1 #ifndef REGINT_H 2 #define REGINT_H 3 /********************************************************************** 4 regint.h - Oniguruma (regular expression library) 5 **********************************************************************/ 6 /*- 7 * Copyright (c) 2002-2013 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__) || \ 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 /* !!! moved to regenc.h. */ /* #define USE_CRNL_AS_LINE_TERMINATOR */ 66 67 /* internal config */ 68 #define USE_OP_PUSH_OR_JUMP_EXACT 69 #define USE_QTFR_PEEK_NEXT 70 #define USE_ST_LIBRARY 71 72 #define INIT_MATCH_STACK_SIZE 160 73 #define DEFAULT_MATCH_STACK_LIMIT_SIZE 0 /* unlimited */ 74 #define DEFAULT_PARSE_DEPTH_LIMIT 4096 75 76 #if defined(__GNUC__) 77 # define ARG_UNUSED __attribute__ ((unused)) 78 #else 79 # define ARG_UNUSED 80 #endif 81 82 /* */ 83 /* escape other system UChar definition */ 84 #include "config.h" 85 #ifdef ONIG_ESCAPE_UCHAR_COLLISION 86 #undef ONIG_ESCAPE_UCHAR_COLLISION 87 #endif 88 89 #define USE_WORD_BEGIN_END /* "\<", "\>" */ 90 #define USE_CAPTURE_HISTORY 91 #define USE_VARIABLE_META_CHARS 92 #define USE_POSIX_API_REGION_OPTION 93 #define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE 94 /* #define USE_COMBINATION_EXPLOSION_CHECK */ /* (X*)* */ 95 96 #define xmalloc malloc 97 #define xrealloc realloc 98 #define xcalloc calloc 99 #define xfree free 100 101 #define CHECK_INTERRUPT_IN_MATCH_AT 102 103 #define st_init_table onig_st_init_table 104 #define st_init_table_with_size onig_st_init_table_with_size 105 #define st_init_numtable onig_st_init_numtable 106 #define st_init_numtable_with_size onig_st_init_numtable_with_size 107 #define st_init_strtable onig_st_init_strtable 108 #define st_init_strtable_with_size onig_st_init_strtable_with_size 109 #define st_delete onig_st_delete 110 #define st_delete_safe onig_st_delete_safe 111 #define st_insert onig_st_insert 112 #define st_lookup onig_st_lookup 113 #define st_foreach onig_st_foreach 114 #define st_add_direct onig_st_add_direct 115 #define st_free_table onig_st_free_table 116 #define st_cleanup_safe onig_st_cleanup_safe 117 #define st_copy onig_st_copy 118 #define st_nothing_key_clone onig_st_nothing_key_clone 119 #define st_nothing_key_free onig_st_nothing_key_free 120 /* */ 121 #define onig_st_is_member st_is_member 122 123 #define STATE_CHECK_STRING_THRESHOLD_LEN 7 124 #define STATE_CHECK_BUFF_MAX_SIZE 0x4000 125 126 #define xmemset memset 127 #define xmemcpy memcpy 128 #define xmemmove memmove 129 130 #if defined(_WIN32) && !defined(__GNUC__) 131 #define xalloca _alloca 132 #define xvsnprintf(buf,size,fmt,args) _vsnprintf_s(buf,size,_TRUNCATE,fmt,args) 133 #define xsnprintf sprintf_s 134 #define xstrcat(dest,src,size) strcat_s(dest,size,src) 135 #else 136 #define xalloca alloca 137 #define xvsnprintf vsnprintf 138 #define xsnprintf snprintf 139 #define xstrcat(dest,src,size) strcat(dest,src) 140 #endif 141 142 143 #ifdef HAVE_STDLIB_H 144 #include <stdlib.h> 145 #endif 146 147 #if defined(HAVE_ALLOCA_H) && !defined(__GNUC__) 148 #include <alloca.h> 149 #endif 150 151 #ifdef HAVE_STRING_H 152 # include <string.h> 153 #else 154 # include <strings.h> 155 #endif 156 157 #include <ctype.h> 158 #ifdef HAVE_SYS_TYPES_H 159 #ifndef __BORLANDC__ 160 #include <sys/types.h> 161 #endif 162 #endif 163 164 #ifdef __BORLANDC__ 165 #include <malloc.h> 166 #endif 167 168 #ifdef ONIG_DEBUG 169 # include <stdio.h> 170 #endif 171 172 #include "regenc.h" 173 174 #ifdef MIN 175 #undef MIN 176 #endif 177 #ifdef MAX 178 #undef MAX 179 #endif 180 #define MIN(a,b) (((a)>(b))?