40 #define ONIGMO_VERSION_MAJOR 6
41 #define ONIGMO_VERSION_MINOR 1
42 #define ONIGMO_VERSION_TEENY 3
46 # define ONIG_EXTERN RUBY_EXTERN
48 # if defined(_WIN32) && !defined(__GNUC__)
49 # if defined(EXPORT) || defined(RUBY_EXPORT)
50 # define ONIG_EXTERN extern __declspec(dllexport)
52 # define ONIG_EXTERN extern __declspec(dllimport)
59 # define ONIG_EXTERN extern
63 # ifndef RUBY_SYMBOL_EXPORT_BEGIN
64 # define RUBY_SYMBOL_EXPORT_BEGIN
65 # define RUBY_SYMBOL_EXPORT_END
69 RUBY_SYMBOL_EXPORT_BEGIN
75 #ifndef ONIG_ESCAPE_UCHAR_COLLISION
76 # define UChar OnigUChar
79 typedef unsigned char OnigUChar;
80 typedef unsigned int OnigCodePoint;
81 typedef unsigned int OnigCtype;
82 typedef size_t OnigDistance;
83 typedef ptrdiff_t OnigPosition;
85 #define ONIG_INFINITE_DISTANCE ~((OnigDistance )0)
95 typedef unsigned int OnigCaseFoldType;
97 ONIG_EXTERN OnigCaseFoldType OnigDefaultCaseFoldFlag;
100 #define OnigCodePointMaskWidth 3
101 #define OnigCodePointMask ((1<<OnigCodePointMaskWidth)-1)
102 #define OnigCodePointCount(n) ((n)&OnigCodePointMask)
103 #define OnigCaseFoldFlags(n) ((n)&~OnigCodePointMask)
110 #define OnigSpecialIndexShift 3
111 #define OnigSpecialIndexWidth 10
113 #define ONIGENC_CASE_UPCASE (1<<13)
114 #define ONIGENC_CASE_DOWNCASE (1<<14)
115 #define ONIGENC_CASE_TITLECASE (1<<15)
116 #define ONIGENC_CASE_SPECIAL_OFFSET 3
117 #define ONIGENC_CASE_UP_SPECIAL (1<<16)
118 #define ONIGENC_CASE_DOWN_SPECIAL (1<<17)
119 #define ONIGENC_CASE_MODIFIED (1<<18)
120 #define ONIGENC_CASE_FOLD (1<<19)
122 #define ONIGENC_CASE_FOLD_TURKISH_AZERI (1<<20)
124 #define ONIGENC_CASE_FOLD_LITHUANIAN (1<<21)
125 #define ONIGENC_CASE_ASCII_ONLY (1<<22)
126 #define ONIGENC_CASE_IS_TITLECASE (1<<23)
128 #define INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR (1<<30)
130 #define ONIGENC_CASE_FOLD_MIN INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR
131 #define ONIGENC_CASE_FOLD_DEFAULT OnigDefaultCaseFoldFlag
134 #define ONIGENC_MAX_COMP_CASE_FOLD_CODE_LEN 3
135 #define ONIGENC_GET_CASE_FOLD_CODES_MAX_NUM 13
139 #define ONIGENC_CODE_RANGE_NUM(range) ((int )range[0])
140 #define ONIGENC_CODE_RANGE_FROM(range,i) range[((i)*2) + 1]
141 #define ONIGENC_CODE_RANGE_TO(range,i) range[((i)*2) + 2]
146 OnigCodePoint code[ONIGENC_MAX_COMP_CASE_FOLD_CODE_LEN];
151 OnigCodePoint anychar;
152 OnigCodePoint anytime;
153 OnigCodePoint zero_or_one_time;
154 OnigCodePoint one_or_more_time;
155 OnigCodePoint anychar_anytime;
158 typedef int (*OnigApplyAllCaseFoldFunc)(OnigCodePoint from, OnigCodePoint* to,
int to_len,
void* arg);
161 int (*precise_mbc_enc_len)(
const OnigUChar* p,
const OnigUChar* e,
const struct OnigEncodingTypeST* enc);
165 int (*is_mbc_newline)(
const OnigUChar* p,
const OnigUChar* end,
const struct OnigEncodingTypeST* enc);
166 OnigCodePoint (*mbc_to_code)(
const OnigUChar* p,
const OnigUChar* end,
const struct OnigEncodingTypeST* enc);
168 int (*code_to_mbc)(OnigCodePoint code, OnigUChar *buf,
const struct OnigEncodingTypeST* enc);
169 int (*mbc_case_fold)(OnigCaseFoldType flag,
const OnigUChar** pp,
const OnigUChar* end, OnigUChar* to,
const struct OnigEncodingTypeST* enc);
170 int (*apply_all_case_fold)(OnigCaseFoldType flag, OnigApplyAllCaseFoldFunc f,
void* arg,
const struct OnigEncodingTypeST* enc);
172 int (*property_name_to_ctype)(
const struct OnigEncodingTypeST* enc,
const OnigUChar* p,
const OnigUChar* end);
173 int (*is_code_ctype)(OnigCodePoint code, OnigCtype ctype,
const struct OnigEncodingTypeST* enc);
174 int (*get_ctype_code_range)(OnigCtype ctype, OnigCodePoint* sb_out,
const OnigCodePoint* ranges[],
const struct OnigEncodingTypeST* enc);
