ICU 66.0.1  66.0.1
uchar.h
Go to the documentation of this file.
1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 **********************************************************************
5 * Copyright (C) 1997-2016, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 **********************************************************************
8 *
9 * File UCHAR.H
10 *
11 * Modification History:
12 *
13 * Date Name Description
14 * 04/02/97 aliu Creation.
15 * 03/29/99 helena Updated for C APIs.
16 * 4/15/99 Madhu Updated for C Implementation and Javadoc
17 * 5/20/99 Madhu Added the function u_getVersion()
18 * 8/19/1999 srl Upgraded scripts to Unicode 3.0
19 * 8/27/1999 schererm UCharDirection constants: U_...
20 * 11/11/1999 weiv added u_isalnum(), cleaned comments
21 * 01/11/2000 helena Renamed u_getVersion to u_getUnicodeVersion().
22 ******************************************************************************
23 */
24 
25 #ifndef UCHAR_H
26 #define UCHAR_H
27 
28 #include "unicode/utypes.h"
29 #include "unicode/stringoptions.h"
30 #include "unicode/ucpmap.h"
31 
32 #if !defined(USET_DEFINED) && !defined(U_IN_DOXYGEN)
33 
34 #define USET_DEFINED
35 
44 typedef struct USet USet;
45 
46 #endif
47 
48 
50 
51 /*==========================================================================*/
52 /* Unicode version number */
53 /*==========================================================================*/
63 #define U_UNICODE_VERSION "13.0"
64 
157 #define UCHAR_MIN_VALUE 0
158 
167 #define UCHAR_MAX_VALUE 0x10ffff
168 
173 #define U_MASK(x) ((uint32_t)1<<(x))
174 
195 typedef enum UProperty {
196  /*
197  * Note: UProperty constants are parsed by preparseucd.py.
198  * It matches lines like
199  * UCHAR_<Unicode property name>=<integer>,
200  */
201 
202  /* Note: Place UCHAR_ALPHABETIC before UCHAR_BINARY_START so that
203  debuggers display UCHAR_ALPHABETIC as the symbolic name for 0,
204  rather than UCHAR_BINARY_START. Likewise for other *_START
205  identifiers. */
206 
486 #ifndef U_HIDE_DEPRECATED_API
487 
492 #endif // U_HIDE_DEPRECATED_API
493 
501  UCHAR_BLOCK=0x1001,
529  UCHAR_SCRIPT=0x100A,
601 #ifndef U_HIDE_DEPRECATED_API
602 
607 #endif // U_HIDE_DEPRECATED_API
608 
620 #ifndef U_HIDE_DEPRECATED_API
621 
626 #endif // U_HIDE_DEPRECATED_API
627 
633 #ifndef U_HIDE_DEPRECATED_API
634 
639 #endif // U_HIDE_DEPRECATED_API
640 
643  UCHAR_AGE=0x4000,
652 #ifndef U_HIDE_DEPRECATED_API
653 
656 #endif /* U_HIDE_DEPRECATED_API */
657 
662  UCHAR_NAME=0x4005,
678 #ifndef U_HIDE_DEPRECATED_API
679 
684 #endif /* U_HIDE_DEPRECATED_API */
685 
691 #ifndef U_HIDE_DEPRECATED_API
692 
697 #endif // U_HIDE_DEPRECATED_API
698 
707 #ifndef U_HIDE_DEPRECATED_API
708 
713 #endif // U_HIDE_DEPRECATED_API
714 
717 } UProperty;
718 
724 typedef enum UCharCategory
725 {
726  /*
727  * Note: UCharCategory constants and their API comments are parsed by preparseucd.py.
728  * It matches pairs of lines like
729  * / ** <Unicode 2-letter General_Category value> comment... * /
730  * U_<[A-Z_]+> = <integer>,
731  */
732 
803 } UCharCategory;
804 
819 #define U_GC_CN_MASK U_MASK(U_GENERAL_OTHER_TYPES)
820 
822 #define U_GC_LU_MASK U_MASK(U_UPPERCASE_LETTER)
823 
824 #define U_GC_LL_MASK U_MASK(U_LOWERCASE_LETTER)
825 
826 #define U_GC_LT_MASK U_MASK(U_TITLECASE_LETTER)
827 
828 #define U_GC_LM_MASK U_MASK(U_MODIFIER_LETTER)
829 
830 #define U_GC_LO_MASK U_MASK(U_OTHER_LETTER)
831 
833 #define U_GC_MN_MASK U_MASK(U_NON_SPACING_MARK)
834 
835 #define U_GC_ME_MASK U_MASK(U_ENCLOSING_MARK)
836 
837 #define U_GC_MC_MASK U_MASK(U_COMBINING_SPACING_MARK)
838 
840 #define U_GC_ND_MASK U_MASK(U_DECIMAL_DIGIT_NUMBER)
841 
842 #define U_GC_NL_MASK U_MASK(U_LETTER_NUMBER)
843 
844 #define U_GC_NO_MASK U_MASK(U_OTHER_NUMBER)
845 
847 #define U_GC_ZS_MASK U_MASK(U_SPACE_SEPARATOR)
848 
849 #define U_GC_ZL_MASK U_MASK(U_LINE_SEPARATOR)
850 
851 #define U_GC_ZP_MASK U_MASK(U_PARAGRAPH_SEPARATOR)
852 
854 #define U_GC_CC_MASK U_MASK(U_CONTROL_CHAR)
855 
856 #define U_GC_CF_MASK U_MASK(U_FORMAT_CHAR)
857 
858 #define U_GC_CO_MASK U_MASK(U_PRIVATE_USE_CHAR)
859 
860 #define U_GC_CS_MASK U_MASK(U_SURROGATE)
861 
863 #define U_GC_PD_MASK U_MASK(U_DASH_PUNCTUATION)
864 
865 #define U_GC_PS_MASK U_MASK(U_START_PUNCTUATION)
866 
867 #define U_GC_PE_MASK U_MASK(U_END_PUNCTUATION)
868 
869 #define U_GC_PC_MASK U_MASK(U_CONNECTOR_PUNCTUATION)
870 
871 #define U_GC_PO_MASK U_MASK(U_OTHER_PUNCTUATION)
872 
874 #define U_GC_SM_MASK U_MASK(U_MATH_SYMBOL)
875 
876 #define U_GC_SC_MASK U_MASK(U_CURRENCY_SYMBOL)
877 
878 #define U_GC_SK_MASK U_MASK(U_MODIFIER_SYMBOL)
879 
880 #define U_GC_SO_MASK U_MASK(U_OTHER_SYMBOL)
881 
883 #define U_GC_PI_MASK U_MASK(U_INITIAL_PUNCTUATION)
884 
885 #define U_GC_PF_MASK U_MASK(U_FINAL_PUNCTUATION)
886 
887 
889 #define U_GC_L_MASK \
890  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK|U_GC_LM_MASK|U_GC_LO_MASK)
891 
893 #define U_GC_LC_MASK \
894  (U_GC_LU_MASK|U_GC_LL_MASK|U_GC_LT_MASK)
895 
897 #define U_GC_M_MASK (U_GC_MN_MASK|U_GC_ME_MASK|U_GC_MC_MASK)
898 
900 #define U_GC_N_MASK (U_GC_ND_MASK|U_GC_NL_MASK|U_GC_NO_MASK)
901 
903 #define U_GC_Z_MASK (U_GC_ZS_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK)
904 
906 #define U_GC_C_MASK \
907  (U_GC_CN_MASK|U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CO_MASK|U_GC_CS_MASK)
908 
910 #define U_GC_P_MASK \
911  (U_GC_PD_MASK|U_GC_PS_MASK|U_GC_PE_MASK|U_GC_PC_MASK|U_GC_PO_MASK| \
912  U_GC_PI_MASK|U_GC_PF_MASK)
913 
915 #define U_GC_S_MASK (U_GC_SM_MASK|U_GC_SC_MASK|U_GC_SK_MASK|U_GC_SO_MASK)
916 
921 typedef enum UCharDirection {
922  /*
923  * Note: UCharDirection constants and their API comments are parsed by preparseucd.py.
924  * It matches pairs of lines like
925  * / ** <Unicode 1..3-letter Bidi_Class value> comment... * /
926  * U_<[A-Z_]+> = <integer>,
927  */
928 
975 #ifndef U_HIDE_DEPRECATED_API
976 
983 #endif // U_HIDE_DEPRECATED_API
985 
993  /*
994  * Note: UBidiPairedBracketType constants are parsed by preparseucd.py.
995  * It matches lines like
996  * U_BPT_<Unicode Bidi_Paired_Bracket_Type value name>
997  */
998 
1005 #ifndef U_HIDE_DEPRECATED_API
1006 
1012  U_BPT_COUNT /* 3 */
1013 #endif // U_HIDE_DEPRECATED_API
1015 
1021  /*
1022  * Note: UBlockCode constants are parsed by preparseucd.py.
