ICU 63.2  63.2
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 "11.0"
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 #ifndef U_HIDE_DEPRECATED_API
1771 
1778 #endif // U_HIDE_DEPRECATED_API
1779 
1782 };
1783 
1785 typedef enum UBlockCode UBlockCode;
1786 
1794 typedef enum UEastAsianWidth {
1795  /*
1796  * Note: UEastAsianWidth constants are parsed by preparseucd.py.
1797  * It matches lines like
1798  * U_EA_<Unicode East_Asian_Width value name>
1799  */
1800 
1801  U_EA_NEUTRAL, /*[N]*/
1802  U_EA_AMBIGUOUS, /*[A]*/
1803  U_EA_HALFWIDTH, /*[H]*/
1804  U_EA_FULLWIDTH, /*[F]*/
1805  U_EA_NARROW, /*[Na]*/
1806  U_EA_WIDE, /*[W]*/
1807 #ifndef U_HIDE_DEPRECATED_API
1808 
1815 #endif // U_HIDE_DEPRECATED_API
1816 } UEastAsianWidth;
1817 
1829 typedef enum UCharNameChoice {
1832 #ifndef U_HIDE_DEPRECATED_API
1833 
1839 #endif /* U_HIDE_DEPRECATED_API */
1840 
1844 #ifndef U_HIDE_DEPRECATED_API
1845 
1850 #endif // U_HIDE_DEPRECATED_API
1851 } UCharNameChoice;
1852 
1866 typedef enum UPropertyNameChoice {
1867  U_SHORT_PROPERTY_NAME,
1868  U_LONG_PROPERTY_NAME,
1869 #ifndef U_HIDE_DEPRECATED_API
1870 
1875 #endif // U_HIDE_DEPRECATED_API
1877 
1884 typedef enum UDecompositionType {
1885  /*
1886  * Note: UDecompositionType constants are parsed by preparseucd.py.
1887  * It matches lines like
1888  * U_DT_<Unicode Decomposition_Type value name>
1889  */
1890 
1891  U_DT_NONE, /*[none]*/
1892  U_DT_CANONICAL, /*[can]*/
1893  U_DT_COMPAT, /*[com]*/
1894  U_DT_CIRCLE, /*[enc]*/
1895  U_DT_FINAL, /*[fin]*/
1896  U_DT_FONT, /*[font]*/
1897  U_DT_FRACTION, /*[fra]*/
1898  U_DT_INITIAL, /*[init]*/
1899  U_DT_ISOLATED, /*[iso]*/
1900  U_DT_MEDIAL, /*[med]*/
1901  U_DT_NARROW, /*[nar]*/
1902  U_DT_NOBREAK, /*[nb]*/
1903  U_DT_SMALL, /*[sml]*/
1904  U_DT_SQUARE, /*[sqr]*/
1905  U_DT_SUB, /*[sub]*/
1906  U_DT_SUPER, /*[sup]*/
1907  U_DT_VERTICAL, /*[vert]*/
1908  U_DT_WIDE, /*[wide]*/
1909 #ifndef U_HIDE_DEPRECATED_API
1910 
1916  U_DT_COUNT /* 18 */
1917 #endif // U_HIDE_DEPRECATED_API
1919 
1926 typedef enum UJoiningType {
1927  /*
1928  * Note: UJoiningType constants are parsed by preparseucd.py.
1929  * It matches lines like
1930  * U_JT_<Unicode Joining_Type value name>
1931  */
1932 
1933  U_JT_NON_JOINING, /*[U]*/
1934  U_JT_JOIN_CAUSING, /*[C]*/
1935  U_JT_DUAL_JOINING, /*[D]*/
1936  U_JT_LEFT_JOINING, /*[L]*/
1937  U_JT_RIGHT_JOINING, /*[R]*/
1938  U_JT_TRANSPARENT, /*[T]*/
1939 #ifndef U_HIDE_DEPRECATED_API
1940 
1946  U_JT_COUNT /* 6 */
1947 #endif // U_HIDE_DEPRECATED_API
1948 } UJoiningType;
1949 
1956 typedef enum UJoiningGroup {
1957  /*
1958  * Note: UJoiningGroup constants are parsed by preparseucd.py.
1959  * It matches lines like
1960  * U_JG_<Unicode Joining_Group value name>
1961  */
1962 
1963  U_JG_NO_JOINING_GROUP,
1964  U_JG_AIN,
1965  U_JG_ALAPH,
1966  U_JG_ALEF,
1967  U_JG_BEH,
1968  U_JG_BETH,
1969  U_JG_DAL,
1970  U_JG_DALATH_RISH,
1971  U_JG_E,
1972  U_JG_FEH,
1973  U_JG_FINAL_SEMKATH,
1974  U_JG_GAF,
1975  U_JG_GAMAL,
1976  U_JG_HAH,
1978  U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
1979  U_JG_HE,
1980  U_JG_HEH,
1981  U_JG_HEH_GOAL,
1982  U_JG_HETH,
1983  U_JG_KAF,
1984  U_JG_KAPH,
1985  U_JG_KNOTTED_HEH,
1986  U_JG_LAM,
1987  U_JG_LAMADH,
1988  U_JG_MEEM,
1989  U_JG_MIM,
1990  U_JG_NOON,
1991  U_JG_NUN,
1992  U_JG_PE,
1993  U_JG_QAF,
1994  U_JG_QAPH,
1995  U_JG_REH,
1996  U_JG_REVERSED_PE,
1997  U_JG_SAD,
1998  U_JG_SADHE,
1999  U_JG_SEEN,
2000  U_JG_SEMKATH,
2001  U_JG_SHIN,
2002  U_JG_SWASH_KAF,
2003  U_JG_SYRIAC_WAW,
2004  U_JG_TAH,
2005  U_JG_TAW,
2006  U_JG_TEH_MARBUTA,
2007  U_JG_TETH,
2008  U_JG_WAW,
2009  U_JG_YEH,
2010  U_JG_YEH_BARREE,
2011  U_JG_YEH_WITH_TAIL,
2012  U_JG_YUDH,
2013  U_JG_YUDH_HE,
2014  U_JG_ZAIN,
2069 #ifndef U_HIDE_DEPRECATED_API
2070 
2077 #endif // U_HIDE_DEPRECATED_API
2078 } UJoiningGroup;
2079 
2087  /*
2088  * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
2089  * It matches lines like
2090  * U_GCB_<Unicode Grapheme_Cluster_Break value name>
2091  */
2092 
2093  U_GCB_OTHER = 0, /*[XX]*/
2094  U_GCB_CONTROL = 1, /*[CN]*/
2095  U_GCB_CR = 2, /*[CR]*/
2096  U_GCB_EXTEND = 3, /*[EX]*/
2097  U_GCB_L = 4, /*[L]*/
2098  U_GCB_LF = 5, /*[LF]*/
2099  U_GCB_LV = 6, /*[LV]*/
2100  U_GCB_LVT = 7, /*[LVT]*/
2101  U_GCB_T = 8, /*[T]*/
2102  U_GCB_V = 9, /*[V]*/
2104  U_GCB_SPACING_MARK = 10, /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2106  U_GCB_PREPEND = 11, /*[PP]*/
2108  U_GCB_REGIONAL_INDICATOR = 12, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2110  U_GCB_E_BASE = 13, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2112  U_GCB_E_BASE_GAZ = 14, /*[EBG]*/
2114  U_GCB_E_MODIFIER = 15, /*[EM]*/
2116  U_GCB_GLUE_AFTER_ZWJ = 16, /*[GAZ]*/
2118  U_GCB_ZWJ = 17, /*[ZWJ]*/
2119 
2120 #ifndef U_HIDE_DEPRECATED_API
2121 
2128 #endif // U_HIDE_DEPRECATED_API
2130 
2138 typedef enum UWordBreakValues {
2139  /*
2140  * Note: UWordBreakValues constants are parsed by preparseucd.py.
2141  * It matches lines like
2142  * U_WB_<Unicode Word_Break value name>
2143  */
2144 
2145  U_WB_OTHER = 0, /*[XX]*/
2146  U_WB_ALETTER = 1, /*[LE]*/
2147  U_WB_FORMAT = 2, /*[FO]*/
2148  U_WB_KATAKANA = 3, /*[KA]*/
2149  U_WB_MIDLETTER = 4, /*[ML]*/
2150  U_WB_MIDNUM = 5, /*[MN]*/
2151  U_WB_NUMERIC = 6, /*[NU]*/
2152  U_WB_EXTENDNUMLET = 7, /*[EX]*/
2154  U_WB_CR = 8, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2156  U_WB_EXTEND = 9, /*[Extend]*/
2158  U_WB_LF = 10, /*[LF]*/
2160  U_WB_MIDNUMLET =11, /*[MB]*/
2162  U_WB_NEWLINE =12, /*[NL]*/
2164  U_WB_REGIONAL_INDICATOR = 13, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2166  U_WB_HEBREW_LETTER = 14, /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */
2168  U_WB_SINGLE_QUOTE = 15, /*[SQ]*/
2170  U_WB_DOUBLE_QUOTE = 16, /*[DQ]*/
2172  U_WB_E_BASE = 17, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2174  U_WB_E_BASE_GAZ = 18, /*[EBG]*/
2176  U_WB_E_MODIFIER = 19, /*[EM]*/
2178  U_WB_GLUE_AFTER_ZWJ = 20, /*[GAZ]*/
2180  U_WB_ZWJ = 21, /*[ZWJ]*/
2182  U_WB_WSEGSPACE = 22, /*[WSEGSPACE]*/
2183 
2184 #ifndef U_HIDE_DEPRECATED_API
2185 
2192 #endif // U_HIDE_DEPRECATED_API
2194 
2201 typedef enum USentenceBreak {
2202  /*
2203  * Note: USentenceBreak constants are parsed by preparseucd.py.
2204  * It matches lines like
2205  * U_SB_<Unicode Sentence_Break value name>
2206  */
2207 
2208  U_SB_OTHER = 0, /*[XX]*/
2209  U_SB_ATERM = 1, /*[AT]*/
2210  U_SB_CLOSE = 2, /*[CL]*/
2211  U_SB_FORMAT = 3, /*[FO]*/
2212  U_SB_LOWER = 4, /*[LO]*/
2213  U_SB_NUMERIC = 5, /*[NU]*/
2214  U_SB_OLETTER = 6, /*[LE]*/
2215  U_SB_SEP = 7, /*[SE]*/
2216  U_SB_SP = 8, /*[SP]*/
2217  U_SB_STERM = 9, /*[ST]*/
2218  U_SB_UPPER = 10, /*[UP]*/
2219  U_SB_CR = 11, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2220  U_SB_EXTEND = 12, /*[EX]*/
2221  U_SB_LF = 13, /*[LF]*/
2222  U_SB_SCONTINUE = 14, /*[SC]*/
2223 #ifndef U_HIDE_DEPRECATED_API
2224 
2231 #endif // U_HIDE_DEPRECATED_API
2232 } USentenceBreak;
2233 
2240 typedef enum ULineBreak {
2241  /*
2242  * Note: ULineBreak constants are parsed by preparseucd.py.
