1 | #ifndef ONIGURUMA_H
|
---|
2 | #define ONIGURUMA_H
|
---|
3 | /**********************************************************************
|
---|
4 | oniguruma.h - Onigmo (Oniguruma-mod) (regular expression library)
|
---|
5 | **********************************************************************/
|
---|
6 | /*-
|
---|
7 | * Copyright (c) 2002-2009 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
|
---|
8 | * Copyright (c) 2011-2014 K.Takata <kentkt AT csc DOT jp>
|
---|
9 | * All rights reserved.
|
---|
10 | *
|
---|
11 | * Redistribution and use in source and binary forms, with or without
|
---|
12 | * modification, are permitted provided that the following conditions
|
---|
13 | * are met:
|
---|
14 | * 1. Redistributions of source code must retain the above copyright
|
---|
15 | * notice, this list of conditions and the following disclaimer.
|
---|
16 | * 2. Redistributions in binary form must reproduce the above copyright
|
---|
17 | * notice, this list of conditions and the following disclaimer in the
|
---|
18 | * documentation and/or other materials provided with the distribution.
|
---|
19 | *
|
---|
20 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
---|
21 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
---|
22 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
---|
23 | * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
---|
24 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
---|
25 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
---|
26 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
---|
27 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
---|
28 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
---|
29 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
---|
30 | * SUCH DAMAGE.
|
---|
31 | */
|
---|
32 |
|
---|
33 | #ifdef __cplusplus
|
---|
34 | extern "C" {
|
---|
35 | #endif
|
---|
36 |
|
---|
37 | #define ONIGURUMA
|
---|
38 | #define ONIGURUMA_VERSION_MAJOR 5
|
---|
39 | #define ONIGURUMA_VERSION_MINOR 15
|
---|
40 | #define ONIGURUMA_VERSION_TEENY 0
|
---|
41 |
|
---|
42 | #ifdef __cplusplus
|
---|
43 | # ifndef HAVE_PROTOTYPES
|
---|
44 | # define HAVE_PROTOTYPES 1
|
---|
45 | # endif
|
---|
46 | # ifndef HAVE_STDARG_PROTOTYPES
|
---|
47 | # define HAVE_STDARG_PROTOTYPES 1
|
---|
48 | # endif
|
---|
49 | #endif
|
---|
50 |
|
---|
51 | /* escape Mac OS X/Xcode 2.4/gcc 4.0.1 problem */
|
---|
52 | #if defined(__APPLE__) && defined(__GNUC__) && __GNUC__ >= 4
|
---|
53 | # ifndef HAVE_STDARG_PROTOTYPES
|
---|
54 | # define HAVE_STDARG_PROTOTYPES 1
|
---|
55 | # endif
|
---|
56 | #endif
|
---|
57 |
|
---|
58 | #if defined(HAVE_STDARG_H) || defined(_WIN32)
|
---|
59 | # ifndef HAVE_STDARG_PROTOTYPES
|
---|
60 | # define HAVE_STDARG_PROTOTYPES 1
|
---|
61 | # endif
|
---|
62 | #endif
|
---|
63 |
|
---|
64 | #ifndef P_
|
---|
65 | #if defined(__STDC__) || defined(_WIN32)
|
---|
66 | # define P_(args) args
|
---|
67 | #else
|
---|
68 | # define P_(args) ()
|
---|
69 | #endif
|
---|
70 | #endif
|
---|
71 |
|
---|
72 | #ifndef PV_
|
---|
73 | #ifdef HAVE_STDARG_PROTOTYPES
|
---|
74 | # define PV_(args) args
|
---|
75 | #else
|
---|
76 | # define PV_(args) ()
|
---|
77 | #endif
|
---|
78 | #endif
|
---|
79 |
|
---|
80 | #ifndef ONIG_EXTERN
|
---|
81 | #if defined(_WIN32) && !defined(__GNUC__)
|
---|
82 | #if defined(EXPORT) || defined(RUBY_EXPORT)
|
---|
83 | #define ONIG_EXTERN extern __declspec(dllexport)
|
---|
84 | #else
|
---|
85 | #define ONIG_EXTERN extern __declspec(dllimport)
|
---|
86 | #endif
|
---|
87 | #endif
|
---|
88 | #endif
|
---|
89 |
|
---|
90 | #ifndef ONIG_EXTERN
|
---|
91 | #define ONIG_EXTERN extern
|
---|
92 | #endif
|
---|
93 |
|
---|
94 | #include <stddef.h> /* for size_t */
|
---|
95 |
|
---|
96 | /* PART: character encoding */
|
---|
97 |
|
---|
98 | #ifndef ONIG_ESCAPE_UCHAR_COLLISION
|
---|
99 | #define UChar OnigUChar
|
---|
100 | #endif
|
---|
101 |
|
---|
102 | typedef unsigned char OnigUChar;
|
---|
103 | typedef unsigned int OnigCodePoint;
|
---|
104 | typedef unsigned int OnigCtype;
|
---|
105 | typedef size_t OnigDistance;
|
---|
106 | typedef ptrdiff_t OnigPosition;
|
---|
107 |
|
---|
108 | #define ONIG_INFINITE_DISTANCE ~((OnigDistance )0)
|
---|
109 |
|
---|
110 | typedef unsigned int OnigCaseFoldType; /* case fold flag */
|
---|
111 |
|
---|
112 | ONIG_EXTERN OnigCaseFoldType OnigDefaultCaseFoldFlag;
|
---|
113 |
|
---|
114 | /* #define ONIGENC_CASE_FOLD_HIRAGANA_KATAKANA (1<<1) */
|
---|
115 | /* #define ONIGENC_CASE_FOLD_KATAKANA_WIDTH (1<<2) */
|
---|
116 | #define ONIGENC_CASE_FOLD_TURKISH_AZERI (1<<20)
|
---|
117 | #define INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR (1<<30)
|
---|
118 |
|
---|
119 | #define ONIGENC_CASE_FOLD_MIN INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR
|
---|
120 | #define ONIGENC_CASE_FOLD_DEFAULT OnigDefaultCaseFoldFlag
|
---|
121 |
|
---|
122 |
|
---|
123 | #define ONIGENC_MAX_COMP_CASE_FOLD_CODE_LEN 3
|
---|
124 | #define ONIGENC_GET_CASE_FOLD_CODES_MAX_NUM 13
|
---|
125 | /* 13 => Unicode:0x1ffc */
|
---|
126 |
|
---|
127 | /* code range */
|
---|
128 | #define ONIGENC_CODE_RANGE_NUM(range) ((int )range[0])
|
---|
129 | #define ONIGENC_CODE_RANGE_FROM(range,i) range[((i)*2) + 1]
|
---|
130 | #define ONIGENC_CODE_RANGE_TO(range,i) range[((i)*2) + 2]
|
---|
131 |
|
---|
132 | typedef struct {
|
---|
133 | int byte_len; /* argument(original) character(s) byte length */
|
---|
134 | int code_len; /* number of code */
|
---|
135 | OnigCodePoint code[ONIGENC_MAX_COMP_CASE_FOLD_CODE_LEN];
|
---|
136 | } OnigCaseFoldCodeItem;
|
---|
137 |
|
---|
138 | typedef struct {
|
---|
139 | OnigCodePoint esc;
|
---|
140 | OnigCodePoint anychar;
|
---|
141 | OnigCodePoint anytime;
|
---|
142 | OnigCodePoint zero_or_one_time;
|
---|
143 | OnigCodePoint one_or_more_time;
|
---|
144 | OnigCodePoint anychar_anytime;
|
