lib/regex_internal.h

   1 /* Extended regular expression matching and search library.
   2    Copyright (C) 2002-2014 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 #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 #ifdef _LIBC
  37 # include <bits/libc-lock.h>
  38 # define lock_define(name) __libc_lock_define (, name)
  39 # define lock_init(lock) (__libc_lock_init (lock), 0)
  40 # define lock_fini(lock) 0
  41 # define lock_lock(lock) __libc_lock_lock (lock)
  42 # define lock_unlock(lock) __libc_lock_unlock (lock)
  43 #elif defined GNULIB_LOCK
  44 # include "glthread/lock.h"
  45   /* Use gl_lock_define if empty macro arguments are known to work.
  46      Otherwise, fall back on less-portable substitutes.  */
  47 # if ((defined __GNUC__ && !defined __STRICT_ANSI__) \
  48       || (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__))
  49 #  define lock_define(name) gl_lock_define (, name)
  50 # elif USE_POSIX_THREADS
  51 #  define lock_define(name) pthread_mutex_t name;
  52 # elif USE_PTH_THREADS
  53 #  define lock_define(name) pth_mutex_t name;
  54 # elif USE_SOLARIS_THREADS
  55 #  define lock_define(name) mutex_t name;
  56 # elif USE_WINDOWS_THREADS
  57 #  define lock_define(name) gl_lock_t name;
  58 # else
  59 #  define lock_define(name)
  60 # endif
  61 # define lock_init(lock) glthread_lock_init (&(lock))
  62 # define lock_fini(lock) glthread_lock_destroy (&(lock))
  63 # define lock_lock(lock) glthread_lock_lock (&(lock))
  64 # define lock_unlock(lock) glthread_lock_unlock (&(lock))
  65 #elif defined GNULIB_PTHREAD
  66 # include <pthread.h>
  67 # define lock_define(name) pthread_mutex_t name;
  68 # define lock_init(lock) pthread_mutex_init (&(lock), 0)
  69 # define lock_fini(lock) pthread_mutex_destroy (&(lock))
  70 # define lock_lock(lock) pthread_mutex_lock (&(lock))
  71 # define lock_unlock(lock) pthread_mutex_unlock (&(lock))
  72 #else
  73 # define lock_define(name)
  74 # define lock_init(lock) 0
  75 # define lock_fini(lock) ((void) 0)
  76   /* The 'dfa' avoids an "unused variable 'dfa'" warning from GCC.  */
  77 # define lock_lock(lock) ((void) dfa)
  78 # define lock_unlock(lock) ((void) 0)
  79 #endif
  80
  81 /* In case that the system doesn't have isblank().  */
  82 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK))
  83 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
  84 #endif
  85
  86 #ifdef _LIBC
  87 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
  88 #  define _RE_DEFINE_LOCALE_FUNCTIONS 1
  89 #   include <locale/localeinfo.h>
  90 #   include <locale/elem-hash.h>
  91 #   include <locale/coll-lookup.h>
  92 # endif
  93 #endif
  94
  95 /* This is for other GNU distributions with internationalized messages.  */
  96 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
  97 # include <libintl.h>
  98 # ifdef _LIBC
  99 #  undef gettext
 100 #  define gettext(msgid) \
 101   __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES)
 102 # endif
 103 #else
 104 # define gettext(msgid) (msgid)
 105 #endif
 106
 107 #ifndef gettext_noop
 108 /* This define is so xgettext can find the internationalizable
 109    strings.  */
 110 # define gettext_noop(String) String
 111 #endif
 112
 113 #if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE) || _LIBC
 114 # define RE_ENABLE_I18N
 115 #endif
 116
 117 #if __GNUC__ >= 3
 118 # define BE(expr, val) __builtin_expect (expr, val)
 119 #else
 120 # define BE(expr, val) (expr)
