 typedef unsigned long int bitset_word_t;
 196 /* All bits set in a bitset_word_t.  */
 197 #define BITSET_WORD_MAX ULONG_MAX
 198
 199 /* Number of bits in a bitset_word_t.  For portability to hosts with
 200    padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)';
 201    instead, deduce it directly from BITSET_WORD_MAX.  Avoid
 202    greater-than-32-bit integers and unconditional shifts by more than
 203    31 bits, as they're not portable.  */
 204 #if BITSET_WORD_MAX == 0xffffffffUL
 205 # define BITSET_WORD_BITS 32
 206 #elif BITSET_WORD_MAX >> 31 >> 4 == 1
 207 # define BITSET_WORD_BITS 36
 208 #elif BITSET_WORD_MAX >> 31 >> 16 == 1
 209 # define BITSET_WORD_BITS 48
 210 #elif BITSET_WORD_MAX >> 31 >> 28 == 1
 211 # define BITSET_WORD_BITS 60
 212 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1
 213 # define BITSET_WORD_BITS 64
 214 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1
 215 # define BITSET_WORD_BITS 72
 216 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1
 217 # define BITSET_WORD_BITS 128
 218 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1
 219 # define BITSET_WORD_BITS 256
 220 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1
 221 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */
 222 # if BITSET_WORD_BITS <= SBC_MAX
 223 #  error "Invalid SBC_MAX"
 224 # endif
 225 #else
 226 # error "Add case for new bitset_word_t size"
 227 #endif
 228
 229 /* Number of bitset_word_t values in a bitset_t.  */
 230 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS)
 231
 232 typedef bitset_word_t bitset_t[BITSET_WORDS];
 233 typedef bitset_word_t *re_bitset_ptr_t;
 234 typedef const bitset_word_t *re_const_bitset_ptr_t;
 235
 236 #define PREV_WORD_CONSTRAINT 0x0001
 237 #define PREV_NOTWORD_CONSTRAINT 0x0002
 238 #define NEXT_WORD_CONSTRAINT 0x0004
 239 #define NEXT_NOTWORD_CONSTRAINT 0x0008
 240 #define PREV_NEWLINE_CONSTRAINT 0x0010
 241 #define NEXT_NEWLINE_CONSTRAINT 0x0020
 242 #define PREV_BEGBUF_CONSTRAINT 0x0040
 243 #define NEXT_ENDBUF_CONSTRAINT 0x0080
 244 #define WORD_DELIM_CONSTRAINT 0x0100
 245 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
 246
 247 typedef enum
 248 {
 249   INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 250   WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 251   WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 252   INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 253   LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
 254   LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
 255   BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
 256   BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
 257   WORD_DELIM = WORD_DELIM_CONSTRAINT,
 258   NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
 259 } re_context_type;
 260
 261 typedef struct
 262 {
 263   Idx alloc;
 264   Idx nelem;
 265   Idx *elems;
 266 } re_node_set;
 267
 268 typedef enum
 269 {
 270   NON_TYPE = 0,
 271
 272   /* Node type, These are used by token, node, tree.  */
 273   CHARACTER = 1,
 274   END_OF_RE = 2,
 275   SIMPLE_BRACKET = 3,
 276   OP_BACK_REF = 4,
 277   OP_PERIOD = 5,
 278 #ifdef RE_ENABLE_I18N
 279   COMPLEX_BRACKET = 6,
 280   OP_UTF8_PERIOD = 7,
 281 #endif /* RE_ENABLE_I18N */
 282
 283   /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
 284      when the debugger shows values of this enum type.  */
 285 #define EPSILON_BIT 8
 286   OP_OPEN_SUBEXP = EPSILON_BIT | 0,
 287   OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
 288   OP_ALT = EPSILON_BIT | 2,
 289   OP_DUP_ASTERISK = EPSILON_BIT | 3,
 290   ANCHOR = EPSILON_BIT | 4,
 291
 292   /* Tree type, these are used only by tree. */
 293   CONCAT = 16,
 294   SUBEXP = 17,
 295