2243  * It matches lines like
2244  * U_LB_<Unicode Line_Break value name>
2245  */
2246 
2247  U_LB_UNKNOWN = 0, /*[XX]*/
2248  U_LB_AMBIGUOUS = 1, /*[AI]*/
2249  U_LB_ALPHABETIC = 2, /*[AL]*/
2250  U_LB_BREAK_BOTH = 3, /*[B2]*/
2251  U_LB_BREAK_AFTER = 4, /*[BA]*/
2252  U_LB_BREAK_BEFORE = 5, /*[BB]*/
2253  U_LB_MANDATORY_BREAK = 6, /*[BK]*/
2254  U_LB_CONTINGENT_BREAK = 7, /*[CB]*/
2255  U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
2256  U_LB_COMBINING_MARK = 9, /*[CM]*/
2257  U_LB_CARRIAGE_RETURN = 10, /*[CR]*/
2258  U_LB_EXCLAMATION = 11, /*[EX]*/
2259  U_LB_GLUE = 12, /*[GL]*/
2260  U_LB_HYPHEN = 13, /*[HY]*/
2261  U_LB_IDEOGRAPHIC = 14, /*[ID]*/
2263  U_LB_INSEPARABLE = 15, /*[IN]*/
2264  U_LB_INSEPERABLE = U_LB_INSEPARABLE,
2265  U_LB_INFIX_NUMERIC = 16, /*[IS]*/
2266  U_LB_LINE_FEED = 17, /*[LF]*/
2267  U_LB_NONSTARTER = 18, /*[NS]*/
2268  U_LB_NUMERIC = 19, /*[NU]*/
2269  U_LB_OPEN_PUNCTUATION = 20, /*[OP]*/
2270  U_LB_POSTFIX_NUMERIC = 21, /*[PO]*/
2271  U_LB_PREFIX_NUMERIC = 22, /*[PR]*/
2272  U_LB_QUOTATION = 23, /*[QU]*/
2273  U_LB_COMPLEX_CONTEXT = 24, /*[SA]*/
2274  U_LB_SURROGATE = 25, /*[SG]*/
2275  U_LB_SPACE = 26, /*[SP]*/
2276  U_LB_BREAK_SYMBOLS = 27, /*[SY]*/
2277  U_LB_ZWSPACE = 28, /*[ZW]*/
2279  U_LB_NEXT_LINE = 29, /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
2281  U_LB_WORD_JOINER = 30, /*[WJ]*/
2283  U_LB_H2 = 31, /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
2285  U_LB_H3 = 32, /*[H3]*/
2287  U_LB_JL = 33, /*[JL]*/
2289  U_LB_JT = 34, /*[JT]*/
2291  U_LB_JV = 35, /*[JV]*/
2293  U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
2295  U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
2297  U_LB_HEBREW_LETTER = 38, /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
2299  U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2301  U_LB_E_BASE = 40, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2303  U_LB_E_MODIFIER = 41, /*[EM]*/
2305  U_LB_ZWJ = 42, /*[ZWJ]*/
2306 #ifndef U_HIDE_DEPRECATED_API
2307 
2314 #endif // U_HIDE_DEPRECATED_API
2315 } ULineBreak;
2316 
2323 typedef enum UNumericType {
2324  /*
2325  * Note: UNumericType constants are parsed by preparseucd.py.
2326  * It matches lines like
2327  * U_NT_<Unicode Numeric_Type value name>
2328  */
2329 
2330  U_NT_NONE, /*[None]*/
2331  U_NT_DECIMAL, /*[de]*/
2332  U_NT_DIGIT, /*[di]*/
2333  U_NT_NUMERIC, /*[nu]*/
2334 #ifndef U_HIDE_DEPRECATED_API
2335 
2342 #endif // U_HIDE_DEPRECATED_API
2343 } UNumericType;
2344 
2351 typedef enum UHangulSyllableType {
2352  /*
2353  * Note: UHangulSyllableType constants are parsed by preparseucd.py.
2354  * It matches lines like
2355  * U_HST_<Unicode Hangul_Syllable_Type value name>
2356  */
2357 
2358  U_HST_NOT_APPLICABLE, /*[NA]*/
2359  U_HST_LEADING_JAMO, /*[L]*/
2360  U_HST_VOWEL_JAMO, /*[V]*/
2361  U_HST_TRAILING_JAMO, /*[T]*/
2362  U_HST_LV_SYLLABLE, /*[LV]*/
2363  U_HST_LVT_SYLLABLE, /*[LVT]*/
2364 #ifndef U_HIDE_DEPRECATED_API
2365 
2372 #endif // U_HIDE_DEPRECATED_API
2374 
2382  /*
2383  * Note: UIndicPositionalCategory constants are parsed by preparseucd.py.
2384  * It matches lines like
2385  * U_INPC_<Unicode Indic_Positional_Category value name>
2386  */
2387 
2419 
2427  /*
2428  * Note: UIndicSyllabicCategory constants are parsed by preparseucd.py.
2429  * It matches lines like
2430  * U_INSC_<Unicode Indic_Syllabic_Category value name>
2431  */
2432 
2506 
2513 typedef enum UVerticalOrientation {
2514  /*
2515  * Note: UVerticalOrientation constants are parsed by preparseucd.py.
2516  * It matches lines like
2517  * U_VO_<Unicode Vertical_Orientation value name>
2518  */
2519 
2529 
2557 U_STABLE UBool U_EXPORT2
2559 
2560 #ifndef U_HIDE_DRAFT_API
2561 
2577 U_CAPI const USet * U_EXPORT2
2578 u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode);
2579 
2580 #endif // U_HIDE_DRAFT_API
2581 
2594 U_STABLE UBool U_EXPORT2
2596 
2609 U_STABLE UBool U_EXPORT2
2611 
2624 U_STABLE UBool U_EXPORT2
2626 
2645 U_STABLE UBool U_EXPORT2
2647 
2686 U_STABLE int32_t U_EXPORT2
2688 
2707 U_STABLE int32_t U_EXPORT2
2709 
2736 U_STABLE int32_t U_EXPORT2
2738 
2739 #ifndef U_HIDE_DRAFT_API
2740 
2756 U_CAPI const UCPMap * U_EXPORT2
2757 u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode);
2758 
2759 #endif // U_HIDE_DRAFT_API
2760 
2783 U_STABLE double U_EXPORT2
2785 
2793 #define U_NO_NUMERIC_VALUE ((double)-123456789.)
2794 
2818 U_STABLE UBool U_EXPORT2
2819 u_islower(UChar32 c);
2820 
2845 U_STABLE UBool U_EXPORT2
2846 u_isupper(UChar32 c);
2847 
2862 U_STABLE UBool U_EXPORT2
2863 u_istitle(UChar32 c);
2864 
2883 U_STABLE UBool U_EXPORT2
2884 u_isdigit(UChar32 c);
2885 
2904 U_STABLE UBool U_EXPORT2
2905 u_isalpha(UChar32 c);
2906 
2925 U_STABLE UBool U_EXPORT2
2926 u_isalnum(UChar32 c);
2927 
2948 U_STABLE UBool U_EXPORT2
2949 u_isxdigit(UChar32 c);
2950 
2964 U_STABLE UBool U_EXPORT2
2965 u_ispunct(UChar32 c);
2966 
2983 U_STABLE UBool U_EXPORT2
2984 u_isgraph(UChar32 c);
2985 
3011 U_STABLE UBool U_EXPORT2
3012 u_isblank(UChar32 c);
3013 
3036 U_STABLE UBool U_EXPORT2
3037 u_isdefined(UChar32 c);
3038 
3057 U_STABLE UBool U_EXPORT2
3058 u_isspace(UChar32 c);
3059 
3078 U_STABLE UBool U_EXPORT2
3080 
3118 U_STABLE UBool U_EXPORT2
3120 
3142 U_STABLE UBool U_EXPORT2
3143 u_iscntrl(UChar32 c);
3144 
3157 U_STABLE UBool U_EXPORT2
3159 
3175 U_STABLE UBool U_EXPORT2
3176 u_isprint(UChar32 c);
3177 
3196 U_STABLE UBool U_EXPORT2
3197 u_isbase(UChar32 c);
3198 
3215 U_STABLE UCharDirection U_EXPORT2
3217 
3233 U_STABLE UBool U_EXPORT2
3235 
3255 U_STABLE UChar32 U_EXPORT2
3257 
3274 U_STABLE UChar32 U_EXPORT2
3276 
3288 U_STABLE int8_t U_EXPORT2
3289 u_charType(UChar32 c);
3290 
3304 #define U_GET_GC_MASK(c) U_MASK(u_charType(c))
3305 
3323 typedef UBool U_CALLCONV
3324 UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
3325 
3345 U_STABLE void U_EXPORT2
3346 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
3347 
3348 #if !UCONFIG_NO_NORMALIZATION
3349 
3357 U_STABLE uint8_t U_EXPORT2
3359 
3360 #endif
3361 
3385 U_STABLE int32_t U_EXPORT2
3387 
3397 U_STABLE UBlockCode U_EXPORT2
3399 
3432 U_STABLE int32_t U_EXPORT2
3433 u_charName(UChar32 code, UCharNameChoice nameChoice,
3434  char *buffer, int32_t bufferLength,
3435  UErrorCode *pErrorCode);
3436 
3437 #ifndef U_HIDE_DEPRECATED_API
3438 
3456 U_DEPRECATED int32_t U_EXPORT2
3458  char *dest, int32_t destCapacity,
3459  UErrorCode *pErrorCode);
3460 #endif /* U_HIDE_DEPRECATED_API */
3461 
3482 U_STABLE UChar32 U_EXPORT2
3483 u_charFromName(UCharNameChoice nameChoice,
3484  const char *name,
3485  UErrorCode *pErrorCode);
3486 
3504 typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
3505  UChar32 code,
3506  UCharNameChoice nameChoice,
3507  const char *name,
3508  int32_t length);
3509 
3531 U_STABLE void U_EXPORT2
3532 u_enumCharNames(UChar32 start, UChar32 limit,
3533  UEnumCharNamesFn *fn,
3534  void *context,
3535  UCharNameChoice nameChoice,
3536  UErrorCode *pErrorCode);
3537 
3569 U_STABLE const char* U_EXPORT2
3570 u_getPropertyName(UProperty property,
3571  UPropertyNameChoice nameChoice);
3572 
3592 U_STABLE UProperty U_EXPORT2
3593 u_getPropertyEnum(const char* alias);
3594 
3642 U_STABLE const char* U_EXPORT2
3644  int32_t value,
3645  UPropertyNameChoice nameChoice);
3646 
3678 U_STABLE int32_t U_EXPORT2
3680  const char* alias);
3681 
3699 U_STABLE UBool U_EXPORT2
3700 u_isIDStart(UChar32 c);
3701 
3723 U_STABLE UBool U_EXPORT2
3724 u_isIDPart(UChar32 c);
3725 
3746 U_STABLE UBool U_EXPORT2
3748 
3765 U_STABLE UBool U_EXPORT2
3767 
3786 U_STABLE UBool U_EXPORT2
3788 
3811 U_STABLE UChar32 U_EXPORT2
3812 u_tolower(UChar32 c);
3813 
3836 U_STABLE UChar32 U_EXPORT2
3837 u_toupper(UChar32 c);
3838 
3861 U_STABLE UChar32 U_EXPORT2
3862 u_totitle(UChar32 c);
3863 
3886 U_STABLE UChar32 U_EXPORT2
3887 u_foldCase(UChar32 c, uint32_t options);
3888 
3927 U_STABLE int32_t U_EXPORT2
3928 u_digit(UChar32 ch, int8_t radix);
3929 
3958 U_STABLE UChar32 U_EXPORT2
3959 u_forDigit(int32_t digit, int8_t radix);
3960 
3975 U_STABLE void U_EXPORT2
3976 u_charAge(UChar32 c, UVersionInfo versionArray);
3977 
3989 U_STABLE void U_EXPORT2
3990 u_getUnicodeVersion(UVersionInfo versionArray);
3991 
3992 #if !UCONFIG_NO_NORMALIZATION
3993 
4014 U_STABLE int32_t U_EXPORT2
4015 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
4016 
4017 #endif
4018 
4019 
4021 
4022 #endif /*_UCHAR*/
4023 /*eof*/
UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS
Definition: uchar.h:1289
U_POP_DIRECTIONAL_ISOLATE
PDI.