 121 #endif
 122
 123 /* Number of ASCII characters.  */
 124 #define ASCII_CHARS 0x80
 125
 126 /* Number of single byte characters.  */
 127 #define SBC_MAX (UCHAR_MAX + 1)
 128
 129 #define COLL_ELEM_LEN_MAX 8
 130
 131 /* The character which represents newline.  */
 132 #define NEWLINE_CHAR '\n'
 133 #define WIDE_NEWLINE_CHAR L'\n'
 134
 135 /* Rename to standard API for using out of glibc.  */
 136 #ifndef _LIBC
 137 # undef __wctype
 138 # undef __iswctype
 139 # define __wctype wctype
 140 # define __iswctype iswctype
 141 # define __btowc btowc
 142 # define __mbrtowc mbrtowc
 143 # define __wcrtomb wcrtomb
 144 # define __regfree regfree
 145 # define attribute_hidden
 146 #endif /* not _LIBC */
 147
 148 #if __GNUC__ < 3 + (__GNUC_MINOR__ < 1)
 149 # define __attribute__(arg)
 150 #endif
 151
 152 typedef __re_idx_t Idx;
 153 #ifdef _REGEX_LARGE_OFFSETS
 154 # define IDX_MAX (SIZE_MAX - 2)
 155 #else
 156 # define IDX_MAX INT_MAX
 157 #endif
 158
 159 /* Special return value for failure to match.  */
 160 #define REG_MISSING ((Idx) -1)
 161
 162 /* Special return value for internal error.  */
 163 #define REG_ERROR ((Idx) -2)
 164
 165 /* Test whether N is a valid index, and is not one of the above.  */
 166 #ifdef _REGEX_LARGE_OFFSETS
 167 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR)
 168 #else
 169 # define REG_VALID_INDEX(n) (0 <= (n))
 170 #endif
 171
 172 /* Test whether N is a valid nonzero index.  */
 173 #ifdef _REGEX_LARGE_OFFSETS
 174 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1))
 175 #else
 176 # define REG_VALID_NONZERO_INDEX(n) (0 < (n))
 177 #endif
 178
 179 /* A hash value, suitable for computing hash tables.  */
 180 typedef __re_size_t re_hashval_t;
 181
 182 /* An integer used to represent a set of bits.  It must be unsigned,
 183    and must be at least as wide as unsigned int.  */
 184 typedef unsigned long int bitset_word_t;
 185 /* All bits set in a bitset_word_t.  */
 186 #define BITSET_WORD_MAX ULONG_MAX
 187
 188 /* Number of bits in a bitset_word_t.  For portability to hosts with
 189    padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)';
 190    instead, deduce it directly from BITSET_WORD_MAX.  Avoid
 191    greater-than-32-bit integers and unconditional shifts by more than
 192    31 bits, as they're not portable.  */
 193 #if BITSET_WORD_MAX == 0xffffffffUL
 194 # define BITSET_WORD_BITS 32
 195 #elif BITSET_WORD_MAX >> 31 >> 4 == 1
 196 # define BITSET_WORD_BITS 36
 197 #elif BITSET_WORD_MAX >> 31 >> 16 == 1
 198 # define BITSET_WORD_BITS 48
 199 #elif BITSET_WORD_MAX >> 31 >> 28 == 1
 200 # define BITSET_WORD_BITS 60
 201 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1
 202 # define BITSET_WORD_BITS 64
 203 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1
 204 # define BITSET_WORD_BITS 72
 205 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1
 206 # define BITSET_WORD_BITS 128
 207 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1
 208 # define BITSET_WORD_BITS 256
 209 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1
 210 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */
 211 # if BITSET_WORD_BITS <= SBC_MAX
 212 #  error "Invalid SBC_MAX"
 213 # endif
 214 #else
 215 # error "Add case for new bitset_word_t size"
 216 #endif
 217
 218 /* Number of bitset_word_t values in a bitset_t.  */
 219 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS)
 220
 221 typedef bitset_word_t bitset_t[BITSET_WORDS];
