[279] | 1 | #ifndef REGINT_H
|
---|
| 2 | #define REGINT_H
|
---|
| 3 | /**********************************************************************
|
---|
| 4 | regint.h - Onigmo (Oniguruma-mod) (regular expression library)
|
---|
| 5 | **********************************************************************/
|
---|
| 6 | /*-
|
---|
| 7 | * Copyright (c) 2002-2013 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
|
---|
| 8 | * Copyright (c) 2011-2014 K.Takata <kentkt AT csc DOT jp>
|
---|
| 9 | * All rights reserved.
|
---|
| 10 | *
|
---|
| 11 | * Redistribution and use in source and binary forms, with or without
|
---|
| 12 | * modification, are permitted provided that the following conditions
|
---|
| 13 | * are met:
|
---|
| 14 | * 1. Redistributions of source code must retain the above copyright
|
---|
| 15 | * notice, this list of conditions and the following disclaimer.
|
---|
| 16 | * 2. Redistributions in binary form must reproduce the above copyright
|
---|
| 17 | * notice, this list of conditions and the following disclaimer in the
|
---|
| 18 | * documentation and/or other materials provided with the distribution.
|
---|
| 19 | *
|
---|
| 20 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
---|
| 21 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
---|
| 22 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
---|
| 23 | * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
---|
| 24 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
---|
| 25 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
---|
| 26 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
---|
| 27 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
---|
| 28 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
---|
| 29 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
---|
| 30 | * SUCH DAMAGE.
|
---|
| 31 | */
|
---|
| 32 |
|
---|
| 33 | /* for debug */
|
---|
| 34 | /* #define ONIG_DEBUG_PARSE_TREE */
|
---|
| 35 | /* #define ONIG_DEBUG_COMPILE */
|
---|
| 36 | /* #define ONIG_DEBUG_SEARCH */
|
---|
| 37 | /* #define ONIG_DEBUG_MATCH */
|
---|
| 38 | /* #define ONIG_DONT_OPTIMIZE */
|
---|
| 39 |
|
---|
| 40 | /* for byte-code statistical data. */
|
---|
| 41 | /* #define ONIG_DEBUG_STATISTICS */
|
---|
| 42 |
|
---|
| 43 | #if defined(ONIG_DEBUG_PARSE_TREE) || defined(ONIG_DEBUG_MATCH) || \
|
---|
| 44 | defined(ONIG_DEBUG_SEARCH) || defined(ONIG_DEBUG_COMPILE) || \
|
---|
| 45 | defined(ONIG_DEBUG_STATISTICS)
|
---|
| 46 | #ifndef ONIG_DEBUG
|
---|
| 47 | #define ONIG_DEBUG
|
---|
| 48 | #endif
|
---|
| 49 | #endif
|
---|
| 50 |
|
---|
| 51 | #ifndef UNALIGNED_WORD_ACCESS
|
---|
| 52 | #if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \
|
---|
| 53 | defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD64) || \
|
---|
| 54 | defined(__powerpc64__) || \
|
---|
| 55 | defined(__mc68020__)
|
---|
| 56 | #define UNALIGNED_WORD_ACCESS 1
|
---|
| 57 | #else
|
---|
| 58 | #define UNALIGNED_WORD_ACCESS 0
|
---|
| 59 | #endif
|
---|
| 60 | #endif
|
---|
| 61 |
|
---|
| 62 | #if UNALIGNED_WORD_ACCESS
|
---|
| 63 | #define PLATFORM_UNALIGNED_WORD_ACCESS
|
---|
| 64 | #endif
|
---|
| 65 |
|
---|
| 66 | /* config */
|
---|
| 67 | /* spec. config */
|
---|
| 68 | #define USE_NAMED_GROUP
|
---|
| 69 | #define USE_SUBEXP_CALL
|
---|
| 70 | #define USE_PERL_SUBEXP_CALL
|
---|
| 71 | #define USE_CAPITAL_P_NAMED_GROUP
|
---|
| 72 | #define USE_BACKREF_WITH_LEVEL /* \k<name+n>, \k<name-n> */
|
---|
| 73 | #define USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT /* /(?:()|())*\2/ */
|
---|
| 74 | #define USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE /* /\n$/ =~ "\n" */
|
---|
| 75 | #define USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR
|
---|
| 76 | /* #define USE_RECOMPILE_API */
|
---|
| 77 | /* !!! moved to regenc.h. */ /* #define USE_CRNL_AS_LINE_TERMINATOR */
|
---|
| 78 | #define USE_NO_INVALID_QUANTIFIER
|
---|
| 79 |
|
---|
| 80 | /* internal config */
|
---|
| 81 | #define USE_PARSE_TREE_NODE_RECYCLE
|
---|
| 82 | #define USE_OP_PUSH_OR_JUMP_EXACT
|
---|
| 83 | #define USE_QTFR_PEEK_NEXT
|
---|
| 84 | #define USE_ST_LIBRARY
|
---|
| 85 | #define USE_SHARED_CCLASS_TABLE
|
---|
| 86 | #define USE_SUNDAY_QUICK_SEARCH
|
---|
| 87 |
|
---|
| 88 | #define INIT_MATCH_STACK_SIZE 160
|
---|
| 89 | #define DEFAULT_MATCH_STACK_LIMIT_SIZE 0 /* unlimited */
|
---|
| 90 |
|
---|
| 91 | /* check config */
|
---|
| 92 | #if defined(USE_PERL_SUBEXP_CALL) || defined(USE_CAPITAL_P_NAMED_GROUP)
|
---|
| 93 | #if !defined(USE_NAMED_GROUP) || !defined(USE_SUBEXP_CALL)
|
---|
| 94 | #error USE_NAMED_GROUP and USE_SUBEXP_CALL must be defined.
