xref: /PHP-7.3/ext/mbstring/oniguruma/src/oniguruma.h (revision 1979c5d1)
1 #ifndef ONIGURUMA_H
2 #define ONIGURUMA_H
3 /**********************************************************************
4   oniguruma.h - Oniguruma (regular expression library)
5 **********************************************************************/
6 /*-
7  * Copyright (c) 2002-2019  K.Kosako
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 #ifdef __cplusplus
33 extern "C" {
34 #endif
35 
36 #define ONIGURUMA
37 #define ONIGURUMA_VERSION_MAJOR   6
38 #define ONIGURUMA_VERSION_MINOR   9
39 #define ONIGURUMA_VERSION_TEENY   4
40 
41 #define ONIGURUMA_VERSION_INT     60904
42 
43 #ifndef P_
44 #if defined(__STDC__) || defined(_WIN32)
45 # define P_(args) args
46 #else
47 # define P_(args) ()
48 #endif
49 #endif
50 
51 #ifndef PV_
52 # define PV_(args) args
53 #endif
54 
55 #ifndef ONIG_STATIC
56 #ifndef ONIG_EXTERN
57 #if defined(_WIN32) && !defined(__GNUC__)
58 #if defined(ONIGURUMA_EXPORT)
59 #define ONIG_EXTERN   extern __declspec(dllexport)
60 #else
61 #define ONIG_EXTERN   extern __declspec(dllimport)
62 #endif
63 #endif
64 #endif
65 
66 #ifndef ONIG_EXTERN
67 #define ONIG_EXTERN   extern
68 #endif
69 #else
70 #define ONIG_EXTERN   extern
71 #endif
72 
73 /* PART: character encoding */
74 
75 #ifndef ONIG_ESCAPE_UCHAR_COLLISION
76 #define UChar OnigUChar
77 #endif
78 
79 typedef unsigned int   OnigCodePoint;
80 typedef unsigned char  OnigUChar;
81 typedef unsigned int   OnigCtype;
82 typedef unsigned int   OnigLen;
83 
84 #define ONIG_INFINITE_DISTANCE  ~((OnigLen )0)
85 
86 typedef unsigned int OnigCaseFoldType; /* case fold flag */
87 
88 ONIG_EXTERN OnigCaseFoldType OnigDefaultCaseFoldFlag;
89 
90 /* #define ONIGENC_CASE_FOLD_HIRAGANA_KATAKANA  (1<<1) */
91 /* #define ONIGENC_CASE_FOLD_KATAKANA_WIDTH     (1<<2) */
92 #define ONIGENC_CASE_FOLD_TURKISH_AZERI         (1<<20)
93 #define INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR   (1<<30)
94 
95 #define ONIGENC_CASE_FOLD_MIN      INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR
96 #define ONIGENC_CASE_FOLD_DEFAULT  OnigDefaultCaseFoldFlag
97 
98 
99 #define ONIGENC_MAX_COMP_CASE_FOLD_CODE_LEN       3
100 #define ONIGENC_GET_CASE_FOLD_CODES_MAX_NUM      13
101 /* 13 => Unicode:0x1ffc */
102 
103 /* code range */
104 #define ONIGENC_CODE_RANGE_NUM(range)     ((int )range[0])
105 #define ONIGENC_CODE_RANGE_FROM(range,i)  range[((i)*2) + 1]
106 #define ONIGENC_CODE_RANGE_TO(range,i)    range[((i)*2) + 2]
107 
108 typedef struct {
109   int byte_len;  /* argument(original) character(s) byte length */
110   int code_len;  /* number of code */
111   OnigCodePoint code[ONIGENC_MAX_COMP_CASE_FOLD_CODE_LEN];
112 } OnigCaseFoldCodeItem;
113 
114 typedef struct {
115   OnigCodePoint esc;
116   OnigCodePoint anychar;
117   OnigCodePoint anytime;
118   OnigCodePoint zero_or_one_time;
119   OnigCodePoint one_or_more_time;
120   OnigCodePoint anychar_anytime;
121 } OnigMetaCharTableType;
122 
123 typedef int (*OnigApplyAllCaseFoldFunc)(OnigCodePoint from, OnigCodePoint* to, int to_len, void* arg);
124 
125 typedef struct OnigEncodingTypeST {
126   int    (*mbc_enc_len)(const OnigUChar* p);
127   const char*   name;
128   int           max_enc_len;
129   int           min_enc_len;
130   int    (*is_mbc_newline)(const OnigUChar* p, const OnigUChar* end);
131   OnigCodePoint (*mbc_to_code)(const OnigUChar* p, const OnigUChar* end);
132   int    (*code_to_mbclen)(OnigCodePoint code);
133   int    (*code_to_mbc)(OnigCodePoint code, OnigUChar *buf);
134   int    (*mbc_case_fold)(OnigCaseFoldType flag, const OnigUChar** pp, const OnigUChar* end, OnigUChar* to);
135   int    (*apply_all_case_fold)(OnigCaseFoldType flag, OnigApplyAllCaseFoldFunc f, void* arg);
136   int    (*get_case_fold_codes_by_str)(OnigCaseFoldType flag, const OnigUChar* p, const OnigUChar* end, OnigCaseFoldCodeItem acs[]);
137   int    (*property_name_to_ctype)(struct OnigEncodingTypeST* enc, OnigUChar* p, OnigUChar* end);
138   int    (*is_code_ctype)(OnigCodePoint code, OnigCtype ctype);
139   int    (*get_ctype_code_range)(OnigCtype ctype, OnigCodePoint* sb_out, const OnigCodePoint* ranges[]);
140   OnigUChar* (*left_adjust_char_head)(const OnigUChar* start, const OnigUChar* p);
141   int    (*is_allowed_reverse_match)(const OnigUChar* p, const OnigUChar* end);
142   int    (*init)(void);
143   int    (*is_initialized)(void);
144   int    (*is_valid_mbc_string)(const OnigUChar* s, const OnigUChar* end);
145   unsigned int flag;
146   OnigCodePoint sb_range;
147   int index;
148 } OnigEncodingType;
149 
150 typedef OnigEncodingType* OnigEncoding;
151 
152 ONIG_EXTERN OnigEncodingType OnigEncodingASCII;
153 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_1;
154 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_2;
155 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_3;
156 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_4;
157 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_5;
158 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_6;
159 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_7;
160 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_8;
161 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_9;
162 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_10;
163 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_11;
164 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_13;
165 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_14;
166 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_15;
167 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_16;
168 ONIG_EXTERN OnigEncodingType OnigEncodingUTF8;
169 ONIG_EXTERN OnigEncodingType OnigEncodingUTF16_BE;
170 ONIG_EXTERN OnigEncodingType OnigEncodingUTF16_LE;
171 ONIG_EXTERN OnigEncodingType OnigEncodingUTF32_BE;
172 ONIG_EXTERN OnigEncodingType OnigEncodingUTF32_LE;
173 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_JP;
174 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_TW;
175 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_KR;
176 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_CN;
177 ONIG_EXTERN OnigEncodingType OnigEncodingSJIS;
178 ONIG_EXTERN OnigEncodingType OnigEncodingKOI8;