 222 typedef bitset_word_t *re_bitset_ptr_t;
 223 typedef const bitset_word_t *re_const_bitset_ptr_t;
 224
 225 #define PREV_WORD_CONSTRAINT 0x0001
 226 #define PREV_NOTWORD_CONSTRAINT 0x0002
 227 #define NEXT_WORD_CONSTRAINT 0x0004
 228 #define NEXT_NOTWORD_CONSTRAINT 0x0008
 229 #define PREV_NEWLINE_CONSTRAINT 0x0010
 230 #define NEXT_NEWLINE_CONSTRAINT 0x0020
 231 #define PREV_BEGBUF_CONSTRAINT 0x0040
 232 #define NEXT_ENDBUF_CONSTRAINT 0x0080
 233 #define WORD_DELIM_CONSTRAINT 0x0100
 234 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
 235
 236 typedef enum
 237 {
 238   INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 239   WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
 240   WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 241   INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
 242   LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
 243   LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
 244   BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
 245   BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
 246   WORD_DELIM = WORD_DELIM_CONSTRAINT,
 247   NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
 248 } re_context_type;
 249
 250 typedef struct
 251 {
 252   Idx alloc;
 253   Idx nelem;
 254   Idx *elems;
 255 } re_node_set;
 256
 257 typedef enum
 258 {
 259   NON_TYPE = 0,
 260
 261   /* Node type, These are used by token, node, tree.  */
 262   CHARACTER = 1,
 263   END_OF_RE = 2,
 264   SIMPLE_BRACKET = 3,
 265   OP_BACK_REF = 4,
 266   OP_PERIOD = 5,
 267 #ifdef RE_ENABLE_I18N
 268   COMPLEX_BRACKET = 6,
 269   OP_UTF8_PERIOD = 7,
 270 #endif /* RE_ENABLE_I18N */
 271
 272   /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
 273      when the debugger shows values of this enum type.  */
 274 #define EPSILON_BIT 8
 275   OP_OPEN_SUBEXP = EPSILON_BIT | 0,
 276   OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
 277   OP_ALT = EPSILON_BIT | 2,
 278   OP_DUP_ASTERISK = EPSILON_BIT | 3,
 279   ANCHOR = EPSILON_BIT | 4,
 280
 281   /* Tree type, these are used only by tree. */
 282   CONCAT = 16,
 283   SUBEXP = 17,
 284
 285   /* Token type, these are used only by token.  */
 286   OP_DUP_PLUS = 18,
 287   OP_DUP_QUESTION,
 288   OP_OPEN_BRACKET,
 289   OP_CLOSE_BRACKET,
 290   OP_CHARSET_RANGE,
 291   OP_OPEN_DUP_NUM,
 292   OP_CLOSE_DUP_NUM,
 293   OP_NON_MATCH_LIST,
 294   OP_OPEN_COLL_ELEM,
 295   OP_CLOSE_COLL_ELEM,
 296   OP_OPEN_EQUIV_CLASS,
 297   OP_CLOSE_EQUIV_CLASS,
 298   OP_OPEN_CHAR_CLASS,
 299   OP_CLOSE_CHAR_CLASS,
 300   OP_WORD,
 301   OP_NOTWORD,
 302   OP_SPACE,
 303   OP_NOTSPACE,
 304   BACK_SLASH
 305
 306 } re_token_type_t;
 307
 308 #ifdef RE_ENABLE_I18N
 309 typedef struct
 310 {
 311   /* Multibyte characters.  */
 312   wchar_t *mbchars;
 313
 314   /* Collating symbols.  */
 315 # ifdef _LIBC
 316   int32_t *coll_syms;
 317 # endif
 318
 319   /* Equivalence classes. */
 320 # ifdef _LIBC
 321   int32_t *equiv_classes;
 322 # endif
 323
 324   /* Range expressions. */
 325 # ifdef _LIBC
 326   uint32_t *range_starts;
 327   uint32_t *range_ends;
 328 # else /* not _LIBC */
 329   wchar_t *range_starts;
 330   wchar_t *range_ends;
 331 # endif /* not _LIBC */
 332
 333   /* Character classes. */
 334   wctype_t *char_classes;
 335
 336   /* If this character set is the non-matching list.  */
 337   unsigned int non_match : 1;
 338
 339   /* # of multibyte characters.  */
 340   Idx nmbchars;