|
---|
| 95 | #endif
|
---|
| 96 | #endif
|
---|
| 97 |
|
---|
| 98 | #if defined(__GNUC__)
|
---|
| 99 | # define ARG_UNUSED __attribute__ ((unused))
|
---|
| 100 | #else
|
---|
| 101 | # define ARG_UNUSED
|
---|
| 102 | #endif
|
---|
| 103 |
|
---|
| 104 | /* */
|
---|
| 105 | /* escape other system UChar definition */
|
---|
| 106 | #include "config.h"
|
---|
| 107 | #ifdef ONIG_ESCAPE_UCHAR_COLLISION
|
---|
| 108 | #undef ONIG_ESCAPE_UCHAR_COLLISION
|
---|
| 109 | #endif
|
---|
| 110 |
|
---|
| 111 | #define USE_WORD_BEGIN_END /* "\<": word-begin, "\>": word-end */
|
---|
| 112 | #define USE_CAPTURE_HISTORY
|
---|
| 113 | #define USE_VARIABLE_META_CHARS
|
---|
| 114 | #define USE_POSIX_API_REGION_OPTION /* needed for POSIX API support */
|
---|
| 115 | #define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
| 116 | /* #define USE_COMBINATION_EXPLOSION_CHECK */ /* (X*)* */
|
---|
| 117 |
|
---|
| 118 | /* multithread config */
|
---|
| 119 | /* #define USE_MULTI_THREAD_SYSTEM */
|
---|
| 120 | /* #define USE_DEFAULT_MULTI_THREAD_SYSTEM */
|
---|
| 121 |
|
---|
| 122 | #if defined(USE_MULTI_THREAD_SYSTEM) \
|
---|
| 123 | && defined(USE_DEFAULT_MULTI_THREAD_SYSTEM)
|
---|
| 124 |
|
---|
| 125 | #if defined(_WIN32) && !defined(MUSL_LIBC)
|
---|
| 126 | #define WIN32_LEAN_AND_MEAN
|
---|
| 127 | #include <windows.h>
|
---|
| 128 | extern CRITICAL_SECTION gOnigMutex;
|
---|
| 129 | #define THREAD_SYSTEM_INIT InitializeCriticalSection(&gOnigMutex)
|
---|
| 130 | #define THREAD_SYSTEM_END DeleteCriticalSection(&gOnigMutex)
|
---|
| 131 | #define THREAD_ATOMIC_START EnterCriticalSection(&gOnigMutex)
|
---|
| 132 | #define THREAD_ATOMIC_END LeaveCriticalSection(&gOnigMutex)
|
---|
| 133 | #define THREAD_PASS Sleep(0)
|
---|
| 134 | #else /* _WIN32 */
|
---|
| 135 | #include <pthread.h>
|
---|
| 136 | #include <sched.h>
|
---|
| 137 | extern pthread_mutex_t gOnigMutex;
|
---|
| 138 | #define THREAD_SYSTEM_INIT pthread_mutex_init(&gOnigMutex, NULL)
|
---|
| 139 | #define THREAD_SYSTEM_END pthread_mutex_destroy(&gOnigMutex)
|
---|
| 140 | #define THREAD_ATOMIC_START pthread_mutex_lock(&gOnigMutex)
|
---|
| 141 | #define THREAD_ATOMIC_END pthread_mutex_unlock(&gOnigMutex)
|
---|
| 142 | #define THREAD_PASS sched_yield()
|
---|
| 143 | #endif /* _WIN32 */
|
---|
| 144 |
|
---|
| 145 | #else /* USE_DEFAULT_MULTI_THREAD_SYSTEM */
|
---|
| 146 |
|
---|
| 147 | #ifndef THREAD_SYSTEM_INIT
|
---|
| 148 | #define THREAD_SYSTEM_INIT /* depend on thread system */
|
---|
| 149 | #endif
|
---|
| 150 | #ifndef THREAD_SYSTEM_END
|
---|
| 151 | #define THREAD_SYSTEM_END /* depend on thread system */
|
---|
| 152 | #endif
|
---|
| 153 | #ifndef THREAD_ATOMIC_START
|
---|
| 154 | #define THREAD_ATOMIC_START /* depend on thread system */
|
---|
| 155 | #endif
|
---|
| 156 | #ifndef THREAD_ATOMIC_END
|
---|
| 157 | #define THREAD_ATOMIC_END /* depend on thread system */
|
---|
| 158 | #endif
|
---|
| 159 | #ifndef THREAD_PASS
|
---|
| 160 | #define THREAD_PASS /* depend on thread system */
|
---|
| 161 | #endif
|
---|
| 162 |
|
---|
| 163 | #endif /* USE_DEFAULT_MULTI_THREAD_SYSTEM */
|
---|
| 164 |
|
---|
| 165 | #ifndef xmalloc
|
---|
| 166 | #define xmalloc malloc
|
---|
| 167 | #define xrealloc realloc
|
---|
| 168 | #define xcalloc calloc
|
---|
| 169 | #define xfree free
|
---|
| 170 | #endif
|
---|
| 171 |
|
---|
| 172 | #define CHECK_INTERRUPT_IN_MATCH_AT
|
---|
| 173 |
|
---|
| 174 | #define st_init_table onig_st_init_table
|
---|
| 175 | #define st_init_table_with_size onig_st_init_table_with_size
|
---|
| 176 | #define st_init_numtable onig_st_init_numtable
|
---|
| 177 | #define st_init_numtable_with_size onig_st_init_numtable_with_size
|
---|
| 178 | #define st_init_strtable onig_st_init_strtable
|
---|
| 179 | #define st_init_strtable_with_size onig_st_init_strtable_with_size
|
---|
| 180 | #define st_delete onig_st_delete
|
---|
| 181 | #define st_delete_safe onig_st_delete_safe
|
---|
| 182 | #define st_insert onig_st_insert
|
---|
| 183 | #define st_lookup onig_st_lookup
|
---|
| 184 | #define st_foreach onig_st_foreach
|
---|
| 185 | #define st_add_direct onig_st_add_direct
|
---|
| 186 | #define st_free_table onig_st_free_table
|
---|
| 187 | #define st_cleanup_safe onig_st_cleanup_safe
|
---|
| 188 | #define st_copy onig_st_copy
|
---|
| 189 | #define st_nothing_key_clone onig_st_nothing_key_clone
|
---|
| 190 | #define st_nothing_key_free onig_st_nothing_key_free
|
---|
| 191 | /* */
|
---|
| 192 | #define onig_st_is_member st_is_member
|
---|
| 193 |
|
---|
| 194 | #define STATE_CHECK_STRING_THRESHOLD_LEN 7
|
---|
| 195 | #define STATE_CHECK_BUFF_MAX_SIZE 0x4000
|
---|
| 196 |
|
---|
| 197 | #define THREAD_PASS_LIMIT_COUNT 8
|
---|
| 198 | #define xmemset memset
|
---|
| 199 | #define xmemcpy memcpy
|
---|
| 200 | #define xmemmove memmove
|
---|
| 201 |
|
---|
| 202 | #if defined(_WIN32) && !defined(__GNUC__) && !defined(MUSL_LIBC)
|
---|
| 203 | #define xalloca _alloca
|
---|
| 204 | #define xvsnprintf _vsnprintf
|
---|
| 205 | #else
|
---|
| 206 | /*#define xalloca alloca*/
|
---|
| 207 | #define xvsnprintf vsnprintf
|
---|
| 208 | #endif
|
---|
| 209 |
|
---|
| 210 |
|
---|
| 211 | #if defined(USE_RECOMPILE_API) && defined(USE_MULTI_THREAD_SYSTEM)
|
---|
| 212 | #define ONIG_STATE_INC(reg) (reg)->state++
|
---|
| 213 | #define ONIG_STATE_DEC(reg) (reg)->state--
|
---|
| 214 |
|
---|
| 215 | #define ONIG_STATE_INC_THREAD(reg) do {\
|
---|
| 216 | THREAD_ATOMIC_START;\
|
---|
| 217 | (reg)->state++;\
|
---|
| 218 | THREAD_ATOMIC_END;\
|
---|
| 219 | } while(0)
|
---|
| 220 | #define ONIG_STATE_DEC_THREAD(reg) do {\
|
---|
| 221 | THREAD_ATOMIC_START;\
|
---|
| 222 | (reg)->state--;\
|
---|
| 223 | THREAD_ATOMIC_END;\
|
---|
| 224 | } while(0)
|
---|
| 225 | #else
|
---|
| 226 | #define ONIG_STATE_INC(reg) /* Nothing */
|
---|
| 227 | #define ONIG_STATE_DEC(reg) /* Nothing */
|
---|
| 228 | #define ONIG_STATE_INC_THREAD(reg) /* Nothing */
|
---|
| 229 | #define ONIG_STATE_DEC_THREAD(reg) /* Nothing */
|
---|
| 230 | #endif /* USE_RECOMPILE_API && USE_MULTI_THREAD_SYSTEM */
|
---|
| 231 |
|
---|
| 232 | #ifdef HAVE_STDLIB_H
|
---|
| 233 | #include <stdlib.h>
|
---|
| 234 | #endif
|
---|
| 235 |
|
---|
| 236 | #if defined(HAVE_ALLOCA_H) && (defined(_AIX) || !defined(__GNUC__))
|
---|
| 237 | #include <alloca.h>
|
---|
| 238 | #endif
|
---|
| 239 |
|
---|
| 240 | #ifdef HAVE_STRING_H
|
---|
| 241 | # include <string.h>
|
---|
| 242 | #else
|
---|
| 243 | # include <strings.h>
|
---|
| 244 | #endif
|
---|
| 245 |
|
---|
| 246 | #include <ctype.h>
|
---|
| 247 | #ifdef HAVE_SYS_TYPES_H
|
---|
| 248 | #include <sys/types.h>
|
---|
| 249 | #endif
|
---|
| 250 |
|
---|
| 251 | #ifdef HAVE_STDINT_H
|
---|
| 252 | # include <stdint.h>
|
---|
| 253 | #endif
|
---|
| 254 |
|
---|
| 255 | #ifdef HAVE_INTTYPES_H
|
---|
| 256 | # include <inttypes.h>