---|
145 | } OnigMetaCharTableType;
|
---|
146 |
|
---|
147 | typedef int (*OnigApplyAllCaseFoldFunc)(OnigCodePoint from, OnigCodePoint* to, int to_len, void* arg);
|
---|
148 |
|
---|
149 | typedef struct OnigEncodingTypeST {
|
---|
150 | int (*mbc_enc_len)(const OnigUChar* p);
|
---|
151 | const char* name;
|
---|
152 | int max_enc_len;
|
---|
153 | int min_enc_len;
|
---|
154 | int (*is_mbc_newline)(const OnigUChar* p, const OnigUChar* end);
|
---|
155 | OnigCodePoint (*mbc_to_code)(const OnigUChar* p, const OnigUChar* end);
|
---|
156 | int (*code_to_mbclen)(OnigCodePoint code);
|
---|
157 | int (*code_to_mbc)(OnigCodePoint code, OnigUChar *buf);
|
---|
158 | int (*mbc_case_fold)(OnigCaseFoldType flag, const OnigUChar** pp, const OnigUChar* end, OnigUChar* to);
|
---|
159 | int (*apply_all_case_fold)(OnigCaseFoldType flag, OnigApplyAllCaseFoldFunc f, void* arg);
|
---|
160 | int (*get_case_fold_codes_by_str)(OnigCaseFoldType flag, const OnigUChar* p, const OnigUChar* end, OnigCaseFoldCodeItem acs[]);
|
---|
161 | int (*property_name_to_ctype)(struct OnigEncodingTypeST* enc, OnigUChar* p, OnigUChar* end);
|
---|
162 | int (*is_code_ctype)(OnigCodePoint code, OnigCtype ctype);
|
---|
163 | int (*get_ctype_code_range)(OnigCtype ctype, OnigCodePoint* sb_out, const OnigCodePoint* ranges[]);
|
---|
164 | OnigUChar* (*left_adjust_char_head)(const OnigUChar* start, const OnigUChar* p);
|
---|
165 | int (*is_allowed_reverse_match)(const OnigUChar* p, const OnigUChar* end);
|
---|
166 | unsigned int flags;
|
---|
167 | } OnigEncodingType;
|
---|
168 |
|
---|
169 | typedef OnigEncodingType* OnigEncoding;
|
---|
170 |
|
---|
171 | ONIG_EXTERN OnigEncodingType OnigEncodingASCII;
|
---|
172 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_1;
|
---|
173 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_2;
|
---|
174 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_3;
|
---|
175 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_4;
|
---|
176 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_5;
|
---|
177 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_6;
|
---|
178 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_7;
|
---|
179 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_8;
|
---|
180 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_9;
|
---|
181 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_10;
|
---|
182 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_11;
|
---|
183 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_13;
|
---|
184 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_14;
|
---|
185 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_15;
|
---|
186 | ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_16;
|
---|
187 | ONIG_EXTERN OnigEncodingType OnigEncodingUTF8;
|
---|
188 | ONIG_EXTERN OnigEncodingType OnigEncodingUTF16_BE;
|
---|
189 | ONIG_EXTERN OnigEncodingType OnigEncodingUTF16_LE;
|
---|
190 | ONIG_EXTERN OnigEncodingType OnigEncodingUTF32_BE;
|
---|
191 | ONIG_EXTERN OnigEncodingType OnigEncodingUTF32_LE;
|
---|
192 | ONIG_EXTERN OnigEncodingType OnigEncodingEUC_JP;
|
---|
193 | ONIG_EXTERN OnigEncodingType OnigEncodingEUC_TW;
|
---|
194 | ONIG_EXTERN OnigEncodingType OnigEncodingEUC_KR;
|
---|
195 | ONIG_EXTERN OnigEncodingType OnigEncodingEUC_CN;
|
---|
196 | ONIG_EXTERN OnigEncodingType OnigEncodingSJIS;
|
---|
197 | ONIG_EXTERN OnigEncodingType OnigEncodingCP932;
|
---|
198 | ONIG_EXTERN OnigEncodingType OnigEncodingKOI8;
|
---|
199 | ONIG_EXTERN OnigEncodingType OnigEncodingKOI8_R;
|
---|
200 | ONIG_EXTERN OnigEncodingType OnigEncodingCP1251;
|
---|
201 | ONIG_EXTERN OnigEncodingType OnigEncodingBIG5;
|
---|
202 | ONIG_EXTERN OnigEncodingType OnigEncodingGB18030;
|
---|
203 |
|
---|
204 | #define ONIG_ENCODING_ASCII (&OnigEncodingASCII)
|
---|
205 | #define ONIG_ENCODING_ISO_8859_1 (&OnigEncodingISO_8859_1)
|
---|
206 | #define ONIG_ENCODING_ISO_8859_2 (&OnigEncodingISO_8859_2)
|
---|
207 | #define ONIG_ENCODING_ISO_8859_3 (&OnigEncodingISO_8859_3)
|
---|
208 | #define ONIG_ENCODING_ISO_8859_4 (&OnigEncodingISO_8859_4)
|
---|
209 | #define ONIG_ENCODING_ISO_8859_5 (&OnigEncodingISO_8859_5)
|
---|
210 | #define ONIG_ENCODING_ISO_8859_6 (&OnigEncodingISO_8859_6)
|
---|
211 | #define ONIG_ENCODING_ISO_8859_7 (&OnigEncodingISO_8859_7)
|
---|
212 | #define ONIG_ENCODING_ISO_8859_8 (&OnigEncodingISO_8859_8)
|
---|
213 | #define ONIG_ENCODING_ISO_8859_9 (&OnigEncodingISO_8859_9)
|
---|
214 | #define ONIG_ENCODING_ISO_8859_10 (&OnigEncodingISO_8859_10)
|
---|
215 | #define ONIG_ENCODING_ISO_8859_11 (&OnigEncodingISO_8859_11)
|
---|
216 | #define ONIG_ENCODING_ISO_8859_13 (&OnigEncodingISO_8859_13)
|
---|
217 | #define ONIG_ENCODING_ISO_8859_14 (&OnigEncodingISO_8859_14)
|
---|
218 | #define ONIG_ENCODING_ISO_8859_15 (&OnigEncodingISO_8859_15)
|
---|
219 | #define ONIG_ENCODING_ISO_8859_16 (&OnigEncodingISO_8859_16)
|
---|
220 | #define ONIG_ENCODING_UTF8 (&OnigEncodingUTF8)
|
---|
221 | #define ONIG_ENCODING_UTF16_BE (&OnigEncodingUTF16_BE)
|
---|
222 | #define ONIG_ENCODING_UTF16_LE (&OnigEncodingUTF16_LE)
|
---|
223 | #define ONIG_ENCODING_UTF32_BE (&OnigEncodingUTF32_BE)
|
---|
224 | #define ONIG_ENCODING_UTF32_LE (&OnigEncodingUTF32_LE)
|
---|
225 | #define ONIG_ENCODING_EUC_JP (&OnigEncodingEUC_JP)
|
---|
226 | #define ONIG_ENCODING_EUC_TW (&OnigEncodingEUC_TW)
|
---|
227 | #define ONIG_ENCODING_EUC_KR (&OnigEncodingEUC_KR)
|
---|
228 | #define ONIG_ENCODING_EUC_CN (&OnigEncodingEUC_CN)
|
---|
229 | #define ONIG_ENCODING_SJIS (&OnigEncodingSJIS)
|
---|
230 | #define ONIG_ENCODING_CP932 (&OnigEncodingCP932)
|
---|
231 | #define ONIG_ENCODING_KOI8 (&OnigEncodingKOI8)
|
---|
232 | #define ONIG_ENCODING_KOI8_R (&OnigEncodingKOI8_R)
|
---|
233 | #define ONIG_ENCODING_CP1251 (&OnigEncodingCP1251)
|
---|
234 | #define ONIG_ENCODING_BIG5 (&OnigEncodingBIG5)
|
---|
