regint.h

   1 #ifndef ONIGMO_REGINT_H
   2 #define ONIGMO_REGINT_H
   3 /**********************************************************************
   4   regint.h -  Onigmo (Oniguruma-mod) (regular expression library)
   5 **********************************************************************/
   6 /*-
   7  * Copyright (c) 2002-2013  K.Kosako  <sndgk393 AT ybb DOT ne DOT jp>
   8  * Copyright (c) 2011-2016  K.Takata  <kentkt AT csc DOT jp>
   9  * All rights reserved.
  10  *
  11  * Redistribution and use in source and binary forms, with or without
  12  * modification, are permitted provided that the following conditions
  13  * are met:
  14  * 1. Redistributions of source code must retain the above copyright
  15  *    notice, this list of conditions and the following disclaimer.
  16  * 2. Redistributions in binary form must reproduce the above copyright
  17  *    notice, this list of conditions and the following disclaimer in the
  18  *    documentation and/or other materials provided with the distribution.
  19  *
  20  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  30  * SUCH DAMAGE.
  31  */
  32
  33 /* for debug */
  34 /* #define ONIG_DEBUG_PARSE_TREE */
  35 /* #define ONIG_DEBUG_COMPILE */
  36 /* #define ONIG_DEBUG_SEARCH */
  37 /* #define ONIG_DEBUG_MATCH */
  38 /* #define ONIG_DEBUG_MATCH_CACHE */
  39 /* #define ONIG_DEBUG_MEMLEAK */
  40 /* #define ONIG_DONT_OPTIMIZE */
  41
  42 /* for byte-code statistical data. */
  43 /* #define ONIG_DEBUG_STATISTICS */
  44
  45 /* enable the match optimization by using a cache. */
  46 #define USE_MATCH_CACHE
  47
  48 #if defined(ONIG_DEBUG_PARSE_TREE) || defined(ONIG_DEBUG_MATCH) || \
  49     defined(ONIG_DEBUG_SEARCH) || defined(ONIG_DEBUG_COMPILE) || \
  50     defined(ONIG_DEBUG_STATISTICS) || defined(ONIG_DEBUG_MEMLEAK)
  51 # ifndef ONIG_DEBUG
  52 #  define ONIG_DEBUG
  53 # endif
  54 #endif
  55
  56 /* __POWERPC__ added to accommodate Darwin case. */
  57 #ifndef UNALIGNED_WORD_ACCESS
  58 # if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \
  59      defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD64) || \
  60      defined(__powerpc64__) || defined(__POWERPC__) || defined(__aarch64__) || \
  61      defined(__mc68020__)
  62 #  define UNALIGNED_WORD_ACCESS 1
  63 # else
  64 #  define UNALIGNED_WORD_ACCESS 0
  65 # endif
  66 #endif
  67
  68 #if UNALIGNED_WORD_ACCESS
  69 # define PLATFORM_UNALIGNED_WORD_ACCESS
  70 #endif
  71
  72 /* config */
  73 /* spec. config */
  74 #define USE_NAMED_GROUP
  75 #define USE_SUBEXP_CALL
  76 #define USE_PERL_SUBEXP_CALL
  77 #define USE_CAPITAL_P_NAMED_GROUP
  78 #define USE_BACKREF_WITH_LEVEL        /* \k<name+n>, \k<name-n> */
  79 #define USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT  /* /(?:()|())*\2/ */
  80 #define USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE     /* /\n$/ =~ "\n" */
  81 #define USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR
  82 /* !!! moved to regenc.h. */ /* #define USE_CRNL_AS_LINE_TERMINATOR */
  83 #define USE_NO_INVALID_QUANTIFIER
  84
  85 /* internal config */
  86 /* #define USE_OP_PUSH_OR_JUMP_EXACT */
  87 #define USE_QTFR_PEEK_NEXT
  88 #define USE_ST_LIBRARY
  89 #define USE_SUNDAY_QUICK_SEARCH
  90
  91 #define INIT_MATCH_STACK_SIZE                     160
  92 #define DEFAULT_MATCH_STACK_LIMIT_SIZE              0 /* unlimited */
  93 #define DEFAULT_PARSE_DEPTH_LIMIT                4096
  94
  95 #define OPT_EXACT_MAXLEN   24
  96
  97 /* check config */
  98 #if defined(USE_PERL_SUBEXP_CALL) || defined(USE_CAPITAL_P_NAMED_GROUP)
  99 # if !defined(USE_NAMED_GROUP) || !defined(USE_SUBEXP_CALL)
 100 #  error USE_NAMED_GROUP and USE_SUBEXP_CALL must be defined.