(b):(a)) 181 #define MAX(a,b) (((a)<(b))?(b):(a)) 182 183 #define IS_NULL(p) (((void*)(p)) == (void*)0) 184 #define IS_NOT_NULL(p) (((void*)(p)) != (void*)0) 185 #define CHECK_NULL_RETURN(p) if (IS_NULL(p)) return NULL 186 #define CHECK_NULL_RETURN_MEMERR(p) if (IS_NULL(p)) return ONIGERR_MEMORY 187 #define NULL_UCHARP ((UChar* )0) 188 189 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS 190 191 #define PLATFORM_GET_INC(val,p,type) do{\ 192 val = *(type* )p;\ 193 (p) += sizeof(type);\ 194 } while(0) 195 196 #else 197 198 #define PLATFORM_GET_INC(val,p,type) do{\ 199 xmemcpy(&val, (p), sizeof(type));\ 200 (p) += sizeof(type);\ 201 } while(0) 202 203 /* sizeof(OnigCodePoint) */ 204 #ifdef SIZEOF_SIZE_T 205 # define WORD_ALIGNMENT_SIZE SIZEOF_SIZE_T 206 #else 207 # define WORD_ALIGNMENT_SIZE SIZEOF_LONG 208 #endif 209 210 #define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\ 211 (pad_size) = WORD_ALIGNMENT_SIZE \ 212 - ((size_t)(addr) % WORD_ALIGNMENT_SIZE);\ 213 if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\ 214 } while (0) 215 216 #define ALIGNMENT_RIGHT(addr) do {\ 217 (addr) += (WORD_ALIGNMENT_SIZE - 1);\ 218 (addr) -= ((size_t)(addr) % WORD_ALIGNMENT_SIZE);\ 219 } while (0) 220 221 #endif /* PLATFORM_UNALIGNED_WORD_ACCESS */ 222 223 /* stack pop level */ 224 #define STACK_POP_LEVEL_FREE 0 225 #define STACK_POP_LEVEL_MEM_START 1 226 #define STACK_POP_LEVEL_ALL 2 227 228 /* optimize flags */ 229 #define ONIG_OPTIMIZE_NONE 0 230 #define ONIG_OPTIMIZE_EXACT 1 /* Slow Search */ 231 #define ONIG_OPTIMIZE_EXACT_BM 2 /* Boyer Moore Search */ 232 #define ONIG_OPTIMIZE_EXACT_BM_NOT_REV 3 /* BM (but not simple match) */ 233 #define ONIG_OPTIMIZE_EXACT_IC 4 /* Slow Search (ignore case) */ 234 #define ONIG_OPTIMIZE_MAP 5 /* char map */ 235 236 /* bit status */ 237 typedef unsigned int BitStatusType; 238 239 #define BIT_STATUS_BITS_NUM (sizeof(BitStatusType) * 8) 240 #define BIT_STATUS_CLEAR(stats) (stats) = 0 241 #define BIT_STATUS_ON_ALL(stats) (stats) = ~((BitStatusType )0) 242 #define BIT_STATUS_AT(stats,n) \ 243 ((n) < (int )BIT_STATUS_BITS_NUM ? ((stats) & (1 << n)) : ((stats) & 1)) 244 245 #define BIT_STATUS_ON_AT(stats,n) do {\ 246 if ((n) < (int )BIT_STATUS_BITS_NUM) \ 247 (stats) |= (1 << (n));\ 248 else\ 249 (stats) |= 1;\ 250 } while (0) 251 252 #define BIT_STATUS_ON_AT_SIMPLE(stats,n) do {\ 253 if ((n) < (int )BIT_STATUS_BITS_NUM)\ 254 (stats) |= (1 << (n));\ 255 } while (0) 256 257 258 #define INT_MAX_LIMIT ((1UL << (SIZEOF_INT * 8 - 1)) - 1) 259 260 #define DIGITVAL(code) ((code) - '0') 261 #define ODIGITVAL(code) DIGITVAL(code) 262 #define XDIGITVAL(enc,code) \ 263 (ONIGENC_IS_CODE_DIGIT(enc,code) ? DIGITVAL(code) \ 264 : (ONIGENC_IS_CODE_UPPER(enc,code) ? (code) - 'A' + 10 : (code) - 'a' + 10)) 265 266 #define IS_SINGLELINE(option) ((option) & ONIG_OPTION_SINGLELINE) 267 #define