235 | #define ONIG_ENCODING_GB18030 (&OnigEncodingGB18030)
|
---|
236 |
|
---|
237 | #define ONIG_ENCODING_UNDEF ((OnigEncoding )0)
|
---|
238 |
|
---|
239 |
|
---|
240 | /* work size */
|
---|
241 | #define ONIGENC_CODE_TO_MBC_MAXLEN 7
|
---|
242 | #define ONIGENC_MBC_CASE_FOLD_MAXLEN 18
|
---|
243 | /* 18: 6(max-byte) * 3(case-fold chars) */
|
---|
244 |
|
---|
245 | /* character types */
|
---|
246 | #define ONIGENC_CTYPE_NEWLINE 0
|
---|
247 | #define ONIGENC_CTYPE_ALPHA 1
|
---|
248 | #define ONIGENC_CTYPE_BLANK 2
|
---|
249 | #define ONIGENC_CTYPE_CNTRL 3
|
---|
250 | #define ONIGENC_CTYPE_DIGIT 4
|
---|
251 | #define ONIGENC_CTYPE_GRAPH 5
|
---|
252 | #define ONIGENC_CTYPE_LOWER 6
|
---|
253 | #define ONIGENC_CTYPE_PRINT 7
|
---|
254 | #define ONIGENC_CTYPE_PUNCT 8
|
---|
255 | #define ONIGENC_CTYPE_SPACE 9
|
---|
256 | #define ONIGENC_CTYPE_UPPER 10
|
---|
257 | #define ONIGENC_CTYPE_XDIGIT 11
|
---|
258 | #define ONIGENC_CTYPE_WORD 12
|
---|
259 | #define ONIGENC_CTYPE_ALNUM 13 /* alpha || digit */
|
---|
260 | #define ONIGENC_CTYPE_ASCII 14
|
---|
261 | #define ONIGENC_MAX_STD_CTYPE ONIGENC_CTYPE_ASCII
|
---|
262 |
|
---|
263 | /* flags */
|
---|
264 | #define ONIGENC_FLAG_NONE 0U
|
---|
265 | #define ONIGENC_FLAG_UNICODE 1U
|
---|
266 |
|
---|
267 |
|
---|
268 | #define onig_enc_len(enc,p,end) ONIGENC_MBC_ENC_LEN(enc,p)
|
---|
269 |
|
---|
270 | #define ONIGENC_IS_UNDEF(enc) ((enc) == ONIG_ENCODING_UNDEF)
|
---|
271 | #define ONIGENC_IS_SINGLEBYTE(enc) (ONIGENC_MBC_MAXLEN(enc) == 1)
|
---|
272 | #define ONIGENC_IS_MBC_HEAD(enc,p) (ONIGENC_MBC_ENC_LEN(enc,p) != 1)
|
---|
273 | #define ONIGENC_IS_MBC_ASCII(p) (*(p) < 128)
|
---|
274 | #define ONIGENC_IS_CODE_ASCII(code) ((code) < 128)
|
---|
275 | #define ONIGENC_IS_MBC_WORD(enc,s,end) \
|
---|
276 | ONIGENC_IS_CODE_WORD(enc,ONIGENC_MBC_TO_CODE(enc,s,end))
|
---|
277 | #define ONIGENC_IS_MBC_ASCII_WORD(enc,s,end) \
|
---|
278 | onigenc_ascii_is_code_ctype( \
|
---|
279 | ONIGENC_MBC_TO_CODE(enc,s,end),ONIGENC_CTYPE_WORD)
|
---|
280 | #define ONIGENC_IS_UNICODE(enc) ((enc)->flags & ONIGENC_FLAG_UNICODE)
|
---|
281 |
|
---|
282 |
|
---|
283 | #define ONIGENC_NAME(enc) ((enc)->name)
|
---|
284 |
|
---|
285 | #define ONIGENC_MBC_CASE_FOLD(enc,flag,pp,end,buf) \
|
---|
286 | (enc)->mbc_case_fold(flag,(const OnigUChar** )pp,end,buf)
|
---|
287 | #define ONIGENC_IS_ALLOWED_REVERSE_MATCH(enc,s,end) \
|
---|
288 | (enc)->is_allowed_reverse_match(s,end)
|
---|
289 | #define ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc,start,s) \
|
---|
290 | (enc)->left_adjust_char_head(start, s)
|
---|
291 | #define ONIGENC_APPLY_ALL_CASE_FOLD(enc,case_fold_flag,f,arg) \
|
---|
292 | (enc)->apply_all_case_fold(case_fold_flag,f,arg)
|
---|
293 | #define ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc,case_fold_flag,p,end,acs) \
|
---|
294 | (enc)->get_case_fold_codes_by_str(case_fold_flag,p,end,acs)
|
---|
295 | #define ONIGENC_STEP_BACK(enc,start,s,n) \
|
---|
296 | onigenc_step_back((enc),(start),(s),(n))
|
---|
297 |
|
---|
298 | #define ONIGENC_MBC_ENC_LEN(enc,p) (enc)->mbc_enc_len(p)
|
---|
299 | #define ONIGENC_MBC_MAXLEN(enc) ((enc)->max_enc_len)
|
---|
300 | #define ONIGENC_MBC_MAXLEN_DIST(enc) ONIGENC_MBC_MAXLEN(enc)
|
---|
301 | #define ONIGENC_MBC_MINLEN(enc) ((enc)->min_enc_len)
|
---|
302 | #define ONIGENC_IS_MBC_NEWLINE(enc,p,end) (enc)->is_mbc_newline((p),(end))
|
---|
303 | #define ONIGENC_MBC_TO_CODE(enc,p,end) (enc)->mbc_to_code((p),(end))
|
---|
304 | #define ONIGENC_CODE_TO_MBCLEN(enc,code) (enc)->code_to_mbclen(code)
|
---|
305 | #define ONIGENC_CODE_TO_MBC(enc,code,buf) (enc)->code_to_mbc(code,buf)
|
---|
306 | #define ONIGENC_PROPERTY_NAME_TO_CTYPE(enc,p,end) \
|
---|
307 | (enc)->property_name_to_ctype(enc,p,end)
|
---|
308 |
|
---|
309 | #define ONIGENC_IS_CODE_CTYPE(enc,code,ctype) (enc)->is_code_ctype(code,ctype)
|
---|
310 |
|
---|
311 | #define ONIGENC_IS_CODE_NEWLINE(enc,code) \
|
---|
312 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_NEWLINE)
|
---|
313 | #define ONIGENC_IS_CODE_GRAPH(enc,code) \
|
---|
314 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_GRAPH)
|
---|
315 | #define ONIGENC_IS_CODE_PRINT(enc,code) \
|
---|
316 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PRINT)
|
---|
317 | #define ONIGENC_IS_CODE_ALNUM(enc,code) \
|
---|
318 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALNUM)
|
---|
319 | #define ONIGENC_IS_CODE_ALPHA(enc,code) \
|
---|
320 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALPHA)
|
---|
321 | #define ONIGENC_IS_CODE_LOWER(enc,code) \
|
---|
322 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_LOWER)
|
---|
323 | #define ONIGENC_IS_CODE_UPPER(enc,code) \
|
---|
324 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_UPPER)
|
---|
325 | #define ONIGENC_IS_CODE_CNTRL(enc,code) \
|
---|
326 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_CNTRL)
|
---|
327 | #define ONIGENC_IS_CODE_PUNCT(enc,code) \
|
---|
328 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PUNCT)
|
---|
329 | #define ONIGENC_IS_CODE_SPACE(enc,code) \
|
---|
330 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_SPACE)
|
---|
331 | #define ONIGENC_IS_CODE_BLANK(enc,code) \
|
---|
332 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_BLANK)
|
---|
333 | #define ONIGENC_IS_CODE_DIGIT(enc,code) \
|
---|
334 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_DIGIT)
|
---|
335 | #define ONIGENC_IS_CODE_XDIGIT(enc,code) \
|
---|
336 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_XDIGIT)
|
---|
337 | #define ONIGENC_IS_CODE_WORD(enc,code) \
|
---|
338 | ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_WORD)
|
---|
339 |
|
---|
340 | #define ONIGENC_GET_CTYPE_CODE_RANGE(enc,ctype,sbout,ranges) \
|
---|
341 | (enc)->get_ctype_code_range(ctype,sbout,ranges)
|
---|
342 |
|
---|
343 | ONIG_EXTERN
|
---|
