ICU 49.1.1  49.1.1
uscript.h
Go to the documentation of this file.
00001 /*
00002  **********************************************************************
00003  *   Copyright (C) 1997-2012, International Business Machines
00004  *   Corporation and others.  All Rights Reserved.
00005  **********************************************************************
00006  *
00007  * File USCRIPT.H
00008  *
00009  * Modification History:
00010  *
00011  *   Date        Name        Description
00012  *   07/06/2001    Ram         Creation.
00013  ******************************************************************************
00014  */
00015 
00016 #ifndef USCRIPT_H
00017 #define USCRIPT_H
00018 #include "unicode/utypes.h"
00019 
00046 typedef enum UScriptCode {
00047     /*
00048      * Note: UScriptCode constants and their ISO script code comments
00049      * are parsed by preparseucd.py.
00050      * It matches lines like
00051      *     USCRIPT_<Unicode Script value name> = <integer>,  / * <ISO script code> * /
00052      */
00053 
00055       USCRIPT_INVALID_CODE = -1,
00057       USCRIPT_COMMON       =  0,  /* Zyyy */
00059       USCRIPT_INHERITED    =  1,  /* Zinh */ /* "Code for inherited script", for non-spacing combining marks; also Qaai */
00061       USCRIPT_ARABIC       =  2,  /* Arab */
00063       USCRIPT_ARMENIAN     =  3,  /* Armn */
00065       USCRIPT_BENGALI      =  4,  /* Beng */
00067       USCRIPT_BOPOMOFO     =  5,  /* Bopo */
00069       USCRIPT_CHEROKEE     =  6,  /* Cher */
00071       USCRIPT_COPTIC       =  7,  /* Copt */
00073       USCRIPT_CYRILLIC     =  8,  /* Cyrl */
00075       USCRIPT_DESERET      =  9,  /* Dsrt */
00077       USCRIPT_DEVANAGARI   = 10,  /* Deva */
00079       USCRIPT_ETHIOPIC     = 11,  /* Ethi */
00081       USCRIPT_GEORGIAN     = 12,  /* Geor */
00083       USCRIPT_GOTHIC       = 13,  /* Goth */
00085       USCRIPT_GREEK        = 14,  /* Grek */
00087       USCRIPT_GUJARATI     = 15,  /* Gujr */
00089       USCRIPT_GURMUKHI     = 16,  /* Guru */
00091       USCRIPT_HAN          = 17,  /* Hani */
00093       USCRIPT_HANGUL       = 18,  /* Hang */
00095       USCRIPT_HEBREW       = 19,  /* Hebr */
00097       USCRIPT_HIRAGANA     = 20,  /* Hira */
00099       USCRIPT_KANNADA      = 21,  /* Knda */
00101       USCRIPT_KATAKANA     = 22,  /* Kana */
00103       USCRIPT_KHMER        = 23,  /* Khmr */
00105       USCRIPT_LAO          = 24,  /* Laoo */
00107       USCRIPT_LATIN        = 25,  /* Latn */
00109       USCRIPT_MALAYALAM    = 26,  /* Mlym */
00111       USCRIPT_MONGOLIAN    = 27,  /* Mong */
00113       USCRIPT_MYANMAR      = 28,  /* Mymr */
00115       USCRIPT_OGHAM        = 29,  /* Ogam */
00117       USCRIPT_OLD_ITALIC   = 30,  /* Ital */
00119       USCRIPT_ORIYA        = 31,  /* Orya */
00121       USCRIPT_RUNIC        = 32,  /* Runr */
00123       USCRIPT_SINHALA      = 33,  /* Sinh */
00125       USCRIPT_SYRIAC       = 34,  /* Syrc */
00127       USCRIPT_TAMIL        = 35,  /* Taml */
00129       USCRIPT_TELUGU       = 36,  /* Telu */
00131       USCRIPT_THAANA       = 37,  /* Thaa */
00133       USCRIPT_THAI         = 38,  /* Thai */