 101 # endif
 102 #endif
 103
 104 #if defined(__GNUC__)
 105 # define ARG_UNUSED  __attribute__ ((unused))
 106 #else
 107 # define ARG_UNUSED
 108 #endif
 109
 110 #if !defined(RUBY) && defined(RUBY_EXPORT)
 111 # define RUBY
 112 #endif
 113 #ifdef RUBY
 114 # ifndef RUBY_DEFINES_H
 115 #  include "ruby/ruby.h"
 116 #  undef xmalloc
 117 #  undef xrealloc
 118 #  undef xcalloc
 119 #  undef xfree
 120 # endif
 121 #else /* RUBY */
 122 # include "config.h"
 123 # if SIZEOF_LONG_LONG > 0
 124 #  define LONG_LONG long long
 125 # endif
 126 #endif /* RUBY */
 127
 128 #include <stdarg.h>
 129
 130 /* */
 131 /* escape other system UChar definition */
 132 #ifdef ONIG_ESCAPE_UCHAR_COLLISION
 133 # undef ONIG_ESCAPE_UCHAR_COLLISION
 134 #endif
 135
 136 #define USE_WORD_BEGIN_END          /* "\<": word-begin, "\>": word-end */
 137 #ifdef RUBY
 138 # undef USE_CAPTURE_HISTORY
 139 #else
 140 # define USE_CAPTURE_HISTORY
 141 #endif
 142 #define USE_VARIABLE_META_CHARS
 143 #define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
 144 /* #define USE_COMBINATION_EXPLOSION_CHECK */     /* (X*)* */
 145
 146
 147 #ifndef xmalloc
 148 # define xmalloc     malloc
 149 # define xrealloc    realloc
 150 # define xcalloc     calloc
 151 # define xfree       free
 152 #endif
 153
 154 #ifdef RUBY
 155
 156 # define CHECK_INTERRUPT_IN_MATCH_AT do { \
 157   msa->counter++; \
 158   if (msa->counter >= 128) { \
 159     msa->counter = 0; \
 160     if (rb_reg_timeout_p(reg, &msa->end_time)) { \
 161       goto timeout; \
 162     } \
 163     rb_thread_check_ints(); \
 164   } \
 165 } while(0)
 166 # define HANDLE_REG_TIMEOUT_IN_MATCH_AT do { \
 167   rb_reg_raise_timeout(); \
 168 } while (0)
 169 # define onig_st_init_table                  st_init_table
 170 # define onig_st_init_table_with_size        st_init_table_with_size
 171 # define onig_st_init_numtable               st_init_numtable
 172 # define onig_st_init_numtable_with_size     st_init_numtable_with_size
 173 # define onig_st_init_strtable               st_init_strtable
 174 # define onig_st_init_strtable_with_size     st_init_strtable_with_size
 175 # define onig_st_delete                      st_delete
 176 # define onig_st_delete_safe                 st_delete_safe
 177 # define onig_st_insert                      st_insert
 178 # define onig_st_lookup                      st_lookup
 179 # define onig_st_foreach                     st_foreach
 180 # define onig_st_add_direct                  st_add_direct
 181 # define onig_st_free_table                  st_free_table
 182 # define onig_st_cleanup_safe                st_cleanup_safe
 183 # define onig_st_copy                        st_copy
 184 # define onig_st_nothing_key_clone           st_nothing_key_clone
 185 # define onig_st_nothing_key_free            st_nothing_key_free
 186 # define onig_st_is_member                   st_is_member
 187
 188 # define USE_UPPER_CASE_TABLE
 189 #else /* RUBY */
 190
 191 # define CHECK_INTERRUPT_IN_MATCH_AT
 192
 193 # define st_init_table                  onig_st_init_table
 194 # define st_init_table_with_size        onig_st_init_table_with_size
 195 # define st_init_numtable               onig_st_init_numtable
 196 # define st_init_numtable_with_size     onig_st_init_numtable_with_size
 197 # define st_init_strtable               onig_st_init_strtable
 198 # define st_init_strtable_with_size     onig_st_init_strtable_with_size
 199 # define st_delete                      onig_st_delete
 200 # define st_delete_safe                 onig_st_delete_safe
 201 # define st_insert                      onig_st_insert
 202 # define st_lookup                      onig_st_lookup
 203 # define st_foreach                     onig_st_foreach
 204 # define st_add_direct                  onig_st_add_direct
 205 # define st_free_table                  onig_st_free_table
 206 # define st_cleanup_safe                onig_st_cleanup_safe
 207 # define st_copy                        onig_st_copy
 208 # define st_nothing_key_clone           onig_st_nothing_key_clone
 209 # define st_nothing_key_free            onig_st_nothing_key_free
 210 /* */
 211 # define onig_st_is_member              st_is_member
 212
 213 #endif /* RUBY */
 214
 215 #define STATE_CHECK_STRING_THRESHOLD_LEN             7
 216 #define STATE_CHECK_BUFF_MAX_SIZE               0x4000
 217
 218 #define xmemset     memset
 219 #define xmemcpy     memcpy
 220 #define xmemmove    memmove
 221
 222 #if ((defined(RUBY_MSVCRT_VERSION) && RUBY_MSVCRT_VERSION >= 90) \
 223         || (!defined(RUBY_MSVCRT_VERSION) && defined(_WIN32))) \
 224     && !defined(__GNUC__)
 225 # define xalloca     _alloca
 226 # define xvsnprintf(buf,size,fmt,args)  _vsnprintf_s(buf,size,_TRUNCATE,fmt,args)
 227 # define xsnprintf   sprintf_s
 228 # define xstrcat(dest,src,size)   strcat_s(dest,size,src)
 229 #else
 230 # define xalloca     alloca
 231 # define xvsnprintf  vsnprintf
 232 # define xsnprintf   snprintf
 233 # define xstrcat(dest,src,size)   strcat(dest,src)
 234 #endif
 235
 236 #if defined(ONIG_DEBUG_MEMLEAK) && defined(_MSC_VER)
 237 # define _CRTDBG_MAP_ALLOC
 238 # include <malloc.h>
 239 # include <crtdbg.h>
 240 #endif
 241
 242 #include <stdlib.h>
 243
 244 #if defined(HAVE_ALLOCA_H) && (defined(_AIX) || !defined(__GNUC__))
 245 # include <alloca.h>
 246 #endif
 247
 248 #include <string.h>
 249
 250 #include <ctype.h>
 251 #ifdef HAVE_SYS_TYPES_H
 252 # include <sys/types.h>
 253 #endif
 254
 255 #ifdef HAVE_STDINT_H
 256 # include <stdint.h>
 257 #endif
 258
 259 #ifdef HAVE_INTTYPES_H
 260 # include <inttypes.h>
 261 #endif
 262
 263 #include <stddef.h>
 264
 265 #ifdef _WIN32
 266 # include <malloc.h>    /* for alloca() */
 267 #endif
 268
 269 #ifdef ONIG_DEBUG
 270 # include <stdio.h>
 271 #endif
 272
 273 #ifdef _WIN32