 341
 342   /* # of collating symbols.  */
 343   Idx ncoll_syms;
 344
 345   /* # of equivalence classes. */
 346   Idx nequiv_classes;
 347
 348   /* # of range expressions. */
 349   Idx nranges;
 350
 351   /* # of character classes. */
 352   Idx nchar_classes;
 353 } re_charset_t;
 354 #endif /* RE_ENABLE_I18N */
 355
 356 typedef struct
 357 {
 358   union
 359   {
 360     unsigned char c;            /* for CHARACTER */
 361     re_bitset_ptr_t sbcset;     /* for SIMPLE_BRACKET */
 362 #ifdef RE_ENABLE_I18N
 363     re_charset_t *mbcset;       /* for COMPLEX_BRACKET */
 364 #endif /* RE_ENABLE_I18N */
 365     Idx idx;                    /* for BACK_REF */
 366     re_context_type ctx_type;   /* for ANCHOR */
 367   } opr;
 368 #if __GNUC__ >= 2 && !defined __STRICT_ANSI__
 369   re_token_type_t type : 8;
 370 #else
 371   re_token_type_t type;
 372 #endif
 373   unsigned int constraint : 10; /* context constraint */
 374   unsigned int duplicated : 1;
 375   unsigned int opt_subexp : 1;
 376 #ifdef RE_ENABLE_I18N
 377   unsigned int accept_mb : 1;
 378   /* These 2 bits can be moved into the union if needed (e.g. if running out
 379      of bits; move opr.c to opr.c.c and move the flags to opr.c.flags).  */
 380   unsigned int mb_partial : 1;
 381 #endif
 382   unsigned int word_char : 1;
 383 } re_token_t;
 384
 385 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
 386
 387 struct re_string_t
 388 {
 389   /* Indicate the raw buffer which is the original string passed as an
 390      argument of regexec(), re_search(), etc..  */
 391   const unsigned char *raw_mbs;
 392   /* Store the multibyte string.  In case of "case insensitive mode" like
 393      REG_ICASE, upper cases of the string are stored, otherwise MBS points
 394      the same address that RAW_MBS points.  */
 395   unsigned char *mbs;
 396 #ifdef RE_ENABLE_I18N
 397   /* Store the wide character string which is corresponding to MBS.  */
 398   wint_t *wcs;
 399   Idx *offsets;
 400   mbstate_t cur_state;
 401 #endif
 402   /* Index in RAW_MBS.  Each character mbs[i] corresponds to
 403      raw_mbs[raw_mbs_idx + i].  */
 404   Idx raw_mbs_idx;
 405   /* The length of the valid characters in the buffers.  */
 406   Idx valid_len;
 407   /* The corresponding number of bytes in raw_mbs array.  */
 408   Idx valid_raw_len;
 409   /* The length of the buffers MBS and WCS.  */
 410   Idx bufs_len;
 411   /* The index in MBS, which is updated by re_string_fetch_byte.  */
 412   Idx cur_idx;
 413   /* length of RAW_MBS array.  */
 414   Idx raw_len;
 415   /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN.  */
 416   Idx len;
 417   /* End of the buffer may be shorter than its length in the cases such
 418      as re_match_2, re_search_2.  Then, we use STOP for end of the buffer
 419      instead of LEN.  */
 420   Idx raw_stop;
 421   /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS.  */
 422   Idx stop;
 423
 424   /* The context of mbs[0].  We store the context independently, since
 425      the context of mbs[0] may be different from raw_mbs[0], which is
 426      the beginning of the input string.  */
 427   unsigned int tip_context;
 428   /* The translation passed as a part of an argument of re_compile_pattern.  */
 429   RE_TRANSLATE_TYPE trans;
 430   /* Copy of re_dfa_t's word_char.  */
 431   re_const_bitset_ptr_t word_char;
 432   /* true if REG_ICASE.  */
 433   unsigned char icase;
 434   unsigned char is_utf8;
