3 /**********************************************************************
\r
4 oniguruma.h - Oniguruma (regular expression library)
\r
5 **********************************************************************/
\r
7 * Copyright (c) 2002-2005 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
\r
8 * All rights reserved.
\r
10 * Redistribution and use in source and binary forms, with or without
\r
11 * modification, are permitted provided that the following conditions
\r
13 * 1. Redistributions of source code must retain the above copyright
\r
14 * notice, this list of conditions and the following disclaimer.
\r
15 * 2. Redistributions in binary form must reproduce the above copyright
\r
16 * notice, this list of conditions and the following disclaimer in the
\r
17 * documentation and/or other materials provided with the distribution.
\r
19 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
\r
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
\r
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
\r
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
\r
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
\r
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
\r
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
\r
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
\r
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
\r
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
\r
37 #define ONIGURUMA_VERSION_MAJOR 3
\r
38 #define ONIGURUMA_VERSION_MINOR 9
\r
39 #define ONIGURUMA_VERSION_TEENY 1
\r
42 # ifndef HAVE_PROTOTYPES
\r
43 # define HAVE_PROTOTYPES 1
\r
45 # ifndef HAVE_STDARG_PROTOTYPES
\r
46 # define HAVE_STDARG_PROTOTYPES 1
\r
51 #if defined(__STDC__) || defined(_WIN32)
\r
52 # define P_(args) args
\r
54 # define P_(args) ()
\r
59 #ifdef HAVE_STDARG_PROTOTYPES
\r
60 # define PV_(args) args
\r
62 # define PV_(args) ()
\r
67 #if defined(_WIN32) && !defined(__GNUC__)
\r
68 #if defined(EXPORT) || defined(RUBY_EXPORT)
\r
69 #define ONIG_EXTERN extern __declspec(dllexport)
\r
71 #define ONIG_EXTERN extern __declspec(dllimport)
\r
77 #define ONIG_EXTERN extern
\r
80 /* PART: character encoding */
\r
82 #ifndef ONIG_ESCAPE_UCHAR_COLLISION
\r
83 #define UChar OnigUChar
\r
86 typedef unsigned char OnigUChar;
\r
87 typedef unsigned long OnigCodePoint;
\r
88 typedef unsigned int OnigDistance;
\r
90 #define ONIG_INFINITE_DISTANCE ~((OnigDistance )0)
\r
92 /* ambiguous match flag */
\r
93 typedef unsigned int OnigAmbigType;
\r
95 ONIG_EXTERN OnigAmbigType OnigDefaultAmbigFlag;
\r
97 #define ONIGENC_AMBIGUOUS_MATCH_NONE 0
\r
98 #define ONIGENC_AMBIGUOUS_MATCH_ASCII_CASE (1<<0)
\r
99 #define ONIGENC_AMBIGUOUS_MATCH_NONASCII_CASE (1<<1)
\r
100 /* #define ONIGENC_AMBIGUOUS_MATCH_ACCENT (1<<2) */
\r
101 /* #define ONIGENC_AMBIGUOUS_MATCH_HIRAGANA_KATAKANA (1<<3) */
\r
102 /* #define ONIGENC_AMBIGUOUS_MATCH_KATAKANA_WIDTH (1<<4) */
\r
104 #define ONIGENC_AMBIGUOUS_MATCH_LIMIT (1<<1)
\r
105 #define ONIGENC_AMBIGUOUS_MATCH_COMPOUND (1<<30)
\r
107 #define ONIGENC_AMBIGUOUS_MATCH_FULL \
\r
108 ( ONIGENC_AMBIGUOUS_MATCH_ASCII_CASE | \
\r
109 ONIGENC_AMBIGUOUS_MATCH_NONASCII_CASE | \
\r
110 ONIGENC_AMBIGUOUS_MATCH_COMPOUND )
\r
111 #define ONIGENC_AMBIGUOUS_MATCH_DEFAULT OnigDefaultAmbigFlag
\r
114 #define ONIGENC_MAX_COMP_AMBIG_CODE_LEN 3
\r
115 #define ONIGENC_MAX_COMP_AMBIG_CODE_ITEM_NUM 4
\r
118 #define ONIGENC_CODE_RANGE_NUM(range) ((int )range[0])
\r
119 #define ONIGENC_CODE_RANGE_FROM(range,i) range[((i)*2) + 1]
\r
120 #define ONIGENC_CODE_RANGE_TO(range,i) range[((i)*2) + 2]
\r
124 OnigCodePoint code[ONIGENC_MAX_COMP_AMBIG_CODE_LEN];
\r
125 } OnigCompAmbigCodeItem;
\r
129 OnigCodePoint code;
\r
130 OnigCompAmbigCodeItem items[ONIGENC_MAX_COMP_AMBIG_CODE_ITEM_NUM];
\r
131 } OnigCompAmbigCodes;
\r
134 OnigCodePoint from;
\r
136 } OnigPairAmbigCodes;
\r
140 OnigCodePoint anychar;
\r
141 OnigCodePoint anytime;
\r
142 OnigCodePoint zero_or_one_time;
\r
143 OnigCodePoint one_or_more_time;
\r
144 OnigCodePoint anychar_anytime;
\r
145 } OnigMetaCharTableType;
\r
148 #if defined(RUBY_PLATFORM) && defined(M17N_H)
\r
150 #define ONIG_RUBY_M17N
\r
151 typedef m17n_encoding* OnigEncoding;
\r
156 int (*mbc_enc_len)(const OnigUChar* p);
\r
160 OnigAmbigType support_ambig_flag;
\r
161 OnigMetaCharTableType meta_char_table;
\r
162 int (*is_mbc_newline)(const OnigUChar* p, const OnigUChar* end);
\r
163 OnigCodePoint (*mbc_to_code)(const OnigUChar* p, const OnigUChar* end);
\r
164 int (*code_to_mbclen)(OnigCodePoint code);
\r
165 int (*code_to_mbc)(OnigCodePoint code, OnigUChar *buf);
\r
166 int (*mbc_to_normalize)(OnigAmbigType flag, const OnigUChar** pp, const OnigUChar* end, OnigUChar* to);
\r
167 int (*is_mbc_ambiguous)(OnigAmbigType flag, const OnigUChar** pp, const OnigUChar* end);