IS_MULTILINE(option) ((option) & ONIG_OPTION_MULTILINE) 268 #define IS_IGNORECASE(option) ((option) & ONIG_OPTION_IGNORECASE) 269 #define IS_EXTEND(option) ((option) & ONIG_OPTION_EXTEND) 270 #define IS_FIND_LONGEST(option) ((option) & ONIG_OPTION_FIND_LONGEST) 271 #define IS_FIND_NOT_EMPTY(option) ((option) & ONIG_OPTION_FIND_NOT_EMPTY) 272 #define IS_FIND_CONDITION(option) ((option) & \ 273 (ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY)) 274 #define IS_NOTBOL(option) ((option) & ONIG_OPTION_NOTBOL) 275 #define IS_NOTEOL(option) ((option) & ONIG_OPTION_NOTEOL) 276 #define IS_POSIX_REGION(option) ((option) & ONIG_OPTION_POSIX_REGION) 277 278 /* OP_SET_OPTION is required for these options. 279 #define IS_DYNAMIC_OPTION(option) \ 280 (((option) & (ONIG_OPTION_MULTILINE | ONIG_OPTION_IGNORECASE)) != 0) 281 */ 282 /* ignore-case and multibyte status are included in compiled code. */ 283 #define IS_DYNAMIC_OPTION(option) 0 284 285 #define DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag) \ 286 ((case_fold_flag) & ~INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) 287 288 #define REPEAT_INFINITE -1 289 #define IS_REPEAT_INFINITE(n) ((n) == REPEAT_INFINITE) 290 291 /* bitset */ 292 #define BITS_PER_BYTE 8 293 #define SINGLE_BYTE_SIZE (1 << BITS_PER_BYTE) 294 #define BITS_IN_ROOM (sizeof(Bits) * BITS_PER_BYTE) 295 #define BITSET_SIZE (SINGLE_BYTE_SIZE / BITS_IN_ROOM) 296 297 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS 298 typedef unsigned int Bits; 299 #else 300 typedef unsigned char Bits; 301 #endif 302 typedef Bits BitSet[BITSET_SIZE]; 303 typedef Bits* BitSetRef; 304 305 #define SIZE_BITSET sizeof(BitSet) 306 307 #define BITSET_CLEAR(bs) do {\ 308 int i;\ 309 for (i = 0; i < (int )BITSET_SIZE; i++) { (bs)[i] = 0; } \ 310 } while (0) 311 312 #define BS_ROOM(bs,pos) (bs)[pos / BITS_IN_ROOM] 313 #define BS_BIT(pos) (1 << (pos % BITS_IN_ROOM)) 314 315 #define BITSET_AT(bs, pos) (BS_ROOM(bs,pos) & BS_BIT(pos)) 316 #define BITSET_SET_BIT(bs, pos) BS_ROOM(bs,pos) |= BS_BIT(pos) 317 #define BITSET_CLEAR_BIT(bs, pos) BS_ROOM(bs,pos) &= ~(BS_BIT(pos)) 318 #define BITSET_INVERT_BIT(bs, pos) BS_ROOM(bs,pos) ^= BS_BIT(pos) 319 320 /* bytes buffer */ 321 typedef struct _BBuf { 322 UChar* p; 323 unsigned int used; 324 unsigned int alloc; 325 } BBuf; 326 327 #define BBUF_INIT(buf,size) onig_bbuf_init((BBuf* )(buf), (size)) 328 329 #define BBUF_SIZE_INC(buf,inc) do{\ 330 (buf)->alloc += (inc);\ 331 (buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\ 332 if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\ 333 } while (0) 334 335 #define BBUF_EXPAND(buf,low) do{\ 336 do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\ 337 (buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\ 338 if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\ 339 } while (0) 340 341 #define BBUF_ENSURE_SIZE(buf,size) do{\ 342 unsigned int new_alloc = (buf)->alloc;\ 343 while (new_alloc < (unsigned int )(size)) { new_alloc *= 2; }\ 344 if ((buf)->alloc != new_alloc) {\ 345 (buf)->p = (UChar* )xrealloc((buf)->p, new_alloc);\ 346 if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\ 347 (buf)->alloc = new_alloc;\ 348 }\ 349 } while (0) 350 351 #define BBUF_WRITE(buf,pos,bytes,n) do{\ 352 int used = (pos) + (n);\ 353 if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\ 354 xmemcpy((buf)->p + (pos), (bytes), (n));\ 355 if ((buf)->used < (unsigned int )used) (buf)->used = used;\ 356 } while (0) 357 358 #define BBUF_WRITE1(buf,pos,byte) do{\ 359 int used = (pos) + 1;\ 360 if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\ 361 (buf)->p[(pos)] = (byte);\ 362 if ((buf)->used < (unsigned int )used) (buf)->used = used;\ 363 } while (0) 364 365 #define BBUF_ADD(buf,bytes,n) BBUF_WRITE((buf),(buf)->used,(bytes),(n)) 366 #define BBUF_ADD1(buf,byte) BBUF_WRITE1((buf),(buf)->used,(byte)) 367 #define BBUF_GET_ADD_ADDRESS(buf) ((buf)->p + (buf)->used) 368 #define BBUF_GET_OFFSET_POS(buf) ((buf)->used) 369 370 /* from < to */ 371 #define BBUF_MOVE_RIGHT(buf,from,to,n) do {\ 372 if ((unsigned int )((to)+(n)) > (buf)->alloc) BBUF_EXPAND((buf),(to) + (n));\ 373 xmemmove((buf)->p + (to), (buf)->p + (from), (n));\ 374 if ((unsigned int )((to)+(n)) > (buf)->used) (buf)->used = (to) + (n);\ 375 } while (0) 376 377 /* from > to */ 378 #define BBUF_MOVE_LEFT(buf,from,to,n) do {\ 379 xmemmove((buf)->p + (to), (buf)->p + (from), (n));\ 380 } while (0) 381 382 /* from > to */ 383 #define BBUF_MOVE_LEFT_REDUCE(buf,from,to) do {\ 384 xmemmove((buf)->p + (to), (buf)->p + (from), (buf)->used - (from));\ 385 (buf)->used -= (from - to);\ 386 } while (0) 387 388 #define BBUF_INSERT(buf,pos,bytes,n) do {\ 389 if (pos >= (buf)->used) {\ 390 BBUF_WRITE(buf,pos,bytes,n);\ 391 }\ 392 else {\ 393 BBUF_MOVE_RIGHT((buf),(pos),(pos) + (n),((buf)->used - (pos)));\ 394 xmemcpy((buf)->p + (pos), (bytes), (n));\ 395 }\ 396 } while (0) 397 398 #define BBUF_GET_BYTE(buf, pos) (buf)->p[(pos)] 399 400 401 #define ANCHOR_BEGIN_BUF (1<<0) 402 #define ANCHOR_BEGIN_LINE (1<<1) 403 #define ANCHOR_BEGIN_POSITION (1<<2) 404 #define ANCHOR_END_BUF (1<<3) 405 #define ANCHOR_SEMI_END_BUF (1<<4) 406 #define ANCHOR_END_LINE (1<<5) 407 408 #define ANCHOR_WORD_BOUND (1<<6) 409 #define ANCHOR_NOT_WORD_BOUND (1<<7) 410 #define ANCHOR_WORD_BEGIN (1<<8) 411 #define ANCHOR_WORD_END (1<<9) 412 #define ANCHOR_PREC_READ (1<<10) 413 #define ANCHOR_PREC_READ_NOT (1<<11) 414 #define ANCHOR_LOOK_BEHIND (1<<12) 415 #define ANCHOR_LOOK_BEHIND_NOT (1<<13) 416 417 #define ANCHOR_ANYCHAR_STAR (1<<14) /* ".*" optimize info */ 418 #define ANCHOR_ANYCHAR_STAR_ML (1<<15) /* ".*" optimize info (multi-line) */ 419 420 /* operation code */ 421 enum OpCode { 422 OP_FINISH = 0, /* matching process terminator (no more alternative) */ 423 OP_END = 1, /* pattern code terminator (success end) */ 424 425 OP_EXACT1 = 2, /* single byte, N = 1 */ 426 OP_EXACT2, /* single byte, N = 2 */ 427 OP_EXACT3, /* single byte, N = 3 */ 428 OP_EXACT4, /* single byte, N = 4 */ 429 OP_EXACT5, /* single byte, N = 5 */ 430 OP_EXACTN, /* single byte */ 431 OP_EXACTMB2N1, /* mb-length = 2 N = 1 */ 432 OP_EXACTMB2N2, /* mb-length = 2 N = 2 */ 433 OP_EXACTMB2N3, /* mb-length = 2 N = 3 */ 434 OP_EXACTMB2N, /* mb-length = 2 */ 435 OP_EXACTMB3N, /* mb-length = 3 */ 436 OP_EXACTMBN, /* other length */ 437 438 OP_EXACT1_IC, /* single byte, N = 1, ignore case */ 439 OP_EXACTN_IC, /* single byte, ignore case */ 440 441 OP_CCLASS, 442 OP_CCLASS_MB, 443 OP_CCLASS_MIX, 444 OP_CCLASS_NOT, 445 OP_CCLASS_MB_NOT, 446 OP_CCLASS_MIX_NOT, 447 OP_CCLASS_NODE, /* pointer to CClassNode node */ 448 449 OP_ANYCHAR, /* "." */ 450 OP_ANYCHAR_ML, /* "." multi-line */ 451 OP_ANYCHAR_STAR, /* ".