|
---|
| 257 | #endif
|
---|
| 258 |
|
---|
| 259 | #ifdef STDC_HEADERS
|
---|
| 260 | # include <stddef.h>
|
---|
| 261 | #endif
|
---|
| 262 |
|
---|
| 263 | #if defined(_WIN32) && !defined(MUSL_LIBC)
|
---|
| 264 | #include <malloc.h> /* for alloca() */
|
---|
| 265 | #endif
|
---|
| 266 |
|
---|
| 267 | #ifdef ONIG_DEBUG
|
---|
| 268 | # include <stdio.h>
|
---|
| 269 | #endif
|
---|
| 270 |
|
---|
| 271 | #if defined(_WIN32) && !defined(MUSL_LIBC)
|
---|
| 272 | #if defined(_MSC_VER) && (_MSC_VER < 1300)
|
---|
| 273 | #ifndef _INTPTR_T_DEFINED
|
---|
| 274 | #define _INTPTR_T_DEFINED
|
---|
| 275 | typedef int intptr_t;
|
---|
| 276 | #endif
|
---|
| 277 | #ifndef _UINTPTR_T_DEFINED
|
---|
| 278 | #define _UINTPTR_T_DEFINED
|
---|
| 279 | typedef unsigned int uintptr_t;
|
---|
| 280 | #endif
|
---|
| 281 | #endif
|
---|
| 282 | #endif /* _WIN32 */
|
---|
| 283 |
|
---|
| 284 | #ifndef PRIdPTR
|
---|
| 285 | #ifdef _WIN64
|
---|
| 286 | #define PRIdPTR "I64d"
|
---|
| 287 | #define PRIuPTR "I64u"
|
---|
| 288 | #define PRIxPTR "I64x"
|
---|
| 289 | #else
|
---|
| 290 | #define PRIdPTR "ld"
|
---|
| 291 | #define PRIuPTR "lu"
|
---|
| 292 | #define PRIxPTR "lx"
|
---|
| 293 | #endif
|
---|
| 294 | #endif
|
---|
| 295 |
|
---|
| 296 | #include "regenc.h"
|
---|
| 297 |
|
---|
| 298 | #ifdef MIN
|
---|
| 299 | #undef MIN
|
---|
| 300 | #endif
|
---|
| 301 | #ifdef MAX
|
---|
| 302 | #undef MAX
|
---|
| 303 | #endif
|
---|
| 304 | #define MIN(a,b) (((a)>(b))?(b):(a))
|
---|
| 305 | #define MAX(a,b) (((a)<(b))?(b):(a))
|
---|
| 306 |
|
---|
| 307 | #define IS_NULL(p) (((void*)(p)) == (void*)0)
|
---|
| 308 | #define IS_NOT_NULL(p) (((void*)(p)) != (void*)0)
|
---|
| 309 | #define CHECK_NULL_RETURN(p) if (IS_NULL(p)) return NULL
|
---|
| 310 | #define CHECK_NULL_RETURN_MEMERR(p) if (IS_NULL(p)) return ONIGERR_MEMORY
|
---|
| 311 | #define NULL_UCHARP ((UChar* )0)
|
---|
| 312 |
|
---|
| 313 | #define ONIG_LAST_CODE_POINT (~((OnigCodePoint )0))
|
---|
| 314 |
|
---|
| 315 | #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
|
---|
| 316 |
|
---|
| 317 | #define PLATFORM_GET_INC(val,p,type) do{\
|
---|
| 318 | val = *(type* )p;\
|
---|
| 319 | (p) += sizeof(type);\
|
---|
| 320 | } while(0)
|
---|
| 321 |
|
---|
| 322 | #else
|
---|
| 323 |
|
---|
| 324 | #define PLATFORM_GET_INC(val,p,type) do{\
|
---|
| 325 | xmemcpy(&val, (p), sizeof(type));\
|
---|
| 326 | (p) += sizeof(type);\
|
---|
| 327 | } while(0)
|
---|
| 328 |
|
---|
| 329 | /* sizeof(OnigCodePoint) */
|
---|
| 330 | #define WORD_ALIGNMENT_SIZE SIZEOF_LONG
|
---|
| 331 |
|
---|
| 332 | #define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\
|
---|
| 333 | (pad_size) = WORD_ALIGNMENT_SIZE \
|
---|
| 334 | - ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
|
---|
| 335 | if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\
|
---|
| 336 | } while (0)
|
---|
| 337 |
|
---|
| 338 | #define ALIGNMENT_RIGHT(addr) do {\
|
---|
| 339 | (addr) += (WORD_ALIGNMENT_SIZE - 1);\
|
---|
| 340 | (addr) -= ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
|
---|
| 341 | } while (0)
|
---|
| 342 |
|
---|
| 343 | #endif /* PLATFORM_UNALIGNED_WORD_ACCESS */
|
---|
| 344 |
|
---|
| 345 | /* stack pop level */
|
---|
| 346 | #define STACK_POP_LEVEL_FREE 0
|
---|
| 347 | #define STACK_POP_LEVEL_MEM_START 1
|
---|
| 348 | #define STACK_POP_LEVEL_ALL 2
|
---|
| 349 |
|
---|
| 350 | /* optimize flags */
|
---|
| 351 | #define ONIG_OPTIMIZE_NONE 0
|
---|
| 352 | #define ONIG_OPTIMIZE_EXACT 1 /* Slow Search */
|
---|
| 353 | #define ONIG_OPTIMIZE_EXACT_BM 2 /* Boyer Moore Search */
|
---|
| 354 | #define ONIG_OPTIMIZE_EXACT_BM_NOT_REV 3 /* BM (applied to a multibyte string) */
|
---|
| 355 | #define ONIG_OPTIMIZE_EXACT_IC 4 /* Slow Search (ignore case) */
|
---|
| 356 | #define ONIG_OPTIMIZE_MAP 5 /* char map */
|
---|
| 357 | #define ONIG_OPTIMIZE_EXACT_BM_IC 6 /* BM (ignore case) */
|
---|
| 358 | #define ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC 7 /* BM (applied to a multibyte string) (ignore case) */
|
---|
| 359 |
|
---|
| 360 | /* bit status */
|
---|
| 361 | typedef unsigned int BitStatusType;
|
---|
| 362 |
|
---|
| 363 | #define BIT_STATUS_BITS_NUM (sizeof(BitStatusType) * 8)
|
---|
| 364 | #define BIT_STATUS_CLEAR(stats) (stats) = 0
|
---|
| 365 | #define BIT_STATUS_ON_ALL(stats) (stats) = ~((BitStatusType )0)
|
---|
| 366 | #define BIT_STATUS_AT(stats,n) \
|
---|
| 367 | ((n) < (int )BIT_STATUS_BITS_NUM ? ((stats) & ((BitStatusType )1 << n)) : ((stats) & 1))
|
---|
| 368 |
|
---|
| 369 | #define BIT_STATUS_ON_AT(stats,n) do {\
|
---|
| 370 | if ((n) < (int )BIT_STATUS_BITS_NUM)\
|
---|
| 371 | (stats) |= (1 << (n));\
|
---|
| 372 | else\
|
---|
| 373 | (stats) |= 1;\
|
---|
| 374 | } while (0)
|
---|
| 375 |
|
---|
| 376 | #define BIT_STATUS_ON_AT_SIMPLE(stats,n) do {\
|
---|
| 377 | if ((n) < (int )BIT_STATUS_BITS_NUM)\
|
---|
| 378 | (stats) |= (1 << (n));\
|
---|
| 379 | } while (0)
|
---|
| 380 |
|
---|
| 381 |
|
---|
| 382 | #define INT_MAX_LIMIT ((1UL << (SIZEOF_INT * 8 - 1)) - 1)
|
---|
| 383 |
|
---|
| 384 | #define DIGITVAL(code) ((code) - '0')
|
---|
| 385 | #define ODIGITVAL(code) DIGITVAL(code)
|
---|
| 386 | #define XDIGITVAL(enc,code) \
|
---|
| 387 | (ONIGENC_IS_CODE_DIGIT(enc,code) ? DIGITVAL(code) \
|
---|
| 388 | : (ONIGENC_IS_CODE_UPPER(enc,code) ? (code) - 'A' + 10 : (code) - 'a' + 10))
|
---|
| 389 |
|
---|
| 390 | #define IS_SINGLELINE(option) ((option) & ONIG_OPTION_SINGLELINE)
|
---|
| 391 | #define IS_MULTILINE(option) ((option) & ONIG_OPTION_MULTILINE)
|
---|
| 392 | #define IS_IGNORECASE(option) ((option) & ONIG_OPTION_IGNORECASE)
|
---|
| 393 | #define IS_EXTEND(option) ((option) & ONIG_OPTION_EXTEND)
|
---|
| 394 | #define IS_FIND_LONGEST(option) ((option) & ONIG_OPTION_FIND_LONGEST)
|
---|
| 395 | #define IS_FIND_NOT_EMPTY(option) ((option) & ONIG_OPTION_FIND_NOT_EMPTY)
|
---|
| 396 | #define IS_FIND_CONDITION(option) ((option) & \
|
---|
| 397 | (ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY))
|
---|
| 398 | #define IS_NOTBOL(option) ((option) & ONIG_OPTION_NOTBOL)
|
---|
| 399 | #define IS_NOTEOL(option) ((option) & ONIG_OPTION_NOTEOL)
|
---|
| 400 | #define IS_NOTBOS(option) ((option) & ONIG_OPTION_NOTBOS)
|
---|
| 401 | #define IS_NOTEOS(option) ((option) & ONIG_OPTION_NOTEOS)
|
---|
| 402 | #define IS_POSIX_REGION(option) ((option) & ONIG_OPTION_POSIX_REGION)
|
---|
| 403 | #define IS_ASCII_RANGE(option) ((option) & ONIG_OPTION_ASCII_RANGE)
|
---|
| 404 | #define IS_POSIX_BRACKET_ALL_RANGE(option) ((option) & ONIG_OPTION_POSIX_BRACKET_ALL_RANGE)
|
---|
| 405 | #define IS_WORD_BOUND_ALL_RANGE(option) ((option) & ONIG_OPTION_WORD_BOUND_ALL_RANGE)
|
---|
| 406 | #define IS_NEWLINE_CRLF(option) ((option) & ONIG_OPTION_NEWLINE_CRLF)
|
---|
| 407 |
|
---|
| 408 | /* OP_SET_OPTION is required for these options.
