regint.h

   1 #ifndef ONIGURUMA_REGINT_H
   2 #define ONIGURUMA_REGINT_H
   3 /**********************************************************************
   4   regint.h -  Oniguruma (regular expression library)
   5 **********************************************************************/
   6 /*-
   7  * Copyright (c) 2002-2007  K.Kosako  <sndgk393 AT ybb DOT ne DOT jp>
   8  * All rights reserved.
   9  *
  10  * Redistribution and use in source and binary forms, with or without
  11  * modification, are permitted provided that the following conditions
  12  * are met:
  13  * 1. Redistributions of source code must retain the above copyright
  14  *    notice, this list of conditions and the following disclaimer.
  15  * 2. Redistributions in binary form must reproduce the above copyright
  16  *    notice, this list of conditions and the following disclaimer in the
  17  *    documentation and/or other materials provided with the distribution.
  18  *
  19  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  29  * SUCH DAMAGE.
  30  */
  31
  32 /* for debug */
  33 /* #define ONIG_DEBUG_PARSE_TREE */
  34 /* #define ONIG_DEBUG_COMPILE */
  35 /* #define ONIG_DEBUG_SEARCH */
  36 /* #define ONIG_DEBUG_MATCH */
  37 /* #define ONIG_DONT_OPTIMIZE */
  38
  39 /* for byte-code statistical data. */
  40 /* #define ONIG_DEBUG_STATISTICS */
  41
  42 #if defined(ONIG_DEBUG_PARSE_TREE) || defined(ONIG_DEBUG_MATCH) || \
  43     defined(ONIG_DEBUG_SEARCH) || defined(ONIG_DEBUG_COMPILE) || \
  44     defined(ONIG_DEBUG_STATISTICS)
  45 #ifndef ONIG_DEBUG
  46 #define ONIG_DEBUG
  47 #endif
  48 #endif
  49
  50 #if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \
  51     (defined(__ppc__) && defined(__APPLE__)) || \
  52     defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD86) || \
  53     defined(__mc68020__)
  54 #define PLATFORM_UNALIGNED_WORD_ACCESS
  55 #endif
  56
  57 /* config */
  58 /* spec. config */
  59 #define USE_NAMED_GROUP
  60 #define USE_SUBEXP_CALL
  61 #define USE_BACKREF_WITH_LEVEL        /* \k<name+n>, \k<name-n> */
  62 #define USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT  /* /(?:()|())*\2/ */
  63 #define USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE     /* /\n$/ =~ "\n" */
  64 #define USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR
  65 /* #define USE_RECOMPILE_API */
  66 /* !!! moved to regenc.h. */ /* #define USE_CRNL_AS_LINE_TERMINATOR */
  67
  68 /* internal config */
  69 #define USE_PARSE_TREE_NODE_RECYCLE
  70 #define USE_OP_PUSH_OR_JUMP_EXACT
  71 #define USE_QTFR_PEEK_NEXT
  72 #define USE_ST_LIBRARY
  73 #define USE_SHARED_CCLASS_TABLE
  74
  75 #define INIT_MATCH_STACK_SIZE                     160
  76 #define DEFAULT_MATCH_STACK_LIMIT_SIZE              0 /* unlimited */
  77
  78 #if defined(__GNUC__)
  79 #  define ARG_UNUSED  __attribute__ ((unused))
  80 #else
  81 #  define ARG_UNUSED
  82 #endif
  83
  84 /* */
  85 /* escape other system UChar definition */
  86 #ifndef RUBY_DEFINES_H
  87 #include "ruby/ruby.h"
  88 #endif
  89 #ifdef ONIG_ESCAPE_UCHAR_COLLISION
  90 #undef ONIG_ESCAPE_UCHAR_COLLISION
  91 #endif
  92 #undef USE_MATCH_RANGE_IS_COMPLETE_RANGE
  93 #undef USE_CAPTURE_HISTORY
  94 #define USE_VARIABLE_META_CHARS
  95 #define USE_WORD_BEGIN_END          /* "\<": word-begin, "\>": word-end */
  96 #define USE_POSIX_REGION_OPTION     /* needed for POSIX API support */
  97 #define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
  98 /* #define USE_COMBINATION_EXPLOSION_CHECK */     /* (X*)* */
  99 /* #define USE_MULTI_THREAD_SYSTEM */
 100 #define THREAD_SYSTEM_INIT      /* depend on thread system */
 101 #define THREAD_SYSTEM_END       /* depend on thread system */
 102 #define THREAD_ATOMIC_START     /* depend on thread system */
 103 #define THREAD_ATOMIC_END       /* depend on thread system */
 104 #define THREAD_PASS             /* depend on thread system */
 105 #ifndef xmalloc
 106 #define xmalloc     malloc
 107 #define xrealloc    realloc
 108 #define xcalloc     calloc
 109 #define xfree       free
 110 #endif
 111
 112 #ifdef RUBY
 113
 114 #include "vm_core.h"
