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 /* #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 #include "config.h" 87 #ifdef ONIG_ESCAPE_UCHAR_COLLISION 88 #undef ONIG_ESCAPE_UCHAR_COLLISION 89 #endif 90 91 #define USE_WORD_BEGIN_END /* "\<", "\>" */ 92 #define USE_CAPTURE_HISTORY 93 #define USE_VARIABLE_META_CHARS 94 #define USE_POSIX_API_REGION_OPTION 95 #define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE 96 /* #define USE_COMBINATION_EXPLOSION_CHECK */ /* (X*)* */ 97 98 /* #define USE_MULTI_THREAD_SYSTEM */ 99 #define THREAD_SYSTEM_INIT /* depend on thread system */ 100 #define THREAD_SYSTEM_END /* depend on thread system */ 101 #define THREAD_ATOMIC_START /* depend on thread system */ 102 #define THREAD_ATOMIC_END /* depend on thread system */ 103 #define THREAD_PASS /* depend on thread system */ 104 #define xmalloc malloc 105 #define xrealloc realloc 106 #define xcalloc calloc 107 #define xfree free 108 109 #define CHECK_INTERRUPT_IN_MATCH_AT 110 111 #define st_init_table onig_st_init_table 112 #define st_init_table_with_size onig_st_init_table_with_size 113 #define st_init_numtable onig_st_init_numtable 114 #define st_init_numtable_with_size onig_st_init_numtable_with_size 115 #define st_init_strtable onig_st_init_strtable 116 #define st_init_strtable_with_size onig_st_init_strtable_with_size 117 #define st_delete onig_st_delete 118 #define st_delete_safe onig_st_delete_safe 119 #define st_insert onig_st_insert 120 #define st_lookup onig_st_lookup 121 #define st_foreach onig_st_foreach 122 #define st_add_direct onig_st_add_direct 123 #define st_free_table onig_st_free_table 124 #define st_cleanup_safe onig_st_cleanup_safe 125 #define st_copy onig_st_copy 126 #define st_nothing_key_clone onig_st_nothing_key_clone 127 #define st_nothing_key_free onig_st_nothing_key_free 128 /* */ 129 #define onig_st_is_member st_is_member 130 131 #define STATE_CHECK_STRING_THRESHOLD_LEN 7 132 #define STATE_CHECK_BUFF_MAX_SIZE 0x4000 133 134 #define THREAD_PASS_LIMIT_COUNT 8 135 #define xmemset memset 136 #define xmemcpy memcpy 137 #define xmemmove memmove 138 139 #if defined(_WIN32) && !defined(__GNUC__) 140 #define xalloca _alloca 141 #define xvsnprintf _vsnprintf 142 #else 143 #define xalloca alloca 144 #define xvsnprintf vsnprintf 145 #endif 146 147 148 #if defined(USE_RECOMPILE_API) && defined(USE_MULTI_THREAD_SYSTEM) 149 #define ONIG_STATE_INC(reg) (reg)->state++ 150 #define ONIG_STATE_DEC(reg) (reg)->state-- 151 152 #define ONIG_STATE_INC_THREAD(reg) do {\ 153 THREAD_ATOMIC_START;\ 154 (reg)->state++;\ 155 THREAD_ATOMIC_END;\ 156 } while(0) 157 #define ONIG_STATE_DEC_THREAD(reg) do {\ 158 THREAD_ATOMIC_START;\ 159 (reg)->state--;\ 160 THREAD_ATOMIC_END;\ 161 } while(0) 162 #else 163 #define ONIG_STATE_INC(reg) /* Nothing */ 164 #define ONIG_STATE_DEC(reg) /* Nothing */ 165 #define ONIG_STATE_INC_THREAD(reg) /* Nothing */ 166 #define ONIG_STATE_DEC_THREAD(reg) /* Nothing */ 167 #endif /* USE_RECOMPILE_API && USE_MULTI_THREAD_SYSTEM */ 168 169 #ifdef HAVE_STDLIB_H 170 #include <stdlib.h> 171 #endif 172 173 #if defined(HAVE_ALLOCA_H) && !defined(__GNUC__) 174 #include <alloca.h> 175 #endif 176 177 #ifdef HAVE_STRING_H 178 # include <string.h> 179 #else 180 # include <strings.h> 181 #endif 182 183 #include <ctype.h> 184 #ifdef HAVE_SYS_TYPES_H 185 #ifndef __BORLANDC__ 186 #include <sys/types.h> 187 #endif 188 #endif 189 190 #ifdef __BORLANDC__ 191 #include <malloc.h> 192 #endif 193 194 #ifdef ONIG_DEBUG 195 # include <stdio.h> 196 #endif 197 198 #include "regenc.h" 199 200 #ifdef MIN 201 #undef MIN 202 #endif 203 #ifdef MAX 204 #undef MAX 205 #endif 206 #define MIN(a,b) (((a)>(b))?