1 /*
2 * Copyright (C) 2012 Grigori Goronzy <greg@kinoho.net>
3 *
4 * Permission to use, copy, modify, and/or distribute this software for any
5 * purpose with or without fee is hereby granted, provided that the above
6 * copyright notice and this permission notice appear in all copies.
7 *
8 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15 */
16
17 #ifndef UCDN_H
18 #define UCDN_H
19
20
21
22 #if defined(__GNUC__) && (__GNUC__ >= 4) && !defined(__MINGW32__)
23 # define HB_BEGIN_VISIBILITY _Pragma ("GCC visibility push(hidden)")
24 # define HB_END_VISIBILITY _Pragma ("GCC visibility pop")
25 #else
26 # define HB_BEGIN_VISIBILITY
27 # define HB_END_VISIBILITY
28 #endif
29 #ifdef __cplusplus
30 # define HB_BEGIN_HEADER extern "C" { HB_BEGIN_VISIBILITY
31 # define HB_END_HEADER HB_END_VISIBILITY }
32 #else
33 # define HB_BEGIN_HEADER HB_BEGIN_VISIBILITY
34 # define HB_END_HEADER HB_END_VISIBILITY
35 #endif
36
37 HB_BEGIN_HEADER
38
39 #if !defined (HB_DONT_DEFINE_STDINT)
40
41 #if defined (_SVR4) || defined (SVR4) || defined (__OpenBSD__) || \
42 defined (_sgi) || defined (__sun) || defined (sun) || \
43 defined (__digital__) || defined (__HP_cc)
44 # include <inttypes.h>
45 #elif defined (_AIX)
46 # include <sys/inttypes.h>
47 /* VS 2010 (_MSC_VER 1600) has stdint.h */
48 #elif defined (_MSC_VER) && _MSC_VER < 1600
49 typedef __int8 int8_t;
50 typedef unsigned __int8 uint8_t;
51 typedef __int16 int16_t;
52 typedef unsigned __int16 uint16_t;
53 typedef __int32 int32_t;
54 typedef unsigned __int32 uint32_t;
55 typedef __int64 int64_t;
56 typedef unsigned __int64 uint64_t;
57 #else
58 # include <stdint.h>
59 #endif
60
61 #endif
62
63
64 #define UCDN_EAST_ASIAN_F 0
65 #define UCDN_EAST_ASIAN_H 1
66 #define UCDN_EAST_ASIAN_W 2
67 #define UCDN_EAST_ASIAN_NA 3
68 #define UCDN_EAST_ASIAN_A 4
69 #define UCDN_EAST_ASIAN_N 5
70
71 #define UCDN_SCRIPT_COMMON 0
72 #define UCDN_SCRIPT_LATIN 1
73 #define UCDN_SCRIPT_GREEK 2
74 #define UCDN_SCRIPT_CYRILLIC 3
75 #define UCDN_SCRIPT_ARMENIAN 4
76 #define UCDN_SCRIPT_HEBREW 5
77 #define UCDN_SCRIPT_ARABIC 6
78 #define UCDN_SCRIPT_SYRIAC 7
79 #define UCDN_SCRIPT_THAANA 8
80 #define UCDN_SCRIPT_DEVANAGARI 9
81 #define UCDN_SCRIPT_BENGALI 10
82 #define UCDN_SCRIPT_GURMUKHI 11
83 #define UCDN_SCRIPT_GUJARATI 12
84 #define UCDN_SCRIPT_ORIYA 13