\r
168 int (*get_all_pair_ambig_codes)(OnigAmbigType flag, const OnigPairAmbigCodes** acs);
\r
169 int (*get_all_comp_ambig_codes)(OnigAmbigType flag, const OnigCompAmbigCodes** acs);
\r
170 int (*is_code_ctype)(OnigCodePoint code, unsigned int ctype);
\r
171 int (*get_ctype_code_range)(int ctype, const OnigCodePoint* sb_range[], const OnigCodePoint* mb_range[]);
\r
172 OnigUChar* (*left_adjust_char_head)(const OnigUChar* start, const OnigUChar* p);
\r
173 int (*is_allowed_reverse_match)(const OnigUChar* p, const OnigUChar* end);
\r
174 } OnigEncodingType;
\r
176 typedef OnigEncodingType* OnigEncoding;
\r
178 ONIG_EXTERN OnigEncodingType OnigEncodingASCII;
\r
179 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_1;
\r
180 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_2;
\r
181 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_3;
\r
182 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_4;
\r
183 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_5;
\r
184 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_6;
\r
185 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_7;
\r
186 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_8;
\r
187 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_9;
\r
188 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_10;
\r
189 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_11;
\r
190 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_13;
\r
191 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_14;
\r
192 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_15;
\r
193 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_16;
\r
194 ONIG_EXTERN OnigEncodingType OnigEncodingUTF8;
\r
195 ONIG_EXTERN OnigEncodingType OnigEncodingUTF16_BE;
\r
196 ONIG_EXTERN OnigEncodingType OnigEncodingUTF16_LE;
\r
197 ONIG_EXTERN OnigEncodingType OnigEncodingUTF32_BE;
\r
198 ONIG_EXTERN OnigEncodingType OnigEncodingUTF32_LE;
\r
199 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_JP;
\r
200 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_TW;
\r
201 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_KR;
\r
202 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_CN;
\r
203 ONIG_EXTERN OnigEncodingType OnigEncodingSJIS;
\r
204 ONIG_EXTERN OnigEncodingType OnigEncodingKOI8;
\r
205 ONIG_EXTERN OnigEncodingType OnigEncodingKOI8_R;
\r
206 ONIG_EXTERN OnigEncodingType OnigEncodingBIG5;
\r
207 ONIG_EXTERN OnigEncodingType OnigEncodingGB18030;
\r
209 #define ONIG_ENCODING_ASCII (&OnigEncodingASCII)
\r
210 #define ONIG_ENCODING_ISO_8859_1 (&OnigEncodingISO_8859_1)
\r
211 #define ONIG_ENCODING_ISO_8859_2 (&OnigEncodingISO_8859_2)
\r
212 #define ONIG_ENCODING_ISO_8859_3 (&OnigEncodingISO_8859_3)
\r
213 #define ONIG_ENCODING_ISO_8859_4 (&OnigEncodingISO_8859_4)
\r
214 #define ONIG_ENCODING_ISO_8859_5 (&OnigEncodingISO_8859_5)
\r
215 #define ONIG_ENCODING_ISO_8859_6 (&OnigEncodingISO_8859_6)
\r
216 #define ONIG_ENCODING_ISO_8859_7 (&OnigEncodingISO_8859_7)
\r
217 #define ONIG_ENCODING_ISO_8859_8 (&OnigEncodingISO_8859_8)
\r
218 #define ONIG_ENCODING_ISO_8859_9 (&OnigEncodingISO_8859_9)
\r
219 #define ONIG_ENCODING_ISO_8859_10 (&OnigEncodingISO_8859_10)
\r
220 #define ONIG_ENCODING_ISO_8859_11 (&OnigEncodingISO_8859_11)
\r
221 #define ONIG_ENCODING_ISO_8859_13 (&OnigEncodingISO_8859_13)
\r
222 #define ONIG_ENCODING_ISO_8859_14 (&OnigEncodingISO_8859_14)
\r
223 #define ONIG_ENCODING_ISO_8859_15 (&OnigEncodingISO_8859_15)
\r
224 #define ONIG_ENCODING_ISO_8859_16 (&OnigEncodingISO_8859_16)
\r
225 #define ONIG_ENCODING_UTF8 (&OnigEncodingUTF8)
\r
226 #define ONIG_ENCODING_UTF16_BE (&OnigEncodingUTF16_BE)
\r
227 #define ONIG_ENCODING_UTF16_LE (&OnigEncodingUTF16_LE)
\r
228 #define ONIG_ENCODING_UTF32_BE (&OnigEncodingUTF32_BE)
\r
229 #define ONIG_ENCODING_UTF32_LE (&OnigEncodingUTF32_LE)
\r
230 #define ONIG_ENCODING_EUC_JP (&OnigEncodingEUC_JP)
\r
231 #define ONIG_ENCODING_EUC_TW (&OnigEncodingEUC_TW)
\r
232 #define ONIG_ENCODING_EUC_KR (&OnigEncodingEUC_KR)
\r
233 #define ONIG_ENCODING_EUC_CN (&OnigEncodingEUC_CN)
\r
234 #define ONIG_ENCODING_SJIS (&OnigEncodingSJIS)
\r
235 #define ONIG_ENCODING_KOI8 (&OnigEncodingKOI8)
\r
236 #define ONIG_ENCODING_KOI8_R (&OnigEncodingKOI8_R)
\r
237 #define ONIG_ENCODING_BIG5 (&OnigEncodingBIG5)
\r
238 #define ONIG_ENCODING_GB18030 (&OnigEncodingGB18030)
\r
240 #endif /* else RUBY && M17N */
\r
242 #define ONIG_ENCODING_UNDEF ((OnigEncoding )0)
\r
246 #define ONIGENC_CODE_TO_MBC_MAXLEN 7
\r
247 #define ONIGENC_MBC_NORMALIZE_MAXLEN ONIGENC_CODE_TO_MBC_MAXLEN
\r
249 /* character types */
\r
250 #define ONIGENC_CTYPE_NEWLINE (1<< 0)
\r
251 #define ONIGENC_CTYPE_ALPHA (1<< 1)
\r
252 #define ONIGENC_CTYPE_BLANK (1<< 2)