Definition: uchar.h:974
U_JG_TEH_MARBUTA_GOAL
Definition: uchar.h:1977
U_INSC_BRAHMI_JOINING_NUMBER
Definition: uchar.h:2440
U_GCB_GLUE_AFTER_ZWJ
Definition: uchar.h:2116
u_getPropertyValueEnum
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...
UBLOCK_HANIFI_ROHINGYA
Definition: uchar.h:1756
U_JG_NYA
Definition: uchar.h:2020
UBLOCK_MEDEFAIDRIN
Definition: uchar.h:1764
UCHAR_INDIC_SYLLABIC_CATEGORY
Enumerated property Indic_Syllabic_Category.
Definition: uchar.h:593
UCHAR_JOINING_TYPE
Enumerated property Joining_Type.
Definition: uchar.h:520
UCHAR_POSIX_XDIGIT
Binary property xdigit (a C/POSIX character class).
Definition: uchar.h:417
UCHAR_ALPHABETIC
Binary property Alphabetic.
Definition: uchar.h:209
u_isalnum
UBool u_isalnum(UChar32 c)
Determines whether the specified code point is an alphanumeric character (letter or digit) according ...
U_ENCLOSING_MARK
Me.
Definition: uchar.h:750
UBLOCK_EARLY_DYNASTIC_CUNEIFORM
Definition: uchar.h:1691
UCHAR_ASCII_HEX_DIGIT
Binary property ASCII_Hex_Digit.
Definition: uchar.h:213
UBLOCK_SUPPLEMENTAL_SYMBOLS_AND_PICTOGRAPHS
Definition: uchar.h:1699
UCHAR_RADICAL
Binary property Radical (new in Unicode 3.2).
Definition: uchar.h:301
U_NT_COUNT
One more than the highest normal UNumericType value.
Definition: uchar.h:2341
UBLOCK_DINGBATS
Definition: uchar.h:1202
UBLOCK_GREEK
Unicode 3.2 renames this block to "Greek and Coptic".
Definition: uchar.h:1055
U_LB_ZWJ
Definition: uchar.h:2305
U_INSC_NON_JOINER
Definition: uchar.h:2478
U_INPC_LEFT
Definition: uchar.h:2397
UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_F
Definition: uchar.h:1731
UCHAR_BIDI_PAIRED_BRACKET
String property Bidi_Paired_Bracket (new in Unicode 6.3).
Definition: uchar.h:690
UCHAR_NFD_QUICK_CHECK
Enumerated property NFD_Quick_Check.
Definition: uchar.h:535
UCHAR_SIMPLE_UPPERCASE_MAPPING
String property Simple_Uppercase_Mapping.
Definition: uchar.h:674
UBLOCK_UGARITIC
Definition: uchar.h:1393
U_JG_MANICHAEAN_QOPH
Definition: uchar.h:2037
U_JG_AFRICAN_QAF
Definition: uchar.h:2052
U_ARABIC_NUMBER
AN.
Definition: uchar.h:940
UBLOCK_LIMBU
Definition: uchar.h:1375
UBLOCK_SORA_SOMPENG
Definition: uchar.h:1607
u_charAge
void u_charAge(UChar32 c, UVersionInfo versionArray)
Get the "age" of the code point.
UCHAR_INDIC_POSITIONAL_CATEGORY
Enumerated property Indic_Positional_Category.
Definition: uchar.h:586
UBLOCK_CJK_STROKES
Definition: uchar.h:1416
UBLOCK_MISCELLANEOUS_TECHNICAL
Definition: uchar.h:1178
U_LB_WORD_JOINER
Definition: uchar.h:2281
UBLOCK_PHAGS_PA
Definition: uchar.h:1459
UCHAR_CASE_FOLDING
String property Case_Folding.
Definition: uchar.h:651
UBLOCK_CUNEIFORM_NUMBERS_AND_PUNCTUATION
Definition: uchar.h:1465
UBLOCK_MEETEI_MAYEK_EXTENSIONS
Definition: uchar.h:1597
UBLOCK_ADLAM
Definition: uchar.h:1706
UBLOCK_SUTTON_SIGNWRITING
Definition: uchar.h:1701
UBLOCK_MISCELLANEOUS_SYMBOLS
Definition: uchar.h:1199
u_charDirection
UCharDirection u_charDirection(UChar32 c)
Returns the bidirectional category value for the code point, which is used in the Unicode bidirection...
U_INSC_CONSONANT_PREFIXED
Definition: uchar.h:2462
UBLOCK_DESERET
Definition: uchar.h:1322
U_TITLECASE_LETTER
Lt.
Definition: uchar.h:742
UCHAR_MASK_LIMIT
One more than the last constant for bit-mask Unicode properties.
Definition: uchar.h:625
U_INPC_BOTTOM_AND_LEFT
Definition: uchar.h:2393
UCHAR_SIMPLE_TITLECASE_MAPPING
String property Simple_Titlecase_Mapping.
Definition: uchar.h:671
UBLOCK_OLD_SOUTH_ARABIAN
Definition: uchar.h:1539
UCHAR_DOUBLE_LIMIT
One more than the last constant for double Unicode properties.
Definition: uchar.h:638
UBLOCK_MODIFIER_TONE_LETTERS
Definition: uchar.h:1432
UCHAR_EMOJI_MODIFIER
Binary property Emoji_Modifier.
Definition: uchar.h:454
u_getIntPropertyValue
int32_t u_getIntPropertyValue(UChar32 c, UProperty which)
Get the property value for an enumerated or integer Unicode property for a code point.
UBLOCK_MANDAIC
Definition: uchar.h:1564
UBLOCK_GREEK_EXTENDED
Definition: uchar.h:1148
UBLOCK_MARCHEN
Definition: uchar.h:1716
U_LOWERCASE_LETTER
Ll.
Definition: uchar.h:740
UCHAR_HYPHEN
Binary property Hyphen.
Definition: uchar.h:259
UCHAR_NUMERIC_VALUE
Double property Numeric_Value.
Definition: uchar.h:630
UBLOCK_NUMBER_FORMS
Definition: uchar.h:1169
UBLOCK_SHORTHAND_FORMAT_CONTROLS
Definition: uchar.h:1668
UBLOCK_MEROITIC_CURSIVE
Definition: uchar.h:1599
U_VO_TRANSFORMED_ROTATED
Definition: uchar.h:2523
utypes.h
Basic definitions for ICU, for both C and C++ APIs.
UBLOCK_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT
Definition: uchar.h:1557
u_charName
int32_t u_charName(UChar32 code, UCharNameChoice nameChoice, char *buffer, int32_t bufferLength, UErrorCode *pErrorCode)
Retrieve the name of a Unicode character.
UCHAR_NFKD_INERT
Binary property NFKD_Inert.
Definition: uchar.h:357
UBLOCK_OSAGE
Definition: uchar.h:1722
U_INSC_MODIFYING_LETTER
Definition: uchar.h:2476
UBLOCK_LETTERLIKE_SYMBOLS
Definition: uchar.h:1166
UCHAR_LOGICAL_ORDER_EXCEPTION
Binary property Logical_Order_Exception (new in Unicode 3.2).
Definition: uchar.h:286
U_LINE_SEPARATOR
Zl.
Definition: uchar.h:762
UCHAR_TITLECASE_MAPPING
String property Titlecase_Mapping.
Definition: uchar.h:677
u_isspace
UBool u_isspace(UChar32 c)
Determines if the specified character is a space character or not.
U_VO_TRANSFORMED_UPRIGHT
Definition: uchar.h:2525
UBLOCK_COUNT
One more than the highest normal UBlockCode value.
Definition: uchar.h:1777
U_JG_STRAIGHT_WAW
Definition: uchar.h:2049
UCHAR_SENTENCE_BREAK
Enumerated property Sentence_Break (new in Unicode 4.1).
Definition: uchar.h:568
U_COMBINING_SPACING_MARK
Mc.
Definition: uchar.h:752
U_EUROPEAN_NUMBER_TERMINATOR
ET.
Definition: uchar.h:938
UBLOCK_VARIATION_SELECTORS
Definition: uchar.h:1366
U_WB_COUNT
One more than the highest normal UWordBreakValues value.
Definition: uchar.h:2191
U_INSC_VOWEL_INDEPENDENT
Definition: uchar.h:2504
U_LB_REGIONAL_INDICATOR
Definition: uchar.h:2299
U_CHAR_CATEGORY_COUNT
One higher than the last enum UCharCategory constant.
Definition: uchar.h:802
U_PROPERTY_NAME_CHOICE_COUNT
One more than the highest normal UPropertyNameChoice value.
Definition: uchar.h:1874
UBool
int8_t UBool
The ICU boolean type.
Definition: umachine.h:225
UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED
Definition: uchar.h:1511
UCHAR_NFKC_INERT
Binary property NFKC_Inert.
Definition: uchar.h:371
UCHAR_VERTICAL_ORIENTATION
Enumerated property Vertical_Orientation.
Definition: uchar.h:600
UBLOCK_TANGUT
Definition: uchar.h:1724
UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT
Definition: uchar.h:1332
UBLOCK_CYRILLIC_EXTENDED_B
Definition: uchar.h:1482
UBLOCK_MAHAJANI
Definition: uchar.h:1640
U_FINAL_PUNCTUATION
Pf.
Definition: uchar.h:794
U_INSC_JOINER
Definition: uchar.h:2474
u_getISOComment
int32_t u_getISOComment(UChar32 c, char *dest, int32_t destCapacity, UErrorCode *pErrorCode)
Returns an empty string.
U_JG_MALAYALAM_NNNA
Definition: uchar.h:2060
UBLOCK_LAO
Definition: uchar.h:1109
UBLOCK_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS
Definition: uchar.h:1593
u_enumCharTypes
void u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context)
Enumerate efficiently all code points with their Unicode general categories.
u_getIntPropertyMaxValue
int32_t u_getIntPropertyMaxValue(UProperty which)
Get the maximum value for an enumerated/integer/binary Unicode property.
U_WB_LF
Definition: uchar.h:2158
U_INSC_CONSONANT_KILLER
Definition: uchar.h:2454
UCHAR_ISO_COMMENT
Deprecated string property ISO_Comment.