85 #define UCDN_SCRIPT_TAMIL 14
86 #define UCDN_SCRIPT_TELUGU 15
87 #define UCDN_SCRIPT_KANNADA 16
88 #define UCDN_SCRIPT_MALAYALAM 17
89 #define UCDN_SCRIPT_SINHALA 18
90 #define UCDN_SCRIPT_THAI 19
91 #define UCDN_SCRIPT_LAO 20
92 #define UCDN_SCRIPT_TIBETAN 21
93 #define UCDN_SCRIPT_MYANMAR 22
94 #define UCDN_SCRIPT_GEORGIAN 23
95 #define UCDN_SCRIPT_HANGUL 24
96 #define UCDN_SCRIPT_ETHIOPIC 25
97 #define UCDN_SCRIPT_CHEROKEE 26
98 #define UCDN_SCRIPT_CANADIAN_ABORIGINAL 27
99 #define UCDN_SCRIPT_OGHAM 28
100 #define UCDN_SCRIPT_RUNIC 29
101 #define UCDN_SCRIPT_KHMER 30
102 #define UCDN_SCRIPT_MONGOLIAN 31
103 #define UCDN_SCRIPT_HIRAGANA 32
104 #define UCDN_SCRIPT_KATAKANA 33
105 #define UCDN_SCRIPT_BOPOMOFO 34
106 #define UCDN_SCRIPT_HAN 35
107 #define UCDN_SCRIPT_YI 36
108 #define UCDN_SCRIPT_OLD_ITALIC 37
109 #define UCDN_SCRIPT_GOTHIC 38
110 #define UCDN_SCRIPT_DESERET 39
111 #define UCDN_SCRIPT_INHERITED 40
112 #define UCDN_SCRIPT_TAGALOG 41
113 #define UCDN_SCRIPT_HANUNOO 42
114 #define UCDN_SCRIPT_BUHID 43
115 #define UCDN_SCRIPT_TAGBANWA 44
116 #define UCDN_SCRIPT_LIMBU 45
117 #define UCDN_SCRIPT_TAI_LE 46
118 #define UCDN_SCRIPT_LINEAR_B 47
119 #define UCDN_SCRIPT_UGARITIC 48
120 #define UCDN_SCRIPT_SHAVIAN 49
121 #define UCDN_SCRIPT_OSMANYA 50
122 #define UCDN_SCRIPT_CYPRIOT 51
123 #define UCDN_SCRIPT_BRAILLE 52
124 #define UCDN_SCRIPT_BUGINESE 53
125 #define UCDN_SCRIPT_COPTIC 54
126 #define UCDN_SCRIPT_NEW_TAI_LUE 55
127 #define UCDN_SCRIPT_GLAGOLITIC 56
128 #define UCDN_SCRIPT_TIFINAGH 57
129 #define UCDN_SCRIPT_SYLOTI_NAGRI 58
130 #define UCDN_SCRIPT_OLD_PERSIAN 59
131 #define UCDN_SCRIPT_KHAROSHTHI 60
132 #define UCDN_SCRIPT_BALINESE 61
133 #define UCDN_SCRIPT_CUNEIFORM 62
134 #define UCDN_SCRIPT_PHOENICIAN 63
135 #define UCDN_SCRIPT_PHAGS_PA 64
136 #define UCDN_SCRIPT_NKO 65
137 #define UCDN_SCRIPT_SUNDANESE 66
138 #define UCDN_SCRIPT_LEPCHA 67
139 #define UCDN_SCRIPT_OL_CHIKI 68
140 #define UCDN_SCRIPT_VAI 69
141 #define UCDN_SCRIPT_SAURASHTRA 70
142 #define UCDN_SCRIPT_KAYAH_LI 71
143 #define UCDN_SCRIPT_REJANG 72
144 #define UCDN_SCRIPT_LYCIAN 73
145 #define UCDN_SCRIPT_CARIAN 74
146 #define UCDN_SCRIPT_LYDIAN 75
147 #define UCDN_SCRIPT_CHAM 76
148 #define UCDN_SCRIPT_TAI_THAM 77
149 #define UCDN_SCRIPT_TAI_VIET 78
150 #define UCDN_SCRIPT_AVESTAN 79
151 #define UCDN_SCRIPT_EGYPTIAN_HIEROGLYPHS 80