175 OnigUChar* (*left_adjust_char_head)(
const OnigUChar* start,
const OnigUChar* p,
const OnigUChar* end,
const struct OnigEncodingTypeST* enc);
176 int (*is_allowed_reverse_match)(
const OnigUChar* p,
const OnigUChar* end,
const struct OnigEncodingTypeST* enc);
177 int (*case_map)(OnigCaseFoldType* flagP,
const OnigUChar** pp,
const OnigUChar* end, OnigUChar* to, OnigUChar* to_end,
const struct OnigEncodingTypeST* enc);
178 int ruby_encoding_index;
225 #define ONIG_ENCODING_ASCII (&OnigEncodingASCII)
227 # define ONIG_ENCODING_ISO_8859_1 (&OnigEncodingISO_8859_1)
228 # define ONIG_ENCODING_ISO_8859_2 (&OnigEncodingISO_8859_2)
229 # define ONIG_ENCODING_ISO_8859_3 (&OnigEncodingISO_8859_3)
230 # define ONIG_ENCODING_ISO_8859_4 (&OnigEncodingISO_8859_4)
231 # define ONIG_ENCODING_ISO_8859_5 (&OnigEncodingISO_8859_5)
232 # define ONIG_ENCODING_ISO_8859_6 (&OnigEncodingISO_8859_6)
233 # define ONIG_ENCODING_ISO_8859_7 (&OnigEncodingISO_8859_7)
234 # define ONIG_ENCODING_ISO_8859_8 (&OnigEncodingISO_8859_8)
235 # define ONIG_ENCODING_ISO_8859_9 (&OnigEncodingISO_8859_9)
236 # define ONIG_ENCODING_ISO_8859_10 (&OnigEncodingISO_8859_10)
237 # define ONIG_ENCODING_ISO_8859_11 (&OnigEncodingISO_8859_11)
238 # define ONIG_ENCODING_ISO_8859_13 (&OnigEncodingISO_8859_13)
239 # define ONIG_ENCODING_ISO_8859_14 (&OnigEncodingISO_8859_14)
240 # define ONIG_ENCODING_ISO_8859_15 (&OnigEncodingISO_8859_15)
241 # define ONIG_ENCODING_ISO_8859_16 (&OnigEncodingISO_8859_16)
242 # define ONIG_ENCODING_UTF_8 (&OnigEncodingUTF_8)
243 # define ONIG_ENCODING_UTF_16BE (&OnigEncodingUTF_16BE)
244 # define ONIG_ENCODING_UTF_16LE (&OnigEncodingUTF_16LE)
245 # define ONIG_ENCODING_UTF_32BE (&OnigEncodingUTF_32BE)
246 # define ONIG_ENCODING_UTF_32LE (&OnigEncodingUTF_32LE)
247 # define ONIG_ENCODING_EUC_JP (&OnigEncodingEUC_JP)
248 # define ONIG_ENCODING_EUC_TW (&OnigEncodingEUC_TW)
249 # define ONIG_ENCODING_EUC_KR (&OnigEncodingEUC_KR)
250 # define ONIG_ENCODING_EUC_CN (&OnigEncodingEUC_CN)
251 # define ONIG_ENCODING_SHIFT_JIS (&OnigEncodingShift_JIS)
252 # define ONIG_ENCODING_WINDOWS_31J (&OnigEncodingWindows_31J)
254 # define ONIG_ENCODING_KOI8_R (&OnigEncodingKOI8_R)
255 # define ONIG_ENCODING_KOI8_U (&OnigEncodingKOI8_U)
256 # define ONIG_ENCODING_WINDOWS_1250 (&OnigEncodingWindows_1250)
257 # define ONIG_ENCODING_WINDOWS_1251 (&OnigEncodingWindows_1251)
258 # define ONIG_ENCODING_WINDOWS_1252 (&OnigEncodingWindows_1252)
259 # define ONIG_ENCODING_WINDOWS_1253 (&OnigEncodingWindows_1253)
260 # define ONIG_ENCODING_WINDOWS_1254 (&OnigEncodingWindows_1254)
261 # define ONIG_ENCODING_WINDOWS_1257 (&OnigEncodingWindows_1257)
262 # define ONIG_ENCODING_BIG5 (&OnigEncodingBIG5)
263 # define ONIG_ENCODING_GB18030 (&OnigEncodingGB18030)
266 # define ONIG_ENCODING_SJIS ONIG_ENCODING_SHIFT_JIS
267 # define ONIG_ENCODING_CP932 ONIG_ENCODING_WINDOWS_31J
268 # define ONIG_ENCODING_CP1250 ONIG_ENCODING_WINDOWS_1250
269 # define ONIG_ENCODING_CP1251 ONIG_ENCODING_WINDOWS_1251
270 # define ONIG_ENCODING_CP1252 ONIG_ENCODING_WINDOWS_1252
271 # define ONIG_ENCODING_CP1253 ONIG_ENCODING_WINDOWS_1253
272 # define ONIG_ENCODING_CP1254 ONIG_ENCODING_WINDOWS_1254
273 # define ONIG_ENCODING_CP1257 ONIG_ENCODING_WINDOWS_1257
274 # define ONIG_ENCODING_UTF8 ONIG_ENCODING_UTF_8
275 # define ONIG_ENCODING_UTF16_BE ONIG_ENCODING_UTF_16BE
276 # define ONIG_ENCODING_UTF16_LE ONIG_ENCODING_UTF_16LE
277 # define ONIG_ENCODING_UTF32_BE ONIG_ENCODING_UTF_32BE
278 # define ONIG_ENCODING_UTF32_LE ONIG_ENCODING_UTF_32LE
281 #define ONIG_ENCODING_UNDEF ((OnigEncoding )0)
285 int onigenc_ascii_only_case_map(OnigCaseFoldType* flagP,
