1 #ifndef ONIGURUMA_H 2 #define ONIGURUMA_H 3 /********************************************************************** 4 oniguruma.h - Oniguruma (regular expression library) 5 **********************************************************************/ 6 /*- 7 * Copyright (c) 2002-2020 K.Kosako 8 * All rights reserved. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 #ifdef __cplusplus 33 extern "C" { 34 #endif 35 36 #define ONIGURUMA 37 #define ONIGURUMA_VERSION_MAJOR 6 38 #define ONIGURUMA_VERSION_MINOR 9 39 #define ONIGURUMA_VERSION_TEENY 4 40 41 #define ONIGURUMA_VERSION_INT 60904 42 43 #ifndef P_ 44 #if defined(__STDC__) || defined(_WIN32) 45 # define P_(args) args 46 #else 47 # define P_(args) () 48 #endif 49 #endif 50 51 #ifndef PV_ 52 # define PV_(args) args 53 #endif 54 55 #ifndef ONIG_STATIC 56 #ifndef ONIG_EXTERN 57 #if defined(_WIN32) && !defined(__GNUC__) 58 #if defined(ONIGURUMA_EXPORT) 59 #define ONIG_EXTERN extern __declspec(dllexport) 60 #else 61 #define ONIG_EXTERN extern __declspec(dllimport) 62 #endif 63 #endif 64 #endif 65 66 #ifndef ONIG_EXTERN 67 #define ONIG_EXTERN extern 68 #endif 69 #else 70 #define ONIG_EXTERN extern 71 #endif 72 73 #ifndef ONIG_VARIADIC_FUNC_ATTR 74 #define ONIG_VARIADIC_FUNC_ATTR 75 #endif 76 77 /* PART: character encoding */ 78 79 #ifndef ONIG_ESCAPE_UCHAR_COLLISION 80 #define UChar OnigUChar 81 #endif 82 83 typedef unsigned int OnigCodePoint; 84 typedef unsigned char OnigUChar; 85 typedef unsigned int OnigCtype; 86 typedef unsigned int OnigLen; 87 88 #define ONIG_INFINITE_DISTANCE ~((OnigLen )0) 89 90 typedef unsigned int OnigCaseFoldType; /* case fold flag */ 91 92 ONIG_EXTERN OnigCaseFoldType OnigDefaultCaseFoldFlag; 93 94 /* #define ONIGENC_CASE_FOLD_HIRAGANA_KATAKANA (1<<1) */ 95 /* #define ONIGENC_CASE_FOLD_KATAKANA_WIDTH (1<<2) */ 96 #define ONIGENC_CASE_FOLD_TURKISH_AZERI (1<<20) 97 #define INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR (1<<30) 98 99 #define ONIGENC_CASE_FOLD_MIN INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR 100 #define ONIGENC_CASE_FOLD_DEFAULT OnigDefaultCaseFoldFlag 101 102 103 #define ONIGENC_MAX_COMP_CASE_FOLD_CODE_LEN 3 104 #define ONIGENC_GET_CASE_FOLD_CODES_MAX_NUM 13 105 /* 13 => Unicode:0x1ffc */ 106 107 /* code range */ 108 #define ONIGENC_CODE_RANGE_NUM(range) ((int )range[0]) 109 #define ONIGENC_CODE_RANGE_FROM(range,i) range[((i)*2) + 1] 110 #define ONIGENC_CODE_RANGE_TO(range,i) range[((i)*2) + 2] 111 112 typedef struct { 113 int byte_len; /* argument(original) character(s) byte length */ 114 int code_len; /* number of code */ 115 OnigCodePoint code[ONIGENC_MAX_COMP_CASE_FOLD_CODE_LEN]; 116 } OnigCaseFoldCodeItem; 117 118 typedef struct { 119 OnigCodePoint esc; 120 OnigCodePoint anychar; 121 OnigCodePoint anytime; 122 OnigCodePoint zero_or_one_time; 123 OnigCodePoint one_or_more_time; 124 OnigCodePoint anychar_anytime; 125 } OnigMetaCharTableType; 126 127 typedef int (*OnigApplyAllCaseFoldFunc)(OnigCodePoint from, OnigCodePoint* to, int to_len, void* arg); 128 129 typedef struct OnigEncodingTypeST { 130 int (*mbc_enc_len)(const OnigUChar* p); 131 const char* name; 132 int max_enc_len; 133 int min_enc_len; 134 int (*is_mbc_newline)(const OnigUChar* p, const OnigUChar* end); 135 OnigCodePoint (*mbc_to_code)(const OnigUChar* p, const OnigUChar* end); 136 int (*code_to_mbclen)(OnigCodePoint code); 137 int (*code_to_mbc)(OnigCodePoint code, OnigUChar *buf); 138 int (*mbc_case_fold)(OnigCaseFoldType flag, const OnigUChar** pp, const OnigUChar* end, OnigUChar* to); 139 int (*apply_all_case_fold)(OnigCaseFoldType flag, OnigApplyAllCaseFoldFunc f, void* arg); 140 int (*get_case_fold_codes_by_str)(OnigCaseFoldType flag, const OnigUChar* p, const OnigUChar* end, OnigCaseFoldCodeItem acs[]); 141 int (*property_name_to_ctype)(struct OnigEncodingTypeST* enc, OnigUChar* p, OnigUChar* end); 142 int (*is_code_ctype)(OnigCodePoint code, OnigCtype ctype); 143 int (*get_ctype_code_range)(OnigCtype ctype, OnigCodePoint* sb_out, const OnigCodePoint* ranges[]); 144 OnigUChar* (*left_adjust_char_head)(const OnigUChar* start, const OnigUChar* p); 145 int (*is_allowed_reverse_match)(const OnigUChar* p, const OnigUChar* end); 146 int (*init)(void); 147 int (*is_initialized)(void); 148 int (*is_valid_mbc_string)(const OnigUChar* s, const OnigUChar* end); 149 unsigned int flag; 150 OnigCodePoint sb_range; 151 int index; 152 } OnigEncodingType; 153 154 typedef OnigEncodingType* OnigEncoding; 155 156 ONIG_EXTERN OnigEncodingType OnigEncodingASCII; 157 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_1; 158 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_2; 159 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_3; 160 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_4; 161 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_5; 162 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_6; 163 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_7; 164 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_8; 165 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_9; 166 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_10; 167 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_11; 168 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_13; 169 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_14; 170 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_15; 171 ONIG_EXTERN OnigEncodingType OnigEncodingISO_8859_16; 172 ONIG_EXTERN OnigEncodingType OnigEncodingUTF8; 173 ONIG_EXTERN OnigEncodingType OnigEncodingUTF16_BE; 174 ONIG_EXTERN OnigEncodingType OnigEncodingUTF16_LE; 175 ONIG_EXTERN OnigEncodingType OnigEncodingUTF32_BE; 176 ONIG_EXTERN OnigEncodingType OnigEncodingUTF32_LE; 177 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_JP; 