|
---|
| 409 | #define IS_DYNAMIC_OPTION(option) \
|
---|
| 410 | (((option) & (ONIG_OPTION_MULTILINE | ONIG_OPTION_IGNORECASE)) != 0)
|
---|
| 411 | */
|
---|
| 412 | /* ignore-case and multibyte status are included in compiled code. */
|
---|
| 413 | #define IS_DYNAMIC_OPTION(option) 0
|
---|
| 414 |
|
---|
| 415 | #define DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag) \
|
---|
| 416 | ((case_fold_flag) & ~INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR)
|
---|
| 417 |
|
---|
| 418 | #define REPEAT_INFINITE -1
|
---|
| 419 | #define IS_REPEAT_INFINITE(n) ((n) == REPEAT_INFINITE)
|
---|
| 420 |
|
---|
| 421 | /* bitset */
|
---|
| 422 | #define BITS_PER_BYTE 8
|
---|
| 423 | #define SINGLE_BYTE_SIZE (1 << BITS_PER_BYTE)
|
---|
| 424 | #define BITS_IN_ROOM ((int )sizeof(Bits) * BITS_PER_BYTE)
|
---|
| 425 | #define BITSET_SIZE (SINGLE_BYTE_SIZE / BITS_IN_ROOM)
|
---|
| 426 |
|
---|
| 427 | #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
|
---|
| 428 | typedef unsigned int Bits;
|
---|
| 429 | #else
|
---|
| 430 | typedef unsigned char Bits;
|
---|
| 431 | #endif
|
---|
| 432 | typedef Bits BitSet[BITSET_SIZE];
|
---|
| 433 | typedef Bits* BitSetRef;
|
---|
| 434 |
|
---|
| 435 | #define SIZE_BITSET (int )sizeof(BitSet)
|
---|
| 436 |
|
---|
| 437 | #define BITSET_CLEAR(bs) do {\
|
---|
| 438 | int i;\
|
---|
| 439 | for (i = 0; i < BITSET_SIZE; i++) { (bs)[i] = 0; } \
|
---|
| 440 | } while (0)
|
---|
| 441 |
|
---|
| 442 | #define BS_ROOM(bs,pos) (bs)[(int )(pos) / BITS_IN_ROOM]
|
---|
| 443 | #define BS_BIT(pos) (1U << ((int )(pos) % BITS_IN_ROOM))
|
---|
| 444 |
|
---|
| 445 | #define BITSET_AT(bs, pos) (BS_ROOM(bs,pos) & BS_BIT(pos))
|
---|
| 446 | #define BITSET_SET_BIT(bs, pos) BS_ROOM(bs,pos) |= BS_BIT(pos)
|
---|
| 447 | #define BITSET_CLEAR_BIT(bs, pos) BS_ROOM(bs,pos) &= ~(BS_BIT(pos))
|
---|
| 448 | #define BITSET_INVERT_BIT(bs, pos) BS_ROOM(bs,pos) ^= BS_BIT(pos)
|
---|
| 449 |
|
---|
| 450 | /* bytes buffer */
|
---|
| 451 | typedef struct _BBuf {
|
---|
| 452 | UChar* p;
|
---|
| 453 | unsigned int used;
|
---|
| 454 | unsigned int alloc;
|
---|
| 455 | } BBuf;
|
---|
| 456 |
|
---|
| 457 | #define BBUF_INIT(buf,size) onig_bbuf_init((BBuf* )(buf), (size))
|
---|
| 458 |
|
---|
| 459 | #define BBUF_SIZE_INC(buf,inc) do{\
|
---|
| 460 | UChar *tmp;\
|
---|
| 461 | (buf)->alloc += (inc);\
|
---|
| 462 | tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
|
---|
| 463 | if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
|
---|
| 464 | (buf)->p = tmp;\
|
---|
| 465 | } while (0)
|
---|
| 466 |
|
---|
| 467 | #define BBUF_EXPAND(buf,low) do{\
|
---|
| 468 | UChar *tmp;\
|
---|
| 469 | do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\
|
---|
| 470 | tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
|
---|
| 471 | if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
|
---|
| 472 | (buf)->p = tmp;\
|
---|
| 473 | } while (0)
|
---|
| 474 |
|
---|
| 475 | #define BBUF_ENSURE_SIZE(buf,size) do{\
|
---|
| 476 | unsigned int new_alloc = (buf)->alloc;\
|
---|
| 477 | while (new_alloc < (unsigned int )(size)) { new_alloc *= 2; }\
|
---|
| 478 | if ((buf)->alloc != new_alloc) {\
|
---|
| 479 | UChar *tmp;\
|
---|
| 480 | tmp = (UChar* )xrealloc((buf)->p, new_alloc);\
|
---|
| 481 | if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
|
---|
| 482 | (buf)->p = tmp;\
|
---|
| 483 | (buf)->alloc = new_alloc;\
|
---|
| 484 | }\
|
---|
| 485 | } while (0)
|
---|
| 486 |
|
---|
| 487 | #define BBUF_WRITE(buf,pos,bytes,n) do{\
|
---|
| 488 | int used = (pos) + (int )(n);\
|
---|
| 489 | if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
|
---|
| 490 | xmemcpy((buf)->p + (pos), (bytes), (n));\
|
---|
| 491 | if ((buf)->used < (unsigned int )used) (buf)->used = used;\
|
---|
| 492 | } while (0)
|
---|
| 493 |
|
---|
| 494 | #define BBUF_WRITE1(buf,pos,byte) do{\
|
---|
| 495 | int used = (pos) + 1;\
|
---|
| 496 | if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
|
---|
| 497 | (buf)->p[(pos)] = (UChar )(byte);\
|
---|
| 498 | if ((buf)->used < (unsigned int )used) (buf)->used = used;\
|
---|
| 499 | } while (0)
|
---|
| 500 |
|
---|
| 501 | #define BBUF_ADD(buf,bytes,n) BBUF_WRITE((buf),(buf)->used,(bytes),(n))
|
---|
| 502 | #define BBUF_ADD1(buf,byte) BBUF_WRITE1((buf),(buf)->used,(byte))
|
---|
| 503 | #define BBUF_GET_ADD_ADDRESS(buf) ((buf)->p + (buf)->used)
|
---|
| 504 | #define BBUF_GET_OFFSET_POS(buf) ((buf)->used)
|
---|
| 505 |
|
---|
| 506 | /* from < to */
|
---|
| 507 | #define BBUF_MOVE_RIGHT(buf,from,to,n) do {\
|
---|
| 508 | if ((unsigned int )((to)+(n)) > (buf)->alloc) BBUF_EXPAND((buf),(to) + (n));\
|
---|
| 509 | xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
|
---|
| 510 | if ((unsigned int )((to)+(n)) > (buf)->used) (buf)->used = (to) + (n);\
|
---|
| 511 | } while (0)
|
---|
| 512 |
|
---|
| 513 | /* from > to */
|
---|
| 514 | #define BBUF_MOVE_LEFT(buf,from,to,n) do {\
|
---|
| 515 | xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
|
---|
| 516 | } while (0)
|
---|
| 517 |
|
---|
| 518 | /* from > to */
|
---|
| 519 | #define BBUF_MOVE_LEFT_REDUCE(buf,from,to) do {\
|
---|
| 520 | xmemmove((buf)->p + (to), (buf)->p + (from), (buf)->used - (from));\