179 ONIG_EXTERN OnigEncodingType OnigEncodingKOI8_R;
180 ONIG_EXTERN OnigEncodingType OnigEncodingCP1251;
181 ONIG_EXTERN OnigEncodingType OnigEncodingBIG5;
182 ONIG_EXTERN OnigEncodingType OnigEncodingGB18030;
183 
184 #define ONIG_ENCODING_ASCII        (&OnigEncodingASCII)
185 #define ONIG_ENCODING_ISO_8859_1   (&OnigEncodingISO_8859_1)
186 #define ONIG_ENCODING_ISO_8859_2   (&OnigEncodingISO_8859_2)
187 #define ONIG_ENCODING_ISO_8859_3   (&OnigEncodingISO_8859_3)
188 #define ONIG_ENCODING_ISO_8859_4   (&OnigEncodingISO_8859_4)
189 #define ONIG_ENCODING_ISO_8859_5   (&OnigEncodingISO_8859_5)
190 #define ONIG_ENCODING_ISO_8859_6   (&OnigEncodingISO_8859_6)
191 #define ONIG_ENCODING_ISO_8859_7   (&OnigEncodingISO_8859_7)
192 #define ONIG_ENCODING_ISO_8859_8   (&OnigEncodingISO_8859_8)
193 #define ONIG_ENCODING_ISO_8859_9   (&OnigEncodingISO_8859_9)
194 #define ONIG_ENCODING_ISO_8859_10  (&OnigEncodingISO_8859_10)
195 #define ONIG_ENCODING_ISO_8859_11  (&OnigEncodingISO_8859_11)
196 #define ONIG_ENCODING_ISO_8859_13  (&OnigEncodingISO_8859_13)
197 #define ONIG_ENCODING_ISO_8859_14  (&OnigEncodingISO_8859_14)
198 #define ONIG_ENCODING_ISO_8859_15  (&OnigEncodingISO_8859_15)
199 #define ONIG_ENCODING_ISO_8859_16  (&OnigEncodingISO_8859_16)
200 #define ONIG_ENCODING_UTF8         (&OnigEncodingUTF8)
201 #define ONIG_ENCODING_UTF16_BE     (&OnigEncodingUTF16_BE)
202 #define ONIG_ENCODING_UTF16_LE     (&OnigEncodingUTF16_LE)
203 #define ONIG_ENCODING_UTF32_BE     (&OnigEncodingUTF32_BE)
204 #define ONIG_ENCODING_UTF32_LE     (&OnigEncodingUTF32_LE)
205 #define ONIG_ENCODING_EUC_JP       (&OnigEncodingEUC_JP)
206 #define ONIG_ENCODING_EUC_TW       (&OnigEncodingEUC_TW)
207 #define ONIG_ENCODING_EUC_KR       (&OnigEncodingEUC_KR)
208 #define ONIG_ENCODING_EUC_CN       (&OnigEncodingEUC_CN)
209 #define ONIG_ENCODING_SJIS         (&OnigEncodingSJIS)
210 #define ONIG_ENCODING_KOI8         (&OnigEncodingKOI8)
211 #define ONIG_ENCODING_KOI8_R       (&OnigEncodingKOI8_R)
212 #define ONIG_ENCODING_CP1251       (&OnigEncodingCP1251)
213 #define ONIG_ENCODING_BIG5         (&OnigEncodingBIG5)
214 #define ONIG_ENCODING_GB18030      (&OnigEncodingGB18030)
215 
216 #define ONIG_ENCODING_UNDEF    ((OnigEncoding )0)
217 
218 
219 /* work size */
220 #define ONIGENC_CODE_TO_MBC_MAXLEN       7
221 #define ONIGENC_MBC_CASE_FOLD_MAXLEN    18
222 /* 18: 6(max-byte) * 3(case-fold chars) */
223 
224 /* character types */
225 typedef enum {
226   ONIGENC_CTYPE_NEWLINE = 0,
227   ONIGENC_CTYPE_ALPHA   = 1,
228   ONIGENC_CTYPE_BLANK   = 2,
229   ONIGENC_CTYPE_CNTRL   = 3,
230   ONIGENC_CTYPE_DIGIT   = 4,
231   ONIGENC_CTYPE_GRAPH   = 5,
232   ONIGENC_CTYPE_LOWER   = 6,
233   ONIGENC_CTYPE_PRINT   = 7,
234   ONIGENC_CTYPE_PUNCT   = 8,
235   ONIGENC_CTYPE_SPACE   = 9,
236   ONIGENC_CTYPE_UPPER   = 10,
237   ONIGENC_CTYPE_XDIGIT  = 11,
238   ONIGENC_CTYPE_WORD    = 12,
239   ONIGENC_CTYPE_ALNUM   = 13,  /* alpha || digit */
240   ONIGENC_CTYPE_ASCII   = 14
241 } OnigEncCtype;
242 
243 #define ONIGENC_MAX_STD_CTYPE  ONIGENC_CTYPE_ASCII
244 
245 
246 #define onig_enc_len(enc,p,end)        ONIGENC_MBC_ENC_LEN(enc,p)
247 
248 #define ONIGENC_IS_UNDEF(enc)          ((enc) == ONIG_ENCODING_UNDEF)
249 #define ONIGENC_IS_SINGLEBYTE(enc)     (ONIGENC_MBC_MAXLEN(enc) == 1)
250 #define ONIGENC_IS_MBC_HEAD(enc,p)     (ONIGENC_MBC_ENC_LEN(enc,p) != 1)
251 #define ONIGENC_IS_MBC_ASCII(p)           (*(p)   < 128)
252 #define ONIGENC_IS_CODE_ASCII(code)       ((code) < 128)
253 #define ONIGENC_IS_MBC_WORD(enc,s,end) \
254    ONIGENC_IS_CODE_WORD(enc,ONIGENC_MBC_TO_CODE(enc,s,end))
255 #define ONIGENC_IS_MBC_WORD_ASCII(enc,s,end) onigenc_is_mbc_word_ascii(enc,s,end)
256 
257 #define ONIGENC_NAME(enc)                      ((enc)->name)
258 
259 #define ONIGENC_MBC_CASE_FOLD(enc,flag,pp,end,buf) \
260   (enc)->mbc_case_fold(flag,(const OnigUChar** )pp,end,buf)
261 #define ONIGENC_IS_ALLOWED_REVERSE_MATCH(enc,s,end) \
262         (enc)->is_allowed_reverse_match(s,end)
263 #define ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc,start,s) \
264         (enc)->left_adjust_char_head(start, s)
265 #define ONIGENC_IS_VALID_MBC_STRING(enc,s,end) \
266         (enc)->is_valid_mbc_string(s,end)
267 #define ONIGENC_APPLY_ALL_CASE_FOLD(enc,case_fold_flag,f,arg) \
268         (enc)->apply_all_case_fold(case_fold_flag,f,arg)
269 #define ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc,case_fold_flag,p,end,acs) \
270        (enc)->get_case_fold_codes_by_str(case_fold_flag,p,end,acs)
271 #define ONIGENC_STEP_BACK(enc,start,s,n) \
272         onigenc_step_back((enc),(start),(s),(n))
273 
274 #define ONIGENC_MBC_ENC_LEN(enc,p)             (enc)->mbc_enc_len(p)
275 #define ONIGENC_MBC_MAXLEN(enc)               ((enc)->max_enc_len)
276 #define ONIGENC_MBC_MAXLEN_DIST(enc)           ONIGENC_MBC_MAXLEN(enc)
277 #define ONIGENC_MBC_MINLEN(enc)               ((enc)->min_enc_len)
278 #define ONIGENC_IS_MBC_NEWLINE(enc,p,end)      (enc)->is_mbc_newline((p),(end))
279 #define ONIGENC_MBC_TO_CODE(enc,p,end)         (enc)->mbc_to_code((p),(end))
280 #define ONIGENC_CODE_TO_MBCLEN(enc,code)       (enc)->code_to_mbclen(code)
281 #define ONIGENC_CODE_TO_MBC(enc,code,buf)      (enc)->code_to_mbc(code,buf)
282 #define ONIGENC_PROPERTY_NAME_TO_CTYPE(enc,p,end) \
283   (enc)->property_name_to_ctype(enc,p,end)
284 
285 #define ONIGENC_IS_CODE_CTYPE(enc,code,ctype)  (enc)->is_code_ctype(code,ctype)
286 
287 #define ONIGENC_IS_CODE_NEWLINE(enc,code) \
288         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_NEWLINE)
289 #define ONIGENC_IS_CODE_GRAPH(enc,code) \
290         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_GRAPH)
291 #define ONIGENC_IS_CODE_PRINT(enc,code) \
292         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PRINT)
293 #define ONIGENC_IS_CODE_ALNUM(enc,code) \
294         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALNUM)
295 #define ONIGENC_IS_CODE_ALPHA(enc,code) \
296         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALPHA)
297 #define ONIGENC_IS_CODE_LOWER(enc,code) \
298         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_LOWER)
299 #define ONIGENC_IS_CODE_UPPER(enc,code) \
