onigmo 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.github/workflows/main.yml +31 -0
- data/.gitignore +10 -0
- data/Gemfile +9 -0
- data/Gemfile.lock +29 -0
- data/LICENSE +7 -0
- data/README.md +101 -0
- data/Rakefile +45 -0
- data/bin/console +10 -0
- data/ext/onigmo/extconf.rb +7 -0
- data/ext/onigmo/onigmo.c +985 -0
- data/ext/onigmo/regenc.h +256 -0
- data/ext/onigmo/regint.h +945 -0
- data/ext/onigmo/regparse.h +370 -0
- data/lib/onigmo/deconstruct_visitor.rb +121 -0
- data/lib/onigmo/json_visitor.rb +123 -0
- data/lib/onigmo/node.rb +266 -0
- data/lib/onigmo/pretty_print_visitor.rb +284 -0
- data/lib/onigmo/visitor.rb +339 -0
- data/lib/onigmo.rb +11 -0
- data/onigmo.gemspec +16 -0
- metadata +64 -0
data/ext/onigmo/regint.h
ADDED
@@ -0,0 +1,945 @@
|
|
1
|
+
#ifndef ONIGMO_REGINT_H
|
2
|
+
#define ONIGMO_REGINT_H
|
3
|
+
/**********************************************************************
|
4
|
+
regint.h - Onigmo (Oniguruma-mod) (regular expression library)
|
5
|
+
**********************************************************************/
|
6
|
+
/*-
|
7
|
+
* Copyright (c) 2002-2013 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
|
8
|
+
* Copyright (c) 2011-2019 K.Takata <kentkt AT csc DOT jp>
|
9
|
+
* All rights reserved.
|
10
|
+
*
|
11
|
+
* Redistribution and use in source and binary forms, with or without
|
12
|
+
* modification, are permitted provided that the following conditions
|
13
|
+
* are met:
|
14
|
+
* 1. Redistributions of source code must retain the above copyright
|
15
|
+
* notice, this list of conditions and the following disclaimer.
|
16
|
+
* 2. Redistributions in binary form must reproduce the above copyright
|
17
|
+
* notice, this list of conditions and the following disclaimer in the
|
18
|
+
* documentation and/or other materials provided with the distribution.
|
19
|
+
*
|
20
|
+
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
21
|
+
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
22
|
+
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
23
|
+
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
24
|
+
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
25
|
+
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
26
|
+
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
27
|
+
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
28
|
+
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
29
|
+
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
30
|
+
* SUCH DAMAGE.
|
31
|
+
*/
|
32
|
+
|
33
|
+
/* for debug */
|
34
|
+
/* #define ONIG_DEBUG_PARSE_TREE */
|
35
|
+
/* #define ONIG_DEBUG_COMPILE */
|
36
|
+
/* #define ONIG_DEBUG_SEARCH */
|
37
|
+
/* #define ONIG_DEBUG_MATCH */
|
38
|
+
/* #define ONIG_DEBUG_MEMLEAK */
|
39
|
+
/* #define ONIG_DONT_OPTIMIZE */
|
40
|
+
|
41
|
+
/* for byte-code statistical data. */
|
42
|
+
/* #define ONIG_DEBUG_STATISTICS */
|
43
|
+
|
44
|
+
#if defined(ONIG_DEBUG_PARSE_TREE) || defined(ONIG_DEBUG_MATCH) || \
|
45
|
+
defined(ONIG_DEBUG_SEARCH) || defined(ONIG_DEBUG_COMPILE) || \
|
46
|
+
defined(ONIG_DEBUG_STATISTICS) || defined(ONIG_DEBUG_MEMLEAK)
|
47
|
+
# ifndef ONIG_DEBUG
|
48
|
+
# define ONIG_DEBUG
|
49
|
+
# endif
|
50
|
+
#endif
|
51
|
+
|
52
|
+
#ifndef UNALIGNED_WORD_ACCESS
|
53
|
+
# if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \
|
54
|
+
defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD64) || \
|
55
|
+
defined(__powerpc64__) || \
|
56
|
+
defined(__mc68020__)
|
57
|
+
# define UNALIGNED_WORD_ACCESS 1
|
58
|
+
# else
|
59
|
+
# define UNALIGNED_WORD_ACCESS 0
|
60
|
+
# endif
|
61
|
+
#endif
|
62
|
+
|
63
|
+
#if UNALIGNED_WORD_ACCESS
|
64
|
+
# define PLATFORM_UNALIGNED_WORD_ACCESS
|
65
|
+
#endif
|
66
|
+
|
67
|
+
/* config */
|
68
|
+
/* spec. config */
|
69
|
+
#define USE_NAMED_GROUP
|
70
|
+
#define USE_SUBEXP_CALL
|
71
|
+
#define USE_PERL_SUBEXP_CALL
|
72
|
+
#define USE_CAPITAL_P_NAMED_GROUP
|
73
|
+
#define USE_BACKREF_WITH_LEVEL /* \k<name+n>, \k<name-n> */
|
74
|
+
#define USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT /* /(?:()|())*\2/ */
|
75
|
+
#define USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE /* /\n$/ =~ "\n" */
|
76
|
+
#define USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR
|
77
|
+
/* !!! moved to regenc.h. */ /* #define USE_CRNL_AS_LINE_TERMINATOR */
|
78
|
+
#define USE_NO_INVALID_QUANTIFIER
|
79
|
+
|
80
|
+
/* internal config */
|
81
|
+
/* #define USE_OP_PUSH_OR_JUMP_EXACT */
|
82
|
+
#define USE_QTFR_PEEK_NEXT
|
83
|
+
#define USE_ST_LIBRARY
|
84
|
+
|
85
|
+
#define INIT_MATCH_STACK_SIZE 160
|
86
|
+
#define DEFAULT_MATCH_STACK_LIMIT_SIZE 0 /* unlimited */
|
87
|
+
#define DEFAULT_PARSE_DEPTH_LIMIT 4096
|
88
|
+
|
89
|
+
#define OPT_EXACT_MAXLEN 24 /* This must be smaller than ONIG_CHAR_TABLE_SIZE. */
|
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
|
+
#if !defined(RUBY) && defined(RUBY_EXPORT)
|
105
|
+
# define RUBY
|
106
|
+
#endif
|
107
|
+
#ifdef RUBY
|
108
|
+
# ifndef RUBY_DEFINES_H
|
109
|
+
# include "ruby/ruby.h"
|
110
|
+
# undef xmalloc
|
111
|
+
# undef xrealloc
|
112
|
+
# undef xcalloc
|
113
|
+
# undef xfree
|
114
|
+
# endif
|
115
|
+
#else /* RUBY */
|
116
|
+
# include "config.h"
|
117
|
+
# if SIZEOF_LONG_LONG > 0
|
118
|
+
# define LONG_LONG long long
|
119
|
+
# endif
|
120
|
+
#endif /* RUBY */
|
121
|
+
|
122
|
+
#include <stdarg.h>
|
123
|
+
|
124
|
+
/* */
|
125
|
+
/* escape other system UChar definition */
|