 274 # if defined(_MSC_VER) && (_MSC_VER < 1300)
 275 #  ifndef _INTPTR_T_DEFINED
 276 #   define _INTPTR_T_DEFINED
 277 typedef int intptr_t;
 278 #  endif
 279 #  ifndef _UINTPTR_T_DEFINED
 280 #   define _UINTPTR_T_DEFINED
 281 typedef unsigned int uintptr_t;
 282 #  endif
 283 # endif
 284 #endif /* _WIN32 */
 285
 286 #ifndef PRIdPTR
 287 # ifdef _WIN64
 288 #  define PRIdPTR       "I64d"
 289 #  define PRIuPTR       "I64u"
 290 #  define PRIxPTR       "I64x"
 291 # else
 292 #  define PRIdPTR       "ld"
 293 #  define PRIuPTR       "lu"
 294 #  define PRIxPTR       "lx"
 295 # endif
 296 #endif
 297
 298 #ifndef PRIdPTRDIFF
 299 # define PRIdPTRDIFF PRIdPTR
 300 #endif
 301
 302 #include "regenc.h"
 303
 304 RUBY_SYMBOL_EXPORT_BEGIN
 305
 306 #ifdef MIN
 307 # undef MIN
 308 #endif
 309 #ifdef MAX
 310 # undef MAX
 311 #endif
 312 #define MIN(a,b) (((a)>(b))?(b):(a))
 313 #define MAX(a,b) (((a)<(b))?(b):(a))
 314
 315 #define IS_NULL(p)                    (((void*)(p)) == (void*)0)
 316 #define IS_NOT_NULL(p)                (((void*)(p)) != (void*)0)
 317 #define CHECK_NULL_RETURN(p)          if (IS_NULL(p)) return NULL
 318 #define CHECK_NULL_RETURN_MEMERR(p)   if (IS_NULL(p)) return ONIGERR_MEMORY
 319 #define NULL_UCHARP                   ((UChar* )0)
 320
 321 #define ONIG_LAST_CODE_POINT    (~((OnigCodePoint )0))
 322
 323 #define PLATFORM_GET_INC_ARGUMENTS_ASSERT(val, type) \
 324   ((void)sizeof(char[2 * (sizeof(val) == sizeof(type)) - 1]))
 325
 326 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
 327
 328 # define PLATFORM_GET_INC(val,p,type) do{\
 329   PLATFORM_GET_INC_ARGUMENTS_ASSERT(val, type);\
 330   val  = *(type* )p;\
 331   (p) += sizeof(type);\
 332 } while(0)
 333
 334 #else
 335
 336 # define PLATFORM_GET_INC(val,p,type) do{\
 337   PLATFORM_GET_INC_ARGUMENTS_ASSERT(val, type);\
 338   type platform_get_value;\
 339   xmemcpy(&platform_get_value, (p), sizeof(type));\
 340   val = platform_get_value;\
 341   (p) += sizeof(type);\
 342 } while(0)
 343
 344 /* sizeof(OnigCodePoint) */
 345 # define WORD_ALIGNMENT_SIZE     SIZEOF_LONG
 346
 347 # define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\
 348   (pad_size) = WORD_ALIGNMENT_SIZE \
 349                - ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
 350   if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\
 351 } while (0)
 352
 353 # define ALIGNMENT_RIGHT(addr) do {\
 354   (addr) += (WORD_ALIGNMENT_SIZE - 1);\
 355   (addr) -= ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
 356 } while (0)
 357
 358 #endif /* PLATFORM_UNALIGNED_WORD_ACCESS */
 359
 360 /* stack pop level */
 361 #define STACK_POP_LEVEL_FREE        0
 362 #define STACK_POP_LEVEL_MEM_START   1
 363 #define STACK_POP_LEVEL_ALL         2
 364
 365 /* optimize flags */
 366 #define ONIG_OPTIMIZE_NONE              0
 367 #define ONIG_OPTIMIZE_EXACT             1   /* Slow Search */
 368 #define ONIG_OPTIMIZE_EXACT_BM          2   /* Boyer Moore Search */
 369 #define ONIG_OPTIMIZE_EXACT_BM_NOT_REV  3   /* BM (applied to a multibyte string) */
 370 #define ONIG_OPTIMIZE_EXACT_IC          4   /* Slow Search (ignore case) */
 371 #define ONIG_OPTIMIZE_MAP               5   /* char map */
 372 #define ONIG_OPTIMIZE_EXACT_BM_IC         6 /* BM (ignore case) */
 373 #define ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC 7 /* BM (applied to a multibyte string) (ignore case) */
 374
 375 /* bit status */
 376 typedef unsigned int  BitStatusType;
 377
 378 #define BIT_STATUS_BITS_NUM          (sizeof(BitStatusType) * 8)
 379 #define BIT_STATUS_CLEAR(stats)      (stats) = 0
 380 #define BIT_STATUS_ON_ALL(stats)     (stats) = ~((BitStatusType )0)
 381 #define BIT_STATUS_AT(stats,n) \
 382   ((n) < (int )BIT_STATUS_BITS_NUM  ?  ((stats) & ((BitStatusType )1 << n)) : ((stats) & 1))
 383
 384 #define BIT_STATUS_ON_AT(stats,n) do {\
 385   if ((n) < (int )BIT_STATUS_BITS_NUM)\
 386     (stats) |= (1 << (n));\
 387   else\
 388     (stats) |= 1;\
 389 } while (0)
 390
 391 #define BIT_STATUS_ON_AT_SIMPLE(stats,n) do {\
 392   if ((n) < (int )BIT_STATUS_BITS_NUM)\
 393     (stats) |= (1 << (n));\
 394 } while (0)
 395
 396
 397 #define INT_MAX_LIMIT           ((1UL << (SIZEOF_INT * 8 - 1)) - 1)
 398 #define LONG_MAX_LIMIT           ((1UL << (SIZEOF_LONG * 8 - 1)) - 1)
 399
 400 #define DIGITVAL(code)    ((code) - '0')
 401 #define ODIGITVAL(code)   DIGITVAL(code)
 402 #define XDIGITVAL(enc,code) \
 403   (ONIGENC_IS_CODE_DIGIT(enc,code) ? DIGITVAL(code) \
 404    : (ONIGENC_IS_CODE_UPPER(enc,code) ? (code) - 'A' + 10 : (code) - 'a' + 10))
 405
 406 #define IS_SINGLELINE(option)     ((option) & ONIG_OPTION_SINGLELINE)
 407 #define IS_MULTILINE(option)      ((option) & ONIG_OPTION_MULTILINE)
 408 #define IS_IGNORECASE(option)     ((option) & ONIG_OPTION_IGNORECASE)
 409 #define IS_EXTEND(option)         ((option) & ONIG_OPTION_EXTEND)
 410 #define IS_FIND_LONGEST(option)   ((option) & ONIG_OPTION_FIND_LONGEST)
 411 #define IS_FIND_NOT_EMPTY(option) ((option) & ONIG_OPTION_FIND_NOT_EMPTY)
 412 #define IS_FIND_CONDITION(option) ((option) & \
 413           (ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY))
 414 #define IS_NOTBOL(option)         ((option) & ONIG_OPTION_NOTBOL)
 415 #define IS_NOTEOL(option)         ((option) & ONIG_OPTION_NOTEOL)
 416 #define IS_NOTBOS(option)         ((option) & ONIG_OPTION_NOTBOS)
 417 #define IS_NOTEOS(option)         ((option) & ONIG_OPTION_NOTEOS)
 418 #define IS_ASCII_RANGE(option)    ((option) & ONIG_OPTION_ASCII_RANGE)
 419 #define IS_POSIX_BRACKET_ALL_RANGE(option)  ((option) & ONIG_OPTION_POSIX_BRACKET_ALL_RANGE)
 420 #define IS_WORD_BOUND_ALL_RANGE(option)     ((option) & ONIG_OPTION_WORD_BOUND_ALL_RANGE)
 421 #define IS_NEWLINE_CRLF(option)   ((option) & ONIG_OPTION_NEWLINE_CRLF)
 422
 423 /* OP_SET_OPTION is required for these options.
