source: EcnlProtoTool/trunk/onigmo-6.1.3/src/regint.h@ 331

Last change on this file since 331 was 331, checked in by coas-nagasima, 6 years ago

prototoolに関連するプロジェクトをnewlibからmuslを使うよう変更・更新
ntshellをnewlibの下位の実装から、muslのsyscallの実装に変更・更新
以下のOSSをアップデート
・mruby-1.3.0
・musl-1.1.18
・onigmo-6.1.3
・tcc-0.9.27
以下のOSSを追加
・openssl-1.1.0e
・curl-7.57.0
・zlib-1.2.11
以下のmrbgemsを追加
・iij/mruby-digest
・iij/mruby-env
・iij/mruby-errno
・iij/mruby-iijson
・iij/mruby-ipaddr
・iij/mruby-mock
・iij/mruby-require
・iij/mruby-tls-openssl
  • Property svn:eol-style set to native
  • Property svn:mime-type set to text/x-chdr;charset=UTF-8
File size: 31.9 KB
Line 
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) 2011-2016 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#define USE_SUNDAY_QUICK_SEARCH
85
86#define INIT_MATCH_STACK_SIZE 160
87#define DEFAULT_MATCH_STACK_LIMIT_SIZE 0 /* unlimited */
88#define DEFAULT_PARSE_DEPTH_LIMIT 4096
89
90#define OPT_EXACT_MAXLEN 24
91
92/* check config */
93#if defined(USE_PERL_SUBEXP_CALL) || defined(USE_CAPITAL_P_NAMED_GROUP)
94# if !defined(USE_NAMED_GROUP) || !defined(USE_SUBEXP_CALL)
95# error USE_NAMED_GROUP and USE_SUBEXP_CALL must be defined.
96# endif
97#endif
98
99#if defined(__GNUC__)
100# define ARG_UNUSED __attribute__ ((unused))
101#else
102# define ARG_UNUSED
103#endif
104
105#if !defined(RUBY) && defined(RUBY_EXPORT)
106# define RUBY
107#endif
108#ifdef RUBY
109# ifndef RUBY_DEFINES_H
110# include "ruby/ruby.h"
111# undef xmalloc
112# undef xrealloc
113# undef xcalloc
114# undef xfree
115# endif
116#else /* RUBY */
117# include "config.h"
118# if SIZEOF_LONG_LONG > 0
119# define LONG_LONG long long
120# endif
121#endif /* RUBY */
122
123#include <stdarg.h>
124
125/* */
126/* escape other system UChar definition */
127#ifdef ONIG_ESCAPE_UCHAR_COLLISION
128# undef ONIG_ESCAPE_UCHAR_COLLISION
129#endif
130
131#define USE_WORD_BEGIN_END /* "\<": word-begin, "\>": word-end */
132#ifdef RUBY
133# undef USE_CAPTURE_HISTORY
134#else
135# define USE_CAPTURE_HISTORY
136#endif
137#define USE_VARIABLE_META_CHARS
138#define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
139/* #define USE_COMBINATION_EXPLOSION_CHECK */ /* (X*)* */
140
141
142#ifndef xmalloc
143# define xmalloc malloc
144# define xrealloc realloc
145# define xcalloc calloc
146# define xfree free
147#endif
148
149#ifdef RUBY
150
151# define CHECK_INTERRUPT_IN_MATCH_AT rb_thread_check_ints()
152# define onig_st_init_table st_init_table
153# define onig_st_init_table_with_size st_init_table_with_size
154# define onig_st_init_numtable st_init_numtable
155# define onig_st_init_numtable_with_size st_init_numtable_with_size
156# define onig_st_init_strtable st_init_strtable
157# define onig_st_init_strtable_with_size st_init_strtable_with_size
158# define onig_st_delete st_delete
159# define onig_st_delete_safe st_delete_safe
160# define onig_st_insert st_insert
161# define onig_st_lookup st_lookup
162# define onig_st_foreach st_foreach
163# define onig_st_add_direct st_add_direct
164# define onig_st_free_table st_free_table
165# define onig_st_cleanup_safe st_cleanup_safe
166# define onig_st_copy st_copy
167# define onig_st_nothing_key_clone st_nothing_key_clone
168# define onig_st_nothing_key_free st_nothing_key_free
169# define onig_st_is_member st_is_member
170
171# define USE_UPPER_CASE_TABLE
172#else /* RUBY */
173
174# define CHECK_INTERRUPT_IN_MATCH_AT
175
176# define st_init_table onig_st_init_table
177# define st_init_table_with_size onig_st_init_table_with_size