300         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_UPPER)
301 #define ONIGENC_IS_CODE_CNTRL(enc,code) \
302         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_CNTRL)
303 #define ONIGENC_IS_CODE_PUNCT(enc,code) \
304         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PUNCT)
305 #define ONIGENC_IS_CODE_SPACE(enc,code) \
306         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_SPACE)
307 #define ONIGENC_IS_CODE_BLANK(enc,code) \
308         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_BLANK)
309 #define ONIGENC_IS_CODE_DIGIT(enc,code) \
310         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_DIGIT)
311 #define ONIGENC_IS_CODE_XDIGIT(enc,code) \
312         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_XDIGIT)
313 #define ONIGENC_IS_CODE_WORD(enc,code) \
314         ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_WORD)
315 
316 #define ONIGENC_GET_CTYPE_CODE_RANGE(enc,ctype,sbout,ranges) \
317         (enc)->get_ctype_code_range(ctype,sbout,ranges)
318 
319 ONIG_EXTERN
320 OnigUChar* onigenc_step_back P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s, int n));
321 
322 
323 /* encoding API */
324 ONIG_EXTERN
325 int onigenc_init P_((void));
326 ONIG_EXTERN
327 int onig_initialize_encoding P_((OnigEncoding enc));
328 ONIG_EXTERN
329 int onigenc_set_default_encoding P_((OnigEncoding enc));
330 ONIG_EXTERN
331 OnigEncoding onigenc_get_default_encoding P_((void));
332 ONIG_EXTERN
333 void  onigenc_set_default_caseconv_table P_((const OnigUChar* table));
334 ONIG_EXTERN
335 OnigUChar* onigenc_get_right_adjust_char_head_with_prev P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s, const OnigUChar** prev));
336 ONIG_EXTERN
337 OnigUChar* onigenc_get_prev_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
338 ONIG_EXTERN
339 OnigUChar* onigenc_get_left_adjust_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
340 ONIG_EXTERN
341 OnigUChar* onigenc_get_right_adjust_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
342 ONIG_EXTERN
343 int onigenc_strlen P_((OnigEncoding enc, const OnigUChar* p, const OnigUChar* end));
344 ONIG_EXTERN
345 int onigenc_strlen_null P_((OnigEncoding enc, const OnigUChar* p));
346 ONIG_EXTERN
347 int onigenc_str_bytelen_null P_((OnigEncoding enc, const OnigUChar* p));
348 ONIG_EXTERN
349 int onigenc_is_valid_mbc_string P_((OnigEncoding enc, const OnigUChar* s, const OnigUChar* end));
350 ONIG_EXTERN
351 OnigUChar* onigenc_strdup P_((OnigEncoding enc, const OnigUChar* s, const OnigUChar* end));
352 
353 
354 /* PART: regular expression */
355 
356 /* config parameters */
357 #define ONIG_NREGION                          10
358 #define ONIG_MAX_CAPTURE_NUM          2147483647  /* 2**31 - 1 */
359 #define ONIG_MAX_BACKREF_NUM                1000
360 #define ONIG_MAX_REPEAT_NUM               100000
361 #define ONIG_MAX_MULTI_BYTE_RANGES_NUM     10000
362 /* constants */
363 #define ONIG_MAX_ERROR_MESSAGE_LEN            90
364 
365 typedef unsigned int        OnigOptionType;
366 
367 #define ONIG_OPTION_DEFAULT            ONIG_OPTION_NONE
368 
369 /* options */
370 #define ONIG_OPTION_NONE                 0U
371 /* options (compile time) */
372 #define ONIG_OPTION_IGNORECASE           1U
373 #define ONIG_OPTION_EXTEND               (ONIG_OPTION_IGNORECASE         << 1)
374 #define ONIG_OPTION_MULTILINE            (ONIG_OPTION_EXTEND             << 1)
375 #define ONIG_OPTION_SINGLELINE           (ONIG_OPTION_MULTILINE          << 1)
376 #define ONIG_OPTION_FIND_LONGEST         (ONIG_OPTION_SINGLELINE         << 1)
377 #define ONIG_OPTION_FIND_NOT_EMPTY       (ONIG_OPTION_FIND_LONGEST       << 1)
378 #define ONIG_OPTION_NEGATE_SINGLELINE    (ONIG_OPTION_FIND_NOT_EMPTY     << 1)
379 #define ONIG_OPTION_DONT_CAPTURE_GROUP   (ONIG_OPTION_NEGATE_SINGLELINE  << 1)
380 #define ONIG_OPTION_CAPTURE_GROUP        (ONIG_OPTION_DONT_CAPTURE_GROUP << 1)
381 /* options (search time) */
382 #define ONIG_OPTION_NOTBOL                    (ONIG_OPTION_CAPTURE_GROUP << 1)
383 #define ONIG_OPTION_NOTEOL                    (ONIG_OPTION_NOTBOL << 1)
384 #define ONIG_OPTION_POSIX_REGION              (ONIG_OPTION_NOTEOL << 1)
385 #define ONIG_OPTION_CHECK_VALIDITY_OF_STRING  (ONIG_OPTION_POSIX_REGION << 1)
386 /* #define ONIG_OPTION_CRLF_AS_LINE_SEPARATOR    (ONIG_OPTION_CHECK_VALIDITY_OF_STRING << 1) */
387 /* options (compile time) */
388 #define ONIG_OPTION_WORD_IS_ASCII        (ONIG_OPTION_CHECK_VALIDITY_OF_STRING << 4)
389 #define ONIG_OPTION_DIGIT_IS_ASCII       (ONIG_OPTION_WORD_IS_ASCII << 1)
390 #define ONIG_OPTION_SPACE_IS_ASCII       (ONIG_OPTION_DIGIT_IS_ASCII << 1)
391 #define ONIG_OPTION_POSIX_IS_ASCII       (ONIG_OPTION_SPACE_IS_ASCII << 1)
392 #define ONIG_OPTION_TEXT_SEGMENT_EXTENDED_GRAPHEME_CLUSTER  (ONIG_OPTION_POSIX_IS_ASCII << 1)
393 #define ONIG_OPTION_TEXT_SEGMENT_WORD    (ONIG_OPTION_TEXT_SEGMENT_EXTENDED_GRAPHEME_CLUSTER << 1)
394 
395 #define ONIG_OPTION_MAXBIT               ONIG_OPTION_TEXT_SEGMENT_WORD  /* limit */
396 
397 #define ONIG_OPTION_ON(options,regopt)      ((options) |= (regopt))
398 #define ONIG_OPTION_OFF(options,regopt)     ((options) &= ~(regopt))
399 #define ONIG_IS_OPTION_ON(options,option)   ((options) & (option))
400 
401 /* syntax */
402 typedef struct {
403   unsigned int   op;
404   unsigned int   op2;
405   unsigned int   behavior;
406   OnigOptionType options;   /* default option */
407   OnigMetaCharTableType meta_char_table;
408 } OnigSyntaxType;
409 
410 ONIG_EXTERN OnigSyntaxType OnigSyntaxASIS;
411 ONIG_EXTERN OnigSyntaxType OnigSyntaxPosixBasic;
412 ONIG_EXTERN OnigSyntaxType OnigSyntaxPosixExtended;
413 ONIG_EXTERN OnigSyntaxType OnigSyntaxEmacs;
414 ONIG_EXTERN OnigSyntaxType OnigSyntaxGrep;
415 ONIG_EXTERN OnigSyntaxType OnigSyntaxGnuRegex;
416 ONIG_EXTERN OnigSyntaxType OnigSyntaxJava;
417 ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl;
418 ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl_NG;
419 ONIG_EXTERN OnigSyntaxType OnigSyntaxRuby;
420 ONIG_EXTERN OnigSyntaxType OnigSyntaxOniguruma;
421 
422 /* predefined syntaxes (see regsyntax.c) */
423 #define ONIG_SYNTAX_ASIS               (&OnigSyntaxASIS)
424 #define ONIG_SYNTAX_POSIX_BASIC        (&OnigSyntaxPosixBasic)