\r
253 #define ONIGENC_CTYPE_CNTRL (1<< 3)
\r
254 #define ONIGENC_CTYPE_DIGIT (1<< 4)
\r
255 #define ONIGENC_CTYPE_GRAPH (1<< 5)
\r
256 #define ONIGENC_CTYPE_LOWER (1<< 6)
\r
257 #define ONIGENC_CTYPE_PRINT (1<< 7)
\r
258 #define ONIGENC_CTYPE_PUNCT (1<< 8)
\r
259 #define ONIGENC_CTYPE_SPACE (1<< 9)
\r
260 #define ONIGENC_CTYPE_UPPER (1<<10)
\r
261 #define ONIGENC_CTYPE_XDIGIT (1<<11)
\r
262 #define ONIGENC_CTYPE_WORD (1<<12)
\r
263 #define ONIGENC_CTYPE_ASCII (1<<13)
\r
264 #define ONIGENC_CTYPE_ALNUM (ONIGENC_CTYPE_ALPHA | ONIGENC_CTYPE_DIGIT)
\r
266 #define enc_len(enc,p) ONIGENC_MBC_ENC_LEN(enc, p)
\r
268 #define ONIGENC_IS_UNDEF(enc) ((enc) == ONIG_ENCODING_UNDEF)
\r
269 #define ONIGENC_IS_SINGLEBYTE(enc) (ONIGENC_MBC_MAXLEN(enc) == 1)
\r
270 #define ONIGENC_IS_MBC_HEAD(enc,p) (ONIGENC_MBC_ENC_LEN(enc,p) != 1)
\r
271 #define ONIGENC_IS_MBC_ASCII(p) (*(p) < 128)
\r
272 #define ONIGENC_IS_CODE_ASCII(code) ((code) < 128)
\r
273 #define ONIGENC_IS_CODE_SB_WORD(enc,code) \
\r
274 (ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code))
\r
275 #define ONIGENC_IS_MBC_WORD(enc,s,end) \
\r
276 ONIGENC_IS_CODE_WORD(enc,ONIGENC_MBC_TO_CODE(enc,s,end))
\r
279 #ifdef ONIG_RUBY_M17N
\r
281 #include <ctype.h> /* for isblank(), isgraph() */
\r
283 #define ONIGENC_MBC_TO_NORMALIZE(enc,flag,pp,end,buf) \
\r
284 onigenc_mbc_to_normalize(enc,flag,pp,end,buf)
\r
285 #define ONIGENC_IS_MBC_AMBIGUOUS(enc,flag,pp,end) \
\r
286 onigenc_is_mbc_ambiguous(enc,flag,pp,end)
\r
288 #define ONIGENC_SUPPORT_AMBIG_FLAG(enc) ONIGENC_AMBIGUOUS_MATCH_ASCII_CASE
\r
289 #define ONIGENC_IS_ALLOWED_REVERSE_MATCH(enc,s,end) \
\r
290 onigenc_is_allowed_reverse_match(enc, s, end)
\r
291 #define ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc,start,s) \
\r
292 onigenc_get_left_adjust_char_head(enc, start, s)
\r
293 #define ONIGENC_GET_ALL_PAIR_AMBIG_CODES(enc, ambig_flag, acs) 0
\r
294 #define ONIGENC_GET_ALL_COMP_AMBIG_CODES(enc, ambig_flag, acs) 0
\r
295 #define ONIGENC_GET_CTYPE_CODE_RANGE(enc,ctype,sbr,mbr) \
\r
296 ONIG_NO_SUPPORT_CONFIG
\r
297 #define ONIGENC_MBC_ENC_LEN(enc,p) m17n_mbclen(enc,(int )(*p))
\r
298 #define ONIGENC_MBC_MAXLEN(enc) m17n_mbmaxlen(enc)
\r
299 #define ONIGENC_MBC_MAXLEN_DIST(enc) \
\r
300 (ONIGENC_MBC_MAXLEN(enc) > 0 ? ONIGENC_MBC_MAXLEN(enc) \
\r
301 : ONIG_INFINITE_DISTANCE)
\r
302 #define ONIGENC_MBC_MINLEN(enc) 1
\r
303 #define ONIGENC_MBC_TO_CODE(enc,p,e) m17n_codepoint((enc),(p),(e))
\r
304 #define ONIGENC_CODE_TO_MBCLEN(enc,code) m17n_codelen((enc),(code))
\r
305 #define ONIGENC_CODE_TO_MBC(enc,code,buf) onigenc_code_to_mbc(enc, code, buf)
\r
307 #if 0 /* !! not supported !! */
\r
308 #define ONIGENC_IS_MBC_NEWLINE(enc,p,end)
\r
309 #define ONIGENC_STEP_BACK(enc,start,s,n)
\r
312 #define ONIGENC_IS_CODE_CTYPE(enc,code,ctype) \
\r
313 onigenc_is_code_ctype(enc,code,ctype)
\r
316 # define ONIGENC_IS_CODE_BLANK(enc,code) isblank((int )code)
\r
318 # define ONIGENC_IS_CODE_BLANK(enc,code) ((code) == ' ' || (code) == '\t')
\r
321 # define ONIGENC_IS_CODE_GRAPH(enc,code) isgraph((int )code)
\r
323 # define ONIGENC_IS_CODE_GRAPH(enc,code) \
\r
324 (isprint((int )code) && !isspace((int )code))
\r
327 #define ONIGENC_IS_CODE_PRINT(enc,code) m17n_isprint(enc,code)
\r
328 #define ONIGENC_IS_CODE_ALNUM(enc,code) m17n_isalnum(enc,code)
\r
329 #define ONIGENC_IS_CODE_ALPHA(enc,code) m17n_isalpha(enc,code)
\r
330 #define ONIGENC_IS_CODE_LOWER(enc,code) m17n_islower(enc,code)
\r
331 #define ONIGENC_IS_CODE_UPPER(enc,code) m17n_isupper(enc,code)
\r
332 #define ONIGENC_IS_CODE_CNTRL(enc,code) m17n_iscntrl(enc,code)
\r
333 #define ONIGENC_IS_CODE_PUNCT(enc,code) m17n_ispunct(enc,code)
\r
334 #define ONIGENC_IS_CODE_SPACE(enc,code) m17n_isspace(enc,code)
\r
335 #define ONIGENC_IS_CODE_DIGIT(enc,code) m17n_isdigit(enc,code)
\r
336 #define ONIGENC_IS_CODE_XDIGIT(enc,code) m17n_isxdigit(enc,code)
\r
337 #define ONIGENC_IS_CODE_WORD(enc,code) m17n_iswchar(enc,code)
\r
340 int onigenc_is_code_ctype P_((OnigEncoding enc, OnigCodePoint code, int ctype));
\r
342 int onigenc_code_to_mbc P_((OnigEncoding enc, OnigCodePoint code, OnigUChar *buf));
\r
344 int onigenc_mbc_to_normalize P_((OnigEncoding enc, OnigAmbigType flag, const OnigUChar** pp, const OnigUChar* end, OnigUChar* buf));
\r
346 int onigenc_is_mbc_ambiguous P_((OnigEncoding enc, OnigAmbigType flag, const OnigUChar** pp, const OnigUChar* end));
\r
348 int onigenc_is_allowed_reverse_match P_((OnigEncoding enc, const OnigUChar* s, const OnigUChar* end));
\r
350 #else /* ONIG_RUBY_M17N */
\r
352 #define ONIGENC_NAME(enc) ((enc)->name)
\r
354 #define ONIGENC_MBC_TO_NORMALIZE(enc,flag,pp,end,buf) \
\r
355 (enc)->mbc_to_normalize(flag,(const OnigUChar** )pp,end,buf)
\r