126
|
+
#ifdef ONIG_ESCAPE_UCHAR_COLLISION
|
127
|
+
# undef ONIG_ESCAPE_UCHAR_COLLISION
|
128
|
+
#endif
|
129
|
+
|
130
|
+
#define USE_WORD_BEGIN_END /* "\<": word-begin, "\>": word-end */
|
131
|
+
#ifdef RUBY
|
132
|
+
# undef USE_CAPTURE_HISTORY
|
133
|
+
#else
|
134
|
+
# define USE_CAPTURE_HISTORY
|
135
|
+
#endif
|
136
|
+
#define USE_VARIABLE_META_CHARS
|
137
|
+
#define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
138
|
+
/* #define USE_COMBINATION_EXPLOSION_CHECK */ /* (X*)* */
|
139
|
+
|
140
|
+
|
141
|
+
#ifndef xmalloc
|
142
|
+
# define xmalloc malloc
|
143
|
+
# define xrealloc realloc
|
144
|
+
# define xcalloc calloc
|
145
|
+
# define xfree free
|
146
|
+
#endif
|
147
|
+
|
148
|
+
#ifdef RUBY
|
149
|
+
|
150
|
+
# define CHECK_INTERRUPT_IN_MATCH_AT rb_thread_check_ints()
|
151
|
+
# define onig_st_init_table st_init_table
|
152
|
+
# define onig_st_init_table_with_size st_init_table_with_size
|
153
|
+
# define onig_st_init_numtable st_init_numtable
|
154
|
+
# define onig_st_init_numtable_with_size st_init_numtable_with_size
|
155
|
+
# define onig_st_init_strtable st_init_strtable
|
156
|
+
# define onig_st_init_strtable_with_size st_init_strtable_with_size
|
157
|
+
# define onig_st_delete st_delete
|
158
|
+
# define onig_st_delete_safe st_delete_safe
|
159
|
+
# define onig_st_insert st_insert
|
160
|
+
# define onig_st_lookup st_lookup
|
161
|
+
# define onig_st_foreach st_foreach
|
162
|
+
# define onig_st_add_direct st_add_direct
|
163
|
+
# define onig_st_free_table st_free_table
|
164
|
+
# define onig_st_cleanup_safe st_cleanup_safe
|
165
|
+
# define onig_st_copy st_copy
|
166
|
+
# define onig_st_nothing_key_clone st_nothing_key_clone
|
167
|
+
# define onig_st_nothing_key_free st_nothing_key_free
|
168
|
+
# define onig_st_is_member st_is_member
|
169
|
+
|
170
|
+
# define USE_UPPER_CASE_TABLE
|
171
|
+
#else /* RUBY */
|
172
|
+
|
173
|
+
# define CHECK_INTERRUPT_IN_MATCH_AT
|
174
|
+
|
175
|
+
# define st_init_table onig_st_init_table
|
176
|
+
# define st_init_table_with_size onig_st_init_table_with_size
|
177
|
+
# define st_init_numtable onig_st_init_numtable
|
178
|
+
# define st_init_numtable_with_size onig_st_init_numtable_with_size
|
179
|
+
# define st_init_strtable onig_st_init_strtable
|
180
|
+
# define st_init_strtable_with_size onig_st_init_strtable_with_size
|
181
|
+
# define st_delete onig_st_delete
|
182
|
+
# define st_delete_safe onig_st_delete_safe
|
183
|
+
# define st_insert onig_st_insert
|
184
|
+
# define st_lookup onig_st_lookup
|
185
|
+
# define st_foreach onig_st_foreach
|
186
|
+
# define st_add_direct onig_st_add_direct
|
187
|
+
# define st_free_table onig_st_free_table
|
188
|
+
# define st_cleanup_safe onig_st_cleanup_safe
|
189
|
+
# define st_copy onig_st_copy
|
190
|
+
# define st_nothing_key_clone onig_st_nothing_key_clone
|
191
|
+
# define st_nothing_key_free onig_st_nothing_key_free
|
192
|
+
/* */
|
193
|
+
# define onig_st_is_member st_is_member
|
194
|
+
|
195
|
+
# ifdef __GNUC__
|
196
|
+
# define RB_GNUC_EXTENSION __extension__
|
197
|
+
# define RB_GNUC_EXTENSION_BLOCK(x) __extension__ ({ x; })
|
198
|
+
# else
|
199
|
+
# define RB_GNUC_EXTENSION
|
200
|
+
# define RB_GNUC_EXTENSION_BLOCK(x) (x)
|
201
|
+
# endif
|
202
|
+
|
203
|
+
#endif /* RUBY */
|
204
|
+
|
205
|
+
#define STATE_CHECK_STRING_THRESHOLD_LEN 7
|
206
|
+
#define STATE_CHECK_BUFF_MAX_SIZE 0x4000
|
207
|
+
|
208
|
+
#define xmemset memset
|
209
|
+
#define xmemcpy memcpy
|
210
|
+
#define xmemmove memmove
|
211
|
+
|
212
|
+
#if ((defined(RUBY_MSVCRT_VERSION) && RUBY_MSVCRT_VERSION >= 90) \
|
213
|
+
|| (!defined(RUBY_MSVCRT_VERSION) && defined(_WIN32))) \
|
214
|
+
&& !defined(__GNUC__)
|
215
|
+
# define xalloca _alloca
|
216
|
+
# define xvsnprintf(buf,size,fmt,args) _vsnprintf_s(buf,size,_TRUNCATE,fmt,args)
|
217
|
+
# define xsnprintf sprintf_s
|
218
|
+
# define xstrcat(dest,src,size) strcat_s(dest,size,src)
|
219
|
+
#else
|
220
|
+
# define xalloca alloca
|
221
|
+
# define xvsnprintf vsnprintf
|
222
|
+
# define xsnprintf snprintf
|
223
|
+
# define xstrcat(dest,src,size) strcat(dest,src)
|
224
|
+
#endif
|
225
|
+
|
226
|
+
#if defined(ONIG_DEBUG_MEMLEAK) && defined(_MSC_VER)
|
227
|
+
# define _CRTDBG_MAP_ALLOC
|
228
|
+
# include <malloc.h>
|
229
|
+
# include <crtdbg.h>
|
230
|
+
#endif
|
231
|
+
|
232
|
+
#include <stdlib.h>
|
233
|
+
|
234
|
+
#if defined(HAVE_ALLOCA_H) && (defined(_AIX) || !defined(__GNUC__))
|
235
|
+
# include <alloca.h>
|
236
|
+
#endif
|
237
|
+
|
238
|
+
#include <string.h>
|
239
|
+
|
240
|
+
#include <ctype.h>
|
241
|
+
#ifdef HAVE_SYS_TYPES_H
|
242
|
+
# include <sys/types.h>
|
243
|
+
#endif
|
244
|
+
|
245
|
+
#ifdef HAVE_STDINT_H
|
246
|
+
# include <stdint.h>
|
247
|
+
#endif
|
248
|
+
|
249
|
+
#ifdef HAVE_INTTYPES_H
|
250
|
+
# include <inttypes.h>
|
251
|
+
#endif
|
252
|
+
|
253
|
+
#include <stddef.h>
|
254
|
+
|
255
|
+
#ifdef _WIN32
|
256
|
+
# include <malloc.h> /* for alloca() */
|
257
|
+
#endif
|
258
|
+
|
259
|
+
#ifdef ONIG_DEBUG
|
260
|
+
# include <stdio.h>
|
261
|
+
#endif
|
262
|
+
|
263
|
+
#ifdef _WIN32
|
264
|
+
# if defined(_MSC_VER) && (_MSC_VER < 1300)
|
265
|
+
# ifndef _INTPTR_T_DEFINED
|
266
|
+
# define _INTPTR_T_DEFINED
|
267
|
+
typedef int intptr_t;
|
268
|
+
# endif
|
269
|
+
# ifndef _UINTPTR_T_DEFINED
|
270
|
+
# define _UINTPTR_T_DEFINED
|
271
|
+
typedef unsigned int uintptr_t;
|
272
|
+
# endif
|
273
|
+
# endif
|