344 | OnigUChar* onigenc_step_back P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s, int n));
|
---|
345 |
|
---|
346 |
|
---|
347 | /* encoding API */
|
---|
348 | ONIG_EXTERN
|
---|
349 | int onigenc_init P_((void));
|
---|
350 | ONIG_EXTERN
|
---|
351 | int onigenc_set_default_encoding P_((OnigEncoding enc));
|
---|
352 | ONIG_EXTERN
|
---|
353 | OnigEncoding onigenc_get_default_encoding P_((void));
|
---|
354 | ONIG_EXTERN
|
---|
355 | void onigenc_set_default_caseconv_table P_((const OnigUChar* table));
|
---|
356 | ONIG_EXTERN
|
---|
357 | OnigUChar* onigenc_get_right_adjust_char_head_with_prev P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s, const OnigUChar** prev));
|
---|
358 | ONIG_EXTERN
|
---|
359 | OnigUChar* onigenc_get_prev_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
|
---|
360 | ONIG_EXTERN
|
---|
361 | OnigUChar* onigenc_get_left_adjust_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
|
---|
362 | ONIG_EXTERN
|
---|
363 | OnigUChar* onigenc_get_right_adjust_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
|
---|
364 | ONIG_EXTERN
|
---|
365 | int onigenc_strlen P_((OnigEncoding enc, const OnigUChar* p, const OnigUChar* end));
|
---|
366 | ONIG_EXTERN
|
---|
367 | int onigenc_strlen_null P_((OnigEncoding enc, const OnigUChar* p));
|
---|
368 | ONIG_EXTERN
|
---|
369 | int onigenc_str_bytelen_null P_((OnigEncoding enc, const OnigUChar* p));
|
---|
370 |
|
---|
371 |
|
---|
372 |
|
---|
373 | /* PART: regular expression */
|
---|
374 |
|
---|
375 | /* config parameters */
|
---|
376 | #define ONIG_NREGION 10
|
---|
377 | #define ONIG_MAX_BACKREF_NUM 1000
|
---|
378 | #define ONIG_MAX_CAPTURE_GROUP_NUM 32767
|
---|
379 | #define ONIG_MAX_REPEAT_NUM 100000
|
---|
380 | #define ONIG_MAX_MULTI_BYTE_RANGES_NUM 10000
|
---|
381 | /* constants */
|
---|
382 | #define ONIG_MAX_ERROR_MESSAGE_LEN 90
|
---|
383 |
|
---|
384 | typedef unsigned int OnigOptionType;
|
---|
385 |
|
---|
386 | #define ONIG_OPTION_DEFAULT ONIG_OPTION_NONE
|
---|
387 |
|
---|
388 | /* options */
|
---|
389 | #define ONIG_OPTION_NONE 0U
|
---|
390 | #define ONIG_OPTION_IGNORECASE 1U
|
---|
391 | #define ONIG_OPTION_EXTEND (ONIG_OPTION_IGNORECASE << 1)
|
---|
392 | #define ONIG_OPTION_MULTILINE (ONIG_OPTION_EXTEND << 1)
|
---|
393 | #define ONIG_OPTION_DOTALL ONIG_OPTION_MULTILINE
|
---|
394 | #define ONIG_OPTION_SINGLELINE (ONIG_OPTION_MULTILINE << 1)
|
---|
395 | #define ONIG_OPTION_FIND_LONGEST (ONIG_OPTION_SINGLELINE << 1)
|
---|
396 | #define ONIG_OPTION_FIND_NOT_EMPTY (ONIG_OPTION_FIND_LONGEST << 1)
|
---|
397 | #define ONIG_OPTION_NEGATE_SINGLELINE (ONIG_OPTION_FIND_NOT_EMPTY << 1)
|
---|
398 | #define ONIG_OPTION_DONT_CAPTURE_GROUP (ONIG_OPTION_NEGATE_SINGLELINE << 1)
|
---|
399 | #define ONIG_OPTION_CAPTURE_GROUP (ONIG_OPTION_DONT_CAPTURE_GROUP << 1)
|
---|
400 | /* options (search time) */
|
---|
401 | #define ONIG_OPTION_NOTBOL (ONIG_OPTION_CAPTURE_GROUP << 1)
|
---|
402 | #define ONIG_OPTION_NOTEOL (ONIG_OPTION_NOTBOL << 1)
|
---|
403 | #define ONIG_OPTION_POSIX_REGION (ONIG_OPTION_NOTEOL << 1)
|
---|
404 | /* options (ctype range) */
|
---|
405 | #define ONIG_OPTION_ASCII_RANGE (ONIG_OPTION_POSIX_REGION << 1)
|
---|
406 | #define ONIG_OPTION_POSIX_BRACKET_ALL_RANGE (ONIG_OPTION_ASCII_RANGE << 1)
|
---|
407 | #define ONIG_OPTION_WORD_BOUND_ALL_RANGE (ONIG_OPTION_POSIX_BRACKET_ALL_RANGE << 1)
|
---|
408 | /* options (newline) */
|
---|
409 | #define ONIG_OPTION_NEWLINE_CRLF (ONIG_OPTION_WORD_BOUND_ALL_RANGE << 1)
|
---|
410 | #define ONIG_OPTION_NOTBOS (ONIG_OPTION_NEWLINE_CRLF << 1)
|
---|
411 | #define ONIG_OPTION_NOTEOS (ONIG_OPTION_NOTBOS << 1)
|
---|
412 | #define ONIG_OPTION_MAXBIT ONIG_OPTION_NOTEOS /* limit */
|
---|
413 |
|
---|
414 | #define ONIG_OPTION_ON(options,regopt) ((options) |= (regopt))
|
---|
415 | #define ONIG_OPTION_OFF(options,regopt) ((options) &= ~(regopt))
|
---|
416 | #define ONIG_IS_OPTION_ON(options,option) ((options) & (option))
|
---|
417 |
|
---|
418 | /* syntax */
|
---|
419 | typedef struct {
|
---|
420 | unsigned int op;
|
---|
421 | unsigned int op2;
|
---|
422 | unsigned int behavior;
|
---|
423 | OnigOptionType options; /* default option */
|
---|
424 | OnigMetaCharTableType meta_char_table;
|
---|
425 | } OnigSyntaxType;
|
---|
426 |
|
---|
427 | ONIG_EXTERN OnigSyntaxType OnigSyntaxASIS;
|
---|
428 | ONIG_EXTERN OnigSyntaxType OnigSyntaxPosixBasic;
|
---|
429 | ONIG_EXTERN OnigSyntaxType OnigSyntaxPosixExtended;
|
---|
430 | ONIG_EXTERN OnigSyntaxType OnigSyntaxEmacs;
|
---|
431 | ONIG_EXTERN OnigSyntaxType OnigSyntaxGrep;
|
---|
432 | ONIG_EXTERN OnigSyntaxType OnigSyntaxGnuRegex;
|
---|
433 | ONIG_EXTERN OnigSyntaxType OnigSyntaxJava;
|
---|
434 | ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl58;
|
---|
435 | ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl58_NG;
|
---|
436 | ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl;
|
---|
437 | ONIG_EXTERN OnigSyntaxType OnigSyntaxRuby;
|
---|
438 | ONIG_EXTERN OnigSyntaxType OnigSyntaxPython;
|
---|
439 |
|
---|
440 | /* predefined syntaxes (see regsyntax.c) */
|
---|
441 | #define ONIG_SYNTAX_ASIS (&OnigSyntaxASIS)
|
---|
442 | #define ONIG_SYNTAX_POSIX_BASIC (&OnigSyntaxPosixBasic)
|
---|
443 | #define ONIG_SYNTAX_POSIX_EXTENDED (&OnigSyntaxPosixExtended)
|
---|
444 | #define ONIG_SYNTAX_EMACS (&OnigSyntaxEmacs)
|
---|
445 | #define ONIG_SYNTAX_GREP (&OnigSyntaxGrep)
|
---|
446 | #define ONIG_SYNTAX_GNU_REGEX (&OnigSyntaxGnuRegex)
|
---|
447 | #define ONIG_SYNTAX_JAVA (&OnigSyntaxJava)
|
---|
448 | #define ONIG_SYNTAX_PERL58 (&OnigSyntaxPerl58)
|
---|
449 | #define ONIG_SYNTAX_PERL58_NG (&OnigSyntaxPerl58_NG)
|
---|
450 | #define ONIG_SYNTAX_PERL (&OnigSyntaxPerl)
|
---|
451 | #define ONIG_SYNTAX_RUBY (&OnigSyntaxRuby)
|
---|
452 | #define ONIG_SYNTAX_PYTHON (&OnigSyntaxPython)
|
---|
453 |
|
---|
454 | /* default syntax */
|
---|
455 | ONIG_EXTERN OnigSyntaxType* OnigDefaultSyntax;