 435   unsigned char map_notascii;
 436   unsigned char mbs_allocated;
 437   unsigned char offsets_needed;
 438   unsigned char newline_anchor;
 439   unsigned char word_ops_used;
 440   int mb_cur_max;
 441 };
 442 typedef struct re_string_t re_string_t;
 443
 444
 445 struct re_dfa_t;
 446 typedef struct re_dfa_t re_dfa_t;
 447
 448 #ifndef _LIBC
 449 # define internal_function
 450 #endif
 451
 452 #ifndef NOT_IN_libc
 453 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
 454                                                 Idx new_buf_len)
 455      internal_function;
 456 # ifdef RE_ENABLE_I18N
 457 static void build_wcs_buffer (re_string_t *pstr) internal_function;
 458 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
 459   internal_function;
 460 # endif /* RE_ENABLE_I18N */
 461 static void build_upper_buffer (re_string_t *pstr) internal_function;
 462 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
 463 static unsigned int re_string_context_at (const re_string_t *input, Idx idx,
 464                                           int eflags)
 465      internal_function __attribute__ ((pure));
 466 #endif
 467 #define re_string_peek_byte(pstr, offset) \
 468   ((pstr)->mbs[(pstr)->cur_idx + offset])
 469 #define re_string_fetch_byte(pstr) \
 470   ((pstr)->mbs[(pstr)->cur_idx++])
 471 #define re_string_first_byte(pstr, idx) \
 472   ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
 473 #define re_string_is_single_byte_char(pstr, idx) \
 474   ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
 475                                 || (pstr)->wcs[(idx) + 1] != WEOF))
 476 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
 477 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
 478 #define re_string_get_buffer(pstr) ((pstr)->mbs)
 479 #define re_string_length(pstr) ((pstr)->len)
 480 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
 481 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
 482 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
 483
 484 #if defined _LIBC || HAVE_ALLOCA
 485 # include <alloca.h>
 486 #endif
 487
 488 #ifndef _LIBC
 489 # if HAVE_ALLOCA
 490 /* The OS usually guarantees only one guard page at the bottom of the stack,
 491    and a page size can be as small as 4096 bytes.  So we cannot safely
 492    allocate anything larger than 4096 bytes.  Also care for the possibility
 493    of a few compiler-allocated temporary stack slots.  */
 494 #  define __libc_use_alloca(n) ((n) < 4032)
 495 # else
 496 /* alloca is implemented with malloc, so just use malloc.  */
 497 #  define __libc_use_alloca(n) 0
 498 #  undef alloca
 499 #  define alloca(n) malloc (n)
 500 # endif
 501 #endif
 502
 503 #ifdef _LIBC
 504 # define MALLOC_0_IS_NONNULL 1
 505 #elif !defined MALLOC_0_IS_NONNULL
 506 # define MALLOC_0_IS_NONNULL 0
 507 #endif
 508
 509 #ifndef MAX
 510 # define MAX(a,b) ((a) < (b) ? (b) : (a))
 511 #endif
 512 #ifndef MIN
 513 # define MIN(a,b) ((a) < (b) ? (a) : (b))
 514 #endif
 515
 516 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
 517 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
 518 #define re_free(p) free (p)
 519
 520 struct bin_tree_t
 521 {
 522   struct bin_tree_t *parent;
 523   struct bin_tree_t *left;
 524   struct bin_tree_t *right;
 525   struct bin_tree_t *first;
 526   struct bin_tree_t *next;
 527
 528   re_token_t token;
 529
 530   /* 'node_idx' is the index in dfa->nodes, if 'type' == 0.