356 #define ONIGENC_IS_MBC_AMBIGUOUS(enc,flag,pp,end) \
\r
357 (enc)->is_mbc_ambiguous(flag,(const OnigUChar** )pp,end)
\r
358 #define ONIGENC_SUPPORT_AMBIG_FLAG(enc) ((enc)->support_ambig_flag)
\r
359 #define ONIGENC_IS_ALLOWED_REVERSE_MATCH(enc,s,end) \
\r
360 (enc)->is_allowed_reverse_match(s,end)
\r
361 #define ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc,start,s) \
\r
362 (enc)->left_adjust_char_head(start, s)
\r
363 #define ONIGENC_GET_ALL_PAIR_AMBIG_CODES(enc,ambig_flag,acs) \
\r
364 (enc)->get_all_pair_ambig_codes(ambig_flag,acs)
\r
365 #define ONIGENC_GET_ALL_COMP_AMBIG_CODES(enc,ambig_flag,acs) \
\r
366 (enc)->get_all_comp_ambig_codes(ambig_flag,acs)
\r
367 #define ONIGENC_STEP_BACK(enc,start,s,n) \
\r
368 onigenc_step_back((enc),(start),(s),(n))
\r
370 #define ONIGENC_MBC_ENC_LEN(enc,p) (enc)->mbc_enc_len(p)
\r
371 #define ONIGENC_MBC_MAXLEN(enc) ((enc)->max_enc_len)
\r
372 #define ONIGENC_MBC_MAXLEN_DIST(enc) ONIGENC_MBC_MAXLEN(enc)
\r
373 #define ONIGENC_MBC_MINLEN(enc) ((enc)->min_enc_len)
\r
374 #define ONIGENC_IS_MBC_NEWLINE(enc,p,end) (enc)->is_mbc_newline((p),(end))
\r
375 #define ONIGENC_MBC_TO_CODE(enc,p,end) (enc)->mbc_to_code((p),(end))
\r
376 #define ONIGENC_CODE_TO_MBCLEN(enc,code) (enc)->code_to_mbclen(code)
\r
377 #define ONIGENC_CODE_TO_MBC(enc,code,buf) (enc)->code_to_mbc(code,buf)
\r
379 #define ONIGENC_IS_CODE_CTYPE(enc,code,ctype) (enc)->is_code_ctype(code,ctype)
\r
381 #define ONIGENC_IS_CODE_NEWLINE(enc,code) \
\r
382 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_NEWLINE)
\r
383 #define ONIGENC_IS_CODE_GRAPH(enc,code) \
\r
384 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_GRAPH)
\r
385 #define ONIGENC_IS_CODE_PRINT(enc,code) \
\r
386 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PRINT)
\r
387 #define ONIGENC_IS_CODE_ALNUM(enc,code) \
\r
388 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALNUM)
\r
389 #define ONIGENC_IS_CODE_ALPHA(enc,code) \
\r
390 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALPHA)
\r
391 #define ONIGENC_IS_CODE_LOWER(enc,code) \
\r
392 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_LOWER)
\r
393 #define ONIGENC_IS_CODE_UPPER(enc,code) \
\r
394 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_UPPER)
\r
395 #define ONIGENC_IS_CODE_CNTRL(enc,code) \
\r
396 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_CNTRL)
\r
397 #define ONIGENC_IS_CODE_PUNCT(enc,code) \
\r
398 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PUNCT)
\r
399 #define ONIGENC_IS_CODE_SPACE(enc,code) \
\r
400 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_SPACE)
\r
401 #define ONIGENC_IS_CODE_BLANK(enc,code) \
\r
402 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_BLANK)
\r
403 #define ONIGENC_IS_CODE_DIGIT(enc,code) \
\r
404 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_DIGIT)
\r
405 #define ONIGENC_IS_CODE_XDIGIT(enc,code) \
\r
406 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_XDIGIT)
\r
407 #define ONIGENC_IS_CODE_WORD(enc,code) \
\r
408 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_WORD)
\r
410 #define ONIGENC_GET_CTYPE_CODE_RANGE(enc,ctype,sbr,mbr) \
\r
411 (enc)->get_ctype_code_range(ctype,sbr,mbr)
\r
414 OnigUChar* onigenc_step_back P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s, int n));
\r
416 #endif /* is not ONIG_RUBY_M17N */
\r
421 int onigenc_init P_(());
\r
423 int onigenc_set_default_encoding P_((OnigEncoding enc));
\r
425 OnigEncoding onigenc_get_default_encoding P_(());
\r
427 void onigenc_set_default_caseconv_table P_((const OnigUChar* table));
\r
429 OnigUChar* onigenc_get_right_adjust_char_head_with_prev P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s, const OnigUChar** prev));
\r
431 OnigUChar* onigenc_get_prev_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
\r
433 OnigUChar* onigenc_get_left_adjust_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
\r
435 OnigUChar* onigenc_get_right_adjust_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s));
\r
437 int onigenc_strlen P_((OnigEncoding enc, const OnigUChar* p, const OnigUChar* end));
\r
439 int onigenc_strlen_null P_((OnigEncoding enc, const OnigUChar* p));
\r
441 int onigenc_str_bytelen_null P_((OnigEncoding enc, const OnigUChar* p));
\r
445 /* PART: regular expression */
\r
447 /* config parameters */
\r
448 #define ONIG_NREGION 10
\r
449 #define ONIG_MAX_BACKREF_NUM 1000
\r
450 #define ONIG_MAX_REPEAT_NUM 100000
\r
451 #define ONIG_MAX_MULTI_BYTE_RANGES_NUM 1000
\r
453 #define ONIG_MAX_ERROR_MESSAGE_LEN 90
\r
455 typedef unsigned int OnigOptionType;
\r
457 #define ONIG_OPTION_DEFAULT ONIG_OPTION_NONE
\r
460 #define ONIG_OPTION_NONE 0
\r
461 #define ONIG_OPTION_IGNORECASE 1L
\r
462 #define ONIG_OPTION_EXTEND (ONIG_OPTION_IGNORECASE << 1)