 296   /* Token type, these are used only by token.  */
 297   OP_DUP_PLUS = 18,
 298   OP_DUP_QUESTION,
 299   OP_OPEN_BRACKET,
 300   OP_CLOSE_BRACKET,
 301   OP_CHARSET_RANGE,
 302   OP_OPEN_DUP_NUM,
 303   OP_CLOSE_DUP_NUM,
 304   OP_NON_MATCH_LIST,
 305   OP_OPEN_COLL_ELEM,
 306   OP_CLOSE_COLL_ELEM,
 307   OP_OPEN_EQUIV_CLASS,
 308   OP_CLOSE_EQUIV_CLASS,
 309   OP_OPEN_CHAR_CLASS,
 310   OP_CLOSE_CHAR_CLASS,
 311   OP_WORD,
 312   OP_NOTWORD,
 313   OP_SPACE,
 314   OP_NOTSPACE,
 315   BACK_SLASH
 316
 317 } re_token_type_t;
 318
 319 #ifdef RE_ENABLE_I18N
 320 typedef struct
 321 {
 322   /* Multibyte characters.  */
 323   wchar_t *mbchars;
 324
 325   /* Collating symbols.  */
 326 # ifdef _LIBC
 327   int32_t *coll_syms;
 328 # endif
 329
 330   /* Equivalence classes. */
 331 # ifdef _LIBC
 332   int32_t *equiv_classes;
 333 # endif
 334
 335   /* Range expressions. */
 336 # ifdef _LIBC
 337   uint32_t *range_starts;
 338   uint32_t *range_ends;
 339 # else /* not _LIBC */
 340   wchar_t *range_starts;
 341   wchar_t *range_ends;
 342 # endif /* not _LIBC */
 343
 344   /* Character classes. */
 345   wctype_t *char_classes;
 346
 347   /* If this character set is the non-matching list.  */
 348   unsigned int non_match : 1;
 349
 350   /* # of multibyte characters.  */
 351   Idx nmbchars;
 352
 353   /* # of collating symbols.  */
 354   Idx ncoll_syms;
 355
 356   /* # of equivalence classes. */
 357   Idx nequiv_classes;
 358
 359   /* # of range expressions. */
 360   Idx nranges;
 361
 362   /* # of character classes. */
 363   Idx nchar_classes;
 364 } re_charset_t;
 365 #endif /* RE_ENABLE_I18N */
 366
 367 typedef struct
 368 {
 369   union
 370   {
 371     unsigned char c;            /* for CHARACTER */
 372     re_bitset_ptr_t sbcset;     /* for SIMPLE_BRACKET */
 373 #ifdef RE_ENABLE_I18N
 374     re_charset_t *mbcset;       /* for COMPLEX_BRACKET */
 375 #endif /* RE_ENABLE_I18N */
 376     Idx idx;                    /* for BACK_REF */
 377     re_context_type ctx_type;   /* for ANCHOR */
 378   } opr;
 379 #if __GNUC__ >= 2 && !defined __STRICT_ANSI__
 380   re_token_type_t type : 8;
 381 #else
 382   re_token_type_t type;
 383 #endif
 384   unsigned int constraint : 10; /* context constraint */
 385   unsigned int duplicated : 1;
 386   unsigned int opt_subexp : 1;
 387 #ifdef RE_ENABLE_I18N
 388   unsigned int accept_mb : 1;
 389   /* These 2 bits can be moved into the union if needed (e.g. if running out
 390      of bits; move opr.c to opr.c.c and move the flags to opr.c.flags).  */
 391   unsigned int mb_partial : 1;
 392 #endif
 393   unsigned int word_char : 1;
 394 } re_token_t;
 395
 396 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
 397
 398 struct re_string_t
 399 {
 400   /* Indicate the raw buffer which is the original string passed as an
 401      argument of regexec(), re_search(), etc..  */
 402   const unsigned char *raw_mbs;
 403   /* Store the multibyte string.  In case of "case insensitive mode" like
 404      REG_ICASE, upper cases of the string are stored, otherwise MBS points
 405      the same address that RAW_MBS points.  */
 406   unsigned char *mbs;
 407 #ifdef RE_ENABLE_I18N
 408   /* Store the wide character string which is corresponding to MBS.  */
 409   wint_t *wcs;
 410   Idx *offsets;
 411   mbstate_t cur_state;
 412 #endif
 413   /* Index in RAW_MBS.  Each character mbs[i] corresponds to
 414      raw_mbs[raw_mbs_idx + i].  */
 415   Idx raw_mbs_idx;
 416   /* The length of the valid characters in the buffers.  */
 417   Idx valid_len;
 418   /* The corresponding number of bytes in raw_mbs array.  */
 419   Idx valid_raw_len;
 420   /* The length of the buffers MBS and WCS.  */
 421   Idx bufs_len;