 531      Otherwise 'type' indicate the type of this node.  */
 532   Idx node_idx;
 533 };
 534 typedef struct bin_tree_t bin_tree_t;
 535
 536 #define BIN_TREE_STORAGE_SIZE \
 537   ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
 538
 539 struct bin_tree_storage_t
 540 {
 541   struct bin_tree_storage_t *next;
 542   bin_tree_t data[BIN_TREE_STORAGE_SIZE];
 543 };
 544 typedef struct bin_tree_storage_t bin_tree_storage_t;
 545
 546 #define CONTEXT_WORD 1
 547 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
 548 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
 549 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
 550
 551 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
 552 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
 553 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
 554 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
 555 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
 556
 557 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
 558 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
 559 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
 560 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
 561
 562 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
 563  ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 564   || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 565   || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
 566   || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
 567
 568 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
 569  ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
 570   || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
 571   || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
 572   || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
 573
 574 struct re_dfastate_t
 575 {
 576   re_hashval_t hash;
 577   re_node_set nodes;
 578   re_node_set non_eps_nodes;
 579   re_node_set inveclosure;
 580   re_node_set *entrance_nodes;
 581   struct re_dfastate_t **trtable, **word_trtable;
 582   unsigned int context : 4;
 583   unsigned int halt : 1;
 584   /* If this state can accept "multi byte".
 585      Note that we refer to multibyte characters, and multi character
 586      collating elements as "multi byte".  */
 587   unsigned int accept_mb : 1;
 588   /* If this state has backreference node(s).  */
 589   unsigned int has_backref : 1;
 590   unsigned int has_constraint : 1;
 591 };
 592 typedef struct re_dfastate_t re_dfastate_t;
 593
 594 struct re_state_table_entry
 595 {
 596   Idx num;
 597   Idx alloc;
 598   re_dfastate_t **array;
 599 };
 600
 601 /* Array type used in re_sub_match_last_t and re_sub_match_top_t.  */
 602
 603 typedef struct
 604 {
 605   Idx next_idx;
 606   Idx alloc;
 607   re_dfastate_t **array;
 608 } state_array_t;
 609
 610 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP.  */
 611
 612 typedef struct
 613 {
 614   Idx node;
 615   Idx str_idx; /* The position NODE match at.  */
 616   state_array_t path;
 617 } re_sub_match_last_t;
 618
 619 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
 620    And information about the node, whose type is OP_CLOSE_SUBEXP,
 621    corresponding to NODE is stored in LASTS.  */
 622
 623 typedef struct
 624 {
 625   Idx str_idx;
 626   Idx node;
 627   state_array_t *path;
 628   Idx alasts; /* Allocation size of LASTS.  */
 629   Idx nlasts; /* The number of LASTS.  */
 630   re_sub_match_last_t **lasts;
 631 } re_sub_match_top_t;
 632
 633 struct re_backref_cache_entry
 634 {
 635   Idx node;
 636   Idx str_idx;
 637   Idx subexp_from;
 638   Idx subexp_to;
 639   char more;
 640   char unused;
 641   unsigned short int eps_reachable_subexps_map;
 642 };
 643
 644 typedef struct