178# define st_init_numtable onig_st_init_numtable
179# define st_init_numtable_with_size onig_st_init_numtable_with_size
180# define st_init_strtable onig_st_init_strtable
181# define st_init_strtable_with_size onig_st_init_strtable_with_size
182# define st_delete onig_st_delete
183# define st_delete_safe onig_st_delete_safe
184# define st_insert onig_st_insert
185# define st_lookup onig_st_lookup
186# define st_foreach onig_st_foreach
187# define st_add_direct onig_st_add_direct
188# define st_free_table onig_st_free_table
189# define st_cleanup_safe onig_st_cleanup_safe
190# define st_copy onig_st_copy
191# define st_nothing_key_clone onig_st_nothing_key_clone
192# define st_nothing_key_free onig_st_nothing_key_free
193/* */
194# define onig_st_is_member st_is_member
195
196#endif /* RUBY */
197
198#define STATE_CHECK_STRING_THRESHOLD_LEN 7
199#define STATE_CHECK_BUFF_MAX_SIZE 0x4000
200
201#define xmemset memset
202#define xmemcpy memcpy
203#define xmemmove memmove
204
205#if ((defined(RUBY_MSVCRT_VERSION) && RUBY_MSVCRT_VERSION >= 90) \
206 || (!defined(RUBY_MSVCRT_VERSION) && defined(_WIN32))) \
207 && !defined(__GNUC__)
208# define xalloca _alloca
209# define xvsnprintf(buf,size,fmt,args) _vsnprintf_s(buf,size,_TRUNCATE,fmt,args)
210# define xsnprintf sprintf_s
211# define xstrcat(dest,src,size) strcat_s(dest,size,src)
212#else
213# define xalloca alloca
214# define xvsnprintf vsnprintf
215# define xsnprintf snprintf
216# define xstrcat(dest,src,size) strcat(dest,src)
217#endif
218
219#if defined(ONIG_DEBUG_MEMLEAK) && defined(_MSC_VER)
220# define _CRTDBG_MAP_ALLOC
221# include <malloc.h>
222# include <crtdbg.h>
223#endif
224
225#include <stdlib.h>
226
227#if defined(HAVE_ALLOCA_H) && (defined(_AIX) || !defined(__GNUC__))
228# include <alloca.h>
229#endif
230
231#include <string.h>
232
233#include <ctype.h>
234#ifdef HAVE_SYS_TYPES_H
235# include <sys/types.h>
236#endif
237
238#ifdef HAVE_STDINT_H
239# include <stdint.h>
240#endif
241
242#ifdef HAVE_INTTYPES_H
243# include <inttypes.h>
244#endif
245
246#include <stddef.h>
247
248#ifdef _WIN32
249# include <malloc.h> /* for alloca() */
250#endif
251
252#ifdef ONIG_DEBUG
253# include <stdio.h>
254#endif
255
256#ifdef _WIN32
257# if defined(_MSC_VER) && (_MSC_VER < 1300)
258# ifndef _INTPTR_T_DEFINED
259# define _INTPTR_T_DEFINED
260typedef int intptr_t;
261# endif
262# ifndef _UINTPTR_T_DEFINED
263# define _UINTPTR_T_DEFINED
264typedef unsigned int uintptr_t;
265# endif
266# endif
267#endif /* _WIN32 */
268
269#ifndef PRIdPTR
270# ifdef _WIN64
271# define PRIdPTR "I64d"
272# define PRIuPTR "I64u"
273# define PRIxPTR "I64x"
274# else
275# define PRIdPTR "ld"
276# define PRIuPTR "lu"
277# define PRIxPTR "lx"
278# endif
279#endif
280
281#ifndef PRIdPTRDIFF
282# define PRIdPTRDIFF PRIdPTR
283#endif
284
285#include "regenc.h"
286
287RUBY_SYMBOL_EXPORT_BEGIN
288
289#ifdef MIN
290# undef MIN
291#endif
292#ifdef MAX
293# undef MAX
294#endif
295#define MIN(a,b) (((a)>(b))?(b):(a))
296#define MAX(a,b) (((a)<(b))?(b):(a))
297
298#define IS_NULL(p) (((void*)(p)) == (void*)0)
299#define IS_NOT_NULL(p) (((void*)(p)) != (void*)0)
300#define CHECK_NULL_RETURN(p) if (IS_NULL(p)) return NULL
301#define CHECK_NULL_RETURN_MEMERR(p) if (IS_NULL(p)) return ONIGERR_MEMORY
302#define NULL_UCHARP ((UChar* )0)
303
304#define ONIG_LAST_CODE_POINT (~((OnigCodePoint )0))
305
306#ifdef PLATFORM_UNALIGNED_WORD_ACCESS
307
308# define PLATFORM_GET_INC(val,p,type) do{\
309 val = *(type* )p;\
310 (p) += sizeof(type);\
311} while(0)
312
313#else
