xref: /PHP-5.6/ext/mbstring/oniguruma/regint.h (revision fe92d64a)
1 #ifndef REGINT_H
2 #define REGINT_H
3 /**********************************************************************
4   regint.h -  Oniguruma (regular expression library)
5 **********************************************************************/
6 /*-
7  * Copyright (c) 2002-2008  K.Kosako  <sndgk393 AT ybb DOT ne DOT jp>
8  * All rights reserved.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 /* for debug */
33 /* #define ONIG_DEBUG_PARSE_TREE */
34 /* #define ONIG_DEBUG_COMPILE */
35 /* #define ONIG_DEBUG_SEARCH */
36 /* #define ONIG_DEBUG_MATCH */
37 /* #define ONIG_DONT_OPTIMIZE */
38 
39 /* for byte-code statistical data. */
40 /* #define ONIG_DEBUG_STATISTICS */
41 
42 #if defined(ONIG_DEBUG_PARSE_TREE) || defined(ONIG_DEBUG_MATCH) || \
43     defined(ONIG_DEBUG_SEARCH) || defined(ONIG_DEBUG_COMPILE) || \
44     defined(ONIG_DEBUG_STATISTICS)
45 #ifndef ONIG_DEBUG
46 #define ONIG_DEBUG
47 #endif
48 #endif
49 
50 #if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \
51     (defined(__ppc__) && defined(__APPLE__)) || \
52     defined(__x86_64) || defined(__x86_64__) || \
53     defined(__mc68020__)
54 #define PLATFORM_UNALIGNED_WORD_ACCESS
55 #endif
56 
57 /* config */
58 /* spec. config */
59 #define USE_NAMED_GROUP
60 #define USE_SUBEXP_CALL
61 #define USE_BACKREF_WITH_LEVEL        /* \k<name+n>, \k<name-n> */
62 #define USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT  /* /(?:()|())*\2/ */
63 #define USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE     /* /\n$/ =~ "\n" */
64 #define USE_WARNING_REDUNDANT_NESTED_REPEAT_OPERATOR
65 /* #define USE_RECOMPILE_API */
66 /* !!! moved to regenc.h. */ /* #define USE_CRNL_AS_LINE_TERMINATOR */
67 
68 /* internal config */
69 #define USE_PARSE_TREE_NODE_RECYCLE
70 #define USE_OP_PUSH_OR_JUMP_EXACT
71 #define USE_QTFR_PEEK_NEXT
72 #define USE_ST_LIBRARY
73 #define USE_SHARED_CCLASS_TABLE
74 
75 #define INIT_MATCH_STACK_SIZE                     160
76 #define DEFAULT_MATCH_STACK_LIMIT_SIZE              0 /* unlimited */
77 
78 #if defined(__GNUC__)
79 #  define ARG_UNUSED  __attribute__ ((unused))
80 #else
81 #  define ARG_UNUSED
82 #endif
83 
84 /* */
85 /* escape other system UChar definition */
86 #include "config.h"
87 #ifdef ONIG_ESCAPE_UCHAR_COLLISION
88 #undef ONIG_ESCAPE_UCHAR_COLLISION
89 #endif
90 
91 #define USE_WORD_BEGIN_END        /* "\<", "\>" */
92 #define USE_CAPTURE_HISTORY
93 #define USE_VARIABLE_META_CHARS
94 #define USE_POSIX_API_REGION_OPTION
95 #define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
96 /* #define USE_COMBINATION_EXPLOSION_CHECK */     /* (X*)* */
97 
98 /* #define USE_MULTI_THREAD_SYSTEM */
99 #define THREAD_SYSTEM_INIT      /* depend on thread system */
100 #define THREAD_SYSTEM_END       /* depend on thread system */
101 #define THREAD_ATOMIC_START     /* depend on thread system */
102 #define THREAD_ATOMIC_END       /* depend on thread system */
103 #define THREAD_PASS             /* depend on thread system */
104 #define xmalloc     malloc
105 #define xrealloc    realloc
106 #define xcalloc     calloc
107 #define xfree       free
108 
109 #define CHECK_INTERRUPT_IN_MATCH_AT
110 
111 #define st_init_table                  onig_st_init_table
112 #define st_init_table_with_size        onig_st_init_table_with_size
113 #define st_init_numtable               onig_st_init_numtable
114 #define st_init_numtable_with_size     onig_st_init_numtable_with_size
115 #define st_init_strtable               onig_st_init_strtable
116 #define st_init_strtable_with_size     onig_st_init_strtable_with_size
117 #define st_delete                      onig_st_delete
118 #define st_delete_safe                 onig_st_delete_safe
119 #define st_insert                      onig_st_insert
120 #define st_lookup                      onig_st_lookup
121 #define st_foreach                     onig_st_foreach