|
---|
456 | #define ONIG_SYNTAX_DEFAULT OnigDefaultSyntax
|
---|
457 |
|
---|
458 | /* syntax (operators) */
|
---|
459 | #define ONIG_SYN_OP_VARIABLE_META_CHARACTERS (1U<<0)
|
---|
460 | #define ONIG_SYN_OP_DOT_ANYCHAR (1U<<1) /* . */
|
---|
461 | #define ONIG_SYN_OP_ASTERISK_ZERO_INF (1U<<2) /* * */
|
---|
462 | #define ONIG_SYN_OP_ESC_ASTERISK_ZERO_INF (1U<<3)
|
---|
463 | #define ONIG_SYN_OP_PLUS_ONE_INF (1U<<4) /* + */
|
---|
464 | #define ONIG_SYN_OP_ESC_PLUS_ONE_INF (1U<<5)
|
---|
465 | #define ONIG_SYN_OP_QMARK_ZERO_ONE (1U<<6) /* ? */
|
---|
466 | #define ONIG_SYN_OP_ESC_QMARK_ZERO_ONE (1U<<7)
|
---|
467 | #define ONIG_SYN_OP_BRACE_INTERVAL (1U<<8) /* {lower,upper} */
|
---|
468 | #define ONIG_SYN_OP_ESC_BRACE_INTERVAL (1U<<9) /* \{lower,upper\} */
|
---|
469 | #define ONIG_SYN_OP_VBAR_ALT (1U<<10) /* | */
|
---|
470 | #define ONIG_SYN_OP_ESC_VBAR_ALT (1U<<11) /* \| */
|
---|
471 | #define ONIG_SYN_OP_LPAREN_SUBEXP (1U<<12) /* (...) */
|
---|
472 | #define ONIG_SYN_OP_ESC_LPAREN_SUBEXP (1U<<13) /* \(...\) */
|
---|
473 | #define ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR (1U<<14) /* \A, \Z, \z */
|
---|
474 | #define ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR (1U<<15) /* \G */
|
---|
475 | #define ONIG_SYN_OP_DECIMAL_BACKREF (1U<<16) /* \num */
|
---|
476 | #define ONIG_SYN_OP_BRACKET_CC (1U<<17) /* [...] */
|
---|
477 | #define ONIG_SYN_OP_ESC_W_WORD (1U<<18) /* \w, \W */
|
---|
478 | #define ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END (1U<<19) /* \<. \> */
|
---|
479 | #define ONIG_SYN_OP_ESC_B_WORD_BOUND (1U<<20) /* \b, \B */
|
---|
480 | #define ONIG_SYN_OP_ESC_S_WHITE_SPACE (1U<<21) /* \s, \S */
|
---|
481 | #define ONIG_SYN_OP_ESC_D_DIGIT (1U<<22) /* \d, \D */
|
---|
482 | #define ONIG_SYN_OP_LINE_ANCHOR (1U<<23) /* ^, $ */
|
---|
483 | #define ONIG_SYN_OP_POSIX_BRACKET (1U<<24) /* [:xxxx:] */
|
---|
484 | #define ONIG_SYN_OP_QMARK_NON_GREEDY (1U<<25) /* ??,*?,+?,{n,m}? */
|
---|
485 | #define ONIG_SYN_OP_ESC_CONTROL_CHARS (1U<<26) /* \n,\r,\t,\a ... */
|
---|
486 | #define ONIG_SYN_OP_ESC_C_CONTROL (1U<<27) /* \cx */
|
---|
487 | #define ONIG_SYN_OP_ESC_OCTAL3 (1U<<28) /* \OOO */
|
---|
488 | #define ONIG_SYN_OP_ESC_X_HEX2 (1U<<29) /* \xHH */
|
---|
489 | #define ONIG_SYN_OP_ESC_X_BRACE_HEX8 (1U<<30) /* \x{7HHHHHHH} */
|
---|
490 | #define ONIG_SYN_OP_ESC_O_BRACE_OCTAL (1U<<31) /* \o{OOO} */ /* NOTIMPL */
|
---|
491 |
|
---|
492 | #define ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE (1U<<0) /* \Q...\E */
|
---|
493 | #define ONIG_SYN_OP2_QMARK_GROUP_EFFECT (1U<<1) /* (?...) */
|
---|
494 | #define ONIG_SYN_OP2_OPTION_PERL (1U<<2) /* (?imsxadlu), (?-imsx), (?^imsxalu) */
|
---|
495 | #define ONIG_SYN_OP2_OPTION_RUBY (1U<<3) /* (?imxadu), (?-imx) */
|
---|
496 | #define ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT (1U<<4) /* ?+,*+,++ */
|
---|
497 | #define ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL (1U<<5) /* {n,m}+ */
|
---|
498 | #define ONIG_SYN_OP2_CCLASS_SET_OP (1U<<6) /* [...&&..[..]..] */
|
---|
499 | #define ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP (1U<<7) /* (?<name>...) */
|
---|
500 | #define ONIG_SYN_OP2_ESC_K_NAMED_BACKREF (1U<<8) /* \k<name> */
|
---|
501 | #define ONIG_SYN_OP2_ESC_G_SUBEXP_CALL (1U<<9) /* \g<name>, \g<n> */
|
---|
502 | #define ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY (1U<<10) /* (?@..),(?@<x>..) */
|
---|
503 | #define ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL (1U<<11) /* \C-x */
|
---|
504 | #define ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META (1U<<12) /* \M-x */
|
---|
505 | #define ONIG_SYN_OP2_ESC_V_VTAB (1U<<13) /* \v as VTAB */
|
---|
506 | #define ONIG_SYN_OP2_ESC_U_HEX4 (1U<<14) /* \uHHHH */
|
---|
507 | #define ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR (1U<<15) /* \`, \' */
|
---|
508 | #define ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY (1U<<16) /* \p{...}, \P{...} */
|
---|
509 | #define ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT (1U<<17) /* \p{^..}, \P{^..} */
|
---|
510 | /* #define ONIG_SYN_OP2_CHAR_PROPERTY_PREFIX_IS (1U<<18) */
|
---|
511 | #define ONIG_SYN_OP2_ESC_H_XDIGIT (1U<<19) /* \h, \H */
|
---|
512 | #define ONIG_SYN_OP2_INEFFECTIVE_ESCAPE (1U<<20) /* \ */
|
---|
513 | #define ONIG_SYN_OP2_ESC_CAPITAL_R_LINEBREAK (1U<<21) /* \R as (?>\x0D\x0A|[\x0A-\x0D\x{85}\x{2028}\x{2029}]) */
|
---|
514 | #define ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER (1U<<22) /* \X as (?>\P{M}\p{M}*) */
|
---|
515 | #define ONIG_SYN_OP2_ESC_V_VERTICAL_WHITESPACE (1U<<23) /* \v, \V -- Perl */ /* NOTIMPL */
|
---|
516 | #define ONIG_SYN_OP2_ESC_H_HORIZONTAL_WHITESPACE (1U<<24) /* \h, \H -- Perl */ /* NOTIMPL */
|
---|
517 | #define ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP (1U<<25) /* \K */
|
---|
518 | #define ONIG_SYN_OP2_ESC_G_BRACE_BACKREF (1U<<26) /* \g{name}, \g{n} */
|
---|
519 | #define ONIG_SYN_OP2_QMARK_SUBEXP_CALL (1U<<27) /* (?&name), (?n), (?R), (?0) */
|
---|
520 | #define ONIG_SYN_OP2_QMARK_VBAR_BRANCH_RESET (1U<<28) /* (?|...) */ /* NOTIMPL */
|
---|
521 | #define ONIG_SYN_OP2_QMARK_LPAREN_CONDITION (1U<<29) /* (?(cond)yes...|no...) */
|
---|
522 | #define ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP (1U<<30) /* (?P<name>...), (?P=name), (?P>name) -- Python/PCRE */
|
---|
523 | #define ONIG_SYN_OP2_OPTION_JAVA (1U<<31) /* (?idmsux), (?-idmsux) */ /* NOTIMPL */
|
---|
524 |
|
---|
525 | /* syntax (behavior) */
|
---|
526 | #define ONIG_SYN_CONTEXT_INDEP_ANCHORS (1U<<31) /* not implemented */
|
---|
527 | #define ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS (1U<<0) /* ?, *, +, {n,m} */
|
---|
528 | #define ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS (1U<<1) /* error or ignore */
|
---|
529 | #define ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP (1U<<2) /* ...)... */