Definition: uchar.h:655
UCHAR_CHANGES_WHEN_CASEMAPPED
Binary property Changes_When_Casemapped.
Definition: uchar.h:431
UBLOCK_CARIAN
Definition: uchar.h:1498
UBLOCK_HIGH_SURROGATES
Definition: uchar.h:1259
U_RIGHT_TO_LEFT_ARABIC
AL.
Definition: uchar.h:956
UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS
Definition: uchar.h:1130
U_LB_H2
Definition: uchar.h:2283
UBLOCK_RUNIC
Definition: uchar.h:1136
UCHAR_POSIX_BLANK
Binary property blank (a C/POSIX character class).
Definition: uchar.h:402
U_LB_NEXT_LINE
Definition: uchar.h:2279
u_iscntrl
UBool u_iscntrl(UChar32 c)
Determines whether the specified code point is a control character (as defined by this function).
UBLOCK_MALAYALAM
Definition: uchar.h:1100
UCHAR_POSIX_GRAPH
Binary property graph (a C/POSIX character class).
Definition: uchar.h:407
UCHAR_NONCHARACTER_CODE_POINT
Binary property Noncharacter_Code_Point.
Definition: uchar.h:295
UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A
Definition: uchar.h:1244
u_isUUppercase
UBool u_isUUppercase(UChar32 c)
Check if a code point has the Uppercase Unicode property.
UBLOCK_OLD_PERMIC
Definition: uchar.h:1656
U_JG_MANICHAEAN_WAW
Definition: uchar.h:2046
UBLOCK_SUPPLEMENTAL_ARROWS_A
Definition: uchar.h:1356
u_getBidiPairedBracket
UChar32 u_getBidiPairedBracket(UChar32 c)
Maps the specified character to its paired bracket character.
UBLOCK_CJK_COMPATIBILITY
Definition: uchar.h:1241
U_CALLCONV
#define U_CALLCONV
Definition: platform.h:840
UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D
Definition: uchar.h:1586
UBLOCK_PLAYING_CARDS
Definition: uchar.h:1576
U_LB_CLOSE_PARENTHESIS
Definition: uchar.h:2293
U_JG_MANICHAEAN_ZAYIN
Definition: uchar.h:2048
U_OTHER_PUNCTUATION
Po.
Definition: uchar.h:782
UBLOCK_HANGUL_SYLLABLES
Definition: uchar.h:1256
UBLOCK_NEW_TAI_LUE
Definition: uchar.h:1434
UBLOCK_BOPOMOFO_EXTENDED
Definition: uchar.h:1235
UCHAR_AGE
String property Age.
Definition: uchar.h:643
UBLOCK_ETHIOPIC_SUPPLEMENT
Definition: uchar.h:1424
UBLOCK_MANICHAEAN
Definition: uchar.h:1642
UBLOCK_GURMUKHI
Definition: uchar.h:1082
UCHAR_BIDI_PAIRED_BRACKET_TYPE
Enumerated property Bidi_Paired_Bracket_Type (new in Unicode 6.3).
Definition: uchar.h:578
UBLOCK_REJANG
Definition: uchar.h:1488
UCHAR_INVALID_CODE
Represents a nonexistent or invalid property or property value.
Definition: uchar.h:716
U_JG_MANICHAEAN_RESH
Definition: uchar.h:2038
UBLOCK_BUHID
Definition: uchar.h:1350
U_LB_JL
Definition: uchar.h:2287
U_CHAR_DIRECTION_COUNT
One more than the highest UCharDirection value.
Definition: uchar.h:982
UBLOCK_COMBINING_MARKS_FOR_SYMBOLS
Unicode 3.2 renames this block to "Combining Diacritical Marks for Symbols".
Definition: uchar.h:1163
U_INSC_VISARGA
Definition: uchar.h:2498
U_UNICODE_10_CHAR_NAME
The Unicode_1_Name property value which is of little practical value.
Definition: uchar.h:1838
UBLOCK_LATIN_EXTENDED_E
Definition: uchar.h:1636
UBLOCK_DOGRA
Definition: uchar.h:1750
u_isUAlphabetic
UBool u_isUAlphabetic(UChar32 c)
Check if a code point has the Alphabetic Unicode property.
U_OTHER_NUMBER
No.
Definition: uchar.h:758
UBLOCK_CYPRIOT_SYLLABARY
Definition: uchar.h:1399
UBLOCK_CHEROKEE_SUPPLEMENT
Definition: uchar.h:1687
UBlockCode
UBlockCode
Constants for Unicode blocks, see the Unicode Data file Blocks.txt.
Definition: uchar.h:1020
UCHAR_WHITE_SPACE
Binary property White_Space.
Definition: uchar.h:321
UBLOCK_EMOTICONS
Definition: uchar.h:1580
UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C
Definition: uchar.h:1559
U_CURRENCY_SYMBOL
Sc.
Definition: uchar.h:786
U_LETTER_NUMBER
Nl.
Definition: uchar.h:756
UProperty
UProperty
Selection constants for Unicode properties.
Definition: uchar.h:195
UBLOCK_MATHEMATICAL_ALPHANUMERIC_SYMBOLS
Definition: uchar.h:1328
UBLOCK_KHUDAWADI
Definition: uchar.h:1634
u_enumCharNames
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,...
UBLOCK_KANA_EXTENDED_A
Definition: uchar.h:1733
U_WB_REGIONAL_INDICATOR
Definition: uchar.h:2164
U_EUROPEAN_NUMBER
EN.
Definition: uchar.h:934
U_DEPRECATED
#define U_DEPRECATED
This is used to declare a function as a deprecated public ICU C API
Definition: umachine.h:115
UBLOCK_LYCIAN
Definition: uchar.h:1496
UBLOCK_SOGDIAN
Definition: uchar.h:1768
UBLOCK_INSCRIPTIONAL_PARTHIAN
Definition: uchar.h:1543
U_COMMON_NUMBER_SEPARATOR
CS.
Definition: uchar.h:942
UBLOCK_IDEOGRAPHIC_DESCRIPTION_CHARACTERS
Definition: uchar.h:1214
u_isJavaIDStart
UBool u_isJavaIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in a Java identifier.
u_isMirrored
UBool u_isMirrored(UChar32 c)
Determines whether the code point has the Bidi_Mirrored property.
U_WB_GLUE_AFTER_ZWJ
Definition: uchar.h:2178
UBLOCK_VEDIC_EXTENSIONS
Definition: uchar.h:1515
UBLOCK_VAI
Definition: uchar.h:1480
U_INPC_TOP_AND_LEFT
Definition: uchar.h:2411
UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_B
Definition: uchar.h:1370
UBLOCK_EGYPTIAN_HIEROGLYPHS
Definition: uchar.h:1553
UBLOCK_PRIVATE_USE
Same as UBLOCK_PRIVATE_USE_AREA.
Definition: uchar.h:1286
U_INPC_TOP_AND_BOTTOM
Definition: uchar.h:2407
UBLOCK_ANCIENT_GREEK_NUMBERS
Definition: uchar.h:1410
U_CHAR_NAME_ALIAS
Corrected name from NameAliases.txt.
Definition: uchar.h:1843
U_DT_COUNT
One more than the highest normal UDecompositionType value.
Definition: uchar.h:1916
UBLOCK_SOYOMBO
Definition: uchar.h:1739
UBLOCK_YIJING_HEXAGRAM_SYMBOLS
Definition: uchar.h:1385
UBLOCK_ZANABAZAR_SQUARE
Definition: uchar.h:1743
U_SURROGATE
Cs.
Definition: uchar.h:772
UCHAR_CASE_IGNORABLE
Binary property Case_Ignorable.
Definition: uchar.h:421
UBLOCK_ARABIC_PRESENTATION_FORMS_B
Definition: uchar.h:1307
UCHAR_CANONICAL_COMBINING_CLASS
Enumerated property Canonical_Combining_Class.
Definition: uchar.h:504
UCHAR_REGIONAL_INDICATOR
Binary property Regional_Indicator.
Definition: uchar.h:473
UCHAR_LINE_BREAK
Enumerated property Line_Break.
Definition: uchar.h:523
UBLOCK_BALINESE
Definition: uchar.h:1453
UBLOCK_AEGEAN_NUMBERS
Definition: uchar.h:1391
u_digit
int32_t u_digit(UChar32 ch, int8_t radix)
Returns the decimal digit value of the code point in the specified radix.
U_INPC_BOTTOM_AND_RIGHT
Definition: uchar.h:2395
UBLOCK_ANCIENT_SYMBOLS
Definition: uchar.h:1492
U_PARAGRAPH_SEPARATOR
Zp.
Definition: uchar.h:764
UBLOCK_COMBINING_DIACRITICAL_MARKS_EXTENDED
Definition: uchar.h:1622
UBLOCK_BOPOMOFO
Definition: uchar.h:1226
U_WB_ZWJ
Definition: uchar.h:2180
UBLOCK_NEWA
Definition: uchar.h:1720
UBLOCK_GEOMETRIC_SHAPES_EXTENDED
Definition: uchar.h:1628
UEnumCharNamesFn
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:3504
UBLOCK_DEVANAGARI
Definition: uchar.h:1076
U_INSC_CANTILLATION_MARK
Definition: uchar.h:2442
UCHAR_GENERAL_CATEGORY_MASK
Bitmask property General_Category_Mask.
Definition: uchar.h:617
U_JG_AFRICAN_NOON
Definition: uchar.h:2051
UCHAR_UPPERCASE_MAPPING
String property Uppercase_Mapping.
Definition: uchar.h:687
UBLOCK_TIBETAN
Definition: uchar.h:1112
UHangulSyllableType
UHangulSyllableType
Hangul Syllable Type constants.
Definition: uchar.h:2351
u_tolower
UChar32 u_tolower(UChar32 c)
The given character is mapped to its lowercase equivalent according to UnicodeData....
U_INSC_CONSONANT_HEAD_LETTER
Definition: uchar.h:2450
U_WB_E_BASE
Definition: uchar.h:2172
UBLOCK_GEORGIAN_EXTENDED
Definition: uchar.h:1752
UBLOCK_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT
Definition: uchar.h:1418
u_isbase
UBool u_isbase(UChar32 c)
Determines whether the specified code point is a base character.
UIndicSyllabicCategory
UIndicSyllabicCategory
Indic Syllabic Category constants.
Definition: uchar.h:2426
UBLOCK_MAHJONG_TILES
Definition: uchar.h:1502
UBLOCK_OGHAM
Definition: uchar.h:1133
U_JG_MANICHAEAN_TAW
Definition: uchar.h:2041
UBLOCK_ARABIC_SUPPLEMENT
Definition: uchar.h:1412
UCHAR_DASH
Binary property Dash.
Definition: uchar.h:224
UBLOCK_SPECIALS
Definition: uchar.h:1310
U_NON_SPACING_MARK
Mn.
Definition: uchar.h:748
UBLOCK_TAI_LE
Definition: uchar.h:1377
u_isdefined
UBool u_isdefined(UChar32 c)
Determines whether the specified code point is "defined", which usually means that it is assigned a c...
U_WHITE_SPACE_NEUTRAL
WS.
Definition: uchar.h:948
u_charType
int8_t u_charType(UChar32 c)
Returns the general category value for the code point.
U_FIRST_STRONG_ISOLATE
FSI.