1023  * It matches lines like
1024  * UBLOCK_<Unicode Block value name> = <integer>,
1025  */
1026 
1028  UBLOCK_NO_BLOCK = 0, /*[none]*/ /* Special range indicating No_Block */
1029 
1031  UBLOCK_BASIC_LATIN = 1, /*[0000]*/
1032 
1035 
1038 
1041 
1043  UBLOCK_IPA_EXTENSIONS =5, /*[0250]*/
1044 
1047 
1050 
1055  UBLOCK_GREEK =8, /*[0370]*/
1056 
1058  UBLOCK_CYRILLIC =9, /*[0400]*/
1059 
1061  UBLOCK_ARMENIAN =10, /*[0530]*/
1062 
1064  UBLOCK_HEBREW =11, /*[0590]*/
1065 
1067  UBLOCK_ARABIC =12, /*[0600]*/
1068 
1070  UBLOCK_SYRIAC =13, /*[0700]*/
1071 
1073  UBLOCK_THAANA =14, /*[0780]*/
1074 
1076  UBLOCK_DEVANAGARI =15, /*[0900]*/
1077 
1079  UBLOCK_BENGALI =16, /*[0980]*/
1080 
1082  UBLOCK_GURMUKHI =17, /*[0A00]*/
1083 
1085  UBLOCK_GUJARATI =18, /*[0A80]*/
1086 
1088  UBLOCK_ORIYA =19, /*[0B00]*/
1089 
1091  UBLOCK_TAMIL =20, /*[0B80]*/
1092 
1094  UBLOCK_TELUGU =21, /*[0C00]*/
1095 
1097  UBLOCK_KANNADA =22, /*[0C80]*/
1098 
1100  UBLOCK_MALAYALAM =23, /*[0D00]*/
1101 
1103  UBLOCK_SINHALA =24, /*[0D80]*/
1104 
1106  UBLOCK_THAI =25, /*[0E00]*/
1107 
1109  UBLOCK_LAO =26, /*[0E80]*/
1110 
1112  UBLOCK_TIBETAN =27, /*[0F00]*/
1113 
1115  UBLOCK_MYANMAR =28, /*[1000]*/
1116 
1118  UBLOCK_GEORGIAN =29, /*[10A0]*/
1119 
1121  UBLOCK_HANGUL_JAMO =30, /*[1100]*/
1122 
1124  UBLOCK_ETHIOPIC =31, /*[1200]*/
1125 
1127  UBLOCK_CHEROKEE =32, /*[13A0]*/
1128 
1131 
1133  UBLOCK_OGHAM =34, /*[1680]*/
1134 
1136  UBLOCK_RUNIC =35, /*[16A0]*/
1137 
1139  UBLOCK_KHMER =36, /*[1780]*/
1140 
1142  UBLOCK_MONGOLIAN =37, /*[1800]*/
1143 
1146 
1148  UBLOCK_GREEK_EXTENDED =39, /*[1F00]*/
1149 
1152 
1155 
1157  UBLOCK_CURRENCY_SYMBOLS =42, /*[20A0]*/
1158 
1164 
1167 
1169  UBLOCK_NUMBER_FORMS =45, /*[2150]*/
1170 
1172  UBLOCK_ARROWS =46, /*[2190]*/
1173 
1176 
1179 
1181  UBLOCK_CONTROL_PICTURES =49, /*[2400]*/
1182 
1185 
1188 
1190  UBLOCK_BOX_DRAWING =52, /*[2500]*/
1191 
1193  UBLOCK_BLOCK_ELEMENTS =53, /*[2580]*/
1194 
1196  UBLOCK_GEOMETRIC_SHAPES =54, /*[25A0]*/
1197 
1200 
1202  UBLOCK_DINGBATS =56, /*[2700]*/
1203 
1205  UBLOCK_BRAILLE_PATTERNS =57, /*[2800]*/
1206 
1209 
1211  UBLOCK_KANGXI_RADICALS =59, /*[2F00]*/
1212 
1215 
1218 
1220  UBLOCK_HIRAGANA =62, /*[3040]*/
1221 
1223  UBLOCK_KATAKANA =63, /*[30A0]*/
1224 
1226  UBLOCK_BOPOMOFO =64, /*[3100]*/
1227 
1230 
1232  UBLOCK_KANBUN =66, /*[3190]*/
1233 
1236 
1239 
1242 
1245 
1248 
1250  UBLOCK_YI_SYLLABLES =72, /*[A000]*/
1251 
1253  UBLOCK_YI_RADICALS =73, /*[A490]*/
1254 
1256  UBLOCK_HANGUL_SYLLABLES =74, /*[AC00]*/
1257 
1259  UBLOCK_HIGH_SURROGATES =75, /*[D800]*/
1260 
1263 
1265  UBLOCK_LOW_SURROGATES =77, /*[DC00]*/
1266 
1276  UBLOCK_PRIVATE_USE_AREA =78, /*[E000]*/
1287 
1290 
1293 
1296 
1299 
1302 
1305 
1308 
1310  UBLOCK_SPECIALS =86, /*[FFF0]*/
1311 
1314 
1315  /* New blocks in Unicode 3.1 */
1316 
1318  UBLOCK_OLD_ITALIC = 88, /*[10300]*/
1320  UBLOCK_GOTHIC = 89, /*[10330]*/
1322  UBLOCK_DESERET = 90, /*[10400]*/
1326  UBLOCK_MUSICAL_SYMBOLS = 92, /*[1D100]*/
1334  UBLOCK_TAGS = 96, /*[E0000]*/
1335 
1336  /* New blocks in Unicode 3.2 */
1337 
1346  UBLOCK_TAGALOG = 98, /*[1700]*/
1348  UBLOCK_HANUNOO = 99, /*[1720]*/
1350  UBLOCK_BUHID = 100, /*[1740]*/
1352  UBLOCK_TAGBANWA = 101, /*[1760]*/
1366  UBLOCK_VARIATION_SELECTORS = 108, /*[FE00]*/
1371 
1372  /* New blocks in Unicode 4 */
1373 
1375  UBLOCK_LIMBU = 111, /*[1900]*/
1377  UBLOCK_TAI_LE = 112, /*[1950]*/
1379  UBLOCK_KHMER_SYMBOLS = 113, /*[19E0]*/
1381  UBLOCK_PHONETIC_EXTENSIONS = 114, /*[1D00]*/
1387  UBLOCK_LINEAR_B_SYLLABARY = 117, /*[10000]*/
1389  UBLOCK_LINEAR_B_IDEOGRAMS = 118, /*[10080]*/
1391  UBLOCK_AEGEAN_NUMBERS = 119, /*[10100]*/
1393  UBLOCK_UGARITIC = 120, /*[10380]*/
1395  UBLOCK_SHAVIAN = 121, /*[10450]*/
1397  UBLOCK_OSMANYA = 122, /*[10480]*/
1399  UBLOCK_CYPRIOT_SYLLABARY = 123, /*[10800]*/
1401  UBLOCK_TAI_XUAN_JING_SYMBOLS = 124, /*[1D300]*/
1404 
1405  /* New blocks in Unicode 4.1 */
1406 
1410  UBLOCK_ANCIENT_GREEK_NUMBERS = 127, /*[10140]*/
1412  UBLOCK_ARABIC_SUPPLEMENT = 128, /*[0750]*/
1414  UBLOCK_BUGINESE = 129, /*[1A00]*/
1416  UBLOCK_CJK_STROKES = 130, /*[31C0]*/
1420  UBLOCK_COPTIC = 132, /*[2C80]*/
1422  UBLOCK_ETHIOPIC_EXTENDED = 133, /*[2D80]*/
1424  UBLOCK_ETHIOPIC_SUPPLEMENT = 134, /*[1380]*/
1426  UBLOCK_GEORGIAN_SUPPLEMENT = 135, /*[2D00]*/
1428  UBLOCK_GLAGOLITIC = 136, /*[2C00]*/
1430  UBLOCK_KHAROSHTHI = 137, /*[10A00]*/
1434  UBLOCK_NEW_TAI_LUE = 139, /*[1980]*/
1436  UBLOCK_OLD_PERSIAN = 140, /*[103A0]*/
1442  UBLOCK_SYLOTI_NAGRI = 143, /*[A800]*/
1444  UBLOCK_TIFINAGH = 144, /*[2D30]*/
1446  UBLOCK_VERTICAL_FORMS = 145, /*[FE10]*/
1447 
1448  /* New blocks in Unicode 5.0 */
1449 
1451  UBLOCK_NKO = 146, /*[07C0]*/
1453  UBLOCK_BALINESE = 147, /*[1B00]*/
1455  UBLOCK_LATIN_EXTENDED_C = 148, /*[2C60]*/
1457  UBLOCK_LATIN_EXTENDED_D = 149, /*[A720]*/
1459  UBLOCK_PHAGS_PA = 150, /*[A840]*/
1461  UBLOCK_PHOENICIAN = 151, /*[10900]*/
1463  UBLOCK_CUNEIFORM = 152, /*[12000]*/
1467  UBLOCK_COUNTING_ROD_NUMERALS = 154, /*[1D360]*/
1468 
1469  /* New blocks in Unicode 5.1 */
1470 
1472  UBLOCK_SUNDANESE = 155, /*[1B80]*/
1474  UBLOCK_LEPCHA = 156, /*[1C00]*/
1476  UBLOCK_OL_CHIKI = 157, /*[1C50]*/
1478  UBLOCK_CYRILLIC_EXTENDED_A = 158, /*[2DE0]*/
1480  UBLOCK_VAI = 159, /*[A500]*/
1482  UBLOCK_CYRILLIC_EXTENDED_B = 160, /*[A640]*/
1484  UBLOCK_SAURASHTRA = 161, /*[A880]*/
1486  UBLOCK_KAYAH_LI = 162, /*[A900]*/
1488  UBLOCK_REJANG = 163, /*[A930]*/
1490  UBLOCK_CHAM = 164, /*[AA00]*/
1492  UBLOCK_ANCIENT_SYMBOLS = 165, /*[10190]*/
1494  UBLOCK_PHAISTOS_DISC = 166, /*[101D0]*/
1496  UBLOCK_LYCIAN = 167, /*[10280]*/
1498  UBLOCK_CARIAN = 168, /*[102A0]*/
1500  UBLOCK_LYDIAN = 169, /*[10920]*/
1502  UBLOCK_MAHJONG_TILES = 170, /*[1F000]*/