 424 #define IS_DYNAMIC_OPTION(option) \
 425   (((option) & (ONIG_OPTION_MULTILINE | ONIG_OPTION_IGNORECASE)) != 0)
 426 */
 427 /* ignore-case and multibyte status are included in compiled code. */
 428 #define IS_DYNAMIC_OPTION(option)  0
 429
 430 #define DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag) \
 431   ((case_fold_flag) & ~INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR)
 432
 433 #define REPEAT_INFINITE         -1
 434 #define IS_REPEAT_INFINITE(n)   ((n) == REPEAT_INFINITE)
 435
 436 /* bitset */
 437 #define BITS_PER_BYTE      8
 438 #define SINGLE_BYTE_SIZE   (1 << BITS_PER_BYTE)
 439 #define BITS_IN_ROOM       ((int )sizeof(Bits) * BITS_PER_BYTE)
 440 #define BITSET_SIZE        (SINGLE_BYTE_SIZE / BITS_IN_ROOM)
 441
 442 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
 443 typedef unsigned int   Bits;
 444 #else
 445 typedef unsigned char  Bits;
 446 #endif
 447 typedef Bits           BitSet[BITSET_SIZE];
 448 typedef Bits*          BitSetRef;
 449
 450 #define SIZE_BITSET        (int )sizeof(BitSet)
 451
 452 #define BITSET_CLEAR(bs) do {\
 453   int i;\
 454   for (i = 0; i < BITSET_SIZE; i++) { (bs)[i] = 0; }    \
 455 } while (0)
 456
 457 #define BS_ROOM(bs,pos)            (bs)[(int )(pos) / BITS_IN_ROOM]
 458 #define BS_BIT(pos)                (1U << ((int )(pos) % BITS_IN_ROOM))
 459
 460 #define BITSET_AT(bs, pos)         (BS_ROOM(bs,pos) & BS_BIT(pos))
 461 #define BITSET_SET_BIT(bs, pos)     BS_ROOM(bs,pos) |= BS_BIT(pos)
 462 #define BITSET_CLEAR_BIT(bs, pos)   BS_ROOM(bs,pos) &= ~(BS_BIT(pos))
 463 #define BITSET_INVERT_BIT(bs, pos)  BS_ROOM(bs,pos) ^= BS_BIT(pos)
 464
 465 /* bytes buffer */
 466 typedef struct _BBuf {
 467   UChar* p;
 468   unsigned int used;
 469   unsigned int alloc;
 470 } BBuf;
 471
 472 #define BBUF_INIT(buf,size)    onig_bbuf_init((BBuf* )(buf), (size))
 473
 474 #define BBUF_SIZE_INC(buf,inc) do{\
 475   UChar *tmp;\
 476   (buf)->alloc += (inc);\
 477   tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
 478   if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
 479   (buf)->p = tmp;\
 480 } while (0)
 481
 482 #define BBUF_EXPAND(buf,low) do{\
 483   UChar *tmp;\
 484   do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\
 485   tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
 486   if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
 487   (buf)->p = tmp;\
 488 } while (0)
 489
 490 #define BBUF_ENSURE_SIZE(buf,size) do{\
 491   unsigned int new_alloc = (buf)->alloc;\
 492   while (new_alloc < (unsigned int )(size)) { new_alloc *= 2; }\
 493   if ((buf)->alloc != new_alloc) {\
 494     UChar *tmp;\
 495     tmp = (UChar* )xrealloc((buf)->p, new_alloc);\
 496     if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
 497     (buf)->p = tmp;\
 498     (buf)->alloc = new_alloc;\
 499   }\
 500 } while (0)
 501
 502 #define BBUF_WRITE(buf,pos,bytes,n) do{\
 503   int used = (pos) + (int )(n);\
 504   if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
 505   xmemcpy((buf)->p + (pos), (bytes), (n));\
 506   if ((buf)->used < (unsigned int )used) (buf)->used = used;\
 507 } while (0)
 508
 509 #define BBUF_WRITE1(buf,pos,byte) do{\
 510   int used = (pos) + 1;\
 511   if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
 512   (buf)->p[(pos)] = (UChar )(byte);\
 513   if ((buf)->used < (unsigned int )used) (buf)->used = used;\
 514 } while (0)
 515
 516 #define BBUF_ADD(buf,bytes,n)       BBUF_WRITE((buf),(buf)->used,(bytes),(n))
 517 #define BBUF_ADD1(buf,byte)         BBUF_WRITE1((buf),(buf)->used,(byte))
 518 #define BBUF_GET_ADD_ADDRESS(buf)   ((buf)->p + (buf)->used)
 519 #define BBUF_GET_OFFSET_POS(buf)    ((buf)->used)
 520
 521 /* from < to */
 522 #define BBUF_MOVE_RIGHT(buf,from,to,n) do {\
 523   if ((unsigned int )((to)+(n)) > (buf)->alloc) BBUF_EXPAND((buf),(to) + (n));\
 524   xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
 525   if ((unsigned int )((to)+(n)) > (buf)->used) (buf)->used = (to) + (n);\
 526 } while (0)
 527
 528 /* from > to */
 529 #define BBUF_MOVE_LEFT(buf,from,to,n) do {\
 530   xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
 531 } while (0)
 532
 533 /* from > to */
 534 #define BBUF_MOVE_LEFT_REDUCE(buf,from,to) do {\
 535   xmemmove((buf)->p + (to), (buf)->p + (from), (buf)->used - (from));\
 536   (buf)->used -= (from - to);\
 537 } while (0)
 538
 539 #define BBUF_INSERT(buf,pos,bytes,n) do {\
 540   if (pos >= (buf)->used) {\
 541     BBUF_WRITE(buf,pos,bytes,n);\
 542   }\
 543   else {\
 544     BBUF_MOVE_RIGHT((buf),(pos),(pos) + (n),((buf)->used - (pos)));\
 545     xmemcpy((buf)->p + (pos), (bytes), (n));\
 546   }\
 547 } while (0)
 548
 549 #define BBUF_GET_BYTE(buf, pos) (buf)->p[(pos)]
 550
 551