Definition: uchar.h:968
U_WB_NEWLINE
Definition: uchar.h:2162
U_MODIFIER_LETTER
Lm.
Definition: uchar.h:744
UCHAR_BIDI_MIRRORED
Binary property Bidi_Mirrored.
Definition: uchar.h:222
UCHAR_BIDI_MIRRORING_GLYPH
String property Bidi_Mirroring_Glyph.
Definition: uchar.h:648
U_JG_MANICHAEAN_DHAMEDH
Definition: uchar.h:2026
UBLOCK_MONGOLIAN_SUPPLEMENT
Definition: uchar.h:1718
UBLOCK_OLD_TURKIC
Definition: uchar.h:1547
U_JG_COUNT
One more than the highest normal UJoiningGroup value.
Definition: uchar.h:2076
U_INPC_RIGHT
Definition: uchar.h:2403
UBLOCK_TAKRI
Definition: uchar.h:1611
UBLOCK_DUPLOYAN
Definition: uchar.h:1624
UCHAR_INT_START
First constant for enumerated/integer Unicode properties.
Definition: uchar.h:498
UBLOCK_NKO
Definition: uchar.h:1451
UBLOCK_BHAIKSUKI
Definition: uchar.h:1708
UBLOCK_IPA_EXTENSIONS
Definition: uchar.h:1043
U_INSC_SYLLABLE_MODIFIER
Definition: uchar.h:2490
U_LEFT_TO_RIGHT_ISOLATE
LRI.
Definition: uchar.h:970
UBLOCK_TRANSPORT_AND_MAP_SYMBOLS
Definition: uchar.h:1582
UCHAR_EXTENDER
Binary property Extender.
Definition: uchar.h:238
UCHAR_IDS_TRINARY_OPERATOR
Binary property IDS_Trinary_Operator (new in Unicode 3.2).
Definition: uchar.h:279
u_isalpha
UBool u_isalpha(UChar32 c)
Determines whether the specified code point is a letter character.
UCHAR_CHANGES_WHEN_UPPERCASED
Binary property Changes_When_Uppercased.
Definition: uchar.h:425
U_LB_E_MODIFIER
Definition: uchar.h:2303
UBLOCK_KANGXI_RADICALS
Definition: uchar.h:1211
UBLOCK_ARABIC
Definition: uchar.h:1067
U_JG_MALAYALAM_NGA
Definition: uchar.h:2058
u_getBinaryPropertySet
const U_CAPI USet * u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode)
Returns a frozen USet for a binary property.
UBLOCK_SAURASHTRA
Definition: uchar.h:1484
UBLOCK_NUSHU
Definition: uchar.h:1737
UCHAR_DECOMPOSITION_TYPE
Enumerated property Decomposition_Type.
Definition: uchar.h:507
UCHAR_IDS_BINARY_OPERATOR
Binary property IDS_Binary_Operator (new in Unicode 3.2).
Definition: uchar.h:275
U_RIGHT_TO_LEFT_ISOLATE
RLI.
Definition: uchar.h:972
UBLOCK_BENGALI
Definition: uchar.h:1079
UBLOCK_PHAISTOS_DISC
Definition: uchar.h:1494
U_PRIVATE_USE_CHAR
Co.
Definition: uchar.h:770
UBLOCK_NO_BLOCK
New No_Block value in Unicode 4.
Definition: uchar.h:1028
UBLOCK_HIRAGANA
Definition: uchar.h:1220
UBLOCK_IDEOGRAPHIC_SYMBOLS_AND_PUNCTUATION
Definition: uchar.h:1714
UJoiningType
UJoiningType
Joining Type constants.
Definition: uchar.h:1926
UCHAR_EXTENDED_PICTOGRAPHIC
Binary property Extended_Pictographic.
Definition: uchar.h:485
UBLOCK_CYRILLIC_EXTENDED_A
Definition: uchar.h:1478
U_JG_MANICHAEAN_PE
Definition: uchar.h:2036
u_getCombiningClass
uint8_t u_getCombiningClass(UChar32 c)
Returns the combining class of the code point as specified in UnicodeData.txt.
UCHAR_MATH
Binary property Math.
Definition: uchar.h:291
UCHAR_MASK_START
First constant for bit-mask Unicode properties.
Definition: uchar.h:619
U_JG_MANICHAEAN_LAMEDH
Definition: uchar.h:2032
UCHAR_DIACRITIC
Binary property Diacritic.
Definition: uchar.h:234
u_isblank
UBool u_isblank(UChar32 c)
Determines whether the specified code point is a "blank" or "horizontal space", a character that visi...
UBLOCK_ARABIC_EXTENDED_A
Definition: uchar.h:1591
UBLOCK_ORIYA
Definition: uchar.h:1088
UBLOCK_BUGINESE
Definition: uchar.h:1414
UCHAR_IDEOGRAPHIC
Binary property Ideographic.
Definition: uchar.h:271
UPropertyNameChoice
UPropertyNameChoice
Selector constants for u_getPropertyName() and u_getPropertyValueName().
Definition: uchar.h:1866
U_INSC_VIRAMA
Definition: uchar.h:2496
UChar32
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:389
UBLOCK_COMMON_INDIC_NUMBER_FORMS
Definition: uchar.h:1521
U_OTHER_NEUTRAL
ON.
Definition: uchar.h:950
U_INPC_TOP_AND_BOTTOM_AND_RIGHT
Definition: uchar.h:2409
UBLOCK_GENERAL_PUNCTUATION
Definition: uchar.h:1151
UBLOCK_ENCLOSED_ALPHANUMERIC_SUPPLEMENT
Definition: uchar.h:1555
UCHAR_GRAPHEME_BASE
Binary property Grapheme_Base (new in Unicode 3.2).
Definition: uchar.h:246
U_JG_MANICHAEAN_GIMEL
Definition: uchar.h:2028
UBLOCK_MATHEMATICAL_OPERATORS
Definition: uchar.h:1175
U_INSC_TONE_MARK
Definition: uchar.h:2494
UCHAR_ID_START
Binary property ID_Start.
Definition: uchar.h:268
U_OTHER_LETTER
Lo.
Definition: uchar.h:746
UCHAR_QUOTATION_MARK
Binary property Quotation_Mark.
Definition: uchar.h:297
UBLOCK_ALPHABETIC_PRESENTATION_FORMS
Definition: uchar.h:1292
UCHAR_DEFAULT_IGNORABLE_CODE_POINT
Binary property Default_Ignorable_Code_Point (new in Unicode 3.2).
Definition: uchar.h:228
U_DECIMAL_DIGIT_NUMBER
Nd.
Definition: uchar.h:754
u_isdigit
UBool u_isdigit(UChar32 c)
Determines whether the specified code point is a digit character according to Java.
UErrorCode
UErrorCode
Error code to replace exception handling, so that the code is compatible with all C++ compilers,...
Definition: utypes.h:401
UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_A
Definition: uchar.h:1368
UBLOCK_PALMYRENE
Definition: uchar.h:1662
UBLOCK_TANGUT_COMPONENTS
Definition: uchar.h:1726
U_SPACE_SEPARATOR
Zs.
Definition: uchar.h:760
UBLOCK_LYDIAN
Definition: uchar.h:1500
UBLOCK_CJK_COMPATIBILITY_FORMS
Definition: uchar.h:1301
UCharDirection
UCharDirection
This specifies the language directional property of a character set.
Definition: uchar.h:921
UBLOCK_TIFINAGH
Definition: uchar.h:1444
U_SB_COUNT
One more than the highest normal USentenceBreak value.
Definition: uchar.h:2230
U_GCB_ZWJ
Definition: uchar.h:2118
UCHAR_BINARY_START
First constant for binary Unicode properties.
Definition: uchar.h:211
U_INPC_OVERSTRUCK
Definition: uchar.h:2401
UCHAR_CHANGES_WHEN_LOWERCASED
Binary property Changes_When_Lowercased.
Definition: uchar.h:423
UBLOCK_MASARAM_GONDI
Definition: uchar.h:1735
UCHAR_EMOJI_MODIFIER_BASE
Binary property Emoji_Modifier_Base.
Definition: uchar.h:461
U_JG_MANICHAEAN_SAMEKH
Definition: uchar.h:2040
UBLOCK_GEORGIAN
Definition: uchar.h:1118
UBLOCK_KATAKANA_PHONETIC_EXTENSIONS
Definition: uchar.h:1364
u_isIDIgnorable
UBool u_isIDIgnorable(UChar32 c)
Determines if the specified character should be regarded as an ignorable character in an identifier,...
UBLOCK_KANNADA
Definition: uchar.h:1097
u_charMirror
UChar32 u_charMirror(UChar32 c)
Maps the specified character to a "mirror-image" character.
UBLOCK_SUNDANESE
Definition: uchar.h:1472
UBLOCK_HANGUL_JAMO_EXTENDED_A
Definition: uchar.h:1525
UBLOCK_COMBINING_DIACRITICAL_MARKS
Definition: uchar.h:1049
U_EUROPEAN_NUMBER_SEPARATOR
ES.
Definition: uchar.h:936
UCHAR_HANGUL_SYLLABLE_TYPE
Enumerated property Hangul_Syllable_Type, new in Unicode 4.
Definition: uchar.h:532
U_GCB_PREPEND
Definition: uchar.h:2106
UBLOCK_SMALL_FORM_VARIANTS
Definition: uchar.h:1304
UCHAR_NFD_INERT
Binary property NFD_Inert.
Definition: uchar.h:350
UBLOCK_RUMI_NUMERAL_SYMBOLS
Definition: uchar.h:1549
UBLOCK_BRAHMI
Definition: uchar.h:1570
ublock_getCode
UBlockCode ublock_getCode(UChar32 c)
Returns the Unicode allocation block that contains the character.
UCHAR_NUMERIC_TYPE
Enumerated property Numeric_Type.
Definition: uchar.h:526
U_JG_MANICHAEAN_TEN
Definition: uchar.h:2042
UBLOCK_OL_CHIKI
Definition: uchar.h:1476
UBLOCK_MEROITIC_HIEROGLYPHS
Definition: uchar.h:1601
UBLOCK_THAANA
Definition: uchar.h:1073
UBLOCK_KAYAH_LI
Definition: uchar.h:1486
UBLOCK_ETHIOPIC_EXTENDED_A
Definition: uchar.h:1568
U_JG_MALAYALAM_LLA
Definition: uchar.h:2056
U_LB_INSEPARABLE
Renamed from the misspelled "inseperable" in Unicode 4.0.1/ICU 3.0.
Definition: uchar.h:2263
U_DIR_NON_SPACING_MARK
NSM.
Definition: uchar.h:964
U_INPC_VISUAL_ORDER_LEFT
Definition: uchar.h:2417
U_LB_H3
Definition: uchar.h:2285
UBLOCK_ORNAMENTAL_DINGBATS
Definition: uchar.h:1658
UCHAR_GRAPHEME_CLUSTER_BREAK
Enumerated property Grapheme_Cluster_Break (new in Unicode 4.1).
Definition: uchar.h:563
U_JG_MANICHAEAN_MEM
Definition: uchar.h:2033
UGraphemeClusterBreak
UGraphemeClusterBreak
Grapheme Cluster Break constants.
