1 #ifndef ONIGMO_REGINT_H
2 #define ONIGMO_REGINT_H
3 /**********************************************************************
4 regint.h - Onigmo (Oniguruma-mod) (regular expression library)
5 **********************************************************************/
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>
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
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.
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
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_MATCH_CACHE */
39 /* #define ONIG_DEBUG_MEMLEAK */
40 /* #define ONIG_DONT_OPTIMIZE */
42 /* for byte-code statistical data. */
43 /* #define ONIG_DEBUG_STATISTICS */
45 /* enable the match optimization by using a cache. */
46 #define USE_MATCH_CACHE
48 #if defined(ONIG_DEBUG_PARSE_TREE) || defined(ONIG_DEBUG_MATCH) || \
49 defined(ONIG_DEBUG_SEARCH) || defined(ONIG_DEBUG_COMPILE) || \
50 defined(ONIG_DEBUG_STATISTICS) || defined(ONIG_DEBUG_MEMLEAK)
56 /* __POWERPC__ added to accommodate Darwin case. */
57 #ifndef UNALIGNED_WORD_ACCESS
58 # if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \
59 defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD64) || \
60 defined(__powerpc64__) || defined(__POWERPC__) || defined(__aarch64__) || \
62 # define UNALIGNED_WORD_ACCESS 1
64 # define UNALIGNED_WORD_ACCESS 0
68 #if UNALIGNED_WORD_ACCESS
69 # define PLATFORM_UNALIGNED_WORD_ACCESS
74 #define USE_NAMED_GROUP
75 #define USE_SUBEXP_CALL
76 #define USE_PERL_SUBEXP_CALL
77 #define USE_CAPITAL_P_NAMED_GROUP
78 #define USE_BACKREF_WITH_LEVEL /* \k<name+n>, \k<name-n> */
79 #define USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT /* /(?:()|())*\2/ */
80 #define USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE /* /\n$/ =~ "\n" */
81 #define USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR
82 /* !!! moved to regenc.h. */ /* #define USE_CRNL_AS_LINE_TERMINATOR */
83 #define USE_NO_INVALID_QUANTIFIER
86 /* #define USE_OP_PUSH_OR_JUMP_EXACT */
87 #define USE_QTFR_PEEK_NEXT
88 #define USE_ST_LIBRARY
89 #define USE_SUNDAY_QUICK_SEARCH
91 #define INIT_MATCH_STACK_SIZE 160
92 #define DEFAULT_MATCH_STACK_LIMIT_SIZE 0 /* unlimited */
93 #define DEFAULT_PARSE_DEPTH_LIMIT 4096
95 #define OPT_EXACT_MAXLEN 24
98 #if defined(USE_PERL_SUBEXP_CALL) || defined(USE_CAPITAL_P_NAMED_GROUP)
99 # if !defined(USE_NAMED_GROUP) || !defined(USE_SUBEXP_CALL)
100 # error USE_NAMED_GROUP and USE_SUBEXP_CALL must be defined.
104 #if defined(__GNUC__)
105 # define ARG_UNUSED __attribute__ ((unused))
110 #if !defined(RUBY) && defined(RUBY_EXPORT)
114 # ifndef RUBY_DEFINES_H
115 # include "ruby/ruby.h"
123 # if SIZEOF_LONG_LONG > 0
124 # define LONG_LONG long long
131 /* escape other system UChar definition */
132 #ifdef ONIG_ESCAPE_UCHAR_COLLISION
133 # undef ONIG_ESCAPE_UCHAR_COLLISION
136 #define USE_WORD_BEGIN_END /* "\<": word-begin, "\>": word-end */
138 # undef USE_CAPTURE_HISTORY
140 # define USE_CAPTURE_HISTORY
142 #define USE_VARIABLE_META_CHARS
143 #define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
144 /* #define USE_COMBINATION_EXPLOSION_CHECK */ /* (X*)* */
148 # define xmalloc malloc
149 # define xrealloc realloc
150 # define xcalloc calloc
156 # define CHECK_INTERRUPT_IN_MATCH_AT do { \
158 if (msa->counter >= 128) { \
160 if (rb_reg_timeout_p(reg, &msa->end_time)) { \
163 rb_thread_check_ints(); \
166 # define onig_st_init_table st_init_table
167 # define onig_st_init_table_with_size st_init_table_with_size
168 # define onig_st_init_numtable st_init_numtable
169 # define onig_st_init_numtable_with_size st_init_numtable_with_size
170 # define onig_st_init_strtable st_init_strtable