152 #define UCDN_SCRIPT_SAMARITAN 81
153 #define UCDN_SCRIPT_LISU 82
154 #define UCDN_SCRIPT_BAMUM 83
155 #define UCDN_SCRIPT_JAVANESE 84
156 #define UCDN_SCRIPT_MEETEI_MAYEK 85
157 #define UCDN_SCRIPT_IMPERIAL_ARAMAIC 86
158 #define UCDN_SCRIPT_OLD_SOUTH_ARABIAN 87
159 #define UCDN_SCRIPT_INSCRIPTIONAL_PARTHIAN 88
160 #define UCDN_SCRIPT_INSCRIPTIONAL_PAHLAVI 89
161 #define UCDN_SCRIPT_OLD_TURKIC 90
162 #define UCDN_SCRIPT_KAITHI 91
163 #define UCDN_SCRIPT_BATAK 92
164 #define UCDN_SCRIPT_BRAHMI 93
165 #define UCDN_SCRIPT_MANDAIC 94
166 #define UCDN_SCRIPT_CHAKMA 95
167 #define UCDN_SCRIPT_MEROITIC_CURSIVE 96
168 #define UCDN_SCRIPT_MEROITIC_HIEROGLYPHS 97
169 #define UCDN_SCRIPT_MIAO 98
170 #define UCDN_SCRIPT_SHARADA 99
171 #define UCDN_SCRIPT_SORA_SOMPENG 100
172 #define UCDN_SCRIPT_TAKRI 101
173 #define UCDN_SCRIPT_UNKNOWN 102
174 #define UCDN_SCRIPT_BASSA_VAH 103
175 #define UCDN_SCRIPT_CAUCASIAN_ALBANIAN 104
176 #define UCDN_SCRIPT_DUPLOYAN 105
177 #define UCDN_SCRIPT_ELBASAN 106
178 #define UCDN_SCRIPT_GRANTHA 107
179 #define UCDN_SCRIPT_KHOJKI 108
180 #define UCDN_SCRIPT_KHUDAWADI 109
181 #define UCDN_SCRIPT_LINEAR_A 110
182 #define UCDN_SCRIPT_MAHAJANI 111
183 #define UCDN_SCRIPT_MANICHAEAN 112
184 #define UCDN_SCRIPT_MENDE_KIKAKUI 113
185 #define UCDN_SCRIPT_MODI 114
186 #define UCDN_SCRIPT_MRO 115
187 #define UCDN_SCRIPT_NABATAEAN 116
188 #define UCDN_SCRIPT_OLD_NORTH_ARABIAN 117
189 #define UCDN_SCRIPT_OLD_PERMIC 118
190 #define UCDN_SCRIPT_PAHAWH_HMONG 119
191 #define UCDN_SCRIPT_PALMYRENE 120
192 #define UCDN_SCRIPT_PAU_CIN_HAU 121
193 #define UCDN_SCRIPT_PSALTER_PAHLAVI 122
194 #define UCDN_SCRIPT_SIDDHAM 123
195 #define UCDN_SCRIPT_TIRHUTA 124
196 #define UCDN_SCRIPT_WARANG_CITI 125
197 #define UCDN_SCRIPT_AHOM 126
198 #define UCDN_SCRIPT_ANATOLIAN_HIEROGLYPHS 127
199 #define UCDN_SCRIPT_HATRAN 128
200 #define UCDN_SCRIPT_MULTANI 129
201 #define UCDN_SCRIPT_OLD_HUNGARIAN 130
202 #define UCDN_SCRIPT_SIGNWRITING 131
203 #define UCDN_SCRIPT_ADLAM 132
204 #define UCDN_SCRIPT_BHAIKSUKI 133
205 #define UCDN_SCRIPT_MARCHEN 134
206 #define UCDN_SCRIPT_NEWA 135
207 #define UCDN_SCRIPT_OSAGE 136
208 #define UCDN_SCRIPT_TANGUT 137
209
210 #define UCDN_LINEBREAK_CLASS_OP 0
211 #define UCDN_LINEBREAK_CLASS_CL 1
212 #define UCDN_LINEBREAK_CLASS_CP 2
213 #define UCDN_LINEBREAK_CLASS_QU 3