(b):(a)) 207 #define MAX(a,b) (((a)<(b))?(b):(a)) 208 209 #define IS_NULL(p) (((void*)(p)) == (void*)0) 210 #define IS_NOT_NULL(p) (((void*)(p)) != (void*)0) 211 #define CHECK_NULL_RETURN(p) if (IS_NULL(p)) return NULL 212 #define CHECK_NULL_RETURN_MEMERR(p) if (IS_NULL(p)) return ONIGERR_MEMORY 213 #define NULL_UCHARP ((UChar* )0) 214 215 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS 216 217 #define PLATFORM_GET_INC(val,p,type) do{\ 218 val = *(type* )p;\ 219 (p) += sizeof(type);\ 220 } while(0) 221 222 #else 223 224 #define PLATFORM_GET_INC(val,p,type) do{\ 225 xmemcpy(&val, (p), sizeof(type));\ 226 (p) += sizeof(type);\ 227 } while(0) 228 229 /* sizeof(OnigCodePoint) */ 230 #ifdef SIZEOF_SIZE_T 231 # define WORD_ALIGNMENT_SIZE SIZEOF_SIZE_T 232 #else 233 # define WORD_ALIGNMENT_SIZE SIZEOF_LONG 234 #endif 235 236 #define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\ 237 (pad_size) = WORD_ALIGNMENT_SIZE \ 238 - ((size_t)(addr) % WORD_ALIGNMENT_SIZE);\ 239 if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\ 240 } while (0) 241 242 #define ALIGNMENT_RIGHT(addr) do {\ 243 (addr) += (WORD_ALIGNMENT_SIZE - 1);\ 244 (addr) -= ((size_t)(addr) % WORD_ALIGNMENT_SIZE);\ 245 } while (0) 246 247 #endif /* PLATFORM_UNALIGNED_WORD_ACCESS */ 248 249 /* stack pop level */ 250 #define STACK_POP_LEVEL_FREE 0 251 #define STACK_POP_LEVEL_MEM_START 1 252 #define STACK_POP_LEVEL_ALL 2 253 254 /* optimize flags */ 255 #define ONIG_OPTIMIZE_NONE 0 256 #define ONIG_OPTIMIZE_EXACT 1 /* Slow Search */ 257 #define ONIG_OPTIMIZE_EXACT_BM 2 /* Boyer Moore Search */ 258 #define ONIG_OPTIMIZE_EXACT_BM_NOT_REV 3 /* BM (but not simple match) */ 259 #define ONIG_OPTIMIZE_EXACT_IC 4 /* Slow Search (ignore case) */ 260 #define ONIG_OPTIMIZE_MAP 5 /* char map */ 261 262 /* bit status */ 263 typedef unsigned int BitStatusType; 264 265 #define BIT_STATUS_BITS_NUM (sizeof(BitStatusType) * 8) 266 #define BIT_STATUS_CLEAR(stats) (stats) = 0 267 #define BIT_STATUS_ON_ALL(stats) (stats) = ~((BitStatusType )0) 268 #define BIT_STATUS_AT(stats,n) \ 269 ((n) < (int )BIT_STATUS_BITS_NUM ? ((stats) & (1 << n)) : ((stats) & 1)) 270 271 #define BIT_STATUS_ON_AT(stats,n) do {\ 272 if ((n) < (int )BIT_STATUS_BITS_NUM) \ 273 (stats) |= (1 << (n));\ 274 else\ 275 (stats) |= 1;\ 276 } while (0) 277 278 #define BIT_STATUS_ON_AT_SIMPLE(stats,n) do {\ 279 if ((n) < (int )BIT_STATUS_BITS_NUM)\ 280 (stats) |= (1 << (n));\ 281 } while (0) 282 283 284 #define INT_MAX_LIMIT ((1UL << (SIZEOF_INT * 8 - 1)) - 1) 285 286 #define DIGITVAL(code) ((code) - '0') 287 #define ODIGITVAL(code) DIGITVAL(code) 288 #define XDIGITVAL(enc,code) \ 289 (ONIGENC_IS_CODE_DIGIT(enc,code) ? DIGITVAL(code) \ 290 : (ONIGENC_IS_CODE_UPPER(enc,code) ? (code) - 'A' + 10 : (code) - 'a' + 10)) 291 292 #define IS_SINGLELINE(option) ((option) & ONIG_OPTION_SINGLELINE) 293 #define IS_MULTILINE(option) ((option) & ONIG_OPTION_MULTILINE) 294 #define IS_IGNORECASE(option) ((option) & ONIG_OPTION_IGNORECASE) 295 #define IS_EXTEND(option) ((option) & ONIG_OPTION_EXTEND) 