122 #define st_add_direct                  onig_st_add_direct
123 #define st_free_table                  onig_st_free_table
124 #define st_cleanup_safe                onig_st_cleanup_safe
125 #define st_copy                        onig_st_copy
126 #define st_nothing_key_clone           onig_st_nothing_key_clone
127 #define st_nothing_key_free            onig_st_nothing_key_free
128 /* */
129 #define onig_st_is_member              st_is_member
130 
131 #define STATE_CHECK_STRING_THRESHOLD_LEN             7
132 #define STATE_CHECK_BUFF_MAX_SIZE               0x4000
133 
134 #define THREAD_PASS_LIMIT_COUNT     8
135 #define xmemset     memset
136 #define xmemcpy     memcpy
137 #define xmemmove    memmove
138 
139 #if defined(_WIN32) && !defined(__GNUC__)
140 #define xalloca     _alloca
141 #define xvsnprintf  _vsnprintf
142 #else
143 #define xalloca     alloca
144 #define xvsnprintf  vsnprintf
145 #endif
146 
147 
148 #if defined(USE_RECOMPILE_API) && defined(USE_MULTI_THREAD_SYSTEM)
149 #define ONIG_STATE_INC(reg) (reg)->state++
150 #define ONIG_STATE_DEC(reg) (reg)->state--
151 
152 #define ONIG_STATE_INC_THREAD(reg) do {\
153   THREAD_ATOMIC_START;\
154   (reg)->state++;\
155   THREAD_ATOMIC_END;\
156 } while(0)
157 #define ONIG_STATE_DEC_THREAD(reg) do {\
158   THREAD_ATOMIC_START;\
159   (reg)->state--;\
160   THREAD_ATOMIC_END;\
161 } while(0)
162 #else
163 #define ONIG_STATE_INC(reg)         /* Nothing */
164 #define ONIG_STATE_DEC(reg)         /* Nothing */
165 #define ONIG_STATE_INC_THREAD(reg)  /* Nothing */
166 #define ONIG_STATE_DEC_THREAD(reg)  /* Nothing */
167 #endif /* USE_RECOMPILE_API && USE_MULTI_THREAD_SYSTEM */
168 
169 #ifdef HAVE_STDLIB_H
170 #include <stdlib.h>
171 #endif
172 
173 #if defined(HAVE_ALLOCA_H) && !defined(__GNUC__)
174 #include <alloca.h>
175 #endif
176 
177 #ifdef HAVE_STRING_H
178 # include <string.h>
179 #else
180 # include <strings.h>
181 #endif
182 
183 #include <ctype.h>
184 #ifdef HAVE_SYS_TYPES_H
185 #ifndef __BORLANDC__
186 #include <sys/types.h>
187 #endif
188 #endif
189 
190 #ifdef __BORLANDC__
191 #include <malloc.h>
192 #endif
193 
194 #ifdef ONIG_DEBUG
195 # include <stdio.h>
196 #endif
197 
198 #include "regenc.h"
199 
200 #ifdef MIN
201 #undef MIN
202 #endif
203 #ifdef MAX
204 #undef MAX
205 #endif
206 #define MIN(a,b) (((a)>(b))?(b):(a))
207 #define MAX(a,b) (((a)<(b))?(b):(a))
208 
209 #define IS_NULL(p)                    (((void*)(p)) == (void*)0)
210 #define IS_NOT_NULL(p)                (((void*)(p)) != (void*)0)
211 #define CHECK_NULL_RETURN(p)          if (IS_NULL(p)) return NULL
212 #define CHECK_NULL_RETURN_MEMERR(p)   if (IS_NULL(p)) return ONIGERR_MEMORY
213 #define NULL_UCHARP                   ((UChar* )0)
214 
215 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
216 
217 #define PLATFORM_GET_INC(val,p,type) do{\
218   val  = *(type* )p;\
219   (p) += sizeof(type);\
220 } while(0)
221 
222 #else
223 
224 #define PLATFORM_GET_INC(val,p,type) do{\
225   xmemcpy(&val, (p), sizeof(type));\
226   (p) += sizeof(type);\
227 } while(0)
228 
229 /* sizeof(OnigCodePoint) */
230 #define WORD_ALIGNMENT_SIZE     SIZEOF_LONG
231 
232 #define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\
233   (pad_size) = WORD_ALIGNMENT_SIZE \
234                - ((unsigned int )(addr) % WORD_ALIGNMENT_SIZE);\
235   if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\
236 } while (0)
237 
238 #define ALIGNMENT_RIGHT(addr) do {\
239   (addr) += (WORD_ALIGNMENT_SIZE - 1);\
240   (addr) -= ((unsigned int )(addr) % WORD_ALIGNMENT_SIZE);\
241 } while (0)
242 
243 #endif /* PLATFORM_UNALIGNED_WORD_ACCESS */
244 
245 /* stack pop level */
246 #define STACK_POP_LEVEL_FREE        0
247 #define STACK_POP_LEVEL_MEM_START   1
248 #define STACK_POP_LEVEL_ALL         2
249 
250 /* optimize flags */
251 #define ONIG_OPTIMIZE_NONE              0
252 #define ONIG_OPTIMIZE_EXACT             1   /* Slow Search */