const OnigUChar** pp,
const OnigUChar* end, OnigUChar* to, OnigUChar* to_end,
const struct OnigEncodingTypeST* enc);
289 #define ONIGENC_CODE_TO_MBC_MAXLEN 7
290 #define ONIGENC_MBC_CASE_FOLD_MAXLEN 18
294 #define ONIGENC_CTYPE_NEWLINE 0
295 #define ONIGENC_CTYPE_ALPHA 1
296 #define ONIGENC_CTYPE_BLANK 2
297 #define ONIGENC_CTYPE_CNTRL 3
298 #define ONIGENC_CTYPE_DIGIT 4
299 #define ONIGENC_CTYPE_GRAPH 5
300 #define ONIGENC_CTYPE_LOWER 6
301 #define ONIGENC_CTYPE_PRINT 7
302 #define ONIGENC_CTYPE_PUNCT 8
303 #define ONIGENC_CTYPE_SPACE 9
304 #define ONIGENC_CTYPE_UPPER 10
305 #define ONIGENC_CTYPE_XDIGIT 11
306 #define ONIGENC_CTYPE_WORD 12
307 #define ONIGENC_CTYPE_ALNUM 13
308 #define ONIGENC_CTYPE_ASCII 14
309 #define ONIGENC_MAX_STD_CTYPE ONIGENC_CTYPE_ASCII
312 #define ONIGENC_FLAG_NONE 0U
313 #define ONIGENC_FLAG_UNICODE 1U
315 #define onig_enc_len(enc,p,e) ONIGENC_MBC_ENC_LEN(enc, p, e)
317 #define ONIGENC_IS_UNDEF(enc) ((enc) == ONIG_ENCODING_UNDEF)
318 #define ONIGENC_IS_SINGLEBYTE(enc) (ONIGENC_MBC_MAXLEN(enc) == 1)
319 #define ONIGENC_IS_MBC_HEAD(enc,p,e) (ONIGENC_MBC_ENC_LEN(enc,p,e) != 1)
320 #define ONIGENC_IS_MBC_ASCII(p) (*(p) < 128)
321 #define ONIGENC_IS_CODE_ASCII(code) ((code) < 128)
322 #define ONIGENC_IS_MBC_WORD(enc,s,end) \
323 ONIGENC_IS_CODE_WORD(enc,ONIGENC_MBC_TO_CODE(enc,s,end))
324 #define ONIGENC_IS_MBC_ASCII_WORD(enc,s,end) \
325 onigenc_ascii_is_code_ctype( \
326 ONIGENC_MBC_TO_CODE(enc,s,end),ONIGENC_CTYPE_WORD,enc)
327 #define ONIGENC_IS_UNICODE(enc) ((enc)->flags & ONIGENC_FLAG_UNICODE)
330 #define ONIGENC_NAME(enc) ((enc)->name)
332 #define ONIGENC_MBC_CASE_FOLD(enc,flag,pp,end,buf) \
333 (enc)->mbc_case_fold(flag,(const OnigUChar** )pp,end,buf,enc)
334 #define ONIGENC_IS_ALLOWED_REVERSE_MATCH(enc,s,end) \
335 (enc)->is_allowed_reverse_match(s,end,enc)
336 #define ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc,start,s,end) \
337 (enc)->left_adjust_char_head(start, s, end, enc)
338 #define ONIGENC_APPLY_ALL_CASE_FOLD(enc,case_fold_flag,f,arg) \
339 (enc)->apply_all_case_fold(case_fold_flag,f,arg,enc)
340 #define ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc,case_fold_flag,p,end,acs) \
341 (enc)->get_case_fold_codes_by_str(case_fold_flag,p,end,acs,enc)
342 #define ONIGENC_STEP_BACK(enc,start,s,end,n) \
343 onigenc_step_back((enc),(start),(s),(end),(n))
345 #define ONIGENC_CONSTRUCT_MBCLEN_CHARFOUND(n) (n)
346 #define ONIGENC_MBCLEN_CHARFOUND_P(r) (0 < (r))
347 #define ONIGENC_MBCLEN_CHARFOUND_LEN(r) (r)
349 #define ONIGENC_CONSTRUCT_MBCLEN_INVALID() (-1)
350 #define ONIGENC_MBCLEN_INVALID_P(r) ((r) == -1)
352 #define ONIGENC_CONSTRUCT_MBCLEN_NEEDMORE(n) (-1-(n))
353 #define ONIGENC_MBCLEN_NEEDMORE_P(r) ((r) < -1)
354 #define ONIGENC_MBCLEN_NEEDMORE_LEN(r) (-1-(r))
356 #define ONIGENC_PRECISE_MBC_ENC_LEN(enc,p,e) (enc)->precise_mbc_enc_len(p,e,enc)
359 int onigenc_mbclen(
const OnigUChar* p,
const OnigUChar* e,
const struct OnigEncodingTypeST* enc);
361 #define ONIGENC_MBC_ENC_LEN(enc,p,e) onigenc_mbclen(p,e,enc)
362 #define ONIGENC_MBC_MAXLEN(enc) ((enc)->max_enc_len)
363 #define ONIGENC_MBC_MAXLEN_DIST(enc) ONIGENC_MBC_MAXLEN(enc)
364 #define ONIGENC_MBC_MINLEN(enc) ((enc)->min_enc_len)
365 #define ONIGENC_IS_MBC_NEWLINE(enc,p,end) (enc)->is_mbc_newline((p),(end),enc)
366 #define ONIGENC_MBC_TO_CODE(enc,p,end) (enc)->mbc_to_code((p),(end),enc)
367 #define ONIGENC_CODE_TO_MBCLEN(enc,code) (enc)->code_to_mbclen(code,enc)
368 #define ONIGENC_CODE_TO_MBC(enc,code,buf) (enc)->code_to_mbc(code,buf,enc)
369 #define ONIGENC_PROPERTY_NAME_TO_CTYPE(enc,p,end) \
370 (enc)->property_name_to_ctype(enc,p,end)