171 # define onig_st_init_strtable_with_size st_init_strtable_with_size
172 # define onig_st_delete st_delete
173 # define onig_st_delete_safe st_delete_safe
174 # define onig_st_insert st_insert
175 # define onig_st_lookup st_lookup
176 # define onig_st_foreach st_foreach
177 # define onig_st_add_direct st_add_direct
178 # define onig_st_free_table st_free_table
179 # define onig_st_cleanup_safe st_cleanup_safe
180 # define onig_st_copy st_copy
181 # define onig_st_nothing_key_clone st_nothing_key_clone
182 # define onig_st_nothing_key_free st_nothing_key_free
183 # define onig_st_is_member st_is_member
185 # define USE_UPPER_CASE_TABLE
188 # define CHECK_INTERRUPT_IN_MATCH_AT
190 # define st_init_table onig_st_init_table
191 # define st_init_table_with_size onig_st_init_table_with_size
192 # define st_init_numtable onig_st_init_numtable
193 # define st_init_numtable_with_size onig_st_init_numtable_with_size
194 # define st_init_strtable onig_st_init_strtable
195 # define st_init_strtable_with_size onig_st_init_strtable_with_size
196 # define st_delete onig_st_delete
197 # define st_delete_safe onig_st_delete_safe
198 # define st_insert onig_st_insert
199 # define st_lookup onig_st_lookup
200 # define st_foreach onig_st_foreach
201 # define st_add_direct onig_st_add_direct
202 # define st_free_table onig_st_free_table
203 # define st_cleanup_safe onig_st_cleanup_safe
204 # define st_copy onig_st_copy
205 # define st_nothing_key_clone onig_st_nothing_key_clone
206 # define st_nothing_key_free onig_st_nothing_key_free
208 # define onig_st_is_member st_is_member
212 #define STATE_CHECK_STRING_THRESHOLD_LEN 7
213 #define STATE_CHECK_BUFF_MAX_SIZE 0x4000
215 #define xmemset memset
216 #define xmemcpy memcpy
217 #define xmemmove memmove
219 #if ((defined(RUBY_MSVCRT_VERSION) && RUBY_MSVCRT_VERSION >= 90) \
220 || (!defined(RUBY_MSVCRT_VERSION) && defined(_WIN32))) \
221 && !defined(__GNUC__)
222 # define xalloca _alloca
223 # define xvsnprintf(buf,size,fmt,args) _vsnprintf_s(buf,size,_TRUNCATE,fmt,args)
224 # define xsnprintf sprintf_s
225 # define xstrcat(dest,src,size) strcat_s(dest,size,src)
227 # define xalloca alloca
228 # define xvsnprintf vsnprintf
229 # define xsnprintf snprintf
230 # define xstrcat(dest,src,size) strcat(dest,src)
233 #if defined(ONIG_DEBUG_MEMLEAK) && defined(_MSC_VER)
234 # define _CRTDBG_MAP_ALLOC
241 #if defined(HAVE_ALLOCA_H) && (defined(_AIX) || !defined(__GNUC__))
248 #ifdef HAVE_SYS_TYPES_H
249 # include <sys/types.h>
256 #ifdef HAVE_INTTYPES_H
257 # include <inttypes.h>
263 # include <malloc.h> /* for alloca() */
271 # if defined(_MSC_VER) && (_MSC_VER < 1300)
272 # ifndef _INTPTR_T_DEFINED
273 # define _INTPTR_T_DEFINED
274 typedef int intptr_t;
276 # ifndef _UINTPTR_T_DEFINED
277 # define _UINTPTR_T_DEFINED
278 typedef unsigned int uintptr_t;
285 # define PRIdPTR "I64d"
286 # define PRIuPTR "I64u"
287 # define PRIxPTR "I64x"
289 # define PRIdPTR "ld"
290 # define PRIuPTR "lu"
291 # define PRIxPTR "lx"
296 # define PRIdPTRDIFF PRIdPTR
301 RUBY_SYMBOL_EXPORT_BEGIN
309 #define MIN(a,b) (((a)>(b))?(b):(a))
310 #define MAX(a,b) (((a)<(b))?(b):(a))
312 #define IS_NULL(p) (((void*)(p)) == (void*)0)
313 #define IS_NOT_NULL(p) (((void*)(p)) != (void*)0)
314 #define CHECK_NULL_RETURN(p) if (IS_NULL(p)) return NULL
315 #define CHECK_NULL_RETURN_MEMERR(p) if (IS_NULL(p)) return ONIGERR_MEMORY
316 #define NULL_UCHARP ((UChar* )0)
318 #define ONIG_LAST_CODE_POINT (~((OnigCodePoint )0))
320 #define PLATFORM_GET_INC_ARGUMENTS_ASSERT(val, type) \
321 ((void)sizeof(char[2 * (sizeof(val) == sizeof(type)) - 1]))
323 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