425 #define ONIG_SYNTAX_POSIX_EXTENDED     (&OnigSyntaxPosixExtended)
426 #define ONIG_SYNTAX_EMACS              (&OnigSyntaxEmacs)
427 #define ONIG_SYNTAX_GREP               (&OnigSyntaxGrep)
428 #define ONIG_SYNTAX_GNU_REGEX          (&OnigSyntaxGnuRegex)
429 #define ONIG_SYNTAX_JAVA               (&OnigSyntaxJava)
430 #define ONIG_SYNTAX_PERL               (&OnigSyntaxPerl)
431 #define ONIG_SYNTAX_PERL_NG            (&OnigSyntaxPerl_NG)
432 #define ONIG_SYNTAX_RUBY               (&OnigSyntaxRuby)
433 #define ONIG_SYNTAX_ONIGURUMA          (&OnigSyntaxOniguruma)
434 
435 /* default syntax */
436 ONIG_EXTERN OnigSyntaxType*   OnigDefaultSyntax;
437 #define ONIG_SYNTAX_DEFAULT   OnigDefaultSyntax
438 
439 /* syntax (operators) */
440 #define ONIG_SYN_OP_VARIABLE_META_CHARACTERS    (1U<<0)
441 #define ONIG_SYN_OP_DOT_ANYCHAR                 (1U<<1)   /* . */
442 #define ONIG_SYN_OP_ASTERISK_ZERO_INF           (1U<<2)   /* * */
443 #define ONIG_SYN_OP_ESC_ASTERISK_ZERO_INF       (1U<<3)
444 #define ONIG_SYN_OP_PLUS_ONE_INF                (1U<<4)   /* + */
445 #define ONIG_SYN_OP_ESC_PLUS_ONE_INF            (1U<<5)
446 #define ONIG_SYN_OP_QMARK_ZERO_ONE              (1U<<6)   /* ? */
447 #define ONIG_SYN_OP_ESC_QMARK_ZERO_ONE          (1U<<7)
448 #define ONIG_SYN_OP_BRACE_INTERVAL              (1U<<8)   /* {lower,upper} */
449 #define ONIG_SYN_OP_ESC_BRACE_INTERVAL          (1U<<9)   /* \{lower,upper\} */
450 #define ONIG_SYN_OP_VBAR_ALT                    (1U<<10)   /* | */
451 #define ONIG_SYN_OP_ESC_VBAR_ALT                (1U<<11)  /* \| */
452 #define ONIG_SYN_OP_LPAREN_SUBEXP               (1U<<12)  /* (...)   */
453 #define ONIG_SYN_OP_ESC_LPAREN_SUBEXP           (1U<<13)  /* \(...\) */
454 #define ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR           (1U<<14)  /* \A, \Z, \z */
455 #define ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR  (1U<<15)  /* \G     */
456 #define ONIG_SYN_OP_DECIMAL_BACKREF             (1U<<16)  /* \num   */
457 #define ONIG_SYN_OP_BRACKET_CC                  (1U<<17)  /* [...]  */
458 #define ONIG_SYN_OP_ESC_W_WORD                  (1U<<18)  /* \w, \W */
459 #define ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END     (1U<<19)  /* \<. \> */
460 #define ONIG_SYN_OP_ESC_B_WORD_BOUND            (1U<<20)  /* \b, \B */
461 #define ONIG_SYN_OP_ESC_S_WHITE_SPACE           (1U<<21)  /* \s, \S */
462 #define ONIG_SYN_OP_ESC_D_DIGIT                 (1U<<22)  /* \d, \D */
463 #define ONIG_SYN_OP_LINE_ANCHOR                 (1U<<23)  /* ^, $   */
464 #define ONIG_SYN_OP_POSIX_BRACKET               (1U<<24)  /* [:xxxx:] */
465 #define ONIG_SYN_OP_QMARK_NON_GREEDY            (1U<<25)  /* ??,*?,+?,{n,m}? */
466 #define ONIG_SYN_OP_ESC_CONTROL_CHARS           (1U<<26)  /* \n,\r,\t,\a ... */
467 #define ONIG_SYN_OP_ESC_C_CONTROL               (1U<<27)  /* \cx  */
468 #define ONIG_SYN_OP_ESC_OCTAL3                  (1U<<28)  /* \OOO */
469 #define ONIG_SYN_OP_ESC_X_HEX2                  (1U<<29)  /* \xHH */
470 #define ONIG_SYN_OP_ESC_X_BRACE_HEX8            (1U<<30)  /* \x{7HHHHHHH} */
471 #define ONIG_SYN_OP_ESC_O_BRACE_OCTAL           (1U<<31)  /* \o{1OOOOOOOOOO} */
472 
473 #define ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE        (1U<<0)  /* \Q...\E */
474 #define ONIG_SYN_OP2_QMARK_GROUP_EFFECT         (1U<<1)  /* (?...) */
475 #define ONIG_SYN_OP2_OPTION_PERL                (1U<<2)  /* (?imsx),(?-imsx) */
476 #define ONIG_SYN_OP2_OPTION_RUBY                (1U<<3)  /* (?imx), (?-imx)  */
477 #define ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT     (1U<<4)  /* ?+,*+,++ */
478 #define ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL   (1U<<5)  /* {n,m}+   */
479 #define ONIG_SYN_OP2_CCLASS_SET_OP              (1U<<6)  /* [...&&..[..]..] */
480 #define ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP       (1U<<7)  /* (?<name>...) */
481 #define ONIG_SYN_OP2_ESC_K_NAMED_BACKREF        (1U<<8)  /* \k<name> */
482 #define ONIG_SYN_OP2_ESC_G_SUBEXP_CALL          (1U<<9)  /* \g<name>, \g<n> */
483 #define ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY     (1U<<10) /* (?@..),(?@<x>..) */
484 #define ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL  (1U<<11) /* \C-x */
485 #define ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META     (1U<<12) /* \M-x */
486 #define ONIG_SYN_OP2_ESC_V_VTAB                 (1U<<13) /* \v as VTAB */
487 #define ONIG_SYN_OP2_ESC_U_HEX4                 (1U<<14) /* \uHHHH */
488 #define ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR         (1U<<15) /* \`, \' */
489 #define ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY  (1U<<16) /* \p{...}, \P{...} */
490 #define ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT (1U<<17) /* \p{^..}, \P{^..} */
491 /* #define ONIG_SYN_OP2_CHAR_PROPERTY_PREFIX_IS (1U<<18) */
492 #define ONIG_SYN_OP2_ESC_H_XDIGIT               (1U<<19) /* \h, \H */
493 #define ONIG_SYN_OP2_INEFFECTIVE_ESCAPE         (1U<<20) /* \ */
494 #define ONIG_SYN_OP2_QMARK_LPAREN_IF_ELSE       (1U<<21) /* (?(n)) (?(...)...|...) */
495 #define ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP         (1U<<22) /* \K */
496 #define ONIG_SYN_OP2_ESC_CAPITAL_R_GENERAL_NEWLINE (1U<<23) /* \R \r\n else [\x0a-\x0d] */
497 #define ONIG_SYN_OP2_ESC_CAPITAL_N_O_SUPER_DOT  (1U<<24) /* \N (?-m:.), \O (?m:.) */
498 #define ONIG_SYN_OP2_QMARK_TILDE_ABSENT_GROUP   (1U<<25) /* (?~...) */
499 #define ONIG_SYN_OP2_ESC_X_Y_GRAPHEME_CLUSTER   (1U<<26) /* obsoleted: use next */
500 #define ONIG_SYN_OP2_ESC_X_Y_TEXT_SEGMENT       (1U<<26) /* \X \y \Y */
501 #define ONIG_SYN_OP2_QMARK_PERL_SUBEXP_CALL     (1U<<27) /* (?R), (?&name)... */
502 #define ONIG_SYN_OP2_QMARK_BRACE_CALLOUT_CONTENTS (1U<<28) /* (?{...}) (?{{...}}) */
503 #define ONIG_SYN_OP2_ASTERISK_CALLOUT_NAME      (1U<<29) /* (*name) (*name{a,..}) */
504 #define ONIG_SYN_OP2_OPTION_ONIGURUMA           (1U<<30) /* (?imxWDSPy) */
505 
506 /* syntax (behavior) */
507 #define ONIG_SYN_CONTEXT_INDEP_ANCHORS           (1U<<31) /* not implemented */
508 #define ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS        (1U<<0)  /* ?, *, +, {n,m} */
509 #define ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS      (1U<<1)  /* error or ignore */