\r
463 #define ONIG_OPTION_MULTILINE (ONIG_OPTION_EXTEND << 1)
\r
464 #define ONIG_OPTION_SINGLELINE (ONIG_OPTION_MULTILINE << 1)
\r
465 #define ONIG_OPTION_FIND_LONGEST (ONIG_OPTION_SINGLELINE << 1)
\r
466 #define ONIG_OPTION_FIND_NOT_EMPTY (ONIG_OPTION_FIND_LONGEST << 1)
\r
467 #define ONIG_OPTION_NEGATE_SINGLELINE (ONIG_OPTION_FIND_NOT_EMPTY << 1)
\r
468 #define ONIG_OPTION_DONT_CAPTURE_GROUP (ONIG_OPTION_NEGATE_SINGLELINE << 1)
\r
469 #define ONIG_OPTION_CAPTURE_GROUP (ONIG_OPTION_DONT_CAPTURE_GROUP << 1)
\r
470 /* options (search time) */
\r
471 #define ONIG_OPTION_NOTBOL (ONIG_OPTION_CAPTURE_GROUP << 1)
\r
472 #define ONIG_OPTION_NOTEOL (ONIG_OPTION_NOTBOL << 1)
\r
473 #define ONIG_OPTION_POSIX_REGION (ONIG_OPTION_NOTEOL << 1)
\r
474 #define ONIG_OPTION_MAXBIT ONIG_OPTION_POSIX_REGION /* limit */
\r
476 #define ONIG_OPTION_ON(options,regopt) ((options) |= (regopt))
\r
477 #define ONIG_OPTION_OFF(options,regopt) ((options) &= ~(regopt))
\r
478 #define ONIG_IS_OPTION_ON(options,option) ((options) & (option))
\r
484 unsigned int behavior;
\r
485 OnigOptionType options; /* default option */
\r
488 ONIG_EXTERN OnigSyntaxType OnigSyntaxASIS;
\r
489 ONIG_EXTERN OnigSyntaxType OnigSyntaxPosixBasic;
\r
490 ONIG_EXTERN OnigSyntaxType OnigSyntaxPosixExtended;
\r
491 ONIG_EXTERN OnigSyntaxType OnigSyntaxEmacs;
\r
492 ONIG_EXTERN OnigSyntaxType OnigSyntaxGrep;
\r
493 ONIG_EXTERN OnigSyntaxType OnigSyntaxGnuRegex;
\r
494 ONIG_EXTERN OnigSyntaxType OnigSyntaxJava;
\r
495 ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl;
\r
496 ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl_NG;
\r
497 ONIG_EXTERN OnigSyntaxType OnigSyntaxRuby;
\r
499 /* predefined syntaxes (see regsyntax.c) */
\r
500 #define ONIG_SYNTAX_ASIS (&OnigSyntaxASIS)
\r
501 #define ONIG_SYNTAX_POSIX_BASIC (&OnigSyntaxPosixBasic)
\r
502 #define ONIG_SYNTAX_POSIX_EXTENDED (&OnigSyntaxPosixExtended)
\r
503 #define ONIG_SYNTAX_EMACS (&OnigSyntaxEmacs)
\r
504 #define ONIG_SYNTAX_GREP (&OnigSyntaxGrep)
\r
505 #define ONIG_SYNTAX_GNU_REGEX (&OnigSyntaxGnuRegex)
\r
506 #define ONIG_SYNTAX_JAVA (&OnigSyntaxJava)
\r
507 #define ONIG_SYNTAX_PERL (&OnigSyntaxPerl)
\r
508 #define ONIG_SYNTAX_PERL_NG (&OnigSyntaxPerl_NG)
\r
509 #define ONIG_SYNTAX_RUBY (&OnigSyntaxRuby)
\r
511 /* default syntax */
\r
512 ONIG_EXTERN OnigSyntaxType* OnigDefaultSyntax;
\r
513 #define ONIG_SYNTAX_DEFAULT OnigDefaultSyntax
\r
515 /* syntax (operators) */
\r
516 #define ONIG_SYN_OP_VARIABLE_META_CHARACTERS (1<<0)
\r
517 #define ONIG_SYN_OP_DOT_ANYCHAR (1<<1) /* . */
\r
518 #define ONIG_SYN_OP_ASTERISK_ZERO_INF (1<<2) /* * */
\r
519 #define ONIG_SYN_OP_ESC_ASTERISK_ZERO_INF (1<<3)
\r
520 #define ONIG_SYN_OP_PLUS_ONE_INF (1<<4) /* + */
\r
521 #define ONIG_SYN_OP_ESC_PLUS_ONE_INF (1<<5)
\r
522 #define ONIG_SYN_OP_QMARK_ZERO_ONE (1<<6) /* ? */
\r
523 #define ONIG_SYN_OP_ESC_QMARK_ZERO_ONE (1<<7)
\r
524 #define ONIG_SYN_OP_BRACE_INTERVAL (1<<8) /* {lower,upper} */
\r
525 #define ONIG_SYN_OP_ESC_BRACE_INTERVAL (1<<9) /* \{lower,upper\} */
\r
526 #define ONIG_SYN_OP_VBAR_ALT (1<<10) /* | */
\r
527 #define ONIG_SYN_OP_ESC_VBAR_ALT (1<<11) /* \| */
\r
528 #define ONIG_SYN_OP_LPAREN_SUBEXP (1<<12) /* (...) */
\r
529 #define ONIG_SYN_OP_ESC_LPAREN_SUBEXP (1<<13) /* \(...\) */
\r
530 #define ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR (1<<14) /* \A, \Z, \z */
\r
531 #define ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR (1<<15) /* \G */
\r
532 #define ONIG_SYN_OP_DECIMAL_BACKREF (1<<16) /* \num */
\r
533 #define ONIG_SYN_OP_BRACKET_CC (1<<17) /* [...] */
\r
534 #define ONIG_SYN_OP_ESC_W_WORD (1<<18) /* \w, \W */
\r
535 #define ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END (1<<19) /* \<. \> */
\r
536 #define ONIG_SYN_OP_ESC_B_WORD_BOUND (1<<20) /* \b, \B */
\r
537 #define ONIG_SYN_OP_ESC_S_WHITE_SPACE (1<<21) /* \s, \S */
\r
538 #define ONIG_SYN_OP_ESC_D_DIGIT (1<<22) /* \d, \D */
\r
539 #define ONIG_SYN_OP_LINE_ANCHOR (1<<23) /* ^, $ */
\r
540 #define ONIG_SYN_OP_POSIX_BRACKET (1<<24) /* [:xxxx:] */
\r
541 #define ONIG_SYN_OP_QMARK_NON_GREEDY (1<<25) /* ??,*?,+?,{n,m}? */
\r
542 #define ONIG_SYN_OP_ESC_CONTROL_CHARS (1<<26) /* \n,\r,\t,\a ... */
\r
543 #define ONIG_SYN_OP_ESC_C_CONTROL (1<<27) /* \cx */
\r
544 #define ONIG_SYN_OP_ESC_OCTAL3 (1<<28) /* \OOO */
\r
545 #define ONIG_SYN_OP_ESC_X_HEX2 (1<<29) /* \xHH */
\r
546 #define ONIG_SYN_OP_ESC_X_BRACE_HEX8 (1<<30) /* \x{7HHHHHHH} */
\r
548 #define ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE (1<<0) /* \Q...\E */
\r
549 #define ONIG_SYN_OP2_QMARK_GROUP_EFFECT (1<<1) /* (?...) */
\r
550 #define ONIG_SYN_OP2_OPTION_PERL (1<<2) /* (?imsx),(?-imsx) */
\r
551 #define ONIG_SYN_OP2_OPTION_RUBY (1<<3) /* (?imx), (?-imx) */
\r