325 # define PLATFORM_GET_INC(val,p,type) do{\
326 PLATFORM_GET_INC_ARGUMENTS_ASSERT(val, type);\
328 (p) += sizeof(type);\
333 # define PLATFORM_GET_INC(val,p,type) do{\
334 PLATFORM_GET_INC_ARGUMENTS_ASSERT(val, type);\
335 type platform_get_value;\
336 xmemcpy(&platform_get_value, (p), sizeof(type));\
337 val = platform_get_value;\
338 (p) += sizeof(type);\
341 /* sizeof(OnigCodePoint) */
342 # define WORD_ALIGNMENT_SIZE SIZEOF_LONG
344 # define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\
345 (pad_size) = WORD_ALIGNMENT_SIZE \
346 - ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
347 if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\
350 # define ALIGNMENT_RIGHT(addr) do {\
351 (addr) += (WORD_ALIGNMENT_SIZE - 1);\
352 (addr) -= ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
355 #endif /* PLATFORM_UNALIGNED_WORD_ACCESS */
357 /* stack pop level */
358 #define STACK_POP_LEVEL_FREE 0
359 #define STACK_POP_LEVEL_MEM_START 1
360 #define STACK_POP_LEVEL_ALL 2
363 #define ONIG_OPTIMIZE_NONE 0
364 #define ONIG_OPTIMIZE_EXACT 1 /* Slow Search */
365 #define ONIG_OPTIMIZE_EXACT_BM 2 /* Boyer Moore Search */
366 #define ONIG_OPTIMIZE_EXACT_BM_NOT_REV 3 /* BM (applied to a multibyte string) */
367 #define ONIG_OPTIMIZE_EXACT_IC 4 /* Slow Search (ignore case) */
368 #define ONIG_OPTIMIZE_MAP 5 /* char map */
369 #define ONIG_OPTIMIZE_EXACT_BM_IC 6 /* BM (ignore case) */
370 #define ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC 7 /* BM (applied to a multibyte string) (ignore case) */
373 typedef unsigned int BitStatusType
;
375 #define BIT_STATUS_BITS_NUM (sizeof(BitStatusType) * 8)
376 #define BIT_STATUS_CLEAR(stats) (stats) = 0
377 #define BIT_STATUS_ON_ALL(stats) (stats) = ~((BitStatusType )0)
378 #define BIT_STATUS_AT(stats,n) \
379 ((n) < (int )BIT_STATUS_BITS_NUM ? ((stats) & ((BitStatusType )1 << n)) : ((stats) & 1))
381 #define BIT_STATUS_ON_AT(stats,n) do {\
382 if ((n) < (int )BIT_STATUS_BITS_NUM)\
383 (stats) |= (1 << (n));\
388 #define BIT_STATUS_ON_AT_SIMPLE(stats,n) do {\
389 if ((n) < (int )BIT_STATUS_BITS_NUM)\
390 (stats) |= (1 << (n));\
394 #define INT_MAX_LIMIT ((1UL << (SIZEOF_INT * 8 - 1)) - 1)
395 #define LONG_MAX_LIMIT ((1UL << (SIZEOF_LONG * 8 - 1)) - 1)
397 #define DIGITVAL(code) ((code) - '0')
398 #define ODIGITVAL(code) DIGITVAL(code)
399 #define XDIGITVAL(enc,code) \
400 (ONIGENC_IS_CODE_DIGIT(enc,code) ? DIGITVAL(code) \
401 : (ONIGENC_IS_CODE_UPPER(enc,code) ? (code) - 'A' + 10 : (code) - 'a' + 10))
403 #define IS_SINGLELINE(option) ((option) & ONIG_OPTION_SINGLELINE)
404 #define IS_MULTILINE(option) ((option) & ONIG_OPTION_MULTILINE)
405 #define IS_IGNORECASE(option) ((option) & ONIG_OPTION_IGNORECASE)
406 #define IS_EXTEND(option) ((option) & ONIG_OPTION_EXTEND)
407 #define IS_FIND_LONGEST(option) ((option) & ONIG_OPTION_FIND_LONGEST)
408 #define IS_FIND_NOT_EMPTY(option) ((option) & ONIG_OPTION_FIND_NOT_EMPTY)
409 #define IS_FIND_CONDITION(option) ((option) & \
410 (ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY))
411 #define IS_NOTBOL(option) ((option) & ONIG_OPTION_NOTBOL)
412 #define IS_NOTEOL(option) ((option) & ONIG_OPTION_NOTEOL)
413 #define IS_NOTBOS(option) ((option) & ONIG_OPTION_NOTBOS)
414 #define IS_NOTEOS(option) ((option) & ONIG_OPTION_NOTEOS)
415 #define IS_ASCII_RANGE(option) ((option) & ONIG_OPTION_ASCII_RANGE)
416 #define IS_POSIX_BRACKET_ALL_RANGE(option) ((option) & ONIG_OPTION_POSIX_BRACKET_ALL_RANGE)
417 #define IS_WORD_BOUND_ALL_RANGE(option) ((option) & ONIG_OPTION_WORD_BOUND_ALL_RANGE)
418 #define IS_NEWLINE_CRLF(option) ((option) & ONIG_OPTION_NEWLINE_CRLF)
420 /* OP_SET_OPTION is required for these options.