372 #define ONIGENC_IS_CODE_CTYPE(enc,code,ctype) (enc)->is_code_ctype(code,ctype,enc)
374 #define ONIGENC_IS_CODE_NEWLINE(enc,code) \
375 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_NEWLINE)
376 #define ONIGENC_IS_CODE_GRAPH(enc,code) \
377 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_GRAPH)
378 #define ONIGENC_IS_CODE_PRINT(enc,code) \
379 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PRINT)
380 #define ONIGENC_IS_CODE_ALNUM(enc,code) \
381 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALNUM)
382 #define ONIGENC_IS_CODE_ALPHA(enc,code) \
383 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALPHA)
384 #define ONIGENC_IS_CODE_LOWER(enc,code) \
385 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_LOWER)
386 #define ONIGENC_IS_CODE_UPPER(enc,code) \
387 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_UPPER)
388 #define ONIGENC_IS_CODE_CNTRL(enc,code) \
389 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_CNTRL)
390 #define ONIGENC_IS_CODE_PUNCT(enc,code) \
391 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PUNCT)
392 #define ONIGENC_IS_CODE_SPACE(enc,code) \
393 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_SPACE)
394 #define ONIGENC_IS_CODE_BLANK(enc,code) \
395 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_BLANK)
396 #define ONIGENC_IS_CODE_DIGIT(enc,code) \
397 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_DIGIT)
398 #define ONIGENC_IS_CODE_XDIGIT(enc,code) \
399 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_XDIGIT)
400 #define ONIGENC_IS_CODE_WORD(enc,code) \
401 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_WORD)
403 #define ONIGENC_GET_CTYPE_CODE_RANGE(enc,ctype,sbout,ranges) \
404 (enc)->get_ctype_code_range(ctype,sbout,ranges,enc)
407 OnigUChar* onigenc_step_back(
OnigEncoding enc,
const OnigUChar* start,
const OnigUChar* s,
const OnigUChar* end,
int n);
412 int onigenc_init(
void);
418 OnigUChar* onigenc_get_right_adjust_char_head_with_prev(
OnigEncoding enc,
const OnigUChar* start,
const OnigUChar* s,
const OnigUChar* end,
const OnigUChar** prev);
420 OnigUChar* onigenc_get_prev_char_head(
OnigEncoding enc,
const OnigUChar* start,
const OnigUChar* s,
const OnigUChar* end);
422 OnigUChar* onigenc_get_left_adjust_char_head(
OnigEncoding enc,
const OnigUChar* start,
const OnigUChar* s,
const OnigUChar* end);
424 OnigUChar* onigenc_get_right_adjust_char_head(
OnigEncoding enc,
const OnigUChar* start,
const OnigUChar* s,
const OnigUChar* end);
426 int onigenc_strlen(
OnigEncoding enc,
const OnigUChar* p,
const OnigUChar* end);
428 int onigenc_strlen_null(
OnigEncoding enc,
const OnigUChar* p);
430 int onigenc_str_bytelen_null(
OnigEncoding enc,
const OnigUChar* p);
437 #define ONIG_NREGION 4
438 #define ONIG_MAX_CAPTURE_GROUP_NUM 32767
439 #define ONIG_MAX_BACKREF_NUM 1000
440 #define ONIG_MAX_REPEAT_NUM 100000
441 #define ONIG_MAX_MULTI_BYTE_RANGES_NUM 10000
443 #define ONIG_MAX_ERROR_MESSAGE_LEN 90
445 typedef unsigned int OnigOptionType;
447 #define ONIG_OPTION_DEFAULT ONIG_OPTION_NONE
450 #define ONIG_OPTION_NONE 0U
451 #define ONIG_OPTION_IGNORECASE 1U
452 #define ONIG_OPTION_EXTEND (ONIG_OPTION_IGNORECASE << 1)
453 #define ONIG_OPTION_MULTILINE (ONIG_OPTION_EXTEND << 1)
454 #define ONIG_OPTION_DOTALL ONIG_OPTION_MULTILINE
455 #define ONIG_OPTION_SINGLELINE (ONIG_OPTION_MULTILINE << 1)
456 #define ONIG_OPTION_FIND_LONGEST (ONIG_OPTION_SINGLELINE << 1)
457 #define ONIG_OPTION_FIND_NOT_EMPTY (ONIG_OPTION_FIND_LONGEST << 1)
458 #define ONIG_OPTION_NEGATE_SINGLELINE (ONIG_OPTION_FIND_NOT_EMPTY << 1)
459 #define ONIG_OPTION_DONT_CAPTURE_GROUP (ONIG_OPTION_NEGATE_SINGLELINE << 1)