Definition: uchar.h:2086
u_charFromName
UChar32 u_charFromName(UCharNameChoice nameChoice, const char *name, UErrorCode *pErrorCode)
Find a Unicode character by its name and return its code point value.
UCHAR_LOWERCASE_MAPPING
String property Lowercase_Mapping.
Definition: uchar.h:659
U_LEFT_TO_RIGHT_OVERRIDE
LRO.
Definition: uchar.h:954
U_MODIFIER_SYMBOL
Sk.
Definition: uchar.h:788
UBLOCK_NABATAEAN
Definition: uchar.h:1652
U_BPT_CLOSE
Close paired bracket.
Definition: uchar.h:1004
UCHAR_POSIX_PRINT
Binary property print (a C/POSIX character class).
Definition: uchar.h:412
UBLOCK_LATIN_EXTENDED_C
Definition: uchar.h:1455
UBLOCK_ETHIOPIC_EXTENDED
Definition: uchar.h:1422
U_WB_WSEGSPACE
Definition: uchar.h:2182
U_JG_MANICHAEAN_YODH
Definition: uchar.h:2047
UBLOCK_INDIC_SIYAQ_NUMBERS
Definition: uchar.h:1758
UBLOCK_SYRIAC
Definition: uchar.h:1070
U_WB_EXTEND
Definition: uchar.h:2156
UBLOCK_SUPPLEMENTAL_ARROWS_C
Definition: uchar.h:1674
U_RIGHT_TO_LEFT
R.
Definition: uchar.h:932
USentenceBreak
USentenceBreak
Sentence Break constants.
Definition: uchar.h:2201
u_getPropertyValueName
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...
U_INSC_CONSONANT_PRECEDING_REPHA
Definition: uchar.h:2460
UCHAR_JOINING_GROUP
Enumerated property Joining_Group.
Definition: uchar.h:517
UBLOCK_KANA_SUPPLEMENT
Definition: uchar.h:1574
U_INSC_CONSONANT_SUCCEEDING_REPHA
Definition: uchar.h:2466
UBLOCK_BASIC_LATIN
Definition: uchar.h:1031
U_CONNECTOR_PUNCTUATION
Pc.
Definition: uchar.h:780
UCharCategory
UCharCategory
Data for enumerated Unicode general category types.
Definition: uchar.h:724
UCHAR_NFKC_QUICK_CHECK
Enumerated property NFKC_Quick_Check.
Definition: uchar.h:544
u_isJavaSpaceChar
UBool u_isJavaSpaceChar(UChar32 c)
Determine if the specified code point is a space character according to Java.
U_LEFT_TO_RIGHT_EMBEDDING
LRE.
Definition: uchar.h:952
U_VO_ROTATED
Definition: uchar.h:2521
UCHAR_WORD_BREAK
Enumerated property Word_Break (new in Unicode 4.1).
Definition: uchar.h:573
UCHAR_DOUBLE_START
First constant for double Unicode properties.
Definition: uchar.h:632
UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED
Binary property Changes_When_NFKC_Casefolded.
Definition: uchar.h:433
U_JG_MANICHAEAN_FIVE
Definition: uchar.h:2027
UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_E
Definition: uchar.h:1689
u_foldCase
UChar32 u_foldCase(UChar32 c, uint32_t options)
The given character is mapped to its case folding equivalent according to UnicodeData....
UBLOCK_CAUCASIAN_ALBANIAN
Definition: uchar.h:1618
U_POP_DIRECTIONAL_FORMAT
PDF.
Definition: uchar.h:962
U_CDECL_END
#define U_CDECL_END
Definition: umachine.h:85
UCHAR_CASED
Binary property Cased.
Definition: uchar.h:419
UBLOCK_MRO
Definition: uchar.h:1648
UCPMap
struct UCPMap UCPMap
Abstract map from Unicode code points (U+0000..U+10FFFF) to integer values.
Definition: ucpmap.h:33
U_JG_HANIFI_ROHINGYA_KINNA_YA
Definition: uchar.h:2066
U_CDECL_BEGIN
#define U_CDECL_BEGIN
Definition: umachine.h:84
UBLOCK_ANATOLIAN_HIEROGLYPHS
Definition: uchar.h:1685
U_INPC_NA
Definition: uchar.h:2389
UBLOCK_HANUNOO
Definition: uchar.h:1348
UBLOCK_BYZANTINE_MUSICAL_SYMBOLS
Definition: uchar.h:1324
UBLOCK_COMBINING_HALF_MARKS
Definition: uchar.h:1298
UCHAR_CHANGES_WHEN_CASEFOLDED
Binary property Changes_When_Casefolded.
Definition: uchar.h:429
UBLOCK_THAI
Definition: uchar.h:1106
U_JG_MALAYALAM_TTA
Definition: uchar.h:2064
UBLOCK_KATAKANA
Definition: uchar.h:1223
UBLOCK_ANCIENT_GREEK_MUSICAL_NOTATION
Definition: uchar.h:1408
U_CHAR_NAME_CHOICE_COUNT
One more than the highest normal UCharNameChoice value.
Definition: uchar.h:1849
UBLOCK_BOX_DRAWING
Definition: uchar.h:1190
UCHAR_POSIX_ALNUM
Binary property alnum (a C/POSIX character class).
Definition: uchar.h:397
u_isUWhiteSpace
UBool u_isUWhiteSpace(UChar32 c)
Check if a code point has the White_Space Unicode property.
UBLOCK_CHESS_SYMBOLS
Definition: uchar.h:1748
u_isJavaIDPart
UBool u_isJavaIDPart(UChar32 c)
Determines if the specified character is permissible in a Java identifier.
UBLOCK_VERTICAL_FORMS
Definition: uchar.h:1446
U_GCB_E_BASE
Definition: uchar.h:2110
UBLOCK_AVESTAN
Definition: uchar.h:1541
UBLOCK_TAI_XUAN_JING_SYMBOLS
Definition: uchar.h:1401
UCHAR_TRAIL_CANONICAL_COMBINING_CLASS
Enumerated property Trail_Canonical_Combining_Class.
Definition: uchar.h:558
U_INSC_CONSONANT_SUBJOINED
Definition: uchar.h:2464
UBLOCK_BATAK
Definition: uchar.h:1566
U_RIGHT_TO_LEFT_OVERRIDE
RLO.
Definition: uchar.h:960
UCHAR_DEPRECATED
Binary property Deprecated (new in Unicode 3.2).
Definition: uchar.h:231
U_BPT_NONE
Not a paired bracket.
Definition: uchar.h:1000
UCHAR_UPPERCASE
Binary property Uppercase.
Definition: uchar.h:317
U_LB_JT
Definition: uchar.h:2289
UBLOCK_SYRIAC_SUPPLEMENT
Definition: uchar.h:1741
UBLOCK_LATIN_EXTENDED_A
Definition: uchar.h:1037
UDecompositionType
UDecompositionType
Decomposition Type constants.
Definition: uchar.h:1884
U_UPPERCASE_LETTER
Lu.
Definition: uchar.h:738
UBLOCK_IMPERIAL_ARAMAIC
Definition: uchar.h:1537
UBLOCK_GUJARATI
Definition: uchar.h:1085
u_toupper
UChar32 u_toupper(UChar32 c)
The given character is mapped to its uppercase equivalent according to UnicodeData....
UBLOCK_TAI_THAM
Definition: uchar.h:1513
U_LEFT_TO_RIGHT
L.
Definition: uchar.h:930
UBLOCK_PRIVATE_USE_AREA
Same as UBLOCK_PRIVATE_USE.
Definition: uchar.h:1276
U_WB_E_BASE_GAZ
Definition: uchar.h:2174
U_START_PUNCTUATION
Ps.
Definition: uchar.h:776
UBLOCK_GRANTHA
Definition: uchar.h:1630
U_LB_HEBREW_LETTER
Definition: uchar.h:2297
U_GENERAL_OTHER_TYPES
Cn "Other, Not Assigned (no characters in [UnicodeData.txt] have this property)" (same as U_UNASSIGNE...
Definition: uchar.h:736
UBLOCK_HIGH_PRIVATE_USE_SURROGATES
Definition: uchar.h:1262
U_JG_FE
Definition: uchar.h:2015
UBLOCK_INVALID_CODE
Definition: uchar.h:1781
UCHAR_S_TERM
Binary property STerm (new in Unicode 4.0.1).
Definition: uchar.h:337
UBLOCK_CHEROKEE
Definition: uchar.h:1127
u_getPropertyEnum
UProperty u_getPropertyEnum(const char *alias)
Return the UProperty enum for a given property name, as specified in the Unicode database file Proper...
U_UNASSIGNED
Non-category for unassigned and non-character code points.
Definition: uchar.h:734
UCHAR_BIDI_CLASS
Enumerated property Bidi_Class.
Definition: uchar.h:496
U_JG_HANIFI_ROHINGYA_PA
Definition: uchar.h:2067
UCHAR_EMOJI
Binary property Emoji.
Definition: uchar.h:440
UBLOCK_SINHALA
Definition: uchar.h:1103
U_JG_MALAYALAM_NNA
Definition: uchar.h:2059
UBLOCK_SYLOTI_NAGRI
Definition: uchar.h:1442
U_INSC_VOWEL_DEPENDENT
Definition: uchar.h:2502
u_istitle
UBool u_istitle(UChar32 c)
Determines whether the specified code point is a titlecase letter.
UBLOCK_MAYAN_NUMERALS
Definition: uchar.h:1762
UBLOCK_MEETEI_MAYEK
Definition: uchar.h:1533
UCHAR_TERMINAL_PUNCTUATION
Binary property Terminal_Punctuation.
Definition: uchar.h:310
UBLOCK_MULTANI
Definition: uchar.h:1695
UBLOCK_PSALTER_PAHLAVI
Definition: uchar.h:1666
UBLOCK_OLD_ITALIC
Definition: uchar.h:1318
U_INSC_AVAGRAHA
Definition: uchar.h:2436
UBLOCK_BAMUM_SUPPLEMENT
Definition: uchar.h:1572
UCHAR_NFC_QUICK_CHECK
Enumerated property NFC_Quick_Check.
Definition: uchar.h:541
UBLOCK_CYRILLIC
Definition: uchar.h:1058
UCHAR_JOIN_CONTROL
Binary property Join_Control.
Definition: uchar.h:282
UBLOCK_ALCHEMICAL_SYMBOLS
Definition: uchar.h:1584
UBLOCK_MUSICAL_SYMBOLS
Definition: uchar.h:1326
UBLOCK_BRAILLE_PATTERNS
Definition: uchar.h:1205
U_JG_MALAYALAM_BHA
Definition: uchar.h:2054
U_JG_MANICHAEAN_TETH
Definition: uchar.h:2043
U_INSC_CONSONANT_INITIAL_POSTFIXED
Definition: uchar.h:2452
u_isprint
UBool u_isprint(UChar32 c)
Determines whether the specified code point is a printable character.
UCHAR_SCRIPT
Enumerated property Script.
Definition: uchar.h:529
UBLOCK_CUNEIFORM
Definition: uchar.h:1463
UBLOCK_CURRENCY_SYMBOLS
Definition: uchar.h:1157
u_getIntPropertyMap
const U_CAPI UCPMap * u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode)
Returns an immutable UCPMap for an enumerated/catalog/int-valued property.