421 #define IS_DYNAMIC_OPTION(option) \
422 (((option) & (ONIG_OPTION_MULTILINE | ONIG_OPTION_IGNORECASE)) != 0)
424 /* ignore-case and multibyte status are included in compiled code. */
425 #define IS_DYNAMIC_OPTION(option) 0
427 #define DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag) \
428 ((case_fold_flag) & ~INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR)
430 #define REPEAT_INFINITE -1
431 #define IS_REPEAT_INFINITE(n) ((n) == REPEAT_INFINITE)
434 #define BITS_PER_BYTE 8
435 #define SINGLE_BYTE_SIZE (1 << BITS_PER_BYTE)
436 #define BITS_IN_ROOM ((int )sizeof(Bits) * BITS_PER_BYTE)
437 #define BITSET_SIZE (SINGLE_BYTE_SIZE / BITS_IN_ROOM)
439 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
440 typedef unsigned int Bits
;
442 typedef unsigned char Bits
;
444 typedef Bits BitSet
[BITSET_SIZE
];
445 typedef Bits
* BitSetRef
;
447 #define SIZE_BITSET (int )sizeof(BitSet)
449 #define BITSET_CLEAR(bs) do {\
451 for (i = 0; i < BITSET_SIZE; i++) { (bs)[i] = 0; } \
454 #define BS_ROOM(bs,pos) (bs)[(int )(pos) / BITS_IN_ROOM]
455 #define BS_BIT(pos) (1U << ((int )(pos) % BITS_IN_ROOM))
457 #define BITSET_AT(bs, pos) (BS_ROOM(bs,pos) & BS_BIT(pos))
458 #define BITSET_SET_BIT(bs, pos) BS_ROOM(bs,pos) |= BS_BIT(pos)
459 #define BITSET_CLEAR_BIT(bs, pos) BS_ROOM(bs,pos) &= ~(BS_BIT(pos))
460 #define BITSET_INVERT_BIT(bs, pos) BS_ROOM(bs,pos) ^= BS_BIT(pos)
463 typedef struct _BBuf
{
469 #define BBUF_INIT(buf,size) onig_bbuf_init((BBuf* )(buf), (size))
471 #define BBUF_SIZE_INC(buf,inc) do{\
473 (buf)->alloc += (inc);\
474 tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
475 if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
479 #define BBUF_EXPAND(buf,low) do{\
481 do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\
482 tmp = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
483 if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
487 #define BBUF_ENSURE_SIZE(buf,size) do{\
488 unsigned int new_alloc = (buf)->alloc;\
489 while (new_alloc < (unsigned int )(size)) { new_alloc *= 2; }\
490 if ((buf)->alloc != new_alloc) {\
492 tmp = (UChar* )xrealloc((buf)->p, new_alloc);\
493 if (IS_NULL(tmp)) return(ONIGERR_MEMORY);\
495 (buf)->alloc = new_alloc;\
499 #define BBUF_WRITE(buf,pos,bytes,n) do{\
500 int used = (pos) + (int )(n);\
501 if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
502 xmemcpy((buf)->p + (pos), (bytes), (n));\
503 if ((buf)->used < (unsigned int )used) (buf)->used = used;\
506 #define BBUF_WRITE1(buf,pos,byte) do{\
507 int used = (pos) + 1;\
508 if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
509 (buf)->p[(pos)] = (UChar )(byte);\
510 if ((buf)->used < (unsigned int )used) (buf)->used = used;\
513 #define BBUF_ADD(buf,bytes,n) BBUF_WRITE((buf),(buf)->used,(bytes),(n))
514 #define BBUF_ADD1(buf,byte) BBUF_WRITE1((buf),(buf)->used,(byte))
515 #define BBUF_GET_ADD_ADDRESS(buf) ((buf)->p + (buf)->used)
516 #define BBUF_GET_OFFSET_POS(buf) ((buf)->used)
519 #define BBUF_MOVE_RIGHT(buf,from,to,n) do {\
520 if ((unsigned int )((to)+(n)) > (buf)->alloc) BBUF_EXPAND((buf),(to) + (n));\
521 xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
522 if ((unsigned int )((to)+(n)) > (buf)->used) (buf)->used = (to) + (n);\
526 #define BBUF_MOVE_LEFT(buf,from,to,n) do {\
527 xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
531 #define BBUF_MOVE_LEFT_REDUCE(buf,from,to) do {\
532 xmemmove((buf)->p + (to), (buf)->p + (from), (buf)->used - (from));\
533 (buf)->used -= (from - to);\
536 #define BBUF_INSERT(buf,pos,bytes,n) do {\
537 if (pos >= (buf)->used) {\
538 BBUF_WRITE(buf,pos,bytes,n);\
541 BBUF_MOVE_RIGHT((buf),(pos),(pos) + (n),((buf)->used - (pos)));\