1504  UBLOCK_DOMINO_TILES = 171, /*[1F030]*/
1505 
1506  /* New blocks in Unicode 5.2 */
1507 
1509  UBLOCK_SAMARITAN = 172, /*[0800]*/
1513  UBLOCK_TAI_THAM = 174, /*[1A20]*/
1515  UBLOCK_VEDIC_EXTENSIONS = 175, /*[1CD0]*/
1517  UBLOCK_LISU = 176, /*[A4D0]*/
1519  UBLOCK_BAMUM = 177, /*[A6A0]*/
1523  UBLOCK_DEVANAGARI_EXTENDED = 179, /*[A8E0]*/
1527  UBLOCK_JAVANESE = 181, /*[A980]*/
1529  UBLOCK_MYANMAR_EXTENDED_A = 182, /*[AA60]*/
1531  UBLOCK_TAI_VIET = 183, /*[AA80]*/
1533  UBLOCK_MEETEI_MAYEK = 184, /*[ABC0]*/
1537  UBLOCK_IMPERIAL_ARAMAIC = 186, /*[10840]*/
1539  UBLOCK_OLD_SOUTH_ARABIAN = 187, /*[10A60]*/
1541  UBLOCK_AVESTAN = 188, /*[10B00]*/
1545  UBLOCK_INSCRIPTIONAL_PAHLAVI = 190, /*[10B60]*/
1547  UBLOCK_OLD_TURKIC = 191, /*[10C00]*/
1549  UBLOCK_RUMI_NUMERAL_SYMBOLS = 192, /*[10E60]*/
1551  UBLOCK_KAITHI = 193, /*[11080]*/
1553  UBLOCK_EGYPTIAN_HIEROGLYPHS = 194, /*[13000]*/
1560 
1561  /* New blocks in Unicode 6.0 */
1562 
1564  UBLOCK_MANDAIC = 198, /*[0840]*/
1566  UBLOCK_BATAK = 199, /*[1BC0]*/
1568  UBLOCK_ETHIOPIC_EXTENDED_A = 200, /*[AB00]*/
1570  UBLOCK_BRAHMI = 201, /*[11000]*/
1572  UBLOCK_BAMUM_SUPPLEMENT = 202, /*[16800]*/
1574  UBLOCK_KANA_SUPPLEMENT = 203, /*[1B000]*/
1576  UBLOCK_PLAYING_CARDS = 204, /*[1F0A0]*/
1580  UBLOCK_EMOTICONS = 206, /*[1F600]*/
1584  UBLOCK_ALCHEMICAL_SYMBOLS = 208, /*[1F700]*/
1587 
1588  /* New blocks in Unicode 6.1 */
1589 
1591  UBLOCK_ARABIC_EXTENDED_A = 210, /*[08A0]*/
1595  UBLOCK_CHAKMA = 212, /*[11100]*/
1599  UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/
1601  UBLOCK_MEROITIC_HIEROGLYPHS = 215, /*[10980]*/
1603  UBLOCK_MIAO = 216, /*[16F00]*/
1605  UBLOCK_SHARADA = 217, /*[11180]*/
1607  UBLOCK_SORA_SOMPENG = 218, /*[110D0]*/
1611  UBLOCK_TAKRI = 220, /*[11680]*/
1612 
1613  /* New blocks in Unicode 7.0 */
1614 
1616  UBLOCK_BASSA_VAH = 221, /*[16AD0]*/
1618  UBLOCK_CAUCASIAN_ALBANIAN = 222, /*[10530]*/
1620  UBLOCK_COPTIC_EPACT_NUMBERS = 223, /*[102E0]*/
1624  UBLOCK_DUPLOYAN = 225, /*[1BC00]*/
1626  UBLOCK_ELBASAN = 226, /*[10500]*/
1630  UBLOCK_GRANTHA = 228, /*[11300]*/
1632  UBLOCK_KHOJKI = 229, /*[11200]*/
1634  UBLOCK_KHUDAWADI = 230, /*[112B0]*/
1636  UBLOCK_LATIN_EXTENDED_E = 231, /*[AB30]*/
1638  UBLOCK_LINEAR_A = 232, /*[10600]*/
1640  UBLOCK_MAHAJANI = 233, /*[11150]*/
1642  UBLOCK_MANICHAEAN = 234, /*[10AC0]*/
1644  UBLOCK_MENDE_KIKAKUI = 235, /*[1E800]*/
1646  UBLOCK_MODI = 236, /*[11600]*/
1648  UBLOCK_MRO = 237, /*[16A40]*/
1650  UBLOCK_MYANMAR_EXTENDED_B = 238, /*[A9E0]*/
1652  UBLOCK_NABATAEAN = 239, /*[10880]*/
1654  UBLOCK_OLD_NORTH_ARABIAN = 240, /*[10A80]*/
1656  UBLOCK_OLD_PERMIC = 241, /*[10350]*/
1658  UBLOCK_ORNAMENTAL_DINGBATS = 242, /*[1F650]*/
1660  UBLOCK_PAHAWH_HMONG = 243, /*[16B00]*/
1662  UBLOCK_PALMYRENE = 244, /*[10860]*/
1664  UBLOCK_PAU_CIN_HAU = 245, /*[11AC0]*/
1666  UBLOCK_PSALTER_PAHLAVI = 246, /*[10B80]*/
1670  UBLOCK_SIDDHAM = 248, /*[11580]*/
1674  UBLOCK_SUPPLEMENTAL_ARROWS_C = 250, /*[1F800]*/
1676  UBLOCK_TIRHUTA = 251, /*[11480]*/
1678  UBLOCK_WARANG_CITI = 252, /*[118A0]*/
1679 
1680  /* New blocks in Unicode 8.0 */
1681 
1683  UBLOCK_AHOM = 253, /*[11700]*/
1685  UBLOCK_ANATOLIAN_HIEROGLYPHS = 254, /*[14400]*/
1687  UBLOCK_CHEROKEE_SUPPLEMENT = 255, /*[AB70]*/
1693  UBLOCK_HATRAN = 258, /*[108E0]*/
1695  UBLOCK_MULTANI = 259, /*[11280]*/
1697  UBLOCK_OLD_HUNGARIAN = 260, /*[10C80]*/
1701  UBLOCK_SUTTON_SIGNWRITING = 262, /*[1D800]*/
1702 
1703  /* New blocks in Unicode 9.0 */
1704 
1706  UBLOCK_ADLAM = 263, /*[1E900]*/
1708  UBLOCK_BHAIKSUKI = 264, /*[11C00]*/
1710  UBLOCK_CYRILLIC_EXTENDED_C = 265, /*[1C80]*/
1712  UBLOCK_GLAGOLITIC_SUPPLEMENT = 266, /*[1E000]*/
1716  UBLOCK_MARCHEN = 268, /*[11C70]*/
1718  UBLOCK_MONGOLIAN_SUPPLEMENT = 269, /*[11660]*/
1720  UBLOCK_NEWA = 270, /*[11400]*/
1722  UBLOCK_OSAGE = 271, /*[104B0]*/
1724  UBLOCK_TANGUT = 272, /*[17000]*/
1726  UBLOCK_TANGUT_COMPONENTS = 273, /*[18800]*/
1727 
1728  // New blocks in Unicode 10.0
1729 
1733  UBLOCK_KANA_EXTENDED_A = 275, /*[1B100]*/
1735  UBLOCK_MASARAM_GONDI = 276, /*[11D00]*/
1737  UBLOCK_NUSHU = 277, /*[1B170]*/
1739  UBLOCK_SOYOMBO = 278, /*[11A50]*/
1741  UBLOCK_SYRIAC_SUPPLEMENT = 279, /*[0860]*/
1743  UBLOCK_ZANABAZAR_SQUARE = 280, /*[11A00]*/
1744 
1745  // New blocks in Unicode 11.0
1746 
1748  UBLOCK_CHESS_SYMBOLS = 281, /*[1FA00]*/
1750  UBLOCK_DOGRA = 282, /*[11800]*/
1752  UBLOCK_GEORGIAN_EXTENDED = 283, /*[1C90]*/
1754  UBLOCK_GUNJALA_GONDI = 284, /*[11D60]*/
1756  UBLOCK_HANIFI_ROHINGYA = 285, /*[10D00]*/
1758  UBLOCK_INDIC_SIYAQ_NUMBERS = 286, /*[1EC70]*/
1760  UBLOCK_MAKASAR = 287, /*[11EE0]*/
1762  UBLOCK_MAYAN_NUMERALS = 288, /*[1D2E0]*/
1764  UBLOCK_MEDEFAIDRIN = 289, /*[16E40]*/
1766  UBLOCK_OLD_SOGDIAN = 290, /*[10F00]*/
1768  UBLOCK_SOGDIAN = 291, /*[10F30]*/
1769 
1770  // New blocks in Unicode 12.0
1771 
1775  UBLOCK_ELYMAIC = 293, /*[10FE0]*/
1777  UBLOCK_NANDINAGARI = 294, /*[119A0]*/
1781  UBLOCK_OTTOMAN_SIYAQ_NUMBERS = 296, /*[1ED00]*/
1783  UBLOCK_SMALL_KANA_EXTENSION = 297, /*[1B130]*/
1787  UBLOCK_TAMIL_SUPPLEMENT = 299, /*[11FC0]*/
1789  UBLOCK_WANCHO = 300, /*[1E2C0]*/
1790 
1791  // New blocks in Unicode 13.0
1792 
1794  UBLOCK_CHORASMIAN = 301, /*[10FB0]*/
1798  UBLOCK_DIVES_AKURU = 303, /*[11900]*/
1800  UBLOCK_KHITAN_SMALL_SCRIPT = 304, /*[18B00]*/
1802  UBLOCK_LISU_SUPPLEMENT = 305, /*[11FB0]*/
1806  UBLOCK_TANGUT_SUPPLEMENT = 307, /*[18D00]*/
1808  UBLOCK_YEZIDI = 308, /*[10E80]*/
1809 
1810 #ifndef U_HIDE_DEPRECATED_API
1811 
1818 #endif // U_HIDE_DEPRECATED_API
1819 
1822 };
1823 
1825 typedef enum UBlockCode UBlockCode;
1826 
1834 typedef enum UEastAsianWidth {
1835  /*
1836  * Note: UEastAsianWidth constants are parsed by preparseucd.py.