 115 #define CHECK_INTERRUPT_IN_MATCH_AT RUBY_VM_CHECK_INTS()
 116 #define onig_st_init_table                  st_init_table
 117 #define onig_st_init_table_with_size        st_init_table_with_size
 118 #define onig_st_init_numtable               st_init_numtable
 119 #define onig_st_init_numtable_with_size     st_init_numtable_with_size
 120 #define onig_st_init_strtable               st_init_strtable
 121 #define onig_st_init_strtable_with_size     st_init_strtable_with_size
 122 #define onig_st_delete                      st_delete
 123 #define onig_st_delete_safe                 st_delete_safe
 124 #define onig_st_insert                      st_insert
 125 #define onig_st_lookup                      st_lookup
 126 #define onig_st_foreach                     st_foreach
 127 #define onig_st_add_direct                  st_add_direct
 128 #define onig_st_free_table                  st_free_table
 129 #define onig_st_cleanup_safe                st_cleanup_safe
 130 #define onig_st_copy                        st_copy
 131 #define onig_st_nothing_key_clone           st_nothing_key_clone
 132 #define onig_st_nothing_key_free            st_nothing_key_free
 133 #define onig_st_is_member                   st_is_member
 134
 135 #define USE_UPPER_CASE_TABLE
 136 #else
 137
 138 #define st_init_table                  onig_st_init_table
 139 #define st_init_table_with_size        onig_st_init_table_with_size
 140 #define st_init_numtable               onig_st_init_numtable
 141 #define st_init_numtable_with_size     onig_st_init_numtable_with_size
 142 #define st_init_strtable               onig_st_init_strtable
 143 #define st_init_strtable_with_size     onig_st_init_strtable_with_size
 144 #define st_delete                      onig_st_delete
 145 #define st_delete_safe                 onig_st_delete_safe
 146 #define st_insert                      onig_st_insert
 147 #define st_lookup                      onig_st_lookup
 148 #define st_foreach                     onig_st_foreach
 149 #define st_add_direct                  onig_st_add_direct
 150 #define st_free_table                  onig_st_free_table
 151 #define st_cleanup_safe                onig_st_cleanup_safe
 152 #define st_copy                        onig_st_copy
 153 #define st_nothing_key_clone           onig_st_nothing_key_clone
 154 #define st_nothing_key_free            onig_st_nothing_key_free
 155 /* */
 156 #define onig_st_is_member              st_is_member
 157
 158 #define CHECK_INTERRUPT_IN_MATCH_AT
 159
 160 #endif
 161
 162 #define STATE_CHECK_STRING_THRESHOLD_LEN             7
 163 #define STATE_CHECK_BUFF_MAX_SIZE               0x4000
 164
 165 #define THREAD_PASS_LIMIT_COUNT     8
 166 #define xmemset     memset
 167 #define xmemcpy     memcpy
 168 #define xmemmove    memmove
 169
 170 #if defined(_WIN32) && !defined(__GNUC__)
 171 #define xalloca     _alloca
 172 #define xvsnprintf  _vsnprintf
 173 #else
 174 #define xalloca     alloca
 175 #define xvsnprintf  vsnprintf
 176 #endif
 177
 178
 179 #if defined(USE_RECOMPILE_API) && defined(USE_MULTI_THREAD_SYSTEM)
 180 #define ONIG_STATE_INC(reg) (reg)->state++
 181 #define ONIG_STATE_DEC(reg) (reg)->state--
 182
 183 #define ONIG_STATE_INC_THREAD(reg) do {\
 184   THREAD_ATOMIC_START;\
 185   (reg)->state++;\
 186   THREAD_ATOMIC_END;\
 187 } while(0)
 188 #define ONIG_STATE_DEC_THREAD(reg) do {\
 189   THREAD_ATOMIC_START;\
 190   (reg)->state--;\
 191   THREAD_ATOMIC_END;\
 192 } while(0)
 193 #else
 194 #define ONIG_STATE_INC(reg)         /* Nothing */
 195 #define ONIG_STATE_DEC(reg)         /* Nothing */
 196 #define ONIG_STATE_INC_THREAD(reg)  /* Nothing */
 197 #define ONIG_STATE_DEC_THREAD(reg)  /* Nothing */
 198 #endif /* USE_RECOMPILE_API && USE_MULTI_THREAD_SYSTEM */
 199
 200 #ifdef HAVE_STDLIB_H
 201 #include <stdlib.h>
 202 #endif
 203
 204 #if defined(HAVE_ALLOCA_H) && (defined(_AIX) || !defined(__GNUC__))
 205 #include <alloca.h>
 206 #endif
 207
 208 #ifdef HAVE_STRING_H
 209 # include <string.h>
 210 #else
 211 # include <strings.h>
 212 #endif
 213
 214 #include <ctype.h>
 215 #ifdef HAVE_SYS_TYPES_H
 216 #ifndef __BORLANDC__
 217 #include <sys/types.h>
 218 #endif
 219 #endif
 220