296 #define IS_FIND_LONGEST(option) ((option) & ONIG_OPTION_FIND_LONGEST) 297 #define IS_FIND_NOT_EMPTY(option) ((option) & ONIG_OPTION_FIND_NOT_EMPTY) 298 #define IS_FIND_CONDITION(option) ((option) & \ 299 (ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY)) 300 #define IS_NOTBOL(option) ((option) & ONIG_OPTION_NOTBOL) 301 #define IS_NOTEOL(option) ((option) & ONIG_OPTION_NOTEOL) 302 #define IS_POSIX_REGION(option) ((option) & ONIG_OPTION_POSIX_REGION) 303 304 /* OP_SET_OPTION is required for these options. 305 #define IS_DYNAMIC_OPTION(option) \ 306 (((option) & (ONIG_OPTION_MULTILINE | ONIG_OPTION_IGNORECASE)) != 0) 307 */ 308 /* ignore-case and multibyte status are included in compiled code. */ 309 #define IS_DYNAMIC_OPTION(option) 0 310 311 #define DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag) \ 312 ((case_fold_flag) & ~INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) 313 314 #define REPEAT_INFINITE -1 315 #define IS_REPEAT_INFINITE(n) ((n) == REPEAT_INFINITE) 316 317 /* bitset */ 318 #define BITS_PER_BYTE 8 319 #define SINGLE_BYTE_SIZE (1 << BITS_PER_BYTE) 320 #define BITS_IN_ROOM (sizeof(Bits) * BITS_PER_BYTE) 321 #define BITSET_SIZE (SINGLE_BYTE_SIZE / BITS_IN_ROOM) 322 323 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS 324 typedef unsigned int Bits; 325 #else 326 typedef unsigned char Bits; 327 #endif 328 typedef Bits BitSet[BITSET_SIZE]; 329 typedef Bits* BitSetRef; 330 331 #define SIZE_BITSET sizeof(BitSet) 332 333 #define BITSET_CLEAR(bs) do {\ 334 int i;\ 335 for (i = 0; i < (int )BITSET_SIZE; i++) { (bs)[i] = 0; } \ 336 } while (0) 337 338 #define BS_ROOM(bs,pos) (bs)[pos / BITS_IN_ROOM] 339 #define BS_BIT(pos) (1 << (pos % BITS_IN_ROOM)) 340 341 #define BITSET_AT(bs, pos) (BS_ROOM(bs,pos) & BS_BIT(pos)) 342 #define BITSET_SET_BIT(bs, pos) BS_ROOM(bs,pos) |= BS_BIT(pos) 343 #define BITSET_CLEAR_BIT(bs, pos) BS_ROOM(bs,pos) &= ~(BS_BIT(pos)) 344 #define BITSET_INVERT_BIT(bs, pos) BS_ROOM(bs,pos) ^= BS_BIT(pos) 345 346 /* bytes buffer */ 347 typedef struct _BBuf { 348 UChar* p; 349 unsigned int used; 350 unsigned int alloc; 351 } BBuf; 352 353 #define BBUF_INIT(buf,size) onig_bbuf_init((BBuf* )(buf), (size)) 354 355 #define BBUF_SIZE_INC(buf,inc) do{\ 356 (buf)->alloc += (inc);\ 357 (buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\ 358 if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\ 359 } while (0) 360 361 #define BBUF_EXPAND(buf,low) do{\ 362 do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\ 363 (buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\ 364 if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\ 365 } while (0) 366 367 #define BBUF_ENSURE_SIZE(buf,size) do{\ 368 unsigned int new_alloc = (buf)->alloc;\ 369 while (new_alloc < (unsigned int )(size)) { new_alloc *= 2; }\ 370 if ((buf)->alloc != new_alloc) {\ 371 (buf)->p = (UChar* )xrealloc((buf)->p, new_alloc);\ 372 if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\ 373 (buf)->alloc = new_alloc;\ 374 }\ 375 } while (0) 376 377 #define BBUF_WRITE(buf,pos,bytes,n) do{\ 378 int used = (pos) + (n);\ 379 if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\ 380 xmemcpy((buf)->p + (pos), (bytes), (n));\ 381 if ((buf)->used < (unsigned int )used) (buf)->used = used;\ 382 } while (0) 383 384 #define BBUF_WRITE1(buf,pos,byte) do{\ 