552 #define ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT (1<<4) /* ?+,*+,++ */
\r
553 #define ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL (1<<5) /* {n,m}+ */
\r
554 #define ONIG_SYN_OP2_CCLASS_SET_OP (1<<6) /* [...&&..[..]..] */
\r
555 #define ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP (1<<7) /* (?<name>...) */
\r
556 #define ONIG_SYN_OP2_ESC_K_NAMED_BACKREF (1<<8) /* \k<name> */
\r
557 #define ONIG_SYN_OP2_ESC_G_SUBEXP_CALL (1<<9) /* \g<name>, \g<n> */
\r
558 #define ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY (1<<10) /* (?@..),(?@<x>..) */
\r
559 #define ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL (1<<11) /* \C-x */
\r
560 #define ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META (1<<12) /* \M-x */
\r
561 #define ONIG_SYN_OP2_ESC_V_VTAB (1<<13) /* \v as VTAB */
\r
562 #define ONIG_SYN_OP2_ESC_U_HEX4 (1<<14) /* \uHHHH */
\r
563 #define ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR (1<<15) /* \`, \' */
\r
564 #define ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY (1<<16) /* \p{...}, \P{...} */
\r
565 #define ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT (1<<17) /* \p{^..}, \P{^..} */
\r
566 #define ONIG_SYN_OP2_CHAR_PROPERTY_PREFIX_IS (1<<18) /* \p{IsXDigit} */
\r
567 #define ONIG_SYN_OP2_ESC_H_XDIGIT (1<<19) /* \h, \H */
\r
568 #define ONIG_SYN_OP2_INEFFECTIVE_ESCAPE (1<<20) /* \ */
\r
570 /* syntax (behavior) */
\r
571 #define ONIG_SYN_CONTEXT_INDEP_ANCHORS (1<<31) /* not implemented */
\r
572 #define ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS (1<<0) /* ?, *, +, {n,m} */
\r
573 #define ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS (1<<1) /* error or ignore */
\r
574 #define ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP (1<<2) /* ...)... */
\r
575 #define ONIG_SYN_ALLOW_INVALID_INTERVAL (1<<3) /* {??? */
\r
576 #define ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV (1<<4) /* {,n} => {0,n} */
\r
577 #define ONIG_SYN_STRICT_CHECK_BACKREF (1<<5) /* /(\1)/,/\1()/ ..*/
\r
578 #define ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND (1<<6) /* (?<=a|bc) */
\r
579 #define ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP (1<<7) /* see doc/RE */
\r
580 #define ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME (1<<8) /* (?<x>)(?<x>) */
\r
581 #define ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY (1<<9) /* a{n}?=(?:a{n})? */
\r
583 /* syntax (behavior) in char class [...] */
\r
584 #define ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC (1<<20) /* [^...] */
\r
585 #define ONIG_SYN_BACKSLASH_ESCAPE_IN_CC (1<<21) /* [..\w..] etc.. */
\r
586 #define ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC (1<<22)
\r
587 #define ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC (1<<23) /* [0-9-a]=[0-9\-a] */
\r
588 /* syntax (behavior) warning */
\r
589 #define ONIG_SYN_WARN_CC_OP_NOT_ESCAPED (1<<24) /* [,-,] */
\r
590 #define ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT (1<<25) /* (?:a*)+ */
\r
592 /* meta character specifiers (onig_set_meta_char()) */
\r
593 #define ONIG_META_CHAR_ESCAPE 0
\r
594 #define ONIG_META_CHAR_ANYCHAR 1
\r
595 #define ONIG_META_CHAR_ANYTIME 2
\r
596 #define ONIG_META_CHAR_ZERO_OR_ONE_TIME 3
\r
597 #define ONIG_META_CHAR_ONE_OR_MORE_TIME 4
\r
598 #define ONIG_META_CHAR_ANYCHAR_ANYTIME 5
\r
600 #define ONIG_INEFFECTIVE_META_CHAR 0
\r
603 #define ONIG_IS_PATTERN_ERROR(ecode) ((ecode) <= -100 && (ecode) > -1000)
\r
604 /* normal return */
\r
605 #define ONIG_NORMAL 0
\r
606 #define ONIG_MISMATCH -1
\r
607 #define ONIG_NO_SUPPORT_CONFIG -2
\r
609 /* internal error */
\r
610 #define ONIGERR_MEMORY -5
\r
611 #define ONIGERR_TYPE_BUG -6
\r
612 #define ONIGERR_PARSER_BUG -11
\r
613 #define ONIGERR_STACK_BUG -12
\r
614 #define ONIGERR_UNDEFINED_BYTECODE -13
\r
615 #define ONIGERR_UNEXPECTED_BYTECODE -14
\r
616 #define ONIGERR_MATCH_STACK_LIMIT_OVER -15
\r
617 #define ONIGERR_DEFAULT_ENCODING_IS_NOT_SETTED -21
\r
618 #define ONIGERR_SPECIFIED_ENCODING_CANT_CONVERT_TO_WIDE_CHAR -22
\r
619 /* general error */
\r
620 #define ONIGERR_INVALID_ARGUMENT -30
\r
622 #define ONIGERR_END_PATTERN_AT_LEFT_BRACE -100
\r
623 #define ONIGERR_END_PATTERN_AT_LEFT_BRACKET -101
\r
624 #define ONIGERR_EMPTY_CHAR_CLASS -102
\r
625 #define ONIGERR_PREMATURE_END_OF_CHAR_CLASS -103
\r
626 #define ONIGERR_END_PATTERN_AT_ESCAPE -104
\r
627 #define ONIGERR_END_PATTERN_AT_META -105
\r
628 #define ONIGERR_END_PATTERN_AT_CONTROL -106
\r
629 #define ONIGERR_META_CODE_SYNTAX -108
\r
630 #define ONIGERR_CONTROL_CODE_SYNTAX -109
\r
631 #define ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE -110
\r
632 #define ONIGERR_CHAR_CLASS_VALUE_AT_START_OF_RANGE -111
\r
633 #define ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS -112