274
|
+
#endif /* _WIN32 */
|
275
|
+
|
276
|
+
#ifndef PRIdPTR
|
277
|
+
# ifdef _WIN64
|
278
|
+
# define PRIdPTR "I64d"
|
279
|
+
# define PRIuPTR "I64u"
|
280
|
+
# define PRIxPTR "I64x"
|
281
|
+
# else
|
282
|
+
# define PRIdPTR "ld"
|
283
|
+
# define PRIuPTR "lu"
|
284
|
+
# define PRIxPTR "lx"
|
285
|
+
# endif
|
286
|
+
#endif
|
287
|
+
|
288
|
+
#ifndef PRIdPTRDIFF
|
289
|
+
# define PRIdPTRDIFF PRIdPTR
|
290
|
+
#endif
|
291
|
+
|
292
|
+
#include "regenc.h"
|
293
|
+
|
294
|
+
RUBY_SYMBOL_EXPORT_BEGIN
|
295
|
+
|
296
|
+
#ifdef MIN
|
297
|
+
# undef MIN
|
298
|
+
#endif
|
299
|
+
#ifdef MAX
|
300
|
+
# undef MAX
|
301
|
+
#endif
|
302
|
+
#define MIN(a,b) (((a)>(b))?(b):(a))
|
303
|
+
#define MAX(a,b) (((a)<(b))?(b):(a))
|
304
|
+
|
305
|
+
#define IS_NULL(p) (((void*)(p)) == (void*)0)
|
306
|
+
#define IS_NOT_NULL(p) (((void*)(p)) != (void*)0)
|
307
|
+
#define CHECK_NULL_RETURN(p) if (IS_NULL(p)) return NULL
|
308
|
+
#define CHECK_NULL_RETURN_MEMERR(p) if (IS_NULL(p)) return ONIGERR_MEMORY
|
309
|
+
#define NULL_UCHARP ((UChar* )0)
|
310
|
+
|
311
|
+
#define ONIG_LAST_CODE_POINT (~((OnigCodePoint )0))
|
312
|
+
|
313
|
+
#ifdef PLATFORM_UNALIGNED_WORD_ACCESS
|
314
|
+
|
315
|
+
# define PLATFORM_GET_INC(val,p,type) do{\
|
316
|
+
val = *(type* )p;\
|
317
|
+
(p) += sizeof(type);\
|
318
|
+
} while(0)
|
319
|
+
|
320
|
+
#else
|
321
|
+
|
322
|
+
# define PLATFORM_GET_INC(val,p,type) do{\
|
323
|
+
xmemcpy(&val, (p), sizeof(type));\
|
324
|
+
(p) += sizeof(type);\
|
325
|
+
} while(0)
|
326
|
+
|
327
|
+
/* sizeof(OnigCodePoint) */
|
328
|
+
# define WORD_ALIGNMENT_SIZE SIZEOF_LONG
|
329
|
+
|
330
|
+
# define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\
|
331
|
+
(pad_size) = WORD_ALIGNMENT_SIZE \
|
332
|
+
- ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
|
333
|
+
if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\
|
334
|
+
} while (0)
|
335
|
+
|
336
|
+
# define ALIGNMENT_RIGHT(addr) do {\
|
337
|
+
(addr) += (WORD_ALIGNMENT_SIZE - 1);\
|
338
|
+
(addr) -= ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
|
339
|
+
} while (0)
|
340
|
+
|
341
|
+
#endif /* PLATFORM_UNALIGNED_WORD_ACCESS */
|
342
|
+
|
343
|
+
/* stack pop level */
|
344
|
+
#define STACK_POP_LEVEL_FREE 0
|
345
|
+
#define STACK_POP_LEVEL_MEM_START 1
|
346
|
+
#define STACK_POP_LEVEL_ALL 2
|
347
|
+
|
348
|
+
/* optimize flags */
|
349
|
+
#define ONIG_OPTIMIZE_NONE 0
|
350
|
+
#define ONIG_OPTIMIZE_EXACT 1 /* Slow Search */
|
351
|
+
#define ONIG_OPTIMIZE_EXACT_BM 2 /* Boyer Moore Search */
|
352
|
+
#define ONIG_OPTIMIZE_EXACT_BM_NOT_REV 3 /* BM (applied to a multibyte string) */
|
353
|
+
#define ONIG_OPTIMIZE_EXACT_IC 4 /* Slow Search (ignore case) */
|
354
|
+
#define ONIG_OPTIMIZE_MAP 5 /* char map */
|
355
|
+
#define ONIG_OPTIMIZE_EXACT_BM_IC 6 /* BM (ignore case) */
|
356
|
+
#define ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC 7 /* BM (applied to a multibyte string) (ignore case) */
|
357
|
+
|
358
|
+
/* bit status */
|
359
|
+
typedef unsigned int BitStatusType;
|
360
|
+
|
361
|
+
#define BIT_STATUS_BITS_NUM (sizeof(BitStatusType) * 8)
|
362
|
+
#define BIT_STATUS_CLEAR(stats) (stats) = 0
|
363
|
+
#define BIT_STATUS_ON_ALL(stats) (stats) = ~((BitStatusType )0)
|
364
|
+
#define BIT_STATUS_AT(stats,n) \
|
365
|
+
((n) < (int )BIT_STATUS_BITS_NUM ? ((stats) & ((BitStatusType )1 << n)) : ((stats) & 1))
|
366
|
+
|
367
|
+
#define BIT_STATUS_ON_AT(stats,n) do {\
|
368
|
+
if ((n) < (int )BIT_STATUS_BITS_NUM)\
|
369
|
+
(stats) |= (1 << (n));\
|
370
|
+
else\
|
371
|
+
(stats) |= 1;\
|
372
|
+
} while (0)
|
373
|
+
|
374
|
+
#define BIT_STATUS_ON_AT_SIMPLE(stats,n) do {\
|
375
|
+
if ((n) < (int )BIT_STATUS_BITS_NUM)\
|
376
|
+
(stats) |= (1 << (n));\
|
377
|
+
} while (0)
|
378
|
+
|
379
|
+
|
380
|
+
#define INT_MAX_LIMIT ((1UL << (SIZEOF_INT * 8 - 1)) - 1)
|
381
|
+
|
382
|
+
#define DIGITVAL(code) ((code) - '0')
|
383
|
+
#define ODIGITVAL(code) DIGITVAL(code)
|
384
|
+
#define XDIGITVAL(enc,code) \
|
385
|
+
(ONIGENC_IS_CODE_DIGIT(enc,code) ? DIGITVAL(code) \
|
386
|
+
: (ONIGENC_IS_CODE_UPPER(enc,code) ? (code) - 'A' + 10 : (code) - 'a' + 10))
|
387
|
+
|
388
|
+
#define IS_SINGLELINE(option) ((option) & ONIG_OPTION_SINGLELINE)
|
389
|
+
#define IS_MULTILINE(option) ((option) & ONIG_OPTION_MULTILINE)
|
390
|
+
#define IS_IGNORECASE(option) ((option) & ONIG_OPTION_IGNORECASE)
|
391
|
+
#define IS_EXTEND(option) ((option) & ONIG_OPTION_EXTEND)
|
392
|
+
#define IS_FIND_LONGEST(option) ((option) & ONIG_OPTION_FIND_LONGEST)
|
393
|
+
#define IS_FIND_NOT_EMPTY(option) ((option) & ONIG_OPTION_FIND_NOT_EMPTY)
|
394
|
+
#define IS_FIND_CONDITION(option) ((option) & \
|
395
|
+
(ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY))
|
396
|
+
#define IS_NOTBOL(option) ((option) & ONIG_OPTION_NOTBOL)
|
397
|
+
#define IS_NOTEOL(option) ((option) & ONIG_OPTION_NOTEOL)
|
398
|
+
#define IS_NOTBOS(option) ((option) & ONIG_OPTION_NOTBOS)
|
399
|
+
#define IS_NOTEOS(option) ((option) & ONIG_OPTION_NOTEOS)
|
400
|
+
#define IS_ASCII_RANGE(option) ((option) & ONIG_OPTION_ASCII_RANGE)
|
401
|
+
#define IS_POSIX_BRACKET_ALL_RANGE(option) ((option) & ONIG_OPTION_POSIX_BRACKET_ALL_RANGE)
|
402
|
+
#define IS_WORD_BOUND_ALL_RANGE(option) ((option) & ONIG_OPTION_WORD_BOUND_ALL_RANGE)
|
403
|
+
#define IS_NEWLINE_CRLF(option) ((option) & ONIG_OPTION_NEWLINE_CRLF)
|
404
|
+
|
405
|
+
/* OP_SET_OPTION is required for these options.