U_GCB_E_BASE_GAZ
Definition: uchar.h:2112
UCHAR_SOFT_DOTTED
Binary property Soft_Dotted (new in Unicode 3.2).
Definition: uchar.h:306
UBLOCK_BLOCK_ELEMENTS
Definition: uchar.h:1193
UBLOCK_VARIATION_SELECTORS_SUPPLEMENT
Definition: uchar.h:1403
UCharEnumTypeRange
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:3324
UBLOCK_ENCLOSED_CJK_LETTERS_AND_MONTHS
Definition: uchar.h:1238
UCHAR_GENERAL_CATEGORY
Enumerated property General_Category.
Definition: uchar.h:514
UCHAR_PREPENDED_CONCATENATION_MARK
Binary property Prepended_Concatenation_Mark.
Definition: uchar.h:478
U_INPC_TOP_AND_LEFT_AND_RIGHT
Definition: uchar.h:2413
U_BPT_OPEN
Open paired bracket.
Definition: uchar.h:1002
UBLOCK_MISCELLANEOUS_SYMBOLS_AND_ARROWS
Definition: uchar.h:1383
U_WB_HEBREW_LETTER
Definition: uchar.h:2166
u_isxdigit
UBool u_isxdigit(UChar32 c)
Determines whether the specified code point is a hexadecimal digit.
UCHAR_EAST_ASIAN_WIDTH
Enumerated property East_Asian_Width.
Definition: uchar.h:511
U_VO_UPRIGHT
Definition: uchar.h:2527
U_JG_AFRICAN_FEH
Definition: uchar.h:2050
U_LB_CONDITIONAL_JAPANESE_STARTER
Definition: uchar.h:2295
UWordBreakValues
UWordBreakValues
Word Break constants.
Definition: uchar.h:2138
U_JG_MANICHAEAN_SADHE
Definition: uchar.h:2039
U_END_PUNCTUATION
Pe.
Definition: uchar.h:778
UBLOCK_PAU_CIN_HAU
Definition: uchar.h:1664
UBLOCK_SINHALA_ARCHAIC_NUMBERS
Definition: uchar.h:1672
ucpmap.h
UBLOCK_ETHIOPIC
Definition: uchar.h:1124
U_MATH_SYMBOL
Sm.
Definition: uchar.h:784
U_INSC_CONSONANT_MEDIAL
Definition: uchar.h:2456
UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS
Definition: uchar.h:1313
UCHAR_SIMPLE_LOWERCASE_MAPPING
String property Simple_Lowercase_Mapping.
Definition: uchar.h:668
UBLOCK_CHAKMA
Definition: uchar.h:1595
UCHAR_CHANGES_WHEN_TITLECASED
Binary property Changes_When_Titlecased.
Definition: uchar.h:427
UEastAsianWidth
UEastAsianWidth
East Asian Width constants.
Definition: uchar.h:1794
u_isIDPart
UBool u_isIDPart(UChar32 c)
Determines if the specified character is permissible in an identifier according to Java.
U_GCB_E_MODIFIER
Definition: uchar.h:2114
UBLOCK_SHAVIAN
Definition: uchar.h:1395
u_isgraph
UBool u_isgraph(UChar32 c)
Determines whether the specified code point is a "graphic" character (printable, excluding spaces).
UBLOCK_DEVANAGARI_EXTENDED
Definition: uchar.h:1523
U_INITIAL_PUNCTUATION
Pi.
Definition: uchar.h:792
U_INSC_OTHER
Definition: uchar.h:2434
U_GCB_SPACING_MARK
Definition: uchar.h:2104
UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B
Definition: uchar.h:1360
UBLOCK_LATIN_EXTENDED_ADDITIONAL
Definition: uchar.h:1145
u_ispunct
UBool u_ispunct(UChar32 c)
Determines whether the specified code point is a punctuation character.
UBLOCK_PHONETIC_EXTENSIONS_SUPPLEMENT
Definition: uchar.h:1438
UBLOCK_KHAROSHTHI
Definition: uchar.h:1430
U_JG_MANICHAEAN_TWENTY
Definition: uchar.h:2045
UBLOCK_ELBASAN
Definition: uchar.h:1626
U_INSC_CONSONANT_PLACEHOLDER
Definition: uchar.h:2458
UBLOCK_BASSA_VAH
Definition: uchar.h:1616
UBLOCK_TAMIL
Definition: uchar.h:1091
U_INSC_REGISTER_SHIFTER
Definition: uchar.h:2488
U_JG_MANICHAEAN_HUNDRED
Definition: uchar.h:2030
UBLOCK_TAGS
Definition: uchar.h:1334
UBLOCK_MYANMAR
Definition: uchar.h:1115
UBLOCK_LATIN_1_SUPPLEMENT
Definition: uchar.h:1034
UBLOCK_LATIN_EXTENDED_B
Definition: uchar.h:1040
UBLOCK_ENCLOSED_ALPHANUMERICS
Definition: uchar.h:1187
UCHAR_INT_LIMIT
One more than the last constant for enumerated/integer Unicode properties.
Definition: uchar.h:606
UBLOCK_LINEAR_B_SYLLABARY
Definition: uchar.h:1387
UCHAR_EMOJI_PRESENTATION
Binary property Emoji_Presentation.
Definition: uchar.h:447
U_INSC_VOWEL
Definition: uchar.h:2500
UBLOCK_GLAGOLITIC_SUPPLEMENT
Definition: uchar.h:1712
U_JG_MALAYALAM_RA
Definition: uchar.h:2062
UCHAR_STRING_START
First constant for string Unicode properties.
Definition: uchar.h:645
U_CONTROL_CHAR
Cc.
Definition: uchar.h:766
UBLOCK_LINEAR_B_IDEOGRAMS
Definition: uchar.h:1389
U_JG_MANICHAEAN_DALETH
Definition: uchar.h:2025
UBLOCK_GEORGIAN_SUPPLEMENT
Definition: uchar.h:1426
UBLOCK_OLD_SOGDIAN
Definition: uchar.h:1766
UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B
Definition: uchar.h:1330
UBLOCK_COPTIC
Definition: uchar.h:1420
U_UNICODE_CHAR_NAME
Unicode character name (Name property).
Definition: uchar.h:1831
u_getPropertyName
const char * u_getPropertyName(UProperty property, UPropertyNameChoice nameChoice)
Return the Unicode name for a given property, as given in the Unicode database file PropertyAliases....
UBLOCK_HANGUL_JAMO_EXTENDED_B
Definition: uchar.h:1535
UBLOCK_INSCRIPTIONAL_PAHLAVI
Definition: uchar.h:1545
UBLOCK_SUPPLEMENTAL_PUNCTUATION
Definition: uchar.h:1440
U_INSC_CONSONANT_DEAD
Definition: uchar.h:2446
U_JG_BURUSHASKI_YEH_BARREE
Definition: uchar.h:2018
UCHAR_SEGMENT_STARTER
Binary Property Segment_Starter.
Definition: uchar.h:382
UBLOCK_SHARADA
Definition: uchar.h:1605
U_INSC_CONSONANT_FINAL
Definition: uchar.h:2448
UBLOCK_OLD_PERSIAN
Definition: uchar.h:1436
UBLOCK_KHOJKI
Definition: uchar.h:1632
UBLOCK_YI_RADICALS
Definition: uchar.h:1253
UBLOCK_DOMINO_TILES
Definition: uchar.h:1504
U_LB_E_BASE
Definition: uchar.h:2301
U_JG_KHAPH
Definition: uchar.h:2016
UCHAR_NFKD_QUICK_CHECK
Enumerated property NFKD_Quick_Check.
Definition: uchar.h:538
U_LB_COUNT
One more than the highest normal ULineBreak value.
Definition: uchar.h:2313
U_EA_COUNT
One more than the highest normal UEastAsianWidth value.
Definition: uchar.h:1814
u_isISOControl
UBool u_isISOControl(UChar32 c)
Determines whether the specified code point is an ISO control code.
UCHAR_VARIATION_SELECTOR
Binary property Variation_Selector (new in Unicode 4.0.1).
Definition: uchar.h:343
UCHAR_ID_CONTINUE
Binary property ID_Continue.
Definition: uchar.h:264
UBLOCK_ARABIC_PRESENTATION_FORMS_A
Definition: uchar.h:1295
UBLOCK_TAI_VIET
Definition: uchar.h:1531
U_HST_COUNT
One more than the highest normal UHangulSyllableType value.
Definition: uchar.h:2371
UBLOCK_SUPPLEMENTAL_MATHEMATICAL_OPERATORS
Definition: uchar.h:1362
U_DASH_PUNCTUATION
Pd.
Definition: uchar.h:774
U_EXTENDED_CHAR_NAME
Standard or synthetic character name.
Definition: uchar.h:1841
UBLOCK_SIDDHAM
Definition: uchar.h:1670
UCHAR_BIDI_CONTROL
Binary property Bidi_Control.
Definition: uchar.h:217
U_STABLE
#define U_STABLE
This is used to declare a function as a stable public ICU C API.
Definition: umachine.h:111
UCHAR_PATTERN_SYNTAX
Binary property Pattern_Syntax (new in Unicode 4.1).
Definition: uchar.h:387
U_JG_MANICHAEAN_ONE
Definition: uchar.h:2035
UCHAR_LOWERCASE
Binary property Lowercase.
Definition: uchar.h:289
u_isWhitespace
UBool u_isWhitespace(UChar32 c)
Determines if the specified code point is a whitespace character according to Java/ICU.
U_JG_MALAYALAM_JA
Definition: uchar.h:2055
UBLOCK_PAHAWH_HMONG
Definition: uchar.h:1660
UBLOCK_MONGOLIAN
Definition: uchar.h:1142
UBLOCK_COUNTING_ROD_NUMERALS
Definition: uchar.h:1467
UBLOCK_OLD_NORTH_ARABIAN
Definition: uchar.h:1654
UBLOCK_HANGUL_COMPATIBILITY_JAMO
Definition: uchar.h:1229
u_islower
UBool u_islower(UChar32 c)
Determines whether the specified code point has the general category "Ll" (lowercase letter).
UBLOCK_PHOENICIAN
Definition: uchar.h:1461
u_getIntPropertyMinValue
int32_t u_getIntPropertyMinValue(UProperty which)
Get the minimum value for an enumerated/integer/binary Unicode property.
U_INPC_BOTTOM
Definition: uchar.h:2391
U_INSC_CONSONANT
Definition: uchar.h:2444
UBLOCK_KANBUN
Definition: uchar.h:1232
UBLOCK_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS
Definition: uchar.h:1578
U_WB_DOUBLE_QUOTE
Definition: uchar.h:2170
UBLOCK_BAMUM
Definition: uchar.h:1519
UBLOCK_ARROWS
Definition: uchar.h:1172
U_JG_MALAYALAM_LLLA
Definition: uchar.h:2057
UBLOCK_LATIN_EXTENDED_D
Definition: uchar.h:1457
UBidiPairedBracketType
UBidiPairedBracketType
Bidi Paired Bracket Type constants.
Definition: uchar.h:992
U_INSC_GEMINATION_MARK
Definition: uchar.h:2470
UCHAR_BLOCK
Enumerated property Block.