314
315# define PLATFORM_GET_INC(val,p,type) do{\
316 xmemcpy(&val, (p), sizeof(type));\
317 (p) += sizeof(type);\
318} while(0)
319
320/* sizeof(OnigCodePoint) */
321# define WORD_ALIGNMENT_SIZE SIZEOF_LONG
322
323# define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\
324 (pad_size) = WORD_ALIGNMENT_SIZE \
325 - ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
326 if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\
327} while (0)
328
329# define ALIGNMENT_RIGHT(addr) do {\
330 (addr) += (WORD_ALIGNMENT_SIZE - 1);\
331 (addr) -= ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
332} while (0)
333
334#endif /* PLATFORM_UNALIGNED_WORD_ACCESS */
335
336/* stack pop level */
337#define STACK_POP_LEVEL_FREE 0
338#define STACK_POP_LEVEL_MEM_START 1
339#define STACK_POP_LEVEL_ALL 2
340
341/* optimize flags */
342#define ONIG_OPTIMIZE_NONE 0
343#define ONIG_OPTIMIZE_EXACT 1 /* Slow Search */
344#define ONIG_OPTIMIZE_EXACT_BM 2 /* Boyer Moore Search */
345#define ONIG_OPTIMIZE_EXACT_BM_NOT_REV 3 /* BM (applied to a multibyte string) */
346#define ONIG_OPTIMIZE_EXACT_IC 4 /* Slow Search (ignore case) */
347#define ONIG_OPTIMIZE_MAP 5 /* char map */
348#define ONIG_OPTIMIZE_EXACT_BM_IC 6 /* BM (ignore case) */
349#define ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC 7 /* BM (applied to a multibyte string) (ignore case) */
350
351/* bit status */
352typedef unsigned int BitStatusType;
353
354#define BIT_STATUS_BITS_NUM (sizeof(BitStatusType) * 8)
355#define BIT_STATUS_CLEAR(stats) (stats) = 0
356#define BIT_STATUS_ON_ALL(stats) (stats) = ~((BitStatusType )0)
357#define BIT_STATUS_AT(stats,n) \
358 ((n) < (int )BIT_STATUS_BITS_NUM ? ((stats) & ((BitStatusType )1 << n)) : ((stats) & 1))
359
360#define BIT_STATUS_ON_AT(stats,n) do {\
361 if ((n) < (int )BIT_STATUS_BITS_NUM)\
362 (stats) |= (1 << (n));\
363 else\
364 (stats) |= 1;\
365} while (0)
366
367#define BIT_STATUS_ON_AT_SIMPLE(stats,n) do {\
368 if ((n) < (int )BIT_STATUS_BITS_NUM)\
369 (stats) |= (1 << (n));\
370} while (0)
371
372
373#define INT_MAX_LIMIT ((1UL << (SIZEOF_INT * 8 - 1)) - 1)
374
375#define DIGITVAL(code) ((code) - '0')
376#define ODIGITVAL(code) DIGITVAL(code)
377#define XDIGITVAL(enc,code) \
378 (ONIGENC_IS_CODE_DIGIT(enc,code) ? DIGITVAL(code) \
379 : (ONIGENC_IS_CODE_UPPER(enc,code) ? (code) - 'A' + 10 : (code) - 'a' + 10))
380
381#define IS_SINGLELINE(option) ((option) & ONIG_OPTION_SINGLELINE)
382#define IS_MULTILINE(option) ((option) & ONIG_OPTION_MULTILINE)
383#define IS_IGNORECASE(option) ((option) & ONIG_OPTION_IGNORECASE)
384#define IS_EXTEND(option) ((option) & ONIG_OPTION_EXTEND)
385#define IS_FIND_LONGEST(option) ((option) & ONIG_OPTION_FIND_LONGEST)
386#define IS_FIND_NOT_EMPTY(option) ((option) & ONIG_OPTION_FIND_NOT_EMPTY)
387#define IS_FIND_CONDITION(option) ((option) & \
388 (ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY))
389#define IS_NOTBOL(option) ((option) & ONIG_OPTION_NOTBOL)
390#define IS_NOTEOL(option) ((option) & ONIG_OPTION_NOTEOL)
391#define IS_NOTBOS(option) ((option) & ONIG_OPTION_NOTBOS)
392#define IS_NOTEOS(option) ((option) & ONIG_OPTION_NOTEOS)
393#define IS_ASCII_RANGE(option) ((option) & ONIG_OPTION_ASCII_RANGE)
394#define IS_POSIX_BRACKET_ALL_RANGE(option) ((option) & ONIG_OPTION_POSIX_BRACKET_ALL_RANGE)
395#define IS_WORD_BOUND_ALL_RANGE(option) ((option) & ONIG_OPTION_WORD_BOUND_ALL_RANGE)
396#define IS_NEWLINE_CRLF(option) ((option) & ONIG_OPTION_NEWLINE_CRLF)
397
398/* OP_SET_OPTION is required for these options.