|
406
|
+
#define IS_DYNAMIC_OPTION(option) \
|
407
|
+
(((option) & (ONIG_OPTION_MULTILINE | ONIG_OPTION_IGNORECASE)) != 0)
|
408
|
+
*/
|
409
|
+
/* ignore-case and multibyte status are included in compiled code. */
|
410
|
+
#define IS_DYNAMIC_OPTION(option) 0
|
411
|
+
|
412
|
+
#define DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag) \
|
413
|
+
((case_fold_flag) & ~INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR)
|
414
|
+
|
415
|
+
#define REPEAT_INFINITE -1
|
416
|
+
#define IS_REPEAT_INFINITE(n) ((n) == REPEAT_INFINITE)
|
417
|
+
|
418
|
+
/* bitset */
|
419
|
+
#define BITS_PER_BYTE 8
|
420
|
+
#define SINGLE_BYTE_SIZE (1 << BITS_PER_BYTE)
|
421
|
+
#define BITS_IN_ROOM ((int )sizeof(Bits) * BITS_PER_BYTE)
|
422
|
+
#define BITSET_SIZE (SINGLE_BYTE_SIZE / BITS_IN_ROOM)
|
423
|
+
|
424
|
+
#ifdef PLATFORM_UNALIGNED_WORD_ACCESS
|
425
|
+
typedef unsigned int Bits;
|
426
|
+
#else
|
427
|
+
typedef unsigned char Bits;
|
428
|
+
#endif
|
429
|
+
typedef Bits BitSet[BITSET_SIZE];
|
430
|
+
typedef Bits* BitSetRef;
|
431
|
+
|
432
|
+
#define SIZE_BITSET (int )sizeof(BitSet)
|
433
|
+
|
434
|
+
#define BITSET_CLEAR(bs) do {\
|
435
|
+
int i;\
|
436
|
+
for (i = 0; i < BITSET_SIZE; i++) { (bs)[i] = 0; } \
|
437
|
+
} while (0)
|
438
|
+
|
439
|
+
#define BS_ROOM(bs,pos) (bs)[(int )(pos) / BITS_IN_ROOM]
|
440
|
+
#define BS_BIT(pos) (1U << ((int )(pos) % BITS_IN_ROOM))
|
441
|
+
|
442
|
+
#define BITSET_AT(bs, pos) (BS_ROOM(bs,pos) & BS_BIT(pos))
|
443
|
+
#define BITSET_SET_BIT(bs, pos) BS_ROOM(bs,pos) |= BS_BIT(pos)
|
444
|
+
#define BITSET_CLEAR_BIT(bs, pos) BS_ROOM(bs,pos) &= ~(BS_BIT(pos))
|
445
|
+
#define BITSET_INVERT_BIT(bs, pos) BS_ROOM(bs,pos) ^= BS_BIT(pos)
|
446
|
+
|
447
|
+
/* bytes buffer */
|
448
|
+
typedef struct _BBuf {
|
449
|
+
UChar* p;
|
450
|
+
unsigned int used;
|
451
|
+
unsigned int alloc;
|
452
|
+
} BBuf;
|
453
|
+
|
454
|
+
#define BBUF_INIT(buf,size) onig_bbuf_init((BBuf* )(buf), (size))
|
455
|
+
|
456
|
+
#define BBUF_SIZE_INC(buf,inc) do{\
|
457
|
+
UChar *tmp;\
|
458
|
+
(buf)->alloc += (inc);\
|
459
|
+
tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
|
460
|
+
if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
|
461
|
+
(buf)->p = tmp;\
|
462
|
+
} while (0)
|
463
|
+
|
464
|
+
#define BBUF_EXPAND(buf,low) do{\
|
465
|
+
UChar *tmp;\
|
466
|
+
do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\
|
467
|
+
tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
|
468
|
+
if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
|
469
|
+
(buf)->p = tmp;\
|
470
|
+
} while (0)
|
471
|
+
|
472
|
+
#define BBUF_ENSURE_SIZE(buf,size) do{\
|
473
|
+
unsigned int new_alloc = (buf)->alloc;\
|
474
|
+
while (new_alloc < (unsigned int )(size)) { new_alloc *= 2; }\
|
475
|
+
if ((buf)->alloc != new_alloc) {\
|
476
|
+
UChar *tmp;\
|
477
|
+
tmp = (UChar* )xrealloc((buf)->p, new_alloc);\
|
478
|
+
if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
|
479
|
+
(buf)->p = tmp;\
|
480
|
+
(buf)->alloc = new_alloc;\
|
481
|
+
}\
|
482
|
+
} while (0)
|
483
|
+
|
484
|
+
#define BBUF_WRITE(buf,pos,bytes,n) do{\
|
485
|
+
int used = (pos) + (int )(n);\
|
486
|
+
if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
|
487
|
+
xmemcpy((buf)->p + (pos), (bytes), (n));\
|
488
|
+
if ((buf)->used < (unsigned int )used) (buf)->used = used;\
|
489
|
+
} while (0)
|
490
|
+
|
491
|
+
#define BBUF_WRITE1(buf,pos,byte) do{\
|
492
|
+
int used = (pos) + 1;\
|
493
|
+
if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
|
494
|
+
(buf)->p[(pos)] = (UChar )(byte);\
|
495
|
+
if ((buf)->used < (unsigned int )used) (buf)->used = used;\
|
496
|
+
} while (0)
|
497
|
+
|
498
|
+
#define BBUF_ADD(buf,bytes,n) BBUF_WRITE((buf),(buf)->used,(bytes),(n))
|
499
|
+
#define BBUF_ADD1(buf,byte) BBUF_WRITE1((buf),(buf)->used,(byte))
|
500
|
+
#define BBUF_GET_ADD_ADDRESS(buf) ((buf)->p + (buf)->used)
|
501
|
+
#define BBUF_GET_OFFSET_POS(buf) ((buf)->used)
|
502
|
+
|
503
|
+
/* from < to */
|
504
|
+
#define BBUF_MOVE_RIGHT(buf,from,to,n) do {\
|
505
|
+
if ((unsigned int )((to)+(n)) > (buf)->alloc) BBUF_EXPAND((buf),(to) + (n));\
|
506
|
+
xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
|
507
|
+
if ((unsigned int )((to)+(n)) > (buf)->used) (buf)->used = (to) + (n);\
|
508
|
+
} while (0)
|
509
|
+
|
510
|
+
/* from > to */
|
511
|
+
#define BBUF_MOVE_LEFT(buf,from,to,n) do {\
|
512
|
+
xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
|
513
|
+
} while (0)
|
514
|
+
|
515
|
+
/* from > to */
|
516
|
+
#define BBUF_MOVE_LEFT_REDUCE(buf,from,to) do {\
|
517
|
+