253 #define ONIG_OPTIMIZE_EXACT_BM          2   /* Boyer Moore Search */
254 #define ONIG_OPTIMIZE_EXACT_BM_NOT_REV  3   /* BM   (but not simple match) */
255 #define ONIG_OPTIMIZE_EXACT_IC          4   /* Slow Search (ignore case) */
256 #define ONIG_OPTIMIZE_MAP               5   /* char map */
257 
258 /* bit status */
259 typedef unsigned int  BitStatusType;
260 
261 #define BIT_STATUS_BITS_NUM          (sizeof(BitStatusType) * 8)
262 #define BIT_STATUS_CLEAR(stats)      (stats) = 0
263 #define BIT_STATUS_ON_ALL(stats)     (stats) = ~((BitStatusType )0)
264 #define BIT_STATUS_AT(stats,n) \
265   ((n) < (int )BIT_STATUS_BITS_NUM  ?  ((stats) & (1 << n)) : ((stats) & 1))
266 
267 #define BIT_STATUS_ON_AT(stats,n) do {\
268     if ((n) < (int )BIT_STATUS_BITS_NUM)	\
269     (stats) |= (1 << (n));\
270   else\
271     (stats) |= 1;\
272 } while (0)
273 
274 #define BIT_STATUS_ON_AT_SIMPLE(stats,n) do {\
275     if ((n) < (int )BIT_STATUS_BITS_NUM)\
276     (stats) |= (1 << (n));\
277 } while (0)
278 
279 
280 #define INT_MAX_LIMIT           ((1UL << (SIZEOF_INT * 8 - 1)) - 1)
281 
282 #define DIGITVAL(code)    ((code) - '0')
283 #define ODIGITVAL(code)   DIGITVAL(code)
284 #define XDIGITVAL(enc,code) \
285   (ONIGENC_IS_CODE_DIGIT(enc,code) ? DIGITVAL(code) \
286    : (ONIGENC_IS_CODE_UPPER(enc,code) ? (code) - 'A' + 10 : (code) - 'a' + 10))
287 
288 #define IS_SINGLELINE(option)     ((option) & ONIG_OPTION_SINGLELINE)
289 #define IS_MULTILINE(option)      ((option) & ONIG_OPTION_MULTILINE)
290 #define IS_IGNORECASE(option)     ((option) & ONIG_OPTION_IGNORECASE)
291 #define IS_EXTEND(option)         ((option) & ONIG_OPTION_EXTEND)
292 #define IS_FIND_LONGEST(option)   ((option) & ONIG_OPTION_FIND_LONGEST)
293 #define IS_FIND_NOT_EMPTY(option) ((option) & ONIG_OPTION_FIND_NOT_EMPTY)
294 #define IS_FIND_CONDITION(option) ((option) & \
295           (ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY))
296 #define IS_NOTBOL(option)         ((option) & ONIG_OPTION_NOTBOL)
297 #define IS_NOTEOL(option)         ((option) & ONIG_OPTION_NOTEOL)
298 #define IS_POSIX_REGION(option)   ((option) & ONIG_OPTION_POSIX_REGION)
299 
300 /* OP_SET_OPTION is required for these options.
301 #define IS_DYNAMIC_OPTION(option) \
302   (((option) & (ONIG_OPTION_MULTILINE | ONIG_OPTION_IGNORECASE)) != 0)
303 */
304 /* ignore-case and multibyte status are included in compiled code. */
305 #define IS_DYNAMIC_OPTION(option)  0
306 
307 #define DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag) \
308   ((case_fold_flag) & ~INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR)
309 
310 #define REPEAT_INFINITE         -1
311 #define IS_REPEAT_INFINITE(n)   ((n) == REPEAT_INFINITE)
312 
313 /* bitset */
314 #define BITS_PER_BYTE      8
315 #define SINGLE_BYTE_SIZE   (1 << BITS_PER_BYTE)
316 #define BITS_IN_ROOM       (sizeof(Bits) * BITS_PER_BYTE)
317 #define BITSET_SIZE        (SINGLE_BYTE_SIZE / BITS_IN_ROOM)
318 
319 #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
320 typedef unsigned int   Bits;
321 #else
322 typedef unsigned char  Bits;
323 #endif
324 typedef Bits           BitSet[BITSET_SIZE];
325 typedef Bits*          BitSetRef;
326 
327 #define SIZE_BITSET        sizeof(BitSet)
328 
329 #define BITSET_CLEAR(bs) do {\
330   int i;\
331   for (i = 0; i < (int )BITSET_SIZE; i++) { (bs)[i] = 0; }	\
332 } while (0)
333 
334 #define BS_ROOM(bs,pos)            (bs)[pos / BITS_IN_ROOM]
335 #define BS_BIT(pos)                (1 << (pos % BITS_IN_ROOM))
336 
337 #define BITSET_AT(bs, pos)         (BS_ROOM(bs,pos) & BS_BIT(pos))
338 #define BITSET_SET_BIT(bs, pos)     BS_ROOM(bs,pos) |= BS_BIT(pos)
339 #define BITSET_CLEAR_BIT(bs, pos)   BS_ROOM(bs,pos) &= ~(BS_BIT(pos))
340 #define BITSET_INVERT_BIT(bs, pos)  BS_ROOM(bs,pos) ^= BS_BIT(pos)
341 
342 /* bytes buffer */
343 typedef struct _BBuf {
344   UChar* p;