\r
634 #define ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED -113
\r
635 #define ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID -114
\r
636 #define ONIGERR_NESTED_REPEAT_OPERATOR -115
\r
637 #define ONIGERR_UNMATCHED_CLOSE_PARENTHESIS -116
\r
638 #define ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS -117
\r
639 #define ONIGERR_END_PATTERN_IN_GROUP -118
\r
640 #define ONIGERR_UNDEFINED_GROUP_OPTION -119
\r
641 #define ONIGERR_INVALID_POSIX_BRACKET_TYPE -121
\r
642 #define ONIGERR_INVALID_LOOK_BEHIND_PATTERN -122
\r
643 #define ONIGERR_INVALID_REPEAT_RANGE_PATTERN -123
\r
644 /* values error (syntax error) */
\r
645 #define ONIGERR_TOO_BIG_NUMBER -200
\r
646 #define ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE -201
\r
647 #define ONIGERR_UPPER_SMALLER_THAN_LOWER_IN_REPEAT_RANGE -202
\r
648 #define ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS -203
\r
649 #define ONIGERR_MISMATCH_CODE_LENGTH_IN_CLASS_RANGE -204
\r
650 #define ONIGERR_TOO_MANY_MULTI_BYTE_RANGES -205
\r
651 #define ONIGERR_TOO_SHORT_MULTI_BYTE_STRING -206
\r
652 #define ONIGERR_TOO_BIG_BACKREF_NUMBER -207
\r
653 #define ONIGERR_INVALID_BACKREF -208
\r
654 #define ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED -209
\r
655 #define ONIGERR_TOO_LONG_WIDE_CHAR_VALUE -212
\r
656 #define ONIGERR_EMPTY_GROUP_NAME -214
\r
657 #define ONIGERR_INVALID_GROUP_NAME -215
\r
658 #define ONIGERR_INVALID_CHAR_IN_GROUP_NAME -216
\r
659 #define ONIGERR_UNDEFINED_NAME_REFERENCE -217
\r
660 #define ONIGERR_UNDEFINED_GROUP_REFERENCE -218
\r
661 #define ONIGERR_MULTIPLEX_DEFINED_NAME -219
\r
662 #define ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL -220
\r
663 #define ONIGERR_NEVER_ENDING_RECURSION -221
\r
664 #define ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY -222
\r
665 #define ONIGERR_INVALID_CHAR_PROPERTY_NAME -223
\r
666 #define ONIGERR_INVALID_WIDE_CHAR_VALUE -400
\r
667 #define ONIGERR_TOO_BIG_WIDE_CHAR_VALUE -401
\r
668 #define ONIGERR_NOT_SUPPORTED_ENCODING_COMBINATION -402
\r
670 /* errors related to thread */
\r
671 #define ONIGERR_OVER_THREAD_PASS_LIMIT_COUNT -1001
\r
674 /* must be smaller than BIT_STATUS_BITS_NUM (unsigned int * 8) */
\r
675 #define ONIG_MAX_CAPTURE_HISTORY_GROUP 31
\r
676 #define ONIG_IS_CAPTURE_HISTORY_GROUP(r, i) \
\r
677 ((i) <= ONIG_MAX_CAPTURE_HISTORY_GROUP && (r)->list && (r)->list[i])
\r
679 typedef struct OnigCaptureTreeNodeStruct {
\r
680 int group; /* group number */
\r
685 struct OnigCaptureTreeNodeStruct** childs;
\r
686 } OnigCaptureTreeNode;
\r
688 /* match result region type */
\r
689 struct re_registers {
\r
695 OnigCaptureTreeNode* history_root; /* capture history tree root */
\r
698 /* capture tree traverse */
\r
699 #define ONIG_TRAVERSE_CALLBACK_AT_FIRST 1
\r
700 #define ONIG_TRAVERSE_CALLBACK_AT_LAST 2
\r
701 #define ONIG_TRAVERSE_CALLBACK_AT_BOTH \
\r
702 ( ONIG_TRAVERSE_CALLBACK_AT_FIRST | ONIG_TRAVERSE_CALLBACK_AT_LAST )
\r
705 #define ONIG_REGION_NOTPOS -1
\r
707 typedef struct re_registers OnigRegion;
\r
711 OnigUChar* par_end;
\r
719 typedef void (*OnigWarnFunc) P_((const char* s));
\r
720 extern void onig_null_warn P_((const char* s));
\r
721 #define ONIG_NULL_WARN onig_null_warn
\r
723 #define ONIG_CHAR_TABLE_SIZE 256
\r
725 /* regex_t state */
\r
726 #define ONIG_STATE_NORMAL 0
\r
727 #define ONIG_STATE_SEARCHING 1
\r
728 #define ONIG_STATE_COMPILING -1
\r
729 #define ONIG_STATE_MODIFY -2
\r
731 #define ONIG_STATE(reg) \
\r
732 ((reg)->state > 0 ? ONIG_STATE_SEARCHING : (reg)->state)
\r
734 typedef struct re_pattern_buffer {
\r
735 /* common members of BBuf(bytes-buffer) */
\r
736 unsigned char* p; /* compiled pattern */
\r
737 unsigned int used; /* used space for p */
\r
738 unsigned int alloc; /* allocated space for p */
\r
740 int state; /* normal, searching, compiling */
\r
741 int num_mem; /* used memory(...) num counted from 1 */
\r
742 int num_repeat; /* OP_REPEAT/OP_REPEAT_NG id-counter */
\r
743 int num_null_check; /* OP_NULL_CHECK_START/END id counter */
\r
744 int num_call; /* number of subexp call */
\r
745 unsigned int capture_history; /* (?@...) flag (1-31) */
\r
746 unsigned int bt_mem_start; /* need backtrack flag */
\r
747 unsigned int bt_mem_end; /* need backtrack flag */
\r
748 int stack_pop_level;
\r
749 int repeat_range_alloc;
\r
750 OnigRepeatRange* repeat_range;
\r
753 OnigOptionType options;
\r
754 OnigSyntaxType* syntax;
\r
755 OnigAmbigType ambig_flag;
\r
758 /* optimization info (string search, char-map and anchors) */
\r
759 int optimize; /* optimize flag */
\r
760 int threshold_len; /* search str-length for apply optimize */
\r
761 int anchor; /* BEGIN_BUF, BEGIN_POS, (SEMI_)END_BUF */
\r