460 #define ONIG_OPTION_CAPTURE_GROUP (ONIG_OPTION_DONT_CAPTURE_GROUP << 1)
462 #define ONIG_OPTION_NOTBOL (ONIG_OPTION_CAPTURE_GROUP << 1)
463 #define ONIG_OPTION_NOTEOL (ONIG_OPTION_NOTBOL << 1)
464 #define ONIG_OPTION_NOTBOS (ONIG_OPTION_NOTEOL << 1)
465 #define ONIG_OPTION_NOTEOS (ONIG_OPTION_NOTBOS << 1)
467 #define ONIG_OPTION_ASCII_RANGE (ONIG_OPTION_NOTEOS << 1)
468 #define ONIG_OPTION_POSIX_BRACKET_ALL_RANGE (ONIG_OPTION_ASCII_RANGE << 1)
469 #define ONIG_OPTION_WORD_BOUND_ALL_RANGE (ONIG_OPTION_POSIX_BRACKET_ALL_RANGE << 1)
471 #define ONIG_OPTION_NEWLINE_CRLF (ONIG_OPTION_WORD_BOUND_ALL_RANGE << 1)
472 #define ONIG_OPTION_MAXBIT ONIG_OPTION_NEWLINE_CRLF
474 #define ONIG_OPTION_ON(options,regopt) ((options) |= (regopt))
475 #define ONIG_OPTION_OFF(options,regopt) ((options) &= ~(regopt))
476 #define ONIG_IS_OPTION_ON(options,option) ((options) & (option))
482 unsigned int behavior;
483 OnigOptionType options;
501 #define ONIG_SYNTAX_ASIS (&OnigSyntaxASIS)
502 #define ONIG_SYNTAX_POSIX_BASIC (&OnigSyntaxPosixBasic)
503 #define ONIG_SYNTAX_POSIX_EXTENDED (&OnigSyntaxPosixExtended)
504 #define ONIG_SYNTAX_EMACS (&OnigSyntaxEmacs)
505 #define ONIG_SYNTAX_GREP (&OnigSyntaxGrep)
506 #define ONIG_SYNTAX_GNU_REGEX (&OnigSyntaxGnuRegex)
507 #define ONIG_SYNTAX_JAVA (&OnigSyntaxJava)
508 #define ONIG_SYNTAX_PERL58 (&OnigSyntaxPerl58)
509 #define ONIG_SYNTAX_PERL58_NG (&OnigSyntaxPerl58_NG)
510 #define ONIG_SYNTAX_PERL (&OnigSyntaxPerl)
511 #define ONIG_SYNTAX_RUBY (&OnigSyntaxRuby)
512 #define ONIG_SYNTAX_PYTHON (&OnigSyntaxPython)
516 #define ONIG_SYNTAX_DEFAULT OnigDefaultSyntax
519 #define ONIG_SYN_OP_VARIABLE_META_CHARACTERS (1U<<0)
520 #define ONIG_SYN_OP_DOT_ANYCHAR (1U<<1)
521 #define ONIG_SYN_OP_ASTERISK_ZERO_INF (1U<<2)
522 #define ONIG_SYN_OP_ESC_ASTERISK_ZERO_INF (1U<<3)
523 #define ONIG_SYN_OP_PLUS_ONE_INF (1U<<4)
524 #define ONIG_SYN_OP_ESC_PLUS_ONE_INF (1U<<5)
525 #define ONIG_SYN_OP_QMARK_ZERO_ONE (1U<<6)
526 #define ONIG_SYN_OP_ESC_QMARK_ZERO_ONE (1U<<7)
527 #define ONIG_SYN_OP_BRACE_INTERVAL (1U<<8)
528 #define ONIG_SYN_OP_ESC_BRACE_INTERVAL (1U<<9)
529 #define ONIG_SYN_OP_VBAR_ALT (1U<<10)
530 #define ONIG_SYN_OP_ESC_VBAR_ALT (1U<<11)
531 #define ONIG_SYN_OP_LPAREN_SUBEXP (1U<<12)
532 #define ONIG_SYN_OP_ESC_LPAREN_SUBEXP (1U<<13)
533 #define ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR (1U<<14)
534 #define ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR (1U<<15)
535 #define ONIG_SYN_OP_DECIMAL_BACKREF (1U<<16)
536 #define ONIG_SYN_OP_BRACKET_CC (1U<<17)
537 #define ONIG_SYN_OP_ESC_W_WORD (1U<<18)
538 #define ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END (1U<<19)
539 #define ONIG_SYN_OP_ESC_B_WORD_BOUND (1U<<20)
540 #define ONIG_SYN_OP_ESC_S_WHITE_SPACE (1U<<21)
541 #define ONIG_SYN_OP_ESC_D_DIGIT (1U<<22)
542 #define ONIG_SYN_OP_LINE_ANCHOR (1U<<23)
543 #define ONIG_SYN_OP_POSIX_BRACKET (1U<<24)
544 #define ONIG_SYN_OP_QMARK_NON_GREEDY (1U<<25)
545 #define ONIG_SYN_OP_ESC_CONTROL_CHARS (1U<<26)
546 #define ONIG_SYN_OP_ESC_C_CONTROL (1U<<27)
547 #define ONIG_SYN_OP_ESC_OCTAL3 (1U<<28)
548 #define ONIG_SYN_OP_ESC_X_HEX2 (1U<<29)
549 #define ONIG_SYN_OP_ESC_X_BRACE_HEX8 (1U<<30)
550 #define ONIG_SYN_OP_ESC_O_BRACE_OCTAL (1U<<31)
552 #define ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE (1U<<0)
553 #define ONIG_SYN_OP2_QMARK_GROUP_EFFECT (1U<<1)
554 #define ONIG_SYN_OP2_OPTION_PERL (1U<<2)
555 #define ONIG_SYN_OP2_OPTION_RUBY (1U<<3)
556 #define ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT (1U<<4)
557 #define ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL (1U<<5)
558 #define ONIG_SYN_OP2_CCLASS_SET_OP (1U<<6)