 645 {
 646   /* The string object corresponding to the input string.  */
 647   re_string_t input;
 648 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
 649   const re_dfa_t *const dfa;
 650 #else
 651   const re_dfa_t *dfa;
 652 #endif
 653   /* EFLAGS of the argument of regexec.  */
 654   int eflags;
 655   /* Where the matching ends.  */
 656   Idx match_last;
 657   Idx last_node;
 658   /* The state log used by the matcher.  */
 659   re_dfastate_t **state_log;
 660   Idx state_log_top;
 661   /* Back reference cache.  */
 662   Idx nbkref_ents;
 663   Idx abkref_ents;
 664   struct re_backref_cache_entry *bkref_ents;
 665   int max_mb_elem_len;
 666   Idx nsub_tops;
 667   Idx asub_tops;
 668   re_sub_match_top_t **sub_tops;
 669 } re_match_context_t;
 670
 671 typedef struct
 672 {
 673   re_dfastate_t **sifted_states;
 674   re_dfastate_t **limited_states;
 675   Idx last_node;
 676   Idx last_str_idx;
 677   re_node_set limits;
 678 } re_sift_context_t;
 679
 680 struct re_fail_stack_ent_t
 681 {
 682   Idx idx;
 683   Idx node;
 684   regmatch_t *regs;
 685   re_node_set eps_via_nodes;
 686 };
 687
 688 struct re_fail_stack_t
 689 {
 690   Idx num;
 691   Idx alloc;
 692   struct re_fail_stack_ent_t *stack;
 693 };
 694
 695 struct re_dfa_t
 696 {
 697   re_token_t *nodes;
 698   size_t nodes_alloc;
 699   size_t nodes_len;
 700   Idx *nexts;
 701   Idx *org_indices;
 702   re_node_set *edests;
 703   re_node_set *eclosures;
 704   re_node_set *inveclosures;
 705   struct re_state_table_entry *state_table;
 706   re_dfastate_t *init_state;
 707   re_dfastate_t *init_state_word;
 708   re_dfastate_t *init_state_nl;
 709   re_dfastate_t *init_state_begbuf;
 710   bin_tree_t *str_tree;
 711   bin_tree_storage_t *str_tree_storage;
 712   re_bitset_ptr_t sb_char;
 713   int str_tree_storage_idx;
 714
 715   /* number of subexpressions 're_nsub' is in regex_t.  */
 716   re_hashval_t state_hash_mask;
 717   Idx init_node;
 718   Idx nbackref; /* The number of backreference in this dfa.  */
 719
 720   /* Bitmap expressing which backreference is used.  */
 721   bitset_word_t used_bkref_map;
 722   bitset_word_t completed_bkref_map;
 723
 724   unsigned int has_plural_match : 1;
 725   /* If this dfa has "multibyte node", which is a backreference or
 726      a node which can accept multibyte character or multi character
 727      collating element.  */
 728   unsigned int has_mb_node : 1;
 729   unsigned int is_utf8 : 1;
 730   unsigned int map_notascii : 1;
 731   unsigned int word_ops_used : 1;
 732   int mb_cur_max;
 733   bitset_t word_char;
 734   reg_syntax_t syntax;
 735   Idx *subexp_map;
 736 #ifdef DEBUG
 737   char* re_str;
 738 #endif
 739   lock_define (lock)
 740 };
 741
 742 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
 743 #define re_node_set_remove(set,id) \
 744   (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
 745 #define re_node_set_empty(p) ((p)->nelem = 0)
 746 #define re_node_set_free(set) re_free ((set)->elems)
 747 \f
 748
 749 typedef enum
 750 {
 751   SB_CHAR,
 752   MB_CHAR,
 753   EQUIV_CLASS,
 754   COLL_SYM,
 755   CHAR_CLASS
 756 } bracket_elem_type;
 757
 758 typedef struct
 759 {
 760   bracket_elem_type type;
 761   union
 762   {
 763     unsigned char ch;
 764     unsigned char *name;
 765     wchar_t wch;
 766   } opr;
 767 } bracket_elem_t;
 768
 769
 770 /* Functions for bitset_t operation.  */
 771
 772 static void
 773 bitset_set (bitset_t set, Idx i)
 774 {
 775   set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS;
 776 }
 777
 778 static void
 779 bitset_clear (bitset_t set, Idx i)
 780 {
 781   set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS);
 782 }
 783
 784 static bool
 785 bitset_contain (const bitset_t set, Idx i)