385 int used = (pos) + 1;\ 386 if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\ 387 (buf)->p[(pos)] = (byte);\ 388 if ((buf)->used < (unsigned int )used) (buf)->used = used;\ 389 } while (0) 390 391 #define BBUF_ADD(buf,bytes,n) BBUF_WRITE((buf),(buf)->used,(bytes),(n)) 392 #define BBUF_ADD1(buf,byte) BBUF_WRITE1((buf),(buf)->used,(byte)) 393 #define BBUF_GET_ADD_ADDRESS(buf) ((buf)->p + (buf)->used) 394 #define BBUF_GET_OFFSET_POS(buf) ((buf)->used) 395 396 /* from < to */ 397 #define BBUF_MOVE_RIGHT(buf,from,to,n) do {\ 398 if ((unsigned int )((to)+(n)) > (buf)->alloc) BBUF_EXPAND((buf),(to) + (n));\ 399 xmemmove((buf)->p + (to), (buf)->p + (from), (n));\ 400 if ((unsigned int )((to)+(n)) > (buf)->used) (buf)->used = (to) + (n);\ 401 } while (0) 402 403 /* from > to */ 404 #define BBUF_MOVE_LEFT(buf,from,to,n) do {\ 405 xmemmove((buf)->p + (to), (buf)->p + (from), (n));\ 406 } while (0) 407 408 /* from > to */ 409 #define BBUF_MOVE_LEFT_REDUCE(buf,from,to) do {\ 410 xmemmove((buf)->p + (to), (buf)->p + (from), (buf)->used - (from));\ 411 (buf)->used -= (from - to);\ 412 } while (0) 413 414 #define BBUF_INSERT(buf,pos,bytes,n) do {\ 415 if (pos >= (buf)->used) {\ 416 BBUF_WRITE(buf,pos,bytes,n);\ 417 }\ 418 else {\ 419 BBUF_MOVE_RIGHT((buf),(pos),(pos) + (n),((buf)->used - (pos)));\ 420 xmemcpy((buf)->p + (pos), (bytes), (n));\ 421 }\ 422 } while (0) 423 424 #define BBUF_GET_BYTE(buf, pos) (buf)->p[(pos)] 425 426 427 #define ANCHOR_BEGIN_BUF (1<<0) 428 #define ANCHOR_BEGIN_LINE (1<<1) 429 #define ANCHOR_BEGIN_POSITION (1<<2) 430 #define ANCHOR_END_BUF (1<<3) 431 #define ANCHOR_SEMI_END_BUF (1<<4) 432 #define ANCHOR_END_LINE (1<<5) 433 434 #define ANCHOR_WORD_BOUND (1<<6) 435 #define ANCHOR_NOT_WORD_BOUND (1<<7) 436 #define ANCHOR_WORD_BEGIN (1<<8) 437 #define ANCHOR_WORD_END (1<<9) 438 #define ANCHOR_PREC_READ (1<<10) 439 #define ANCHOR_PREC_READ_NOT (1<<11) 440 #define ANCHOR_LOOK_BEHIND (1<<12) 441 #define ANCHOR_LOOK_BEHIND_NOT (1<<13) 442 443 #define ANCHOR_ANYCHAR_STAR (1<<14) /* ".*" optimize info */ 444 #define ANCHOR_ANYCHAR_STAR_ML (1<<15) /* ".*" optimize info (multi-line) */ 445 446 /* operation code */ 447 enum OpCode { 448 OP_FINISH = 0, /* matching process terminator (no more alternative) */ 449 OP_END = 1, /* pattern code terminator (success end) */ 450 451 OP_EXACT1 = 2, /* single byte, N = 1 */ 452 OP_EXACT2, /* single byte, N = 2 */ 453 OP_EXACT3, /* single byte, N = 3 */ 454 OP_EXACT4, /* single byte, N = 4 */ 455 OP_EXACT5, /* single byte, N = 5 */ 456 OP_EXACTN, /* single byte */ 457 OP_EXACTMB2N1, /* mb-length = 2 N = 1 */ 458 OP_EXACTMB2N2, /* mb-length = 2 N = 2 */ 459 OP_EXACTMB2N3, /* mb-length = 2 N = 3 */ 460 OP_EXACTMB2N, /* mb-length = 2 */ 461 OP_EXACTMB3N, /* mb-length = 3 */ 462 OP_EXACTMBN, /* other length */ 463 464 OP_EXACT1_IC, /* single byte, N = 1, ignore case */ 465 OP_EXACTN_IC, /* single byte, ignore case */ 466 467 OP_CCLASS, 468 OP_CCLASS_MB, 469 OP_CCLASS_MIX, 470 OP_CCLASS_NOT, 471 OP_CCLASS_MB_NOT, 472 OP_CCLASS_MIX_NOT, 473 OP_CCLASS_NODE, /* pointer to CClassNode node */ 474 475 OP_ANYCHAR, /* "." */ 476 OP_ANYCHAR_ML, /* "." multi-line */ 477 OP_ANYCHAR_STAR, /* ".*" */ 478 OP_ANYCHAR_ML_STAR, /* ".