*" */ 452 OP_ANYCHAR_ML_STAR, /* ".*" multi-line */ 453 OP_ANYCHAR_STAR_PEEK_NEXT, 454 OP_ANYCHAR_ML_STAR_PEEK_NEXT, 455 456 OP_WORD, 457 OP_NOT_WORD, 458 OP_WORD_BOUND, 459 OP_NOT_WORD_BOUND, 460 OP_WORD_BEGIN, 461 OP_WORD_END, 462 463 OP_BEGIN_BUF, 464 OP_END_BUF, 465 OP_BEGIN_LINE, 466 OP_END_LINE, 467 OP_SEMI_END_BUF, 468 OP_BEGIN_POSITION, 469 470 OP_BACKREF1, 471 OP_BACKREF2, 472 OP_BACKREFN, 473 OP_BACKREFN_IC, 474 OP_BACKREF_MULTI, 475 OP_BACKREF_MULTI_IC, 476 OP_BACKREF_WITH_LEVEL, /* \k<xxx+n>, \k<xxx-n> */ 477 478 OP_MEMORY_START, 479 OP_MEMORY_START_PUSH, /* push back-tracker to stack */ 480 OP_MEMORY_END_PUSH, /* push back-tracker to stack */ 481 OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */ 482 OP_MEMORY_END, 483 OP_MEMORY_END_REC, /* push marker to stack */ 484 485 OP_FAIL, /* pop stack and move */ 486 OP_JUMP, 487 OP_PUSH, 488 OP_POP, 489 OP_PUSH_OR_JUMP_EXACT1, /* if match exact then push, else jump. */ 490 OP_PUSH_IF_PEEK_NEXT, /* if match exact then push, else none. */ 491 OP_REPEAT, /* {n,m} */ 492 OP_REPEAT_NG, /* {n,m}? (non greedy) */ 493 OP_REPEAT_INC, 494 OP_REPEAT_INC_NG, /* non greedy */ 495 OP_REPEAT_INC_SG, /* search and get in stack */ 496 OP_REPEAT_INC_NG_SG, /* search and get in stack (non greedy) */ 497 OP_NULL_CHECK_START, /* null loop checker start */ 498 OP_NULL_CHECK_END, /* null loop checker end */ 499 OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */ 500 OP_NULL_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */ 501 502 OP_PUSH_POS, /* (?=...) start */ 503 OP_POP_POS, /* (?=...) end */ 504 OP_PUSH_POS_NOT, /* (?!...) start */ 505 OP_FAIL_POS, /* (?!...) end */ 506 OP_PUSH_STOP_BT, /* (?>...) start */ 507 OP_POP_STOP_BT, /* (?>...) end */ 508 OP_LOOK_BEHIND, /* (?<=...) start (no needs end opcode) */ 509 OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */ 510 OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end */ 511 512 OP_CALL, /* \g<name> */ 513 OP_RETURN, 514 515 OP_STATE_CHECK_PUSH, /* combination explosion check and push */ 516 OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump */ 517 OP_STATE_CHECK, /* check only */ 518 OP_STATE_CHECK_ANYCHAR_STAR, 519 OP_STATE_CHECK_ANYCHAR_ML_STAR, 520 521 /* no need: IS_DYNAMIC_OPTION() == 0 */ 522 OP_SET_OPTION_PUSH, /* set option and push recover option */ 523 OP_SET_OPTION /* set option */ 524 }; 525 526 typedef int RelAddrType; 527 typedef int AbsAddrType; 528 typedef int LengthType; 529 typedef int RepeatNumType; 530 typedef int MemNumType; 531 typedef short int StateCheckNumType; 532 typedef void* PointerType; 533 534 #define SIZE_OPCODE 1 535 #define SIZE_RELADDR sizeof(RelAddrType) 536 #define SIZE_ABSADDR sizeof(AbsAddrType) 537 #define SIZE_LENGTH sizeof(LengthType) 538 #define SIZE_MEMNUM sizeof(MemNumType) 539 #define SIZE_STATE_CHECK_NUM sizeof(StateCheckNumType) 540 #define SIZE_REPEATNUM sizeof(RepeatNumType) 541 #define SIZE_OPTION sizeof(OnigOptionType) 542 #define SIZE_CODE_POINT sizeof(OnigCodePoint) 543 #define SIZE_POINTER sizeof(PointerType) 544 545 546 #define GET_RELADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, RelAddrType) 547 #define GET_ABSADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, AbsAddrType) 548 #define GET_LENGTH_INC(len,p) PLATFORM_GET_INC(len, p, LengthType) 549 #define GET_MEMNUM_INC(num,p) PLATFORM_GET_INC(num, p, MemNumType) 550 #define GET_REPEATNUM_INC(num,p) PLATFORM_GET_INC(num, p, RepeatNumType) 551 #define GET_OPTION_INC(option,p) PLATFORM_GET_INC(option, p, OnigOptionType) 552 #define GET_POINTER_INC(ptr,p) PLATFORM_GET_INC(ptr, p, PointerType) 553 #define GET_STATE_CHECK_NUM_INC(num,p) PLATFORM_GET_INC(num, p, StateCheckNumType) 554 555 /* code point's address must be aligned address. */ 556 #define GET_CODE_POINT(code,p) code = *((OnigCodePoint* )(p)) 557 #define GET_BYTE_INC(byte,p) do{\ 558 byte = *(p);\ 559 (p)++;\ 560 } while(0) 561 562 563 /* op-code + arg size */ 564 #define SIZE_OP_ANYCHAR_STAR SIZE_OPCODE 565 #define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT (SIZE_OPCODE + 1) 566 #define SIZE_OP_JUMP (SIZE_OPCODE + SIZE_RELADDR) 567 #define SIZE_OP_PUSH (SIZE_OPCODE + SIZE_RELADDR) 568 #define SIZE_OP_POP SIZE_OPCODE 569 #define SIZE_OP_PUSH_OR_JUMP_EXACT1 (SIZE_OPCODE + SIZE_RELADDR + 1) 570 #define SIZE_OP_PUSH_IF_PEEK_NEXT (SIZE_OPCODE + SIZE_RELADDR + 1) 571 #define SIZE_OP_REPEAT_INC (SIZE_OPCODE + SIZE_MEMNUM) 572 #define SIZE_OP_REPEAT_INC_NG (SIZE_OPCODE + SIZE_MEMNUM) 573 #define SIZE_OP_PUSH_POS SIZE_OPCODE 574 #define SIZE_OP_PUSH_POS_NOT (SIZE_OPCODE + SIZE_RELADDR) 575 #define SIZE_OP_POP_POS SIZE_OPCODE 576 #define SIZE_OP_FAIL_POS SIZE_OPCODE 577 #define SIZE_OP_SET_OPTION (SIZE_OPCODE + SIZE_OPTION) 578 #define SIZE_OP_SET_OPTION_PUSH (SIZE_OPCODE + SIZE_OPTION) 579 #define SIZE_OP_FAIL SIZE_OPCODE 580 #define SIZE_OP_MEMORY_START (SIZE_OPCODE + SIZE_MEMNUM) 581 #define SIZE_OP_MEMORY_START_PUSH (SIZE_OPCODE + SIZE_MEMNUM) 582 #define SIZE_OP_MEMORY_END_PUSH (SIZE_OPCODE + SIZE_MEMNUM) 583 #define SIZE_OP_MEMORY_END_PUSH_REC (SIZE_OPCODE + SIZE_MEMNUM) 584 #define SIZE_OP_MEMORY_END (SIZE_OPCODE + SIZE_MEMNUM) 585 #define SIZE_OP_MEMORY_END_REC (SIZE_OPCODE + SIZE_MEMNUM) 586 #define SIZE_OP_PUSH_STOP_BT SIZE_OPCODE 587 #define SIZE_OP_POP_STOP_BT SIZE_OPCODE 588 #define SIZE_OP_NULL_CHECK_START (SIZE_OPCODE + SIZE_MEMNUM) 589 #define SIZE_OP_NULL_CHECK_END (SIZE_OPCODE + SIZE_MEMNUM) 590 #define SIZE_OP_LOOK_BEHIND (SIZE_OPCODE + SIZE_LENGTH) 591 #define SIZE_OP_PUSH_LOOK_BEHIND_NOT (SIZE_OPCODE + SIZE_RELADDR + SIZE_LENGTH) 592 #define SIZE_OP_FAIL_LOOK_BEHIND_NOT SIZE_OPCODE 593 #define SIZE_OP_CALL (SIZE_OPCODE + SIZE_ABSADDR) 594 #define SIZE_OP_RETURN SIZE_OPCODE 595 596 #ifdef USE_COMBINATION_EXPLOSION_CHECK 597 #define SIZE_OP_STATE_CHECK (SIZE_OPCODE + SIZE_STATE_CHECK_NUM) 598 #define SIZE_OP_STATE_CHECK_PUSH (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR) 599 #define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR) 600 #define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM) 601 #endif 602 603 #define MC_ESC(syn) (syn)->meta_char_table.esc 604 #define MC_ANYCHAR(syn) (syn)->meta_char_table.anychar 605 #define MC_ANYTIME(syn) (syn)->meta_char_table.anytime 606 #define MC_ZERO_OR_ONE_TIME(syn) (syn)->meta_char_table.zero_or_one_time 