178 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_TW; 179 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_KR; 180 ONIG_EXTERN OnigEncodingType OnigEncodingEUC_CN; 181 ONIG_EXTERN OnigEncodingType OnigEncodingSJIS; 182 ONIG_EXTERN OnigEncodingType OnigEncodingKOI8; 183 ONIG_EXTERN OnigEncodingType OnigEncodingKOI8_R; 184 ONIG_EXTERN OnigEncodingType OnigEncodingCP1251; 185 ONIG_EXTERN OnigEncodingType OnigEncodingBIG5; 186 ONIG_EXTERN OnigEncodingType OnigEncodingGB18030; 187 188 #define ONIG_ENCODING_ASCII (&OnigEncodingASCII) 189 #define ONIG_ENCODING_ISO_8859_1 (&OnigEncodingISO_8859_1) 190 #define ONIG_ENCODING_ISO_8859_2 (&OnigEncodingISO_8859_2) 191 #define ONIG_ENCODING_ISO_8859_3 (&OnigEncodingISO_8859_3) 192 #define ONIG_ENCODING_ISO_8859_4 (&OnigEncodingISO_8859_4) 193 #define ONIG_ENCODING_ISO_8859_5 (&OnigEncodingISO_8859_5) 194 #define ONIG_ENCODING_ISO_8859_6 (&OnigEncodingISO_8859_6) 195 #define ONIG_ENCODING_ISO_8859_7 (&OnigEncodingISO_8859_7) 196 #define ONIG_ENCODING_ISO_8859_8 (&OnigEncodingISO_8859_8) 197 #define ONIG_ENCODING_ISO_8859_9 (&OnigEncodingISO_8859_9) 198 #define ONIG_ENCODING_ISO_8859_10 (&OnigEncodingISO_8859_10) 199 #define ONIG_ENCODING_ISO_8859_11 (&OnigEncodingISO_8859_11) 200 #define ONIG_ENCODING_ISO_8859_13 (&OnigEncodingISO_8859_13) 201 #define ONIG_ENCODING_ISO_8859_14 (&OnigEncodingISO_8859_14) 202 #define ONIG_ENCODING_ISO_8859_15 (&OnigEncodingISO_8859_15) 203 #define ONIG_ENCODING_ISO_8859_16 (&OnigEncodingISO_8859_16) 204 #define ONIG_ENCODING_UTF8 (&OnigEncodingUTF8) 205 #define ONIG_ENCODING_UTF16_BE (&OnigEncodingUTF16_BE) 206 #define ONIG_ENCODING_UTF16_LE (&OnigEncodingUTF16_LE) 207 #define ONIG_ENCODING_UTF32_BE (&OnigEncodingUTF32_BE) 208 #define ONIG_ENCODING_UTF32_LE (&OnigEncodingUTF32_LE) 209 #define ONIG_ENCODING_EUC_JP (&OnigEncodingEUC_JP) 210 #define ONIG_ENCODING_EUC_TW (&OnigEncodingEUC_TW) 211 #define ONIG_ENCODING_EUC_KR (&OnigEncodingEUC_KR) 212 #define ONIG_ENCODING_EUC_CN (&OnigEncodingEUC_CN) 213 #define ONIG_ENCODING_SJIS (&OnigEncodingSJIS) 214 #define ONIG_ENCODING_KOI8 (&OnigEncodingKOI8) 215 #define ONIG_ENCODING_KOI8_R (&OnigEncodingKOI8_R) 216 #define ONIG_ENCODING_CP1251 (&OnigEncodingCP1251) 217 #define ONIG_ENCODING_BIG5 (&OnigEncodingBIG5) 218 #define ONIG_ENCODING_GB18030 (&OnigEncodingGB18030) 219 220 #define ONIG_ENCODING_UNDEF ((OnigEncoding )0) 221 222 223 /* work size */ 224 #define ONIGENC_CODE_TO_MBC_MAXLEN 7 225 #define ONIGENC_MBC_CASE_FOLD_MAXLEN 18 226 /* 18: 6(max-byte) * 3(case-fold chars) */ 227 228 /* character types */ 229 typedef enum { 230 ONIGENC_CTYPE_NEWLINE = 0, 231 ONIGENC_CTYPE_ALPHA = 1, 232 ONIGENC_CTYPE_BLANK = 2, 233 ONIGENC_CTYPE_CNTRL = 3, 234 ONIGENC_CTYPE_DIGIT = 4, 235 ONIGENC_CTYPE_GRAPH = 5, 236 ONIGENC_CTYPE_LOWER = 6, 237 ONIGENC_CTYPE_PRINT = 7, 238 ONIGENC_CTYPE_PUNCT = 8, 239 ONIGENC_CTYPE_SPACE = 9, 240 ONIGENC_CTYPE_UPPER = 10, 241 ONIGENC_CTYPE_XDIGIT = 11, 242 ONIGENC_CTYPE_WORD = 12, 243 ONIGENC_CTYPE_ALNUM = 13, /* alpha || digit */ 244 ONIGENC_CTYPE_ASCII = 14 245 } OnigEncCtype; 246 247 #define ONIGENC_MAX_STD_CTYPE ONIGENC_CTYPE_ASCII 248 249 250 #define onig_enc_len(enc,p,end) ONIGENC_MBC_ENC_LEN(enc,p) 251 252 #define ONIGENC_IS_UNDEF(enc) ((enc) == ONIG_ENCODING_UNDEF) 253 #define ONIGENC_IS_SINGLEBYTE(enc) (ONIGENC_MBC_MAXLEN(enc) == 1) 254 #define ONIGENC_IS_MBC_HEAD(enc,p) (ONIGENC_MBC_ENC_LEN(enc,p) != 1) 255 #define ONIGENC_IS_MBC_ASCII(p) (*(p) < 128) 256 #define ONIGENC_IS_CODE_ASCII(code) ((code) < 128) 257 #define ONIGENC_IS_MBC_WORD(enc,s,end) \ 258 ONIGENC_IS_CODE_WORD(enc,ONIGENC_MBC_TO_CODE(enc,s,end)) 259 #define ONIGENC_IS_MBC_WORD_ASCII(enc,s,end) onigenc_is_mbc_word_ascii(enc,s,end) 260 261 #define ONIGENC_NAME(enc) ((enc)->name) 262 263 #define ONIGENC_MBC_CASE_FOLD(enc,flag,pp,end,buf) \ 264 (enc)->mbc_case_fold(flag,(const OnigUChar** )pp,end,buf) 265 #define ONIGENC_IS_ALLOWED_REVERSE_MATCH(enc,s,end) \ 266 (enc)->is_allowed_reverse_match(s,end) 267 #define ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc,start,s) \ 268 (enc)->left_adjust_char_head(start, s) 269 #define ONIGENC_IS_VALID_MBC_STRING(enc,s,end) \ 270 (enc)->is_valid_mbc_string(s,end) 271 #define ONIGENC_APPLY_ALL_CASE_FOLD(enc,case_fold_flag,f,arg) \ 272 (enc)->apply_all_case_fold(case_fold_flag,f,arg) 273 #define ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc,case_fold_flag,p,end,acs) \ 274 (enc)->get_case_fold_codes_by_str(case_fold_flag,p,end,acs) 275 #define ONIGENC_STEP_BACK(enc,start,s,n) \ 276 onigenc_step_back((enc),(start),(s),(n)) 277 278 #define ONIGENC_MBC_ENC_LEN(enc,p) (enc)->mbc_enc_len(p) 279 #define ONIGENC_MBC_MAXLEN(enc) ((enc)->max_enc_len) 280 #define ONIGENC_MBC_MAXLEN_DIST(enc) ONIGENC_MBC_MAXLEN(enc) 281 #define ONIGENC_MBC_MINLEN(enc) ((enc)->min_enc_len) 282 #define ONIGENC_IS_MBC_NEWLINE(enc,p,end) (enc)->is_mbc_newline((p),(end)) 283 #define ONIGENC_MBC_TO_CODE(enc,p,end) (enc)->mbc_to_code((p),(end)) 284 #define ONIGENC_CODE_TO_MBCLEN(enc,code) (enc)->code_to_mbclen(code) 285 #define ONIGENC_CODE_TO_MBC(enc,code,buf) (enc)->code_to_mbc(code,buf) 286 #define ONIGENC_PROPERTY_NAME_TO_CTYPE(enc,p,end) \ 287 (enc)->property_name_to_ctype(enc,p,end) 288 289 #define ONIGENC_IS_CODE_CTYPE(enc,code,ctype) (enc)->is_code_ctype(code,ctype) 290 291 #define ONIGENC_IS_CODE_NEWLINE(enc,code) \ 292 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_NEWLINE) 293 #define ONIGENC_IS_CODE_GRAPH(enc,code) \ 294 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_GRAPH) 295 #define ONIGENC_IS_CODE_PRINT(enc,code) \ 296 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PRINT) 297 #define ONIGENC_IS_CODE_ALNUM(enc,code) \ 298 