 552 #define ANCHOR_BEGIN_BUF        (1<<0)
 553 #define ANCHOR_BEGIN_LINE       (1<<1)
 554 #define ANCHOR_BEGIN_POSITION   (1<<2)
 555 #define ANCHOR_END_BUF          (1<<3)
 556 #define ANCHOR_SEMI_END_BUF     (1<<4)
 557 #define ANCHOR_END_LINE         (1<<5)
 558
 559 #define ANCHOR_WORD_BOUND       (1<<6)
 560 #define ANCHOR_NOT_WORD_BOUND   (1<<7)
 561 #define ANCHOR_WORD_BEGIN       (1<<8)
 562 #define ANCHOR_WORD_END         (1<<9)
 563 #define ANCHOR_PREC_READ        (1<<10)
 564 #define ANCHOR_PREC_READ_NOT    (1<<11)
 565 #define ANCHOR_LOOK_BEHIND      (1<<12)
 566 #define ANCHOR_LOOK_BEHIND_NOT  (1<<13)
 567
 568 #define ANCHOR_ANYCHAR_STAR     (1<<14)   /* ".*" optimize info */
 569 #define ANCHOR_ANYCHAR_STAR_ML  (1<<15)   /* ".*" optimize info (multi-line) */
 570
 571 #define ANCHOR_KEEP             (1<<16)
 572
 573 /* operation code */
 574 enum OpCode {
 575   OP_FINISH = 0,        /* matching process terminator (no more alternative) */
 576   OP_END    = 1,        /* pattern code terminator (success end) */
 577
 578   OP_EXACT1 = 2,        /* single byte, N = 1 */
 579   OP_EXACT2,            /* single byte, N = 2 */
 580   OP_EXACT3,            /* single byte, N = 3 */
 581   OP_EXACT4,            /* single byte, N = 4 */
 582   OP_EXACT5,            /* single byte, N = 5 */
 583   OP_EXACTN,            /* single byte */
 584   OP_EXACTMB2N1,        /* mb-length = 2 N = 1 */
 585   OP_EXACTMB2N2,        /* mb-length = 2 N = 2 */
 586   OP_EXACTMB2N3,        /* mb-length = 2 N = 3 */
 587   OP_EXACTMB2N,         /* mb-length = 2 */
 588   OP_EXACTMB3N,         /* mb-length = 3 */
 589   OP_EXACTMBN,          /* other length */
 590
 591   OP_EXACT1_IC,         /* single byte, N = 1, ignore case */
 592   OP_EXACTN_IC,         /* single byte,        ignore case */
 593
 594   OP_CCLASS,
 595   OP_CCLASS_MB,
 596   OP_CCLASS_MIX,
 597   OP_CCLASS_NOT,
 598   OP_CCLASS_MB_NOT,
 599   OP_CCLASS_MIX_NOT,
 600
 601   OP_ANYCHAR,                 /* "."  */
 602   OP_ANYCHAR_ML,              /* "."  multi-line */
 603   OP_ANYCHAR_STAR,            /* ".*" */
 604   OP_ANYCHAR_ML_STAR,         /* ".*" multi-line */
 605   OP_ANYCHAR_STAR_PEEK_NEXT,
 606   OP_ANYCHAR_ML_STAR_PEEK_NEXT,
 607
 608   OP_WORD,
 609   OP_NOT_WORD,
 610   OP_WORD_BOUND,
 611   OP_NOT_WORD_BOUND,
 612   OP_WORD_BEGIN,
 613   OP_WORD_END,
 614
 615   OP_ASCII_WORD,
 616   OP_NOT_ASCII_WORD,
 617   OP_ASCII_WORD_BOUND,
 618   OP_NOT_ASCII_WORD_BOUND,
 619   OP_ASCII_WORD_BEGIN,
 620   OP_ASCII_WORD_END,
 621
 622   OP_BEGIN_BUF,
 623   OP_END_BUF,
 624   OP_BEGIN_LINE,
 625   OP_END_LINE,
 626   OP_SEMI_END_BUF,
 627   OP_BEGIN_POSITION,
 628
 629   OP_BACKREF1,
 630   OP_BACKREF2,
 631   OP_BACKREFN,
 632   OP_BACKREFN_IC,
 633   OP_BACKREF_MULTI,
 634   OP_BACKREF_MULTI_IC,
 635   OP_BACKREF_WITH_LEVEL,    /* \k<xxx+n>, \k<xxx-n> */
 636
 637   OP_MEMORY_START,
 638   OP_MEMORY_START_PUSH,   /* push back-tracker to stack */
 639   OP_MEMORY_END_PUSH,     /* push back-tracker to stack */
 640   OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */
 641   OP_MEMORY_END,
 642   OP_MEMORY_END_REC,      /* push marker to stack */
 643
 644   OP_KEEP,
 645
 646   OP_FAIL,               /* pop stack and move */
 647   OP_JUMP,
 648   OP_PUSH,
 649   OP_POP,
 650   OP_PUSH_OR_JUMP_EXACT1,  /* if match exact then push, else jump. */
 651   OP_PUSH_IF_PEEK_NEXT,    /* if match exact then push, else none. */
 652   OP_REPEAT,               /* {n,m} */
 653   OP_REPEAT_NG,            /* {n,m}? (non greedy) */
 654   OP_REPEAT_INC,
 655   OP_REPEAT_INC_NG,        /* non greedy */
 656   OP_REPEAT_INC_SG,        /* search and get in stack */
 657   OP_REPEAT_INC_NG_SG,     /* search and get in stack (non greedy) */
 658   OP_NULL_CHECK_START,     /* null loop checker start */
 659   OP_NULL_CHECK_END,       /* null loop checker end   */
 660   OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */
 661   OP_NULL_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */
 662
 663   OP_PUSH_POS,             /* (?=...)  start */
 664   OP_POP_POS,              /* (?=...)  end   */
 665   OP_PUSH_POS_NOT,         /* (?!...)  start */
 666   OP_FAIL_POS,             /* (?!...)  end   */
 667   OP_PUSH_STOP_BT,         /* (?>...)  start */
 668   OP_POP_STOP_BT,          /* (?>...)  end   */
 669   OP_LOOK_BEHIND,          /* (?<=...) start (no needs end opcode) */
 670   OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */
 671   OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end   */
 672   OP_PUSH_ABSENT_POS,      /* (?~...)  start */
 673   OP_ABSENT,               /* (?~...)  start of inner loop */
 674   OP_ABSENT_END,           /* (?~...)  end   */
 675
 676   OP_CALL,                 /* \g<name> */