|
---|
| 521 | (buf)->used -= (from - to);\
|
---|
| 522 | } while (0)
|
---|
| 523 |
|
---|
| 524 | #define BBUF_INSERT(buf,pos,bytes,n) do {\
|
---|
| 525 | if (pos >= (buf)->used) {\
|
---|
| 526 | BBUF_WRITE(buf,pos,bytes,n);\
|
---|
| 527 | }\
|
---|
| 528 | else {\
|
---|
| 529 | BBUF_MOVE_RIGHT((buf),(pos),(pos) + (n),((buf)->used - (pos)));\
|
---|
| 530 | xmemcpy((buf)->p + (pos), (bytes), (n));\
|
---|
| 531 | }\
|
---|
| 532 | } while (0)
|
---|
| 533 |
|
---|
| 534 | #define BBUF_GET_BYTE(buf, pos) (buf)->p[(pos)]
|
---|
| 535 |
|
---|
| 536 |
|
---|
| 537 | #define ANCHOR_BEGIN_BUF (1<<0)
|
---|
| 538 | #define ANCHOR_BEGIN_LINE (1<<1)
|
---|
| 539 | #define ANCHOR_BEGIN_POSITION (1<<2)
|
---|
| 540 | #define ANCHOR_END_BUF (1<<3)
|
---|
| 541 | #define ANCHOR_SEMI_END_BUF (1<<4)
|
---|
| 542 | #define ANCHOR_END_LINE (1<<5)
|
---|
| 543 |
|
---|
| 544 | #define ANCHOR_WORD_BOUND (1<<6)
|
---|
| 545 | #define ANCHOR_NOT_WORD_BOUND (1<<7)
|
---|
| 546 | #define ANCHOR_WORD_BEGIN (1<<8)
|
---|
| 547 | #define ANCHOR_WORD_END (1<<9)
|
---|
| 548 | #define ANCHOR_PREC_READ (1<<10)
|
---|
| 549 | #define ANCHOR_PREC_READ_NOT (1<<11)
|
---|
| 550 | #define ANCHOR_LOOK_BEHIND (1<<12)
|
---|
| 551 | #define ANCHOR_LOOK_BEHIND_NOT (1<<13)
|
---|
| 552 |
|
---|
| 553 | #define ANCHOR_ANYCHAR_STAR (1<<14) /* ".*" optimize info */
|
---|
| 554 | #define ANCHOR_ANYCHAR_STAR_ML (1<<15) /* ".*" optimize info (multi-line) */
|
---|
| 555 |
|
---|
| 556 | #define ANCHOR_KEEP (1<<16)
|
---|
| 557 |
|
---|
| 558 | /* operation code */
|
---|
| 559 | enum OpCode {
|
---|
| 560 | OP_FINISH = 0, /* matching process terminator (no more alternative) */
|
---|
| 561 | OP_END = 1, /* pattern code terminator (success end) */
|
---|
| 562 |
|
---|
| 563 | OP_EXACT1 = 2, /* single byte, N = 1 */
|
---|
| 564 | OP_EXACT2, /* single byte, N = 2 */
|
---|
| 565 | OP_EXACT3, /* single byte, N = 3 */
|
---|
| 566 | OP_EXACT4, /* single byte, N = 4 */
|
---|
| 567 | OP_EXACT5, /* single byte, N = 5 */
|
---|
| 568 | OP_EXACTN, /* single byte */
|
---|
| 569 | OP_EXACTMB2N1, /* mb-length = 2 N = 1 */
|
---|
| 570 | OP_EXACTMB2N2, /* mb-length = 2 N = 2 */
|
---|
| 571 | OP_EXACTMB2N3, /* mb-length = 2 N = 3 */
|
---|
| 572 | OP_EXACTMB2N, /* mb-length = 2 */
|
---|
| 573 | OP_EXACTMB3N, /* mb-length = 3 */
|
---|
| 574 | OP_EXACTMBN, /* other length */
|
---|
| 575 |
|
---|
| 576 | OP_EXACT1_IC, /* single byte, N = 1, ignore case */
|
---|
| 577 | OP_EXACTN_IC, /* single byte, ignore case */
|
---|
| 578 |
|
---|
| 579 | OP_CCLASS,
|
---|
| 580 | OP_CCLASS_MB,
|
---|
| 581 | OP_CCLASS_MIX,
|
---|
| 582 | OP_CCLASS_NOT,
|
---|
| 583 | OP_CCLASS_MB_NOT,
|
---|
| 584 | OP_CCLASS_MIX_NOT,
|
---|
| 585 | OP_CCLASS_NODE, /* pointer to CClassNode node */
|
---|
| 586 |
|
---|
| 587 | OP_ANYCHAR, /* "." */
|
---|
| 588 | OP_ANYCHAR_ML, /* "." multi-line */
|
---|
| 589 | OP_ANYCHAR_STAR, /* ".*" */
|
---|
| 590 | OP_ANYCHAR_ML_STAR, /* ".*" multi-line */
|
---|
| 591 | OP_ANYCHAR_STAR_PEEK_NEXT,
|
---|
| 592 | OP_ANYCHAR_ML_STAR_PEEK_NEXT,
|
---|
| 593 |
|
---|
| 594 | OP_WORD,
|
---|
| 595 | OP_NOT_WORD,
|
---|
| 596 | OP_WORD_BOUND,
|
---|
| 597 | OP_NOT_WORD_BOUND,
|
---|
| 598 | OP_WORD_BEGIN,
|
---|
| 599 | OP_WORD_END,
|
---|
| 600 |
|
---|
| 601 | OP_ASCII_WORD,
|
---|
| 602 | OP_NOT_ASCII_WORD,
|
---|
| 603 | OP_ASCII_WORD_BOUND,
|
---|
| 604 | OP_NOT_ASCII_WORD_BOUND,
|
---|
| 605 | OP_ASCII_WORD_BEGIN,
|
---|
| 606 | OP_ASCII_WORD_END,
|
---|
| 607 |
|
---|
| 608 | OP_BEGIN_BUF,
|
---|
| 609 | OP_END_BUF,
|
---|
| 610 | OP_BEGIN_LINE,
|
---|
| 611 | OP_END_LINE,
|
---|
| 612 | OP_SEMI_END_BUF,
|
---|
| 613 | OP_BEGIN_POSITION,
|
---|
| 614 | OP_BEGIN_POS_OR_LINE, /* used for implicit anchor optimization */
|
---|
| 615 |
|
---|
| 616 | OP_BACKREF1,
|
---|
| 617 | OP_BACKREF2,
|
---|
| 618 | OP_BACKREFN,
|
---|
| 619 | OP_BACKREFN_IC,
|
---|
| 620 | OP_BACKREF_MULTI,
|
---|
| 621 | OP_BACKREF_MULTI_IC,
|
---|
| 622 | OP_BACKREF_WITH_LEVEL, /* \k<xxx+n>, \k<xxx-n> */
|
---|
| 623 |
|
---|
| 624 | OP_MEMORY_START,
|
---|
| 625 | OP_MEMORY_START_PUSH, /* push back-tracker to stack */
|
---|
| 626 | OP_MEMORY_END_PUSH, /* push back-tracker to stack */
|
---|
| 627 | OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */
|
---|
| 628 | OP_MEMORY_END,
|
---|
| 629 | OP_MEMORY_END_REC, /* push marker to stack */
|
---|
| 630 |
|
---|
| 631 | OP_KEEP,
|
---|
| 632 |
|
---|
| 633 | OP_FAIL, /* pop stack and move */
|
---|
| 634 | OP_JUMP,
|
---|
| 635 | OP_PUSH,
|
---|
| 636 | OP_POP,
|
---|
| 637 | OP_PUSH_OR_JUMP_EXACT1, /* if match exact then push, else jump. */
|
---|
| 638 | OP_PUSH_IF_PEEK_NEXT, /* if match exact then push, else none. */
|
---|
| 639 | OP_REPEAT, /* {n,m} */
|
---|
| 640 | OP_REPEAT_NG, /* {n,m}? (non greedy) */
|
---|
| 641 | OP_REPEAT_INC,
|
---|
| 642 | OP_REPEAT_INC_NG, /* non greedy */
|
---|
| 643 | OP_REPEAT_INC_SG, /* search and get in stack */
|
---|
| 644 | OP_REPEAT_INC_NG_SG, /* search and get in stack (non greedy) */
|
---|
| 645 | OP_NULL_CHECK_START, /* null loop checker start */
|
---|
| 646 | OP_NULL_CHECK_END, /* null loop checker end */
|
---|
| 647 | OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */
|
---|