00135       USCRIPT_TIBETAN      = 39,  /* Tibt */
00137       USCRIPT_CANADIAN_ABORIGINAL = 40,  /* Cans */
00139       USCRIPT_UCAS         = USCRIPT_CANADIAN_ABORIGINAL,
00141       USCRIPT_YI           = 41,  /* Yiii */
00142       /* New scripts in Unicode 3.2 */
00144       USCRIPT_TAGALOG      = 42,  /* Tglg */
00146       USCRIPT_HANUNOO      = 43,  /* Hano */
00148       USCRIPT_BUHID        = 44,  /* Buhd */
00150       USCRIPT_TAGBANWA     = 45,  /* Tagb */
00151 
00152       /* New scripts in Unicode 4 */
00154       USCRIPT_BRAILLE      = 46,  /* Brai */
00156       USCRIPT_CYPRIOT      = 47,  /* Cprt */
00158       USCRIPT_LIMBU        = 48,  /* Limb */
00160       USCRIPT_LINEAR_B     = 49,  /* Linb */
00162       USCRIPT_OSMANYA      = 50,  /* Osma */
00164       USCRIPT_SHAVIAN      = 51,  /* Shaw */
00166       USCRIPT_TAI_LE       = 52,  /* Tale */
00168       USCRIPT_UGARITIC     = 53,  /* Ugar */
00169 
00171       USCRIPT_KATAKANA_OR_HIRAGANA = 54,/*Hrkt */
00172 
00173       /* New scripts in Unicode 4.1 */
00175       USCRIPT_BUGINESE      = 55, /* Bugi */
00177       USCRIPT_GLAGOLITIC    = 56, /* Glag */
00179       USCRIPT_KHAROSHTHI    = 57, /* Khar */
00181       USCRIPT_SYLOTI_NAGRI  = 58, /* Sylo */
00183       USCRIPT_NEW_TAI_LUE   = 59, /* Talu */
00185       USCRIPT_TIFINAGH      = 60, /* Tfng */
00187       USCRIPT_OLD_PERSIAN   = 61, /* Xpeo */
00188 
00189       /* New script codes from ISO 15924 */
00191       USCRIPT_BALINESE                      = 62, /* Bali */
00193       USCRIPT_BATAK                         = 63, /* Batk */
00195       USCRIPT_BLISSYMBOLS                   = 64, /* Blis */
00197       USCRIPT_BRAHMI                        = 65, /* Brah */
00199       USCRIPT_CHAM                          = 66, /* Cham */
00201       USCRIPT_CIRTH                         = 67, /* Cirt */
00203       USCRIPT_OLD_CHURCH_SLAVONIC_CYRILLIC  = 68, /* Cyrs */
00205       USCRIPT_DEMOTIC_EGYPTIAN              = 69, /* Egyd */
00207       USCRIPT_HIERATIC_EGYPTIAN             = 70, /* Egyh */
00209       USCRIPT_EGYPTIAN_HIEROGLYPHS          = 71, /* Egyp */
00211       USCRIPT_KHUTSURI                      = 72, /* Geok */
00213       USCRIPT_SIMPLIFIED_HAN                = 73, /* Hans */
00215       USCRIPT_TRADITIONAL_HAN               = 74, /* Hant */
00217       USCRIPT_PAHAWH_HMONG                  = 75, /* Hmng */
00219       USCRIPT_OLD_HUNGARIAN                 = 76, /* Hung */
00221       USCRIPT_HARAPPAN_INDUS                = 77, /* Inds */
00223       USCRIPT_JAVANESE                      = 78, /* Java */
00225       USCRIPT_KAYAH_LI                      = 79, /* Kali */
00227       USCRIPT_LATIN_FRAKTUR                 = 80, /* Latf */
00229       USCRIPT_LATIN_GAELIC                  = 81, /* Latg */
00231       USCRIPT_LEPCHA                        = 82, /* Lepc */
00233       USCRIPT_LINEAR_A                      = 83, /* Lina */
00235       USCRIPT_MANDAIC                       = 84, /* Mand */
00237       USCRIPT_MANDAEAN                      = USCRIPT_MANDAIC,
00239       USCRIPT_MAYAN_HIEROGLYPHS             = 85, /* Maya */