762 OnigDistance anchor_dmin; /* (SEMI_)END_BUF anchor distance */
\r
763 OnigDistance anchor_dmax; /* (SEMI_)END_BUF anchor distance */
\r
764 int sub_anchor; /* start-anchor for exact or map */
\r
765 unsigned char *exact;
\r
766 unsigned char *exact_end;
\r
767 unsigned char map[ONIG_CHAR_TABLE_SIZE]; /* used as BM skip or char-map */
\r
768 int *int_map; /* BM skip for exact_len > 255 */
\r
769 int *int_map_backward; /* BM skip for backward search */
\r
770 OnigDistance dmin; /* min-distance of exact or map */
\r
771 OnigDistance dmax; /* max-distance of exact or map */
\r
773 /* regex_t link chain */
\r
774 struct re_pattern_buffer* chain; /* escape compile-conflict */
\r
779 int num_of_elements;
\r
780 OnigEncoding pattern_enc;
\r
781 OnigEncoding target_enc;
\r
782 OnigSyntaxType* syntax;
\r
783 OnigOptionType option;
\r
784 OnigAmbigType ambig_flag;
\r
787 /* Oniguruma Native API */
\r
789 int onig_init P_((void));
\r
791 int onig_error_code_to_str PV_((OnigUChar* s, int err_code, ...));
\r
793 void onig_set_warn_func P_((OnigWarnFunc f));
\r
795 void onig_set_verb_warn_func P_((OnigWarnFunc f));
\r
797 int onig_new P_((regex_t**, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax, OnigErrorInfo* einfo));
\r
799 int onig_new_deluxe P_((regex_t** reg, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigCompileInfo* ci, OnigErrorInfo* einfo));
\r
801 void onig_free P_((regex_t*));
\r
803 int onig_recompile P_((regex_t*, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax, OnigErrorInfo* einfo));
\r
805 int onig_recompile_deluxe P_((regex_t* reg, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigCompileInfo* ci, OnigErrorInfo* einfo));
\r
807 int onig_search P_((regex_t*, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegion* region, OnigOptionType option));
\r
809 int onig_match P_((regex_t*, const OnigUChar* str, const OnigUChar* end, const OnigUChar* at, OnigRegion* region, OnigOptionType option));
\r
811 OnigRegion* onig_region_new P_((void));
\r
813 void onig_region_init P_((OnigRegion* region));
\r
815 void onig_region_free P_((OnigRegion* region, int free_self));
\r
817 void onig_region_copy P_((OnigRegion* to, OnigRegion* from));
\r
819 void onig_region_clear P_((OnigRegion* region));
\r
821 int onig_region_resize P_((OnigRegion* region, int n));
\r
823 int onig_region_set P_((OnigRegion* region, int at, int beg, int end));
\r
825 int onig_name_to_group_numbers P_((regex_t* reg, const OnigUChar* name, const OnigUChar* name_end, int** nums));
\r
827 int onig_name_to_backref_number P_((regex_t* reg, const OnigUChar* name, const OnigUChar* name_end, OnigRegion *region));
\r
829 int onig_foreach_name P_((regex_t* reg, int (*func)(const OnigUChar*, const OnigUChar*,int,int*,regex_t*,void*), void* arg));
\r
831 int onig_number_of_names P_((regex_t* reg));
\r
833 int onig_number_of_captures P_((regex_t* reg));
\r
835 int onig_number_of_capture_histories P_((regex_t* reg));
\r
837 OnigCaptureTreeNode* onig_get_capture_tree P_((OnigRegion* region));
\r
839 int onig_capture_tree_traverse P_((OnigRegion* region, int at, int(*callback_func)(int,int,int,int,int,void*), void* arg));
\r
841 OnigEncoding onig_get_encoding P_((regex_t* reg));
\r
843 OnigOptionType onig_get_options P_((regex_t* reg));
\r
845 OnigAmbigType onig_get_ambig_flag P_((regex_t* reg));
\r
847 OnigSyntaxType* onig_get_syntax P_((regex_t* reg));
\r
849 int onig_set_default_syntax P_((OnigSyntaxType* syntax));
\r
851 void onig_copy_syntax P_((OnigSyntaxType* to, OnigSyntaxType* from));
\r
853 unsigned int onig_get_syntax_op P_((OnigSyntaxType* syntax));
\r
855 unsigned int onig_get_syntax_op2 P_((OnigSyntaxType* syntax));
\r
857 unsigned int onig_get_syntax_behavior P_((OnigSyntaxType* syntax));
\r
859 OnigOptionType onig_get_syntax_options P_((OnigSyntaxType* syntax));
\r
861 void onig_set_syntax_op P_((OnigSyntaxType* syntax, unsigned int op));
\r
863 void onig_set_syntax_op2 P_((OnigSyntaxType* syntax, unsigned int op2));
\r
865 void onig_set_syntax_behavior P_((OnigSyntaxType* syntax, unsigned int behavior));
\r
867 void onig_set_syntax_options P_((OnigSyntaxType* syntax, OnigOptionType options));
\r
869 int onig_set_meta_char P_((OnigEncoding enc, unsigned int what, OnigCodePoint code));
\r
871 void onig_copy_encoding P_((OnigEncoding to, OnigEncoding from));
\r
873 OnigAmbigType onig_get_default_ambig_flag P_(());
\r
875 int onig_set_default_ambig_flag P_((OnigAmbigType ambig_flag));
\r
877 unsigned int onig_get_match_stack_limit_size P_((void));
\r
879 int onig_set_match_stack_limit_size P_((unsigned int size));
\r
881 int onig_end P_((void));
\r
883 const char* onig_version P_((void));
\r
885 const char* onig_copyright P_((void));
\r
891 #endif /* ONIGURUMA_H */
\r