xmemmove((buf)->p + (to), (buf)->p + (from), (buf)->used - (from));\
|
518
|
+
(buf)->used -= (from - to);\
|
519
|
+
} while (0)
|
520
|
+
|
521
|
+
#define BBUF_INSERT(buf,pos,bytes,n) do {\
|
522
|
+
if (pos >= (buf)->used) {\
|
523
|
+
BBUF_WRITE(buf,pos,bytes,n);\
|
524
|
+
}\
|
525
|
+
else {\
|
526
|
+
BBUF_MOVE_RIGHT((buf),(pos),(pos) + (n),((buf)->used - (pos)));\
|
527
|
+
xmemcpy((buf)->p + (pos), (bytes), (n));\
|
528
|
+
}\
|
529
|
+
} while (0)
|
530
|
+
|
531
|
+
#define BBUF_GET_BYTE(buf, pos) (buf)->p[(pos)]
|
532
|
+
|
533
|
+
|
534
|
+
#define ANCHOR_BEGIN_BUF (1<<0)
|
535
|
+
#define ANCHOR_BEGIN_LINE (1<<1)
|
536
|
+
#define ANCHOR_BEGIN_POSITION (1<<2)
|
537
|
+
#define ANCHOR_END_BUF (1<<3)
|
538
|
+
#define ANCHOR_SEMI_END_BUF (1<<4)
|
539
|
+
#define ANCHOR_END_LINE (1<<5)
|
540
|
+
|
541
|
+
#define ANCHOR_WORD_BOUND (1<<6)
|
542
|
+
#define ANCHOR_NOT_WORD_BOUND (1<<7)
|
543
|
+
#define ANCHOR_WORD_BEGIN (1<<8)
|
544
|
+
#define ANCHOR_WORD_END (1<<9)
|
545
|
+
#define ANCHOR_PREC_READ (1<<10)
|
546
|
+
#define ANCHOR_PREC_READ_NOT (1<<11)
|
547
|
+
#define ANCHOR_LOOK_BEHIND (1<<12)
|
548
|
+
#define ANCHOR_LOOK_BEHIND_NOT (1<<13)
|
549
|
+
|
550
|
+
#define ANCHOR_ANYCHAR_STAR (1<<14) /* ".*" optimize info */
|
551
|
+
#define ANCHOR_ANYCHAR_STAR_ML (1<<15) /* ".*" optimize info (multi-line) */
|
552
|
+
|
553
|
+
#define ANCHOR_KEEP (1<<16)
|
554
|
+
|
555
|
+
/* operation code */
|
556
|
+
enum OpCode {
|
557
|
+
OP_FINISH = 0, /* matching process terminator (no more alternative) */
|
558
|
+
OP_END = 1, /* pattern code terminator (success end) */
|
559
|
+
|
560
|
+
OP_EXACT1 = 2, /* single byte, N = 1 */
|
561
|
+
OP_EXACT2, /* single byte, N = 2 */
|
562
|
+
OP_EXACT3, /* single byte, N = 3 */
|
563
|
+
OP_EXACT4, /* single byte, N = 4 */
|
564
|
+
OP_EXACT5, /* single byte, N = 5 */
|
565
|
+
OP_EXACTN, /* single byte */
|
566
|
+
OP_EXACTMB2N1, /* mb-length = 2 N = 1 */
|
567
|
+
OP_EXACTMB2N2, /* mb-length = 2 N = 2 */
|
568
|
+
OP_EXACTMB2N3, /* mb-length = 2 N = 3 */
|
569
|
+
OP_EXACTMB2N, /* mb-length = 2 */
|
570
|
+
OP_EXACTMB3N, /* mb-length = 3 */
|
571
|
+
OP_EXACTMBN, /* other length */
|
572
|
+
|
573
|
+
OP_EXACT1_IC, /* single byte, N = 1, ignore case */
|
574
|
+
OP_EXACTN_IC, /* single byte, ignore case */
|
575
|
+
|
576
|
+
OP_CCLASS,
|
577
|
+
OP_CCLASS_MB,
|
578
|
+
OP_CCLASS_MIX,
|
579
|
+
OP_CCLASS_NOT,
|
580
|
+
OP_CCLASS_MB_NOT,
|
581
|
+
OP_CCLASS_MIX_NOT,
|
582
|
+
|
583
|
+
OP_ANYCHAR, /* "." */
|
584
|
+
OP_ANYCHAR_ML, /* "." multi-line */
|
585
|
+
OP_ANYCHAR_STAR, /* ".*" */
|
586
|
+
OP_ANYCHAR_ML_STAR, /* ".*" multi-line */
|
587
|
+
OP_ANYCHAR_STAR_PEEK_NEXT,
|
588
|
+
OP_ANYCHAR_ML_STAR_PEEK_NEXT,
|
589
|
+
|
590
|
+
OP_WORD,
|
591
|
+
OP_NOT_WORD,
|
592
|
+
OP_WORD_BOUND,
|
593
|
+
OP_NOT_WORD_BOUND,
|
594
|
+
OP_WORD_BEGIN,
|
595
|
+
OP_WORD_END,
|
596
|
+
|
597
|
+
OP_ASCII_WORD,
|
598
|
+
OP_NOT_ASCII_WORD,
|
599
|
+
OP_ASCII_WORD_BOUND,
|
600
|
+
OP_NOT_ASCII_WORD_BOUND,
|
601
|
+
OP_ASCII_WORD_BEGIN,
|
602
|
+
OP_ASCII_WORD_END,
|
603
|
+
|
604
|
+
OP_BEGIN_BUF,
|
605
|
+
OP_END_BUF,
|
606
|
+
OP_BEGIN_LINE,
|
607
|
+
OP_END_LINE,
|
608
|
+
OP_SEMI_END_BUF,
|
609
|
+
OP_BEGIN_POSITION,
|
610
|
+
|
611
|
+
OP_BACKREF1,
|
612
|
+
OP_BACKREF2,
|
613
|
+
OP_BACKREFN,
|
614
|
+
OP_BACKREFN_IC,
|
615
|
+
OP_BACKREF_MULTI,
|
616
|
+
OP_BACKREF_MULTI_IC,
|
617
|
+
OP_BACKREF_WITH_LEVEL, /* \k<xxx+n>, \k<xxx-n> */
|
618
|
+
|
619
|
+
OP_MEMORY_START,
|
620
|
+
OP_MEMORY_START_PUSH, /* push back-tracker to stack */
|
621
|
+
OP_MEMORY_END_PUSH, /* push back-tracker to stack */
|
622
|
+
OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */
|
623
|
+
OP_MEMORY_END,
|
624
|
+
OP_MEMORY_END_REC, /* push marker to stack */
|
625
|
+
|
626
|
+
OP_KEEP,
|
627
|
+
|
628
|
+
OP_FAIL, /* pop stack and move */
|
629
|
+
OP_JUMP,
|
630
|
+
OP_PUSH,
|
631
|
+
OP_POP,
|
632
|
+
OP_PUSH_OR_JUMP_EXACT1, /* if match exact then push, else jump. */
|
633
|
+
OP_PUSH_IF_PEEK_NEXT, /* if match exact then push, else none. */
|
634
|
+
OP_REPEAT, /* {n,m} */
|
635
|
+
OP_REPEAT_NG, /* {n,m}? (non greedy) */
|
636
|
+
OP_REPEAT_INC,
|
637
|
+
OP_REPEAT_INC_NG, /* non greedy */
|
638
|
+
OP_REPEAT_INC_SG, /* search and get in stack */
|
639
|
+
OP_REPEAT_INC_NG_SG, /* search and get in stack (non greedy) */
|
640
|
+
OP_NULL_CHECK_START, /* null loop checker start */
|
641
|
+
OP_NULL_CHECK_END, /* null loop checker end */
|
642
|
+
OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */
|
643
|
+
OP_NULL_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */
|
644