1837  * It matches lines like
1838  * U_EA_<Unicode East_Asian_Width value name>
1839  */
1840 
1841  U_EA_NEUTRAL, /*[N]*/
1842  U_EA_AMBIGUOUS, /*[A]*/
1843  U_EA_HALFWIDTH, /*[H]*/
1844  U_EA_FULLWIDTH, /*[F]*/
1845  U_EA_NARROW, /*[Na]*/
1846  U_EA_WIDE, /*[W]*/
1847 #ifndef U_HIDE_DEPRECATED_API
1848 
1855 #endif // U_HIDE_DEPRECATED_API
1856 } UEastAsianWidth;
1857 
1869 typedef enum UCharNameChoice {
1872 #ifndef U_HIDE_DEPRECATED_API
1873 
1879 #endif /* U_HIDE_DEPRECATED_API */
1880 
1884 #ifndef U_HIDE_DEPRECATED_API
1885 
1890 #endif // U_HIDE_DEPRECATED_API
1891 } UCharNameChoice;
1892 
1906 typedef enum UPropertyNameChoice {
1907  U_SHORT_PROPERTY_NAME,
1908  U_LONG_PROPERTY_NAME,
1909 #ifndef U_HIDE_DEPRECATED_API
1910 
1915 #endif // U_HIDE_DEPRECATED_API
1917 
1924 typedef enum UDecompositionType {
1925  /*
1926  * Note: UDecompositionType constants are parsed by preparseucd.py.
1927  * It matches lines like
1928  * U_DT_<Unicode Decomposition_Type value name>
1929  */
1930 
1931  U_DT_NONE, /*[none]*/
1932  U_DT_CANONICAL, /*[can]*/
1933  U_DT_COMPAT, /*[com]*/
1934  U_DT_CIRCLE, /*[enc]*/
1935  U_DT_FINAL, /*[fin]*/
1936  U_DT_FONT, /*[font]*/
1937  U_DT_FRACTION, /*[fra]*/
1938  U_DT_INITIAL, /*[init]*/
1939  U_DT_ISOLATED, /*[iso]*/
1940  U_DT_MEDIAL, /*[med]*/
1941  U_DT_NARROW, /*[nar]*/
1942  U_DT_NOBREAK, /*[nb]*/
1943  U_DT_SMALL, /*[sml]*/
1944  U_DT_SQUARE, /*[sqr]*/
1945  U_DT_SUB, /*[sub]*/
1946  U_DT_SUPER, /*[sup]*/
1947  U_DT_VERTICAL, /*[vert]*/
1948  U_DT_WIDE, /*[wide]*/
1949 #ifndef U_HIDE_DEPRECATED_API
1950 
1956  U_DT_COUNT /* 18 */
1957 #endif // U_HIDE_DEPRECATED_API
1959 
1966 typedef enum UJoiningType {
1967  /*
1968  * Note: UJoiningType constants are parsed by preparseucd.py.
1969  * It matches lines like
1970  * U_JT_<Unicode Joining_Type value name>
1971  */
1972 
1973  U_JT_NON_JOINING, /*[U]*/
1974  U_JT_JOIN_CAUSING, /*[C]*/
1975  U_JT_DUAL_JOINING, /*[D]*/
1976  U_JT_LEFT_JOINING, /*[L]*/
1977  U_JT_RIGHT_JOINING, /*[R]*/
1978  U_JT_TRANSPARENT, /*[T]*/
1979 #ifndef U_HIDE_DEPRECATED_API
1980 
1986  U_JT_COUNT /* 6 */
1987 #endif // U_HIDE_DEPRECATED_API
1988 } UJoiningType;
1989 
1996 typedef enum UJoiningGroup {
1997  /*
1998  * Note: UJoiningGroup constants are parsed by preparseucd.py.
1999  * It matches lines like
2000  * U_JG_<Unicode Joining_Group value name>
2001  */
2002 
2003  U_JG_NO_JOINING_GROUP,
2004  U_JG_AIN,
2005  U_JG_ALAPH,
2006  U_JG_ALEF,
2007  U_JG_BEH,
2008  U_JG_BETH,
2009  U_JG_DAL,
2010  U_JG_DALATH_RISH,
2011  U_JG_E,
2012  U_JG_FEH,
2013  U_JG_FINAL_SEMKATH,
2014  U_JG_GAF,
2015  U_JG_GAMAL,
2016  U_JG_HAH,
2018  U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
2019  U_JG_HE,
2020  U_JG_HEH,
2021  U_JG_HEH_GOAL,
2022  U_JG_HETH,
2023  U_JG_KAF,
2024  U_JG_KAPH,
2025  U_JG_KNOTTED_HEH,
2026  U_JG_LAM,
2027  U_JG_LAMADH,
2028  U_JG_MEEM,
2029  U_JG_MIM,
2030  U_JG_NOON,
2031  U_JG_NUN,
2032  U_JG_PE,
2033  U_JG_QAF,
2034  U_JG_QAPH,
2035  U_JG_REH,
2036  U_JG_REVERSED_PE,
2037  U_JG_SAD,
2038  U_JG_SADHE,
2039  U_JG_SEEN,
2040  U_JG_SEMKATH,
2041  U_JG_SHIN,
2042  U_JG_SWASH_KAF,
2043  U_JG_SYRIAC_WAW,
2044  U_JG_TAH,
2045  U_JG_TAW,
2046  U_JG_TEH_MARBUTA,
2047  U_JG_TETH,
2048  U_JG_WAW,
2049  U_JG_YEH,
2050  U_JG_YEH_BARREE,
2051  U_JG_YEH_WITH_TAIL,
2052  U_JG_YUDH,
2053  U_JG_YUDH_HE,
2054  U_JG_ZAIN,
2109 #ifndef U_HIDE_DEPRECATED_API
2110 
2117 #endif // U_HIDE_DEPRECATED_API
2118 } UJoiningGroup;
2119 
2127  /*
2128  * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
2129  * It matches lines like
2130  * U_GCB_<Unicode Grapheme_Cluster_Break value name>
2131  */
2132 
2133  U_GCB_OTHER = 0, /*[XX]*/
2134  U_GCB_CONTROL = 1, /*[CN]*/
2135  U_GCB_CR = 2, /*[CR]*/
2136  U_GCB_EXTEND = 3, /*[EX]*/
2137  U_GCB_L = 4, /*[L]*/
2138  U_GCB_LF = 5, /*[LF]*/
2139  U_GCB_LV = 6, /*[LV]*/
2140  U_GCB_LVT = 7, /*[LVT]*/
2141  U_GCB_T = 8, /*[T]*/
2142  U_GCB_V = 9, /*[V]*/
2144  U_GCB_SPACING_MARK = 10, /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2146  U_GCB_PREPEND = 11, /*[PP]*/
2148  U_GCB_REGIONAL_INDICATOR = 12, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2150  U_GCB_E_BASE = 13, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2152  U_GCB_E_BASE_GAZ = 14, /*[EBG]*/
2154  U_GCB_E_MODIFIER = 15, /*[EM]*/
2156  U_GCB_GLUE_AFTER_ZWJ = 16, /*[GAZ]*/
2158  U_GCB_ZWJ = 17, /*[ZWJ]*/
2159 
2160 #ifndef U_HIDE_DEPRECATED_API
2161 
2168 #endif // U_HIDE_DEPRECATED_API
2170 
2178 typedef enum UWordBreakValues {
2179  /*
2180  * Note: UWordBreakValues constants are parsed by preparseucd.py.