|
---|
530 | #define ONIG_SYN_ALLOW_INVALID_INTERVAL (1U<<3) /* {??? */
|
---|
531 | #define ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV (1U<<4) /* {,n} => {0,n} */
|
---|
532 | #define ONIG_SYN_STRICT_CHECK_BACKREF (1U<<5) /* /(\1)/,/\1()/ ..*/
|
---|
533 | #define ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND (1U<<6) /* (?<=a|bc) */
|
---|
534 | #define ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP (1U<<7) /* see doc/RE */
|
---|
535 | #define ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME (1U<<8) /* (?<x>)(?<x>) */
|
---|
536 | #define ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY (1U<<9) /* a{n}?=(?:a{n})? */
|
---|
537 | #define ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME_CALL (1U<<10) /* (?<x>)(?<x>)(?&x) */
|
---|
538 |
|
---|
539 | /* syntax (behavior) in char class [...] */
|
---|
540 | #define ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC (1U<<20) /* [^...] */
|
---|
541 | #define ONIG_SYN_BACKSLASH_ESCAPE_IN_CC (1U<<21) /* [..\w..] etc.. */
|
---|
542 | #define ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC (1U<<22)
|
---|
543 | #define ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC (1U<<23) /* [0-9-a]=[0-9\-a] */
|
---|
544 | /* syntax (behavior) warning */
|
---|
545 | #define ONIG_SYN_WARN_CC_OP_NOT_ESCAPED (1U<<24) /* [,-,] */
|
---|
546 | #define ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT (1U<<25) /* (?:a*)+ */
|
---|
547 |
|
---|
548 | /* meta character specifiers (onig_set_meta_char()) */
|
---|
549 | #define ONIG_META_CHAR_ESCAPE 0
|
---|
550 | #define ONIG_META_CHAR_ANYCHAR 1
|
---|
551 | #define ONIG_META_CHAR_ANYTIME 2
|
---|
552 | #define ONIG_META_CHAR_ZERO_OR_ONE_TIME 3
|
---|
553 | #define ONIG_META_CHAR_ONE_OR_MORE_TIME 4
|
---|
554 | #define ONIG_META_CHAR_ANYCHAR_ANYTIME 5
|
---|
555 |
|
---|
556 | #define ONIG_INEFFECTIVE_META_CHAR 0
|
---|
557 |
|
---|
558 | /* error codes */
|
---|
559 | #define ONIG_IS_PATTERN_ERROR(ecode) ((ecode) <= -100 && (ecode) > -1000)
|
---|
560 | /* normal return */
|
---|
561 | #define ONIG_NORMAL 0
|
---|
562 | #define ONIG_MISMATCH -1
|
---|
563 | #define ONIG_NO_SUPPORT_CONFIG -2
|
---|
564 |
|
---|
565 | /* internal error */
|
---|
566 | #define ONIGERR_MEMORY -5
|
---|
567 | #define ONIGERR_TYPE_BUG -6
|
---|
568 | #define ONIGERR_PARSER_BUG -11
|
---|
569 | #define ONIGERR_STACK_BUG -12
|
---|
570 | #define ONIGERR_UNDEFINED_BYTECODE -13
|
---|
571 | #define ONIGERR_UNEXPECTED_BYTECODE -14
|
---|
572 | #define ONIGERR_MATCH_STACK_LIMIT_OVER -15
|
---|
573 | #define ONIGERR_DEFAULT_ENCODING_IS_NOT_SET -21
|
---|
574 | #define ONIGERR_SPECIFIED_ENCODING_CANT_CONVERT_TO_WIDE_CHAR -22
|
---|
575 | /* general error */
|
---|
576 | #define ONIGERR_INVALID_ARGUMENT -30
|
---|
577 | /* syntax error */
|
---|
578 | #define ONIGERR_END_PATTERN_AT_LEFT_BRACE -100
|
---|
579 | #define ONIGERR_END_PATTERN_AT_LEFT_BRACKET -101
|
---|
580 | #define ONIGERR_EMPTY_CHAR_CLASS -102
|
---|
581 | #define ONIGERR_PREMATURE_END_OF_CHAR_CLASS -103
|
---|
582 | #define ONIGERR_END_PATTERN_AT_ESCAPE -104
|
---|
583 | #define ONIGERR_END_PATTERN_AT_META -105
|
---|
584 | #define ONIGERR_END_PATTERN_AT_CONTROL -106
|
---|
585 | #define ONIGERR_META_CODE_SYNTAX -108
|
---|
586 | #define ONIGERR_CONTROL_CODE_SYNTAX -109
|
---|
587 | #define ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE -110
|
---|
588 | #define ONIGERR_CHAR_CLASS_VALUE_AT_START_OF_RANGE -111
|
---|
589 | #define ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS -112
|
---|
590 | #define ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED -113
|
---|
591 | #define ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID -114
|
---|
592 | #define ONIGERR_NESTED_REPEAT_OPERATOR -115
|
---|
593 | #define ONIGERR_UNMATCHED_CLOSE_PARENTHESIS -116
|
---|
594 | #define ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS -117
|
---|
595 | #define ONIGERR_END_PATTERN_IN_GROUP -118
|
---|
596 | #define ONIGERR_UNDEFINED_GROUP_OPTION -119
|
---|
597 | #define ONIGERR_INVALID_POSIX_BRACKET_TYPE -121
|
---|
598 | #define ONIGERR_INVALID_LOOK_BEHIND_PATTERN -122
|
---|
599 | #define ONIGERR_INVALID_REPEAT_RANGE_PATTERN -123
|
---|
600 | #define ONIGERR_INVALID_CONDITION_PATTERN -124
|
---|
601 | /* values error (syntax error) */
|
---|
602 | #define ONIGERR_TOO_BIG_NUMBER -200
|
---|
603 | #define ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE -201
|
---|
604 | #define ONIGERR_UPPER_SMALLER_THAN_LOWER_IN_REPEAT_RANGE -202
|
---|
605 | #define ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS -203
|
---|
606 | #define ONIGERR_MISMATCH_CODE_LENGTH_IN_CLASS_RANGE -204
|
---|
607 | #define ONIGERR_TOO_MANY_MULTI_BYTE_RANGES -205
|
---|
608 | #define ONIGERR_TOO_SHORT_MULTI_BYTE_STRING -206
|
---|
609 | #define ONIGERR_TOO_BIG_BACKREF_NUMBER -207
|
---|
610 | #define ONIGERR_INVALID_BACKREF -208
|
---|
611 | #define ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED -209
|
---|
612 | #define ONIGERR_TOO_SHORT_DIGITS -210
|
---|
613 | #define ONIGERR_TOO_LONG_WIDE_CHAR_VALUE -212
|
---|
614 | #define ONIGERR_EMPTY_GROUP_NAME -214
|
---|
615 | #define ONIGERR_INVALID_GROUP_NAME -215
|
---|
616 | #define ONIGERR_INVALID_CHAR_IN_GROUP_NAME -216
|
---|
617 | #define ONIGERR_UNDEFINED_NAME_REFERENCE -217
|
---|
618 | #define ONIGERR_UNDEFINED_GROUP_REFERENCE -218
|
---|
619 | #define ONIGERR_MULTIPLEX_DEFINED_NAME -219
|
---|
620 | #define ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL -220
|
---|
621 | #define ONIGERR_NEVER_ENDING_RECURSION -221
|
---|
622 | #define ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY -222
|
---|
623 | #define ONIGERR_INVALID_CHAR_PROPERTY_NAME -223
|
---|
624 | #define ONIGERR_TOO_MANY_CAPTURE_GROUPS -224
|
---|
625 | #define ONIGERR_INVALID_CODE_POINT_VALUE -400
|
---|