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALNUM) 299 #define ONIGENC_IS_CODE_ALPHA(enc,code) \ 300 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_ALPHA) 301 #define ONIGENC_IS_CODE_LOWER(enc,code) \ 302 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_LOWER) 303 #define ONIGENC_IS_CODE_UPPER(enc,code) \ 304 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_UPPER) 305 #define ONIGENC_IS_CODE_CNTRL(enc,code) \ 306 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_CNTRL) 307 #define ONIGENC_IS_CODE_PUNCT(enc,code) \ 308 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_PUNCT) 309 #define ONIGENC_IS_CODE_SPACE(enc,code) \ 310 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_SPACE) 311 #define ONIGENC_IS_CODE_BLANK(enc,code) \ 312 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_BLANK) 313 #define ONIGENC_IS_CODE_DIGIT(enc,code) \ 314 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_DIGIT) 315 #define ONIGENC_IS_CODE_XDIGIT(enc,code) \ 316 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_XDIGIT) 317 #define ONIGENC_IS_CODE_WORD(enc,code) \ 318 ONIGENC_IS_CODE_CTYPE(enc,code,ONIGENC_CTYPE_WORD) 319 320 #define ONIGENC_GET_CTYPE_CODE_RANGE(enc,ctype,sbout,ranges) \ 321 (enc)->get_ctype_code_range(ctype,sbout,ranges) 322 323 ONIG_EXTERN 324 OnigUChar* onigenc_step_back P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s, int n)); 325 326 327 /* encoding API */ 328 ONIG_EXTERN 329 int onigenc_init P_((void)); 330 ONIG_EXTERN 331 int onig_initialize_encoding P_((OnigEncoding enc)); 332 ONIG_EXTERN 333 int onigenc_set_default_encoding P_((OnigEncoding enc)); 334 ONIG_EXTERN 335 OnigEncoding onigenc_get_default_encoding P_((void)); 336 ONIG_EXTERN 337 void onigenc_set_default_caseconv_table P_((const OnigUChar* table)); 338 ONIG_EXTERN 339 OnigUChar* onigenc_get_right_adjust_char_head_with_prev P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s, const OnigUChar** prev)); 340 ONIG_EXTERN 341 OnigUChar* onigenc_get_prev_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s)); 342 ONIG_EXTERN 343 OnigUChar* onigenc_get_left_adjust_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s)); 344 ONIG_EXTERN 345 OnigUChar* onigenc_get_right_adjust_char_head P_((OnigEncoding enc, const OnigUChar* start, const OnigUChar* s)); 346 ONIG_EXTERN 347 int onigenc_strlen P_((OnigEncoding enc, const OnigUChar* p, const OnigUChar* end)); 348 ONIG_EXTERN 349 int onigenc_strlen_null P_((OnigEncoding enc, const OnigUChar* p)); 350 ONIG_EXTERN 351 int onigenc_str_bytelen_null P_((OnigEncoding enc, const OnigUChar* p)); 352 ONIG_EXTERN 353 int onigenc_is_valid_mbc_string P_((OnigEncoding enc, const OnigUChar* s, const OnigUChar* end)); 354 ONIG_EXTERN 355 OnigUChar* onigenc_strdup P_((OnigEncoding enc, const OnigUChar* s, const OnigUChar* end)); 356 357 358 /* PART: regular expression */ 359 360 /* config parameters */ 361 #define ONIG_NREGION 10 362 #define ONIG_MAX_CAPTURE_NUM 2147483647 /* 2**31 - 1 */ 363 #define ONIG_MAX_BACKREF_NUM 1000 364 #define ONIG_MAX_REPEAT_NUM 100000 365 #define ONIG_MAX_MULTI_BYTE_RANGES_NUM 10000 366 /* constants */ 367 #define ONIG_MAX_ERROR_MESSAGE_LEN 90 368 369 typedef unsigned int OnigOptionType; 370 371 #define ONIG_OPTION_DEFAULT ONIG_OPTION_NONE 372 373 /* options */ 374 #define ONIG_OPTION_NONE 0U 375 /* options (compile time) */ 376 #define ONIG_OPTION_IGNORECASE 1U 377 #define ONIG_OPTION_EXTEND (ONIG_OPTION_IGNORECASE << 1) 378 #define ONIG_OPTION_MULTILINE (ONIG_OPTION_EXTEND << 1) 379 #define ONIG_OPTION_SINGLELINE (ONIG_OPTION_MULTILINE << 1) 380 #define ONIG_OPTION_FIND_LONGEST (ONIG_OPTION_SINGLELINE << 1) 381 #define ONIG_OPTION_FIND_NOT_EMPTY (ONIG_OPTION_FIND_LONGEST << 1) 382 #define ONIG_OPTION_NEGATE_SINGLELINE (ONIG_OPTION_FIND_NOT_EMPTY << 1) 383 #define ONIG_OPTION_DONT_CAPTURE_GROUP (ONIG_OPTION_NEGATE_SINGLELINE << 1) 384 #define ONIG_OPTION_CAPTURE_GROUP (ONIG_OPTION_DONT_CAPTURE_GROUP << 1) 385 /* options (search time) */ 386 #define ONIG_OPTION_NOTBOL (ONIG_OPTION_CAPTURE_GROUP << 1) 387 #define ONIG_OPTION_NOTEOL (ONIG_OPTION_NOTBOL << 1) 388 #define ONIG_OPTION_POSIX_REGION (ONIG_OPTION_NOTEOL << 1) 389 #define ONIG_OPTION_CHECK_VALIDITY_OF_STRING (ONIG_OPTION_POSIX_REGION << 1) 390 /* #define ONIG_OPTION_CRLF_AS_LINE_SEPARATOR (ONIG_OPTION_CHECK_VALIDITY_OF_STRING << 1) */ 391 /* options (compile time) */ 392 #define ONIG_OPTION_WORD_IS_ASCII (ONIG_OPTION_CHECK_VALIDITY_OF_STRING << 4) 393 #define ONIG_OPTION_DIGIT_IS_ASCII (ONIG_OPTION_WORD_IS_ASCII << 1) 394 #define ONIG_OPTION_SPACE_IS_ASCII (ONIG_OPTION_DIGIT_IS_ASCII << 1) 395 #define ONIG_OPTION_POSIX_IS_ASCII (ONIG_OPTION_SPACE_IS_ASCII << 1) 396 #define ONIG_OPTION_TEXT_SEGMENT_EXTENDED_GRAPHEME_CLUSTER (ONIG_OPTION_POSIX_IS_ASCII << 1) 397 #define ONIG_OPTION_TEXT_SEGMENT_WORD (ONIG_OPTION_TEXT_SEGMENT_EXTENDED_GRAPHEME_CLUSTER << 1) 398 399 #define ONIG_OPTION_MAXBIT ONIG_OPTION_TEXT_SEGMENT_WORD /* limit */ 400 401 #define ONIG_OPTION_ON(options,regopt) ((options) |= (regopt)) 402 #define ONIG_OPTION_OFF(options,regopt) ((options) &= ~(regopt)) 403 #define ONIG_IS_OPTION_ON(options,option) ((options) & (option)) 404 405 /* syntax */ 406 typedef struct { 407 unsigned int op; 408 unsigned int op2; 409 unsigned int behavior; 410 OnigOptionType options; /* default option */ 411 OnigMetaCharTableType meta_char_table; 412 } OnigSyntaxType; 413 414 ONIG_EXTERN OnigSyntaxType OnigSyntaxASIS; 415 ONIG_EXTERN OnigSyntaxType OnigSyntaxPosixBasic; 416 ONIG_EXTERN OnigSyntaxType OnigSyntaxPosixExtended; 417 ONIG_EXTERN OnigSyntaxType OnigSyntaxEmacs; 418 ONIG_EXTERN OnigSyntaxType OnigSyntaxGrep; 419 ONIG_EXTERN OnigSyntaxType OnigSyntaxGnuRegex; 420 ONIG_EXTERN OnigSyntaxType OnigSyntaxJava; 421 ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl; 422 ONIG_EXTERN OnigSyntaxType OnigSyntaxPerl_NG; 423 ONIG_EXTERN OnigSyntaxType OnigSyntaxRuby; 424 ONIG_EXTERN OnigSyntaxType OnigSyntaxOniguruma; 425 426 /* predefined syntaxes (see regsyntax.c) */ 427 #define ONIG_SYNTAX_ASIS (&OnigSyntaxASIS) 428 #define ONIG_SYNTAX_POSIX_BASIC (&OnigSyntaxPosixBasic) 429 #define ONIG_SYNTAX_POSIX_EXTENDED (&OnigSyntaxPosixExtended) 430 #define ONIG_SYNTAX_EMACS (&OnigSyntaxEmacs) 431 #define ONIG_SYNTAX_GREP (&OnigSyntaxGrep) 432 #define ONIG_SYNTAX_GNU_REGEX (&OnigSyntaxGnuRegex) 433 #define ONIG_SYNTAX_JAVA (&OnigSyntaxJava) 434 #define ONIG_SYNTAX_PERL (&OnigSyntaxPerl) 435 #define ONIG_SYNTAX_PERL_NG (&OnigSyntaxPerl_NG) 436 #define ONIG_SYNTAX_RUBY (&OnigSyntaxRuby) 437 #define ONIG_SYNTAX_ONIGURUMA (&OnigSyntaxOniguruma) 438 439 /* default syntax */ 440 ONIG_EXTERN OnigSyntaxType* OnigDefaultSyntax; 441 #define ONIG_SYNTAX_DEFAULT OnigDefaultSyntax 442 443 /* syntax (operators) */ 444 #define ONIG_SYN_OP_VARIABLE_META_CHARACTERS (1U<<0) 445 #define ONIG_SYN_OP_DOT_ANYCHAR (1U<<1) /* . */ 446 #define ONIG_SYN_OP_ASTERISK_ZERO_INF (1U<<2) /* * */ 447 #define ONIG_SYN_OP_ESC_ASTERISK_ZERO_INF (1U<<3) 448 #define ONIG_SYN_OP_PLUS_ONE_INF (1U<<4) /* + */ 449 #define ONIG_SYN_OP_ESC_PLUS_ONE_INF (1U<<5) 450 #define ONIG_SYN_OP_QMARK_ZERO_ONE (1U<<6) /* ? */ 451 #define ONIG_SYN_OP_ESC_QMARK_ZERO_ONE (1U<<7) 452 #define ONIG_SYN_OP_BRACE_INTERVAL (1U<<8) /* {lower,upper} */ 453 #define ONIG_SYN_OP_ESC_BRACE_INTERVAL (1U<<9) /* \{lower,upper\} */ 454 #define ONIG_SYN_OP_VBAR_ALT (1U<<10) /* | */ 455 #define ONIG_SYN_OP_ESC_VBAR_ALT (1U<<11) /* \| */ 456 #define ONIG_SYN_OP_LPAREN_SUBEXP (1U<<12) /* (...) */ 457 #define ONIG_SYN_OP_ESC_LPAREN_SUBEXP (1U<<13) /* \(...\) */ 458 #define ONIG_SYN_OP_ESC_AZ_BUF_ANCHOR (1U<<14) /* \A, \Z, \z */ 459 #define ONIG_SYN_OP_ESC_CAPITAL_G_BEGIN_ANCHOR (1U<<15) /* \G */ 460 #define ONIG_SYN_OP_DECIMAL_BACKREF (1U<<16) /* \num */ 461 #define ONIG_SYN_OP_BRACKET_CC (1U<<17) /* [...] */ 462 #define ONIG_SYN_OP_ESC_W_WORD (1U<<18) /* \w, \W */ 463 #define ONIG_SYN_OP_ESC_LTGT_WORD_BEGIN_END (1U<<19) /* \<. \> */ 464 #define ONIG_SYN_OP_ESC_B_WORD_BOUND (1U<<20) /* \b, \B */ 465 #define ONIG_SYN_OP_ESC_S_WHITE_SPACE (1U<<21) /* \s, \S */ 466 #define ONIG_SYN_OP_ESC_D_DIGIT (1U<<22) /* \d, \D */ 467 #define ONIG_SYN_OP_LINE_ANCHOR (1U<<23) /* ^, $ */ 468 #define ONIG_SYN_OP_POSIX_BRACKET (1U<<24) /* [:xxxx:] */ 469 #define ONIG_SYN_OP_QMARK_NON_GREEDY (1U<<25) /* ??,*?,+?,{n,m}? */ 470 #define ONIG_SYN_OP_ESC_CONTROL_CHARS (1U<<26) /* \n,\r,\t,\a ... */ 471 #define ONIG_SYN_OP_ESC_C_CONTROL (1U<<27) /* \cx */ 472 #define ONIG_SYN_OP_ESC_OCTAL3 (1U<<28) /* \OOO */ 473 #define ONIG_SYN_OP_ESC_X_HEX2 (1U<<29) /* \xHH */ 474 #define ONIG_SYN_OP_ESC_X_BRACE_HEX8 (1U<<30) /* \x{7HHHHHHH} */ 475 #define ONIG_SYN_OP_ESC_O_BRACE_OCTAL (1U<<31) /* \o{1OOOOOOOOOO} */ 476 477 #define ONIG_SYN_OP2_ESC_CAPITAL_Q_QUOTE (1U<<0) /* \Q...\E */ 478 #define ONIG_SYN_OP2_QMARK_GROUP_EFFECT (1U<<1) /* (?...) */ 479 #define ONIG_SYN_OP2_OPTION_PERL (1U<<2) /* (?imsx),(?-imsx) */ 480 #define ONIG_SYN_OP2_OPTION_RUBY (1U<<3) /* (?imx), (?-imx) */ 481 #define ONIG_SYN_OP2_PLUS_POSSESSIVE_REPEAT (1U<<4) /* ?+,*+,++ */ 482 #define ONIG_SYN_OP2_PLUS_POSSESSIVE_INTERVAL (1U<<5) /* {n,m}+ */ 483 #define ONIG_SYN_OP2_CCLASS_SET_OP (1U<<6) /* [...&&..[..]..] */ 484 #define ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP (1U<<7) /* (?<name>...) */ 485 #define ONIG_SYN_OP2_ESC_K_NAMED_BACKREF (1U<<8) /* \k<name> */ 486 #define ONIG_SYN_OP2_ESC_G_SUBEXP_CALL (1U<<9) /* \g<name>, \g<n> */ 487 #define ONIG_SYN_OP2_ATMARK_CAPTURE_HISTORY (1U<<10) /* (?@..),(?@<x>..) */ 488 #define ONIG_SYN_OP2_ESC_CAPITAL_C_BAR_CONTROL (1U<<11) /* \C-x */ 489 #define ONIG_SYN_OP2_ESC_CAPITAL_M_BAR_META (1U<<12) /* \M-x */ 490 #define ONIG_SYN_OP2_ESC_V_VTAB (1U<<13) /* \v as VTAB */ 491 #define ONIG_SYN_OP2_ESC_U_HEX4 (1U<<14) /* \uHHHH */ 492 #define ONIG_SYN_OP2_ESC_GNU_BUF_ANCHOR (1U<<15) /* \`, \' */ 493 #define ONIG_SYN_OP2_ESC_P_BRACE_CHAR_PROPERTY (1U<<16) /* \p{...}, \P{...} */ 494 #define ONIG_SYN_OP2_ESC_P_BRACE_CIRCUMFLEX_NOT (1U<<17) /* \p{^..}, \P{^..} */ 495 /* #define ONIG_SYN_OP2_CHAR_PROPERTY_PREFIX_IS (1U<<18) */ 496 #define ONIG_SYN_OP2_ESC_H_XDIGIT (1U<<19) /* \h, \H */ 497 #define ONIG_SYN_OP2_INEFFECTIVE_ESCAPE (1U<<20) /* \ */ 498 #define ONIG_SYN_OP2_QMARK_LPAREN_IF_ELSE (1U<<21) /* (?(n)) (?(...)...|...) */ 499 #define ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP (1U<<22) /* \K */ 500 #define ONIG_SYN_OP2_ESC_CAPITAL_R_GENERAL_NEWLINE (1U<<23) /* \R \r\n else [\x0a-\x0d] */ 501 #define ONIG_SYN_OP2_ESC_CAPITAL_N_O_SUPER_DOT (1U<<24) /* \N (?-m:.), \O (?m:.) */ 502 #define ONIG_SYN_OP2_QMARK_TILDE_ABSENT_GROUP (1U<<25) /* (?~...) */ 503 #define ONIG_SYN_OP2_ESC_X_Y_GRAPHEME_CLUSTER (1U<<26) /* obsoleted: use next */ 504 #define ONIG_SYN_OP2_ESC_X_Y_TEXT_SEGMENT (1U<<26) /* \X \y \Y */ 505 #define ONIG_SYN_OP2_QMARK_PERL_SUBEXP_CALL (1U<<27) /* (?R), (?&name)... */ 506 #define ONIG_SYN_OP2_QMARK_BRACE_CALLOUT_CONTENTS (1U<<28) /* (?{...}) (?{{...}}) */ 507 #define ONIG_SYN_OP2_ASTERISK_CALLOUT_NAME (1U<<29) /* (*name) (*name{a,..}) */ 508 #define ONIG_SYN_OP2_OPTION_ONIGURUMA (1U<<30) /* (?imxWDSPy) */ 509 510 /* syntax (behavior) */ 511 #define ONIG_SYN_CONTEXT_INDEP_ANCHORS (1U<<31) /* not implemented */ 512 #define ONIG_SYN_CONTEXT_INDEP_REPEAT_OPS (1U<<0) /* ?, *, +, {n,m} */ 513 #define ONIG_SYN_CONTEXT_INVALID_REPEAT_OPS (1U<<1) /* error or ignore */ 514 #define ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP (1U<<2) /* ...)