214 #define UCDN_LINEBREAK_CLASS_GL 4
215 #define UCDN_LINEBREAK_CLASS_NS 5
216 #define UCDN_LINEBREAK_CLASS_EX 6
217 #define UCDN_LINEBREAK_CLASS_SY 7
218 #define UCDN_LINEBREAK_CLASS_IS 8
219 #define UCDN_LINEBREAK_CLASS_PR 9
220 #define UCDN_LINEBREAK_CLASS_PO 10
221 #define UCDN_LINEBREAK_CLASS_NU 11
222 #define UCDN_LINEBREAK_CLASS_AL 12
223 #define UCDN_LINEBREAK_CLASS_HL 13
224 #define UCDN_LINEBREAK_CLASS_ID 14
225 #define UCDN_LINEBREAK_CLASS_IN 15
226 #define UCDN_LINEBREAK_CLASS_HY 16
227 #define UCDN_LINEBREAK_CLASS_BA 17
228 #define UCDN_LINEBREAK_CLASS_BB 18
229 #define UCDN_LINEBREAK_CLASS_B2 19
230 #define UCDN_LINEBREAK_CLASS_ZW 20
231 #define UCDN_LINEBREAK_CLASS_CM 21
232 #define UCDN_LINEBREAK_CLASS_WJ 22
233 #define UCDN_LINEBREAK_CLASS_H2 23
234 #define UCDN_LINEBREAK_CLASS_H3 24
235 #define UCDN_LINEBREAK_CLASS_JL 25
236 #define UCDN_LINEBREAK_CLASS_JV 26
237 #define UCDN_LINEBREAK_CLASS_JT 27
238 #define UCDN_LINEBREAK_CLASS_RI 28
239 #define UCDN_LINEBREAK_CLASS_AI 29
240 #define UCDN_LINEBREAK_CLASS_BK 30
241 #define UCDN_LINEBREAK_CLASS_CB 31
242 #define UCDN_LINEBREAK_CLASS_CJ 32
243 #define UCDN_LINEBREAK_CLASS_CR 33
244 #define UCDN_LINEBREAK_CLASS_LF 34
245 #define UCDN_LINEBREAK_CLASS_NL 35
246 #define UCDN_LINEBREAK_CLASS_SA 36
247 #define UCDN_LINEBREAK_CLASS_SG 37
248 #define UCDN_LINEBREAK_CLASS_SP 38
249 #define UCDN_LINEBREAK_CLASS_XX 39
250
251 #define UCDN_GENERAL_CATEGORY_CC 0
252 #define UCDN_GENERAL_CATEGORY_CF 1
253 #define UCDN_GENERAL_CATEGORY_CN 2
254 #define UCDN_GENERAL_CATEGORY_CO 3
255 #define UCDN_GENERAL_CATEGORY_CS 4
256 #define UCDN_GENERAL_CATEGORY_LL 5
257 #define UCDN_GENERAL_CATEGORY_LM 6
258 #define UCDN_GENERAL_CATEGORY_LO 7
259 #define UCDN_GENERAL_CATEGORY_LT 8
260 #define UCDN_GENERAL_CATEGORY_LU 9
261 #define UCDN_GENERAL_CATEGORY_MC 10
262 #define UCDN_GENERAL_CATEGORY_ME 11
263 #define UCDN_GENERAL_CATEGORY_MN 12
264 #define UCDN_GENERAL_CATEGORY_ND 13
265 #define UCDN_GENERAL_CATEGORY_NL 14
266 #define UCDN_GENERAL_CATEGORY_NO 15
267 #define UCDN_GENERAL_CATEGORY_PC 16
268 #define UCDN_GENERAL_CATEGORY_PD 17
269 #define UCDN_GENERAL_CATEGORY_PE 18
270 #define UCDN_GENERAL_CATEGORY_PF 19
271 #define UCDN_GENERAL_CATEGORY_PI 20
272 #define UCDN_GENERAL_CATEGORY_PO 21
273 #define UCDN_GENERAL_CATEGORY_PS 22