510 #define ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP    (1U<<2)  /* ...)... */
511 #define ONIG_SYN_ALLOW_INVALID_INTERVAL          (1U<<3)  /* {??? */
512 #define ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV       (1U<<4)  /* {,n} => {0,n} */
513 #define ONIG_SYN_STRICT_CHECK_BACKREF            (1U<<5)  /* /(\1)/,/\1()/ ..*/
514 #define ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND   (1U<<6)  /* (?<=a|bc) */
515 #define ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP        (1U<<7)  /* see doc/RE */
516 #define ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME (1U<<8)  /* (?<x>)(?<x>) */
517 #define ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY   (1U<<9)  /* a{n}?=(?:a{n})? */
518 
519 /* syntax (behavior) in char class [...] */
520 #define ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC      (1U<<20) /* [^...] */
521 #define ONIG_SYN_BACKSLASH_ESCAPE_IN_CC          (1U<<21) /* [..\w..] etc.. */
522 #define ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC         (1U<<22)
523 #define ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC     (1U<<23) /* [0-9-a]=[0-9\-a] */
524 #define ONIG_SYN_ALLOW_INVALID_CODE_END_OF_RANGE_IN_CC (1U<<26)
525 /* syntax (behavior) warning */
526 #define ONIG_SYN_WARN_CC_OP_NOT_ESCAPED          (1U<<24) /* [,-,] */
527 #define ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT    (1U<<25) /* (?:a*)+ */
528 
529 /* meta character specifiers (onig_set_meta_char()) */
530 #define ONIG_META_CHAR_ESCAPE               0
531 #define ONIG_META_CHAR_ANYCHAR              1
532 #define ONIG_META_CHAR_ANYTIME              2
533 #define ONIG_META_CHAR_ZERO_OR_ONE_TIME     3
534 #define ONIG_META_CHAR_ONE_OR_MORE_TIME     4
535 #define ONIG_META_CHAR_ANYCHAR_ANYTIME      5
536 
537 #define ONIG_INEFFECTIVE_META_CHAR          0
538 
539 /* error codes */
540 #define ONIG_IS_PATTERN_ERROR(ecode)   ((ecode) <= -100 && (ecode) > -1000)
541 /* normal return */
542 #define ONIG_NORMAL                                            0
543 #define ONIG_MISMATCH                                         -1
544 #define ONIG_NO_SUPPORT_CONFIG                                -2
545 #define ONIG_ABORT                                            -3
546 
547 /* internal error */
548 #define ONIGERR_MEMORY                                         -5
549 #define ONIGERR_TYPE_BUG                                       -6
550 #define ONIGERR_PARSER_BUG                                    -11
551 #define ONIGERR_STACK_BUG                                     -12
552 #define ONIGERR_UNDEFINED_BYTECODE                            -13
553 #define ONIGERR_UNEXPECTED_BYTECODE                           -14
554 #define ONIGERR_MATCH_STACK_LIMIT_OVER                        -15
555 #define ONIGERR_PARSE_DEPTH_LIMIT_OVER                        -16
556 #define ONIGERR_RETRY_LIMIT_IN_MATCH_OVER                     -17
557 #define ONIGERR_DEFAULT_ENCODING_IS_NOT_SETTED                -21
558 #define ONIGERR_SPECIFIED_ENCODING_CANT_CONVERT_TO_WIDE_CHAR  -22
559 #define ONIGERR_FAIL_TO_INITIALIZE                            -23
560 /* general error */
561 #define ONIGERR_INVALID_ARGUMENT                              -30
562 /* syntax error */
563 #define ONIGERR_END_PATTERN_AT_LEFT_BRACE                    -100
564 #define ONIGERR_END_PATTERN_AT_LEFT_BRACKET                  -101
565 #define ONIGERR_EMPTY_CHAR_CLASS                             -102
566 #define ONIGERR_PREMATURE_END_OF_CHAR_CLASS                  -103
567 #define ONIGERR_END_PATTERN_AT_ESCAPE                        -104
568 #define ONIGERR_END_PATTERN_AT_META                          -105
569 #define ONIGERR_END_PATTERN_AT_CONTROL                       -106
570 #define ONIGERR_META_CODE_SYNTAX                             -108
571 #define ONIGERR_CONTROL_CODE_SYNTAX                          -109
572 #define ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE             -110
573 #define ONIGERR_CHAR_CLASS_VALUE_AT_START_OF_RANGE           -111
574 #define ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS      -112
575 #define ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED      -113
576 #define ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID            -114
577 #define ONIGERR_NESTED_REPEAT_OPERATOR                       -115
578 #define ONIGERR_UNMATCHED_CLOSE_PARENTHESIS                  -116
579 #define ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS       -117
580 #define ONIGERR_END_PATTERN_IN_GROUP                         -118
581 #define ONIGERR_UNDEFINED_GROUP_OPTION                       -119
582 #define ONIGERR_INVALID_POSIX_BRACKET_TYPE                   -121
583 #define ONIGERR_INVALID_LOOK_BEHIND_PATTERN                  -122
584 #define ONIGERR_INVALID_REPEAT_RANGE_PATTERN                 -123
585 /* values error (syntax error) */
586 #define ONIGERR_TOO_BIG_NUMBER                               -200
587 #define ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE              -201
588 #define ONIGERR_UPPER_SMALLER_THAN_LOWER_IN_REPEAT_RANGE     -202
589 #define ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS                    -203
590 #define ONIGERR_MISMATCH_CODE_LENGTH_IN_CLASS_RANGE          -204
591 #define ONIGERR_TOO_MANY_MULTI_BYTE_RANGES                   -205
592 #define ONIGERR_TOO_SHORT_MULTI_BYTE_STRING                  -206
593 #define ONIGERR_TOO_BIG_BACKREF_NUMBER                       -207
594 #define ONIGERR_INVALID_BACKREF                              -208
595 #define ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED         -209
596 #define ONIGERR_TOO_MANY_CAPTURES                            -210
597 #define ONIGERR_TOO_LONG_WIDE_CHAR_VALUE                     -212
598 #define ONIGERR_EMPTY_GROUP_NAME                             -214
599 #define ONIGERR_INVALID_GROUP_NAME                           -215
600 #define ONIGERR_INVALID_CHAR_IN_GROUP_NAME                   -216
601 #define ONIGERR_UNDEFINED_NAME_REFERENCE                     -217
602 #define ONIGERR_UNDEFINED_GROUP_REFERENCE                    -218
603 #define ONIGERR_MULTIPLEX_DEFINED_NAME                       -219