 221 #ifdef __BORLANDC__
 222 #include <malloc.h>
 223 #endif
 224
 225 #ifdef ONIG_DEBUG
 226 # include <stdio.h>
 227 #endif
 228
 229 #include "regenc.h"
 230
 231 #ifdef MIN
 232 #undef MIN
 233 #endif
 234 #ifdef MAX
 235 #undef MAX
 236 #endif
 237 #define MIN(a,b) (((a)>(b))?(b):(a))
 238 #define MAX(a,b) (((a)<(b))?(b):(a))
 239
 240 #define IS_NULL(p)                    (((void*)(p)) == (void*)0)
 241 #define IS_NOT_NULL(p)                (((void*)(p)) != (void*)0)
 242 #define CHECK_NULL_RETURN(p)          if (IS_NULL(p)) return NULL
 243 #define CHECK_NULL_RETURN_MEMERR(p)   if (IS_NULL(p)) return ONIGERR_MEMORY
 244 #define NULL_UCHARP                   ((UChar* )0)
 245
 246 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
 247
 248 #define PLATFORM_GET_INC(val,p,type) do{\
 249   val  = *(type* )p;\
 250   (p) += sizeof(type);\
 251 } while(0)
 252
 253 #else
 254
 255 #define PLATFORM_GET_INC(val,p,type) do{\
 256   xmemcpy(&val, (p), sizeof(type));\
 257   (p) += sizeof(type);\
 258 } while(0)
 259
 260 /* sizeof(OnigCodePoint) */
 261 #define WORD_ALIGNMENT_SIZE     SIZEOF_LONG
 262
 263 #define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\
 264   (pad_size) = WORD_ALIGNMENT_SIZE \
 265                - ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
 266   if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\
 267 } while (0)
 268
 269 #define ALIGNMENT_RIGHT(addr) do {\
 270   (addr) += (WORD_ALIGNMENT_SIZE - 1);\
 271   (addr) -= ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
 272 } while (0)
 273
 274 #endif /* PLATFORM_UNALIGNED_WORD_ACCESS */
 275
 276 /* stack pop level */
 277 #define STACK_POP_LEVEL_FREE        0
 278 #define STACK_POP_LEVEL_MEM_START   1
 279 #define STACK_POP_LEVEL_ALL         2
 280
 281 /* optimize flags */
 282 #define ONIG_OPTIMIZE_NONE              0
 283 #define ONIG_OPTIMIZE_EXACT             1   /* Slow Search */
 284 #define ONIG_OPTIMIZE_EXACT_BM          2   /* Boyer Moore Search */
 285 #define ONIG_OPTIMIZE_EXACT_BM_NOT_REV  3   /* BM   (but not simple match) */
 286 #define ONIG_OPTIMIZE_EXACT_IC          4   /* Slow Search (ignore case) */
 287 #define ONIG_OPTIMIZE_MAP               5   /* char map */
 288
 289 /* bit status */
 290 typedef unsigned int  BitStatusType;
 291
 292 #define BIT_STATUS_BITS_NUM          (sizeof(BitStatusType) * 8)
 293 #define BIT_STATUS_CLEAR(stats)      (stats) = 0
 294 #define BIT_STATUS_ON_ALL(stats)     (stats) = ~((BitStatusType )0)
 295 #define BIT_STATUS_AT(stats,n) \
 296   ((n) < (int )BIT_STATUS_BITS_NUM  ?  ((stats) & (1 << n)) : ((stats) & 1))
 297
 298 #define BIT_STATUS_ON_AT(stats,n) do {\
 299     if ((n) < (int )BIT_STATUS_BITS_NUM)        \
 300     (stats) |= (1 << (n));\
 301   else\
 302     (stats) |= 1;\
 303 } while (0)
 304
 305 #define BIT_STATUS_ON_AT_SIMPLE(stats,n) do {\
 306     if ((n) < (int )BIT_STATUS_BITS_NUM)\
 307     (stats) |= (1 << (n));\
 308 } while (0)
 309
 310
 311 #define INT_MAX_LIMIT           ((1UL << (SIZEOF_INT * 8 - 1)) - 1)
 312
 313 #define DIGITVAL(code)    ((code) - '0')
 314 #define ODIGITVAL(code)   DIGITVAL(code)
 315 #define XDIGITVAL(enc,code) \
 316   (ONIGENC_IS_CODE_DIGIT(enc,code) ? DIGITVAL(code) \
 317    : (ONIGENC_IS_CODE_UPPER(enc,code) ? (code) - 'A' + 10 : (code) - 'a' + 10))
 318
 319 #define IS_SINGLELINE(option)     ((option) & ONIG_OPTION_SINGLELINE)
 320 #define IS_MULTILINE(option)      ((option) & ONIG_OPTION_MULTILINE)
 321 #define IS_IGNORECASE(option)     ((option) & ONIG_OPTION_IGNORECASE)
 322 #define IS_EXTEND(option)         ((option) & ONIG_OPTION_EXTEND)
 323 #define IS_FIND_LONGEST(option)   ((option) & ONIG_OPTION_FIND_LONGEST)
 324 #define IS_FIND_NOT_EMPTY(option) ((option) & ONIG_OPTION_FIND_NOT_EMPTY)
 325 #define IS_FIND_CONDITION(option) ((option) & \
 326           (ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY))
 327 #define IS_NOTBOL(option)         ((option) & ONIG_OPTION_NOTBOL)
 328 #define IS_NOTEOL(option)         ((option) & ONIG_OPTION_NOTEOL)
 329 #define IS_POSIX_REGION(option)   ((option) & ONIG_OPTION_POSIX_REGION)
 330
 331 /* OP_SET_OPTION is required for these options.