|
+
|
645
|
+
OP_PUSH_POS, /* (?=...) start */
|
646
|
+
OP_POP_POS, /* (?=...) end */
|
647
|
+
OP_PUSH_POS_NOT, /* (?!...) start */
|
648
|
+
OP_FAIL_POS, /* (?!...) end */
|
649
|
+
OP_PUSH_STOP_BT, /* (?>...) start */
|
650
|
+
OP_POP_STOP_BT, /* (?>...) end */
|
651
|
+
OP_LOOK_BEHIND, /* (?<=...) start (no needs end opcode) */
|
652
|
+
OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */
|
653
|
+
OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end */
|
654
|
+
OP_PUSH_ABSENT_POS, /* (?~...) start */
|
655
|
+
OP_ABSENT, /* (?~...) start of inner loop */
|
656
|
+
OP_ABSENT_END, /* (?~...) end */
|
657
|
+
|
658
|
+
OP_CALL, /* \g<name> */
|
659
|
+
OP_RETURN,
|
660
|
+
|
661
|
+
OP_CONDITION,
|
662
|
+
|
663
|
+
OP_STATE_CHECK_PUSH, /* combination explosion check and push */
|
664
|
+
OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump */
|
665
|
+
OP_STATE_CHECK, /* check only */
|
666
|
+
OP_STATE_CHECK_ANYCHAR_STAR,
|
667
|
+
OP_STATE_CHECK_ANYCHAR_ML_STAR,
|
668
|
+
|
669
|
+
/* no need: IS_DYNAMIC_OPTION() == 0 */
|
670
|
+
OP_SET_OPTION_PUSH, /* set option and push recover option */
|
671
|
+
OP_SET_OPTION /* set option */
|
672
|
+
};
|
673
|
+
|
674
|
+
typedef int RelAddrType;
|
675
|
+
typedef int AbsAddrType;
|
676
|
+
typedef int LengthType;
|
677
|
+
typedef int RepeatNumType;
|
678
|
+
typedef short int MemNumType;
|
679
|
+
typedef short int StateCheckNumType;
|
680
|
+
typedef void* PointerType;
|
681
|
+
|
682
|
+
#define SIZE_OPCODE 1
|
683
|
+
#define SIZE_RELADDR (int )sizeof(RelAddrType)
|
684
|
+
#define SIZE_ABSADDR (int )sizeof(AbsAddrType)
|
685
|
+
#define SIZE_LENGTH (int )sizeof(LengthType)
|
686
|
+
#define SIZE_MEMNUM (int )sizeof(MemNumType)
|
687
|
+
#define SIZE_STATE_CHECK_NUM (int )sizeof(StateCheckNumType)
|
688
|
+
#define SIZE_REPEATNUM (int )sizeof(RepeatNumType)
|
689
|
+
#define SIZE_OPTION (int )sizeof(OnigOptionType)
|
690
|
+
#define SIZE_CODE_POINT (int )sizeof(OnigCodePoint)
|
691
|
+
#define SIZE_POINTER (int )sizeof(PointerType)
|
692
|
+
|
693
|
+
|
694
|
+
#define GET_RELADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, RelAddrType)
|
695
|
+
#define GET_ABSADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, AbsAddrType)
|
696
|
+
#define GET_LENGTH_INC(len,p) PLATFORM_GET_INC(len, p, LengthType)
|
697
|
+
#define GET_MEMNUM_INC(num,p) PLATFORM_GET_INC(num, p, MemNumType)
|
698
|
+
#define GET_REPEATNUM_INC(num,p) PLATFORM_GET_INC(num, p, RepeatNumType)
|
699
|
+
#define GET_OPTION_INC(option,p) PLATFORM_GET_INC(option, p, OnigOptionType)
|
700
|
+
#define GET_POINTER_INC(ptr,p) PLATFORM_GET_INC(ptr, p, PointerType)
|
701
|
+
#define GET_STATE_CHECK_NUM_INC(num,p) PLATFORM_GET_INC(num, p, StateCheckNumType)
|
702
|
+
|
703
|
+
/* code point's address must be aligned address. */
|
704
|
+
#define GET_CODE_POINT(code,p) code = *((OnigCodePoint* )(p))
|
705
|
+
#define GET_BYTE_INC(byte,p) do{\
|
706
|
+
byte = *(p);\
|
707
|
+
(p)++;\
|
708
|
+
} while(0)
|
709
|
+
|
710
|
+
|
711
|
+
/* op-code + arg size */
|
712
|
+
#define SIZE_OP_ANYCHAR_STAR SIZE_OPCODE
|
713
|
+
#define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT (SIZE_OPCODE + 1)
|
714
|
+
#define SIZE_OP_JUMP (SIZE_OPCODE + SIZE_RELADDR)
|
715
|
+
#define SIZE_OP_PUSH (SIZE_OPCODE + SIZE_RELADDR)
|
716
|
+
#define SIZE_OP_POP SIZE_OPCODE
|
717
|
+
#define SIZE_OP_PUSH_OR_JUMP_EXACT1 (SIZE_OPCODE + SIZE_RELADDR + 1)
|
718
|
+
#define SIZE_OP_PUSH_IF_PEEK_NEXT (SIZE_OPCODE + SIZE_RELADDR + 1)
|
719
|
+
#define SIZE_OP_REPEAT_INC (SIZE_OPCODE + SIZE_MEMNUM)
|
720
|
+
#define SIZE_OP_REPEAT_INC_NG (SIZE_OPCODE + SIZE_MEMNUM)
|
721
|
+
#define SIZE_OP_PUSH_POS SIZE_OPCODE
|
722
|
+
#define SIZE_OP_PUSH_POS_NOT (SIZE_OPCODE + SIZE_RELADDR)
|
723
|
+
#define SIZE_OP_POP_POS SIZE_OPCODE
|
724
|
+
#define SIZE_OP_FAIL_POS SIZE_OPCODE
|
725
|
+
#define SIZE_OP_SET_OPTION (SIZE_OPCODE + SIZE_OPTION)
|
726
|
+
#define SIZE_OP_SET_OPTION_PUSH (SIZE_OPCODE + SIZE_OPTION)
|
727
|
+
#define SIZE_OP_FAIL SIZE_OPCODE
|
728
|
+
#define SIZE_OP_MEMORY_START (SIZE_OPCODE + SIZE_MEMNUM)
|
729
|
+
#define SIZE_OP_MEMORY_START_PUSH (SIZE_OPCODE + SIZE_MEMNUM)
|
730
|
+
#define SIZE_OP_MEMORY_END_PUSH (SIZE_OPCODE + SIZE_MEMNUM)
|
731
|
+
#define SIZE_OP_MEMORY_END_PUSH_REC (SIZE_OPCODE + SIZE_MEMNUM)
|
732
|
+
#define SIZE_OP_MEMORY_END (SIZE_OPCODE + SIZE_MEMNUM)
|
733
|
+
#define SIZE_OP_MEMORY_END_REC (SIZE_OPCODE + SIZE_MEMNUM)
|
734
|
+
#define SIZE_OP_PUSH_STOP_BT SIZE_OPCODE
|
735
|
+
#define SIZE_OP_POP_STOP_BT SIZE_OPCODE
|
736
|
+
#define SIZE_OP_NULL_CHECK_START (SIZE_OPCODE + SIZE_MEMNUM)