542 xmemcpy((buf)->p + (pos), (bytes), (n));\
546 #define BBUF_GET_BYTE(buf, pos) (buf)->p[(pos)]
549 #define ANCHOR_BEGIN_BUF (1<<0)
550 #define ANCHOR_BEGIN_LINE (1<<1)
551 #define ANCHOR_BEGIN_POSITION (1<<2)
552 #define ANCHOR_END_BUF (1<<3)
553 #define ANCHOR_SEMI_END_BUF (1<<4)
554 #define ANCHOR_END_LINE (1<<5)
556 #define ANCHOR_WORD_BOUND (1<<6)
557 #define ANCHOR_NOT_WORD_BOUND (1<<7)
558 #define ANCHOR_WORD_BEGIN (1<<8)
559 #define ANCHOR_WORD_END (1<<9)
560 #define ANCHOR_PREC_READ (1<<10)
561 #define ANCHOR_PREC_READ_NOT (1<<11)
562 #define ANCHOR_LOOK_BEHIND (1<<12)
563 #define ANCHOR_LOOK_BEHIND_NOT (1<<13)
565 #define ANCHOR_ANYCHAR_STAR (1<<14) /* ".*" optimize info */
566 #define ANCHOR_ANYCHAR_STAR_ML (1<<15) /* ".*" optimize info (multi-line) */
568 #define ANCHOR_KEEP (1<<16)
572 OP_FINISH
= 0, /* matching process terminator (no more alternative) */
573 OP_END
= 1, /* pattern code terminator (success end) */
575 OP_EXACT1
= 2, /* single byte, N = 1 */
576 OP_EXACT2
, /* single byte, N = 2 */
577 OP_EXACT3
, /* single byte, N = 3 */
578 OP_EXACT4
, /* single byte, N = 4 */
579 OP_EXACT5
, /* single byte, N = 5 */
580 OP_EXACTN
, /* single byte */
581 OP_EXACTMB2N1
, /* mb-length = 2 N = 1 */
582 OP_EXACTMB2N2
, /* mb-length = 2 N = 2 */
583 OP_EXACTMB2N3
, /* mb-length = 2 N = 3 */
584 OP_EXACTMB2N
, /* mb-length = 2 */
585 OP_EXACTMB3N
, /* mb-length = 3 */
586 OP_EXACTMBN
, /* other length */
588 OP_EXACT1_IC
, /* single byte, N = 1, ignore case */
589 OP_EXACTN_IC
, /* single byte, ignore case */
598 OP_ANYCHAR
, /* "." */
599 OP_ANYCHAR_ML
, /* "." multi-line */
600 OP_ANYCHAR_STAR
, /* ".*" */
601 OP_ANYCHAR_ML_STAR
, /* ".*" multi-line */
602 OP_ANYCHAR_STAR_PEEK_NEXT
,
603 OP_ANYCHAR_ML_STAR_PEEK_NEXT
,
615 OP_NOT_ASCII_WORD_BOUND
,
632 OP_BACKREF_WITH_LEVEL
, /* \k<xxx+n>, \k<xxx-n> */
635 OP_MEMORY_START_PUSH
, /* push back-tracker to stack */
636 OP_MEMORY_END_PUSH
, /* push back-tracker to stack */
637 OP_MEMORY_END_PUSH_REC
, /* push back-tracker to stack */
639 OP_MEMORY_END_REC
, /* push marker to stack */
643 OP_FAIL
, /* pop stack and move */
647 OP_PUSH_OR_JUMP_EXACT1
, /* if match exact then push, else jump. */
648 OP_PUSH_IF_PEEK_NEXT
, /* if match exact then push, else none. */
649 OP_REPEAT
, /* {n,m} */
650 OP_REPEAT_NG
, /* {n,m}? (non greedy) */
652 OP_REPEAT_INC_NG
, /* non greedy */
653 OP_REPEAT_INC_SG
, /* search and get in stack */
654 OP_REPEAT_INC_NG_SG
, /* search and get in stack (non greedy) */
655 OP_NULL_CHECK_START
, /* null loop checker start */
656 OP_NULL_CHECK_END
, /* null loop checker end */
657 OP_NULL_CHECK_END_MEMST
, /* null loop checker end (with capture status) */
658 OP_NULL_CHECK_END_MEMST_PUSH
, /* with capture status and push check-end */
660 OP_PUSH_POS
, /* (?=...) start */
661 OP_POP_POS
, /* (?=...) end */
662 OP_PUSH_POS_NOT
, /* (?!...) start */
663 OP_FAIL_POS
, /* (?!...) end */
664 OP_PUSH_STOP_BT
, /* (?>...) start */
665 OP_POP_STOP_BT
, /* (?>...) end */
666 OP_LOOK_BEHIND
, /* (?<=...) start (no needs end opcode) */
667 OP_PUSH_LOOK_BEHIND_NOT
, /* (?<!...) start */
668 OP_FAIL_LOOK_BEHIND_NOT
, /* (?<!...) end */
669 OP_PUSH_ABSENT_POS
, /* (?~...) start */
670 OP_ABSENT
, /* (?~...) start of inner loop */
671 OP_ABSENT_END
, /* (?~...) end */
673 OP_CALL
, /* \g<name> */
678 OP_STATE_CHECK_PUSH
, /* combination explosion check and push */
679 OP_STATE_CHECK_PUSH_OR_JUMP
, /* check ok -> push, else jump */
680 OP_STATE_CHECK
, /* check only */
681 OP_STATE_CHECK_ANYCHAR_STAR
,
682 OP_STATE_CHECK_ANYCHAR_ML_STAR
,
684 /* no need: IS_DYNAMIC_OPTION() == 0 */
685 OP_SET_OPTION_PUSH
, /* set option and push recover option */
686 OP_SET_OPTION
/* set option */