399#define IS_DYNAMIC_OPTION(option) \
400 (((option) & (ONIG_OPTION_MULTILINE | ONIG_OPTION_IGNORECASE)) != 0)
401*/
402/* ignore-case and multibyte status are included in compiled code. */
403#define IS_DYNAMIC_OPTION(option) 0
404
405#define DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag) \
406 ((case_fold_flag) & ~INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR)
407
408#define REPEAT_INFINITE -1
409#define IS_REPEAT_INFINITE(n) ((n) == REPEAT_INFINITE)
410
411/* bitset */
412#define BITS_PER_BYTE 8
413#define SINGLE_BYTE_SIZE (1 << BITS_PER_BYTE)
414#define BITS_IN_ROOM ((int )sizeof(Bits) * BITS_PER_BYTE)
415#define BITSET_SIZE (SINGLE_BYTE_SIZE / BITS_IN_ROOM)
416
417#ifdef PLATFORM_UNALIGNED_WORD_ACCESS
418typedef unsigned int Bits;
419#else
420typedef unsigned char Bits;
421#endif
422typedef Bits BitSet[BITSET_SIZE];
423typedef Bits* BitSetRef;
424
425#define SIZE_BITSET (int )sizeof(BitSet)
426
427#define BITSET_CLEAR(bs) do {\
428 int i;\
429 for (i = 0; i < BITSET_SIZE; i++) { (bs)[i] = 0; } \
430} while (0)
431
432#define BS_ROOM(bs,pos) (bs)[(int )(pos) / BITS_IN_ROOM]
433#define BS_BIT(pos) (1U << ((int )(pos) % BITS_IN_ROOM))
434
435#define BITSET_AT(bs, pos) (BS_ROOM(bs,pos) & BS_BIT(pos))
436#define BITSET_SET_BIT(bs, pos) BS_ROOM(bs,pos) |= BS_BIT(pos)
437#define BITSET_CLEAR_BIT(bs, pos) BS_ROOM(bs,pos) &= ~(BS_BIT(pos))
438#define BITSET_INVERT_BIT(bs, pos) BS_ROOM(bs,pos) ^= BS_BIT(pos)
439
440/* bytes buffer */
441typedef struct _BBuf {
442 UChar* p;
443 unsigned int used;
444 unsigned int alloc;
445} BBuf;
446
447#define BBUF_INIT(buf,size) onig_bbuf_init((BBuf* )(buf), (size))
448
449#define BBUF_SIZE_INC(buf,inc) do{\
450 UChar *tmp;\
451 (buf)->alloc += (inc);\
452 tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
453 if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
454 (buf)->p = tmp;\
455} while (0)
456
457#define BBUF_EXPAND(buf,low) do{\
458 UChar *tmp;\
459 do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\
460 tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
461 if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
462 (buf)->p = tmp;\
463} while (0)
464
465#define BBUF_ENSURE_SIZE(buf,size) do{\
466 unsigned int new_alloc = (buf)->alloc;\
467 while (new_alloc < (unsigned int )(size)) { new_alloc *= 2; }\
468 if ((buf)->alloc != new_alloc) {\
469 UChar *tmp;\
470 tmp = (UChar* )xrealloc((buf)->p, new_alloc);\
471 if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
472 (buf)->p = tmp;\
473 (buf)->alloc = new_alloc;\
474 }\
475} while (0)
476
477#define BBUF_WRITE(buf,pos,bytes,n) do{\
478 int used = (pos) + (int )(n);\
479 if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
480 xmemcpy((buf)->p + (pos), (bytes), (n));\
481 if ((buf)->used < (unsigned int )used) (buf)->used = used;\
482} while (0)
483
484#define BBUF_WRITE1(buf,pos,byte) do{\
485 int used = (pos) + 1;\
486 if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
487 (buf)->p[(pos)] = (UChar )(byte);\
488 if ((buf)->used < (unsigned int )used) (buf)->used = used;\
489} while (0)
490
491#define BBUF_ADD(buf,bytes,n) BBUF_WRITE((buf),(buf)->used,(bytes),(n))
492#define BBUF_ADD1(buf,byte) BBUF_WRITE1((buf),(buf)->used,(byte))
493#define BBUF_GET_ADD_ADDRESS(buf) ((buf)->p + (buf)->used)
494#define BBUF_GET_OFFSET_POS(buf) ((buf)->used)
495
496/* from < to */
497#define BBUF_MOVE_RIGHT(buf,from,to,n) do {\
498 if ((unsigned int )((to)+(n)) > (buf)->alloc) BBUF_EXPAND((buf),(to) + (n));\
499 xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
500 if ((unsigned int )((to)+(n)) > (buf)->used) (buf)->used = (to) + (n);\
501} while (0)
502
503/* from > to */
504#define BBUF_MOVE_LEFT(buf,from,to,n) do {\
505 xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
506} while (0)
507
508/* from > to */
509#define BBUF_MOVE_LEFT_REDUCE(buf,from,to) do {\
510 xmemmove((buf)->p + (to), (buf)->p + (from), (buf)->used - (from));\