|
737
|
+
#define SIZE_OP_NULL_CHECK_END (SIZE_OPCODE + SIZE_MEMNUM)
|
738
|
+
#define SIZE_OP_LOOK_BEHIND (SIZE_OPCODE + SIZE_LENGTH)
|
739
|
+
#define SIZE_OP_PUSH_LOOK_BEHIND_NOT (SIZE_OPCODE + SIZE_RELADDR + SIZE_LENGTH)
|
740
|
+
#define SIZE_OP_FAIL_LOOK_BEHIND_NOT SIZE_OPCODE
|
741
|
+
#define SIZE_OP_CALL (SIZE_OPCODE + SIZE_ABSADDR)
|
742
|
+
#define SIZE_OP_RETURN SIZE_OPCODE
|
743
|
+
#define SIZE_OP_CONDITION (SIZE_OPCODE + SIZE_MEMNUM + SIZE_RELADDR)
|
744
|
+
#define SIZE_OP_PUSH_ABSENT_POS SIZE_OPCODE
|
745
|
+
#define SIZE_OP_ABSENT (SIZE_OPCODE + SIZE_RELADDR)
|
746
|
+
#define SIZE_OP_ABSENT_END SIZE_OPCODE
|
747
|
+
|
748
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
749
|
+
# define SIZE_OP_STATE_CHECK (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
|
750
|
+
# define SIZE_OP_STATE_CHECK_PUSH (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
|
751
|
+
# define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
|
752
|
+
# define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
|
753
|
+
#endif
|
754
|
+
|
755
|
+
#define MC_ESC(syn) (syn)->meta_char_table.esc
|
756
|
+
#define MC_ANYCHAR(syn) (syn)->meta_char_table.anychar
|
757
|
+
#define MC_ANYTIME(syn) (syn)->meta_char_table.anytime
|
758
|
+
#define MC_ZERO_OR_ONE_TIME(syn) (syn)->meta_char_table.zero_or_one_time
|
759
|
+
#define MC_ONE_OR_MORE_TIME(syn) (syn)->meta_char_table.one_or_more_time
|
760
|
+
#define MC_ANYCHAR_ANYTIME(syn) (syn)->meta_char_table.anychar_anytime
|
761
|
+
|
762
|
+
#define IS_MC_ESC_CODE(code, syn) \
|
763
|
+
((code) == MC_ESC(syn) && \
|
764
|
+
!IS_SYNTAX_OP2((syn), ONIG_SYN_OP2_INEFFECTIVE_ESCAPE))
|
765
|
+
|
766
|
+
|
767
|
+
#define SYN_POSIX_COMMON_OP \
|
768
|
+
( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_POSIX_BRACKET | \
|
769
|
+
ONIG_SYN_OP_DECIMAL_BACKREF | \
|
770
|
+
ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_ASTERISK_ZERO_INF | \
|
771
|
+
ONIG_SYN_OP_LINE_ANCHOR | \
|
772
|
+
ONIG_SYN_OP_ESC_CONTROL_CHARS )
|
773
|
+
|
774
|
+
#define SYN_GNU_REGEX_OP \
|
775
|
+
( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | \
|
776
|
+
ONIG_SYN_OP_POSIX_BRACKET | ONIG_SYN_OP_DECIMAL_BACKREF | \
|
777
|
+
ONIG_SYN_OP_BRACE_INTERVAL | ONIG_SYN_OP_LPAREN_SUBEXP | \
|
778
|
+
ONIG_SYN_OP_VBAR_ALT | \
|
779
|
+
ONIG_SYN_OP_ASTERISK_ZERO_INF | ONIG_SYN_OP_PLUS_ONE_INF | \
|
780
|
+
ONIG_SYN_OP_QMARK_ZERO_ONE | \
|
781
|
+
ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR | ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR | \
|
782
|
+
ONIG_SYN_OP_ESC_W_WORD | \
|
783
|
+
ONIG_SYN_OP_ESC_B_WORD_BOUND | ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END | \
|
784
|
+
ONIG_SYN_OP_ESC_S_WHITE_SPACE | ONIG_SYN_OP_ESC_D_DIGIT | \
|
785
|
+
ONIG_SYN_OP_LINE_ANCHOR )
|
786
|
+
|
787
|
+
#define SYN_GNU_REGEX_BV \
|
788
|
+
( ONIG_SYN_CONTEXT_INDEP_ANCHORS | ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | \
|
789
|
+
ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | ONIG_SYN_ALLOW_INVALID_INTERVAL | \
|
790
|
+
ONIG_SYN_BACKSLASH_ESCAPE_IN_CC | ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC )
|
791
|
+
|
792
|
+
|
793
|
+
#define NCCLASS_FLAGS(cc) ((cc)->flags)
|
794
|
+
#define NCCLASS_FLAG_SET(cc,flag) (NCCLASS_FLAGS(cc) |= (flag))
|
795
|
+
#define NCCLASS_FLAG_CLEAR(cc,flag) (NCCLASS_FLAGS(cc) &= ~(flag))
|
796
|
+
#define IS_NCCLASS_FLAG_ON(cc,flag) ((NCCLASS_FLAGS(cc) & (flag)) != 0)
|
797
|
+
|
798
|
+
/* cclass node */
|
799
|
+
#define FLAG_NCCLASS_NOT (1<<0)
|
800
|
+
|
801
|
+
#define NCCLASS_SET_NOT(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_NOT)
|
802
|
+
#define NCCLASS_CLEAR_NOT(nd) NCCLASS_FLAG_CLEAR(nd, FLAG_NCCLASS_NOT)
|
803
|
+
#define IS_NCCLASS_NOT(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_NOT)
|
804
|
+
|
805
|
+
typedef struct {
|
806
|
+
int type;
|
807
|
+
/* struct _Node* next; */
|
808
|
+
/* unsigned int flags; */
|
809
|
+
} NodeBase;
|
810
|
+
|
811
|
+
typedef struct {
|
812
|
+
NodeBase base;
|
813
|
+
unsigned int flags;
|
814
|
+
BitSet bs;
|
815
|
+
BBuf* mbuf; /* multi-byte info or NULL */
|
816
|
+
} CClassNode;
|
817
|
+
|
818
|
+
typedef intptr_t OnigStackIndex;
|
819
|
+
|
820
|
+
typedef struct _OnigStackType {
|
821
|
+
unsigned int type;
|
822
|
+
union {
|
823
|
+
struct {
|
824
|
+
UChar *pcode; /* byte code position */
|
825
|
+
UChar *pstr; /* string position */
|
826
|
+
UChar *pstr_prev; /* previous char position of pstr */
|
827
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
828
|
+
unsigned int state_check;
|
829
|
+
#endif
|
