ICU 54.1
54.1
Main Page
Related Pages
Modules
Namespaces
Data Structures
Files
File List
Globals
All
Data Structures
Namespaces
Files
Functions
Variables
Typedefs
Enumerations
Enumerator
Friends
Macros
Groups
Pages
common
unicode
uscript.h
Go to the documentation of this file.
1
/*
2
**********************************************************************
3
* Copyright (C) 1997-2014, International Business Machines
4
* Corporation and others. All Rights Reserved.
5
**********************************************************************
6
*
7
* File USCRIPT.H
8
*
9
* Modification History:
10
*
11
* Date Name Description
12
* 07/06/2001 Ram Creation.
13
******************************************************************************
14
*/
15
16
#ifndef USCRIPT_H
17
#define USCRIPT_H
18
#include "
unicode/utypes.h
"
19
46
typedef
enum
UScriptCode
{
47
/*
48
* Note: UScriptCode constants and their ISO script code comments
49
* are parsed by preparseucd.py.
50
* It matches lines like
51
* USCRIPT_<Unicode Script value name> = <integer>, / * <ISO script code> * /
52
*/
53
55
USCRIPT_INVALID_CODE
= -1,
57
USCRIPT_COMMON
= 0,
/* Zyyy */
59
USCRIPT_INHERITED
= 1,
/* Zinh */
/* "Code for inherited script", for non-spacing combining marks; also Qaai */
61
USCRIPT_ARABIC
= 2,
/* Arab */
63
USCRIPT_ARMENIAN
= 3,
/* Armn */
65
USCRIPT_BENGALI
= 4,
/* Beng */
67
USCRIPT_BOPOMOFO
= 5,
/* Bopo */
69
USCRIPT_CHEROKEE
= 6,
/* Cher */
71
USCRIPT_COPTIC
= 7,
/* Copt */
73
USCRIPT_CYRILLIC
= 8,
/* Cyrl */
75
USCRIPT_DESERET
= 9,
/* Dsrt */
77
USCRIPT_DEVANAGARI
= 10,
/* Deva */
79
USCRIPT_ETHIOPIC
= 11,
/* Ethi */
81
USCRIPT_GEORGIAN
= 12,
/* Geor */
83
USCRIPT_GOTHIC
= 13,
/* Goth */
85
USCRIPT_GREEK
= 14,
/* Grek */
87
USCRIPT_GUJARATI
= 15,
/* Gujr */
89
USCRIPT_GURMUKHI
= 16,
/* Guru */
91
USCRIPT_HAN
= 17,
/* Hani */
93
USCRIPT_HANGUL
= 18,
/* Hang */
95
USCRIPT_HEBREW
= 19,
/* Hebr */
97
USCRIPT_HIRAGANA
= 20,
/* Hira */
99
USCRIPT_KANNADA
= 21,
/* Knda */
101
USCRIPT_KATAKANA
= 22,
/* Kana */
103
USCRIPT_KHMER
= 23,
/* Khmr */
105
USCRIPT_LAO
= 24,
/* Laoo */
107
USCRIPT_LATIN
= 25,
/* Latn */
109
USCRIPT_MALAYALAM
= 26,
/* Mlym */
111
USCRIPT_MONGOLIAN
= 27,
/* Mong */
113
USCRIPT_MYANMAR
= 28,
/* Mymr */
115
USCRIPT_OGHAM
= 29,
/* Ogam */
117
USCRIPT_OLD_ITALIC
= 30,
/* Ital */
119
USCRIPT_ORIYA
= 31,
/* Orya */
121
USCRIPT_RUNIC
= 32,
/* Runr */
123
USCRIPT_SINHALA
= 33,
/* Sinh */
125
USCRIPT_SYRIAC
= 34,
/* Syrc */
127
USCRIPT_TAMIL
= 35,
/* Taml */
129
USCRIPT_TELUGU
= 36,
/* Telu */
131
USCRIPT_THAANA
= 37,