345   unsigned int used;
346   unsigned int alloc;
347 } BBuf;
348 
349 #define BBUF_INIT(buf,size)    onig_bbuf_init((BBuf* )(buf), (size))
350 
351 #define BBUF_SIZE_INC(buf,inc) do{\
352   (buf)->alloc += (inc);\
353   (buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
354   if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\
355 } while (0)
356 
357 #define BBUF_EXPAND(buf,low) do{\
358   do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\
359   (buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
360   if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\
361 } while (0)
362 
363 #define BBUF_ENSURE_SIZE(buf,size) do{\
364   unsigned int new_alloc = (buf)->alloc;\
365   while (new_alloc < (unsigned int )(size)) { new_alloc *= 2; }\
366   if ((buf)->alloc != new_alloc) {\
367     (buf)->p = (UChar* )xrealloc((buf)->p, new_alloc);\
368     if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\
369     (buf)->alloc = new_alloc;\
370   }\
371 } while (0)
372 
373 #define BBUF_WRITE(buf,pos,bytes,n) do{\
374   int used = (pos) + (n);\
375   if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
376   xmemcpy((buf)->p + (pos), (bytes), (n));\
377   if ((buf)->used < (unsigned int )used) (buf)->used = used;\
378 } while (0)
379 
380 #define BBUF_WRITE1(buf,pos,byte) do{\
381   int used = (pos) + 1;\
382   if ((buf)->alloc < (unsigned int )used) BBUF_EXPAND((buf),used);\
383   (buf)->p[(pos)] = (byte);\
384   if ((buf)->used < (unsigned int )used) (buf)->used = used;\
385 } while (0)
386 
387 #define BBUF_ADD(buf,bytes,n)       BBUF_WRITE((buf),(buf)->used,(bytes),(n))
388 #define BBUF_ADD1(buf,byte)         BBUF_WRITE1((buf),(buf)->used,(byte))
389 #define BBUF_GET_ADD_ADDRESS(buf)   ((buf)->p + (buf)->used)
390 #define BBUF_GET_OFFSET_POS(buf)    ((buf)->used)
391 
392 /* from < to */
393 #define BBUF_MOVE_RIGHT(buf,from,to,n) do {\
394   if ((unsigned int )((to)+(n)) > (buf)->alloc) BBUF_EXPAND((buf),(to) + (n));\
395   xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
396   if ((unsigned int )((to)+(n)) > (buf)->used) (buf)->used = (to) + (n);\
397 } while (0)
398 
399 /* from > to */
400 #define BBUF_MOVE_LEFT(buf,from,to,n) do {\
401   xmemmove((buf)->p + (to), (buf)->p + (from), (n));\
402 } while (0)
403 
404 /* from > to */
405 #define BBUF_MOVE_LEFT_REDUCE(buf,from,to) do {\
406   xmemmove((buf)->p + (to), (buf)->p + (from), (buf)->used - (from));\
407   (buf)->used -= (from - to);\
408 } while (0)
409 
410 #define BBUF_INSERT(buf,pos,bytes,n) do {\
411   if (pos >= (buf)->used) {\
412     BBUF_WRITE(buf,pos,bytes,n);\
413   }\
414   else {\
415     BBUF_MOVE_RIGHT((buf),(pos),(pos) + (n),((buf)->used - (pos)));\
416     xmemcpy((buf)->p + (pos), (bytes), (n));\
417   }\
418 } while (0)
419 
420 #define BBUF_GET_BYTE(buf, pos) (buf)->p[(pos)]
421 
422 
423 #define ANCHOR_BEGIN_BUF        (1<<0)
424 #define ANCHOR_BEGIN_LINE       (1<<1)
425 #define ANCHOR_BEGIN_POSITION   (1<<2)
426 #define ANCHOR_END_BUF          (1<<3)
427 #define ANCHOR_SEMI_END_BUF     (1<<4)
428 #define ANCHOR_END_LINE         (1<<5)
429 
430 #define ANCHOR_WORD_BOUND       (1<<6)
431 #define ANCHOR_NOT_WORD_BOUND   (1<<7)
432 #define ANCHOR_WORD_BEGIN       (1<<8)
433 #define ANCHOR_WORD_END         (1<<9)
434 #define ANCHOR_PREC_READ        (1<<10)
435 #define ANCHOR_PREC_READ_NOT    (1<<11)
436 #define ANCHOR_LOOK_BEHIND      (1<<12)
437 #define ANCHOR_LOOK_BEHIND_NOT  (1<<13)
438 
439 #define ANCHOR_ANYCHAR_STAR     (1<<14)   /* ".*" optimize info */
440 #define ANCHOR_ANYCHAR_STAR_ML  (1<<15)   /* ".*" optimize info (multi-line) */
441 
442 /* operation code */
443 enum OpCode {
444   OP_FINISH = 0,        /* matching process terminator (no more alternative) */
445   OP_END    = 1,        /* pattern code terminator (success end) */
446 
447   OP_EXACT1 = 2,        /* single byte, N = 1 */
448   OP_EXACT2,            /* single byte, N = 2 */
