ICU 65.1 65.1
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"
30#include "unicode/ucpmap.h"
31
32#if !defined(USET_DEFINED) && !defined(U_IN_DOXYGEN)
33
34#define USET_DEFINED
35
44typedef struct USet USet;
45
46#endif
47
48
50
51/*==========================================================================*/
52/* Unicode version number */
53/*==========================================================================*/
63#define U_UNICODE_VERSION "12.1"
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
195typedef 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
492#endif // U_HIDE_DEPRECATED_API
493
601#ifndef U_HIDE_DEPRECATED_API
607#endif // U_HIDE_DEPRECATED_API
608
620#ifndef U_HIDE_DEPRECATED_API
626#endif // U_HIDE_DEPRECATED_API
627
633#ifndef U_HIDE_DEPRECATED_API
639#endif // U_HIDE_DEPRECATED_API
640
643 UCHAR_AGE=0x4000,
652#ifndef U_HIDE_DEPRECATED_API
656#endif /* U_HIDE_DEPRECATED_API */
678#ifndef U_HIDE_DEPRECATED_API
684#endif /* U_HIDE_DEPRECATED_API */
691#ifndef U_HIDE_DEPRECATED_API
697#endif // U_HIDE_DEPRECATED_API
698
707#ifndef U_HIDE_DEPRECATED_API
713#endif // U_HIDE_DEPRECATED_API
714
718
724typedef 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
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)
824#define U_GC_LL_MASK U_MASK(U_LOWERCASE_LETTER)
826#define U_GC_LT_MASK U_MASK(U_TITLECASE_LETTER)
828#define U_GC_LM_MASK U_MASK(U_MODIFIER_LETTER)
830#define U_GC_LO_MASK U_MASK(U_OTHER_LETTER)
831
833#define U_GC_MN_MASK U_MASK(U_NON_SPACING_MARK)
835#define U_GC_ME_MASK U_MASK(U_ENCLOSING_MARK)
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)
842#define U_GC_NL_MASK U_MASK(U_LETTER_NUMBER)
844#define U_GC_NO_MASK U_MASK(U_OTHER_NUMBER)
845
847#define U_GC_ZS_MASK U_MASK(U_SPACE_SEPARATOR)
849#define U_GC_ZL_MASK U_MASK(U_LINE_SEPARATOR)
851#define U_GC_ZP_MASK U_MASK(U_PARAGRAPH_SEPARATOR)
852
854#define U_GC_CC_MASK U_MASK(U_CONTROL_CHAR)
856#define U_GC_CF_MASK U_MASK(U_FORMAT_CHAR)
858#define U_GC_CO_MASK U_MASK(U_PRIVATE_USE_CHAR)
860#define U_GC_CS_MASK U_MASK(U_SURROGATE)
861
863#define U_GC_PD_MASK U_MASK(U_DASH_PUNCTUATION)
865#define U_GC_PS_MASK U_MASK(U_START_PUNCTUATION)
867#define U_GC_PE_MASK U_MASK(U_END_PUNCTUATION)
869#define U_GC_PC_MASK U_MASK(U_CONNECTOR_PUNCTUATION)
871#define U_GC_PO_MASK U_MASK(U_OTHER_PUNCTUATION)
872
874#define U_GC_SM_MASK U_MASK(U_MATH_SYMBOL)
876#define U_GC_SC_MASK U_MASK(U_CURRENCY_SYMBOL)
878#define U_GC_SK_MASK U_MASK(U_MODIFIER_SYMBOL)
880#define U_GC_SO_MASK U_MASK(U_OTHER_SYMBOL)
881
883#define U_GC_PI_MASK U_MASK(U_INITIAL_PUNCTUATION)
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
921typedef 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
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
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
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
1149
1152
1155
1158
1164
1167
1169 UBLOCK_NUMBER_FORMS =45, /*[2150]*/
1170
1172 UBLOCK_ARROWS =46, /*[2190]*/
1173
1176
1179
1182
1185
1188
1190 UBLOCK_BOX_DRAWING =52, /*[2500]*/
1191
1194
1197
1200
1202 UBLOCK_DINGBATS =56, /*[2700]*/
1203
1206
1209
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
1257
1260
1263
1266
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]*/
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]*/
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]*/
1404
1405 /* New blocks in Unicode 4.1 */
1406