| 648 | OP_NULL_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */
|
---|
| 649 |
|
---|
| 650 | OP_PUSH_POS, /* (?=...) start */
|
---|
| 651 | OP_POP_POS, /* (?=...) end */
|
---|
| 652 | OP_PUSH_POS_NOT, /* (?!...) start */
|
---|
| 653 | OP_FAIL_POS, /* (?!...) end */
|
---|
| 654 | OP_PUSH_STOP_BT, /* (?>...) start */
|
---|
| 655 | OP_POP_STOP_BT, /* (?>...) end */
|
---|
| 656 | OP_LOOK_BEHIND, /* (?<=...) start (no needs end opcode) */
|
---|
| 657 | OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */
|
---|
| 658 | OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end */
|
---|
| 659 |
|
---|
| 660 | OP_CALL, /* \g<name> */
|
---|
| 661 | OP_RETURN,
|
---|
| 662 |
|
---|
| 663 | OP_CONDITION,
|
---|
| 664 |
|
---|
| 665 | OP_STATE_CHECK_PUSH, /* combination explosion check and push */
|
---|
| 666 | OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump */
|
---|
| 667 | OP_STATE_CHECK, /* check only */
|
---|
| 668 | OP_STATE_CHECK_ANYCHAR_STAR,
|
---|
| 669 | OP_STATE_CHECK_ANYCHAR_ML_STAR,
|
---|
| 670 |
|
---|
| 671 | /* no need: IS_DYNAMIC_OPTION() == 0 */
|
---|
| 672 | OP_SET_OPTION_PUSH, /* set option and push recover option */
|
---|
| 673 | OP_SET_OPTION /* set option */
|
---|
| 674 | };
|
---|
| 675 |
|
---|
| 676 | typedef int RelAddrType;
|
---|
| 677 | typedef int AbsAddrType;
|
---|
| 678 | typedef int LengthType;
|
---|
| 679 | typedef int RepeatNumType;
|
---|
| 680 | typedef short int MemNumType;
|
---|
| 681 | typedef short int StateCheckNumType;
|
---|
| 682 | typedef void* PointerType;
|
---|
| 683 |
|
---|
| 684 | #define SIZE_OPCODE 1
|
---|
| 685 | #define SIZE_RELADDR (int )sizeof(RelAddrType)
|
---|
| 686 | #define SIZE_ABSADDR (int )sizeof(AbsAddrType)
|
---|
| 687 | #define SIZE_LENGTH (int )sizeof(LengthType)
|
---|
| 688 | #define SIZE_MEMNUM (int )sizeof(MemNumType)
|
---|
| 689 | #define SIZE_STATE_CHECK_NUM (int )sizeof(StateCheckNumType)
|
---|
| 690 | #define SIZE_REPEATNUM (int )sizeof(RepeatNumType)
|
---|
| 691 | #define SIZE_OPTION (int )sizeof(OnigOptionType)
|
---|
| 692 | #define SIZE_CODE_POINT (int )sizeof(OnigCodePoint)
|
---|
| 693 | #define SIZE_POINTER (int )sizeof(PointerType)
|
---|
| 694 |
|
---|
| 695 |
|
---|
| 696 | #define GET_RELADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, RelAddrType)
|
---|
| 697 | #define GET_ABSADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, AbsAddrType)
|
---|
| 698 | #define GET_LENGTH_INC(len,p) PLATFORM_GET_INC(len, p, LengthType)
|
---|
| 699 | #define GET_MEMNUM_INC(num,p) PLATFORM_GET_INC(num, p, MemNumType)
|
---|
| 700 | #define GET_REPEATNUM_INC(num,p) PLATFORM_GET_INC(num, p, RepeatNumType)
|
---|
| 701 | #define GET_OPTION_INC(option,p) PLATFORM_GET_INC(option, p, OnigOptionType)
|
---|
| 702 | #define GET_POINTER_INC(ptr,p) PLATFORM_GET_INC(ptr, p, PointerType)
|
---|
| 703 | #define GET_STATE_CHECK_NUM_INC(num,p) PLATFORM_GET_INC(num, p, StateCheckNumType)
|
---|
| 704 |
|
---|
| 705 | /* code point's address must be aligned address. */
|
---|
| 706 | #define GET_CODE_POINT(code,p) code = *((OnigCodePoint* )(p))
|
---|
| 707 | #define GET_BYTE_INC(byte,p) do{\
|
---|
| 708 | byte = *(p);\
|
---|
| 709 | (p)++;\
|
---|
| 710 | } while(0)
|
---|
| 711 |
|
---|
| 712 |
|
---|
| 713 | /* op-code + arg size */
|
---|
| 714 | #define SIZE_OP_ANYCHAR_STAR SIZE_OPCODE
|
---|
| 715 | #define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT (SIZE_OPCODE + 1)
|
---|
| 716 | #define SIZE_OP_JUMP (SIZE_OPCODE + SIZE_RELADDR)
|
---|
| 717 | #define SIZE_OP_PUSH (SIZE_OPCODE + SIZE_RELADDR)
|
---|
| 718 | #define SIZE_OP_POP SIZE_OPCODE
|
---|
| 719 | #define SIZE_OP_PUSH_OR_JUMP_EXACT1 (SIZE_OPCODE + SIZE_RELADDR + 1)
|
---|
| 720 | #define SIZE_OP_PUSH_IF_PEEK_NEXT (SIZE_OPCODE + SIZE_RELADDR + 1)
|
---|
| 721 | #define SIZE_OP_REPEAT_INC (SIZE_OPCODE + SIZE_MEMNUM)
|
---|
| 722 | #define SIZE_OP_REPEAT_INC_NG (SIZE_OPCODE + SIZE_MEMNUM)
|
---|
| 723 | #define SIZE_OP_PUSH_POS SIZE_OPCODE
|
---|
| 724 | #define SIZE_OP_PUSH_POS_NOT (SIZE_OPCODE + SIZE_RELADDR)
|
---|
| 725 | #define SIZE_OP_POP_POS SIZE_OPCODE
|
---|
| 726 | #define SIZE_OP_FAIL_POS SIZE_OPCODE
|
---|
| 727 | #define SIZE_OP_SET_OPTION (SIZE_OPCODE + SIZE_OPTION)
|
---|
| 728 | #define SIZE_OP_SET_OPTION_PUSH (SIZE_OPCODE + SIZE_OPTION)
|
---|
| 729 | #define SIZE_OP_FAIL SIZE_OPCODE
|
---|
| 730 | #define SIZE_OP_MEMORY_START (SIZE_OPCODE + SIZE_MEMNUM)
|
---|
| 731 | #define SIZE_OP_MEMORY_START_PUSH (SIZE_OPCODE + SIZE_MEMNUM)
|
---|
| 732 | #define SIZE_OP_MEMORY_END_PUSH (SIZE_OPCODE + SIZE_MEMNUM)
|
---|
| 733 | #define SIZE_OP_MEMORY_END_PUSH_REC (SIZE_OPCODE + SIZE_MEMNUM)
|
---|
| 734 | #define SIZE_OP_MEMORY_END (SIZE_OPCODE + SIZE_MEMNUM)
|
---|
| 735 | #define SIZE_OP_MEMORY_END_REC (SIZE_OPCODE + SIZE_MEMNUM)
|
---|
| 736 | #define SIZE_OP_PUSH_STOP_BT SIZE_OPCODE
|
---|
| 737 | #define SIZE_OP_POP_STOP_BT SIZE_OPCODE
|
---|
| 738 | #define SIZE_OP_NULL_CHECK_START (SIZE_OPCODE + SIZE_MEMNUM)
|
---|
| 739 | #define SIZE_OP_NULL_CHECK_END (SIZE_OPCODE + SIZE_MEMNUM)
|
---|
| 740 | #define SIZE_OP_LOOK_BEHIND (SIZE_OPCODE + SIZE_LENGTH)
|
---|
| 741 | #define SIZE_OP_PUSH_LOOK_BEHIND_NOT (SIZE_OPCODE + SIZE_RELADDR + SIZE_LENGTH)