 332 #define IS_DYNAMIC_OPTION(option) \
 333   (((option) & (ONIG_OPTION_MULTILINE | ONIG_OPTION_IGNORECASE)) != 0)
 334 */
 335 /* ignore-case and multibyte status are included in compiled code. */
 336 #define IS_DYNAMIC_OPTION(option)  0
 337
 338 #define DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag) \
 339   ((case_fold_flag) & ~INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR)
 340
 341 #define REPEAT_INFINITE         -1
 342 #define IS_REPEAT_INFINITE(n)   ((n) == REPEAT_INFINITE)
 343
 344 /* bitset */
 345 #define BITS_PER_BYTE      8
 346 #define SINGLE_BYTE_SIZE   (1 << BITS_PER_BYTE)
 347 #define BITS_IN_ROOM       (sizeof(Bits) * BITS_PER_BYTE)
 348 #define BITSET_SIZE        (SINGLE_BYTE_SIZE / BITS_IN_ROOM)
 349
 350 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
 351 typedef unsigned int   Bits;
 352 #else
 353 typedef unsigned char  Bits;
 354 #endif
 355 typedef Bits           BitSet[BITSET_SIZE];
 356 typedef Bits*          BitSetRef;
 357
 358 #define SIZE_BITSET        sizeof(BitSet)
 359
 360 #define BITSET_CLEAR(bs) do {\
 361   int i;\
 362   for (i = 0; i < (int )BITSET_SIZE; i++) { (bs)[i] = 0; }      \
 363 } while (0)
 364
 365 #define BS_ROOM(bs,pos)            (bs)[pos / BITS_IN_ROOM]
 366 #define BS_BIT(pos)                (1 << (pos % BITS_IN_ROOM))
 367
 368 #define BITSET_AT(bs, pos)         (BS_ROOM(bs,pos) & BS_BIT(pos))
 369 #define BITSET_SET_BIT(bs, pos)     BS_ROOM(bs,pos) |= BS_BIT(pos)
 370 #define BITSET_CLEAR_BIT(bs, pos)   BS_ROOM(bs,pos) &= ~(BS_BIT(pos))
 371 #define BITSET_INVERT_BIT(bs, pos)  BS_ROOM(bs,pos) ^= BS_BIT(pos)
 372
 373 /* bytes buffer */
 374 typedef struct _BBuf {
 375   UChar* p;
 376   unsigned int used;
 377   unsigned int alloc;
 378 } BBuf;
 379
 380 #define BBUF_INIT(buf,size)    onig_bbuf_init((BBuf* )(buf), (size))
 381
 382 #define BBUF_SIZE_INC(buf,inc) do{\
 383   (buf)->alloc += (inc);\
 384   (buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
 385   if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\
 386 } while (0)
 387
 388 #define BBUF_EXPAND(buf,low) do{\
 389   do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\
 390   (buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
 391   if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\
 392 } while (0)
 393
 394 #define BBUF_ENSURE_SIZE(buf,size) do{\
 395   unsigned int new_alloc = (buf)->alloc;\
 396   while (new_alloc < (unsigned int )(size)) { new_alloc *= 2; }\
 397   if ((buf)->alloc != new_alloc) {\
 398     (buf)->p = (UChar* )xrealloc((buf)->p, new_alloc);\
 399     if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\
 400     (buf)->alloc = new_alloc;\
 401   }\
 402 } while (0)
 403
 404 #define BBUF_WRITE(buf,pos,bytes,n) do{\
 405   int used = (pos) + (n);\
 406   if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
 407   xmemcpy((buf)->p + (pos), (bytes), (n));\
 408   if ((buf)->used < (unsigned int )used) (buf)->used = used;\
 409 } while (0)
 410
 411 #define BBUF_WRITE1(buf,pos,byte) do{\
 412   int used = (pos) + 1;\
 413   if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
 414   (buf)->p[(pos)] = (byte);\
 415   if ((buf)->used < (unsigned int )used) (buf)->used = used;\
 416 } while (0)
 417
 418 #define BBUF_ADD(buf,bytes,n)       BBUF_WRITE((buf),(buf)->used,(bytes),(n))
 419 #define BBUF_ADD1(buf,byte)         BBUF_WRITE1((buf),(buf)->used,(byte))
 420 #define BBUF_GET_ADD_ADDRESS(buf)   ((buf)->p + (buf)->used)
 421 #define BBUF_GET_OFFSET_POS(buf)    ((buf)->used)
 422
 423 /* from < to */
 424 #define BBUF_MOVE_RIGHT(buf,from,to,n) do {\
 425   if ((unsigned int )((to)+(n)) > (buf)->alloc) BBUF_EXPAND((buf),(to) + (n));\
 426   xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
 427   if ((unsigned int )((to)+(n)) > (buf)->used) (buf)->used = (to) + (n);\
 428 } while (0)
 429
 430 /* from > to */
 431 #define BBUF_MOVE_LEFT(buf,from,to,n) do {\
 432   xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
 433 } while (0)
 434
 435 /* from > to */
 436 #define BBUF_MOVE_LEFT_REDUCE(buf,from,to) do {\
 437   xmemmove((buf)->p + (to), (buf)->p + (from), (buf)->used - (from));\
 438   (buf)->used -= (from - to);\
 439 } while (0)
 440
 441 #define BBUF_INSERT(buf,pos,bytes,n) do {\
 442   if (pos >= (buf)->used) {\
 443     BBUF_WRITE(buf,pos,bytes,n);\
 444   }\