449   OP_EXACT3,            /* single byte, N = 3 */
450   OP_EXACT4,            /* single byte, N = 4 */
451   OP_EXACT5,            /* single byte, N = 5 */
452   OP_EXACTN,            /* single byte */
453   OP_EXACTMB2N1,        /* mb-length = 2 N = 1 */
454   OP_EXACTMB2N2,        /* mb-length = 2 N = 2 */
455   OP_EXACTMB2N3,        /* mb-length = 2 N = 3 */
456   OP_EXACTMB2N,         /* mb-length = 2 */
457   OP_EXACTMB3N,         /* mb-length = 3 */
458   OP_EXACTMBN,          /* other length */
459 
460   OP_EXACT1_IC,         /* single byte, N = 1, ignore case */
461   OP_EXACTN_IC,         /* single byte,        ignore case */
462 
463   OP_CCLASS,
464   OP_CCLASS_MB,
465   OP_CCLASS_MIX,
466   OP_CCLASS_NOT,
467   OP_CCLASS_MB_NOT,
468   OP_CCLASS_MIX_NOT,
469   OP_CCLASS_NODE,       /* pointer to CClassNode node */
470 
471   OP_ANYCHAR,                 /* "."  */
472   OP_ANYCHAR_ML,              /* "."  multi-line */
473   OP_ANYCHAR_STAR,            /* ".*" */
474   OP_ANYCHAR_ML_STAR,         /* ".*" multi-line */
475   OP_ANYCHAR_STAR_PEEK_NEXT,
476   OP_ANYCHAR_ML_STAR_PEEK_NEXT,
477 
478   OP_WORD,
479   OP_NOT_WORD,
480   OP_WORD_BOUND,
481   OP_NOT_WORD_BOUND,
482   OP_WORD_BEGIN,
483   OP_WORD_END,
484 
485   OP_BEGIN_BUF,
486   OP_END_BUF,
487   OP_BEGIN_LINE,
488   OP_END_LINE,
489   OP_SEMI_END_BUF,
490   OP_BEGIN_POSITION,
491 
492   OP_BACKREF1,
493   OP_BACKREF2,
494   OP_BACKREFN,
495   OP_BACKREFN_IC,
496   OP_BACKREF_MULTI,
497   OP_BACKREF_MULTI_IC,
498   OP_BACKREF_WITH_LEVEL,    /* \k<xxx+n>, \k<xxx-n> */
499 
500   OP_MEMORY_START,
501   OP_MEMORY_START_PUSH,   /* push back-tracker to stack */
502   OP_MEMORY_END_PUSH,     /* push back-tracker to stack */
503   OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */
504   OP_MEMORY_END,
505   OP_MEMORY_END_REC,      /* push marker to stack */
506 
507   OP_FAIL,               /* pop stack and move */
508   OP_JUMP,
509   OP_PUSH,
510   OP_POP,
511   OP_PUSH_OR_JUMP_EXACT1,  /* if match exact then push, else jump. */
512   OP_PUSH_IF_PEEK_NEXT,    /* if match exact then push, else none. */
513   OP_REPEAT,               /* {n,m} */
514   OP_REPEAT_NG,            /* {n,m}? (non greedy) */
515   OP_REPEAT_INC,
516   OP_REPEAT_INC_NG,        /* non greedy */
517   OP_REPEAT_INC_SG,        /* search and get in stack */
518   OP_REPEAT_INC_NG_SG,     /* search and get in stack (non greedy) */
519   OP_NULL_CHECK_START,     /* null loop checker start */
520   OP_NULL_CHECK_END,       /* null loop checker end   */
521   OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */
522   OP_NULL_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */
523 
524   OP_PUSH_POS,             /* (?=...)  start */
525   OP_POP_POS,              /* (?=...)  end   */
526   OP_PUSH_POS_NOT,         /* (?!...)  start */
527   OP_FAIL_POS,             /* (?!...)  end   */
528   OP_PUSH_STOP_BT,         /* (?>...)  start */
529   OP_POP_STOP_BT,          /* (?>...)  end   */
530   OP_LOOK_BEHIND,          /* (?<=...) start (no needs end opcode) */
531   OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */
532   OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end   */
533 
534   OP_CALL,                 /* \g<name> */
535   OP_RETURN,
536 
537   OP_STATE_CHECK_PUSH,         /* combination explosion check and push */
538   OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump  */
539   OP_STATE_CHECK,              /* check only */
540   OP_STATE_CHECK_ANYCHAR_STAR,
541   OP_STATE_CHECK_ANYCHAR_ML_STAR,
542 
543   /* no need: IS_DYNAMIC_OPTION() == 0 */
544   OP_SET_OPTION_PUSH,    /* set option and push recover option */
545   OP_SET_OPTION          /* set option */
546 };
547 
548 typedef int RelAddrType;
549 typedef int AbsAddrType;
550 typedef int LengthType;
551 typedef int RepeatNumType;
552 typedef short int MemNumType;