 677   OP_RETURN,
 678
 679   OP_CONDITION,
 680
 681   OP_STATE_CHECK_PUSH,         /* combination explosion check and push */
 682   OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump  */
 683   OP_STATE_CHECK,              /* check only */
 684   OP_STATE_CHECK_ANYCHAR_STAR,
 685   OP_STATE_CHECK_ANYCHAR_ML_STAR,
 686
 687   /* no need: IS_DYNAMIC_OPTION() == 0 */
 688   OP_SET_OPTION_PUSH,    /* set option and push recover option */
 689   OP_SET_OPTION          /* set option */
 690 };
 691
 692 typedef int RelAddrType;
 693 typedef int AbsAddrType;
 694 typedef int LengthType;
 695 typedef int RepeatNumType;
 696 typedef short int MemNumType;
 697 typedef short int StateCheckNumType;
 698 typedef void* PointerType;
 699
 700 #define SIZE_OPCODE           1
 701 #define SIZE_RELADDR          (int )sizeof(RelAddrType)
 702 #define SIZE_ABSADDR          (int )sizeof(AbsAddrType)
 703 #define SIZE_LENGTH           (int )sizeof(LengthType)
 704 #define SIZE_MEMNUM           (int )sizeof(MemNumType)
 705 #define SIZE_STATE_CHECK_NUM  (int )sizeof(StateCheckNumType)
 706 #define SIZE_REPEATNUM        (int )sizeof(RepeatNumType)
 707 #define SIZE_OPTION           (int )sizeof(OnigOptionType)
 708 #define SIZE_CODE_POINT       (int )sizeof(OnigCodePoint)
 709 #define SIZE_POINTER          (int )sizeof(PointerType)
 710
 711
 712 #define GET_RELADDR_INC(addr,p)    PLATFORM_GET_INC(addr,   p, RelAddrType)
 713 #define GET_ABSADDR_INC(addr,p)    PLATFORM_GET_INC(addr,   p, AbsAddrType)
 714 #define GET_LENGTH_INC(len,p)      PLATFORM_GET_INC(len,    p, LengthType)
 715 #define GET_MEMNUM_INC(num,p)      PLATFORM_GET_INC(num,    p, MemNumType)
 716 #define GET_REPEATNUM_INC(num,p)   PLATFORM_GET_INC(num,    p, RepeatNumType)
 717 #define GET_OPTION_INC(option,p)   PLATFORM_GET_INC(option, p, OnigOptionType)
 718 #define GET_POINTER_INC(ptr,p)     PLATFORM_GET_INC(ptr,    p, PointerType)
 719 #define GET_STATE_CHECK_NUM_INC(num,p)  PLATFORM_GET_INC(num, p, StateCheckNumType)
 720
 721 /* code point's address must be aligned address. */
 722 #define GET_CODE_POINT(code,p)   code = *((OnigCodePoint* )(p))
 723 #define GET_BYTE_INC(byte,p) do{\
 724   byte = *(p);\
 725   (p)++;\
 726 } while(0)
 727
 728
 729 /* op-code + arg size */
 730 #define SIZE_OP_ANYCHAR_STAR            SIZE_OPCODE
 731 #define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT (SIZE_OPCODE + 1)
 732 #define SIZE_OP_JUMP                   (SIZE_OPCODE + SIZE_RELADDR)
 733 #define SIZE_OP_PUSH                   (SIZE_OPCODE + SIZE_RELADDR)
 734 #define SIZE_OP_POP                     SIZE_OPCODE
 735 #define SIZE_OP_PUSH_OR_JUMP_EXACT1    (SIZE_OPCODE + SIZE_RELADDR + 1)
 736 #define SIZE_OP_PUSH_IF_PEEK_NEXT      (SIZE_OPCODE + SIZE_RELADDR + 1)
 737 #define SIZE_OP_REPEAT_INC             (SIZE_OPCODE + SIZE_MEMNUM)
 738 #define SIZE_OP_REPEAT_INC_NG          (SIZE_OPCODE + SIZE_MEMNUM)
 739 #define SIZE_OP_PUSH_POS                SIZE_OPCODE
 740 #define SIZE_OP_PUSH_POS_NOT           (SIZE_OPCODE + SIZE_RELADDR)
 741 #define SIZE_OP_POP_POS                 SIZE_OPCODE
 742 #define SIZE_OP_FAIL_POS                SIZE_OPCODE
 743 #define SIZE_OP_SET_OPTION             (SIZE_OPCODE + SIZE_OPTION)
 744 #define SIZE_OP_SET_OPTION_PUSH        (SIZE_OPCODE + SIZE_OPTION)
 745 #define SIZE_OP_FAIL                    SIZE_OPCODE
 746 #define SIZE_OP_MEMORY_START           (SIZE_OPCODE + SIZE_MEMNUM)
 747 #define SIZE_OP_MEMORY_START_PUSH      (SIZE_OPCODE + SIZE_MEMNUM)
 748 #define SIZE_OP_MEMORY_END_PUSH        (SIZE_OPCODE + SIZE_MEMNUM)
 749 #define SIZE_OP_MEMORY_END_PUSH_REC    (SIZE_OPCODE + SIZE_MEMNUM)
 750 #define SIZE_OP_MEMORY_END             (SIZE_OPCODE + SIZE_MEMNUM)
 751 #define SIZE_OP_MEMORY_END_REC         (SIZE_OPCODE + SIZE_MEMNUM)
 752 #define SIZE_OP_PUSH_STOP_BT            SIZE_OPCODE
 753 #define SIZE_OP_POP_STOP_BT             SIZE_OPCODE
 754 #define SIZE_OP_NULL_CHECK_START       (SIZE_OPCODE + SIZE_MEMNUM)
 755 #define SIZE_OP_NULL_CHECK_END         (SIZE_OPCODE + SIZE_MEMNUM)
 756 #define SIZE_OP_LOOK_BEHIND            (SIZE_OPCODE + SIZE_LENGTH)
 757 #define SIZE_OP_PUSH_LOOK_BEHIND_NOT   (SIZE_OPCODE + SIZE_RELADDR + SIZE_LENGTH)
 758 #define SIZE_OP_FAIL_LOOK_BEHIND_NOT    SIZE_OPCODE
 759 #define SIZE_OP_CALL                   (SIZE_OPCODE + SIZE_ABSADDR)
 760 #define SIZE_OP_RETURN                  SIZE_OPCODE
 761 #define SIZE_OP_CONDITION              (SIZE_OPCODE + SIZE_MEMNUM + SIZE_RELADDR)
 762 #define SIZE_OP_PUSH_ABSENT_POS         SIZE_OPCODE
 763 #define SIZE_OP_ABSENT                 (SIZE_OPCODE + SIZE_RELADDR)
 764 #define SIZE_OP_ABSENT_END              SIZE_OPCODE
 765
 766 #ifdef USE_COMBINATION_EXPLOSION_CHECK