Definition: uchar.h:501
U_JG_MANICHAEAN_HETH
Definition: uchar.h:2029
UBLOCK_GEOMETRIC_SHAPES
Definition: uchar.h:1196
UBLOCK_LEPCHA
Definition: uchar.h:1474
UCHAR_FULL_COMPOSITION_EXCLUSION
Binary property Full_Composition_Exclusion.
Definition: uchar.h:242
UCHAR_STRING_LIMIT
One more than the last constant for string Unicode properties.
Definition: uchar.h:696
u_charDigitValue
int32_t u_charDigitValue(UChar32 c)
Returns the decimal digit value of a decimal digit character.
UBLOCK_KHMER
Definition: uchar.h:1139
U_JG_MANICHAEAN_AYIN
Definition: uchar.h:2023
UBLOCK_CYRILLIC_SUPPLEMENTARY
Unicode 4.0.1 renames the "Cyrillic Supplementary" block to "Cyrillic Supplement".
Definition: uchar.h:1344
UChar
uint16_t UChar
Definition: umachine.h:342
U_JG_MALAYALAM_SSA
Definition: uchar.h:2063
u_totitle
UChar32 u_totitle(UChar32 c)
The given character is mapped to its titlecase equivalent according to UnicodeData....
UCHAR_OTHER_PROPERTY_START
First constant for Unicode properties with unusual value types.
Definition: uchar.h:706
UCHAR_OTHER_PROPERTY_LIMIT
One more than the last constant for Unicode properties with unusual value types.
Definition: uchar.h:712
U_BPT_COUNT
One more than the highest normal UBidiPairedBracketType value.
Definition: uchar.h:1012
U_JG_MANICHAEAN_THAMEDH
Definition: uchar.h:2044
UVersionInfo
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
U_JG_MANICHAEAN_BETH
Definition: uchar.h:2024
UBLOCK_LINEAR_A
Definition: uchar.h:1638
UCHAR_UNIFIED_IDEOGRAPH
Binary property Unified_Ideograph (new in Unicode 3.2).
Definition: uchar.h:314
UCHAR_SIMPLE_CASE_FOLDING
String property Simple_Case_Folding.
Definition: uchar.h:665
UVerticalOrientation
UVerticalOrientation
Vertical Orientation constants.
Definition: uchar.h:2513
UBLOCK_SUNDANESE_SUPPLEMENT
Definition: uchar.h:1609
UCHAR_GRAPHEME_LINK
Binary property Grapheme_Link (new in Unicode 3.2).
Definition: uchar.h:253
u_forDigit
UChar32 u_forDigit(int32_t digit, int8_t radix)
Determines the character representation for a specific digit in the specified radix.
UCHAR_SCRIPT_EXTENSIONS
Miscellaneous property Script_Extensions (new in Unicode 6.0).
Definition: uchar.h:704
u_hasBinaryProperty
UBool u_hasBinaryProperty(UChar32 c, UProperty which)
Check a binary Unicode property for a code point.
U_INSC_NUMBER
Definition: uchar.h:2482
UBLOCK_ARMENIAN
Definition: uchar.h:1061
U_JG_MALAYALAM_NYA
Definition: uchar.h:2061
UCHAR_LEAD_CANONICAL_COMBINING_CLASS
Enumerated property Lead_Canonical_Combining_Class.
Definition: uchar.h:551
UBLOCK_GOTHIC
Definition: uchar.h:1320
u_getUnicodeVersion
void u_getUnicodeVersion(UVersionInfo versionArray)
Gets the Unicode version information.
u_isupper
UBool u_isupper(UChar32 c)
Determines whether the specified code point has the general category "Lu" (uppercase letter).
U_JG_MANICHAEAN_NUN
Definition: uchar.h:2034
UCHAR_HEX_DIGIT
Binary property Hex_Digit.
Definition: uchar.h:256
U_WB_CR
Definition: uchar.h:2154
ULineBreak
ULineBreak
Line Break constants.
Definition: uchar.h:2240
UCHAR_XID_START
Binary property XID_Start.
Definition: uchar.h:328
UBLOCK_SPACING_MODIFIER_LETTERS
Definition: uchar.h:1046
UBLOCK_HEBREW
Definition: uchar.h:1064
UCHAR_UNICODE_1_NAME
String property Unicode_1_Name.
Definition: uchar.h:683
U_JG_ZHAIN
Definition: uchar.h:2017
UBLOCK_CHAM
Definition: uchar.h:1490
UBLOCK_GLAGOLITIC
Definition: uchar.h:1428
UBLOCK_MIAO
Definition: uchar.h:1603
UCHAR_BINARY_LIMIT
One more than the last constant for binary Unicode properties.
Definition: uchar.h:491
UCHAR_PATTERN_WHITE_SPACE
Binary property Pattern_White_Space (new in Unicode 4.1).
Definition: uchar.h:392
UBLOCK_LISU
Definition: uchar.h:1517
U_WB_SINGLE_QUOTE
Definition: uchar.h:2168
U_INSC_NUKTA
Definition: uchar.h:2480
UBLOCK_OLD_HUNGARIAN
Definition: uchar.h:1697
UBLOCK_CYRILLIC_SUPPLEMENT
Definition: uchar.h:1339
U_JG_MANICHAEAN_ALEPH
Definition: uchar.h:2022
UBLOCK_MYANMAR_EXTENDED_B
Definition: uchar.h:1650
U_CAPI
#define U_CAPI
This is used to declare a function as a public ICU C API.
Definition: umachine.h:109
UBLOCK_MAKASAR
Definition: uchar.h:1760
U_JG_FARSI_YEH
Definition: uchar.h:2019
U_FORMAT_CHAR
Cf.
Definition: uchar.h:768
UBLOCK_CONTROL_PICTURES
Definition: uchar.h:1181
UBLOCK_OPTICAL_CHARACTER_RECOGNITION
Definition: uchar.h:1184
UBLOCK_GUNJALA_GONDI
Definition: uchar.h:1754
UBLOCK_CJK_UNIFIED_IDEOGRAPHS
Definition: uchar.h:1247
UCHAR_NAME
String property Name.
Definition: uchar.h:662
UBLOCK_CJK_RADICALS_SUPPLEMENT
Definition: uchar.h:1208
UBLOCK_COPTIC_EPACT_NUMBERS
Definition: uchar.h:1620
UJoiningGroup
UJoiningGroup
Joining Group constants.
Definition: uchar.h:1956
UBLOCK_CYRILLIC_EXTENDED_C
Definition: uchar.h:1710
UBLOCK_SAMARITAN
Definition: uchar.h:1509
U_GCB_REGIONAL_INDICATOR
Definition: uchar.h:2108
UBLOCK_SUPERSCRIPTS_AND_SUBSCRIPTS
Definition: uchar.h:1154
UCHAR_GRAPHEME_EXTEND
Binary property Grapheme_Extend (new in Unicode 3.2).
Definition: uchar.h:250
U_INSC_NUMBER_JOINER
Definition: uchar.h:2484
U_INSC_TONE_LETTER
Definition: uchar.h:2492
U_JT_COUNT
One more than the highest normal UJoiningType value.
Definition: uchar.h:1946
UBLOCK_KHMER_SYMBOLS
Definition: uchar.h:1379
UCharNameChoice
UCharNameChoice
Selector constants for u_charName().
Definition: uchar.h:1829
USet
struct USet USet
USet is the C API type corresponding to C++ class UnicodeSet.
Definition: uset.h:47
U_GCB_COUNT
One more than the highest normal UGraphemeClusterBreak value.
Definition: uchar.h:2127
UBLOCK_TAGALOG
Definition: uchar.h:1346
U_INSC_CONSONANT_WITH_STACKER
Definition: uchar.h:2468
UBLOCK_CJK_SYMBOLS_AND_PUNCTUATION
Definition: uchar.h:1217
U_INPC_LEFT_AND_RIGHT
Definition: uchar.h:2399
U_INSC_INVISIBLE_STACKER
Definition: uchar.h:2472
u_getNumericValue
double u_getNumericValue(UChar32 c)
Get the numeric value for a Unicode code point as defined in the Unicode Character Database.
U_JG_ROHINGYA_YEH
Definition: uchar.h:2021
stringoptions.h
C API: Bit set option bit constants for various string and character processing functions.
U_BLOCK_SEPARATOR
B.
Definition: uchar.h:944
UNumericType
UNumericType
Numeric Type constants.
Definition: uchar.h:2323
U_INSC_BINDU
Definition: uchar.h:2438
UBLOCK_TAGBANWA
Definition: uchar.h:1352
u_isIDStart
UBool u_isIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in an identifier accordin...
U_LB_JV
Definition: uchar.h:2291
U_JG_MANICHAEAN_KAPH
Definition: uchar.h:2031
U_INSC_PURE_KILLER
Definition: uchar.h:2486
UBLOCK_YI_SYLLABLES
Definition: uchar.h:1250
UBLOCK_AHOM
Definition: uchar.h:1683
UBLOCK_PHONETIC_EXTENSIONS
Definition: uchar.h:1381
UBLOCK_MODI
Definition: uchar.h:1646
UBLOCK_HANGUL_JAMO
Definition: uchar.h:1121
U_SEGMENT_SEPARATOR
S.
Definition: uchar.h:946
UBLOCK_LOW_SURROGATES
Definition: uchar.h:1265
U_WB_MIDNUMLET
Definition: uchar.h:2160
UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A
Definition: uchar.h:1354
U_BOUNDARY_NEUTRAL
BN.
Definition: uchar.h:966
UBLOCK_SUPPLEMENTAL_ARROWS_B
Definition: uchar.h:1358
UCHAR_CASE_SENSITIVE
Binary property Case_Sensitive.
Definition: uchar.h:332
U_RIGHT_TO_LEFT_EMBEDDING
RLE.
Definition: uchar.h:958
u_getFC_NFKC_Closure
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.
UBLOCK_WARANG_CITI
Definition: uchar.h:1678
UBLOCK_TIRHUTA
Definition: uchar.h:1676
UBLOCK_KAITHI
Definition: uchar.h:1551
u_isULowercase
UBool u_isULowercase(UChar32 c)
Check if a code point has the Lowercase Unicode property.
U_INPC_TOP
Definition: uchar.h:2405
UBLOCK_MENDE_KIKAKUI
Definition: uchar.h:1644
UCHAR_EMOJI_COMPONENT
Binary property Emoji_Component.
Definition: uchar.h:468
UIndicPositionalCategory
UIndicPositionalCategory
Indic Positional Category constants.
Definition: uchar.h:2381
UBLOCK_MYANMAR_EXTENDED_A
Definition: uchar.h:1529
UBLOCK_JAVANESE
Definition: uchar.h:1527
U_INPC_TOP_AND_RIGHT
Definition: uchar.h:2415
UBLOCK_TELUGU
Definition: uchar.h:1094
U_OTHER_SYMBOL
So.
Definition: uchar.h:790
UCHAR_XID_CONTINUE
Binary property XID_Continue.
Definition: uchar.h:325
U_WB_E_MODIFIER
Definition: uchar.h:2176
UBLOCK_OSMANYA
Definition: uchar.h:1397
UBLOCK_HATRAN
Definition: uchar.h:1693
UCHAR_NFC_INERT
Binary property NFC_Inert.
Definition: uchar.h:364