689 typedef int RelAddrType
;
690 typedef int AbsAddrType
;
691 typedef int LengthType
;
692 typedef int RepeatNumType
;
693 typedef short int MemNumType
;
694 typedef short int StateCheckNumType
;
695 typedef void* PointerType
;
697 #define SIZE_OPCODE 1
698 #define SIZE_RELADDR (int )sizeof(RelAddrType)
699 #define SIZE_ABSADDR (int )sizeof(AbsAddrType)
700 #define SIZE_LENGTH (int )sizeof(LengthType)
701 #define SIZE_MEMNUM (int )sizeof(MemNumType)
702 #define SIZE_STATE_CHECK_NUM (int )sizeof(StateCheckNumType)
703 #define SIZE_REPEATNUM (int )sizeof(RepeatNumType)
704 #define SIZE_OPTION (int )sizeof(OnigOptionType)
705 #define SIZE_CODE_POINT (int )sizeof(OnigCodePoint)
706 #define SIZE_POINTER (int )sizeof(PointerType)
709 #define GET_RELADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, RelAddrType)
710 #define GET_ABSADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, AbsAddrType)
711 #define GET_LENGTH_INC(len,p) PLATFORM_GET_INC(len, p, LengthType)
712 #define GET_MEMNUM_INC(num,p) PLATFORM_GET_INC(num, p, MemNumType)
713 #define GET_REPEATNUM_INC(num,p) PLATFORM_GET_INC(num, p, RepeatNumType)
714 #define GET_OPTION_INC(option,p) PLATFORM_GET_INC(option, p, OnigOptionType)
715 #define GET_POINTER_INC(ptr,p) PLATFORM_GET_INC(ptr, p, PointerType)
716 #define GET_STATE_CHECK_NUM_INC(num,p) PLATFORM_GET_INC(num, p, StateCheckNumType)
718 /* code point's address must be aligned address. */
719 #define GET_CODE_POINT(code,p) code = *((OnigCodePoint* )(p))
720 #define GET_BYTE_INC(byte,p) do{\
726 /* op-code + arg size */
727 #define SIZE_OP_ANYCHAR_STAR SIZE_OPCODE
728 #define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT (SIZE_OPCODE + 1)
729 #define SIZE_OP_JUMP (SIZE_OPCODE + SIZE_RELADDR)
730 #define SIZE_OP_PUSH (SIZE_OPCODE + SIZE_RELADDR)
731 #define SIZE_OP_POP SIZE_OPCODE
732 #define SIZE_OP_PUSH_OR_JUMP_EXACT1 (SIZE_OPCODE + SIZE_RELADDR + 1)
733 #define SIZE_OP_PUSH_IF_PEEK_NEXT (SIZE_OPCODE + SIZE_RELADDR + 1)
734 #define SIZE_OP_REPEAT_INC (SIZE_OPCODE + SIZE_MEMNUM)
735 #define SIZE_OP_REPEAT_INC_NG (SIZE_OPCODE + SIZE_MEMNUM)
736 #define SIZE_OP_PUSH_POS SIZE_OPCODE
737 #define SIZE_OP_PUSH_POS_NOT (SIZE_OPCODE + SIZE_RELADDR)
738 #define SIZE_OP_POP_POS SIZE_OPCODE
739 #define SIZE_OP_FAIL_POS SIZE_OPCODE
740 #define SIZE_OP_SET_OPTION (SIZE_OPCODE + SIZE_OPTION)
741 #define SIZE_OP_SET_OPTION_PUSH (SIZE_OPCODE + SIZE_OPTION)
742 #define SIZE_OP_FAIL SIZE_OPCODE
743 #define SIZE_OP_MEMORY_START (SIZE_OPCODE + SIZE_MEMNUM)
744 #define SIZE_OP_MEMORY_START_PUSH (SIZE_OPCODE + SIZE_MEMNUM)
745 #define SIZE_OP_MEMORY_END_PUSH (SIZE_OPCODE + SIZE_MEMNUM)
746 #define SIZE_OP_MEMORY_END_PUSH_REC (SIZE_OPCODE + SIZE_MEMNUM)
747 #define SIZE_OP_MEMORY_END (SIZE_OPCODE + SIZE_MEMNUM)
748 #define SIZE_OP_MEMORY_END_REC (SIZE_OPCODE + SIZE_MEMNUM)
749 #define SIZE_OP_PUSH_STOP_BT SIZE_OPCODE
750 #define SIZE_OP_POP_STOP_BT SIZE_OPCODE
751 #define SIZE_OP_NULL_CHECK_START (SIZE_OPCODE + SIZE_MEMNUM)
752 #define SIZE_OP_NULL_CHECK_END (SIZE_OPCODE + SIZE_MEMNUM)
753 #define SIZE_OP_LOOK_BEHIND (SIZE_OPCODE + SIZE_LENGTH)
754 #define SIZE_OP_PUSH_LOOK_BEHIND_NOT (SIZE_OPCODE + SIZE_RELADDR + SIZE_LENGTH)
755 #define SIZE_OP_FAIL_LOOK_BEHIND_NOT SIZE_OPCODE
756 #define SIZE_OP_CALL (SIZE_OPCODE + SIZE_ABSADDR)
757 #define SIZE_OP_RETURN SIZE_OPCODE
758 #define SIZE_OP_CONDITION (SIZE_OPCODE + SIZE_MEMNUM + SIZE_RELADDR)
759 #define SIZE_OP_PUSH_ABSENT_POS SIZE_OPCODE
760 #define SIZE_OP_ABSENT (SIZE_OPCODE + SIZE_RELADDR)
761 #define SIZE_OP_ABSENT_END SIZE_OPCODE
763 #ifdef USE_COMBINATION_EXPLOSION_CHECK
764 # define SIZE_OP_STATE_CHECK (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