511 (buf)->used -= (from - to);\
512} while (0)
513
514#define BBUF_INSERT(buf,pos,bytes,n) do {\
515 if (pos >= (buf)->used) {\
516 BBUF_WRITE(buf,pos,bytes,n);\
517 }\
518 else {\
519 BBUF_MOVE_RIGHT((buf),(pos),(pos) + (n),((buf)->used - (pos)));\
520 xmemcpy((buf)->p + (pos), (bytes), (n));\
521 }\
522} while (0)
523
524#define BBUF_GET_BYTE(buf, pos) (buf)->p[(pos)]
525
526
527#define ANCHOR_BEGIN_BUF (1<<0)
528#define ANCHOR_BEGIN_LINE (1<<1)
529#define ANCHOR_BEGIN_POSITION (1<<2)
530#define ANCHOR_END_BUF (1<<3)
531#define ANCHOR_SEMI_END_BUF (1<<4)
532#define ANCHOR_END_LINE (1<<5)
533
534#define ANCHOR_WORD_BOUND (1<<6)
535#define ANCHOR_NOT_WORD_BOUND (1<<7)
536#define ANCHOR_WORD_BEGIN (1<<8)
537#define ANCHOR_WORD_END (1<<9)
538#define ANCHOR_PREC_READ (1<<10)
539#define ANCHOR_PREC_READ_NOT (1<<11)
540#define ANCHOR_LOOK_BEHIND (1<<12)
541#define ANCHOR_LOOK_BEHIND_NOT (1<<13)
542
543#define ANCHOR_ANYCHAR_STAR (1<<14) /* ".*" optimize info */
544#define ANCHOR_ANYCHAR_STAR_ML (1<<15) /* ".*" optimize info (multi-line) */
545
546#define ANCHOR_KEEP (1<<16)
547
548/* operation code */
549enum OpCode {
550 OP_FINISH = 0, /* matching process terminator (no more alternative) */
551 OP_END = 1, /* pattern code terminator (success end) */
552
553 OP_EXACT1 = 2, /* single byte, N = 1 */
554 OP_EXACT2, /* single byte, N = 2 */
555 OP_EXACT3, /* single byte, N = 3 */
556 OP_EXACT4, /* single byte, N = 4 */
557 OP_EXACT5, /* single byte, N = 5 */
558 OP_EXACTN, /* single byte */
559 OP_EXACTMB2N1, /* mb-length = 2 N = 1 */
560 OP_EXACTMB2N2, /* mb-length = 2 N = 2 */
561 OP_EXACTMB2N3, /* mb-length = 2 N = 3 */
562 OP_EXACTMB2N, /* mb-length = 2 */
563 OP_EXACTMB3N, /* mb-length = 3 */
564 OP_EXACTMBN, /* other length */
565
566 OP_EXACT1_IC, /* single byte, N = 1, ignore case */
567 OP_EXACTN_IC, /* single byte, ignore case */
568
569 OP_CCLASS,
570 OP_CCLASS_MB,
571 OP_CCLASS_MIX,
572 OP_CCLASS_NOT,
573 OP_CCLASS_MB_NOT,
574 OP_CCLASS_MIX_NOT,
575
576 OP_ANYCHAR, /* "." */
577 OP_ANYCHAR_ML, /* "." multi-line */
578 OP_ANYCHAR_STAR, /* ".*" */
579 OP_ANYCHAR_ML_STAR, /* ".*" multi-line */
580 OP_ANYCHAR_STAR_PEEK_NEXT,
581 OP_ANYCHAR_ML_STAR_PEEK_NEXT,
582
583 OP_WORD,
584 OP_NOT_WORD,
585 OP_WORD_BOUND,
586 OP_NOT_WORD_BOUND,
587 OP_WORD_BEGIN,
588 OP_WORD_END,
589
590 OP_ASCII_WORD,
591 OP_NOT_ASCII_WORD,
592 OP_ASCII_WORD_BOUND,
593 OP_NOT_ASCII_WORD_BOUND,
594 OP_ASCII_WORD_BEGIN,
595 OP_ASCII_WORD_END,
596
597 OP_BEGIN_BUF,
598 OP_END_BUF,
599 OP_BEGIN_LINE,
600 OP_END_LINE,
601 OP_SEMI_END_BUF,
602 OP_BEGIN_POSITION,
603
604 OP_BACKREF1,
605 OP_BACKREF2,
606 OP_BACKREFN,
607 OP_BACKREFN_IC,
608 OP_BACKREF_MULTI,
609 OP_BACKREF_MULTI_IC,
610 OP_BACKREF_WITH_LEVEL, /* \k<xxx+n>, \k<xxx-n> */
611
612 OP_MEMORY_START,
613 OP_MEMORY_START_PUSH, /* push back-tracker to stack */
614 OP_MEMORY_END_PUSH, /* push back-tracker to stack */
615 OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */
616 OP_MEMORY_END,
617 OP_MEMORY_END_REC, /* push marker to stack */
618
619 OP_KEEP,
620
621 OP_FAIL, /* pop stack and move */
622 OP_JUMP,
623 OP_PUSH,
624 OP_POP,
625 OP_PUSH_OR_JUMP_EXACT1, /* if match exact then push, else jump. */
626 OP_PUSH_IF_PEEK_NEXT, /* if match exact then push, else none. */
627 OP_REPEAT, /* {n,m} */
628 OP_REPEAT_NG, /* {n,m}? (non greedy) */
629 OP_REPEAT_INC,
630 OP_REPEAT_INC_NG, /* non greedy */
631 OP_REPEAT_INC_SG, /* search and get in stack */
632 OP_REPEAT_INC_NG_SG, /* search and get in stack (non greedy) */
633 OP_NULL_CHECK_START, /* null loop checker start */
634 OP_NULL_CHECK_END, /* null loop checker end */
635 OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */
636 OP_NULL_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */
637
638 OP_PUSH_POS, /* (?=...) start */
639 OP_POP_POS, /* (?=...) end */
640 OP_PUSH_POS_NOT, /* (?!...) start */
641 OP_FAIL_POS, /* (?!...) end */