/* Thaa */
133
USCRIPT_THAI
= 38,
/* Thai */
135
USCRIPT_TIBETAN
= 39,
/* Tibt */
137
USCRIPT_CANADIAN_ABORIGINAL
= 40,
/* Cans */
139
USCRIPT_UCAS
=
USCRIPT_CANADIAN_ABORIGINAL
,
141
USCRIPT_YI
= 41,
/* Yiii */
142
/* New scripts in Unicode 3.2 */
144
USCRIPT_TAGALOG
= 42,
/* Tglg */
146
USCRIPT_HANUNOO
= 43,
/* Hano */
148
USCRIPT_BUHID
= 44,
/* Buhd */
150
USCRIPT_TAGBANWA
= 45,
/* Tagb */
151
152
/* New scripts in Unicode 4 */
154
USCRIPT_BRAILLE
= 46,
/* Brai */
156
USCRIPT_CYPRIOT
= 47,
/* Cprt */
158
USCRIPT_LIMBU
= 48,
/* Limb */
160
USCRIPT_LINEAR_B
= 49,
/* Linb */
162
USCRIPT_OSMANYA
= 50,
/* Osma */
164
USCRIPT_SHAVIAN
= 51,
/* Shaw */
166
USCRIPT_TAI_LE
= 52,
/* Tale */
168
USCRIPT_UGARITIC
= 53,
/* Ugar */
169
171
USCRIPT_KATAKANA_OR_HIRAGANA
= 54,
/*Hrkt */
172
173
/* New scripts in Unicode 4.1 */
175
USCRIPT_BUGINESE
= 55,
/* Bugi */
177
USCRIPT_GLAGOLITIC
= 56,
/* Glag */
179
USCRIPT_KHAROSHTHI
= 57,
/* Khar */
181
USCRIPT_SYLOTI_NAGRI
= 58,
/* Sylo */
183
USCRIPT_NEW_TAI_LUE
= 59,
/* Talu */
185
USCRIPT_TIFINAGH
= 60,
/* Tfng */
187
USCRIPT_OLD_PERSIAN
= 61,
/* Xpeo */
188
189
/* New script codes from Unicode and ISO 15924 */
191
USCRIPT_BALINESE
= 62,
/* Bali */
193
USCRIPT_BATAK
= 63,
/* Batk */
195
USCRIPT_BLISSYMBOLS
= 64,
/* Blis */
197
USCRIPT_BRAHMI
= 65,
/* Brah */
199
USCRIPT_CHAM
= 66,
/* Cham */
201
USCRIPT_CIRTH
= 67,
/* Cirt */
203
USCRIPT_OLD_CHURCH_SLAVONIC_CYRILLIC
= 68,
/* Cyrs */
205
USCRIPT_DEMOTIC_EGYPTIAN
= 69,
/* Egyd */
207
USCRIPT_HIERATIC_EGYPTIAN
= 70,
/* Egyh */
209
USCRIPT_EGYPTIAN_HIEROGLYPHS
= 71,
/* Egyp */
211
USCRIPT_KHUTSURI
= 72,
/* Geok */
213
USCRIPT_SIMPLIFIED_HAN
= 73,
/* Hans */
215
USCRIPT_TRADITIONAL_HAN
= 74,
/* Hant */
217
USCRIPT_PAHAWH_HMONG
= 75,
/* Hmng */
219
USCRIPT_OLD_HUNGARIAN
= 76,
/* Hung */
221
USCRIPT_HARAPPAN_INDUS
= 77,
/* Inds */
223
USCRIPT_JAVANESE
= 78,
/* Java */
225
USCRIPT_KAYAH_LI
= 79,
/* Kali */
227
USCRIPT_LATIN_FRAKTUR
= 80,
/* Latf */
229
USCRIPT_LATIN_GAELIC
= 81,
/* Latg */
231
USCRIPT_LEPCHA
= 82,
/* Lepc */
233
USCRIPT_LINEAR_A
= 83,
/* Lina */
235
USCRIPT_MANDAIC
= 84,
/* Mand */
237
USCRIPT_MANDAEAN
=
USCRIPT_MANDAIC
,
239
USCRIPT_MAYAN_HIEROGLYPHS
= 85,
/* Maya */
241
USCRIPT_MEROITIC_HIEROGLYPHS
= 86,
/* Mero */
243
USCRIPT_MEROITIC
=
USCRIPT_MEROITIC_HIEROGLYPHS
,
245
USCRIPT_NKO
= 87,
/* Nkoo */
247
USCRIPT_ORKHON
= 88,
/* Orkh */
249
USCRIPT_OLD_PERMIC
= 89,
/* Perm */
251
USCRIPT_PHAGS_PA
= 90,
/* Phag */
253
USCRIPT_PHOENICIAN
= 91,
/* Phnx */
255
USCRIPT_MIAO
= 92,
/* Plrd */
257