626 | #define ONIGERR_INVALID_WIDE_CHAR_VALUE -400
|
---|
627 | #define ONIGERR_TOO_BIG_WIDE_CHAR_VALUE -401
|
---|
628 | #define ONIGERR_NOT_SUPPORTED_ENCODING_COMBINATION -402
|
---|
629 | #define ONIGERR_INVALID_COMBINATION_OF_OPTIONS -403
|
---|
630 |
|
---|
631 | /* errors related to thread */
|
---|
632 | #define ONIGERR_OVER_THREAD_PASS_LIMIT_COUNT -1001
|
---|
633 |
|
---|
634 |
|
---|
635 | /* must be smaller than BIT_STATUS_BITS_NUM (unsigned int * 8) */
|
---|
636 | #define ONIG_MAX_CAPTURE_HISTORY_GROUP 31
|
---|
637 | #define ONIG_IS_CAPTURE_HISTORY_GROUP(r, i) \
|
---|
638 | ((i) <= ONIG_MAX_CAPTURE_HISTORY_GROUP && (r)->list && (r)->list[i])
|
---|
639 |
|
---|
640 | typedef struct OnigCaptureTreeNodeStruct {
|
---|
641 | int group; /* group number */
|
---|
642 | OnigPosition beg;
|
---|
643 | OnigPosition end;
|
---|
644 | int allocated;
|
---|
645 | int num_childs;
|
---|
646 | struct OnigCaptureTreeNodeStruct** childs;
|
---|
647 | } OnigCaptureTreeNode;
|
---|
648 |
|
---|
649 | /* match result region type */
|
---|
650 | struct re_registers {
|
---|
651 | int allocated;
|
---|
652 | int num_regs;
|
---|
653 | OnigPosition* beg;
|
---|
654 | OnigPosition* end;
|
---|
655 | /* extended */
|
---|
656 | OnigCaptureTreeNode* history_root; /* capture history tree root */
|
---|
657 | };
|
---|
658 |
|
---|
659 | /* capture tree traverse */
|
---|
660 | #define ONIG_TRAVERSE_CALLBACK_AT_FIRST 1
|
---|
661 | #define ONIG_TRAVERSE_CALLBACK_AT_LAST 2
|
---|
662 | #define ONIG_TRAVERSE_CALLBACK_AT_BOTH \
|
---|
663 | ( ONIG_TRAVERSE_CALLBACK_AT_FIRST | ONIG_TRAVERSE_CALLBACK_AT_LAST )
|
---|
664 |
|
---|
665 |
|
---|
666 | #define ONIG_REGION_NOTPOS -1
|
---|
667 |
|
---|
668 | typedef struct re_registers OnigRegion;
|
---|
669 |
|
---|
670 | typedef struct {
|
---|
671 | OnigEncoding enc;
|
---|
672 | OnigUChar* par;
|
---|
673 | OnigUChar* par_end;
|
---|
674 | } OnigErrorInfo;
|
---|
675 |
|
---|
676 | typedef struct {
|
---|
677 | int lower;
|
---|
678 | int upper;
|
---|
679 | } OnigRepeatRange;
|
---|
680 |
|
---|
681 | typedef void (*OnigWarnFunc) P_((const char* s));
|
---|
682 | extern void onig_null_warn P_((const char* s));
|
---|
683 | #define ONIG_NULL_WARN onig_null_warn
|
---|
684 |
|
---|
685 | #define ONIG_CHAR_TABLE_SIZE 256
|
---|
686 |
|
---|
687 | /* regex_t state */
|
---|
688 | #define ONIG_STATE_NORMAL 0
|
---|
689 | #define ONIG_STATE_SEARCHING 1
|
---|
690 | #define ONIG_STATE_COMPILING -1
|
---|
691 | #define ONIG_STATE_MODIFY -2
|
---|
692 |
|
---|
693 | #define ONIG_STATE(reg) \
|
---|
694 | ((reg)->state > 0 ? ONIG_STATE_SEARCHING : (reg)->state)
|
---|
695 |
|
---|
696 | typedef struct re_pattern_buffer {
|
---|
697 | /* common members of BBuf(bytes-buffer) */
|
---|
698 | unsigned char* p; /* compiled pattern */
|
---|
699 | unsigned int used; /* used space for p */
|
---|
700 | unsigned int alloc; /* allocated space for p */
|
---|
701 |
|
---|
702 | int state; /* normal, searching, compiling */
|
---|
703 | int num_mem; /* used memory(...) num counted from 1 */
|
---|
704 | int num_repeat; /* OP_REPEAT/OP_REPEAT_NG id-counter */
|
---|
705 | int num_null_check; /* OP_NULL_CHECK_START/END id counter */
|
---|
706 | int num_comb_exp_check; /* combination explosion check */
|
---|
707 | int num_call; /* number of subexp call */
|
---|
708 | unsigned int capture_history; /* (?@...) flag (1-31) */
|
---|
709 | unsigned int bt_mem_start; /* need backtrack flag */
|
---|
710 | unsigned int bt_mem_end; /* need backtrack flag */
|
---|
711 | int stack_pop_level;
|
---|
712 | int repeat_range_alloc;
|
---|
713 | OnigRepeatRange* repeat_range;
|
---|
714 |
|
---|
715 | OnigEncoding enc;
|
---|
716 | OnigOptionType options;
|
---|
717 | OnigSyntaxType* syntax;
|
---|
718 | OnigCaseFoldType case_fold_flag;
|
---|
719 | void* name_table;
|
---|
720 |
|
---|
721 | /* optimization info (string search, char-map and anchors) */
|
---|
722 | int optimize; /* optimize flag */
|
---|
723 | int threshold_len; /* search str-length for apply optimize */
|
---|
724 | int anchor; /* BEGIN_BUF, BEGIN_POS, (SEMI_)END_BUF */
|
---|
725 | OnigDistance anchor_dmin; /* (SEMI_)END_BUF anchor distance */
|
---|
726 | OnigDistance anchor_dmax; /* (SEMI_)END_BUF anchor distance */
|
---|
727 | int sub_anchor; /* start-anchor for exact or map */
|
---|
728 | unsigned char *exact;
|
---|
729 | unsigned char *exact_end;
|
---|
730 | unsigned char map[ONIG_CHAR_TABLE_SIZE]; /* used as BM skip or char-map */
|
---|
731 | int *int_map; /* BM skip for exact_len > 255 */
|
---|
732 | int *int_map_backward; /* BM skip for backward search */
|
---|
733 | OnigDistance dmin; /* min-distance of exact or map */
|
---|
734 | OnigDistance dmax; /* max-distance of exact or map */
|
---|
735 |
|
---|
736 | /* regex_t link chain */
|
---|
737 | struct re_pattern_buffer* chain; /* escape compile-conflict */
|
---|
738 | } OnigRegexType;
|
---|
739 |
|
---|
740 | typedef OnigRegexType* OnigRegex;
|
---|
741 |
|
---|
742 | #ifndef ONIG_ESCAPE_REGEX_T_COLLISION
|
---|
743 | typedef OnigRegexType regex_t;
|
---|
744 | #endif
|
---|
745 |
|
---|
746 |
|
---|
747 | typedef struct {
|
---|
748 | int num_of_elements;
|
---|
749 | OnigEncoding pattern_enc;
|
---|
750 | OnigEncoding target_enc;
|
---|
751 | OnigSyntaxType* syntax;
|
---|
752 | OnigOptionType option;
|
---|
753 | OnigCaseFoldType case_fold_flag;
|
---|
754 | } OnigCompileInfo;
|
---|
755 |
|
---|
756 | /* Oniguruma Native API */
|
---|
757 | ONIG_EXTERN
|
---|
758 | int onig_init P_((void));
|
---|
759 | ONIG_EXTERN
|
---|
760 | int onig_error_code_to_str PV_((OnigUChar* s, OnigPosition err_code, ...));
|
---|
761 | ONIG_EXTERN
|
---|
762 | void onig_set_warn_func P_((OnigWarnFunc f));
|
---|
763 | ONIG_EXTERN
|
---|
764 | void onig_set_verb_warn_func P_((OnigWarnFunc f));
|
---|
765 | ONIG_EXTERN
|
---|