 422   /* The index in MBS, which is updated by re_string_fetch_byte.  */
 423   Idx cur_idx;
 424   /* length of RAW_MBS array.  */
 425   Idx raw_len;
 426   /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN.  */
 427   Idx len;
 428   /* End of the buffer may be shorter than its length in the cases such
 429      as re_match_2, re_search_2.  Then, we use STOP for end of the buffer
 430      instead of LEN.  */
 431   Idx raw_stop;
 432   /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS.  */
 433   Idx stop;
 434
 435   /* The context of mbs[0].  We store the context independently, since
 436      the context of mbs[0] may be different from raw_mbs[0], which is
 437      the beginning of the input string.  */
 438   unsigned int tip_context;
 439   /* The translation passed as a part of an argument of re_compile_pattern.  */
 440   RE_TRANSLATE_TYPE trans;
 441   /* Copy of re_dfa_t's word_char.  */
 442   re_const_bitset_ptr_t word_char;
 443   /* true if REG_ICASE.  */
 444   unsigned char icase;
 445   unsigned char is_utf8;
 446   unsigned char map_notascii;
 447   unsigned char mbs_allocated;
 448   unsigned char offsets_needed;
 449   unsigned char newline_anchor;
 450   unsigned char word_ops_used;
 451   int mb_cur_max;
 452 };
 453 typedef struct re_string_t re_string_t;
 454
 455
 456 struct re_dfa_t;
 457 typedef struct re_dfa_t re_dfa_t;
 458
 459 #ifndef _LIBC
 460 # define IS_IN(libc) false
 461 #endif
 462
 463 #define re_string_peek_byte(pstr, offset) \
 464   ((pstr)->mbs[(pstr)->cur_idx + offset])
 465 #define re_string_fetch_byte(pstr) \
 466   ((pstr)->mbs[(pstr)->cur_idx++])
 467 #define re_string_first_byte(pstr, idx) \
 468   ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
 469 #define re_string_is_single_byte_char(pstr, idx) \
 470   ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
 471                                 || (pstr)->wcs[(idx) + 1] != WEOF))
 472 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
 473 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
 474 #define re_string_get_buffer(pstr) ((pstr)->mbs)
 475 #define re_string_length(pstr) ((pstr)->len)
 476 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
 477 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
 478 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
 479
 480 #if defined _LIBC || HAVE_ALLOCA
 481 # include <alloca.h>
 482 #endif
 483
 484 #ifndef _LIBC
 485 # if HAVE_ALLOCA
 486 /* The OS usually guarantees only one guard page at the bottom of the stack,
 487    and a page size can be as small as 4096 bytes.  So we cannot safely
 488    allocate anything larger than 4096 bytes.  Also care for the possibility
 489    of a few compiler-allocated temporary stack slots.  */
 490 #  define __libc_use_alloca(n) ((n) < 4032)
 491 # else
 492 /* alloca is implemented with malloc, so just use malloc.  */
 493 #  define __libc_use_alloca(n) 0
 494 #  undef alloca
 495 #  define alloca(n) malloc (n)
 496 # endif
 497 #endif
 498
 499 #ifdef _LIBC
 500 # define MALLOC_0_IS_NONNULL 1
 501 #elif !defined MALLOC_0_IS_NONNULL
 502 # define MALLOC_0_IS_NONNULL 0
 503 #endif
 504
 505 #ifndef MAX
 506 # define MAX(a,b) ((a) < (b) ? (b) : (a))
 507 #endif
 508 #ifndef MIN
 509 # define MIN(a,b) ((a) < (b) ? (a) : (b))
 510 #endif
 511
 512 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
 513 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
 514 #define re_free(p) free (p)
 515
 516 struct bin_tree_t
 517 {
 518   struct bin_tree_t *parent;
 519   struct bin_tree_t *left;
 520   struct bin_tree_t *right;
 521   struct bin_tree_t *first;
 522   struct bin_tree_t *next;
 523
 524   re_token_t token;
 525
 526   /* 'node_idx' is the index in dfa->nodes, if 'type' == 0.