*" multi-line */ 479 OP_ANYCHAR_STAR_PEEK_NEXT, 480 OP_ANYCHAR_ML_STAR_PEEK_NEXT, 481 482 OP_WORD, 483 OP_NOT_WORD, 484 OP_WORD_BOUND, 485 OP_NOT_WORD_BOUND, 486 OP_WORD_BEGIN, 487 OP_WORD_END, 488 489 OP_BEGIN_BUF, 490 OP_END_BUF, 491 OP_BEGIN_LINE, 492 OP_END_LINE, 493 OP_SEMI_END_BUF, 494 OP_BEGIN_POSITION, 495 496 OP_BACKREF1, 497 OP_BACKREF2, 498 OP_BACKREFN, 499 OP_BACKREFN_IC, 500 OP_BACKREF_MULTI, 501 OP_BACKREF_MULTI_IC, 502 OP_BACKREF_WITH_LEVEL, /* \k<xxx+n>, \k<xxx-n> */ 503 504 OP_MEMORY_START, 505 OP_MEMORY_START_PUSH, /* push back-tracker to stack */ 506 OP_MEMORY_END_PUSH, /* push back-tracker to stack */ 507 OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */ 508 OP_MEMORY_END, 509 OP_MEMORY_END_REC, /* push marker to stack */ 510 511 OP_FAIL, /* pop stack and move */ 512 OP_JUMP, 513 OP_PUSH, 514 OP_POP, 515 OP_PUSH_OR_JUMP_EXACT1, /* if match exact then push, else jump. */ 516 OP_PUSH_IF_PEEK_NEXT, /* if match exact then push, else none. */ 517 OP_REPEAT, /* {n,m} */ 518 OP_REPEAT_NG, /* {n,m}? (non greedy) */ 519 OP_REPEAT_INC, 520 OP_REPEAT_INC_NG, /* non greedy */ 521 OP_REPEAT_INC_SG, /* search and get in stack */ 522 OP_REPEAT_INC_NG_SG, /* search and get in stack (non greedy) */ 523 OP_NULL_CHECK_START, /* null loop checker start */ 524 OP_NULL_CHECK_END, /* null loop checker end */ 525 OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */ 526 OP_NULL_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */ 527 528 OP_PUSH_POS, /* (?=...) start */ 529 OP_POP_POS, /* (?=...) end */ 530 OP_PUSH_POS_NOT, /* (?!...) start */ 531 OP_FAIL_POS, /* (?!...) end */ 532 OP_PUSH_STOP_BT, /* (?>...) start */ 533 OP_POP_STOP_BT, /* (?>...) end */ 534 OP_LOOK_BEHIND, /* (?<=...) start (no needs end opcode) */ 535 OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */ 536 OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end */ 537 538 OP_CALL, /* \g<name> */ 539 OP_RETURN, 540 541 OP_STATE_CHECK_PUSH, /* combination explosion check and push */ 542 OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump */ 543 OP_STATE_CHECK, /* check only */ 544 OP_STATE_CHECK_ANYCHAR_STAR, 545 OP_STATE_CHECK_ANYCHAR_ML_STAR, 546 547 /* no need: IS_DYNAMIC_OPTION() == 0 */ 548 OP_SET_OPTION_PUSH, /* set option and push recover option */ 549 OP_SET_OPTION /* set option */ 550 }; 551 552 typedef int RelAddrType; 553 typedef int AbsAddrType; 554 typedef int LengthType; 555 typedef int RepeatNumType; 556 typedef short int MemNumType; 557 typedef short int StateCheckNumType; 558 typedef void* PointerType; 559 560 #define SIZE_OPCODE 1 561 #define SIZE_RELADDR sizeof(RelAddrType) 562 #define SIZE_ABSADDR sizeof(AbsAddrType) 563 #define SIZE_LENGTH sizeof(LengthType) 564 #define SIZE_MEMNUM sizeof(MemNumType) 565 #define SIZE_STATE_CHECK_NUM sizeof(StateCheckNumType) 566 #define SIZE_REPEATNUM sizeof(RepeatNumType) 567 #define SIZE_OPTION sizeof(OnigOptionType) 568 #define SIZE_CODE_POINT sizeof(OnigCodePoint) 569 #define SIZE_POINTER sizeof(PointerType) 570 571 572 #define GET_RELADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, RelAddrType) 573 #define GET_ABSADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, AbsAddrType) 574 #define GET_LENGTH_INC(len,p) PLATFORM_GET_INC(len, p, LengthType) 575 #define GET_MEMNUM_INC(num,p) PLATFORM_GET_INC(num, p, MemNumType) 576 #define GET_REPEATNUM_INC(num,p) PLATFORM_GET_INC(num, p, RepeatNumType) 577 #define GET_OPTION_INC(option,p) PLATFORM_GET_INC(option, p, OnigOptionType) 