765 # define SIZE_OP_STATE_CHECK_PUSH (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
766 # define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
767 # define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
770 #define MC_ESC(syn) (syn)->meta_char_table.esc
771 #define MC_ANYCHAR(syn) (syn)->meta_char_table.anychar
772 #define MC_ANYTIME(syn) (syn)->meta_char_table.anytime
773 #define MC_ZERO_OR_ONE_TIME(syn) (syn)->meta_char_table.zero_or_one_time
774 #define MC_ONE_OR_MORE_TIME(syn) (syn)->meta_char_table.one_or_more_time
775 #define MC_ANYCHAR_ANYTIME(syn) (syn)->meta_char_table.anychar_anytime
777 #define IS_MC_ESC_CODE(code, syn) \
778 ((code) == MC_ESC(syn) && \
779 !IS_SYNTAX_OP2((syn), ONIG_SYN_OP2_INEFFECTIVE_ESCAPE))
782 #define SYN_POSIX_COMMON_OP \
783 ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_POSIX_BRACKET | \
784 ONIG_SYN_OP_DECIMAL_BACKREF | \
785 ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_ASTERISK_ZERO_INF | \
786 ONIG_SYN_OP_LINE_ANCHOR | \
787 ONIG_SYN_OP_ESC_CONTROL_CHARS )
789 #define SYN_GNU_REGEX_OP \
790 ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | \
791 ONIG_SYN_OP_POSIX_BRACKET | ONIG_SYN_OP_DECIMAL_BACKREF | \
792 ONIG_SYN_OP_BRACE_INTERVAL | ONIG_SYN_OP_LPAREN_SUBEXP | \
793 ONIG_SYN_OP_VBAR_ALT | \
794 ONIG_SYN_OP_ASTERISK_ZERO_INF | ONIG_SYN_OP_PLUS_ONE_INF | \
795 ONIG_SYN_OP_QMARK_ZERO_ONE | \
796 ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR | ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR | \
797 ONIG_SYN_OP_ESC_W_WORD | \
798 ONIG_SYN_OP_ESC_B_WORD_BOUND | ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END | \
799 ONIG_SYN_OP_ESC_S_WHITE_SPACE | ONIG_SYN_OP_ESC_D_DIGIT | \
800 ONIG_SYN_OP_LINE_ANCHOR )
802 #define SYN_GNU_REGEX_BV \
803 ( ONIG_SYN_CONTEXT_INDEP_ANCHORS | ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | \
804 ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | ONIG_SYN_ALLOW_INVALID_INTERVAL | \
805 ONIG_SYN_BACKSLASH_ESCAPE_IN_CC | ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC )
808 #define NCCLASS_FLAGS(cc) ((cc)->flags)
809 #define NCCLASS_FLAG_SET(cc,flag) (NCCLASS_FLAGS(cc) |= (flag))
810 #define NCCLASS_FLAG_CLEAR(cc,flag) (NCCLASS_FLAGS(cc) &= ~(flag))
811 #define IS_NCCLASS_FLAG_ON(cc,flag) ((NCCLASS_FLAGS(cc) & (flag)) != 0)
814 #define FLAG_NCCLASS_NOT (1<<0)
816 #define NCCLASS_SET_NOT(nd) NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_NOT)
817 #define NCCLASS_CLEAR_NOT(nd) NCCLASS_FLAG_CLEAR(nd, FLAG_NCCLASS_NOT)
818 #define IS_NCCLASS_NOT(nd) IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_NOT)
822 /* struct _Node* next; */
823 /* unsigned int flags; */
830 BBuf
* mbuf
; /* multi-byte info or NULL */
833 typedef intptr_t OnigStackIndex
;
835 typedef struct _OnigStackType
{
837 OnigStackIndex null_check
;
840 UChar
*pcode
; /* byte code position */
841 UChar
*pstr
; /* string position */
842 UChar
*pstr_prev
; /* previous char position of pstr */
843 #ifdef USE_COMBINATION_EXPLOSION_CHECK
844 unsigned int state_check
;
846 UChar
*pkeep
; /* keep pattern position */
849 int count
; /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */
850 UChar
*pcode
; /* byte code position (head of repeated target) */
851 int num
; /* repeat id */
854 OnigStackIndex si
; /* index of stack */
857 int num
; /* memory num */
858 UChar
*pstr
; /* start/end position */
859 /* Following information is set, if this stack type is MEM-START */
860 OnigStackIndex start
; /* prev. info (for backtrack "(...)*" ) */
861 OnigStackIndex end
; /* prev. info (for backtrack "(...)*" ) */
864 int num
; /* null check id */
865 UChar
*pstr
; /* start position */
867 #ifdef USE_SUBEXP_CALL
869 UChar
*ret_addr
; /* byte code position */