 527      Otherwise 'type' indicate the type of this node.  */
 528   Idx node_idx;
 529 };
 530 typedef struct bin_tree_t bin_tree_t;
 531
 532 #define BIN_TREE_STORAGE_SIZE \
 533   ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
 534
 535 struct bin_tree_storage_t
 536 {
 537   struct bin_tree_storage_t *next;
 538   bin_tree_t data[BIN_TREE_STORAGE_SIZE];
 539 };
 540 typedef struct bin_tree_storage_t bin_tree_storage_t;
 541
 542 #define CONTEXT_WORD 1
 543 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
 544 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
 545 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
 546
 547 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
 548 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
 549 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
 550 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
 551 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
 552
 553 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
 554 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
 555 #define IS_WIDE_WORD_CHAR(ch) (__iswalnum (ch) || (ch) == L'_')
 556 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
 557
 558 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
 559  ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 560   || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 561   || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
 562   || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
 563
 564 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
 565  ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 566   || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 567   || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
 568   || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
 569
 570 struct re_dfastate_t
 571 {
 572   re_hashval_t hash;
 573   re_node_set nodes;
 574   re_node_set non_eps_nodes;
 575   re_node_set inveclosure;
 576   re_node_set *entrance_nodes;
 577   struct re_dfastate_t **trtable, **word_trtable;
 578   unsigned int context : 4;
 579   unsigned int halt : 1;
 580   /* If this state can accept "multi byte".
 581      Note that we refer to multibyte characters, and multi character
 582      collating elements as "multi byte".  */
 583   unsigned int accept_mb : 1;
 584   /* If this state has backreference node(s).  */
 585   unsigned int has_backref : 1;
 586   unsigned int has_constraint : 1;
 587 };
 588 typedef struct re_dfastate_t re_dfastate_t;
 589
 590 struct re_state_table_entry
 591 {
 592   Idx num;
 593   Idx alloc;
 594   re_dfastate_t **array;
 595 };
 596
 597 /* Array type used in re_sub_match_last_t and re_sub_match_top_t.  */
 598
 599 typedef struct
 600 {
 601   Idx next_idx;
 602   Idx alloc;
 603   re_dfastate_t **array;
 604 } state_array_t;
 605
 606 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP.  */
 607
 608 typedef struct
 609 {
 610   Idx node;
 611   Idx str_idx; /* The position NODE match at.  */
 612   state_array_t path;
 613 } re_sub_match_last_t;
 614
 615 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
 616    And information about the node, whose type is OP_CLOSE_SUBEXP,
 617    corresponding to NODE is stored in LASTS.  */
 618
 619 typedef struct
 620 {
 621   Idx str_idx;
 622   Idx node;
 623   state_array_t *path;
 624   Idx alasts; /* Allocation size of LASTS.  */
 625   Idx nlasts; /* The number of LASTS.  */
 626   re_sub_match_last_t **lasts;
 627 } re_sub_match_top_t;
 628
 629 struct re_backref_cache_entry
 630 {
 631   Idx node;
 632   Idx str_idx;
 633   Idx subexp_from;
 634   Idx subexp_to;
 635   char more;
 636   char unused;
 637   unsigned short int eps_reachable_subexps_map;
 638 };
 639
 640 typedef struct
 641 {