00241       USCRIPT_MEROITIC_HIEROGLYPHS          = 86, /* Mero */
00243       USCRIPT_MEROITIC                      = USCRIPT_MEROITIC_HIEROGLYPHS,
00245       USCRIPT_NKO                           = 87, /* Nkoo */
00247       USCRIPT_ORKHON                        = 88, /* Orkh */
00249       USCRIPT_OLD_PERMIC                    = 89, /* Perm */
00251       USCRIPT_PHAGS_PA                      = 90, /* Phag */
00253       USCRIPT_PHOENICIAN                    = 91, /* Phnx */
00255       USCRIPT_PHONETIC_POLLARD              = 92, /* Plrd */
00257       USCRIPT_RONGORONGO                    = 93, /* Roro */
00259       USCRIPT_SARATI                        = 94, /* Sara */
00261       USCRIPT_ESTRANGELO_SYRIAC             = 95, /* Syre */
00263       USCRIPT_WESTERN_SYRIAC                = 96, /* Syrj */
00265       USCRIPT_EASTERN_SYRIAC                = 97, /* Syrn */
00267       USCRIPT_TENGWAR                       = 98, /* Teng */
00269       USCRIPT_VAI                           = 99, /* Vaii */
00271       USCRIPT_VISIBLE_SPEECH                = 100,/* Visp */
00273       USCRIPT_CUNEIFORM                     = 101,/* Xsux */
00275       USCRIPT_UNWRITTEN_LANGUAGES           = 102,/* Zxxx */
00277       USCRIPT_UNKNOWN                       = 103,/* Zzzz */ /* Unknown="Code for uncoded script", for unassigned code points */
00278 
00279       /* New script codes from ISO 15924 */
00281       USCRIPT_CARIAN                        = 104,/* Cari */
00283       USCRIPT_JAPANESE                      = 105,/* Jpan */
00285       USCRIPT_LANNA                         = 106,/* Lana */
00287       USCRIPT_LYCIAN                        = 107,/* Lyci */
00289       USCRIPT_LYDIAN                        = 108,/* Lydi */
00291       USCRIPT_OL_CHIKI                      = 109,/* Olck */
00293       USCRIPT_REJANG                        = 110,/* Rjng */
00295       USCRIPT_SAURASHTRA                    = 111,/* Saur */
00297       USCRIPT_SIGN_WRITING                  = 112,/* Sgnw */
00299       USCRIPT_SUNDANESE                     = 113,/* Sund */
00301       USCRIPT_MOON                          = 114,/* Moon */
00303       USCRIPT_MEITEI_MAYEK                  = 115,/* Mtei */
00304 
00305       /* New script codes from ISO 15924 */
00307       USCRIPT_IMPERIAL_ARAMAIC              = 116,/* Armi */
00309       USCRIPT_AVESTAN                       = 117,/* Avst */
00311       USCRIPT_CHAKMA                        = 118,/* Cakm */
00313       USCRIPT_KOREAN                        = 119,/* Kore */
00315       USCRIPT_KAITHI                        = 120,/* Kthi */
00317       USCRIPT_MANICHAEAN                    = 121,/* Mani */
00319       USCRIPT_INSCRIPTIONAL_PAHLAVI         = 122,/* Phli */
00321       USCRIPT_PSALTER_PAHLAVI               = 123,/* Phlp */
00323       USCRIPT_BOOK_PAHLAVI                  = 124,/* Phlv */
00325       USCRIPT_INSCRIPTIONAL_PARTHIAN        = 125,/* Prti */
00327       USCRIPT_SAMARITAN                     = 126,/* Samr */
00329       USCRIPT_TAI_VIET                      = 127,/* Tavt */
00331       USCRIPT_MATHEMATICAL_NOTATION         = 128,/* Zmth */