2181  * It matches lines like
2182  * U_WB_<Unicode Word_Break value name>
2183  */
2184 
2185  U_WB_OTHER = 0, /*[XX]*/
2186  U_WB_ALETTER = 1, /*[LE]*/
2187  U_WB_FORMAT = 2, /*[FO]*/
2188  U_WB_KATAKANA = 3, /*[KA]*/
2189  U_WB_MIDLETTER = 4, /*[ML]*/
2190  U_WB_MIDNUM = 5, /*[MN]*/
2191  U_WB_NUMERIC = 6, /*[NU]*/
2192  U_WB_EXTENDNUMLET = 7, /*[EX]*/
2194  U_WB_CR = 8, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2196  U_WB_EXTEND = 9, /*[Extend]*/
2198  U_WB_LF = 10, /*[LF]*/
2200  U_WB_MIDNUMLET =11, /*[MB]*/
2202  U_WB_NEWLINE =12, /*[NL]*/
2204  U_WB_REGIONAL_INDICATOR = 13, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2206  U_WB_HEBREW_LETTER = 14, /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */
2208  U_WB_SINGLE_QUOTE = 15, /*[SQ]*/
2210  U_WB_DOUBLE_QUOTE = 16, /*[DQ]*/
2212  U_WB_E_BASE = 17, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2214  U_WB_E_BASE_GAZ = 18, /*[EBG]*/
2216  U_WB_E_MODIFIER = 19, /*[EM]*/
2218  U_WB_GLUE_AFTER_ZWJ = 20, /*[GAZ]*/
2220  U_WB_ZWJ = 21, /*[ZWJ]*/
2222  U_WB_WSEGSPACE = 22, /*[WSEGSPACE]*/
2223 
2224 #ifndef U_HIDE_DEPRECATED_API
2225 
2232 #endif // U_HIDE_DEPRECATED_API
2234 
2241 typedef enum USentenceBreak {
2242  /*
2243  * Note: USentenceBreak constants are parsed by preparseucd.py.
2244  * It matches lines like
2245  * U_SB_<Unicode Sentence_Break value name>
2246  */
2247 
2248  U_SB_OTHER = 0, /*[XX]*/
2249  U_SB_ATERM = 1, /*[AT]*/
2250  U_SB_CLOSE = 2, /*[CL]*/
2251  U_SB_FORMAT = 3, /*[FO]*/
2252  U_SB_LOWER = 4, /*[LO]*/
2253  U_SB_NUMERIC = 5, /*[NU]*/
2254  U_SB_OLETTER = 6, /*[LE]*/
2255  U_SB_SEP = 7, /*[SE]*/
2256  U_SB_SP = 8, /*[SP]*/
2257  U_SB_STERM = 9, /*[ST]*/
2258  U_SB_UPPER = 10, /*[UP]*/
2259  U_SB_CR = 11, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2260  U_SB_EXTEND = 12, /*[EX]*/
2261  U_SB_LF = 13, /*[LF]*/
2262  U_SB_SCONTINUE = 14, /*[SC]*/
2263 #ifndef U_HIDE_DEPRECATED_API
2264 
2271 #endif // U_HIDE_DEPRECATED_API
2272 } USentenceBreak;
2273 
2280 typedef enum ULineBreak {
2281  /*
2282  * Note: ULineBreak constants are parsed by preparseucd.py.
2283  * It matches lines like
2284  * U_LB_<Unicode Line_Break value name>
2285  */
2286 
2287  U_LB_UNKNOWN = 0, /*[XX]*/
2288  U_LB_AMBIGUOUS = 1, /*[AI]*/
2289  U_LB_ALPHABETIC = 2, /*[AL]*/
2290  U_LB_BREAK_BOTH = 3, /*[B2]*/
2291  U_LB_BREAK_AFTER = 4, /*[BA]*/
2292  U_LB_BREAK_BEFORE = 5, /*[BB]*/
2293  U_LB_MANDATORY_BREAK = 6, /*[BK]*/
2294  U_LB_CONTINGENT_BREAK = 7, /*[CB]*/
2295  U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
2296  U_LB_COMBINING_MARK = 9, /*[CM]*/
2297  U_LB_CARRIAGE_RETURN = 10, /*[CR]*/
2298  U_LB_EXCLAMATION = 11, /*[EX]*/
2299  U_LB_GLUE = 12, /*[GL]*/
2300  U_LB_HYPHEN = 13, /*[HY]*/
2301  U_LB_IDEOGRAPHIC = 14, /*[ID]*/
2303  U_LB_INSEPARABLE = 15, /*[IN]*/
2304  U_LB_INSEPERABLE = U_LB_INSEPARABLE,
2305  U_LB_INFIX_NUMERIC = 16, /*[IS]*/
2306  U_LB_LINE_FEED = 17, /*[LF]*/
2307  U_LB_NONSTARTER = 18, /*[NS]*/
2308  U_LB_NUMERIC = 19, /*[NU]*/
2309  U_LB_OPEN_PUNCTUATION = 20, /*[OP]*/
2310  U_LB_POSTFIX_NUMERIC = 21, /*[PO]*/
2311  U_LB_PREFIX_NUMERIC = 22, /*[PR]*/
2312  U_LB_QUOTATION = 23, /*[QU]*/
2313  U_LB_COMPLEX_CONTEXT = 24, /*[SA]*/
2314  U_LB_SURROGATE = 25, /*[SG]*/
2315  U_LB_SPACE = 26, /*[SP]*/
2316  U_LB_BREAK_SYMBOLS = 27, /*[SY]*/
2317  U_LB_ZWSPACE = 28, /*[ZW]*/
2319  U_LB_NEXT_LINE = 29, /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
2321  U_LB_WORD_JOINER = 30, /*[WJ]*/
2323  U_LB_H2 = 31, /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
2325  U_LB_H3 = 32, /*[H3]*/
2327  U_LB_JL = 33, /*[JL]*/
2329  U_LB_JT = 34, /*[JT]*/
2331  U_LB_JV = 35, /*[JV]*/
2333  U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
2335  U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
2337  U_LB_HEBREW_LETTER = 38, /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
2339  U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2341  U_LB_E_BASE = 40, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2343  U_LB_E_MODIFIER = 41, /*[EM]*/
2345  U_LB_ZWJ = 42, /*[ZWJ]*/
2346 #ifndef U_HIDE_DEPRECATED_API
2347 
2354 #endif // U_HIDE_DEPRECATED_API
2355 } ULineBreak;
2356 
2363 typedef enum UNumericType {
2364  /*
2365  * Note: UNumericType constants are parsed by preparseucd.py.
2366  * It matches lines like
2367  * U_NT_<Unicode Numeric_Type value name>
2368  */
2369 
2370  U_NT_NONE, /*[None]*/
2371  U_NT_DECIMAL, /*[de]*/
2372  U_NT_DIGIT, /*[di]*/
2373  U_NT_NUMERIC, /*[nu]*/
2374 #ifndef U_HIDE_DEPRECATED_API
2375 
2382 #endif // U_HIDE_DEPRECATED_API
2383 } UNumericType;
2384 
2391 typedef enum UHangulSyllableType {
2392  /*
2393  * Note: UHangulSyllableType constants are parsed by preparseucd.py.
2394  * It matches lines like
2395  * U_HST_<Unicode Hangul_Syllable_Type value name>
2396  */
2397 
2398  U_HST_NOT_APPLICABLE, /*[NA]*/
2399  U_HST_LEADING_JAMO, /*[L]*/
2400  U_HST_VOWEL_JAMO, /*[V]*/
2401  U_HST_TRAILING_JAMO, /*[T]*/
2402  U_HST_LV_SYLLABLE, /*[LV]*/
2403  U_HST_LVT_SYLLABLE, /*[LVT]*/
2404 #ifndef U_HIDE_DEPRECATED_API
2405 
2412 #endif // U_HIDE_DEPRECATED_API
2414 
2422  /*
2423  * Note: UIndicPositionalCategory constants are parsed by preparseucd.py.
2424  * It matches lines like
2425  * U_INPC_<Unicode Indic_Positional_Category value name>
2426  */
2427 
2461 
2469  /*
2470  * Note: UIndicSyllabicCategory constants are parsed by preparseucd.py.
2471  * It matches lines like
2472  * U_INSC_<Unicode Indic_Syllabic_Category value name>
2473  */
2474 
2548 
2555 typedef enum UVerticalOrientation {
2556  /*
2557  * Note: UVerticalOrientation constants are parsed by preparseucd.py.