642 OP_PUSH_STOP_BT, /* (?>...) start */
643 OP_POP_STOP_BT, /* (?>...) end */
644 OP_LOOK_BEHIND, /* (?<=...) start (no needs end opcode) */
645 OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */
646 OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end */
647 OP_PUSH_ABSENT_POS, /* (?~...) start */
648 OP_ABSENT, /* (?~...) start of inner loop */
649 OP_ABSENT_END, /* (?~...) end */
650
651 OP_CALL, /* \g<name> */
652 OP_RETURN,
653
654 OP_CONDITION,
655
656 OP_STATE_CHECK_PUSH, /* combination explosion check and push */
657 OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump */
658 OP_STATE_CHECK, /* check only */
659 OP_STATE_CHECK_ANYCHAR_STAR,
660 OP_STATE_CHECK_ANYCHAR_ML_STAR,
661
662 /* no need: IS_DYNAMIC_OPTION() == 0 */
663 OP_SET_OPTION_PUSH, /* set option and push recover option */
664 OP_SET_OPTION /* set option */
665};
666
667typedef int RelAddrType;
668typedef int AbsAddrType;
669typedef int LengthType;
670typedef int RepeatNumType;
671typedef short int MemNumType;
672typedef short int StateCheckNumType;
673typedef void* PointerType;
674
675#define SIZE_OPCODE 1
676#define SIZE_RELADDR (int )sizeof(RelAddrType)
677#define SIZE_ABSADDR (int )sizeof(AbsAddrType)
678#define SIZE_LENGTH (int )sizeof(LengthType)
679#define SIZE_MEMNUM (int )sizeof(MemNumType)
680#define SIZE_STATE_CHECK_NUM (int )sizeof(StateCheckNumType)
681#define SIZE_REPEATNUM (int )sizeof(RepeatNumType)
682#define SIZE_OPTION (int )sizeof(OnigOptionType)
683#define SIZE_CODE_POINT (int )sizeof(OnigCodePoint)
684#define SIZE_POINTER (int )sizeof(PointerType)
685
686
687#define GET_RELADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, RelAddrType)
688#define GET_ABSADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, AbsAddrType)
689#define GET_LENGTH_INC(len,p) PLATFORM_GET_INC(len, p, LengthType)
690#define GET_MEMNUM_INC(num,p) PLATFORM_GET_INC(num, p, MemNumType)
691#define GET_REPEATNUM_INC(num,p) PLATFORM_GET_INC(num, p, RepeatNumType)
692#define GET_OPTION_INC(option,p) PLATFORM_GET_INC(option, p, OnigOptionType)
693#define GET_POINTER_INC(ptr,p) PLATFORM_GET_INC(ptr, p, PointerType)
694#define GET_STATE_CHECK_NUM_INC(num,p) PLATFORM_GET_INC(num, p, StateCheckNumType)
695
696/* code point's address must be aligned address. */
697#define GET_CODE_POINT(code,p) code = *((OnigCodePoint* )(p))
698#define GET_BYTE_INC(byte,p) do{\
699 byte = *(p);\
700 (p)++;\
701} while(0)
702
703
704/* op-code + arg size */
705#define SIZE_OP_ANYCHAR_STAR SIZE_OPCODE
706#define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT (SIZE_OPCODE + 1)
707#define SIZE_OP_JUMP (SIZE_OPCODE + SIZE_RELADDR)
708#define SIZE_OP_PUSH (SIZE_OPCODE + SIZE_RELADDR)
709#define SIZE_OP_POP SIZE_OPCODE
710#define SIZE_OP_PUSH_OR_JUMP_EXACT1 (SIZE_OPCODE + SIZE_RELADDR + 1)
711#define SIZE_OP_PUSH_IF_PEEK_NEXT (SIZE_OPCODE + SIZE_RELADDR + 1)
712#define SIZE_OP_REPEAT_INC (SIZE_OPCODE + SIZE_MEMNUM)
713#define SIZE_OP_REPEAT_INC_NG (SIZE_OPCODE + SIZE_MEMNUM)
714#define SIZE_OP_PUSH_POS SIZE_OPCODE
715#define SIZE_OP_PUSH_POS_NOT (SIZE_OPCODE + SIZE_RELADDR)
716#define SIZE_OP_POP_POS SIZE_OPCODE
717#define SIZE_OP_FAIL_POS SIZE_OPCODE
718#define SIZE_OP_SET_OPTION (SIZE_OPCODE + SIZE_OPTION)
719#define SIZE_OP_SET_OPTION_PUSH (SIZE_OPCODE + SIZE_OPTION)
720#define SIZE_OP_FAIL SIZE_OPCODE
721#define SIZE_OP_MEMORY_START (SIZE_OPCODE + SIZE_MEMNUM)
722#define SIZE_OP_MEMORY_START_PUSH (SIZE_OPCODE + SIZE_MEMNUM)
723#define SIZE_OP_MEMORY_END_PUSH (SIZE_OPCODE + SIZE_MEMNUM)
724#define SIZE_OP_MEMORY_END_PUSH_REC (SIZE_OPCODE + SIZE_MEMNUM)
725#define SIZE_OP_MEMORY_END (SIZE_OPCODE + SIZE_MEMNUM)
726#define SIZE_OP_MEMORY_END_REC (SIZE_OPCODE + SIZE_MEMNUM)
727#define SIZE_OP_PUSH_STOP_BT SIZE_OPCODE
728#define SIZE_OP_POP_STOP_BT SIZE_OPCODE
729#define SIZE_OP_NULL_CHECK_START (SIZE_OPCODE + SIZE_MEMNUM)
730#define SIZE_OP_NULL_CHECK_END (SIZE_OPCODE + SIZE_MEMNUM)