870 int num
; /* null check id */
871 UChar
*pstr
; /* string position */
875 UChar
*abs_pstr
; /* absent start position */
876 const UChar
*end_pstr
; /* end position */
878 #ifdef USE_MATCH_CACHE
880 long index
; /* index of the match cache buffer */
881 uint8_t mask
; /* bit-mask for the match cache buffer */
887 #ifdef USE_MATCH_CACHE
891 int outer_repeat_mem
;
892 long num_cache_points_at_outer_repeat
;
893 long num_cache_points_in_outer_repeat
;
894 int lookaround_nesting
;
902 OnigOptionType options
;
904 const UChar
* start
; /* search start position */
905 const UChar
* gpos
; /* global position (for \G: BEGIN_POSITION) */
906 #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
907 OnigPosition best_len
; /* for ONIG_OPTION_FIND_LONGEST */
910 #ifdef USE_COMBINATION_EXPLOSION_CHECK
911 void* state_check_buff
;
912 int state_check_buff_size
;
915 /* rb_hrtime_t from hrtime.h */
916 #ifdef MY_RUBY_BUILD_MAY_TIME_TRAVEL
921 #ifdef USE_MATCH_CACHE
922 int match_cache_status
;
924 long num_cache_opcodes
;
925 OnigCacheOpcode
* cache_opcodes
;
926 long num_cache_points
;
927 uint8_t* match_cache_buf
;
931 #define NUM_CACHE_OPCODES_UNINIT 1
932 #define NUM_CACHE_OPCODES_IMPOSSIBLE -1
934 #define MATCH_CACHE_STATUS_UNINIT 1
935 #define MATCH_CACHE_STATUS_INIT 2
936 #define MATCH_CACHE_STATUS_DISABLED -1
937 #define MATCH_CACHE_STATUS_ENABLED 0
939 #define IS_CODE_SB_WORD(enc,code) \
940 (ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code))
942 typedef struct OnigEndCallListItem
{
943 struct OnigEndCallListItem
* next
;
945 } OnigEndCallListItemType
;
947 extern void onig_add_end_call(void (*func
)(void));
958 extern OnigOpInfoType OnigOpInfo
[];
961 extern void onig_print_compiled_byte_code(FILE* f
, UChar
* bp
, UChar
* bpend
, UChar
** nextp
, OnigEncoding enc
);
963 # ifdef ONIG_DEBUG_STATISTICS
964 extern void onig_statistics_init(void);
965 extern void onig_print_statistics(FILE* f
);
970 #define PRINTF_ARGS(func, fmt, vargs) func
973 extern UChar
* onig_error_code_to_format(OnigPosition code
);
974 PRINTF_ARGS(extern void onig_vsnprintf_with_pattern(UChar buf
[], int bufsize
, OnigEncoding enc
, UChar
* pat
, UChar
* pat_end
, const char *fmt
, va_list args
), 6, 0);
975 PRINTF_ARGS(extern void onig_snprintf_with_pattern(UChar buf
[], int bufsize
, OnigEncoding enc
, UChar
* pat
, UChar
* pat_end
, const char *fmt
, ...), 6, 7);
976 extern int onig_bbuf_init(BBuf
* buf
, OnigDistance size
);
977 extern int onig_compile(regex_t
* reg
, const UChar
* pattern
, const UChar
* pattern_end
, OnigErrorInfo
* einfo
);
979 extern int onig_compile_ruby(regex_t
* reg
, const UChar
* pattern
, const UChar
* pattern_end
, OnigErrorInfo
* einfo
, const char *sourcefile
, int sourceline
);
981 extern void onig_transfer(regex_t
* to
, regex_t
* from
);
982 extern int onig_is_code_in_cc(OnigEncoding enc
, OnigCodePoint code
, CClassNode
* cc
);
983 extern int onig_is_code_in_cc_len(int enclen
, OnigCodePoint code
, CClassNode
* cc
);
986 typedef void hash_table_type
;
988 # include "ruby/st.h"
992 typedef st_data_t hash_data_type
;
994 extern hash_table_type
* onig_st_init_strend_table_with_size(st_index_t size
);
995 extern int onig_st_lookup_strend(hash_table_type
* table
, const UChar
* str_key
, const UChar
* end_key
, hash_data_type
*value
);
996 extern int onig_st_insert_strend(hash_table_type
* table
, const UChar
* str_key
, const UChar
* end_key
, hash_data_type value
);
999 extern size_t onig_memsize(const regex_t
*reg
);
1000 extern size_t onig_region_memsize(const struct re_registers
*regs
);
1001 bool rb_reg_timeout_p(regex_t
*reg
, void *end_time
);
1004 RUBY_SYMBOL_EXPORT_END
1006 #endif /* ONIGMO_REGINT_H */