578 #define GET_POINTER_INC(ptr,p) PLATFORM_GET_INC(ptr, p, PointerType) 579 #define GET_STATE_CHECK_NUM_INC(num,p) PLATFORM_GET_INC(num, p, StateCheckNumType) 580 581 /* code point's address must be aligned address. */ 582 #define GET_CODE_POINT(code,p) code = *((OnigCodePoint* )(p)) 583 #define GET_BYTE_INC(byte,p) do{\ 584 byte = *(p);\ 585 (p)++;\ 586 } while(0) 587 588 589 /* op-code + arg size */ 590 #define SIZE_OP_ANYCHAR_STAR SIZE_OPCODE 591 #define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT (SIZE_OPCODE + 1) 592 #define SIZE_OP_JUMP (SIZE_OPCODE + SIZE_RELADDR) 593 #define SIZE_OP_PUSH (SIZE_OPCODE + SIZE_RELADDR) 594 #define SIZE_OP_POP SIZE_OPCODE 595 #define SIZE_OP_PUSH_OR_JUMP_EXACT1 (SIZE_OPCODE + SIZE_RELADDR + 1) 596 #define SIZE_OP_PUSH_IF_PEEK_NEXT (SIZE_OPCODE + SIZE_RELADDR + 1) 597 #define SIZE_OP_REPEAT_INC (SIZE_OPCODE + SIZE_MEMNUM) 598 #define SIZE_OP_REPEAT_INC_NG (SIZE_OPCODE + SIZE_MEMNUM) 599 #define SIZE_OP_PUSH_POS SIZE_OPCODE 600 #define SIZE_OP_PUSH_POS_NOT (SIZE_OPCODE + SIZE_RELADDR) 601 #define SIZE_OP_POP_POS SIZE_OPCODE 602 #define SIZE_OP_FAIL_POS SIZE_OPCODE 603 #define SIZE_OP_SET_OPTION (SIZE_OPCODE + SIZE_OPTION) 604 #define SIZE_OP_SET_OPTION_PUSH (SIZE_OPCODE + SIZE_OPTION) 605 #define SIZE_OP_FAIL SIZE_OPCODE 606 #define SIZE_OP_MEMORY_START (SIZE_OPCODE + SIZE_MEMNUM) 607 #define SIZE_OP_MEMORY_START_PUSH (SIZE_OPCODE + SIZE_MEMNUM) 608 #define SIZE_OP_MEMORY_END_PUSH (SIZE_OPCODE + SIZE_MEMNUM) 609 #define SIZE_OP_MEMORY_END_PUSH_REC (SIZE_OPCODE + SIZE_MEMNUM) 610 #define SIZE_OP_MEMORY_END (SIZE_OPCODE + SIZE_MEMNUM) 611 #define SIZE_OP_MEMORY_END_REC (SIZE_OPCODE + SIZE_MEMNUM) 612 #define SIZE_OP_PUSH_STOP_BT SIZE_OPCODE 613 #define SIZE_OP_POP_STOP_BT SIZE_OPCODE 614 #define SIZE_OP_NULL_CHECK_START (SIZE_OPCODE + SIZE_MEMNUM) 615 #define SIZE_OP_NULL_CHECK_END (SIZE_OPCODE + SIZE_MEMNUM) 616 #define SIZE_OP_LOOK_BEHIND (SIZE_OPCODE + SIZE_LENGTH) 617 #define SIZE_OP_PUSH_LOOK_BEHIND_NOT (SIZE_OPCODE + SIZE_RELADDR + SIZE_LENGTH) 618 #define SIZE_OP_FAIL_LOOK_BEHIND_NOT SIZE_OPCODE 619 #define SIZE_OP_CALL (SIZE_OPCODE + SIZE_ABSADDR) 620 #define SIZE_OP_RETURN SIZE_OPCODE 621 622 #ifdef USE_COMBINATION_EXPLOSION_CHECK 623 #define SIZE_OP_STATE_CHECK (SIZE_OPCODE + SIZE_STATE_CHECK_NUM) 624 #define SIZE_OP_STATE_CHECK_PUSH (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR) 625 #define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR) 626 #define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM) 627 #endif 628 629 #define MC_ESC(syn) (syn)->meta_char_table.esc 630 #define MC_ANYCHAR(syn) (syn)->meta_char_table.anychar 631 #define MC_ANYTIME(syn) (syn)->meta_char_table.anytime 632 #define MC_ZERO_OR_ONE_TIME(syn) (syn)->meta_char_table.zero_or_one_time 633 #define MC_ONE_OR_MORE_TIME(syn) (syn)->meta_char_table.one_or_more_time 634 #define MC_ANYCHAR_ANYTIME(syn) (syn)->meta_char_table.anychar_anytime 635 636 #define IS_MC_ESC_CODE(code, syn) \ 637 ((code) == MC_ESC(syn) && \ 638 !IS_SYNTAX_OP2((syn), ONIG_SYN_OP2_INEFFECTIVE_ESCAPE)) 639 640 641 #define SYN_POSIX_COMMON_OP \ 642 ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_POSIX_BRACKET | \ 643 ONIG_SYN_OP_DECIMAL_BACKREF | \ 644 ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_ASTERISK_ZERO_INF | \ 645 ONIG_SYN_OP_LINE_ANCHOR | \ 646 ONIG_SYN_OP_ESC_CONTROL_CHARS ) 647 648 #define SYN_GNU_REGEX_OP \ 649 ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | \ 650 ONIG_SYN_OP_POSIX_BRACKET | ONIG_SYN_OP_DECIMAL_BACKREF | \ 651 ONIG_SYN_OP_BRACE_INTERVAL | ONIG_SYN_OP_LPAREN_SUBEXP | \ 652 ONIG_SYN_OP_VBAR_ALT | \ 653 ONIG_SYN_OP_ASTERISK_ZERO_INF | ONIG_SYN_OP_PLUS_ONE_INF | \ 654 ONIG_SYN_OP_QMARK_ZERO_ONE | \ 655 ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR | ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR | \ 656 ONIG_SYN_OP_ESC_W_WORD | \ 657 ONIG_SYN_OP_ESC_B_WORD_BOUND | ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END | \ 658 ONIG_SYN_OP_ESC_S_WHITE_SPACE | ONIG_SYN_OP_ESC_D_DIGIT | \ 659 ONIG_SYN_OP_LINE_ANCHOR ) 660 661 #define SYN_GNU_REGEX_BV \ 662 ( ONIG_SYN_CONTEXT_INDEP_ANCHORS | ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | \ 663 ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | ONIG_SYN_ALLOW_INVALID_INTERVAL | \ 664 ONIG_SYN_BACKSLASH_ESCAPE_IN_CC | ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC ) 665 666 667 #define NCCLASS_FLAGS(cc) ((cc)->flags) 668 #define NCCLASS_FLAG_SET(cc,flag) (NCCLASS_FLAGS(cc) |= (flag)) 669 #define NCCLASS_FLAG_CLEAR(cc,flag) (NCCLASS_FLAGS(cc) &= ~(flag)) 670 #define IS_NCCLASS_FLAG_ON(cc,flag) ((NCCLASS_FLAGS(cc) & (flag)) != 0) 671 672 /* cclass node */ 673 #define FLAG_NCCLASS_NOT (1<<0) 674 #define FLAG_NCCLASS_SHARE (1<<1) 675 676 #define NCCLASS_SET_NOT(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_NOT) 677 #define NCCLASS_SET_SHARE(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_SHARE) 678 #define NCCLASS_CLEAR_NOT(nd) NCCLASS_FLAG_CLEAR(nd, FLAG_NCCLASS_NOT) 679 #define IS_NCCLASS_NOT(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_NOT) 680 #define IS_NCCLASS_SHARE(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_SHARE) 681 682 typedef struct { 683 int type; 684 /* struct _Node* next; */ 685 /* unsigned int flags; */ 686 } NodeBase; 687 688 typedef struct { 689 NodeBase base; 690 unsigned int flags; 691 BitSet bs; 692 BBuf* mbuf; /* multi-byte info or NULL */ 693 } CClassNode; 694 695 #ifdef _WIN64 696 typedef __int64 OnigStackIndex; 697 #else 698 typedef long OnigStackIndex; 699 #endif 700 701 typedef struct _OnigStackType { 702 unsigned int type; 703 union { 704 struct { 705 UChar *pcode; /* byte code position */ 706 UChar *pstr; /* string position */ 707 UChar *pstr_prev; /* previous char position of pstr */ 708 #ifdef USE_COMBINATION_EXPLOSION_CHECK 709 unsigned int state_check; 710 #endif 711 } state; 712 struct { 713 int count; /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */ 714 UChar *pcode; /* byte code position (head of repeated target) */ 715 int num; /* repeat id */ 716 } repeat; 717 struct { 718 OnigStackIndex si; /* index of stack */ 719 } repeat_inc; 720 struct { 721 int num; /* memory num */ 722 UChar *pstr; /* start/end position */ 723 /* Following information is setted, if this stack type is MEM-START */ 724 OnigStackIndex start; /* prev. info (for backtrack "(...)*" ) */ 725 OnigStackIndex end; /* prev. info (for backtrack "(...)