830
|
+
UChar *pkeep; /* keep pattern position */
|
831
|
+
} state;
|
832
|
+
struct {
|
833
|
+
int count; /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */
|
834
|
+
UChar *pcode; /* byte code position (head of repeated target) */
|
835
|
+
int num; /* repeat id */
|
836
|
+
} repeat;
|
837
|
+
struct {
|
838
|
+
OnigStackIndex si; /* index of stack */
|
839
|
+
} repeat_inc;
|
840
|
+
struct {
|
841
|
+
int num; /* memory num */
|
842
|
+
UChar *pstr; /* start/end position */
|
843
|
+
/* Following information is set, if this stack type is MEM-START */
|
844
|
+
OnigStackIndex start; /* prev. info (for backtrack "(...)*" ) */
|
845
|
+
OnigStackIndex end; /* prev. info (for backtrack "(...)*" ) */
|
846
|
+
} mem;
|
847
|
+
struct {
|
848
|
+
int num; /* null check id */
|
849
|
+
UChar *pstr; /* start position */
|
850
|
+
} null_check;
|
851
|
+
#ifdef USE_SUBEXP_CALL
|
852
|
+
struct {
|
853
|
+
UChar *ret_addr; /* byte code position */
|
854
|
+
int num; /* null check id */
|
855
|
+
UChar *pstr; /* string position */
|
856
|
+
} call_frame;
|
857
|
+
#endif
|
858
|
+
struct {
|
859
|
+
UChar *abs_pstr; /* absent start position */
|
860
|
+
const UChar *end_pstr; /* end position */
|
861
|
+
} absent_pos;
|
862
|
+
} u;
|
863
|
+
} OnigStackType;
|
864
|
+
|
865
|
+
typedef struct {
|
866
|
+
void* stack_p;
|
867
|
+
size_t stack_n;
|
868
|
+
OnigOptionType options;
|
869
|
+
OnigRegion* region;
|
870
|
+
const UChar* start; /* search start position */
|
871
|
+
const UChar* gpos; /* global position (for \G: BEGIN_POSITION) */
|
872
|
+
#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
873
|
+
OnigPosition best_len; /* for ONIG_OPTION_FIND_LONGEST */
|
874
|
+
UChar* best_s;
|
875
|
+
#endif
|
876
|
+
#ifdef USE_COMBINATION_EXPLOSION_CHECK
|
877
|
+
void* state_check_buff;
|
878
|
+
int state_check_buff_size;
|
879
|
+
#endif
|
880
|
+
} OnigMatchArg;
|
881
|
+
|
882
|
+
|
883
|
+
#define IS_CODE_SB_WORD(enc,code) \
|
884
|
+
(ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code))
|
885
|
+
|
886
|
+
typedef struct OnigEndCallListItem {
|
887
|
+
struct OnigEndCallListItem* next;
|
888
|
+
void (*func)(void);
|
889
|
+
} OnigEndCallListItemType;
|
890
|
+
|
891
|
+
extern void onig_add_end_call(void (*func)(void));
|
892
|
+
|
893
|
+
|
894
|
+
#ifdef ONIG_DEBUG
|
895
|
+
|
896
|
+
typedef struct {
|
897
|
+
short int opcode;
|
898
|
+
const char* name;
|
899
|
+
short int arg_type;
|
900
|
+
} OnigOpInfoType;
|
901
|
+
|
902
|
+
extern OnigOpInfoType OnigOpInfo[];
|
903
|
+
|
904
|
+
|
905
|
+
extern void onig_print_compiled_byte_code(FILE* f, UChar* bp, UChar* bpend, UChar** nextp, OnigEncoding enc);
|
906
|
+
|
907
|
+
# ifdef ONIG_DEBUG_STATISTICS
|
908
|
+
extern void onig_statistics_init(void);
|
909
|
+
extern void onig_print_statistics(FILE* f);
|
910
|
+
# endif
|
911
|
+
#endif
|
912
|
+
|
913
|
+
extern UChar* onig_error_code_to_format(OnigPosition code);
|
914
|
+
extern void onig_vsnprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const UChar *fmt, va_list args);
|
915
|
+
extern void onig_snprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const UChar *fmt, ...);
|
916
|
+
extern int onig_bbuf_init(BBuf* buf, OnigDistance size);
|
917
|
+
extern int onig_compile(regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo);
|
918
|
+
#ifdef RUBY
|
919
|
+
extern int onig_compile_ruby(regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo, const char *sourcefile, int sourceline);
|
920
|
+
#endif
|
921
|
+
extern void onig_transfer(regex_t* to, regex_t* from);
|
922
|
+
extern int onig_is_code_in_cc(OnigEncoding enc, OnigCodePoint code, CClassNode* cc);
|
923
|
+
extern int onig_is_code_in_cc_len(int enclen, OnigCodePoint code, CClassNode* cc);
|
924
|
+
|
925
|
+
/* strend hash */
|
926
|
+
typedef void hash_table_type;
|
927
|
+
#ifdef RUBY
|
928
|
+
# include "ruby/st.h"
|
929
|
+
#else
|
930
|
+
# include "st.h"
|
931
|
+
#endif
|
932
|
+
typedef st_data_t hash_data_type;
|
933
|
+
|
934
|
+
extern hash_table_type* onig_st_init_strend_table_with_size(st_index_t size);
|
935
|
+
extern int onig_st_lookup_strend(hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type *value);
|
936
|
+
extern int onig_st_insert_strend(hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type value);
|
937
|
+
|
938
|
+
#ifdef RUBY
|
939
|
+
extern size_t onig_memsize(const regex_t *reg);
|
940
|
+
extern size_t onig_region_memsize(const struct re_registers *regs);
|
941
|
+
#endif
|
942
|
+
|
943
|
+
RUBY_SYMBOL_EXPORT_END
|
944
|
+
|
945
|
+
#endif /* ONIGMO_REGINT_H */
|