 445   else {\
 446     BBUF_MOVE_RIGHT((buf),(pos),(pos) + (n),((buf)->used - (pos)));\
 447     xmemcpy((buf)->p + (pos), (bytes), (n));\
 448   }\
 449 } while (0)
 450
 451 #define BBUF_GET_BYTE(buf, pos) (buf)->p[(pos)]
 452
 453
 454 #define ANCHOR_BEGIN_BUF        (1<<0)
 455 #define ANCHOR_BEGIN_LINE       (1<<1)
 456 #define ANCHOR_BEGIN_POSITION   (1<<2)
 457 #define ANCHOR_END_BUF          (1<<3)
 458 #define ANCHOR_SEMI_END_BUF     (1<<4)
 459 #define ANCHOR_END_LINE         (1<<5)
 460
 461 #define ANCHOR_WORD_BOUND       (1<<6)
 462 #define ANCHOR_NOT_WORD_BOUND   (1<<7)
 463 #define ANCHOR_WORD_BEGIN       (1<<8)
 464 #define ANCHOR_WORD_END         (1<<9)
 465 #define ANCHOR_PREC_READ        (1<<10)
 466 #define ANCHOR_PREC_READ_NOT    (1<<11)
 467 #define ANCHOR_LOOK_BEHIND      (1<<12)
 468 #define ANCHOR_LOOK_BEHIND_NOT  (1<<13)
 469
 470 #define ANCHOR_ANYCHAR_STAR     (1<<14)   /* ".*" optimize info */
 471 #define ANCHOR_ANYCHAR_STAR_ML  (1<<15)   /* ".*" optimize info (multi-line) */
 472
 473 /* operation code */
 474 enum OpCode {
 475   OP_FINISH = 0,        /* matching process terminator (no more alternative) */
 476   OP_END    = 1,        /* pattern code terminator (success end) */
 477
 478   OP_EXACT1 = 2,        /* single byte, N = 1 */
 479   OP_EXACT2,            /* single byte, N = 2 */
 480   OP_EXACT3,            /* single byte, N = 3 */
 481   OP_EXACT4,            /* single byte, N = 4 */
 482   OP_EXACT5,            /* single byte, N = 5 */
 483   OP_EXACTN,            /* single byte */
 484   OP_EXACTMB2N1,        /* mb-length = 2 N = 1 */
 485   OP_EXACTMB2N2,        /* mb-length = 2 N = 2 */
 486   OP_EXACTMB2N3,        /* mb-length = 2 N = 3 */
 487   OP_EXACTMB2N,         /* mb-length = 2 */
 488   OP_EXACTMB3N,         /* mb-length = 3 */
 489   OP_EXACTMBN,          /* other length */
 490
 491   OP_EXACT1_IC,         /* single byte, N = 1, ignore case */
 492   OP_EXACTN_IC,         /* single byte,        ignore case */
 493
 494   OP_CCLASS,
 495   OP_CCLASS_MB,
 496   OP_CCLASS_MIX,
 497   OP_CCLASS_NOT,
 498   OP_CCLASS_MB_NOT,
 499   OP_CCLASS_MIX_NOT,
 500   OP_CCLASS_NODE,       /* pointer to CClassNode node */
 501
 502   OP_ANYCHAR,                 /* "."  */
 503   OP_ANYCHAR_ML,              /* "."  multi-line */
 504   OP_ANYCHAR_STAR,            /* ".*" */
 505   OP_ANYCHAR_ML_STAR,         /* ".*" multi-line */
 506   OP_ANYCHAR_STAR_PEEK_NEXT,
 507   OP_ANYCHAR_ML_STAR_PEEK_NEXT,
 508
 509   OP_WORD,
 510   OP_NOT_WORD,
 511   OP_WORD_BOUND,
 512   OP_NOT_WORD_BOUND,
 513   OP_WORD_BEGIN,
 514   OP_WORD_END,
 515
 516   OP_BEGIN_BUF,
 517   OP_END_BUF,
 518   OP_BEGIN_LINE,
 519   OP_END_LINE,
 520   OP_SEMI_END_BUF,
 521   OP_BEGIN_POSITION,
 522
 523   OP_BACKREF1,
 524   OP_BACKREF2,
 525   OP_BACKREFN,
 526   OP_BACKREFN_IC,
 527   OP_BACKREF_MULTI,
 528   OP_BACKREF_MULTI_IC,
 529   OP_BACKREF_WITH_LEVEL,    /* \k<xxx+n>, \k<xxx-n> */
 530
 531   OP_MEMORY_START,
 532   OP_MEMORY_START_PUSH,   /* push back-tracker to stack */
 533   OP_MEMORY_END_PUSH,     /* push back-tracker to stack */
 534   OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */
 535   OP_MEMORY_END,
 536   OP_MEMORY_END_REC,      /* push marker to stack */
 537
 538   OP_FAIL,               /* pop stack and move */
 539   OP_JUMP,
 540   OP_PUSH,
 541   OP_POP,
 542   OP_PUSH_OR_JUMP_EXACT1,  /* if match exact then push, else jump. */
 543   OP_PUSH_IF_PEEK_NEXT,    /* if match exact then push, else none. */
 544   OP_REPEAT,               /* {n,m} */
 545   OP_REPEAT_NG,            /* {n,m}? (non greedy) */
 546   OP_REPEAT_INC,
 547   OP_REPEAT_INC_NG,        /* non greedy */
 548   OP_REPEAT_INC_SG,        /* search and get in stack */
 549   OP_REPEAT_INC_NG_SG,     /* search and get in stack (non greedy) */
 550   OP_NULL_CHECK_START,     /* null loop checker start */
 551   OP_NULL_CHECK_END,       /* null loop checker end   */
 552   OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */
 553   OP_NULL_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */
 554
 555   OP_PUSH_POS,             /* (?=...)  start */
 556   OP_POP_POS,              /* (?=...)  end   */
 557   OP_PUSH_POS_NOT,         /* (?!...)  start */
 558   OP_FAIL_POS,             /* (?!...)  end   */
 559   OP_PUSH_STOP_BT,         /* (?>...)  start */