 642   /* The string object corresponding to the input string.  */
 643   re_string_t input;
 644 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
 645   const re_dfa_t *const dfa;
 646 #else
 647   const re_dfa_t *dfa;
 648 #endif
 649   /* EFLAGS of the argument of regexec.  */
 650   int eflags;
 651   /* Where the matching ends.  */
 652   Idx match_last;
 653   Idx last_node;
 654   /* The state log used by the matcher.  */
 655   re_dfastate_t **state_log;
 656   Idx state_log_top;
 657   /* Back reference cache.  */
 658   Idx nbkref_ents;
 659   Idx abkref_ents;
 660   struct re_backref_cache_entry *bkref_ents;
 661   int max_mb_elem_len;
 662   Idx nsub_tops;
 663   Idx asub_tops;
 664   re_sub_match_top_t **sub_tops;
 665 } re_match_context_t;
 666
 667 typedef struct
 668 {
 669   re_dfastate_t **sifted_states;
 670   re_dfastate_t **limited_states;
 671   Idx last_node;
 672   Idx last_str_idx;
 673   re_node_set limits;
 674 } re_sift_context_t;
 675
 676 struct re_fail_stack_ent_t
 677 {
 678   Idx idx;
 679   Idx node;
 680   regmatch_t *regs;
 681   re_node_set eps_via_nodes;
 682 };
 683
 684 struct re_fail_stack_t
 685 {
 686   Idx num;
 687   Idx alloc;
 688   struct re_fail_stack_ent_t *stack;
 689 };
 690
 691 struct re_dfa_t
 692 {
 693   re_token_t *nodes;
 694   size_t nodes_alloc;
 695   size_t nodes_len;
 696   Idx *nexts;
 697   Idx *org_indices;
 698   re_node_set *edests;
 699   re_node_set *eclosures;
 700   re_node_set *inveclosures;
 701   struct re_state_table_entry *state_table;
 702   re_dfastate_t *init_state;
 703   re_dfastate_t *init_state_word;
 704   re_dfastate_t *init_state_nl;
 705   re_dfastate_t *init_state_begbuf;
 706   bin_tree_t *str_tree;
 707   bin_tree_storage_t *str_tree_storage;
 708   re_bitset_ptr_t sb_char;
 709   int str_tree_storage_idx;
 710
 711   /* number of subexpressions 're_nsub' is in regex_t.  */
 712   re_hashval_t state_hash_mask;
 713   Idx init_node;
 714   Idx nbackref; /* The number of backreference in this dfa.  */
 715
 716   /* Bitmap expressing which backreference is used.  */
 717   bitset_word_t used_bkref_map;
 718   bitset_word_t completed_bkref_map;
 719
 720   unsigned int has_plural_match : 1;
 721   /* If this dfa has "multibyte node", which is a backreference or
 722      a node which can accept multibyte character or multi character
 723      collating element.  */
 724   unsigned int has_mb_node : 1;
 725   unsigned int is_utf8 : 1;
 726   unsigned int map_notascii : 1;
 727   unsigned int word_ops_used : 1;
 728   int mb_cur_max;
 729   bitset_t word_char;
 730   reg_syntax_t syntax;
 731   Idx *subexp_map;
 732 #ifdef DEBUG
 733   char* re_str;
 734 #endif
 735   lock_define (lock)
 736 };
 737
 738 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
 739 #define re_node_set_remove(set,id) \
 740   (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
 741 #define re_node_set_empty(p) ((p)->nelem = 0)
 742 #define re_node_set_free(set) re_free ((set)->elems)
 743 \f
 744
 745 typedef enum
 746 {
 747   SB_CHAR,
 748   MB_CHAR,
 749   EQUIV_CLASS,
 750   COLL_SYM,
 751   CHAR_CLASS
 752 } bracket_elem_type;
 753
 754 typedef struct
 755 {
 756   bracket_elem_type type;
 757   union
 758   {
 759     unsigned char ch;
 760     unsigned char *name;
 761     wchar_t wch;
 762   } opr;
 763 } bracket_elem_t;
 764
 765
 766 /* Functions for bitset_t operation.  */
 767
 768 static inline void
 769 bitset_set (bitset_t set, Idx i)
 770 {
 771   set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS;
 772 }
 773
 774 static inline void
 775 bitset_clear (bitset_t set, Idx i)
 776 {
 777   set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS);
 778 }
 779
 780 static inline bool
 781 bitset_contain (const bitset_t set, Idx i)
 782 {
 783   return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1;