... */ 515 #define ONIG_SYN_ALLOW_INVALID_INTERVAL (1U<<3) /* {??? */ 516 #define ONIG_SYN_ALLOW_INTERVAL_LOW_ABBREV (1U<<4) /* {,n} => {0,n} */ 517 #define ONIG_SYN_STRICT_CHECK_BACKREF (1U<<5) /* /(\1)/,/\1()/ ..*/ 518 #define ONIG_SYN_DIFFERENT_LEN_ALT_LOOK_BEHIND (1U<<6) /* (?<=a|bc) */ 519 #define ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP (1U<<7) /* see doc/RE */ 520 #define ONIG_SYN_ALLOW_MULTIPLEX_DEFINITION_NAME (1U<<8) /* (?<x>)(?<x>) */ 521 #define ONIG_SYN_FIXED_INTERVAL_IS_GREEDY_ONLY (1U<<9) /* a{n}?=(?:a{n})? */ 522 #define ONIG_SYN_ISOLATED_OPTION_CONTINUE_BRANCH (1U<<10) /* ..(?i)...|... */ 523 #define ONIG_SYN_VARIABLE_LEN_LOOK_BEHIND (1U<<11) /* (?<=a+|..) */ 524 525 /* syntax (behavior) in char class [...] */ 526 #define ONIG_SYN_NOT_NEWLINE_IN_NEGATIVE_CC (1U<<20) /* [^...] */ 527 #define ONIG_SYN_BACKSLASH_ESCAPE_IN_CC (1U<<21) /* [..\w..] etc.. */ 528 #define ONIG_SYN_ALLOW_EMPTY_RANGE_IN_CC (1U<<22) 529 #define ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC (1U<<23) /* [0-9-a]=[0-9\-a] */ 530 #define ONIG_SYN_ALLOW_INVALID_CODE_END_OF_RANGE_IN_CC (1U<<26) 531 /* syntax (behavior) warning */ 532 #define ONIG_SYN_WARN_CC_OP_NOT_ESCAPED (1U<<24) /* [,-,] */ 533 #define ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT (1U<<25) /* (?:a*)+ */ 534 535 /* meta character specifiers (onig_set_meta_char()) */ 536 #define ONIG_META_CHAR_ESCAPE 0 537 #define ONIG_META_CHAR_ANYCHAR 1 538 #define ONIG_META_CHAR_ANYTIME 2 539 #define ONIG_META_CHAR_ZERO_OR_ONE_TIME 3 540 #define ONIG_META_CHAR_ONE_OR_MORE_TIME 4 541 #define ONIG_META_CHAR_ANYCHAR_ANYTIME 5 542 543 #define ONIG_INEFFECTIVE_META_CHAR 0 544 545 /* error codes */ 546 #define ONIG_IS_PATTERN_ERROR(ecode) ((ecode) <= -100 && (ecode) > -1000) 547 /* normal return */ 548 #define ONIG_NORMAL 0 549 #define ONIG_MISMATCH -1 550 #define ONIG_NO_SUPPORT_CONFIG -2 551 #define ONIG_ABORT -3 552 553 /* internal error */ 554 #define ONIGERR_MEMORY -5 555 #define ONIGERR_TYPE_BUG -6 556 #define ONIGERR_PARSER_BUG -11 557 #define ONIGERR_STACK_BUG -12 558 #define ONIGERR_UNDEFINED_BYTECODE -13 559 #define ONIGERR_UNEXPECTED_BYTECODE -14 560 #define ONIGERR_MATCH_STACK_LIMIT_OVER -15 561 #define ONIGERR_PARSE_DEPTH_LIMIT_OVER -16 562 #define ONIGERR_RETRY_LIMIT_IN_MATCH_OVER -17 563 #define ONIGERR_RETRY_LIMIT_IN_SEARCH_OVER -18 564 #define ONIGERR_DEFAULT_ENCODING_IS_NOT_SETTED -21 565 #define ONIGERR_SPECIFIED_ENCODING_CANT_CONVERT_TO_WIDE_CHAR -22 566 #define ONIGERR_FAIL_TO_INITIALIZE -23 567 /* general error */ 568 #define ONIGERR_INVALID_ARGUMENT -30 569 /* syntax error */ 570 #define ONIGERR_END_PATTERN_AT_LEFT_BRACE -100 571 #define ONIGERR_END_PATTERN_AT_LEFT_BRACKET -101 572 #define ONIGERR_EMPTY_CHAR_CLASS -102 573 #define ONIGERR_PREMATURE_END_OF_CHAR_CLASS -103 574 #define ONIGERR_END_PATTERN_AT_ESCAPE -104 575 #define ONIGERR_END_PATTERN_AT_META -105 576 #define ONIGERR_END_PATTERN_AT_CONTROL -106 577 #define ONIGERR_META_CODE_SYNTAX -108 578 #define ONIGERR_CONTROL_CODE_SYNTAX -109 579 #define ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE -110 580 #define ONIGERR_CHAR_CLASS_VALUE_AT_START_OF_RANGE -111 581 #define ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS -112 582 #define ONIGERR_TARGET_OF_REPEAT_OPERATOR_NOT_SPECIFIED -113 583 #define ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID -114 584 #define ONIGERR_NESTED_REPEAT_OPERATOR -115 585 #define ONIGERR_UNMATCHED_CLOSE_PARENTHESIS -116 586 #define ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS -117 587 #define ONIGERR_END_PATTERN_IN_GROUP -118 588 #define ONIGERR_UNDEFINED_GROUP_OPTION -119 589 #define ONIGERR_INVALID_POSIX_BRACKET_TYPE -121 590 #define ONIGERR_INVALID_LOOK_BEHIND_PATTERN -122 591 #define ONIGERR_INVALID_REPEAT_RANGE_PATTERN -123 592 /* values error (syntax error) */ 593 #define ONIGERR_TOO_BIG_NUMBER -200 594 #define ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE -201 595 #define ONIGERR_UPPER_SMALLER_THAN_LOWER_IN_REPEAT_RANGE -202 596 #define ONIGERR_EMPTY_RANGE_IN_CHAR_CLASS -203 597 #define ONIGERR_MISMATCH_CODE_LENGTH_IN_CLASS_RANGE -204 598 #define ONIGERR_TOO_MANY_MULTI_BYTE_RANGES -205 599 #define ONIGERR_TOO_SHORT_MULTI_BYTE_STRING -206 600 #define ONIGERR_TOO_BIG_BACKREF_NUMBER -207 601 #define ONIGERR_INVALID_BACKREF -208 602 #define ONIGERR_NUMBERED_BACKREF_OR_CALL_NOT_ALLOWED -209 603 #define ONIGERR_TOO_MANY_CAPTURES -210 604 #define ONIGERR_TOO_LONG_WIDE_CHAR_VALUE -212 605 #define ONIGERR_EMPTY_GROUP_NAME -214 606 #define ONIGERR_INVALID_GROUP_NAME -215 607 #define ONIGERR_INVALID_CHAR_IN_GROUP_NAME -216 608 #define ONIGERR_UNDEFINED_NAME_REFERENCE -217 609 #define ONIGERR_UNDEFINED_GROUP_REFERENCE -218 610 #define ONIGERR_MULTIPLEX_DEFINED_NAME -219 611 #define ONIGERR_MULTIPLEX_DEFINITION_NAME_CALL -220 612 #define ONIGERR_NEVER_ENDING_RECURSION -221 613 #define ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY -222 614 #define ONIGERR_INVALID_CHAR_PROPERTY_NAME -223 615 #define ONIGERR_INVALID_IF_ELSE_SYNTAX -224 616 #define ONIGERR_INVALID_ABSENT_GROUP_PATTERN -225 617 #define ONIGERR_INVALID_ABSENT_GROUP_GENERATOR_PATTERN -226 618 #define ONIGERR_INVALID_CALLOUT_PATTERN -227 619 #define ONIGERR_INVALID_CALLOUT_NAME -228 620 #define ONIGERR_UNDEFINED_CALLOUT_NAME -229 621 #define ONIGERR_INVALID_CALLOUT_BODY -230 622 #define ONIGERR_INVALID_CALLOUT_TAG_NAME -231 623 #define ONIGERR_INVALID_CALLOUT_ARG -232 624 #define ONIGERR_INVALID_CODE_POINT_VALUE -400 625 #define ONIGERR_INVALID_WIDE_CHAR_VALUE -400 626 #define ONIGERR_TOO_BIG_WIDE_CHAR_VALUE -401 627 #define ONIGERR_NOT_SUPPORTED_ENCODING_COMBINATION -402 628 #define ONIGERR_INVALID_COMBINATION_OF_OPTIONS -403 629 #define ONIGERR_TOO_MANY_USER_DEFINED_OBJECTS -404 630 #define ONIGERR_TOO_LONG_PROPERTY_NAME -405 631 #define ONIGERR_LIBRARY_IS_NOT_INITIALIZED -500 632 633 /* errors related to thread */ 634 /* #define ONIGERR_OVER_THREAD_PASS_LIMIT_COUNT -1001 */ 635 636 637 /* must be smaller than MEM_STATUS_BITS_NUM (unsigned int * 8) */ 638 #define ONIG_MAX_CAPTURE_HISTORY_GROUP 31 639 #define ONIG_IS_CAPTURE_HISTORY_GROUP(r, i) \ 640 ((i) <= ONIG_MAX_CAPTURE_HISTORY_GROUP && (r)->list && (r)->list[i]) 641 642 typedef struct OnigCaptureTreeNodeStruct { 643 int group; /* group number */ 644 int beg; 645 int end; 646 int allocated; 647 int num_childs; 648 struct OnigCaptureTreeNodeStruct** childs; 649 } OnigCaptureTreeNode; 650 651 /* match result region type */ 652 struct re_registers { 653 int allocated; 654 int num_regs; 655 int* beg; 656 int* end; 657 /* extended */ 658 OnigCaptureTreeNode* history_root; /* capture history tree root */ 659 }; 660 661 /* capture tree traverse */ 662 #define ONIG_TRAVERSE_CALLBACK_AT_FIRST 1 663 #define ONIG_TRAVERSE_CALLBACK_AT_LAST 2 664 #define ONIG_TRAVERSE_CALLBACK_AT_BOTH \ 665 ( ONIG_TRAVERSE_CALLBACK_AT_FIRST | ONIG_TRAVERSE_CALLBACK_AT_LAST ) 666 667 668 #define ONIG_REGION_NOTPOS -1 669 670 typedef struct re_registers OnigRegion; 671 672 typedef struct { 673 OnigEncoding enc; 674 OnigUChar* par; 675 OnigUChar* par_end; 676 } OnigErrorInfo; 677 678 typedef struct { 679 int lower; 680 int upper; 681 } OnigRepeatRange; 682 683 typedef void (*OnigWarnFunc) P_((const char* s)); 684 extern void onig_null_warn P_((const char* s)); 685 #define ONIG_NULL_WARN onig_null_warn 686 687 #define ONIG_CHAR_TABLE_SIZE 256 688 689 struct re_pattern_buffer; 690 typedef struct re_pattern_buffer OnigRegexType; 691 typedef OnigRegexType* OnigRegex; 692 693 #ifndef ONIG_ESCAPE_REGEX_T_COLLISION 694 typedef OnigRegexType regex_t; 695 #endif 696 697 struct OnigRegSetStruct; 698 typedef struct OnigRegSetStruct OnigRegSet; 699 700 typedef enum { 701 ONIG_REGSET_POSITION_LEAD = 0, 702 ONIG_REGSET_REGEX_LEAD = 1, 703 ONIG_REGSET_PRIORITY_TO_REGEX_ORDER = 2 704 } OnigRegSetLead; 705 706 typedef struct { 707 int num_of_elements; 708 OnigEncoding pattern_enc; 709 OnigEncoding target_enc; 710 OnigSyntaxType* syntax; 711 OnigOptionType option; 712 OnigCaseFoldType case_fold_flag; 713 } OnigCompileInfo; 714 715 716 /* types for callout */ 717 typedef enum { 718 ONIG_CALLOUT_IN_PROGRESS = 1, /* 1<<0 */ 719 ONIG_CALLOUT_IN_RETRACTION = 2 /* 1<<1 */ 720 } OnigCalloutIn; 721 722 #define ONIG_CALLOUT_IN_BOTH (ONIG_CALLOUT_IN_PROGRESS | ONIG_CALLOUT_IN_RETRACTION) 723 724 typedef enum { 725 ONIG_CALLOUT_OF_CONTENTS = 0, 726 ONIG_CALLOUT_OF_NAME = 1 727 } OnigCalloutOf; 728 729 typedef enum { 730 ONIG_CALLOUT_TYPE_SINGLE = 0, 731 ONIG_CALLOUT_TYPE_START_CALL = 1, 732 ONIG_CALLOUT_TYPE_BOTH_CALL = 2, 733 ONIG_CALLOUT_TYPE_START_MARK_END_CALL = 3, 734 } OnigCalloutType; 735 736 737 #define ONIG_NON_NAME_ID -1 738 #define ONIG_NON_CALLOUT_NUM 0 739 740 #define ONIG_CALLOUT_MAX_ARGS_NUM 4 741 #define ONIG_CALLOUT_DATA_SLOT_NUM 5 742 743 struct OnigCalloutArgsStruct; 744 typedef struct OnigCalloutArgsStruct OnigCalloutArgs; 745 746 typedef int (*OnigCalloutFunc)(OnigCalloutArgs* args, void* user_data); 747 748 /* callout function return values (less than -1: error code) */ 749 typedef enum { 750 ONIG_CALLOUT_FAIL = 1, 751 ONIG_CALLOUT_SUCCESS = 0 752 } OnigCalloutResult; 753 754 typedef enum { 755 ONIG_TYPE_VOID = 0, 756 ONIG_TYPE_LONG = 1<<0, 757 ONIG_TYPE_CHAR = 1<<1, 758 ONIG_TYPE_STRING = 1<<2, 759 ONIG_TYPE_POINTER = 1<<3, 760 ONIG_TYPE_TAG = 1<<4, 761 } OnigType; 762 763 typedef union { 764 long l; 765 OnigCodePoint c; 766 struct { 767 OnigUChar* start; 768 OnigUChar* end; 769 } s; 770 void* p; 771 int tag; /* tag -> callout_num */ 772 } OnigValue; 773 774 775 struct OnigMatchParamStruct; 776 typedef struct OnigMatchParamStruct OnigMatchParam; 777 778 779 /* Oniguruma Native API */ 780 781 ONIG_EXTERN 782 int onig_initialize P_((OnigEncoding encodings[], int number_of_encodings)); 783 /* onig_init(): deprecated function. Use onig_initialize(). */ 784 ONIG_EXTERN 785 int onig_init P_((void)); 786 ONIG_EXTERN 787 int ONIG_VARIADIC_FUNC_ATTR onig_error_code_to_str PV_((OnigUChar* s, int err_code, ...)); 788 ONIG_EXTERN 789 int onig_is_error_code_needs_param PV_((int code)); 790 ONIG_EXTERN 791 void onig_set_warn_func P_((OnigWarnFunc f)); 792 ONIG_EXTERN 793 void onig_set_verb_warn_func P_((OnigWarnFunc f)); 794 ONIG_EXTERN 795 int onig_new P_((OnigRegex*, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax, OnigErrorInfo* einfo)); 796 ONIG_EXTERN 797 int onig_reg_init P_((OnigRegex reg, OnigOptionType option, OnigCaseFoldType case_fold_flag, OnigEncoding enc, OnigSyntaxType* syntax)); 798 int onig_new_without_alloc P_((OnigRegex, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax, OnigErrorInfo* einfo)); 799 ONIG_EXTERN 800 int onig_new_deluxe P_((OnigRegex* reg, const OnigUChar* pattern, const OnigUChar* pattern_end, OnigCompileInfo* ci, OnigErrorInfo* einfo)); 801 ONIG_EXTERN 802 void onig_free P_((OnigRegex)); 803 ONIG_EXTERN 804 void onig_free_body P_((OnigRegex)); 805 ONIG_EXTERN 806 int onig_scan(OnigRegex reg, const OnigUChar* str, const OnigUChar* end, OnigRegion* region, OnigOptionType option, int (*scan_callback)(int, int, OnigRegion*, void*), void* callback_arg); 807 ONIG_EXTERN 808 int onig_search P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegion* region, OnigOptionType option)); 809 ONIG_EXTERN 810 int onig_search_with_param P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegion* region, OnigOptionType option, OnigMatchParam* mp)); 811 ONIG_EXTERN 812 int onig_match P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* at, OnigRegion* region, OnigOptionType option)); 813 ONIG_EXTERN 814 int