2558  * It matches lines like
2559  * U_VO_<Unicode Vertical_Orientation value name>
2560  */
2561 
2571 
2599 U_STABLE UBool U_EXPORT2
2601 
2617 U_CAPI const USet * U_EXPORT2
2618 u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode);
2619 
2632 U_STABLE UBool U_EXPORT2
2634 
2647 U_STABLE UBool U_EXPORT2
2649 
2662 U_STABLE UBool U_EXPORT2
2664 
2683 U_STABLE UBool U_EXPORT2
2685 
2724 U_STABLE int32_t U_EXPORT2
2726 
2745 U_STABLE int32_t U_EXPORT2
2747 
2774 U_STABLE int32_t U_EXPORT2
2776 
2792 U_CAPI const UCPMap * U_EXPORT2
2793 u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode);
2794 
2817 U_STABLE double U_EXPORT2
2819 
2827 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
2828 
2852 U_STABLE UBool U_EXPORT2
2853 u_islower(UChar32 c);
2854 
2879 U_STABLE UBool U_EXPORT2
2880 u_isupper(UChar32 c);
2881 
2896 U_STABLE UBool U_EXPORT2
2897 u_istitle(UChar32 c);
2898 
2917 U_STABLE UBool U_EXPORT2
2918 u_isdigit(UChar32 c);
2919 
2938 U_STABLE UBool U_EXPORT2
2939 u_isalpha(UChar32 c);
2940 
2959 U_STABLE UBool U_EXPORT2
2960 u_isalnum(UChar32 c);
2961 
2982 U_STABLE UBool U_EXPORT2
2983 u_isxdigit(UChar32 c);
2984 
2998 U_STABLE UBool U_EXPORT2
2999 u_ispunct(UChar32 c);
3000 
3017 U_STABLE UBool U_EXPORT2
3018 u_isgraph(UChar32 c);
3019 
3045 U_STABLE UBool U_EXPORT2
3046 u_isblank(UChar32 c);
3047 
3070 U_STABLE UBool U_EXPORT2
3071 u_isdefined(UChar32 c);
3072 
3091 U_STABLE UBool U_EXPORT2
3092 u_isspace(UChar32 c);
3093 
3112 U_STABLE UBool U_EXPORT2
3114 
3152 U_STABLE UBool U_EXPORT2
3154 
3176 U_STABLE UBool U_EXPORT2
3177 u_iscntrl(UChar32 c);
3178 
3191 U_STABLE UBool U_EXPORT2
3193 
3209 U_STABLE UBool U_EXPORT2
3210 u_isprint(UChar32 c);
3211 
3229 U_STABLE UBool U_EXPORT2
3230 u_isbase(UChar32 c);
3231 
3248 U_STABLE UCharDirection U_EXPORT2
3250 
3266 U_STABLE UBool U_EXPORT2
3268 
3288 U_STABLE UChar32 U_EXPORT2
3290 
3307 U_STABLE UChar32 U_EXPORT2
3309 
3321 U_STABLE int8_t U_EXPORT2
3322 u_charType(UChar32 c);
3323 
3337 #define U_GET_GC_MASK(c) U_MASK(u_charType(c))
3338 
3356 typedef UBool U_CALLCONV
3357 UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
3358 
3378 U_STABLE void U_EXPORT2
3379 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
3380 
3381 #if !UCONFIG_NO_NORMALIZATION
3382 
3390 U_STABLE uint8_t U_EXPORT2
3392 
3393 #endif
3394 
3418 U_STABLE int32_t U_EXPORT2
3420 
3430 U_STABLE UBlockCode U_EXPORT2
3432 
3465 U_STABLE int32_t U_EXPORT2
3466 u_charName(UChar32 code, UCharNameChoice nameChoice,
3467  char *buffer, int32_t bufferLength,
3468  UErrorCode *pErrorCode);
3469 
3470 #ifndef U_HIDE_DEPRECATED_API
3471 
3489 U_DEPRECATED int32_t U_EXPORT2
3491  char *dest, int32_t destCapacity,
3492  UErrorCode *pErrorCode);
3493 #endif /* U_HIDE_DEPRECATED_API */
3494 
3515 U_STABLE UChar32 U_EXPORT2
3516 u_charFromName(UCharNameChoice nameChoice,
3517  const char *name,
3518  UErrorCode *pErrorCode);
3519 
3537 typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
3538  UChar32 code,
3539  UCharNameChoice nameChoice,
3540  const char *name,
3541  int32_t length);
3542 
3564 U_STABLE void U_EXPORT2
3565 u_enumCharNames(UChar32 start, UChar32 limit,
3566  UEnumCharNamesFn *fn,
3567  void *context,
3568  UCharNameChoice nameChoice,
3569  UErrorCode *pErrorCode);
3570 
3602 U_STABLE const char* U_EXPORT2
3603 u_getPropertyName(UProperty property,
3604  UPropertyNameChoice nameChoice);
3605 
3625 U_STABLE UProperty U_EXPORT2
3626 u_getPropertyEnum(const char* alias);
3627 
3675 U_STABLE const char* U_EXPORT2
3677  int32_t value,
3678  UPropertyNameChoice nameChoice);
3679 
3711 U_STABLE int32_t U_EXPORT2
3713  const char* alias);
3714 
3732 U_STABLE UBool U_EXPORT2
3733 u_isIDStart(UChar32 c);
3734 
3756 U_STABLE UBool U_EXPORT2
3757 u_isIDPart(UChar32 c);
3758 
3779 U_STABLE UBool U_EXPORT2
3781 
3798 U_STABLE UBool U_EXPORT2
3800 
3819 U_STABLE UBool U_EXPORT2
3821 
3844 U_STABLE UChar32 U_EXPORT2
3845 u_tolower(UChar32 c);
3846 
3869 U_STABLE UChar32 U_EXPORT2
3870 u_toupper(UChar32 c);
3871 
3894 U_STABLE UChar32 U_EXPORT2
3895 u_totitle(UChar32 c);
3896 
3919 U_STABLE UChar32 U_EXPORT2
3920 u_foldCase(UChar32 c, uint32_t options);
3921 
3960 U_STABLE int32_t U_EXPORT2
3961 u_digit(UChar32 ch, int8_t radix);
3962 
3991 U_STABLE UChar32 U_EXPORT2
3992 u_forDigit(int32_t digit, int8_t radix);
3993 
4008 U_STABLE void U_EXPORT2
4009 u_charAge(UChar32 c, UVersionInfo versionArray);
4010 
4022 U_STABLE void U_EXPORT2
4023 u_getUnicodeVersion(UVersionInfo versionArray);
4024 
4025 #if !UCONFIG_NO_NORMALIZATION
4026 
4047 U_STABLE int32_t U_EXPORT2
4048 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
4049 
4050 #endif
4051 
4052 
4054 
4055 #endif /*_UCHAR*/
4056 /*eof*/
uint8_t UVersionInfo[U_MAX_VERSION_LENGTH]
The binary form of a version on ICU APIs is an array of 4 uint8_t.
Definition: uversion.h:59
Binary property Ideographic.
Definition: uchar.h:271
Unicode 4.0.1 renames the "Cyrillic Supplementary" block to "Cyrillic Supplement".
Definition: uchar.h:1344
Enumerated property Indic_Positional_Category.
Definition: uchar.h:586
Binary property Changes_When_Lowercased.
Definition: uchar.h:423
Binary property IDS_Binary_Operator (new in Unicode 3.2).
Definition: uchar.h:275
Binary property Case_Ignorable.
Definition: uchar.h:421
UBool u_isUWhiteSpace(UChar32 c)
Check if a code point has the White_Space Unicode property.
UBool u_istitle(UChar32 c)
Determines whether the specified code point is a titlecase letter.
Enumerated property NFC_Quick_Check.
Definition: uchar.h:541
UChar32 u_totitle(UChar32 c)
The given character is mapped to its titlecase equivalent according to UnicodeData.txt; if none is defined, the character itself is returned.
One more than the highest normal UJoiningGroup value.
Definition: uchar.h:2116
Miscellaneous property Script_Extensions (new in Unicode 6.0).
Definition: uchar.h:704
const char * u_getPropertyName(UProperty property, UPropertyNameChoice nameChoice)
Return the Unicode name for a given property, as given in the Unicode database file PropertyAliases...
Same as UBLOCK_PRIVATE_USE_AREA.
Definition: uchar.h:1286
UChar32 u_foldCase(UChar32 c, uint32_t options)
The given character is mapped to its case folding equivalent according to UnicodeData.txt and CaseFolding.txt; if the character has no case folding equivalent, the character itself is returned.
First constant for enumerated/integer Unicode properties.
Definition: uchar.h:498
UChar32 u_getBidiPairedBracket(UChar32 c)
Maps the specified character to its paired bracket character.
Binary property XID_Start.
Definition: uchar.h:328
Binary property Join_Control.
Definition: uchar.h:282
Binary property Logical_Order_Exception (new in Unicode 3.2).
Definition: uchar.h:286
struct UCPMap UCPMap
Abstract map from Unicode code points (U+0000..U+10FFFF) to integer values.
Definition: ucpmap.h:31
Binary property White_Space.
Definition: uchar.h:321
String property Titlecase_Mapping.
Definition: uchar.h:677
#define U_CAPI
This is used to declare a function as a public ICU C API.
Definition: umachine.h:109
One more than the highest normal UPropertyNameChoice value.
Definition: uchar.h:1914
One more than the last constant for enumerated/integer Unicode properties.
Definition: uchar.h:606
Enumerated property Numeric_Type.
Definition: uchar.h:526
Binary property xdigit (a C/POSIX character class).
Definition: uchar.h:417
UBlockCode ublock_getCode(UChar32 c)
Returns the Unicode allocation block that contains the character.
Binary property Alphabetic.
Definition: uchar.h:209
First constant for double Unicode properties.
Definition: uchar.h:632
C API: Bit set option bit constants for various string and character processing functions.
UBool UEnumCharNamesFn(void *context, UChar32 code, UCharNameChoice nameChoice, const char *name, int32_t length)
Type of a callback function for u_enumCharNames() that gets called for each Unicode character with th...
Definition: uchar.h:3537
One more than the highest normal USentenceBreak value.