604 #define ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL               -220
605 #define ONIGERR_NEVER_ENDING_RECURSION                       -221
606 #define ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY        -222
607 #define ONIGERR_INVALID_CHAR_PROPERTY_NAME                   -223
608 #define ONIGERR_INVALID_IF_ELSE_SYNTAX                       -224
609 #define ONIGERR_INVALID_ABSENT_GROUP_PATTERN                 -225
610 #define ONIGERR_INVALID_ABSENT_GROUP_GENERATOR_PATTERN       -226
611 #define ONIGERR_INVALID_CALLOUT_PATTERN                      -227
612 #define ONIGERR_INVALID_CALLOUT_NAME                         -228
613 #define ONIGERR_UNDEFINED_CALLOUT_NAME                       -229
614 #define ONIGERR_INVALID_CALLOUT_BODY                         -230
615 #define ONIGERR_INVALID_CALLOUT_TAG_NAME                     -231
616 #define ONIGERR_INVALID_CALLOUT_ARG                          -232
617 #define ONIGERR_INVALID_CODE_POINT_VALUE                     -400
618 #define ONIGERR_INVALID_WIDE_CHAR_VALUE                      -400
619 #define ONIGERR_TOO_BIG_WIDE_CHAR_VALUE                      -401
620 #define ONIGERR_NOT_SUPPORTED_ENCODING_COMBINATION           -402
621 #define ONIGERR_INVALID_COMBINATION_OF_OPTIONS               -403
622 #define ONIGERR_TOO_MANY_USER_DEFINED_OBJECTS                -404
623 #define ONIGERR_TOO_LONG_PROPERTY_NAME                       -405
624 #define ONIGERR_LIBRARY_IS_NOT_INITIALIZED                   -500
625 
626 /* errors related to thread */
627 /* #define ONIGERR_OVER_THREAD_PASS_LIMIT_COUNT                -1001 */
628 
629 
630 /* must be smaller than MEM_STATUS_BITS_NUM (unsigned int * 8) */
631 #define ONIG_MAX_CAPTURE_HISTORY_GROUP   31
632 #define ONIG_IS_CAPTURE_HISTORY_GROUP(r, i) \
633   ((i) <= ONIG_MAX_CAPTURE_HISTORY_GROUP && (r)->list && (r)->list[i])
634 
635 typedef struct OnigCaptureTreeNodeStruct {
636   int group;   /* group number */
637   int beg;
638   int end;
639   int allocated;
640   int num_childs;
641   struct OnigCaptureTreeNodeStruct** childs;
642 } OnigCaptureTreeNode;
643 
644 /* match result region type */
645 struct re_registers {
646   int  allocated;
647   int  num_regs;
648   int* beg;
649   int* end;
650   /* extended */
651   OnigCaptureTreeNode* history_root;  /* capture history tree root */
652 };
653 
654 /* capture tree traverse */
655 #define ONIG_TRAVERSE_CALLBACK_AT_FIRST   1
656 #define ONIG_TRAVERSE_CALLBACK_AT_LAST    2
657 #define ONIG_TRAVERSE_CALLBACK_AT_BOTH \
658   ( ONIG_TRAVERSE_CALLBACK_AT_FIRST | ONIG_TRAVERSE_CALLBACK_AT_LAST )
659 
660 
661 #define ONIG_REGION_NOTPOS            -1
662 
663 typedef struct re_registers   OnigRegion;
664 
665 typedef struct {
666   OnigEncoding enc;
667   OnigUChar* par;
668   OnigUChar* par_end;
669 } OnigErrorInfo;
670 
671 typedef struct {
672   int lower;
673   int upper;
674 } OnigRepeatRange;
675 
676 typedef void (*OnigWarnFunc) P_((const char* s));
677 extern void onig_null_warn P_((const char* s));
678 #define ONIG_NULL_WARN       onig_null_warn
679 
680 #define ONIG_CHAR_TABLE_SIZE   256
681 
682 struct re_pattern_buffer;
683 typedef struct re_pattern_buffer OnigRegexType;
684 typedef OnigRegexType*  OnigRegex;
685 
686 #ifndef ONIG_ESCAPE_REGEX_T_COLLISION
687   typedef OnigRegexType  regex_t;
688 #endif
689 
690 struct OnigRegSetStruct;
691 typedef struct OnigRegSetStruct OnigRegSet;
692 
693 typedef enum {
694   ONIG_REGSET_POSITION_LEAD = 0,
695   ONIG_REGSET_REGEX_LEAD    = 1,
696   ONIG_REGSET_PRIORITY_TO_REGEX_ORDER = 2
697 } OnigRegSetLead;
698 
699 typedef struct {
700   int             num_of_elements;
701   OnigEncoding    pattern_enc;
702   OnigEncoding    target_enc;
703   OnigSyntaxType* syntax;
704   OnigOptionType  option;
705   OnigCaseFoldType   case_fold_flag;
706 } OnigCompileInfo;
707 
708 
709 /* types for callout */
710 typedef enum {
711   ONIG_CALLOUT_IN_PROGRESS   = 1, /* 1<<0 */
712   ONIG_CALLOUT_IN_RETRACTION = 2  /* 1<<1 */
713 } OnigCalloutIn;
714 
715 #define ONIG_CALLOUT_IN_BOTH  (ONIG_CALLOUT_IN_PROGRESS | ONIG_CALLOUT_IN_RETRACTION)
716 
717 typedef enum {
718   ONIG_CALLOUT_OF_CONTENTS = 0,
719   ONIG_CALLOUT_OF_NAME     = 1
720 } OnigCalloutOf;
721 
722 typedef enum {
723   ONIG_CALLOUT_TYPE_SINGLE              = 0,
724   ONIG_CALLOUT_TYPE_START_CALL          = 1,
725   ONIG_CALLOUT_TYPE_BOTH_CALL           = 2,
726   ONIG_CALLOUT_TYPE_START_MARK_END_CALL = 3,
727 } OnigCalloutType;
728 
729 
730 #define ONIG_NON_NAME_ID        -1
731 #define ONIG_NON_CALLOUT_NUM     0
732 
733 #define ONIG_CALLOUT_MAX_ARGS_NUM     4
734 #define ONIG_CALLOUT_DATA_SLOT_NUM    5
735 
736 struct OnigCalloutArgsStruct;
737 typedef struct OnigCalloutArgsStruct OnigCalloutArgs;
738 
739 typedef int (*OnigCalloutFunc)(OnigCalloutArgs* args, void* user_data);
740 
741 /* callout function return values (less than -1: error code) */
742 typedef enum {
743   ONIG_CALLOUT_FAIL     =  1,
744   ONIG_CALLOUT_SUCCESS  =  0
745 } OnigCalloutResult;
746 
747 typedef enum {
748   ONIG_TYPE_VOID     = 0,
749   ONIG_TYPE_LONG     = 1<<0,
750   ONIG_TYPE_CHAR     = 1<<1,
751   ONIG_TYPE_STRING   = 1<<2,
752   ONIG_TYPE_POINTER  = 1<<3,
753   ONIG_TYPE_TAG      = 1<<4,
754 } OnigType;
755 
756 typedef union {
757   long l;
758   OnigCodePoint c;
759   struct {
760     OnigUChar* start;
761     OnigUChar* end;
762   } s;
763   void* p;
764   int tag;  /* tag -> callout_num */
765 } OnigValue;
766 
767 
768 struct OnigMatchParamStruct;
769 typedef struct OnigMatchParamStruct OnigMatchParam;
770 
771 
772 /* Oniguruma Native API */
773 
774 ONIG_EXTERN
775 int onig_initialize P_((OnigEncoding encodings[], int number_of_encodings));
776 /* onig_init(): deprecated function. Use onig_initialize(). */
777 ONIG_EXTERN
778 int onig_init P_((void));
779 ONIG_EXTERN
780 int onig_error_code_to_str PV_((OnigUChar* s, int err_code, ...));
781 ONIG_EXTERN
782 int onig_is_error_code_needs_param PV_((int code));
783 ONIG_EXTERN
784 void onig_set_warn_func P_((OnigWarnFunc f));
785 ONIG_EXTERN
786 void onig_set_verb_warn_func P_((OnigWarnFunc f));
787 ONIG_EXTERN
788 int onig_new P_((OnigRegex*, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax, OnigErrorInfo* einfo));