607 #define MC_ONE_OR_MORE_TIME(syn) (syn)->meta_char_table.one_or_more_time 608 #define MC_ANYCHAR_ANYTIME(syn) (syn)->meta_char_table.anychar_anytime 609 610 #define IS_MC_ESC_CODE(code, syn) \ 611 ((code) == MC_ESC(syn) && \ 612 !IS_SYNTAX_OP2((syn), ONIG_SYN_OP2_INEFFECTIVE_ESCAPE)) 613 614 615 #define SYN_POSIX_COMMON_OP \ 616 ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_POSIX_BRACKET | \ 617 ONIG_SYN_OP_DECIMAL_BACKREF | \ 618 ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_ASTERISK_ZERO_INF | \ 619 ONIG_SYN_OP_LINE_ANCHOR | \ 620 ONIG_SYN_OP_ESC_CONTROL_CHARS ) 621 622 #define SYN_GNU_REGEX_OP \ 623 ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | \ 624 ONIG_SYN_OP_POSIX_BRACKET | ONIG_SYN_OP_DECIMAL_BACKREF | \ 625 ONIG_SYN_OP_BRACE_INTERVAL | ONIG_SYN_OP_LPAREN_SUBEXP | \ 626 ONIG_SYN_OP_VBAR_ALT | \ 627 ONIG_SYN_OP_ASTERISK_ZERO_INF | ONIG_SYN_OP_PLUS_ONE_INF | \ 628 ONIG_SYN_OP_QMARK_ZERO_ONE | \ 629 ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR | ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR | \ 630 ONIG_SYN_OP_ESC_W_WORD | \ 631 ONIG_SYN_OP_ESC_B_WORD_BOUND | ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END | \ 632 ONIG_SYN_OP_ESC_S_WHITE_SPACE | ONIG_SYN_OP_ESC_D_DIGIT | \ 633 ONIG_SYN_OP_LINE_ANCHOR ) 634 635 #define SYN_GNU_REGEX_BV \ 636 ( ONIG_SYN_CONTEXT_INDEP_ANCHORS | ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | \ 637 ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | ONIG_SYN_ALLOW_INVALID_INTERVAL | \ 638 ONIG_SYN_BACKSLASH_ESCAPE_IN_CC | ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC ) 639 640 641 #define NCCLASS_FLAGS(cc) ((cc)->flags) 642 #define NCCLASS_FLAG_SET(cc,flag) (NCCLASS_FLAGS(cc) |= (flag)) 643 #define NCCLASS_FLAG_CLEAR(cc,flag) (NCCLASS_FLAGS(cc) &= ~(flag)) 644 #define IS_NCCLASS_FLAG_ON(cc,flag) ((NCCLASS_FLAGS(cc) & (flag)) != 0) 645 646 /* cclass node */ 647 #define FLAG_NCCLASS_NOT (1<<0) 648 #define FLAG_NCCLASS_SHARE (1<<1) 649 650 #define NCCLASS_SET_NOT(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_NOT) 651 #define NCCLASS_SET_SHARE(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_SHARE) 652 #define NCCLASS_CLEAR_NOT(nd) NCCLASS_FLAG_CLEAR(nd, FLAG_NCCLASS_NOT) 653 #define IS_NCCLASS_NOT(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_NOT) 654 #define IS_NCCLASS_SHARE(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_SHARE) 655 656 typedef struct { 657 int type; 658 /* struct _Node* next; */ 659 /* unsigned int flags; */ 660 } NodeBase; 661 662 typedef struct { 663 NodeBase base; 664 unsigned int flags; 665 BitSet bs; 666 BBuf* mbuf; /* multi-byte info or NULL */ 667 } CClassNode; 668 669 #ifdef _WIN64 670 typedef __int64 OnigStackIndex; 671 #else 672 typedef long OnigStackIndex; 673 #endif 674 675 typedef struct _OnigStackType { 676 unsigned int type; 677 union { 678 struct { 679 UChar *pcode; /* byte code position */ 680 UChar *pstr; /* string position */ 681 UChar *pstr_prev; /* previous char position of pstr */ 682 #ifdef USE_COMBINATION_EXPLOSION_CHECK 683 unsigned int state_check; 684 #endif 685 } state; 686 struct { 687 int count; /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */ 688 UChar *pcode; /* byte code position (head of repeated target) */ 689 int num; /* repeat id */ 690 } repeat; 691 struct { 692 OnigStackIndex si; /* index of stack */ 693 } repeat_inc; 694 struct { 695 int num; /* memory num */ 696 UChar *pstr; /* start/end position */ 697 /* Following information is set, if this stack type is MEM-START */ 698 OnigStackIndex start; /* prev. info (for backtrack "(...)