553 typedef short int StateCheckNumType;
554 typedef void* PointerType;
555 
556 #define SIZE_OPCODE           1
557 #define SIZE_RELADDR          sizeof(RelAddrType)
558 #define SIZE_ABSADDR          sizeof(AbsAddrType)
559 #define SIZE_LENGTH           sizeof(LengthType)
560 #define SIZE_MEMNUM           sizeof(MemNumType)
561 #define SIZE_STATE_CHECK_NUM  sizeof(StateCheckNumType)
562 #define SIZE_REPEATNUM        sizeof(RepeatNumType)
563 #define SIZE_OPTION           sizeof(OnigOptionType)
564 #define SIZE_CODE_POINT       sizeof(OnigCodePoint)
565 #define SIZE_POINTER          sizeof(PointerType)
566 
567 
568 #define GET_RELADDR_INC(addr,p)    PLATFORM_GET_INC(addr,   p, RelAddrType)
569 #define GET_ABSADDR_INC(addr,p)    PLATFORM_GET_INC(addr,   p, AbsAddrType)
570 #define GET_LENGTH_INC(len,p)      PLATFORM_GET_INC(len,    p, LengthType)
571 #define GET_MEMNUM_INC(num,p)      PLATFORM_GET_INC(num,    p, MemNumType)
572 #define GET_REPEATNUM_INC(num,p)   PLATFORM_GET_INC(num,    p, RepeatNumType)
573 #define GET_OPTION_INC(option,p)   PLATFORM_GET_INC(option, p, OnigOptionType)
574 #define GET_POINTER_INC(ptr,p)     PLATFORM_GET_INC(ptr,    p, PointerType)
575 #define GET_STATE_CHECK_NUM_INC(num,p)  PLATFORM_GET_INC(num, p, StateCheckNumType)
576 
577 /* code point's address must be aligned address. */
578 #define GET_CODE_POINT(code,p)   code = *((OnigCodePoint* )(p))
579 #define GET_BYTE_INC(byte,p) do{\
580   byte = *(p);\
581   (p)++;\
582 } while(0)
583 
584 
585 /* op-code + arg size */
586 #define SIZE_OP_ANYCHAR_STAR            SIZE_OPCODE
587 #define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT (SIZE_OPCODE + 1)
588 #define SIZE_OP_JUMP                   (SIZE_OPCODE + SIZE_RELADDR)
589 #define SIZE_OP_PUSH                   (SIZE_OPCODE + SIZE_RELADDR)
590 #define SIZE_OP_POP                     SIZE_OPCODE
591 #define SIZE_OP_PUSH_OR_JUMP_EXACT1    (SIZE_OPCODE + SIZE_RELADDR + 1)
592 #define SIZE_OP_PUSH_IF_PEEK_NEXT      (SIZE_OPCODE + SIZE_RELADDR + 1)
593 #define SIZE_OP_REPEAT_INC             (SIZE_OPCODE + SIZE_MEMNUM)
594 #define SIZE_OP_REPEAT_INC_NG          (SIZE_OPCODE + SIZE_MEMNUM)
595 #define SIZE_OP_PUSH_POS                SIZE_OPCODE
596 #define SIZE_OP_PUSH_POS_NOT           (SIZE_OPCODE + SIZE_RELADDR)
597 #define SIZE_OP_POP_POS                 SIZE_OPCODE
598 #define SIZE_OP_FAIL_POS                SIZE_OPCODE
599 #define SIZE_OP_SET_OPTION             (SIZE_OPCODE + SIZE_OPTION)
600 #define SIZE_OP_SET_OPTION_PUSH        (SIZE_OPCODE + SIZE_OPTION)
601 #define SIZE_OP_FAIL                    SIZE_OPCODE
602 #define SIZE_OP_MEMORY_START           (SIZE_OPCODE + SIZE_MEMNUM)
603 #define SIZE_OP_MEMORY_START_PUSH      (SIZE_OPCODE + SIZE_MEMNUM)
604 #define SIZE_OP_MEMORY_END_PUSH        (SIZE_OPCODE + SIZE_MEMNUM)
605 #define SIZE_OP_MEMORY_END_PUSH_REC    (SIZE_OPCODE + SIZE_MEMNUM)
606 #define SIZE_OP_MEMORY_END             (SIZE_OPCODE + SIZE_MEMNUM)
607 #define SIZE_OP_MEMORY_END_REC         (SIZE_OPCODE + SIZE_MEMNUM)
608 #define SIZE_OP_PUSH_STOP_BT            SIZE_OPCODE
609 #define SIZE_OP_POP_STOP_BT             SIZE_OPCODE
610 #define SIZE_OP_NULL_CHECK_START       (SIZE_OPCODE + SIZE_MEMNUM)
611 #define SIZE_OP_NULL_CHECK_END         (SIZE_OPCODE + SIZE_MEMNUM)
612 #define SIZE_OP_LOOK_BEHIND            (SIZE_OPCODE + SIZE_LENGTH)
613 #define SIZE_OP_PUSH_LOOK_BEHIND_NOT   (SIZE_OPCODE + SIZE_RELADDR + SIZE_LENGTH)
614 #define SIZE_OP_FAIL_LOOK_BEHIND_NOT    SIZE_OPCODE
615 #define SIZE_OP_CALL                   (SIZE_OPCODE + SIZE_ABSADDR)
616 #define SIZE_OP_RETURN                  SIZE_OPCODE
617 
618 #ifdef USE_COMBINATION_EXPLOSION_CHECK
619 #define SIZE_OP_STATE_CHECK            (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
620 #define SIZE_OP_STATE_CHECK_PUSH       (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
621 #define SIZE_OP_STATE_CHECK_PUSH_OR_JUMP (SIZE_OPCODE + SIZE_STATE_CHECK_NUM + SIZE_RELADDR)