onig_match_with_param P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* at, OnigRegion* region, OnigOptionType option, OnigMatchParam* mp)); 815 816 ONIG_EXTERN 817 int onig_regset_new P_((OnigRegSet** rset, int n, regex_t* regs[])); 818 ONIG_EXTERN 819 int onig_regset_add P_((OnigRegSet* set, regex_t* reg)); 820 ONIG_EXTERN 821 int onig_regset_replace P_((OnigRegSet* set, int at, regex_t* reg)); 822 ONIG_EXTERN 823 void onig_regset_free P_((OnigRegSet* set)); 824 ONIG_EXTERN 825 int onig_regset_number_of_regex P_((OnigRegSet* set)); 826 ONIG_EXTERN 827 regex_t* onig_regset_get_regex P_((OnigRegSet* set, int at)); 828 ONIG_EXTERN 829 OnigRegion* onig_regset_get_region P_((OnigRegSet* set, int at)); 830 ONIG_EXTERN 831 int onig_regset_search P_((OnigRegSet* set, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegSetLead lead, OnigOptionType option, int* rmatch_pos)); 832 ONIG_EXTERN 833 int onig_regset_search_with_param P_((OnigRegSet* set, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegSetLead lead, OnigOptionType option, OnigMatchParam* mps[], int* rmatch_pos)); 834 835 ONIG_EXTERN 836 OnigRegion* onig_region_new P_((void)); 837 ONIG_EXTERN 838 void onig_region_init P_((OnigRegion* region)); 839 ONIG_EXTERN 840 void onig_region_free P_((OnigRegion* region, int free_self)); 841 ONIG_EXTERN 842 void onig_region_copy P_((OnigRegion* to, OnigRegion* from)); 843 ONIG_EXTERN 844 void onig_region_clear P_((OnigRegion* region)); 845 ONIG_EXTERN 846 int onig_region_resize P_((OnigRegion* region, int n)); 847 ONIG_EXTERN 848 int onig_region_set P_((OnigRegion* region, int at, int beg, int end)); 849 ONIG_EXTERN 850 int onig_name_to_group_numbers P_((OnigRegex reg, const OnigUChar* name, const OnigUChar* name_end, int** nums)); 851 ONIG_EXTERN 852 int onig_name_to_backref_number P_((OnigRegex reg, const OnigUChar* name, const OnigUChar* name_end, OnigRegion *region)); 853 ONIG_EXTERN 854 int onig_foreach_name P_((OnigRegex reg, int (*func)(const OnigUChar*, const OnigUChar*,int,int*,OnigRegex,void*), void* arg)); 855 ONIG_EXTERN 856 int onig_number_of_names P_((OnigRegex reg)); 857 ONIG_EXTERN 858 int onig_number_of_captures P_((OnigRegex reg)); 859 ONIG_EXTERN 860 int onig_number_of_capture_histories P_((OnigRegex reg)); 861 ONIG_EXTERN 862 OnigCaptureTreeNode* onig_get_capture_tree P_((OnigRegion* region)); 863 ONIG_EXTERN 864 int onig_capture_tree_traverse P_((OnigRegion* region, int at, int(*callback_func)(int,int,int,int,int,void*), void* arg)); 865 ONIG_EXTERN 866 int onig_noname_group_capture_is_active P_((OnigRegex reg)); 867 ONIG_EXTERN 868 OnigEncoding onig_get_encoding P_((OnigRegex reg)); 869 ONIG_EXTERN 870 OnigOptionType onig_get_options P_((OnigRegex reg)); 871 ONIG_EXTERN 872 OnigCaseFoldType onig_get_case_fold_flag P_((OnigRegex reg)); 873 ONIG_EXTERN 874 OnigSyntaxType* onig_get_syntax P_((OnigRegex reg)); 875 ONIG_EXTERN 876 int onig_set_default_syntax P_((OnigSyntaxType* syntax)); 877 ONIG_EXTERN 878 void onig_copy_syntax P_((OnigSyntaxType* to, OnigSyntaxType* from)); 879 ONIG_EXTERN 880 unsigned int onig_get_syntax_op P_((OnigSyntaxType* syntax)); 881 ONIG_EXTERN 882 unsigned int onig_get_syntax_op2 P_((OnigSyntaxType* syntax)); 883 ONIG_EXTERN 884 unsigned int onig_get_syntax_behavior P_((OnigSyntaxType* syntax)); 885 ONIG_EXTERN 886 OnigOptionType onig_get_syntax_options P_((OnigSyntaxType* syntax)); 887 ONIG_EXTERN 888 void onig_set_syntax_op P_((OnigSyntaxType* syntax, unsigned int op)); 889 ONIG_EXTERN 890 void onig_set_syntax_op2 P_((OnigSyntaxType* syntax, unsigned int op2)); 891 ONIG_EXTERN 892 void onig_set_syntax_behavior P_((OnigSyntaxType* syntax, unsigned int behavior)); 893 ONIG_EXTERN 894 void onig_set_syntax_options P_((OnigSyntaxType* syntax, OnigOptionType options)); 895 ONIG_EXTERN 896 int onig_set_meta_char P_((OnigSyntaxType* syntax, unsigned int what, OnigCodePoint code)); 897 ONIG_EXTERN 898 void onig_copy_encoding P_((OnigEncoding to, OnigEncoding from)); 899 ONIG_EXTERN 900 OnigCaseFoldType onig_get_default_case_fold_flag P_((void)); 901 ONIG_EXTERN 902 int onig_set_default_case_fold_flag P_((OnigCaseFoldType case_fold_flag)); 903 ONIG_EXTERN 904 unsigned int onig_get_match_stack_limit_size P_((void)); 905 ONIG_EXTERN 906 int onig_set_match_stack_limit_size P_((unsigned int size)); 907 ONIG_EXTERN 908 unsigned long onig_get_retry_limit_in_match P_((void)); 909 ONIG_EXTERN 910 int onig_set_retry_limit_in_match P_((unsigned long n)); 911 ONIG_EXTERN 912 unsigned long onig_get_retry_limit_in_search P_((void)); 913 ONIG_EXTERN 914 int onig_set_retry_limit_in_search P_((unsigned long n)); 915 ONIG_EXTERN 916 unsigned int onig_get_parse_depth_limit P_((void)); 917 ONIG_EXTERN 918 int onig_set_capture_num_limit P_((int num)); 919 ONIG_EXTERN 920 int onig_set_parse_depth_limit P_((unsigned int depth)); 921 ONIG_EXTERN 922 int onig_unicode_define_user_property P_((const char* name, OnigCodePoint* ranges)); 923 ONIG_EXTERN 924 int onig_end P_((void)); 925 ONIG_EXTERN 926 const char* onig_version P_((void)); 927 ONIG_EXTERN 928 const char* onig_copyright P_((void)); 929 930 /* for OnigMatchParam */ 931 ONIG_EXTERN 932 OnigMatchParam* onig_new_match_param P_((void)); 933 ONIG_EXTERN 934 void onig_free_match_param P_((OnigMatchParam* p)); 935 ONIG_EXTERN 936 void onig_free_match_param_content P_((OnigMatchParam* p)); 937 ONIG_EXTERN 938 int onig_initialize_match_param P_((OnigMatchParam* mp)); 939 ONIG_EXTERN 940 int onig_set_match_stack_limit_size_of_match_param P_((OnigMatchParam* param, unsigned int limit)); 941 ONIG_EXTERN 942 int onig_set_retry_limit_in_match_of_match_param P_((OnigMatchParam* param, unsigned long limit)); 943 ONIG_EXTERN 944 int onig_set_retry_limit_in_search_of_match_param