789 ONIG_EXTERN
790 int  onig_reg_init P_((OnigRegex reg, OnigOptionType option, OnigCaseFoldType case_fold_flag, OnigEncoding enc, OnigSyntaxType* syntax));
791 int onig_new_without_alloc P_((OnigRegex, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax, OnigErrorInfo* einfo));
792 ONIG_EXTERN
793 int onig_new_deluxe P_((OnigRegex* reg, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigCompileInfo* ci, OnigErrorInfo* einfo));
794 ONIG_EXTERN
795 void onig_free P_((OnigRegex));
796 ONIG_EXTERN
797 void onig_free_body P_((OnigRegex));
798 ONIG_EXTERN
799 int onig_scan(OnigRegex reg, const OnigUChar* str, const OnigUChar* end, OnigRegion* region, OnigOptionType option, int (*scan_callback)(int, int, OnigRegion*, void*), void* callback_arg);
800 ONIG_EXTERN
801 int onig_search P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegion* region, OnigOptionType option));
802 ONIG_EXTERN
803 int onig_search_with_param P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegion* region, OnigOptionType option, OnigMatchParam* mp));
804 ONIG_EXTERN
805 int onig_match P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* at, OnigRegion* region, OnigOptionType option));
806 ONIG_EXTERN
807 int onig_match_with_param P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* at, OnigRegion* region, OnigOptionType option, OnigMatchParam* mp));
808 
809 ONIG_EXTERN
810 int onig_regset_new P_((OnigRegSet** rset, int n, regex_t* regs[]));
811 ONIG_EXTERN
812 int onig_regset_add P_((OnigRegSet* set, regex_t* reg));
813 ONIG_EXTERN
814 int onig_regset_replace P_((OnigRegSet* set, int at, regex_t* reg));
815 ONIG_EXTERN
816 void onig_regset_free P_((OnigRegSet* set));
817 ONIG_EXTERN
818 int onig_regset_number_of_regex P_((OnigRegSet* set));
819 ONIG_EXTERN
820 regex_t* onig_regset_get_regex P_((OnigRegSet* set, int at));
821 ONIG_EXTERN
822 OnigRegion* onig_regset_get_region P_((OnigRegSet* set, int at));
823 ONIG_EXTERN
824 int onig_regset_search P_((OnigRegSet* set, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegSetLead lead, OnigOptionType option, int* rmatch_pos));
825 ONIG_EXTERN
826 int onig_regset_search_with_param P_((OnigRegSet* set, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range,  OnigRegSetLead lead, OnigOptionType option, OnigMatchParam* mps[], int* rmatch_pos));
827 
828 ONIG_EXTERN
829 OnigRegion* onig_region_new P_((void));
830 ONIG_EXTERN
831 void onig_region_init P_((OnigRegion* region));
832 ONIG_EXTERN
833 void onig_region_free P_((OnigRegion* region, int free_self));
834 ONIG_EXTERN
835 void onig_region_copy P_((OnigRegion* to, OnigRegion* from));
836 ONIG_EXTERN
837 void onig_region_clear P_((OnigRegion* region));
838 ONIG_EXTERN
839 int onig_region_resize P_((OnigRegion* region, int n));
840 ONIG_EXTERN
841 int onig_region_set P_((OnigRegion* region, int at, int beg, int end));
842 ONIG_EXTERN
843 int onig_name_to_group_numbers P_((OnigRegex reg, const OnigUChar* name, const OnigUChar* name_end, int** nums));
844 ONIG_EXTERN
845 int onig_name_to_backref_number P_((OnigRegex reg, const OnigUChar* name, const OnigUChar* name_end, OnigRegion *region));
846 ONIG_EXTERN
847 int onig_foreach_name P_((OnigRegex reg, int (*func)(const OnigUChar*, const OnigUChar*,int,int*,OnigRegex,void*), void* arg));
848 ONIG_EXTERN
849 int onig_number_of_names P_((OnigRegex reg));
850 ONIG_EXTERN
851 int onig_number_of_captures P_((OnigRegex reg));
852 ONIG_EXTERN
853 int onig_number_of_capture_histories P_((OnigRegex reg));
854 ONIG_EXTERN
855 OnigCaptureTreeNode* onig_get_capture_tree P_((OnigRegion* region));
856 ONIG_EXTERN
857 int onig_capture_tree_traverse P_((OnigRegion* region, int at, int(*callback_func)(int,int,int,int,int,void*), void* arg));
858 ONIG_EXTERN
859 int onig_noname_group_capture_is_active P_((OnigRegex reg));
860 ONIG_EXTERN
861 OnigEncoding onig_get_encoding P_((OnigRegex reg));
862 ONIG_EXTERN
863 OnigOptionType onig_get_options P_((OnigRegex reg));
864 ONIG_EXTERN
865 OnigCaseFoldType onig_get_case_fold_flag P_((OnigRegex reg));
866 ONIG_EXTERN
867 OnigSyntaxType* onig_get_syntax P_((OnigRegex reg));
868 ONIG_EXTERN
869 int onig_set_default_syntax P_((OnigSyntaxType* syntax));
870 ONIG_EXTERN
871 void onig_copy_syntax P_((OnigSyntaxType* to, OnigSyntaxType* from));
872 ONIG_EXTERN
873 unsigned int onig_get_syntax_op P_((OnigSyntaxType* syntax));
874 ONIG_EXTERN
875 unsigned int onig_get_syntax_op2 P_((OnigSyntaxType* syntax));
876 ONIG_EXTERN
877 unsigned int onig_get_syntax_behavior P_((OnigSyntaxType* syntax));
878 ONIG_EXTERN
879 OnigOptionType onig_get_syntax_options P_((OnigSyntaxType* syntax));
880 ONIG_EXTERN
881 void onig_set_syntax_op P_((OnigSyntaxType* syntax, unsigned int op));
882 ONIG_EXTERN
883 void onig_set_syntax_op2 P_((OnigSyntaxType* syntax, unsigned int op2));
884 ONIG_EXTERN
885 void onig_set_syntax_behavior P_((OnigSyntaxType* syntax, unsigned int behavior));
886 ONIG_EXTERN
887 void onig_set_syntax_options P_((OnigSyntaxType* syntax, OnigOptionType options));
888 ONIG_EXTERN
889 int onig_set_meta_char P_((OnigSyntaxType* syntax, unsigned int what, OnigCodePoint code));
890 ONIG_EXTERN
891 void onig_copy_encoding P_((OnigEncoding to, OnigEncoding from));
892 ONIG_EXTERN
893 OnigCaseFoldType onig_get_default_case_fold_flag P_((void));
894 ONIG_EXTERN
895 int onig_set_default_case_fold_flag P_((OnigCaseFoldType case_fold_flag));
896 ONIG_EXTERN
897 unsigned int onig_get_match_stack_limit_size P_((void));
898 ONIG_EXTERN
899 int onig_set_match_stack_limit_size P_((unsigned int size));
900 ONIG_EXTERN
901 unsigned long onig_get_retry_limit_in_match P_((void));
902 ONIG_EXTERN
903 int onig_set_retry_limit_in_match P_((unsigned long n));
904 ONIG_EXTERN
905 unsigned int onig_get_parse_depth_limit P_((void));
906 ONIG_EXTERN
907 int onig_set_capture_num_limit P_((int num));
908 ONIG_EXTERN
909 int onig_set_parse_depth_limit P_((unsigned int depth));
910 ONIG_EXTERN
911 int onig_unicode_define_user_property P_((const char* name, OnigCodePoint* ranges));
912 ONIG_EXTERN
913 int onig_end P_((void));
914 ONIG_EXTERN
915 const char* onig_version P_((void));
916 ONIG_EXTERN
917 const char* onig_copyright P_((void));
918 
919 /* for OnigMatchParam */