766 | int onig_new P_((OnigRegex*, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax, OnigErrorInfo* einfo));
|
---|
767 | ONIG_EXTERN
|
---|
768 | int onig_reg_init P_((OnigRegex reg, OnigOptionType option, OnigCaseFoldType case_fold_flag, OnigEncoding enc, OnigSyntaxType* syntax));
|
---|
769 | ONIG_EXTERN
|
---|
770 | int onig_new_without_alloc P_((OnigRegex, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax, OnigErrorInfo* einfo));
|
---|
771 | ONIG_EXTERN
|
---|
772 | int onig_new_deluxe P_((OnigRegex* reg, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigCompileInfo* ci, OnigErrorInfo* einfo));
|
---|
773 | ONIG_EXTERN
|
---|
774 | void onig_free P_((OnigRegex));
|
---|
775 | ONIG_EXTERN
|
---|
776 | void onig_free_body P_((OnigRegex));
|
---|
777 | ONIG_EXTERN
|
---|
778 | int onig_recompile P_((OnigRegex, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax, OnigErrorInfo* einfo));
|
---|
779 | ONIG_EXTERN
|
---|
780 | int onig_recompile_deluxe P_((OnigRegex reg, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigCompileInfo* ci, OnigErrorInfo* einfo));
|
---|
781 | ONIG_EXTERN
|
---|
782 | OnigPosition onig_search P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegion* region, OnigOptionType option));
|
---|
783 | ONIG_EXTERN
|
---|
784 | OnigPosition onig_search_gpos P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* global_pos, const OnigUChar* start, const OnigUChar* range, OnigRegion* region, OnigOptionType option));
|
---|
785 | ONIG_EXTERN
|
---|
786 | OnigPosition onig_match P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* at, OnigRegion* region, OnigOptionType option));
|
---|
787 | ONIG_EXTERN
|
---|
788 | OnigRegion* onig_region_new P_((void));
|
---|
789 | ONIG_EXTERN
|
---|
790 | void onig_region_init P_((OnigRegion* region));
|
---|
791 | ONIG_EXTERN
|
---|
792 | void onig_region_free P_((OnigRegion* region, int free_self));
|
---|
793 | ONIG_EXTERN
|
---|
794 | void onig_region_copy P_((OnigRegion* to, OnigRegion* from));
|
---|
795 | ONIG_EXTERN
|
---|
796 | void onig_region_clear P_((OnigRegion* region));
|
---|
797 | ONIG_EXTERN
|
---|
798 | int onig_region_resize P_((OnigRegion* region, int n));
|
---|
799 | ONIG_EXTERN
|
---|
800 | int onig_region_set P_((OnigRegion* region, int at, int beg, int end));
|
---|
801 | ONIG_EXTERN
|
---|
802 | int onig_name_to_group_numbers P_((OnigRegex reg, const OnigUChar* name, const OnigUChar* name_end, int** nums));
|
---|
803 | ONIG_EXTERN
|
---|
804 | int onig_name_to_backref_number P_((OnigRegex reg, const OnigUChar* name, const OnigUChar* name_end, OnigRegion *region));
|
---|
805 | ONIG_EXTERN
|
---|
806 | int onig_foreach_name P_((OnigRegex reg, int (*func)(const OnigUChar*, const OnigUChar*,int,int*,OnigRegex,void*), void* arg));
|
---|
807 | ONIG_EXTERN
|
---|
808 | int onig_number_of_names P_((OnigRegex reg));
|
---|
809 | ONIG_EXTERN
|
---|
810 | int onig_number_of_captures P_((OnigRegex reg));
|
---|
811 | ONIG_EXTERN
|
---|
812 | int onig_number_of_capture_histories P_((OnigRegex reg));
|
---|
813 | ONIG_EXTERN
|
---|
814 | OnigCaptureTreeNode* onig_get_capture_tree P_((OnigRegion* region));
|
---|
815 | ONIG_EXTERN
|
---|
816 | int onig_capture_tree_traverse P_((OnigRegion* region, int at, int(*callback_func)(int,OnigPosition,OnigPosition,int,int,void*), void* arg));
|
---|
817 | ONIG_EXTERN
|
---|
818 | int onig_noname_group_capture_is_active P_((OnigRegex reg));
|
---|
819 | ONIG_EXTERN
|
---|
820 | OnigEncoding onig_get_encoding P_((OnigRegex reg));
|
---|
821 | ONIG_EXTERN
|
---|
822 | OnigOptionType onig_get_options P_((OnigRegex reg));
|
---|
823 | ONIG_EXTERN
|
---|
824 | OnigCaseFoldType onig_get_case_fold_flag P_((OnigRegex reg));
|
---|
825 | ONIG_EXTERN
|
---|
826 | OnigSyntaxType* onig_get_syntax P_((OnigRegex reg));
|
---|
827 | ONIG_EXTERN
|
---|
828 | int onig_set_default_syntax P_((OnigSyntaxType* syntax));
|
---|
829 | ONIG_EXTERN
|
---|
830 | void onig_copy_syntax P_((OnigSyntaxType* to, const OnigSyntaxType* from));
|
---|
831 | ONIG_EXTERN
|
---|
832 | unsigned int onig_get_syntax_op P_((OnigSyntaxType* syntax));
|
---|
833 | ONIG_EXTERN
|
---|
834 | unsigned int onig_get_syntax_op2 P_((OnigSyntaxType* syntax));
|
---|
835 | ONIG_EXTERN
|
---|
836 | unsigned int onig_get_syntax_behavior P_((OnigSyntaxType* syntax));
|
---|
837 | ONIG_EXTERN
|
---|
838 | OnigOptionType onig_get_syntax_options P_((OnigSyntaxType* syntax));
|
---|
839 | ONIG_EXTERN
|
---|
840 | void onig_set_syntax_op P_((OnigSyntaxType* syntax, unsigned int op));
|
---|
841 | ONIG_EXTERN
|
---|
842 | void onig_set_syntax_op2 P_((OnigSyntaxType* syntax, unsigned int op2));
|
---|
843 | ONIG_EXTERN
|
---|
844 | void onig_set_syntax_behavior P_((OnigSyntaxType* syntax, unsigned int behavior));
|
---|
845 | ONIG_EXTERN
|
---|
846 | void onig_set_syntax_options P_((OnigSyntaxType* syntax, OnigOptionType options));
|
---|
847 | ONIG_EXTERN
|
---|
848 | int onig_set_meta_char P_((OnigSyntaxType* syntax, unsigned int what, OnigCodePoint code));
|
---|
849 | ONIG_EXTERN
|
---|
850 | void onig_copy_encoding P_((OnigEncoding to, OnigEncoding from));
|
---|
851 | ONIG_EXTERN
|
---|
852 | OnigCaseFoldType onig_get_default_case_fold_flag P_((void));
|
---|
853 | ONIG_EXTERN
|
---|
854 | int onig_set_default_case_fold_flag P_((OnigCaseFoldType case_fold_flag));
|
---|
855 | ONIG_EXTERN
|
---|
856 | unsigned int onig_get_match_stack_limit_size P_((void));
|
---|
857 | ONIG_EXTERN
|
---|
858 | int onig_set_match_stack_limit_size P_((unsigned int size));
|
---|
859 | ONIG_EXTERN
|
---|
860 | int onig_end P_((void));
|
---|
861 | ONIG_EXTERN
|
---|
862 | const char* onig_version P_((void));
|
---|
863 | ONIG_EXTERN
|
---|
864 | const char* onig_copyright P_((void));
|
---|
865 |
|
---|
866 | #ifdef __cplusplus
|
---|
867 | }
|
---|
868 | #endif
|
---|
869 |
|
---|
870 | #endif /* ONIGURUMA_H */
|
---|