Definition: uchar.h:2270
UBool u_isgraph(UChar32 c)
Determines whether the specified code point is a "graphic" character (printable, excluding spaces)...
Binary property Emoji_Component.
Definition: uchar.h:468
Binary property Emoji.
Definition: uchar.h:440
Cn "Other, Not Assigned (no characters in [UnicodeData.txt] have this property)" (same as U_UNASSIGNE...
Definition: uchar.h:736
String property Simple_Case_Folding.
Definition: uchar.h:665
Binary property NFC_Inert.
Definition: uchar.h:364
Binary property graph (a C/POSIX character class).
Definition: uchar.h:407
String property Bidi_Mirroring_Glyph.
Definition: uchar.h:648
One more than the last constant for bit-mask Unicode properties.
Definition: uchar.h:625
UBool u_isdefined(UChar32 c)
Determines whether the specified code point is "defined", which usually means that it is assigned a c...
Enumerated property Block.
Definition: uchar.h:501
Represents a nonexistent or invalid property or property value.
Definition: uchar.h:716
Renamed from the misspelled "inseperable" in Unicode 4.0.1/ICU 3.0.
Definition: uchar.h:2303
Binary property Radical (new in Unicode 3.2).
Definition: uchar.h:301
UCharDirection
This specifies the language directional property of a character set.
Definition: uchar.h:921
Binary property IDS_Trinary_Operator (new in Unicode 3.2).
Definition: uchar.h:279
Binary property Grapheme_Link (new in Unicode 3.2).
Definition: uchar.h:253
Enumerated property Decomposition_Type.
Definition: uchar.h:507
Binary property Regional_Indicator.
Definition: uchar.h:473
#define U_CALLCONV
Similar to U_CDECL_BEGIN/U_CDECL_END, this qualifier is necessary in callback function typedefs to ma...
Definition: platform.h:870
String property Case_Folding.
Definition: uchar.h:651
String property Name.
Definition: uchar.h:662
This file defines an abstract map from Unicode code points to integer values.
String property Simple_Uppercase_Mapping.
Definition: uchar.h:674
UChar32 u_forDigit(int32_t digit, int8_t radix)
Determines the character representation for a specific digit in the specified radix.
Enumerated property Bidi_Class.
Definition: uchar.h:496
int32_t u_charDigitValue(UChar32 c)
Returns the decimal digit value of a decimal digit character.
Enumerated property General_Category.
Definition: uchar.h:514
Binary property Prepended_Concatenation_Mark.
Definition: uchar.h:478
Sm.
Definition: uchar.h:784
String property Unicode_1_Name.
Definition: uchar.h:683
int32_t u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode)
Get the FC_NFKC_Closure property string for a character.
UNumericType
Numeric Type constants.
Definition: uchar.h:2363
One more than the highest normal UBidiPairedBracketType value.
Definition: uchar.h:1012
Binary property Pattern_White_Space (new in Unicode 4.1).
Definition: uchar.h:392
Enumerated property Vertical_Orientation.
Definition: uchar.h:600
Close paired bracket.
Definition: uchar.h:1004
UBool u_iscntrl(UChar32 c)
Determines whether the specified code point is a control character (as defined by this function)...
UBool UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type)
Callback from u_enumCharTypes(), is called for each contiguous range of code points c (where start<=c...
Definition: uchar.h:3357
Binary property Changes_When_Casefolded.
Definition: uchar.h:429
Binary property NFD_Inert.
Definition: uchar.h:350
Binary property Diacritic.
Definition: uchar.h:234
Binary property Terminal_Punctuation.
Definition: uchar.h:310
U_CAPI const UCPMap * u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode)
Returns an immutable UCPMap for an enumerated/catalog/int-valued property.
UChar32 u_charFromName(UCharNameChoice nameChoice, const char *name, UErrorCode *pErrorCode)
Find a Unicode character by its name and return its code point value.
One more than the highest normal UNumericType value.
Definition: uchar.h:2381
UBool u_isUAlphabetic(UChar32 c)
Check if a code point has the Alphabetic Unicode property.
Enumerated property NFD_Quick_Check.
Definition: uchar.h:535
void u_charAge(UChar32 c, UVersionInfo versionArray)
Get the "age" of the code point.
int32_t u_getPropertyValueEnum(UProperty property, const char *alias)
Return the property value integer for a given value name, as specified in the Unicode database file P...
#define U_CDECL_BEGIN
This is used to begin a declaration of a library private ICU C API.
Definition: umachine.h:84
Binary property STerm (new in Unicode 4.0.1).
Definition: uchar.h:337
Enumerated property Joining_Group.
Definition: uchar.h:517
Binary property ID_Continue.
Definition: uchar.h:264
Binary property Emoji_Presentation.
Definition: uchar.h:447
Open paired bracket.
Definition: uchar.h:1002
Binary property blank (a C/POSIX character class).
Definition: uchar.h:402
Binary property Quotation_Mark.
Definition: uchar.h:297
#define U_DEPRECATED
This is used to declare a function as a deprecated public ICU C API.
Definition: umachine.h:115
Binary property Changes_When_NFKC_Casefolded.
Definition: uchar.h:433
First constant for binary Unicode properties.
Definition: uchar.h:211
Binary property Noncharacter_Code_Point.
Definition: uchar.h:295
Binary property Hyphen.
Definition: uchar.h:259
Enumerated property East_Asian_Width.
Definition: uchar.h:511
ULineBreak
Line Break constants.
Definition: uchar.h:2280
One more than the highest UCharDirection value.
Definition: uchar.h:982
Binary property Full_Composition_Exclusion.
Definition: uchar.h:242
Bitmask property General_Category_Mask.
Definition: uchar.h:617
String property Simple_Titlecase_Mapping.
Definition: uchar.h:671
One more than the highest normal UGraphemeClusterBreak value.
Definition: uchar.h:2167
Unicode 3.2 renames this block to "Combining Diacritical Marks for Symbols".
Definition: uchar.h:1163
int32_t u_digit(UChar32 ch, int8_t radix)
Returns the decimal digit value of the code point in the specified radix.
UDecompositionType
Decomposition Type constants.
Definition: uchar.h:1924
UBool u_isprint(UChar32 c)
Determines whether the specified code point is a printable character.
UBool u_isxdigit(UChar32 c)
Determines whether the specified code point is a hexadecimal digit.
One more than the highest normal ULineBreak value.
Definition: uchar.h:2353
UHangulSyllableType
Hangul Syllable Type constants.
Definition: uchar.h:2391
String property Simple_Lowercase_Mapping.
Definition: uchar.h:668
UIndicPositionalCategory
Indic Positional Category constants.
Definition: uchar.h:2421
Binary property print (a C/POSIX character class).
Definition: uchar.h:412
Binary property Case_Sensitive.
Definition: uchar.h:332
Standard or synthetic character name.
Definition: uchar.h:1881
Binary property Bidi_Mirrored.
Definition: uchar.h:222
UVerticalOrientation
Vertical Orientation constants.
Definition: uchar.h:2555
Binary property NFKC_Inert.
Definition: uchar.h:371
int32_t u_getIntPropertyValue(UChar32 c, UProperty which)
Get the property value for an enumerated or integer Unicode property for a code point.
Binary property Changes_When_Casemapped.
Definition: uchar.h:431
First constant for string Unicode properties.
Definition: uchar.h:645
Binary property Grapheme_Extend (new in Unicode 3.2).
Definition: uchar.h:250
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:425
UGraphemeClusterBreak
Grapheme Cluster Break constants.
Definition: uchar.h:2126
New No_Block value in Unicode 4.
Definition: uchar.h:1028
Binary property Extender.
Definition: uchar.h:238
Double property Numeric_Value.
Definition: uchar.h:630
Binary property Math.
Definition: uchar.h:291
Unicode character name (Name property).
Definition: uchar.h:1871
Binary property Grapheme_Base (new in Unicode 3.2).
Definition: uchar.h:246
Binary property NFKD_Inert.
Definition: uchar.h:357
uint8_t u_getCombiningClass(UChar32 c)
Returns the combining class of the code point as specified in UnicodeData.txt.
UBool u_isbase(UChar32 c)
Non-standard: Determines whether the specified code point is a base character.
UCharCategory
Data for enumerated Unicode general category types.
Definition: uchar.h:724
const char * u_getPropertyValueName(UProperty property, int32_t value, UPropertyNameChoice nameChoice)
Return the Unicode name for a given property value, as given in the Unicode database file PropertyVal...
Enumerated property Sentence_Break (new in Unicode 4.1).
Definition: uchar.h:568
double u_getNumericValue(UChar32 c)
Get the numeric value for a Unicode code point as defined in the Unicode Character Database...
Binary property Lowercase.
Definition: uchar.h:289
UBool u_isJavaIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in a Java identifier...
First constant for bit-mask Unicode properties.
Definition: uchar.h:619
Enumerated property Bidi_Paired_Bracket_Type (new in Unicode 6.3).
Definition: uchar.h:578
UBool u_isspace(UChar32 c)
Determines if the specified character is a space character or not.
USentenceBreak
Sentence Break constants.
Definition: uchar.h:2241
Binary property Extended_Pictographic.
Definition: uchar.h:485
Enumerated property Indic_Syllabic_Category.
Definition: uchar.h:593
Binary property Unified_Ideograph (new in Unicode 3.2).
Definition: uchar.h:314
Enumerated property Canonical_Combining_Class.