920 ONIG_EXTERN
921 OnigMatchParam* onig_new_match_param P_((void));
922 ONIG_EXTERN
923 void onig_free_match_param P_((OnigMatchParam* p));
924 ONIG_EXTERN
925 void onig_free_match_param_content P_((OnigMatchParam* p));
926 ONIG_EXTERN
927 int onig_initialize_match_param P_((OnigMatchParam* mp));
928 ONIG_EXTERN
929 int onig_set_match_stack_limit_size_of_match_param P_((OnigMatchParam* param, unsigned int limit));
930 ONIG_EXTERN
931 int onig_set_retry_limit_in_match_of_match_param P_((OnigMatchParam* param, unsigned long limit));
932 ONIG_EXTERN
933 int onig_set_progress_callout_of_match_param P_((OnigMatchParam* param, OnigCalloutFunc f));
934 ONIG_EXTERN
935 int onig_set_retraction_callout_of_match_param P_((OnigMatchParam* param, OnigCalloutFunc f));
936 ONIG_EXTERN
937 int onig_set_callout_user_data_of_match_param P_((OnigMatchParam* param, void* user_data));
938 
939 /* for callout functions */
940 ONIG_EXTERN
941 OnigCalloutFunc onig_get_progress_callout P_((void));
942 ONIG_EXTERN
943 int onig_set_progress_callout P_((OnigCalloutFunc f));
944 ONIG_EXTERN
945 OnigCalloutFunc onig_get_retraction_callout P_((void));
946 ONIG_EXTERN
947 int onig_set_retraction_callout P_((OnigCalloutFunc f));
948 ONIG_EXTERN
949 int onig_set_callout_of_name P_((OnigEncoding enc, OnigCalloutType type, OnigUChar* name, OnigUChar* name_end, int callout_in, OnigCalloutFunc callout, OnigCalloutFunc end_callout, int arg_num, unsigned int arg_types[], int optional_arg_num, OnigValue opt_defaults[]));
950 ONIG_EXTERN
951 OnigUChar* onig_get_callout_name_by_name_id P_((int id));
952 ONIG_EXTERN
953 int onig_get_callout_num_by_tag P_((OnigRegex reg, const OnigUChar* tag, const OnigUChar* tag_end));
954 ONIG_EXTERN
955 int onig_get_callout_data_by_tag P_((OnigRegex reg, OnigMatchParam* mp, const OnigUChar* tag, const OnigUChar* tag_end, int slot, OnigType* type, OnigValue* val));
956 ONIG_EXTERN
957 int onig_set_callout_data_by_tag P_((OnigRegex reg, OnigMatchParam* mp, const OnigUChar* tag, const OnigUChar* tag_end, int slot, OnigType type, OnigValue* val));
958 
959 /* used in callout functions */
960 ONIG_EXTERN
961 int onig_get_callout_num_by_callout_args P_((OnigCalloutArgs* args));
962 ONIG_EXTERN
963 OnigCalloutIn onig_get_callout_in_by_callout_args P_((OnigCalloutArgs* args));
964 ONIG_EXTERN
965 int onig_get_name_id_by_callout_args P_((OnigCalloutArgs* args));
966 ONIG_EXTERN
967 const OnigUChar* onig_get_contents_by_callout_args P_((OnigCalloutArgs* args));
968 ONIG_EXTERN
969 const OnigUChar* onig_get_contents_end_by_callout_args P_((OnigCalloutArgs* args));
970 ONIG_EXTERN
971 int onig_get_args_num_by_callout_args P_((OnigCalloutArgs* args));
972 ONIG_EXTERN
973 int onig_get_passed_args_num_by_callout_args P_((OnigCalloutArgs* args));
974 ONIG_EXTERN
975 int onig_get_arg_by_callout_args P_((OnigCalloutArgs* args, int index, OnigType* type, OnigValue* val));
976 ONIG_EXTERN
977 const OnigUChar* onig_get_string_by_callout_args P_((OnigCalloutArgs* args));
978 ONIG_EXTERN
979 const OnigUChar* onig_get_string_end_by_callout_args P_((OnigCalloutArgs* args));
980 ONIG_EXTERN
981 const OnigUChar* onig_get_start_by_callout_args P_((OnigCalloutArgs* args));
982 ONIG_EXTERN
983 const OnigUChar* onig_get_right_range_by_callout_args P_((OnigCalloutArgs* args));
984 ONIG_EXTERN
985 const OnigUChar* onig_get_current_by_callout_args P_((OnigCalloutArgs* args));
986 ONIG_EXTERN
987 OnigRegex onig_get_regex_by_callout_args P_((OnigCalloutArgs* args));
988 ONIG_EXTERN
989 unsigned long onig_get_retry_counter_by_callout_args P_((OnigCalloutArgs* args));
990 ONIG_EXTERN
991 int onig_callout_tag_is_exist_at_callout_num P_((OnigRegex reg, int callout_num));
992 ONIG_EXTERN
993 const OnigUChar* onig_get_callout_tag_start P_((OnigRegex reg, int callout_num));
994 ONIG_EXTERN
995 const OnigUChar* onig_get_callout_tag_end P_((OnigRegex reg, int callout_num));
996 ONIG_EXTERN
997 int onig_get_callout_data_dont_clear_old P_((OnigRegex reg, OnigMatchParam* mp, int callout_num, int slot, OnigType* type, OnigValue* val));
998 ONIG_EXTERN
999 int onig_get_callout_data_by_callout_args_self_dont_clear_old P_((OnigCalloutArgs* args, int slot, OnigType* type, OnigValue* val));
1000 ONIG_EXTERN
1001 int onig_get_callout_data P_((OnigRegex reg, OnigMatchParam* mp, int callout_num, int slot, OnigType* type, OnigValue* val));
1002 ONIG_EXTERN
1003 int onig_get_callout_data_by_callout_args P_((OnigCalloutArgs* args, int callout_num, int slot, OnigType* type, OnigValue* val));
1004 ONIG_EXTERN
1005 int onig_get_callout_data_by_callout_args_self P_((OnigCalloutArgs* args, int slot, OnigType* type, OnigValue* val));
1006 ONIG_EXTERN
1007 int onig_set_callout_data P_((OnigRegex reg, OnigMatchParam* mp, int callout_num, int slot, OnigType type, OnigValue* val));
1008 ONIG_EXTERN
1009 int onig_set_callout_data_by_callout_args P_((OnigCalloutArgs* args, int callout_num, int slot, OnigType type, OnigValue* val));
1010 ONIG_EXTERN
1011 int onig_set_callout_data_by_callout_args_self P_((OnigCalloutArgs* args, int slot, OnigType type, OnigValue* val));
1012 ONIG_EXTERN
1013 int onig_get_capture_range_in_callout P_((OnigCalloutArgs* args, int mem_num, int* begin, int* end));
1014 ONIG_EXTERN
1015 int onig_get_used_stack_size_in_callout P_((OnigCalloutArgs* args, int* used_num, int* used_bytes));
1016 
1017 /* builtin callout functions */
1018 ONIG_EXTERN
1019 int onig_builtin_fail P_((OnigCalloutArgs* args, void* user_data));
1020 ONIG_EXTERN
1021 int onig_builtin_mismatch P_((OnigCalloutArgs* args, void* user_data));
1022 ONIG_EXTERN
1023 int onig_builtin_error P_((OnigCalloutArgs* args, void* user_data));
1024 ONIG_EXTERN
1025 int onig_builtin_count P_((OnigCalloutArgs* args, void* user_data));
1026 ONIG_EXTERN
1027 int onig_builtin_total_count P_((OnigCalloutArgs* args, void* user_data));
1028 ONIG_EXTERN
1029 int onig_builtin_max P_((OnigCalloutArgs* args, void* user_data));
1030 ONIG_EXTERN
1031 int onig_builtin_cmp P_((OnigCalloutArgs* args, void* user_data));
1032 
1033 ONIG_EXTERN
1034 int onig_setup_builtin_monitors_by_ascii_encoded_name P_((void* fp));
1035 
1036 #ifdef __cplusplus
1037 }
1038 #endif
1039 
1040 #endif /* ONIGURUMA_H */
1041