 784 }
 785
 786 static inline void
 787 bitset_empty (bitset_t set)
 788 {
 789   memset (set, '\0', sizeof (bitset_t));
 790 }
 791
 792 static inline void
 793 bitset_set_all (bitset_t set)
 794 {
 795   memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS));
 796   if (SBC_MAX % BITSET_WORD_BITS != 0)
 797     set[BITSET_WORDS - 1] =
 798       ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1;
 799 }
 800
 801 static inline void
 802 bitset_copy (bitset_t dest, const bitset_t src)
 803 {
 804   memcpy (dest, src, sizeof (bitset_t));
 805 }
 806
 807 static inline void
 808 bitset_not (bitset_t set)
 809 {
 810   int bitset_i;
 811   for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i)
 812     set[bitset_i] = ~set[bitset_i];
 813   if (SBC_MAX % BITSET_WORD_BITS != 0)
 814     set[BITSET_WORDS - 1] =
 815       ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1)
 816        & ~set[BITSET_WORDS - 1]);
 817 }
 818
 819 static inline void
 820 bitset_merge (bitset_t dest, const bitset_t src)
 821 {
 822   int bitset_i;
 823   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 824     dest[bitset_i] |= src[bitset_i];
 825 }
 826
 827 static inline void
 828 bitset_mask (bitset_t dest, const bitset_t src)
 829 {
 830   int bitset_i;
 831   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 832     dest[bitset_i] &= src[bitset_i];
 833 }
 834
 835 #ifdef RE_ENABLE_I18N
 836 /* Functions for re_string.  */
 837 static int
 838 __attribute__ ((pure, unused))
 839 re_string_char_size_at (const re_string_t *pstr, Idx idx)
 840 {
 841   int byte_idx;
 842   if (pstr->mb_cur_max == 1)
 843     return 1;
 844   for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
 845     if (pstr->wcs[idx + byte_idx] != WEOF)
 846       break;
 847   return byte_idx;
 848 }
 849
 850 static wint_t
 851 __attribute__ ((pure, unused))
 852 re_string_wchar_at (const re_string_t *pstr, Idx idx)
 853 {
 854   if (pstr->mb_cur_max == 1)
 855     return (wint_t) pstr->mbs[idx];
 856   return (wint_t) pstr->wcs[idx];
 857 }
 858
 859 # ifdef _LIBC
 860 #  include <locale/weight.h>
 861 # endif
 862
 863 static int
 864 __attribute__ ((pure, unused))
 865 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
 866 {
 867 # ifdef _LIBC
 868   const unsigned char *p, *extra;
 869   const int32_t *table, *indirect;
 870   uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
 871
 872   if (nrules != 0)
 873     {
 874       table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
 875       extra = (const unsigned char *)
 876         _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
 877       indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
 878                                                 _NL_COLLATE_INDIRECTMB);
 879       p = pstr->mbs + idx;
 880       findidx (table, indirect, extra, &p, pstr->len - idx);
 881       return p - pstr->mbs - idx;
 882     }
 883   else
 884 # endif /* _LIBC */
 885     return 1;
 886 }
 887 #endif /* RE_ENABLE_I18N */
 888
 889 #ifndef __GNUC_PREREQ
 890 # if defined __GNUC__ && defined __GNUC_MINOR__
 891 #  define __GNUC_PREREQ(maj, min) \
 892          ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
 893 # else
 894 #  define __GNUC_PREREQ(maj, min) 0
 895 # endif
 896 #endif
 897
 898 #if __GNUC_PREREQ (3,4)
 899 # undef __attribute_warn_unused_result__
 900 # define __attribute_warn_unused_result__ \
 901    __attribute__ ((__warn_unused_result__))
 902 #else
 903 # define __attribute_warn_unused_result__ /* empty */
 904 #endif
 905
 906 #ifndef FALLTHROUGH
 907 # if __GNUC__ < 7
 908 #  define FALLTHROUGH ((void) 0)
 909 # else
 910 #  define FALLTHROUGH __attribute__ ((__fallthrough__))
 911 # endif
 912 #endif
 913
 914 #endif /*  _REGEX_INTERNAL_H */