731#define SIZE_OP_LOOK_BEHIND (SIZE_OPCODE + SIZE_LENGTH)
732#define SIZE_OP_PUSH_LOOK_BEHIND_NOT (SIZE_OPCODE + SIZE_RELADDR + SIZE_LENGTH)
733#define SIZE_OP_FAIL_LOOK_BEHIND_NOT SIZE_OPCODE
734#define SIZE_OP_CALL (SIZE_OPCODE + SIZE_ABSADDR)
735#define SIZE_OP_RETURN SIZE_OPCODE
736#define SIZE_OP_CONDITION (SIZE_OPCODE + SIZE_MEMNUM + SIZE_RELADDR)
737#define SIZE_OP_PUSH_ABSENT_POS SIZE_OPCODE
738#define SIZE_OP_ABSENT (SIZE_OPCODE + SIZE_RELADDR)
739#define SIZE_OP_ABSENT_END SIZE_OPCODE
740
741#ifdef USE_COMBINATION_EXPLOSION_CHECK
742# define SIZE_OP_STATE_CHECK (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
743# define SIZE_OP_STATE_CHECK_PUSH (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
744# define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
745# define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
746#endif
747
748#define MC_ESC(syn) (syn)->meta_char_table.esc
749#define MC_ANYCHAR(syn) (syn)->meta_char_table.anychar
750#define MC_ANYTIME(syn) (syn)->meta_char_table.anytime
751#define MC_ZERO_OR_ONE_TIME(syn) (syn)->meta_char_table.zero_or_one_time
752#define MC_ONE_OR_MORE_TIME(syn) (syn)->meta_char_table.one_or_more_time
753#define MC_ANYCHAR_ANYTIME(syn) (syn)->meta_char_table.anychar_anytime
754
755#define IS_MC_ESC_CODE(code, syn) \
756 ((code) == MC_ESC(syn) && \
757 !IS_SYNTAX_OP2((syn), ONIG_SYN_OP2_INEFFECTIVE_ESCAPE))
758
759
760#define SYN_POSIX_COMMON_OP \
761 ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_POSIX_BRACKET | \
762 ONIG_SYN_OP_DECIMAL_BACKREF | \
763 ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_ASTERISK_ZERO_INF | \
764 ONIG_SYN_OP_LINE_ANCHOR | \
765 ONIG_SYN_OP_ESC_CONTROL_CHARS )
766
767#define SYN_GNU_REGEX_OP \
768 ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | \
769 ONIG_SYN_OP_POSIX_BRACKET | ONIG_SYN_OP_DECIMAL_BACKREF | \
770 ONIG_SYN_OP_BRACE_INTERVAL | ONIG_SYN_OP_LPAREN_SUBEXP | \
771 ONIG_SYN_OP_VBAR_ALT | \
772 ONIG_SYN_OP_ASTERISK_ZERO_INF | ONIG_SYN_OP_PLUS_ONE_INF | \
773 ONIG_SYN_OP_QMARK_ZERO_ONE | \
774 ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR | ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR | \
775 ONIG_SYN_OP_ESC_W_WORD | \
776 ONIG_SYN_OP_ESC_B_WORD_BOUND | ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END | \
777 ONIG_SYN_OP_ESC_S_WHITE_SPACE | ONIG_SYN_OP_ESC_D_DIGIT | \
778 ONIG_SYN_OP_LINE_ANCHOR )
779
780#define SYN_GNU_REGEX_BV \
781 ( ONIG_SYN_CONTEXT_INDEP_ANCHORS | ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | \
782 ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | ONIG_SYN_ALLOW_INVALID_INTERVAL | \
783 ONIG_SYN_BACKSLASH_ESCAPE_IN_CC | ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC )
784
785
786#define NCCLASS_FLAGS(cc) ((cc)->flags)
787#define NCCLASS_FLAG_SET(cc,flag) (NCCLASS_FLAGS(cc) |= (flag))
788#define NCCLASS_FLAG_CLEAR(cc,flag) (NCCLASS_FLAGS(cc) &= ~(flag))
789#define IS_NCCLASS_FLAG_ON(cc,flag) ((NCCLASS_FLAGS(cc) & (flag)) != 0)
790
791/* cclass node */
792#define FLAG_NCCLASS_NOT (1<<0)
793
794#define NCCLASS_SET_NOT(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_NOT)
795#define NCCLASS_CLEAR_NOT(nd) NCCLASS_FLAG_CLEAR(nd, FLAG_NCCLASS_NOT)
796#define IS_NCCLASS_NOT(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_NOT)
797
798typedef struct {
799 int type;
800 /* struct _Node* next; */
801 /* unsigned int flags; */
802} NodeBase;
803
804typedef struct {
805 NodeBase base;
806 unsigned int flags;
807 BitSet bs;
808 BBuf* mbuf; /* multi-byte info or NULL */
809} CClassNode;
810
811typedef intptr_t OnigStackIndex;
812
813typedef struct _OnigStackType {
814 unsigned int type;
815 union {
816 struct {
817 UChar *pcode; /* byte code position */
818 UChar *pstr; /* string position */
819 UChar *pstr_prev; /* previous char position of pstr */
820#ifdef USE_COMBINATION_EXPLOSION_CHECK
821 unsigned int state_check;
822#endif