559 #define ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP (1U<<7)
560 #define ONIG_SYN_OP2_ESC_K_NAMED_BACKREF (1U<<8)
561 #define ONIG_SYN_OP2_ESC_G_SUBEXP_CALL (1U<<9)
562 #define ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY (1U<<10)
563 #define ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL (1U<<11)
564 #define ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META (1U<<12)
565 #define ONIG_SYN_OP2_ESC_V_VTAB (1U<<13)
566 #define ONIG_SYN_OP2_ESC_U_HEX4 (1U<<14)
567 #define ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR (1U<<15)
568 #define ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY (1U<<16)
569 #define ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT (1U<<17)
571 #define ONIG_SYN_OP2_ESC_H_XDIGIT (1U<<19)
572 #define ONIG_SYN_OP2_INEFFECTIVE_ESCAPE (1U<<20)
573 #define ONIG_SYN_OP2_ESC_CAPITAL_R_LINEBREAK (1U<<21)
574 #define ONIG_SYN_OP2_ESC_CAPITAL_X_EXTENDED_GRAPHEME_CLUSTER (1U<<22)
575 #define ONIG_SYN_OP2_ESC_V_VERTICAL_WHITESPACE (1U<<23)
576 #define ONIG_SYN_OP2_ESC_H_HORIZONTAL_WHITESPACE (1U<<24)
577 #define ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP (1U<<25)
578 #define ONIG_SYN_OP2_ESC_G_BRACE_BACKREF (1U<<26)
579 #define ONIG_SYN_OP2_QMARK_SUBEXP_CALL (1U<<27)
580 #define ONIG_SYN_OP2_QMARK_VBAR_BRANCH_RESET (1U<<28)
581 #define ONIG_SYN_OP2_QMARK_LPAREN_CONDITION (1U<<29)
582 #define ONIG_SYN_OP2_QMARK_CAPITAL_P_NAMED_GROUP (1U<<30)
583 #define ONIG_SYN_OP2_QMARK_TILDE_ABSENT (1U<<31)
587 #define ONIG_SYN_CONTEXT_INDEP_ANCHORS (1U<<31)
588 #define ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS (1U<<0)
589 #define ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS (1U<<1)
590 #define ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP (1U<<2)
591 #define ONIG_SYN_ALLOW_INVALID_INTERVAL (1U<<3)
592 #define ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV (1U<<4)
593 #define ONIG_SYN_STRICT_CHECK_BACKREF (1U<<5)
594 #define ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND (1U<<6)
595 #define ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP (1U<<7)
596 #define ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME (1U<<8)
597 #define ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY (1U<<9)
598 #define ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME_CALL (1U<<10)
599 #define ONIG_SYN_USE_LEFT_MOST_NAMED_GROUP (1U<<11)
602 #define ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC (1U<<20)
603 #define ONIG_SYN_BACKSLASH_ESCAPE_IN_CC (1U<<21)
604 #define ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC (1U<<22)
605 #define ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC (1U<<23)
607 #define ONIG_SYN_WARN_CC_OP_NOT_ESCAPED (1U<<24)
608 #define ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT (1U<<25)
609 #define ONIG_SYN_WARN_CC_DUP (1U<<26)
612 #define ONIG_META_CHAR_ESCAPE 0
613 #define ONIG_META_CHAR_ANYCHAR 1
614 #define ONIG_META_CHAR_ANYTIME 2
615 #define ONIG_META_CHAR_ZERO_OR_ONE_TIME 3
616 #define ONIG_META_CHAR_ONE_OR_MORE_TIME 4
617 #define ONIG_META_CHAR_ANYCHAR_ANYTIME 5
619 #define ONIG_INEFFECTIVE_META_CHAR 0
622 #define ONIG_IS_PATTERN_ERROR(ecode) ((ecode) <= -100 && (ecode) > -1000)
624 #define ONIG_NORMAL 0
625 #define ONIG_MISMATCH -1
626 #define ONIG_NO_SUPPORT_CONFIG -2
629 #define ONIGERR_MEMORY -5
630 #define ONIGERR_TYPE_BUG -6
631 #define ONIGERR_PARSER_BUG -11
632 #define ONIGERR_STACK_BUG -12
633 #define ONIGERR_UNDEFINED_BYTECODE -13
634 #define ONIGERR_UNEXPECTED_BYTECODE -14
635 #define ONIGERR_MATCH_STACK_LIMIT_OVER -15
636 #define ONIGERR_PARSE_DEPTH_LIMIT_OVER -16