00333       USCRIPT_SYMBOLS                       = 129,/* Zsym */
00334 
00335       /* New script codes from ISO 15924 */
00337       USCRIPT_BAMUM                         = 130,/* Bamu */
00339       USCRIPT_LISU                          = 131,/* Lisu */
00341       USCRIPT_NAKHI_GEBA                    = 132,/* Nkgb */
00343       USCRIPT_OLD_SOUTH_ARABIAN             = 133,/* Sarb */
00344 
00345       /* New script codes from ISO 15924 */
00347       USCRIPT_BASSA_VAH                     = 134,/* Bass */
00349       USCRIPT_DUPLOYAN_SHORTAND             = 135,/* Dupl */
00351       USCRIPT_ELBASAN                       = 136,/* Elba */
00353       USCRIPT_GRANTHA                       = 137,/* Gran */
00355       USCRIPT_KPELLE                        = 138,/* Kpel */
00357       USCRIPT_LOMA                          = 139,/* Loma */
00359       USCRIPT_MENDE                         = 140,/* Mend */
00361       USCRIPT_MEROITIC_CURSIVE              = 141,/* Merc */
00363       USCRIPT_OLD_NORTH_ARABIAN             = 142,/* Narb */
00365       USCRIPT_NABATAEAN                     = 143,/* Nbat */
00367       USCRIPT_PALMYRENE                     = 144,/* Palm */
00369       USCRIPT_SINDHI                        = 145,/* Sind */
00371       USCRIPT_WARANG_CITI                   = 146,/* Wara */
00372 
00374       USCRIPT_AFAKA                         = 147,/* Afak */
00376       USCRIPT_JURCHEN                       = 148,/* Jurc */
00378       USCRIPT_MRO                           = 149,/* Mroo */
00380       USCRIPT_NUSHU                         = 150,/* Nshu */
00382       USCRIPT_SHARADA                       = 151,/* Shrd */
00384       USCRIPT_SORA_SOMPENG                  = 152,/* Sora */
00386       USCRIPT_TAKRI                         = 153,/* Takr */
00388       USCRIPT_TANGUT                        = 154,/* Tang */
00390       USCRIPT_WOLEAI                        = 155,/* Wole */
00391 
00393       USCRIPT_ANATOLIAN_HIEROGLYPHS         = 156,/* Hluw */
00395       USCRIPT_KHOJKI                        = 157,/* Khoj */
00397       USCRIPT_TIRHUTA                       = 158,/* Tirh */
00398 
00399       /* Private use codes from Qaaa - Qabx are not supported */
00400 
00402       USCRIPT_CODE_LIMIT    = 159
00403 } UScriptCode;
00404 
00423 U_STABLE int32_t  U_EXPORT2 
00424 uscript_getCode(const char* nameOrAbbrOrLocale,UScriptCode* fillIn,int32_t capacity,UErrorCode *err);
00425 
00434 U_STABLE const char*  U_EXPORT2 
00435 uscript_getName(UScriptCode scriptCode);
00436 
00445 U_STABLE const char*  U_EXPORT2 
00446 uscript_getShortName(UScriptCode scriptCode);
00447 
00456 U_STABLE UScriptCode  U_EXPORT2 
00457 uscript_getScript(UChar32 codepoint, UErrorCode *err);
00458 
00459 #ifndef U_HIDE_DRAFT_API
00460 
00475 U_DRAFT UBool U_EXPORT2
00476 uscript_hasScript(UChar32 c, UScriptCode sc);
00477 
00509 U_DRAFT int32_t U_EXPORT2
00510 uscript_getScriptExtensions(UChar32 c,
00511                             UScriptCode *scripts, int32_t capacity,
00512                             UErrorCode *errorCode);
00513 #endif  /* U_HIDE_DRAFT_API */
00514 
00515 #endif