USCRIPT_PHONETIC_POLLARD
=
USCRIPT_MIAO
,
259
USCRIPT_RONGORONGO
= 93,
/* Roro */
261
USCRIPT_SARATI
= 94,
/* Sara */
263
USCRIPT_ESTRANGELO_SYRIAC
= 95,
/* Syre */
265
USCRIPT_WESTERN_SYRIAC
= 96,
/* Syrj */
267
USCRIPT_EASTERN_SYRIAC
= 97,
/* Syrn */
269
USCRIPT_TENGWAR
= 98,
/* Teng */
271
USCRIPT_VAI
= 99,
/* Vaii */
273
USCRIPT_VISIBLE_SPEECH
= 100,
/* Visp */
275
USCRIPT_CUNEIFORM
= 101,
/* Xsux */
277
USCRIPT_UNWRITTEN_LANGUAGES
= 102,
/* Zxxx */
279
USCRIPT_UNKNOWN
= 103,
/* Zzzz */
/* Unknown="Code for uncoded script", for unassigned code points */
280
282
USCRIPT_CARIAN
= 104,
/* Cari */
284
USCRIPT_JAPANESE
= 105,
/* Jpan */
286
USCRIPT_LANNA
= 106,
/* Lana */
288
USCRIPT_LYCIAN
= 107,
/* Lyci */
290
USCRIPT_LYDIAN
= 108,
/* Lydi */
292
USCRIPT_OL_CHIKI
= 109,
/* Olck */
294
USCRIPT_REJANG
= 110,
/* Rjng */
296
USCRIPT_SAURASHTRA
= 111,
/* Saur */
298
USCRIPT_SIGN_WRITING
= 112,
/* Sgnw */
300
USCRIPT_SUNDANESE
= 113,
/* Sund */
302
USCRIPT_MOON
= 114,
/* Moon */
304
USCRIPT_MEITEI_MAYEK
= 115,
/* Mtei */
305
307
USCRIPT_IMPERIAL_ARAMAIC
= 116,
/* Armi */
309
USCRIPT_AVESTAN
= 117,
/* Avst */
311
USCRIPT_CHAKMA
= 118,
/* Cakm */
313
USCRIPT_KOREAN
= 119,
/* Kore */
315
USCRIPT_KAITHI
= 120,
/* Kthi */
317
USCRIPT_MANICHAEAN
= 121,
/* Mani */
319
USCRIPT_INSCRIPTIONAL_PAHLAVI
= 122,
/* Phli */
321
USCRIPT_PSALTER_PAHLAVI
= 123,
/* Phlp */
323
USCRIPT_BOOK_PAHLAVI
= 124,
/* Phlv */
325
USCRIPT_INSCRIPTIONAL_PARTHIAN
= 125,
/* Prti */
327
USCRIPT_SAMARITAN
= 126,
/* Samr */
329
USCRIPT_TAI_VIET
= 127,
/* Tavt */
331
USCRIPT_MATHEMATICAL_NOTATION
= 128,
/* Zmth */
333
USCRIPT_SYMBOLS
= 129,
/* Zsym */
334
336
USCRIPT_BAMUM
= 130,
/* Bamu */
338
USCRIPT_LISU
= 131,
/* Lisu */
340
USCRIPT_NAKHI_GEBA
= 132,
/* Nkgb */
342
USCRIPT_OLD_SOUTH_ARABIAN
= 133,
/* Sarb */
343
345
USCRIPT_BASSA_VAH
= 134,
/* Bass */
347
USCRIPT_DUPLOYAN
= 135,
/* Dupl */
348
#ifndef U_HIDE_DEPRECATED_API
349
350
USCRIPT_DUPLOYAN_SHORTAND
=
USCRIPT_DUPLOYAN
,
351
#endif
/* U_HIDE_DEPRECATED_API */
352
353
USCRIPT_ELBASAN
= 136,
/* Elba */
355
USCRIPT_GRANTHA
= 137,
/* Gran */
357
USCRIPT_KPELLE
= 138,
/* Kpel */
359
USCRIPT_LOMA
= 139,
/* Loma */
361
USCRIPT_MENDE
= 140,
/* Mend */
363
USCRIPT_MEROITIC_CURSIVE
= 141,
/* Merc */
365
USCRIPT_OLD_NORTH_ARABIAN
= 142,
/* Narb */
367
USCRIPT_NABATAEAN
= 143,
/* Nbat */
369
USCRIPT_PALMYRENE
= 144,
/* Palm */
371
USCRIPT_KHUDAWADI
= 145,
/* Sind */
373
USCRIPT_SINDHI
=
USCRIPT_KHUDAWADI
,
375
USCRIPT_WARANG_CITI
= 146,
/* Wara */
376
378
USCRIPT_AFAKA
= 147,
/* Afak */
380