637 #define ONIGERR_DEFAULT_ENCODING_IS_NOT_SET -21
638 #define ONIGERR_SPECIFIED_ENCODING_CANT_CONVERT_TO_WIDE_CHAR -22
639 #define ONIGERR_TIMEOUT -23
641 #define ONIGERR_INVALID_ARGUMENT -30
643 #define ONIGERR_END_PATTERN_AT_LEFT_BRACE -100
644 #define ONIGERR_END_PATTERN_AT_LEFT_BRACKET -101
645 #define ONIGERR_EMPTY_CHAR_CLASS -102
646 #define ONIGERR_PREMATURE_END_OF_CHAR_CLASS -103
647 #define ONIGERR_END_PATTERN_AT_ESCAPE -104
648 #define ONIGERR_END_PATTERN_AT_META -105
649 #define ONIGERR_END_PATTERN_AT_CONTROL -106
650 #define ONIGERR_META_CODE_SYNTAX -108
651 #define ONIGERR_CONTROL_CODE_SYNTAX -109
652 #define ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE -110
653 #define ONIGERR_CHAR_CLASS_VALUE_AT_START_OF_RANGE -111
654 #define ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS -112
655 #define ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED -113
656 #define ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID -114
657 #define ONIGERR_NESTED_REPEAT_OPERATOR -115
658 #define ONIGERR_UNMATCHED_CLOSE_PARENTHESIS -116
659 #define ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS -117
660 #define ONIGERR_END_PATTERN_IN_GROUP -118
661 #define ONIGERR_UNDEFINED_GROUP_OPTION -119
662 #define ONIGERR_INVALID_POSIX_BRACKET_TYPE -121
663 #define ONIGERR_INVALID_LOOK_BEHIND_PATTERN -122
664 #define ONIGERR_INVALID_REPEAT_RANGE_PATTERN -123
665 #define ONIGERR_INVALID_CONDITION_PATTERN -124
667 #define ONIGERR_TOO_BIG_NUMBER -200
668 #define ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE -201
669 #define ONIGERR_UPPER_SMALLER_THAN_LOWER_IN_REPEAT_RANGE -202
670 #define ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS -203
671 #define ONIGERR_MISMATCH_CODE_LENGTH_IN_CLASS_RANGE -204
672 #define ONIGERR_TOO_MANY_MULTI_BYTE_RANGES -205
673 #define ONIGERR_TOO_SHORT_MULTI_BYTE_STRING -206
674 #define ONIGERR_TOO_BIG_BACKREF_NUMBER -207
675 #define ONIGERR_INVALID_BACKREF -208
676 #define ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED -209
677 #define ONIGERR_TOO_MANY_CAPTURE_GROUPS -210
678 #define ONIGERR_TOO_SHORT_DIGITS -211
679 #define ONIGERR_TOO_LONG_WIDE_CHAR_VALUE -212
680 #define ONIGERR_EMPTY_GROUP_NAME -214
681 #define ONIGERR_INVALID_GROUP_NAME -215
682 #define ONIGERR_INVALID_CHAR_IN_GROUP_NAME -216
683 #define ONIGERR_UNDEFINED_NAME_REFERENCE -217
684 #define ONIGERR_UNDEFINED_GROUP_REFERENCE -218
685 #define ONIGERR_MULTIPLEX_DEFINED_NAME -219
686 #define ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL -220
687 #define ONIGERR_NEVER_ENDING_RECURSION -221
688 #define ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY -222
689 #define ONIGERR_INVALID_CHAR_PROPERTY_NAME -223
690 #define ONIGERR_INVALID_CODE_POINT_VALUE -400
691 #define ONIGERR_INVALID_WIDE_CHAR_VALUE -400
692 #define ONIGERR_TOO_BIG_WIDE_CHAR_VALUE -401
693 #define ONIGERR_NOT_SUPPORTED_ENCODING_COMBINATION -402
694 #define ONIGERR_INVALID_COMBINATION_OF_OPTIONS -403
701 #define ONIG_MAX_CAPTURE_HISTORY_GROUP 31
702 #define ONIG_IS_CAPTURE_HISTORY_GROUP(r, i) \
703 ((i) <= ONIG_MAX_CAPTURE_HISTORY_GROUP && (r)->list && (r)->list[i])
705 #ifdef USE_CAPTURE_HISTORY
706 typedef struct OnigCaptureTreeNodeStruct {
712 struct OnigCaptureTreeNodeStruct** childs;
713 } OnigCaptureTreeNode;
722 #ifdef USE_CAPTURE_HISTORY
724 OnigCaptureTreeNode* history_root;
729 #define ONIG_TRAVERSE_CALLBACK_AT_FIRST 1
730 #define ONIG_TRAVERSE_CALLBACK_AT_LAST 2
731 #define ONIG_TRAVERSE_CALLBACK_AT_BOTH \
732 ( ONIG_TRAVERSE_CALLBACK_AT_FIRST | ONIG_TRAVERSE_CALLBACK_AT_LAST )
735 #define ONIG_REGION_NOTPOS -1
750 typedef void (*OnigWarnFunc)(