622 #define SIZE_OP_STATE_CHECK_ANYCHAR_STAR (SIZE_OPCODE + SIZE_STATE_CHECK_NUM)
623 #endif
624 
625 #define MC_ESC(syn)               (syn)->meta_char_table.esc
626 #define MC_ANYCHAR(syn)           (syn)->meta_char_table.anychar
627 #define MC_ANYTIME(syn)           (syn)->meta_char_table.anytime
628 #define MC_ZERO_OR_ONE_TIME(syn)  (syn)->meta_char_table.zero_or_one_time
629 #define MC_ONE_OR_MORE_TIME(syn)  (syn)->meta_char_table.one_or_more_time
630 #define MC_ANYCHAR_ANYTIME(syn)   (syn)->meta_char_table.anychar_anytime
631 
632 #define IS_MC_ESC_CODE(code, syn) \
633   ((code) == MC_ESC(syn) && \
634    !IS_SYNTAX_OP2((syn), ONIG_SYN_OP2_INEFFECTIVE_ESCAPE))
635 
636 
637 #define SYN_POSIX_COMMON_OP \
638  ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_POSIX_BRACKET | \
639    ONIG_SYN_OP_DECIMAL_BACKREF | \
640    ONIG_SYN_OP_BRACKET_CC | ONIG_SYN_OP_ASTERISK_ZERO_INF | \
641    ONIG_SYN_OP_LINE_ANCHOR | \
642    ONIG_SYN_OP_ESC_CONTROL_CHARS )
643 
644 #define SYN_GNU_REGEX_OP \
645   ( ONIG_SYN_OP_DOT_ANYCHAR | ONIG_SYN_OP_BRACKET_CC | \
646     ONIG_SYN_OP_POSIX_BRACKET | ONIG_SYN_OP_DECIMAL_BACKREF | \
647     ONIG_SYN_OP_BRACE_INTERVAL | ONIG_SYN_OP_LPAREN_SUBEXP | \
648     ONIG_SYN_OP_VBAR_ALT | \
649     ONIG_SYN_OP_ASTERISK_ZERO_INF | ONIG_SYN_OP_PLUS_ONE_INF | \
650     ONIG_SYN_OP_QMARK_ZERO_ONE | \
651     ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR | ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR | \
652     ONIG_SYN_OP_ESC_W_WORD | \
653     ONIG_SYN_OP_ESC_B_WORD_BOUND | ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END | \
654     ONIG_SYN_OP_ESC_S_WHITE_SPACE | ONIG_SYN_OP_ESC_D_DIGIT | \
655     ONIG_SYN_OP_LINE_ANCHOR )
656 
657 #define SYN_GNU_REGEX_BV \
658   ( ONIG_SYN_CONTEXT_INDEP_ANCHORS | ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS | \
659     ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS | ONIG_SYN_ALLOW_INVALID_INTERVAL | \
660     ONIG_SYN_BACKSLASH_ESCAPE_IN_CC | ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC )
661 
662 
663 #define NCCLASS_FLAGS(cc)           ((cc)->flags)
664 #define NCCLASS_FLAG_SET(cc,flag)    (NCCLASS_FLAGS(cc) |= (flag))
665 #define NCCLASS_FLAG_CLEAR(cc,flag)  (NCCLASS_FLAGS(cc) &= ~(flag))
666 #define IS_NCCLASS_FLAG_ON(cc,flag) ((NCCLASS_FLAGS(cc) & (flag)) != 0)
667 
668 /* cclass node */
669 #define FLAG_NCCLASS_NOT           (1<<0)
670 #define FLAG_NCCLASS_SHARE         (1<<1)
671 
672 #define NCCLASS_SET_NOT(nd)     NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_NOT)
673 #define NCCLASS_SET_SHARE(nd)   NCCLASS_FLAG_SET(nd, FLAG_NCCLASS_SHARE)
674 #define NCCLASS_CLEAR_NOT(nd)   NCCLASS_FLAG_CLEAR(nd, FLAG_NCCLASS_NOT)
675 #define IS_NCCLASS_NOT(nd)      IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_NOT)
676 #define IS_NCCLASS_SHARE(nd)    IS_NCCLASS_FLAG_ON(nd, FLAG_NCCLASS_SHARE)
677 
678 typedef struct {
679   int type;
680   /* struct _Node* next; */
681   /* unsigned int flags; */
682 } NodeBase;
683 
684 typedef struct {
685   NodeBase base;
686   unsigned int flags;
687   BitSet bs;
688   BBuf*  mbuf;   /* multi-byte info or NULL */
689 } CClassNode;
690 
691 typedef long OnigStackIndex;
692 
693 typedef struct _OnigStackType {
694   unsigned int type;
695   union {
696     struct {
697       UChar *pcode;      /* byte code position */
698       UChar *pstr;       /* string position */
699       UChar *pstr_prev;  /* previous char position of pstr */
700 #ifdef USE_COMBINATION_EXPLOSION_CHECK
701       unsigned int state_check;
702 #endif
703     } state;
704     struct {
705       int   count;       /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */
706       UChar *pcode;      /* byte code position (head of repeated target) */
707       int   num;         /* repeat id */
708     } repeat;
709     struct {
710       OnigStackIndex si;     /* index of stack */