 767 # define SIZE_OP_STATE_CHECK           (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
 768 # define SIZE_OP_STATE_CHECK_PUSH      (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
 769 # define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
 770 # define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
 771 #endif
 772
 773 #define MC_ESC(syn)               (syn)->meta_char_table.esc
 774 #define MC_ANYCHAR(syn)           (syn)->meta_char_table.anychar
 775 #define MC_ANYTIME(syn)           (syn)->meta_char_table.anytime
 776 #define MC_ZERO_OR_ONE_TIME(syn)  (syn)->meta_char_table.zero_or_one_time
 777 #define MC_ONE_OR_MORE_TIME(syn)  (syn)->meta_char_table.one_or_more_time
 778 #define MC_ANYCHAR_ANYTIME(syn)   (syn)->meta_char_table.anychar_anytime
 779
 780 #define IS_MC_ESC_CODE(code, syn) \
 781   ((code) == MC_ESC(syn) && \
 782    !IS_SYNTAX_OP2((syn), ONIG_SYN_OP2_INEFFECTIVE_ESCAPE))
 783
 784
 785 #define SYN_POSIX_COMMON_OP \
 786  ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_POSIX_BRACKET | \
 787    ONIG_SYN_OP_DECIMAL_BACKREF | \
 788    ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_ASTERISK_ZERO_INF | \
 789    ONIG_SYN_OP_LINE_ANCHOR | \
 790    ONIG_SYN_OP_ESC_CONTROL_CHARS )
 791
 792 #define SYN_GNU_REGEX_OP \
 793   ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | \
 794     ONIG_SYN_OP_POSIX_BRACKET | ONIG_SYN_OP_DECIMAL_BACKREF | \
 795     ONIG_SYN_OP_BRACE_INTERVAL | ONIG_SYN_OP_LPAREN_SUBEXP | \
 796     ONIG_SYN_OP_VBAR_ALT | \
 797     ONIG_SYN_OP_ASTERISK_ZERO_INF | ONIG_SYN_OP_PLUS_ONE_INF | \
 798     ONIG_SYN_OP_QMARK_ZERO_ONE | \
 799     ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR | ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR | \
 800     ONIG_SYN_OP_ESC_W_WORD | \
 801     ONIG_SYN_OP_ESC_B_WORD_BOUND | ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END | \
 802     ONIG_SYN_OP_ESC_S_WHITE_SPACE | ONIG_SYN_OP_ESC_D_DIGIT | \
 803     ONIG_SYN_OP_LINE_ANCHOR )
 804
 805 #define SYN_GNU_REGEX_BV \
 806   ( ONIG_SYN_CONTEXT_INDEP_ANCHORS | ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | \
 807     ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | ONIG_SYN_ALLOW_INVALID_INTERVAL | \
 808     ONIG_SYN_BACKSLASH_ESCAPE_IN_CC | ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC )
 809
 810
 811 #define NCCLASS_FLAGS(cc)           ((cc)->flags)
 812 #define NCCLASS_FLAG_SET(cc,flag)    (NCCLASS_FLAGS(cc) |= (flag))
 813 #define NCCLASS_FLAG_CLEAR(cc,flag)  (NCCLASS_FLAGS(cc) &= ~(flag))
 814 #define IS_NCCLASS_FLAG_ON(cc,flag) ((NCCLASS_FLAGS(cc) & (flag)) != 0)
 815
 816 /* cclass node */
 817 #define FLAG_NCCLASS_NOT           (1<<0)
 818
 819 #define NCCLASS_SET_NOT(nd)     NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_NOT)
 820 #define NCCLASS_CLEAR_NOT(nd)   NCCLASS_FLAG_CLEAR(nd, FLAG_NCCLASS_NOT)
 821 #define IS_NCCLASS_NOT(nd)      IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_NOT)
 822
 823 typedef struct {
 824   int type;
 825   /* struct _Node* next; */
 826   /* unsigned int flags; */
 827 } NodeBase;
 828
 829 typedef struct {
 830   NodeBase base;
 831   unsigned int flags;
 832   BitSet bs;
 833   BBuf*  mbuf;   /* multi-byte info or NULL */
 834 } CClassNode;
 835
 836 typedef intptr_t OnigStackIndex;
 837
 838 typedef struct _OnigStackType {
 839   unsigned int type;
 840   OnigStackIndex null_check;
 841   union {
 842     struct {
 843       UChar *pcode;      /* byte code position */
 844       UChar *pstr;       /* string position */
 845       UChar *pstr_prev;  /* previous char position of pstr */
 846 #ifdef USE_COMBINATION_EXPLOSION_CHECK
 847       unsigned int state_check;
 848 #endif
 849       UChar *pkeep;      /* keep pattern position */
 850     } state;
 851     struct {
 852       int   count;       /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */
 853       UChar *pcode;      /* byte code position (head of repeated target) */
 854       int   num;         /* repeat id */
 855     } repeat;
 856     struct {
 857       OnigStackIndex si;     /* index of stack */
 858     } repeat_inc;
 859     struct {
 860       int num;           /* memory num */
 861       UChar *pstr;       /* start/end position */
 862       /* Following information is set, if this stack type is MEM-START */
 863       OnigStackIndex start;  /* prev. info (for backtrack  "(...)*" ) */
 864       OnigStackIndex end;    /* prev. info (for backtrack  "(...)*" ) */
 865     } mem;
 866     struct {
 867       int num;           /* null check id */
 868       UChar *pstr;       /* start position */
 869     } null_check;
 870 #ifdef USE_SUBEXP_CALL
 871     struct {
 872       UChar *ret_addr;   /* byte code position */
 873       int    num;        /* null check id */