274 #define UCDN_GENERAL_CATEGORY_SC 23
275 #define UCDN_GENERAL_CATEGORY_SK 24
276 #define UCDN_GENERAL_CATEGORY_SM 25
277 #define UCDN_GENERAL_CATEGORY_SO 26
278 #define UCDN_GENERAL_CATEGORY_ZL 27
279 #define UCDN_GENERAL_CATEGORY_ZP 28
280 #define UCDN_GENERAL_CATEGORY_ZS 29
281
282 #define UCDN_BIDI_CLASS_L 0
283 #define UCDN_BIDI_CLASS_LRE 1
284 #define UCDN_BIDI_CLASS_LRO 2
285 #define UCDN_BIDI_CLASS_R 3
286 #define UCDN_BIDI_CLASS_AL 4
287 #define UCDN_BIDI_CLASS_RLE 5
288 #define UCDN_BIDI_CLASS_RLO 6
289 #define UCDN_BIDI_CLASS_PDF 7
290 #define UCDN_BIDI_CLASS_EN 8
291 #define UCDN_BIDI_CLASS_ES 9
292 #define UCDN_BIDI_CLASS_ET 10
293 #define UCDN_BIDI_CLASS_AN 11
294 #define UCDN_BIDI_CLASS_CS 12
295 #define UCDN_BIDI_CLASS_NSM 13
296 #define UCDN_BIDI_CLASS_BN 14
297 #define UCDN_BIDI_CLASS_B 15
298 #define UCDN_BIDI_CLASS_S 16
299 #define UCDN_BIDI_CLASS_WS 17
300 #define UCDN_BIDI_CLASS_ON 18
301 #define UCDN_BIDI_CLASS_LRI 19
302 #define UCDN_BIDI_CLASS_RLI 20
303 #define UCDN_BIDI_CLASS_FSI 21
304 #define UCDN_BIDI_CLASS_PDI 22
305
306 #define UCDN_BIDI_PAIRED_BRACKET_TYPE_OPEN 0
307 #define UCDN_BIDI_PAIRED_BRACKET_TYPE_CLOSE 1
308 #define UCDN_BIDI_PAIRED_BRACKET_TYPE_NONE 2
309
310 /**
311 * Return version of the Unicode database.
312 *
313 * @return Unicode database version
314 */
315 const char *ucdn_get_unicode_version(void);
316
317 /**
318 * Get combining class of a codepoint.
319 *
320 * @param code Unicode codepoint
321 * @return combining class value, as defined in UAX#44
322 */
323 int ucdn_get_combining_class(uint32_t code);
324
325 /**
326 * Get east-asian width of a codepoint.
327 *
328 * @param code Unicode codepoint
329 * @return value according to UCDN_EAST_ASIAN_* and as defined in UAX#11.
330 */
331 int ucdn_get_east_asian_width(uint32_t code);
332
333 /**
334 * Get general category of a codepoint.
335 *
336 * @param code Unicode codepoint
337 * @return value according to UCDN_GENERAL_CATEGORY_* and as defined in
338 * UAX#44.
339 */
340 int ucdn_get_general_category(uint32_t code);
341
342 /**
343 * Get bidirectional class of a codepoint.
344 *
345 * @param code Unicode codepoint
346 * @return value according to UCDN_BIDI_CLASS_* and as defined in UAX#44.
347 */
348 int ucdn_get_bidi_class(uint32_t code);
349
350 /**
351 * Get script of a codepoint.