 786 {
 787   return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1;
 788 }
 789
 790 static void
 791 bitset_empty (bitset_t set)
 792 {
 793   memset (set, '\0', sizeof (bitset_t));
 794 }
 795
 796 static void
 797 bitset_set_all (bitset_t set)
 798 {
 799   memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS));
 800   if (SBC_MAX % BITSET_WORD_BITS != 0)
 801     set[BITSET_WORDS - 1] =
 802       ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1;
 803 }
 804
 805 static void
 806 bitset_copy (bitset_t dest, const bitset_t src)
 807 {
 808   memcpy (dest, src, sizeof (bitset_t));
 809 }
 810
 811 static void __attribute__ ((unused))
 812 bitset_not (bitset_t set)
 813 {
 814   int bitset_i;
 815   for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i)
 816     set[bitset_i] = ~set[bitset_i];
 817   if (SBC_MAX % BITSET_WORD_BITS != 0)
 818     set[BITSET_WORDS - 1] =
 819       ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1)
 820        & ~set[BITSET_WORDS - 1]);
 821 }
 822
 823 static void __attribute__ ((unused))
 824 bitset_merge (bitset_t dest, const bitset_t src)
 825 {
 826   int bitset_i;
 827   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 828     dest[bitset_i] |= src[bitset_i];
 829 }
 830
 831 static void __attribute__ ((unused))
 832 bitset_mask (bitset_t dest, const bitset_t src)
 833 {
 834   int bitset_i;
 835   for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
 836     dest[bitset_i] &= src[bitset_i];
 837 }
 838
 839 #ifdef RE_ENABLE_I18N
 840 /* Functions for re_string.  */
 841 static int
 842 internal_function __attribute__ ((pure, unused))
 843 re_string_char_size_at (const re_string_t *pstr, Idx idx)
 844 {
 845   int byte_idx;
 846   if (pstr->mb_cur_max == 1)
 847     return 1;
 848   for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
 849     if (pstr->wcs[idx + byte_idx] != WEOF)
 850       break;
 851   return byte_idx;
 852 }
 853
 854 static wint_t
 855 internal_function __attribute__ ((pure, unused))
 856 re_string_wchar_at (const re_string_t *pstr, Idx idx)
 857 {
 858   if (pstr->mb_cur_max == 1)
 859     return (wint_t) pstr->mbs[idx];
 860   return (wint_t) pstr->wcs[idx];
 861 }
 862
 863 # ifndef NOT_IN_libc
 864 static int
 865 internal_function __attribute__ ((pure, unused))
 866 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
 867 {
 868 #  ifdef _LIBC
 869   const unsigned char *p, *extra;
 870   const int32_t *table, *indirect;
 871 #   include <locale/weight.h>
 872   uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
 873
 874   if (nrules != 0)
 875     {
 876       table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
 877       extra = (const unsigned char *)
 878         _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
 879       indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
 880                                                 _NL_COLLATE_INDIRECTMB);
 881       p = pstr->mbs + idx;
 882       findidx (&p, pstr->len - idx);
 883       return p - pstr->mbs - idx;
 884     }
 885   else
 886 #  endif /* _LIBC */
 887     return 1;
 888 }
 889 # endif
 890 #endif /* RE_ENABLE_I18N */
 891
 892 #ifndef __GNUC_PREREQ
 893 # if defined __GNUC__ && defined __GNUC_MINOR__
 894 #  define __GNUC_PREREQ(maj, min) \
 895          ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
 896 # else
 897 #  define __GNUC_PREREQ(maj, min) 0
 898 # endif
 899 #endif
 900
 901 #if __GNUC_PREREQ (3,4)
 902 # undef __attribute_warn_unused_result__
 903 # define __attribute_warn_unused_result__ \
 904    __attribute__ ((__warn_unused_result__))
 905 #else
 906 # define __attribute_warn_unused_result__ /* empty */
 907 #endif
 908
 909 #endif /*  _REGEX_INTERNAL_H */