 560   OP_POP_STOP_BT,          /* (?>...)  end   */
 561   OP_LOOK_BEHIND,          /* (?<=...) start (no needs end opcode) */
 562   OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */
 563   OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end   */
 564
 565   OP_CALL,                 /* \g<name> */
 566   OP_RETURN,
 567
 568   OP_STATE_CHECK_PUSH,         /* combination explosion check and push */
 569   OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump  */
 570   OP_STATE_CHECK,              /* check only */
 571   OP_STATE_CHECK_ANYCHAR_STAR,
 572   OP_STATE_CHECK_ANYCHAR_ML_STAR,
 573
 574   /* no need: IS_DYNAMIC_OPTION() == 0 */
 575   OP_SET_OPTION_PUSH,    /* set option and push recover option */
 576   OP_SET_OPTION          /* set option */
 577 };
 578
 579 typedef int RelAddrType;
 580 typedef int AbsAddrType;
 581 typedef int LengthType;
 582 typedef int RepeatNumType;
 583 typedef short int MemNumType;
 584 typedef short int StateCheckNumType;
 585 typedef void* PointerType;
 586
 587 #define SIZE_OPCODE           1
 588 #define SIZE_RELADDR          sizeof(RelAddrType)
 589 #define SIZE_ABSADDR          sizeof(AbsAddrType)
 590 #define SIZE_LENGTH           sizeof(LengthType)
 591 #define SIZE_MEMNUM           sizeof(MemNumType)
 592 #define SIZE_STATE_CHECK_NUM  sizeof(StateCheckNumType)
 593 #define SIZE_REPEATNUM        sizeof(RepeatNumType)
 594 #define SIZE_OPTION           sizeof(OnigOptionType)
 595 #define SIZE_CODE_POINT       sizeof(OnigCodePoint)
 596 #define SIZE_POINTER          sizeof(PointerType)
 597
 598
 599 #define GET_RELADDR_INC(addr,p)    PLATFORM_GET_INC(addr,   p, RelAddrType)
 600 #define GET_ABSADDR_INC(addr,p)    PLATFORM_GET_INC(addr,   p, AbsAddrType)
 601 #define GET_LENGTH_INC(len,p)      PLATFORM_GET_INC(len,    p, LengthType)
 602 #define GET_MEMNUM_INC(num,p)      PLATFORM_GET_INC(num,    p, MemNumType)
 603 #define GET_REPEATNUM_INC(num,p)   PLATFORM_GET_INC(num,    p, RepeatNumType)
 604 #define GET_OPTION_INC(option,p)   PLATFORM_GET_INC(option, p, OnigOptionType)
 605 #define GET_POINTER_INC(ptr,p)     PLATFORM_GET_INC(ptr,    p, PointerType)
 606 #define GET_STATE_CHECK_NUM_INC(num,p)  PLATFORM_GET_INC(num, p, StateCheckNumType)
 607
 608 /* code point's address must be aligned address. */
 609 #define GET_CODE_POINT(code,p)   code = *((OnigCodePoint* )(p))
 610 #define GET_BYTE_INC(byte,p) do{\
 611   byte = *(p);\
 612   (p)++;\
 613 } while(0)
 614
 615
 616 /* op-code + arg size */
 617 #define SIZE_OP_ANYCHAR_STAR            SIZE_OPCODE
 618 #define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT (SIZE_OPCODE + 1)
 619 #define SIZE_OP_JUMP                   (SIZE_OPCODE + SIZE_RELADDR)
 620 #define SIZE_OP_PUSH                   (SIZE_OPCODE + SIZE_RELADDR)
 621 #define SIZE_OP_POP                     SIZE_OPCODE
 622 #define SIZE_OP_PUSH_OR_JUMP_EXACT1    (SIZE_OPCODE + SIZE_RELADDR + 1)
 623 #define SIZE_OP_PUSH_IF_PEEK_NEXT      (SIZE_OPCODE + SIZE_RELADDR + 1)
 624 #define SIZE_OP_REPEAT_INC             (SIZE_OPCODE + SIZE_MEMNUM)
 625 #define SIZE_OP_REPEAT_INC_NG          (SIZE_OPCODE + SIZE_MEMNUM)
 626 #define SIZE_OP_PUSH_POS                SIZE_OPCODE
 627 #define SIZE_OP_PUSH_POS_NOT           (SIZE_OPCODE + SIZE_RELADDR)
 628 #define SIZE_OP_POP_POS                 SIZE_OPCODE
 629 #define SIZE_OP_FAIL_POS                SIZE_OPCODE
 630 #define SIZE_OP_SET_OPTION             (SIZE_OPCODE + SIZE_OPTION)
 631 #define SIZE_OP_SET_OPTION_PUSH        (SIZE_OPCODE + SIZE_OPTION)
 632 #define SIZE_OP_FAIL                    SIZE_OPCODE
 633 #define SIZE_OP_MEMORY_START           (SIZE_OPCODE + SIZE_MEMNUM)
 634 #define SIZE_OP_MEMORY_START_PUSH      (SIZE_OPCODE + SIZE_MEMNUM)
 635 #define SIZE_OP_MEMORY_END_PUSH        (SIZE_OPCODE + SIZE_MEMNUM)
 636 #define SIZE_OP_MEMORY_END_PUSH_REC    (SIZE_OPCODE + SIZE_MEMNUM)
 637 #define SIZE_OP_MEMORY_END             (SIZE_OPCODE + SIZE_MEMNUM)
 638 #define SIZE_OP_MEMORY_END_REC         (SIZE_OPCODE + SIZE_MEMNUM)
 639 #define SIZE_OP_PUSH_STOP_BT            SIZE_OPCODE
 640 #define SIZE_OP_POP_STOP_BT             SIZE_OPCODE
 641 #define SIZE_OP_NULL_CHECK_START       (SIZE_OPCODE + SIZE_MEMNUM)
 642 #define SIZE_OP_NULL_CHECK_END         (SIZE_OPCODE + SIZE_MEMNUM)
 643 #define SIZE_OP_LOOK_BEHIND            (SIZE_OPCODE + SIZE_LENGTH)