352 *
353 * @param code Unicode codepoint
354 * @return value according to UCDN_SCRIPT_* and as defined in UAX#24.
355 */
356 int ucdn_get_script(uint32_t code);
357
358 /**
359 * Get unresolved linebreak class of a codepoint. This does not take
360 * rule LB1 of UAX#14 into account. See ucdn_get_resolved_linebreak_class()
361 * for resolved linebreak classes.
362 *
363 * @param code Unicode codepoint
364 * @return value according to UCDN_LINEBREAK_* and as defined in UAX#14.
365 */
366 int ucdn_get_linebreak_class(uint32_t code);
367
368 /**
369 * Get resolved linebreak class of a codepoint. This resolves characters
370 * in the AI, SG, XX, SA and CJ classes according to rule LB1 of UAX#14.
371 * In addition the CB class is resolved as the equivalent B2 class and
372 * the NL class is resolved as the equivalent BK class.
373 *
374 * @param code Unicode codepoint
375 * @return value according to UCDN_LINEBREAK_* and as defined in UAX#14.
376 */
377 int ucdn_get_resolved_linebreak_class(uint32_t code);
378
379 /**
380 * Check if codepoint can be mirrored.
381 *
382 * @param code Unicode codepoint
383 * @return 1 if mirrored character exists, otherwise 0
384 */
385 int ucdn_get_mirrored(uint32_t code);
386
387 /**
388 * Mirror a codepoint.
389 *
390 * @param code Unicode codepoint
391 * @return mirrored codepoint or the original codepoint if no
392 * mirrored character exists
393 */
394 uint32_t ucdn_mirror(uint32_t code);
395
396 /**
397 * Get paired bracket for a codepoint.
398 *
399 * @param code Unicode codepoint
400 * @return paired bracket codepoint or the original codepoint if no
401 * paired bracket character exists
402 */
403 uint32_t ucdn_paired_bracket(uint32_t code);
404
405 /**
406 * Get paired bracket type for a codepoint.
407 *
408 * @param code Unicode codepoint
409 * @return value according to UCDN_BIDI_PAIRED_BRACKET_TYPE_* and as defined
410 * in UAX#9.
411 *
412 */
413 int ucdn_paired_bracket_type(uint32_t code);
414
415 /**
416 * Pairwise canonical decomposition of a codepoint. This includes
417 * Hangul Jamo decomposition (see chapter 3.12 of the Unicode core
418 * specification).
419 *
420 * Hangul is decomposed into L and V jamos for LV forms, and an
421 * LV precomposed syllable and a T jamo for LVT forms.
422 *
423 * @param code Unicode codepoint
424 * @param a filled with first codepoint of decomposition
425 * @param b filled with second codepoint of decomposition, or 0
426 * @return success
427 */
428 int ucdn_decompose(uint32_t code, uint32_t *a, uint32_t *b);
429
430 /**
431 * Compatibility decomposition of a codepoint.
432 *
433 * @param code Unicode codepoint
434 * @param decomposed filled with decomposition, must be able to hold 18
435 * characters
436 * @return length of decomposition or 0 in case none exists
437 */
438 int ucdn_compat_decompose(uint32_t code, uint32_t *decomposed);
439
440 /**
441 * Pairwise canonical composition of two codepoints. This includes
442 * Hangul Jamo composition (see chapter 3.12 of the Unicode core
443 * specification).
444 *
445 * Hangul composition expects either L and V jamos, or an LV
446 * precomposed syllable and a T jamo. This is exactly the inverse
447 * of pairwise Hangul decomposition.
448 *
449 * @param code filled with composition
450 * @param a first codepoint
451 * @param b second codepoint
452 * @return success
453 */
454 int ucdn_compose(uint32_t *code, uint32_t a, uint32_t b);
455
456 #ifdef __cplusplus
457 }
458 #endif
459
460 #endif