Definition: uchar.h:504
UCharNameChoice
Selector constants for u_charName().
Definition: uchar.h:1869
One more than the last constant for binary Unicode properties.
Definition: uchar.h:491
UBool u_isJavaIDPart(UChar32 c)
Determines if the specified character is permissible in a Java identifier.
Enumerated property Script.
Definition: uchar.h:529
Unicode 3.2 renames this block to "Greek and Coptic".
Definition: uchar.h:1055
Binary property Hex_Digit.
Definition: uchar.h:256
String property Uppercase_Mapping.
Definition: uchar.h:687
UPropertyNameChoice
Selector constants for u_getPropertyName() and u_getPropertyValueName().
Definition: uchar.h:1906
String property Lowercase_Mapping.
Definition: uchar.h:659
UCharDirection u_charDirection(UChar32 c)
Returns the bidirectional category value for the code point, which is used in the Unicode bidirection...
UBool u_islower(UChar32 c)
Determines whether the specified code point has the general category "Ll" (lowercase letter)...
uint16_t UChar
The base type for UTF-16 code units and pointers.
Definition: umachine.h:378
#define U_CDECL_END
This is used to end a declaration of a library private ICU C API.
Definition: umachine.h:85
Enumerated property NFKC_Quick_Check.
Definition: uchar.h:544
Not a paired bracket.
Definition: uchar.h:1000
UProperty
Selection constants for Unicode properties.
Definition: uchar.h:195
U_CAPI const USet * u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode)
Returns a frozen USet for a binary property.
void u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context)
Enumerate efficiently all code points with their Unicode general categories.
Enumerated property Hangul_Syllable_Type, new in Unicode 4.
Definition: uchar.h:532
Binary property Dash.
Definition: uchar.h:224
Binary property alnum (a C/POSIX character class).
Definition: uchar.h:397
Cf.
Definition: uchar.h:768
Binary property Variation_Selector (new in Unicode 4.0.1).
Definition: uchar.h:343
One more than the highest normal UHangulSyllableType value.
Definition: uchar.h:2411
UBool u_isUUppercase(UChar32 c)
Check if a code point has the Uppercase Unicode property.
UBlockCode
Constants for Unicode blocks, see the Unicode Data file Blocks.txt.
Definition: uchar.h:1020
Enumerated property Word_Break (new in Unicode 4.1).
Definition: uchar.h:573
Binary property Deprecated (new in Unicode 3.2).
Definition: uchar.h:231
Binary property Bidi_Control.
Definition: uchar.h:217
Binary property XID_Continue.
Definition: uchar.h:325
Same as UBLOCK_PRIVATE_USE.
Definition: uchar.h:1276
UBool u_hasBinaryProperty(UChar32 c, UProperty which)
Check a binary Unicode property for a code point.
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition: utypes.h:415
Binary property Uppercase.
Definition: uchar.h:317
void u_getUnicodeVersion(UVersionInfo versionArray)
Gets the Unicode version information.
Binary property Changes_When_Uppercased.
Definition: uchar.h:425
UJoiningGroup
Joining Group constants.
Definition: uchar.h:1996
Binary property Cased.
Definition: uchar.h:419
struct USet USet
USet is the C API type corresponding to C++ class UnicodeSet.
Definition: uset.h:47
One more than the highest normal UEastAsianWidth value.
Definition: uchar.h:1854
One more than the highest normal UDecompositionType value.
Definition: uchar.h:1956
Cs.
Definition: uchar.h:772
UEastAsianWidth
East Asian Width constants.
Definition: uchar.h:1834
UBool u_isupper(UChar32 c)
Determines whether the specified code point has the general category "Lu" (uppercase letter)...
Enumerated property Trail_Canonical_Combining_Class.
Definition: uchar.h:558
UBool u_isULowercase(UChar32 c)
Check if a code point has the Lowercase Unicode property.
Non-category for unassigned and non-character code points.
Definition: uchar.h:734
UBool u_ispunct(UChar32 c)
Determines whether the specified code point is a punctuation character.
First constant for Unicode properties with unusual value types.
Definition: uchar.h:706
UWordBreakValues
Word Break constants.
Definition: uchar.h:2178
void u_enumCharNames(UChar32 start, UChar32 limit, UEnumCharNamesFn *fn, void *context, UCharNameChoice nameChoice, UErrorCode *pErrorCode)
Enumerate all assigned Unicode characters between the start and limit code points (start inclusive...
int32_t u_getIntPropertyMaxValue(UProperty which)
Get the maximum value for an enumerated/integer/binary Unicode property.
Enumerated property Joining_Type.
Definition: uchar.h:520
One more than the last constant for double Unicode properties.
Definition: uchar.h:638
Binary property Emoji_Modifier.
Definition: uchar.h:454
Basic definitions for ICU, for both C and C++ APIs.
UBool u_isIDPart(UChar32 c)
Determines if the specified character is permissible in an identifier according to Java...
String property Bidi_Paired_Bracket (new in Unicode 6.3).
Definition: uchar.h:690
Enumerated property Lead_Canonical_Combining_Class.
Definition: uchar.h:551
Binary property ASCII_Hex_Digit.
Definition: uchar.h:213
UJoiningType
Joining Type constants.
Definition: uchar.h:1966
UBool u_isIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in an identifier accordin...
Binary property Soft_Dotted (new in Unicode 3.2).
Definition: uchar.h:306
One more than the last constant for string Unicode properties.
Definition: uchar.h:696
Binary property Emoji_Modifier_Base.
Definition: uchar.h:461
Binary Property Segment_Starter.
Definition: uchar.h:382
UChar32 u_toupper(UChar32 c)
The given character is mapped to its uppercase equivalent according to UnicodeData.txt; if the character has no uppercase equivalent, the character itself is returned.
Binary property ID_Start.
Definition: uchar.h:268
int32_t u_getIntPropertyMinValue(UProperty which)
Get the minimum value for an enumerated/integer/binary Unicode property.
UBool u_isJavaSpaceChar(UChar32 c)
Determine if the specified code point is a space character according to Java.
UBool u_isMirrored(UChar32 c)
Determines whether the code point has the Bidi_Mirrored property.
Corrected name from NameAliases.txt.
Definition: uchar.h:1883
Binary property Changes_When_Titlecased.
Definition: uchar.h:427
UChar32 u_tolower(UChar32 c)
The given character is mapped to its lowercase equivalent according to UnicodeData.txt; if the character has no lowercase equivalent, the character itself is returned.
Enumerated property Line_Break.
Definition: uchar.h:523
One more than the highest normal UJoiningType value.
Definition: uchar.h:1986
UBool u_isIDIgnorable(UChar32 c)
Determines if the specified character should be regarded as an ignorable character in an identifier...
Enumerated property Grapheme_Cluster_Break (new in Unicode 4.1).
Definition: uchar.h:563
UBool u_isdigit(UChar32 c)
Determines whether the specified code point is a digit character according to Java.
UProperty u_getPropertyEnum(const char *alias)
Return the UProperty enum for a given property name, as specified in the Unicode database file Proper...
UBool u_isWhitespace(UChar32 c)
Determines if the specified code point is a whitespace character according to Java/ICU.
Deprecated string property ISO_Comment.
Definition: uchar.h:655
UBool u_isblank(UChar32 c)
Determines whether the specified code point is a "blank" or "horizontal space", a character that visi...
UBool u_isalpha(UChar32 c)
Determines whether the specified code point is a letter character.
Binary property Default_Ignorable_Code_Point (new in Unicode 3.2).
Definition: uchar.h:228
One more than the highest normal UBlockCode value.
Definition: uchar.h:1817
The Unicode_1_Name property value which is of little practical value.
Definition: uchar.h:1878
int32_t u_getISOComment(UChar32 c, char *dest, int32_t destCapacity, UErrorCode *pErrorCode)
Returns an empty string.
One higher than the last enum UCharCategory constant.
Definition: uchar.h:802
One more than the highest normal UWordBreakValues value.
Definition: uchar.h:2231
UBool u_isalnum(UChar32 c)
Determines whether the specified code point is an alphanumeric character (letter or digit) according ...
Binary property Pattern_Syntax (new in Unicode 4.1).
Definition: uchar.h:387
One more than the last constant for Unicode properties with unusual value types.
Definition: uchar.h:712
UBidiPairedBracketType
Bidi Paired Bracket Type constants.
Definition: uchar.h:992
One more than the highest normal UCharNameChoice value.
Definition: uchar.h:1889
UChar32 u_charMirror(UChar32 c)
Maps the specified character to a "mirror-image" character.
UBool u_isISOControl(UChar32 c)
Determines whether the specified code point is an ISO control code.
int8_t u_charType(UChar32 c)
Returns the general category value for the code point.
UIndicSyllabicCategory
Indic Syllabic Category constants.
Definition: uchar.h:2468
int32_t u_charName(UChar32 code, UCharNameChoice nameChoice, char *buffer, int32_t bufferLength, UErrorCode *pErrorCode)
Retrieve the name of a Unicode character.
String property Age.
Definition: uchar.h:643
#define U_STABLE
This is used to declare a function as a stable public ICU C API.
Definition: umachine.h:111
int8_t UBool
The ICU boolean type.
Definition: umachine.h:261
Enumerated property NFKD_Quick_Check.
Definition: uchar.h:538