823 UChar *pkeep; /* keep pattern position */
824 } state;
825 struct {
826 int count; /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */
827 UChar *pcode; /* byte code position (head of repeated target) */
828 int num; /* repeat id */
829 } repeat;
830 struct {
831 OnigStackIndex si; /* index of stack */
832 } repeat_inc;
833 struct {
834 int num; /* memory num */
835 UChar *pstr; /* start/end position */
836 /* Following information is set, if this stack type is MEM-START */
837 OnigStackIndex start; /* prev. info (for backtrack "(...)*" ) */
838 OnigStackIndex end; /* prev. info (for backtrack "(...)*" ) */
839 } mem;
840 struct {
841 int num; /* null check id */
842 UChar *pstr; /* start position */
843 } null_check;
844#ifdef USE_SUBEXP_CALL
845 struct {
846 UChar *ret_addr; /* byte code position */
847 int num; /* null check id */
848 UChar *pstr; /* string position */
849 } call_frame;
850#endif
851 struct {
852 UChar *abs_pstr; /* absent start position */
853 const UChar *end_pstr; /* end position */
854 } absent_pos;
855 } u;
856} OnigStackType;
857
858typedef struct {
859 void* stack_p;
860 size_t stack_n;
861 OnigOptionType options;
862 OnigRegion* region;
863 const UChar* start; /* search start position */
864 const UChar* gpos; /* global position (for \G: BEGIN_POSITION) */
865#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
866 OnigPosition best_len; /* for ONIG_OPTION_FIND_LONGEST */
867 UChar* best_s;
868#endif
869#ifdef USE_COMBINATION_EXPLOSION_CHECK
870 void* state_check_buff;
871 int state_check_buff_size;
872#endif
873} OnigMatchArg;
874
875
876#define IS_CODE_SB_WORD(enc,code) \
877 (ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code))
878
879typedef struct OnigEndCallListItem {
880 struct OnigEndCallListItem* next;
881 void (*func)(void);
882} OnigEndCallListItemType;
883
884extern void onig_add_end_call(void (*func)(void));
885
886
887#ifdef ONIG_DEBUG
888
889typedef struct {
890 short int opcode;
891 const char* name;
892 short int arg_type;
893} OnigOpInfoType;
894
895extern OnigOpInfoType OnigOpInfo[];
896
897
898extern void onig_print_compiled_byte_code(FILE* f, UChar* bp, UChar* bpend, UChar** nextp, OnigEncoding enc);
899
900# ifdef ONIG_DEBUG_STATISTICS
901extern void onig_statistics_init(void);
902extern void onig_print_statistics(FILE* f);
903# endif
904#endif
905
906extern UChar* onig_error_code_to_format(OnigPosition code);
907extern void onig_vsnprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const UChar *fmt, va_list args);
908extern void onig_snprintf_with_pattern(UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const UChar *fmt, ...);
909extern int onig_bbuf_init(BBuf* buf, OnigDistance size);
910extern int onig_compile(regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo);
911#ifdef RUBY
912extern int onig_compile_ruby(regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo, const char *sourcefile, int sourceline);
913#endif
914extern void onig_transfer(regex_t* to, regex_t* from);
915extern int onig_is_code_in_cc(OnigEncoding enc, OnigCodePoint code, CClassNode* cc);
916extern int onig_is_code_in_cc_len(int enclen, OnigCodePoint code, CClassNode* cc);
917
918/* strend hash */
919typedef void hash_table_type;
920#ifdef RUBY
921# include "ruby/st.h"
922#else
923# include "st.h"
924#endif
925typedef st_data_t hash_data_type;
926
927extern hash_table_type* onig_st_init_strend_table_with_size(st_index_t size);
928extern int onig_st_lookup_strend(hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type *value);
929extern int onig_st_insert_strend(hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type value);
930
931#ifdef RUBY
932extern size_t onig_memsize(const regex_t *reg);
933extern size_t onig_region_memsize(const struct re_registers *regs);
934#endif
935
936RUBY_SYMBOL_EXPORT_END
937
938#endif /* ONIGMO_REGINT_H */
Note: See TracBrowser for help on using the repository browser.