|
---|
| 742 | #define SIZE_OP_FAIL_LOOK_BEHIND_NOT SIZE_OPCODE
|
---|
| 743 | #define SIZE_OP_CALL (SIZE_OPCODE + SIZE_ABSADDR)
|
---|
| 744 | #define SIZE_OP_RETURN SIZE_OPCODE
|
---|
| 745 | #define SIZE_OP_CONDITION (SIZE_OPCODE + SIZE_MEMNUM + SIZE_RELADDR)
|
---|
| 746 |
|
---|
| 747 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 748 | #define SIZE_OP_STATE_CHECK (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
|
---|
| 749 | #define SIZE_OP_STATE_CHECK_PUSH (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
|
---|
| 750 | #define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
|
---|
| 751 | #define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
|
---|
| 752 | #endif
|
---|
| 753 |
|
---|
| 754 | #define MC_ESC(syn) (syn)->meta_char_table.esc
|
---|
| 755 | #define MC_ANYCHAR(syn) (syn)->meta_char_table.anychar
|
---|
| 756 | #define MC_ANYTIME(syn) (syn)->meta_char_table.anytime
|
---|
| 757 | #define MC_ZERO_OR_ONE_TIME(syn) (syn)->meta_char_table.zero_or_one_time
|
---|
| 758 | #define MC_ONE_OR_MORE_TIME(syn) (syn)->meta_char_table.one_or_more_time
|
---|
| 759 | #define MC_ANYCHAR_ANYTIME(syn) (syn)->meta_char_table.anychar_anytime
|
---|
| 760 |
|
---|
| 761 | #define IS_MC_ESC_CODE(code, syn) \
|
---|
| 762 | ((code) == MC_ESC(syn) && \
|
---|
| 763 | !IS_SYNTAX_OP2((syn), ONIG_SYN_OP2_INEFFECTIVE_ESCAPE))
|
---|
| 764 |
|
---|
| 765 |
|
---|
| 766 | #define SYN_POSIX_COMMON_OP \
|
---|
| 767 | ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_POSIX_BRACKET | \
|
---|
| 768 | ONIG_SYN_OP_DECIMAL_BACKREF | \
|
---|
| 769 | ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_ASTERISK_ZERO_INF | \
|
---|
| 770 | ONIG_SYN_OP_LINE_ANCHOR | \
|
---|
| 771 | ONIG_SYN_OP_ESC_CONTROL_CHARS )
|
---|
| 772 |
|
---|
| 773 | #define SYN_GNU_REGEX_OP \
|
---|
| 774 | ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | \
|
---|
| 775 | ONIG_SYN_OP_POSIX_BRACKET | ONIG_SYN_OP_DECIMAL_BACKREF | \
|
---|
| 776 | ONIG_SYN_OP_BRACE_INTERVAL | ONIG_SYN_OP_LPAREN_SUBEXP | \
|
---|
| 777 | ONIG_SYN_OP_VBAR_ALT | \
|
---|
| 778 | ONIG_SYN_OP_ASTERISK_ZERO_INF | ONIG_SYN_OP_PLUS_ONE_INF | \
|
---|
| 779 | ONIG_SYN_OP_QMARK_ZERO_ONE | \
|
---|
| 780 | ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR | ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR | \
|
---|
| 781 | ONIG_SYN_OP_ESC_W_WORD | \
|
---|
| 782 | ONIG_SYN_OP_ESC_B_WORD_BOUND | ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END | \
|
---|
| 783 | ONIG_SYN_OP_ESC_S_WHITE_SPACE | ONIG_SYN_OP_ESC_D_DIGIT | \
|
---|
| 784 | ONIG_SYN_OP_LINE_ANCHOR )
|
---|
| 785 |
|
---|
| 786 | #define SYN_GNU_REGEX_BV \
|
---|
| 787 | ( ONIG_SYN_CONTEXT_INDEP_ANCHORS | ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | \
|
---|
| 788 | ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | ONIG_SYN_ALLOW_INVALID_INTERVAL | \
|
---|
| 789 | ONIG_SYN_BACKSLASH_ESCAPE_IN_CC | ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC )
|
---|
| 790 |
|
---|
| 791 |
|
---|
| 792 | #define NCCLASS_FLAGS(cc) ((cc)->flags)
|
---|
| 793 | #define NCCLASS_FLAG_SET(cc,flag) (NCCLASS_FLAGS(cc) |= (flag))
|
---|
| 794 | #define NCCLASS_FLAG_CLEAR(cc,flag) (NCCLASS_FLAGS(cc) &= ~(flag))
|
---|
| 795 | #define IS_NCCLASS_FLAG_ON(cc,flag) ((NCCLASS_FLAGS(cc) & (flag)) != 0)
|
---|
| 796 |
|
---|
| 797 | /* cclass node */
|
---|
| 798 | #define FLAG_NCCLASS_NOT (1<<0)
|
---|
| 799 | #define FLAG_NCCLASS_SHARE (1<<1)
|
---|
| 800 |
|
---|
| 801 | #define NCCLASS_SET_NOT(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_NOT)
|
---|
| 802 | #define NCCLASS_SET_SHARE(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_SHARE)
|
---|
| 803 | #define NCCLASS_CLEAR_NOT(nd) NCCLASS_FLAG_CLEAR(nd, FLAG_NCCLASS_NOT)
|
---|
| 804 | #define IS_NCCLASS_NOT(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_NOT)
|
---|
| 805 | #define IS_NCCLASS_SHARE(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_SHARE)
|
---|
| 806 |
|
---|
| 807 | typedef struct {
|
---|
| 808 | int type;
|
---|
| 809 | /* struct _Node* next; */
|
---|
| 810 | /* unsigned int flags; */
|
---|
| 811 | } NodeBase;
|
---|
| 812 |
|
---|
| 813 | typedef struct {
|
---|
| 814 | NodeBase base;
|
---|
| 815 | unsigned int flags;
|
---|
| 816 | BitSet bs;
|
---|
| 817 | BBuf* mbuf; /* multi-byte info or NULL */
|
---|
| 818 | } CClassNode;
|
---|
| 819 |
|
---|
| 820 | typedef intptr_t OnigStackIndex;
|
---|
| 821 |
|
---|
| 822 | typedef struct _OnigStackType {
|
---|
| 823 | unsigned int type;
|
---|
| 824 | union {
|
---|
| 825 | struct {
|
---|
| 826 | UChar *pcode; /* byte code position */
|
---|
| 827 | UChar *pstr; /* string position */
|
---|
| 828 | UChar *pstr_prev; /* previous char position of pstr */
|
---|
| 829 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 830 | unsigned int state_check;
|
---|
| 831 | #endif
|
---|
| 832 | UChar *pkeep; /* keep pattern position */
|
---|
| 833 | } state;
|
---|
| 834 | struct {
|
---|
| 835 | int count; /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */
|
---|
| 836 | UChar *pcode; /* byte code position (head of repeated target) */
|
---|
| 837 | int num; /* repeat id */
|
---|
| 838 | } repeat;
|
---|
| 839 | struct {
|
---|