const char* s);
751 extern void onig_null_warn(
const char* s);
752 #define ONIG_NULL_WARN onig_null_warn
754 #define ONIG_CHAR_TABLE_SIZE 256
765 int num_comb_exp_check;
767 unsigned int capture_history;
768 unsigned int bt_mem_start;
769 unsigned int bt_mem_end;
771 int repeat_range_alloc;
773 OnigOptionType options;
780 OnigCaseFoldType case_fold_flag;
786 OnigDistance anchor_dmin;
787 OnigDistance anchor_dmax;
789 unsigned char *exact;
790 unsigned char *exact_end;
791 unsigned char map[ONIG_CHAR_TABLE_SIZE];
793 int *int_map_backward;
798 #ifdef MY_RUBY_BUILD_MAY_TIME_TRAVEL
810 #ifndef ONIG_ESCAPE_REGEX_T_COLLISION
820 OnigOptionType option;
821 OnigCaseFoldType case_fold_flag;
830 int onig_error_code_to_str(OnigUChar* s, OnigPosition err_code, ...);
832 void onig_set_warn_func(OnigWarnFunc f);
834 void onig_set_verb_warn_func(OnigWarnFunc f);
850 OnigPosition onig_scan(
OnigRegex reg,
const OnigUChar* str,
const OnigUChar* end,
OnigRegion* region, OnigOptionType option,
int (*scan_callback)(OnigPosition, OnigPosition,
OnigRegion*,
void*),
void* callback_arg);
852 OnigPosition onig_search(
OnigRegex,
const OnigUChar* str,
const OnigUChar* end,
const OnigUChar* start,
const OnigUChar* range,
OnigRegion* region, OnigOptionType option);
854 OnigPosition onig_search_gpos(
OnigRegex,
const OnigUChar* str,
const OnigUChar* end,
const OnigUChar* global_pos,
const OnigUChar* start,
const OnigUChar* range,
OnigRegion* region, OnigOptionType option);
856 OnigPosition onig_match(
OnigRegex,
const OnigUChar* str,
const OnigUChar* end,
const OnigUChar* at,
OnigRegion* region, OnigOptionType option);
858 int onig_check_linear_time(
OnigRegex reg);
864 void onig_region_free(
OnigRegion* region,
int free_self);
870 int onig_region_resize(
OnigRegion* region,
int n);
872 int onig_region_set(
OnigRegion* region,
int at,
int beg,
int end);
874 int onig_name_to_group_numbers(
OnigRegex reg,
const OnigUChar* name,
const OnigUChar* name_end,
int** nums);
876 int onig_name_to_backref_number(
OnigRegex reg,
const OnigUChar* name,
const OnigUChar* name_end,
const OnigRegion *region);
878 int onig_foreach_name(
OnigRegex reg,
int (*func)(
const OnigUChar*,
const OnigUChar*,
int,
int*,
OnigRegex,
void*),
void* arg);
884 int onig_number_of_capture_histories(
const OnigRegexType *reg);
885 #ifdef USE_CAPTURE_HISTORY
887 OnigCaptureTreeNode* onig_get_capture_tree(
OnigRegion* region);
890 int onig_capture_tree_traverse(
OnigRegion* region,
int at,
int(*callback_func)(
int,OnigPosition,OnigPosition,
int,
int,
void*),
void* arg);
892 int onig_noname_group_capture_is_active(
const OnigRegexType *reg);
898 OnigCaseFoldType onig_get_case_fold_flag(
const OnigRegexType *reg);
910 unsigned int onig_get_syntax_behavior(
const OnigSyntaxType* syntax);
912 OnigOptionType onig_get_syntax_options(
const OnigSyntaxType* syntax);
916 void onig_set_syntax_op2(
OnigSyntaxType* syntax,
unsigned int op2);
918 void onig_set_syntax_behavior(
OnigSyntaxType* syntax,
unsigned int behavior);
920 void onig_set_syntax_options(
OnigSyntaxType* syntax, OnigOptionType options);
922 int onig_set_meta_char(
OnigSyntaxType* syntax,
unsigned int what, OnigCodePoint code);
926 OnigCaseFoldType onig_get_default_case_fold_flag(
void);
928 int onig_set_default_case_fold_flag(OnigCaseFoldType case_fold_flag);
930 unsigned int onig_get_match_stack_limit_size(
void);
932 int onig_set_match_stack_limit_size(
unsigned int size);
934 unsigned int onig_get_parse_depth_limit(
void);
936 int onig_set_parse_depth_limit(
unsigned int depth);
940 const char* onig_version(
void);
942 const char* onig_copyright(
void);
944 RUBY_SYMBOL_EXPORT_END