*" ) */ 726 } mem; 727 struct { 728 int num; /* null check id */ 729 UChar *pstr; /* start position */ 730 } null_check; 731 #ifdef USE_SUBEXP_CALL 732 struct { 733 UChar *ret_addr; /* byte code position */ 734 int num; /* null check id */ 735 UChar *pstr; /* string position */ 736 } call_frame; 737 #endif 738 } u; 739 } OnigStackType; 740 741 typedef struct { 742 void* stack_p; 743 int stack_n; 744 OnigOptionType options; 745 OnigRegion* region; 746 const UChar* start; /* search start position (for \G: BEGIN_POSITION) */ 747 #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE 748 int best_len; /* for ONIG_OPTION_FIND_LONGEST */ 749 UChar* best_s; 750 #endif 751 #ifdef USE_COMBINATION_EXPLOSION_CHECK 752 void* state_check_buff; 753 int state_check_buff_size; 754 #endif 755 } OnigMatchArg; 756 757 758 #define IS_CODE_SB_WORD(enc,code) \ 759 (ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code)) 760 761 typedef struct OnigEndCallListItem { 762 struct OnigEndCallListItem* next; 763 void (*func)(void); 764 } OnigEndCallListItemType; 765 766 extern void onig_add_end_call(void (*func)(void)); 767 768 769 #ifdef ONIG_DEBUG 770 771 typedef struct { 772 short int opcode; 773 char* name; 774 short int arg_type; 775 } OnigOpInfoType; 776 777 extern OnigOpInfoType OnigOpInfo[]; 778 779 780 extern void onig_print_compiled_byte_code P_((FILE* f, UChar* bp, UChar** nextp, OnigEncoding enc)); 781 782 #ifdef ONIG_DEBUG_STATISTICS 783 extern void onig_statistics_init P_((void)); 784 extern void onig_print_statistics P_((FILE* f)); 785 #endif 786 #endif 787 788 extern UChar* onig_error_code_to_format P_((int code)); 789 extern void onig_snprintf_with_pattern PV_((UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const UChar *fmt, ...)); 790 extern int onig_bbuf_init P_((BBuf* buf, int size)); 791 extern int onig_compile P_((regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo)); 792 extern void onig_chain_reduce P_((regex_t* reg)); 793 extern void onig_chain_link_add P_((regex_t* to, regex_t* add)); 794 extern void onig_transfer P_((regex_t* to, regex_t* from)); 795 extern int onig_is_code_in_cc P_((OnigEncoding enc, OnigCodePoint code, CClassNode* cc)); 796 extern int onig_is_code_in_cc_len P_((int enclen, OnigCodePoint code, CClassNode* cc)); 797 798 /* strend hash */ 799 typedef void hash_table_type; 800 #ifdef _WIN32 801 # include <windows.h> 802 typedef ULONG_PTR hash_data_type; 803 #else 804 typedef unsigned long hash_data_type; 805 #endif 806 807 extern hash_table_type* onig_st_init_strend_table_with_size P_((int size)); 808 extern int onig_st_lookup_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type *value)); 809 extern int onig_st_insert_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type value)); 810 811 /* encoding property management */ 812 #define PROPERTY_LIST_ADD_PROP(Name, CR) \ 813 r = onigenc_property_list_add_property((UChar* )Name, CR,\ 814 &PropertyNameTable, &PropertyList, &PropertyListNum,\ 815 &PropertyListSize);\ 816 if (r != 0) goto end 817 818 #define PROPERTY_LIST_INIT_CHECK \ 819 if (PropertyInited == 0) {\ 820 int r = onigenc_property_list_init(init_property_list);\ 821 if (r != 0) return r;\ 822 } 823 824 extern int onigenc_property_list_add_property P_((UChar* name, const OnigCodePoint* prop, hash_table_type **table, const OnigCodePoint*** plist, int *pnum, int *psize)); 825 826 typedef int (*ONIGENC_INIT_PROPERTY_LIST_FUNC_TYPE)(void); 827 828 extern int onigenc_property_list_init P_((ONIGENC_INIT_PROPERTY_LIST_FUNC_TYPE)); 829 830 #endif /* REGINT_H */ 831