 644 #define SIZE_OP_PUSH_LOOK_BEHIND_NOT   (SIZE_OPCODE + SIZE_RELADDR + SIZE_LENGTH)
 645 #define SIZE_OP_FAIL_LOOK_BEHIND_NOT    SIZE_OPCODE
 646 #define SIZE_OP_CALL                   (SIZE_OPCODE + SIZE_ABSADDR)
 647 #define SIZE_OP_RETURN                  SIZE_OPCODE
 648
 649 #ifdef USE_COMBINATION_EXPLOSION_CHECK
 650 #define SIZE_OP_STATE_CHECK            (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
 651 #define SIZE_OP_STATE_CHECK_PUSH       (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
 652 #define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
 653 #define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
 654 #endif
 655
 656 #define MC_ESC(syn)               (syn)->meta_char_table.esc
 657 #define MC_ANYCHAR(syn)           (syn)->meta_char_table.anychar
 658 #define MC_ANYTIME(syn)           (syn)->meta_char_table.anytime
 659 #define MC_ZERO_OR_ONE_TIME(syn)  (syn)->meta_char_table.zero_or_one_time
 660 #define MC_ONE_OR_MORE_TIME(syn)  (syn)->meta_char_table.one_or_more_time
 661 #define MC_ANYCHAR_ANYTIME(syn)   (syn)->meta_char_table.anychar_anytime
 662
 663 #define IS_MC_ESC_CODE(code, syn) \
 664   ((code) == MC_ESC(syn) && \
 665    !IS_SYNTAX_OP2((syn), ONIG_SYN_OP2_INEFFECTIVE_ESCAPE))
 666
 667
 668 #define SYN_POSIX_COMMON_OP \
 669  ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_POSIX_BRACKET | \
 670    ONIG_SYN_OP_DECIMAL_BACKREF | \
 671    ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_ASTERISK_ZERO_INF | \
 672    ONIG_SYN_OP_LINE_ANCHOR | \
 673    ONIG_SYN_OP_ESC_CONTROL_CHARS )
 674
 675 #define SYN_GNU_REGEX_OP \
 676   ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | \
 677     ONIG_SYN_OP_POSIX_BRACKET | ONIG_SYN_OP_DECIMAL_BACKREF | \
 678     ONIG_SYN_OP_BRACE_INTERVAL | ONIG_SYN_OP_LPAREN_SUBEXP | \
 679     ONIG_SYN_OP_VBAR_ALT | \
 680     ONIG_SYN_OP_ASTERISK_ZERO_INF | ONIG_SYN_OP_PLUS_ONE_INF | \
 681     ONIG_SYN_OP_QMARK_ZERO_ONE | \
 682     ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR | ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR | \
 683     ONIG_SYN_OP_ESC_W_WORD | \
 684     ONIG_SYN_OP_ESC_B_WORD_BOUND | ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END | \
 685     ONIG_SYN_OP_ESC_S_WHITE_SPACE | ONIG_SYN_OP_ESC_D_DIGIT | \
 686     ONIG_SYN_OP_LINE_ANCHOR )
 687
 688 #define SYN_GNU_REGEX_BV \
 689   ( ONIG_SYN_CONTEXT_INDEP_ANCHORS | ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | \
 690     ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | ONIG_SYN_ALLOW_INVALID_INTERVAL | \
 691     ONIG_SYN_BACKSLASH_ESCAPE_IN_CC | ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC )
 692
 693
 694 #define NCCLASS_FLAGS(cc)           ((cc)->flags)
 695 #define NCCLASS_FLAG_SET(cc,flag)    (NCCLASS_FLAGS(cc) |= (flag))
 696 #define NCCLASS_FLAG_CLEAR(cc,flag)  (NCCLASS_FLAGS(cc) &= ~(flag))
 697 #define IS_NCCLASS_FLAG_ON(cc,flag) ((NCCLASS_FLAGS(cc) & (flag)) != 0)
 698
 699 /* cclass node */
 700 #define FLAG_NCCLASS_NOT           (1<<0)
 701 #define FLAG_NCCLASS_SHARE         (1<<1)
 702
 703 #define NCCLASS_SET_NOT(nd)     NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_NOT)
 704 #define NCCLASS_SET_SHARE(nd)   NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_SHARE)
 705 #define NCCLASS_CLEAR_NOT(nd)   NCCLASS_FLAG_CLEAR(nd, FLAG_NCCLASS_NOT)
 706 #define IS_NCCLASS_NOT(nd)      IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_NOT)
 707 #define IS_NCCLASS_SHARE(nd)    IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_SHARE)
 708
 709 typedef struct {
 710   int type;
 711   /* struct _Node* next; */
 712   /* unsigned int flags; */
 713 } NodeBase;
 714
 715 typedef struct {
 716   NodeBase base;
 717   unsigned int flags;
 718   BitSet bs;
 719   BBuf*  mbuf;   /* multi-byte info or NULL */
 720 } CClassNode;
 721
 722 typedef long OnigStackIndex;
 723
 724 typedef struct _OnigStackType {
 725   unsigned int type;
 726   union {
 727     struct {
 728       UChar *pcode;      /* byte code position */
 729       UChar *pstr;       /* string position */
 730       UChar *pstr_prev;  /* previous char position of pstr */
 731 #ifdef USE_COMBINATION_EXPLOSION_CHECK
 732       unsigned int state_check;
 733 #endif
 734     } state;
 735     struct {
 736       int   count;       /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */
 737       UChar *pcode;      /* byte code position (head of repeated target) */
 738       int   num;         /* repeat id */
 739     } repeat;
 740     struct {
 741       OnigStackIndex si;     /* index of stack */
 742     } repeat_inc;
 743     struct {
 744       int num;           /* memory num */
 745       UChar *pstr;       /* start/end position */
 746       /* Following information is setted, if this stack type is MEM-START */
 747       OnigStackIndex start;  /* prev. info (for backtrack  "(...)*" ) */
 748       OnigStackIndex end;    /* prev. info (for backtrack  "(...)*" ) */
 749     } mem;
 750     struct {
 751       int num;           /* null check id */
 752       UChar *pstr;       /* start position */
 753     } null_check;
 754 #ifdef USE_SUBEXP_CALL
 755     struct {
 756       UChar *ret_addr;   /* byte code position */
 757       int    num;        /* null check id */
 758       UChar *pstr;       /* string position */
 759     } call_frame;
 760 #endif
 761   } u;
 762 } OnigStackType;
 763
 764 typedef struct {
 765   void* stack_p;
 766   int   stack_n;
 767   OnigOptionType options;
 768   OnigRegion*    region;
 769   const UChar* start;   /* search start position (for \G: BEGIN_POSITION) */
 770 #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
 771   int    best_len;      /* for ONIG_OPTION_FIND_LONGEST */
 772   UChar* best_s;
 773 #endif
 774 #ifdef USE_COMBINATION_EXPLOSION_CHECK
 775   void* state_check_buff;
 776   int   state_check_buff_size;
 777 #endif
 778 } OnigMatchArg;
 779
 780
 781 #define IS_CODE_SB_WORD(enc,code) \
 782   (ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code))
 783
 784 #ifdef ONIG_DEBUG
 785
 786 typedef struct {
 787   short int opcode;
 788   char*     name;
 789   short int arg_type;
 790 } OnigOpInfoType;
 791
 792 extern OnigOpInfoType OnigOpInfo[];
 793
 794 extern void onig_print_compiled_byte_code P_((FILE* f, UChar* bp, UChar** nextp, OnigEncoding enc));
 795
 796 #ifdef ONIG_DEBUG_STATISTICS
 797 extern void onig_statistics_init P_((void));
 798 extern void onig_print_statistics P_((FILE* f));
 799 #endif
 800 #endif
 801
 802 extern UChar* onig_error_code_to_format P_((int code));
 803 extern void  onig_snprintf_with_pattern PV_((UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const UChar *fmt, ...));
 804 extern int  onig_bbuf_init P_((BBuf* buf, int size));
 805 extern int  onig_alloc_init P_((regex_t** reg, OnigOptionType option, OnigCaseFoldType case_fold_flag, OnigEncoding enc, const OnigSyntaxType* syntax));
 806 extern int  onig_compile P_((regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo));
 807 extern void onig_chain_reduce P_((regex_t* reg));
 808 extern void onig_chain_link_add P_((regex_t* to, regex_t* add));
 809 extern void onig_transfer P_((regex_t* to, regex_t* from));
 810 extern int  onig_is_code_in_cc P_((OnigEncoding enc, OnigCodePoint code, CClassNode* cc));
 811 extern int  onig_is_code_in_cc_len P_((int enclen, OnigCodePoint code, CClassNode* cc));
 812
 813 /* strend hash */
 814 typedef void hash_table_type;
 815 #ifdef RUBY
 816 #include <ruby/st.h>
 817 typedef st_data_t hash_data_type;
 818 #else
 819 typedef unsigned long hash_data_type;
 820 #endif
 821
 822 extern hash_table_type* onig_st_init_strend_table_with_size P_((int size));
 823 extern int onig_st_lookup_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type *value));
 824 extern int onig_st_insert_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type value));
 825
 826 /* encoding property management */
 827 #define PROPERTY_LIST_ADD_PROP(Name, CR) \
 828   r = onigenc_property_list_add_property((UChar* )Name, CR,\
 829               &PropertyNameTable, &PropertyList, &PropertyListNum,\
 830               &PropertyListSize);\
 831   if (r != 0) goto end
 832
 833 #define PROPERTY_LIST_INIT_CHECK \
 834   if (PropertyInited == 0) {\
 835     int r = onigenc_property_list_init(init_property_list);\
 836     if (r != 0) return r;\
 837   }
 838
 839 extern int onigenc_property_list_add_property P_((UChar* name, const OnigCodePoint* prop, hash_table_type **table, const OnigCodePoint*** plist, int *pnum, int *psize));
 840
 841 typedef int (*ONIGENC_INIT_PROPERTY_LIST_FUNC_TYPE)(void);
 842
 843 extern int onigenc_property_list_init P_((ONIGENC_INIT_PROPERTY_LIST_FUNC_TYPE));
 844
 845 #endif /* ONIGURUMA_REGINT_H */