P_((OnigMatchParam* param, unsigned long limit)); 945 ONIG_EXTERN 946 int onig_set_progress_callout_of_match_param P_((OnigMatchParam* param, OnigCalloutFunc f)); 947 ONIG_EXTERN 948 int onig_set_retraction_callout_of_match_param P_((OnigMatchParam* param, OnigCalloutFunc f)); 949 ONIG_EXTERN 950 int onig_set_callout_user_data_of_match_param P_((OnigMatchParam* param, void* user_data)); 951 952 /* for callout functions */ 953 ONIG_EXTERN 954 OnigCalloutFunc onig_get_progress_callout P_((void)); 955 ONIG_EXTERN 956 int onig_set_progress_callout P_((OnigCalloutFunc f)); 957 ONIG_EXTERN 958 OnigCalloutFunc onig_get_retraction_callout P_((void)); 959 ONIG_EXTERN 960 int onig_set_retraction_callout P_((OnigCalloutFunc f)); 961 ONIG_EXTERN 962 int onig_set_callout_of_name P_((OnigEncoding enc, OnigCalloutType type, OnigUChar* name, OnigUChar* name_end, int callout_in, OnigCalloutFunc callout, OnigCalloutFunc end_callout, int arg_num, unsigned int arg_types[], int optional_arg_num, OnigValue opt_defaults[])); 963 ONIG_EXTERN 964 OnigUChar* onig_get_callout_name_by_name_id P_((int id)); 965 ONIG_EXTERN 966 int onig_get_callout_num_by_tag P_((OnigRegex reg, const OnigUChar* tag, const OnigUChar* tag_end)); 967 ONIG_EXTERN 968 int onig_get_callout_data_by_tag P_((OnigRegex reg, OnigMatchParam* mp, const OnigUChar* tag, const OnigUChar* tag_end, int slot, OnigType* type, OnigValue* val)); 969 ONIG_EXTERN 970 int onig_set_callout_data_by_tag P_((OnigRegex reg, OnigMatchParam* mp, const OnigUChar* tag, const OnigUChar* tag_end, int slot, OnigType type, OnigValue* val)); 971 972 /* used in callout functions */ 973 ONIG_EXTERN 974 int onig_get_callout_num_by_callout_args P_((OnigCalloutArgs* args)); 975 ONIG_EXTERN 976 OnigCalloutIn onig_get_callout_in_by_callout_args P_((OnigCalloutArgs* args)); 977 ONIG_EXTERN 978 int onig_get_name_id_by_callout_args P_((OnigCalloutArgs* args)); 979 ONIG_EXTERN 980 const OnigUChar* onig_get_contents_by_callout_args P_((OnigCalloutArgs* args)); 981 ONIG_EXTERN 982 const OnigUChar* onig_get_contents_end_by_callout_args P_((OnigCalloutArgs* args)); 983 ONIG_EXTERN 984 int onig_get_args_num_by_callout_args P_((OnigCalloutArgs* args)); 985 ONIG_EXTERN 986 int onig_get_passed_args_num_by_callout_args P_((OnigCalloutArgs* args)); 987 ONIG_EXTERN 988 int onig_get_arg_by_callout_args P_((OnigCalloutArgs* args, int index, OnigType* type, OnigValue* val)); 989 ONIG_EXTERN 990 const OnigUChar* onig_get_string_by_callout_args P_((OnigCalloutArgs* args)); 991 ONIG_EXTERN 992 const OnigUChar* onig_get_string_end_by_callout_args P_((OnigCalloutArgs* args)); 993 ONIG_EXTERN 994 const OnigUChar* onig_get_start_by_callout_args P_((OnigCalloutArgs* args)); 995 ONIG_EXTERN 996 const OnigUChar* onig_get_right_range_by_callout_args P_((OnigCalloutArgs* args)); 997 ONIG_EXTERN 998 const OnigUChar* onig_get_current_by_callout_args P_((OnigCalloutArgs* args)); 999 ONIG_EXTERN 1000 OnigRegex onig_get_regex_by_callout_args P_((OnigCalloutArgs* args)); 1001 ONIG_EXTERN 1002 unsigned long onig_get_retry_counter_by_callout_args P_((OnigCalloutArgs* args)); 1003 ONIG_EXTERN 1004 int onig_callout_tag_is_exist_at_callout_num P_((OnigRegex reg, int callout_num)); 1005 ONIG_EXTERN 1006 const OnigUChar* onig_get_callout_tag_start P_((OnigRegex reg, int callout_num)); 1007 ONIG_EXTERN 1008 const OnigUChar* onig_get_callout_tag_end P_((OnigRegex reg, int callout_num)); 1009 ONIG_EXTERN 1010 int onig_get_callout_data_dont_clear_old P_((OnigRegex reg, OnigMatchParam* mp, int callout_num, int slot, OnigType* type, OnigValue* val)); 1011 ONIG_EXTERN 1012 int onig_get_callout_data_by_callout_args_self_dont_clear_old P_((OnigCalloutArgs* args, int slot, OnigType* type, OnigValue* val)); 1013 ONIG_EXTERN 1014 int onig_get_callout_data P_((OnigRegex reg, OnigMatchParam* mp, int callout_num, int slot, OnigType* type, OnigValue* val)); 1015 ONIG_EXTERN 1016 int onig_get_callout_data_by_callout_args P_((OnigCalloutArgs* args, int callout_num, int slot, OnigType* type, OnigValue* val)); 1017 ONIG_EXTERN 1018 int onig_get_callout_data_by_callout_args_self P_((OnigCalloutArgs* args, int slot, OnigType* type, OnigValue* val)); 1019 ONIG_EXTERN 1020 int onig_set_callout_data P_((OnigRegex reg, OnigMatchParam* mp, int callout_num, int slot, OnigType type, OnigValue* val)); 1021 ONIG_EXTERN 1022 int onig_set_callout_data_by_callout_args P_((OnigCalloutArgs* args, int callout_num, int slot, OnigType type, OnigValue* val)); 1023 ONIG_EXTERN 1024 int onig_set_callout_data_by_callout_args_self P_((OnigCalloutArgs* args, int slot, OnigType type, OnigValue* val)); 1025 ONIG_EXTERN 1026 int onig_get_capture_range_in_callout P_((OnigCalloutArgs* args, int mem_num, int* begin, int* end)); 1027 ONIG_EXTERN 1028 int onig_get_used_stack_size_in_callout P_((OnigCalloutArgs* args, int* used_num, int* used_bytes)); 1029 1030 /* builtin callout functions */ 1031 ONIG_EXTERN 1032 int onig_builtin_fail P_((OnigCalloutArgs* args, void* user_data)); 1033 ONIG_EXTERN 1034 int onig_builtin_mismatch P_((OnigCalloutArgs* args, void* user_data)); 1035 ONIG_EXTERN 1036 int onig_builtin_error P_((OnigCalloutArgs* args, void* user_data)); 1037 ONIG_EXTERN 1038 int onig_builtin_count P_((OnigCalloutArgs* args, void* user_data)); 1039 ONIG_EXTERN 1040 int onig_builtin_total_count P_((OnigCalloutArgs* args, void* user_data)); 1041 ONIG_EXTERN 1042 int onig_builtin_max P_((OnigCalloutArgs* args, void* user_data)); 1043 ONIG_EXTERN 1044 int onig_builtin_cmp P_((OnigCalloutArgs* args, void* user_data)); 1045 1046 ONIG_EXTERN 1047 int onig_setup_builtin_monitors_by_ascii_encoded_name P_((void* fp)); 1048 1049 #ifdef __cplusplus 1050 } 1051 #endif 1052 1053 #endif /* ONIGURUMA_H */ 1054