*" ) */ 699 OnigStackIndex end; /* prev. info (for backtrack "(...)*" ) */ 700 } mem; 701 struct { 702 int num; /* null check id */ 703 UChar *pstr; /* start position */ 704 } null_check; 705 #ifdef USE_SUBEXP_CALL 706 struct { 707 UChar *ret_addr; /* byte code position */ 708 int num; /* null check id */ 709 UChar *pstr; /* string position */ 710 } call_frame; 711 #endif 712 } u; 713 } OnigStackType; 714 715 typedef struct { 716 void* stack_p; 717 int stack_n; 718 OnigOptionType options; 719 OnigRegion* region; 720 int ptr_num; 721 const UChar* start; /* search start position (for \G: BEGIN_POSITION) */ 722 #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE 723 int best_len; /* for ONIG_OPTION_FIND_LONGEST */ 724 UChar* best_s; 725 #endif 726 #ifdef USE_COMBINATION_EXPLOSION_CHECK 727 void* state_check_buff; 728 int state_check_buff_size; 729 #endif 730 } OnigMatchArg; 731 732 733 #define IS_CODE_SB_WORD(enc,code) \ 734 (ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code)) 735 736 typedef struct OnigEndCallListItem { 737 struct OnigEndCallListItem* next; 738 void (*func)(void); 739 } OnigEndCallListItemType; 740 741 extern void onig_add_end_call(void (*func)(void)); 742 743 744 #ifdef ONIG_DEBUG 745 746 typedef struct { 747 short int opcode; 748 char* name; 749 short int arg_type; 750 } OnigOpInfoType; 751 752 extern OnigOpInfoType OnigOpInfo[]; 753 754 755 extern void onig_print_compiled_byte_code P_((FILE* f, UChar* bp, UChar** nextp, OnigEncoding enc)); 756 757 #ifdef ONIG_DEBUG_STATISTICS 758 extern void onig_statistics_init P_((void)); 759 extern int onig_print_statistics P_((FILE* f)); 760 #endif 761 #endif 762 763 extern void onig_warning(const char* s); 764 extern UChar* onig_error_code_to_format P_((int code)); 765 extern void onig_snprintf_with_pattern PV_((UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const UChar *fmt, ...)); 766 extern int onig_bbuf_init P_((BBuf* buf, int size)); 767 extern int onig_compile P_((regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo)); 768 extern void onig_transfer P_((regex_t* to, regex_t* from)); 769 extern int onig_is_code_in_cc P_((OnigEncoding enc, OnigCodePoint code, CClassNode* cc)); 770 extern int onig_is_code_in_cc_len P_((int enclen, OnigCodePoint code, CClassNode* cc)); 771 772 /* strend hash */ 773 typedef void hash_table_type; 774 #ifdef _WIN32 775 # include <windows.h> 776 typedef ULONG_PTR hash_data_type; 777 #else 778 typedef unsigned long hash_data_type; 779 #endif 780 781 extern hash_table_type* onig_st_init_strend_table_with_size P_((int size)); 782 extern int onig_st_lookup_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type *value)); 783 extern int onig_st_insert_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type value)); 784 785 typedef int (*ONIGENC_INIT_PROPERTY_LIST_FUNC_TYPE)(void); 786 787 #endif /* REGINT_H */ 788
