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