711     } repeat_inc;
712     struct {
713       int num;           /* memory num */
714       UChar *pstr;       /* start/end position */
715       /* Following information is setted, if this stack type is MEM-START */
716       OnigStackIndex start;  /* prev. info (for backtrack  "(...)*" ) */
717       OnigStackIndex end;    /* prev. info (for backtrack  "(...)*" ) */
718     } mem;
719     struct {
720       int num;           /* null check id */
721       UChar *pstr;       /* start position */
722     } null_check;
723 #ifdef USE_SUBEXP_CALL
724     struct {
725       UChar *ret_addr;   /* byte code position */
726       int    num;        /* null check id */
727       UChar *pstr;       /* string position */
728     } call_frame;
729 #endif
730   } u;
731 } OnigStackType;
732 
733 typedef struct {
734   void* stack_p;
735   int   stack_n;
736   OnigOptionType options;
737   OnigRegion*    region;
738   const UChar* start;   /* search start position (for \G: BEGIN_POSITION) */
739 #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
740   int    best_len;      /* for ONIG_OPTION_FIND_LONGEST */
741   UChar* best_s;
742 #endif
743 #ifdef USE_COMBINATION_EXPLOSION_CHECK
744   void* state_check_buff;
745   int   state_check_buff_size;
746 #endif
747 } OnigMatchArg;
748 
749 
750 #define IS_CODE_SB_WORD(enc,code) \
751   (ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code))
752 
753 #ifdef ONIG_DEBUG
754 
755 typedef struct {
756   short int opcode;
757   char*     name;
758   short int arg_type;
759 } OnigOpInfoType;
760 
761 extern OnigOpInfoType OnigOpInfo[];
762 
763 extern void onig_print_compiled_byte_code P_((FILE* f, UChar* bp, UChar** nextp, OnigEncoding enc));
764 
765 #ifdef ONIG_DEBUG_STATISTICS
766 extern void onig_statistics_init P_((void));
767 extern void onig_print_statistics P_((FILE* f));
768 #endif
769 #endif
770 
771 extern UChar* onig_error_code_to_format P_((int code));
772 extern void  onig_snprintf_with_pattern PV_((UChar buf[], int bufsize, OnigEncoding enc, UChar* pat, UChar* pat_end, const UChar *fmt, ...));
773 extern int  onig_bbuf_init P_((BBuf* buf, int size));
774 extern int  onig_compile P_((regex_t* reg, const UChar* pattern, const UChar* pattern_end, OnigErrorInfo* einfo));
775 extern void onig_chain_reduce P_((regex_t* reg));
776 extern void onig_chain_link_add P_((regex_t* to, regex_t* add));
777 extern void onig_transfer P_((regex_t* to, regex_t* from));
778 extern int  onig_is_code_in_cc P_((OnigEncoding enc, OnigCodePoint code, CClassNode* cc));
779 extern int  onig_is_code_in_cc_len P_((int enclen, OnigCodePoint code, CClassNode* cc));
780 
781 /* strend hash */
782 typedef void hash_table_type;
783 typedef unsigned long hash_data_type;
784 
785 extern hash_table_type* onig_st_init_strend_table_with_size P_((int size));
786 extern int onig_st_lookup_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type *value));
787 extern int onig_st_insert_strend P_((hash_table_type* table, const UChar* str_key, const UChar* end_key, hash_data_type value));
788 
789 /* encoding property management */
790 #define PROPERTY_LIST_ADD_PROP(Name, CR) \
791   r = onigenc_property_list_add_property((UChar* )Name, CR,\
792 	      &PropertyNameTable, &PropertyList, &PropertyListNum,\
793 	      &PropertyListSize);\
794   if (r != 0) goto end
795 
796 #define PROPERTY_LIST_INIT_CHECK \
797   if (PropertyInited == 0) {\
798     int r = onigenc_property_list_init(init_property_list);\
799     if (r != 0) return r;\
800   }
801 
802 extern int onigenc_property_list_add_property P_((UChar* name, const OnigCodePoint* prop, hash_table_type **table, const OnigCodePoint*** plist, int *pnum, int *psize));
803 
804 typedef int (*ONIGENC_INIT_PROPERTY_LIST_FUNC_TYPE)(void);
805 
806 extern int onigenc_property_list_init P_((ONIGENC_INIT_PROPERTY_LIST_FUNC_TYPE));
807 
808 #endif /* REGINT_H */
809