| 840 | OnigStackIndex si; /* index of stack */
|
---|
| 841 | } repeat_inc;
|
---|
| 842 | struct {
|
---|
| 843 | int num; /* memory num */
|
---|
| 844 | UChar *pstr; /* start/end position */
|
---|
| 845 | /* Following information is set, if this stack type is MEM-START */
|
---|
| 846 | OnigStackIndex start; /* prev. info (for backtrack "(...)*" ) */
|
---|
| 847 | OnigStackIndex end; /* prev. info (for backtrack "(...)*" ) */
|
---|
| 848 | } mem;
|
---|
| 849 | struct {
|
---|
| 850 | int num; /* null check id */
|
---|
| 851 | UChar *pstr; /* start position */
|
---|
| 852 | } null_check;
|
---|
| 853 | #ifdef USE_SUBEXP_CALL
|
---|
| 854 | struct {
|
---|
| 855 | UChar *ret_addr; /* byte code position */
|
---|
| 856 | int num; /* null check id */
|
---|
| 857 | UChar *pstr; /* string position */
|
---|
| 858 | } call_frame;
|
---|
| 859 | #endif
|
---|
| 860 | } u;
|
---|
| 861 | } OnigStackType;
|
---|
| 862 |
|
---|
| 863 | typedef struct {
|
---|
| 864 | void* stack_p;
|
---|
| 865 | size_t stack_n;
|
---|
| 866 | OnigOptionType options;
|
---|
| 867 | OnigRegion* region;
|
---|
| 868 | const UChar* start; /* search start position */
|
---|
| 869 | const UChar* gpos; /* global position (for \G: BEGIN_POSITION) */
|
---|
| 870 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
| 871 | OnigPosition best_len; /* for ONIG_OPTION_FIND_LONGEST */
|
---|
| 872 | UChar* best_s;
|
---|
| 873 | #endif
|
---|
| 874 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
| 875 | void* state_check_buff;
|
---|
| 876 | int state_check_buff_size;
|
---|
| 877 | #endif
|
---|
| 878 | } OnigMatchArg;
|
---|
| 879 |
|
---|
| 880 |
|
---|
| 881 | #define IS_CODE_SB_WORD(enc,code) \
|
---|
| 882 | (ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code))
|
---|
| 883 |
|
---|
| 884 | typedef struct OnigEndCallListItem {
|
---|
| 885 | struct OnigEndCallListItem* next;
|
---|
| 886 | void (*func)(void);
|
---|
| 887 | } OnigEndCallListItemType;
|
---|
| 888 |
|
---|
| 889 | extern void onig_add_end_call(void (*func)(void));
|
---|
| 890 |
|
---|
| 891 |
|
---|
| 892 | #ifdef ONIG_DEBUG
|
---|
| 893 |
|
---|
| 894 | typedef struct {
|
---|
| 895 | short int opcode;
|
---|
| 896 | const char* name;
|
---|
| 897 | short int arg_type;
|
---|
| 898 | } OnigOpInfoType;
|
---|
| 899 |
|
---|
| 900 | extern OnigOpInfoType OnigOpInfo[];
|
---|
| 901 |
|
---|
| 902 |
|
---|
| 903 | extern void onig_print_compiled_byte_code P_((FILE* f, UChar* bp, UChar** nextp, OnigEncoding enc));
|
---|
| 904 |
|
---|
| 905 | #ifdef ONIG_DEBUG_STATISTICS
|
---|
| 906 | extern void onig_statistics_init P_((void));
|
---|
| 907 | extern void onig_print_statistics P_((FILE* f));
|
---|
| 908 | #endif
|
---|
| 909 | #endif
|
---|
| 910 |
|
---|
| 911 | extern UChar* onig_error_code_to_format P_((OnigPosition code));
|
---|
| 912 | extern void onig_snprintf_with_pattern PV_((UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const UChar *fmt, ...));
|
---|
| 913 | extern int onig_bbuf_init P_((BBuf* buf, OnigDistance size));
|
---|
| 914 | extern int onig_compile P_((regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo));
|
---|
| 915 | extern void onig_chain_reduce P_((regex_t* reg));
|
---|
| 916 | extern void onig_chain_link_add P_((regex_t* to, regex_t* add));
|
---|
| 917 | extern void onig_transfer P_((regex_t* to, regex_t* from));
|
---|
| 918 | extern int onig_is_code_in_cc P_((OnigEncoding enc, OnigCodePoint code, CClassNode* cc));
|
---|
| 919 | extern int onig_is_code_in_cc_len P_((int enclen, OnigCodePoint code, CClassNode* cc));
|
---|
| 920 |
|
---|
| 921 | /* strend hash */
|
---|
| 922 | typedef void hash_table_type;
|
---|
| 923 | typedef uintptr_t hash_data_type;
|
---|
| 924 |
|
---|
| 925 | extern hash_table_type* onig_st_init_strend_table_with_size P_((int size));
|
---|
| 926 | extern int onig_st_lookup_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type *value));
|
---|
| 927 | extern int onig_st_insert_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type value));
|
---|
| 928 |
|
---|
| 929 | /* encoding property management */
|
---|
| 930 | #define PROPERTY_LIST_ADD_PROP(Name, CR) \
|
---|
| 931 | r = onigenc_property_list_add_property((UChar* )Name, CR,\
|
---|
| 932 | &PropertyNameTable, &PropertyList, &PropertyListNum,\
|
---|
| 933 | &PropertyListSize);\
|
---|
| 934 | if (r != 0) goto end
|
---|
| 935 |
|
---|
| 936 | #define PROPERTY_LIST_INIT_CHECK \
|
---|
| 937 | if (PropertyInited == 0) {\
|
---|
| 938 | int r = onigenc_property_list_init(init_property_list);\
|
---|
| 939 | if (r != 0) return r;\
|
---|
| 940 | }
|
---|
| 941 |
|
---|
| 942 | extern int onigenc_property_list_add_property P_((UChar* name, const OnigCodePoint* prop, hash_table_type **table, const OnigCodePoint*** plist, int *pnum, int *psize));
|
---|
| 943 |
|
---|
| 944 | typedef int (*ONIGENC_INIT_PROPERTY_LIST_FUNC_TYPE)(void);
|
---|
| 945 |
|
---|
| 946 | extern int onigenc_property_list_init P_((ONIGENC_INIT_PROPERTY_LIST_FUNC_TYPE));
|
---|
| 947 |
|
---|
| 948 | extern size_t onig_memsize P_((const regex_t *reg));
|
---|
| 949 | extern size_t onig_region_memsize P_((const struct re_registers *regs));
|
---|
| 950 |
|
---|
| 951 | #endif /* REGINT_H */
|
---|