 874       UChar *pstr;       /* string position */
 875     } call_frame;
 876 #endif
 877     struct {
 878       UChar *abs_pstr;        /* absent start position */
 879       const UChar *end_pstr;  /* end position */
 880     } absent_pos;
 881 #ifdef USE_MATCH_CACHE
 882     struct {
 883       long    index;      /* index of the match cache buffer */
 884       uint8_t mask;       /* bit-mask for the match cache buffer */
 885     } match_cache_point;
 886 #endif
 887   } u;
 888 } OnigStackType;
 889
 890 #ifdef USE_MATCH_CACHE
 891 typedef struct {
 892   UChar *addr;
 893   long cache_point;
 894   int outer_repeat_mem;
 895   long num_cache_points_at_outer_repeat;
 896   long num_cache_points_in_outer_repeat;
 897   int lookaround_nesting;
 898   UChar *match_addr;
 899 } OnigCacheOpcode;
 900 #endif
 901
 902 typedef struct {
 903   void* stack_p;
 904   size_t stack_n;
 905   OnigOptionType options;
 906   OnigRegion*    region;
 907   const UChar* start;   /* search start position */
 908   const UChar* gpos;    /* global position (for \G: BEGIN_POSITION) */
 909 #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
 910   OnigPosition best_len;  /* for ONIG_OPTION_FIND_LONGEST */
 911   UChar* best_s;
 912 #endif
 913 #ifdef USE_COMBINATION_EXPLOSION_CHECK
 914   void* state_check_buff;
 915   int   state_check_buff_size;
 916 #endif
 917   int counter;
 918   /* rb_hrtime_t from hrtime.h */
 919 #ifdef MY_RUBY_BUILD_MAY_TIME_TRAVEL
 920   int128_t end_time;
 921 #else
 922   uint64_t end_time;
 923 #endif
 924 #ifdef USE_MATCH_CACHE
 925   int              match_cache_status;
 926   long             num_fails;
 927   long             num_cache_opcodes;
 928   OnigCacheOpcode* cache_opcodes;
 929   long             num_cache_points;
 930   uint8_t*         match_cache_buf;
 931 #endif
 932 } OnigMatchArg;
 933
 934 #define NUM_CACHE_OPCODES_UNINIT      1
 935 #define NUM_CACHE_OPCODES_IMPOSSIBLE -1
 936
 937 #define MATCH_CACHE_STATUS_UNINIT    1
 938 #define MATCH_CACHE_STATUS_INIT      2
 939 #define MATCH_CACHE_STATUS_DISABLED -1
 940 #define MATCH_CACHE_STATUS_ENABLED   0
 941
 942 #define IS_CODE_SB_WORD(enc,code) \
 943   (ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code))
 944
 945 typedef struct OnigEndCallListItem {
 946   struct OnigEndCallListItem* next;
 947   void (*func)(void);
 948 } OnigEndCallListItemType;
 949
 950 extern void onig_add_end_call(void (*func)(void));
 951
 952
 953 #ifdef ONIG_DEBUG
 954
 955 typedef struct {
 956   short int opcode;
 957   const char* name;
 958   short int arg_type;
 959 } OnigOpInfoType;
 960
 961 extern OnigOpInfoType OnigOpInfo[];
 962
 963
 964 extern void onig_print_compiled_byte_code(FILE* f, UChar* bp, UChar* bpend, UChar** nextp, OnigEncoding enc);
 965
 966 # ifdef ONIG_DEBUG_STATISTICS
 967 extern void onig_statistics_init(void);
 968 extern void onig_print_statistics(FILE* f);
 969 # endif
 970 #endif
 971
 972 #ifndef PRINTF_ARGS
 973 #define PRINTF_ARGS(func, fmt, vargs) func
 974 #endif
 975
 976 extern UChar* onig_error_code_to_format(OnigPosition code);
 977 PRINTF_ARGS(extern void onig_vsnprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const char *fmt, va_list args), 6, 0);
 978 PRINTF_ARGS(extern void onig_snprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const char *fmt, ...), 6, 7);
 979 extern int  onig_bbuf_init(BBuf* buf, OnigDistance size);
 980 extern int  onig_compile(regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo);
 981 #ifdef RUBY
 982 extern int  onig_compile_ruby(regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo, const char *sourcefile, int sourceline);
 983 #endif
 984 extern void onig_transfer(regex_t* to, regex_t* from);
 985 extern int  onig_is_code_in_cc(OnigEncoding enc, OnigCodePoint code, CClassNode* cc);
 986 extern int  onig_is_code_in_cc_len(int enclen, OnigCodePoint code, CClassNode* cc);
 987
 988 /* strend hash */
 989 typedef void hash_table_type;
 990 #ifdef RUBY
 991 # include "ruby/st.h"
 992 #else
 993 # include "st.h"
 994 #endif
 995 typedef st_data_t hash_data_type;
 996
 997 extern hash_table_type* onig_st_init_strend_table_with_size(st_index_t size);
 998 extern int onig_st_lookup_strend(hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type *value);
 999 extern int onig_st_insert_strend(hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type value);
1000
1001 #ifdef RUBY
1002 extern size_t onig_memsize(const regex_t *reg);
1003 extern size_t onig_region_memsize(const struct re_registers *regs);
1004 bool rb_reg_timeout_p(regex_t *reg, void *end_time);
1005 NORETURN(void rb_reg_raise_timeout(void));
1006 #endif
1007
1008 RUBY_SYMBOL_EXPORT_END
1009
1010 #endif /* ONIGMO_REGINT_H */