USCRIPT_JURCHEN
= 148,
/* Jurc */
382
USCRIPT_MRO
= 149,
/* Mroo */
384
USCRIPT_NUSHU
= 150,
/* Nshu */
386
USCRIPT_SHARADA
= 151,
/* Shrd */
388
USCRIPT_SORA_SOMPENG
= 152,
/* Sora */
390
USCRIPT_TAKRI
= 153,
/* Takr */
392
USCRIPT_TANGUT
= 154,
/* Tang */
394
USCRIPT_WOLEAI
= 155,
/* Wole */
395
397
USCRIPT_ANATOLIAN_HIEROGLYPHS
= 156,
/* Hluw */
399
USCRIPT_KHOJKI
= 157,
/* Khoj */
401
USCRIPT_TIRHUTA
= 158,
/* Tirh */
402
404
USCRIPT_CAUCASIAN_ALBANIAN
= 159,
/* Aghb */
406
USCRIPT_MAHAJANI
= 160,
/* Mahj */
407
409
USCRIPT_AHOM
= 161,
/* Ahom */
411
USCRIPT_HATRAN
= 162,
/* Hatr */
413
USCRIPT_MODI
= 163,
/* Modi */
415
USCRIPT_MULTANI
= 164,
/* Mult */
417
USCRIPT_PAU_CIN_HAU
= 165,
/* Pauc */
419
USCRIPT_SIDDHAM
= 166,
/* Sidd */
420
421
/* Private use codes from Qaaa - Qabx are not supported */
422
424
USCRIPT_CODE_LIMIT
= 167
425
}
UScriptCode
;
426
446
U_STABLE
int32_t U_EXPORT2
447
uscript_getCode
(
const
char
* nameOrAbbrOrLocale,
UScriptCode
* fillIn,int32_t capacity,
UErrorCode
*err);
448
459
U_STABLE
const
char
* U_EXPORT2
460
uscript_getName
(
UScriptCode
scriptCode);
461
471
U_STABLE
const
char
* U_EXPORT2
472
uscript_getShortName
(
UScriptCode
scriptCode);
473
482
U_STABLE
UScriptCode
U_EXPORT2
483
uscript_getScript
(
UChar32
codepoint,
UErrorCode
*err);
484
500
U_STABLE
UBool
U_EXPORT2
501
uscript_hasScript
(
UChar32
c,
UScriptCode
sc);
502
534
U_STABLE
int32_t U_EXPORT2
535
uscript_getScriptExtensions
(
UChar32
c,
536
UScriptCode
*scripts, int32_t capacity,
537
UErrorCode
*errorCode);
538
546
typedef
enum
UScriptUsage
{
548
USCRIPT_USAGE_NOT_ENCODED
,
550
USCRIPT_USAGE_UNKNOWN
,
552
USCRIPT_USAGE_EXCLUDED
,
554
USCRIPT_USAGE_LIMITED_USE
,
556
USCRIPT_USAGE_ASPIRATIONAL
,
558
USCRIPT_USAGE_RECOMMENDED
559
}
UScriptUsage
;
560
573
U_STABLE
int32_t U_EXPORT2
574
uscript_getSampleString
(
UScriptCode
script,
UChar
*dest, int32_t capacity,
UErrorCode
*pErrorCode);
575
576
#if U_SHOW_CPLUSPLUS_API
577
578
U_NAMESPACE_BEGIN
579
class
UnicodeString;
580
U_NAMESPACE_END
581
591
U_COMMON_API
icu::UnicodeString
U_EXPORT2
592
uscript_getSampleUnicodeString
(
UScriptCode
script);
593
594
#endif
595
605
U_STABLE
UScriptUsage
U_EXPORT2
606
uscript_getUsage
(
UScriptCode
script);
607
616
U_STABLE
UBool
U_EXPORT2
617
uscript_isRightToLeft
(
UScriptCode
script);
618
628
U_STABLE
UBool
U_EXPORT2
629
uscript_breaksBetweenLetters
(
UScriptCode
script);
630
639
U_STABLE
UBool
U_EXPORT2
640
uscript_isCased
(
UScriptCode
script);
641
642
#endif
Generated on Mon Sep 29 2014 13:32:06 for ICU 54.1 by
1.8.1.2