1414 UBLOCK_BUGINESE = 129, /*[1A00]*/
1416 UBLOCK_CJK_STROKES = 130, /*[31C0]*/
1420 UBLOCK_COPTIC = 132, /*[2C80]*/
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]*/
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]*/
1480 UBLOCK_VAI = 159, /*[A500]*/
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]*/
1527 UBLOCK_JAVANESE = 181, /*[A980]*/
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]*/
1547 UBLOCK_OLD_TURKIC = 191, /*[10C00]*/
1551 UBLOCK_KAITHI = 193, /*[11080]*/
1560
1561 /* New blocks in Unicode 6.0 */
1562
1564 UBLOCK_MANDAIC = 198, /*[0840]*/
1566 UBLOCK_BATAK = 199, /*[1BC0]*/
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
1595 UBLOCK_CHAKMA = 212, /*[11100]*/
1599 UBLOCK_MEROITIC_CURSIVE = 214, /*[109A0]*/
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]*/
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]*/
1652 UBLOCK_NABATAEAN = 239, /*[10880]*/
1654 UBLOCK_OLD_NORTH_ARABIAN = 240, /*[10A80]*/
1656 UBLOCK_OLD_PERMIC = 241, /*[10350]*/
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]*/
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]*/
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]*/
1716 UBLOCK_MARCHEN = 268, /*[11C70]*/
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]*/
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]*/
1754 UBLOCK_GUNJALA_GONDI = 284, /*[11D60]*/
1756 UBLOCK_HANIFI_ROHINGYA = 285, /*[10D00]*/
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]*/
1787 UBLOCK_TAMIL_SUPPLEMENT = 299, /*[11FC0]*/
1789 UBLOCK_WANCHO = 300, /*[1E2C0]*/
1790
1791#ifndef U_HIDE_DEPRECATED_API
1799#endif // U_HIDE_DEPRECATED_API
1800
1804
1807
1815typedef enum UEastAsianWidth {
1816 /*
1817 * Note: UEastAsianWidth constants are parsed by preparseucd.py.
1818 * It matches lines like
1819 * U_EA_<Unicode East_Asian_Width value name>
1820 */
1821
1822 U_EA_NEUTRAL, /*[N]*/
1823 U_EA_AMBIGUOUS, /*[A]*/
1824 U_EA_HALFWIDTH, /*[H]*/
1825 U_EA_FULLWIDTH, /*[F]*/
1826 U_EA_NARROW, /*[Na]*/
1827 U_EA_WIDE, /*[W]*/
1828#ifndef U_HIDE_DEPRECATED_API
1836#endif // U_HIDE_DEPRECATED_API
1838
1850typedef enum UCharNameChoice {
1853#ifndef U_HIDE_DEPRECATED_API
1860#endif /* U_HIDE_DEPRECATED_API */
1865#ifndef U_HIDE_DEPRECATED_API
1871#endif // U_HIDE_DEPRECATED_API
1873
1888 U_SHORT_PROPERTY_NAME,
1889 U_LONG_PROPERTY_NAME,
1890#ifndef U_HIDE_DEPRECATED_API
1896#endif // U_HIDE_DEPRECATED_API
1898
1906 /*
1907 * Note: UDecompositionType constants are parsed by preparseucd.py.
1908 * It matches lines like
1909 * U_DT_<Unicode Decomposition_Type value name>
1910 */
1911
1912 U_DT_NONE, /*[none]*/
1913 U_DT_CANONICAL, /*[can]*/
1914 U_DT_COMPAT, /*[com]*/
1915 U_DT_CIRCLE, /*[enc]*/
1916 U_DT_FINAL, /*[fin]*/
1917 U_DT_FONT, /*[font]*/
1918 U_DT_FRACTION, /*[fra]*/
1919 U_DT_INITIAL, /*[init]*/
1920 U_DT_ISOLATED, /*[iso]*/
1921 U_DT_MEDIAL, /*[med]*/
1922 U_DT_NARROW, /*[nar]*/
1923 U_DT_NOBREAK, /*[nb]*/
1924 U_DT_SMALL, /*[sml]*/
1925 U_DT_SQUARE, /*[sqr]*/
1926 U_DT_SUB, /*[sub]*/
1927 U_DT_SUPER, /*[sup]*/
1928 U_DT_VERTICAL, /*[vert]*/
1929 U_DT_WIDE, /*[wide]*/
1930#ifndef U_HIDE_DEPRECATED_API
1937 U_DT_COUNT /* 18 */
1938#endif // U_HIDE_DEPRECATED_API
1940
1947typedef enum UJoiningType {
1948 /*
1949 * Note: UJoiningType constants are parsed by preparseucd.py.
1950 * It matches lines like
1951 * U_JT_<Unicode Joining_Type value name>
1952 */
1953
1954 U_JT_NON_JOINING, /*[U]*/
1955 U_JT_JOIN_CAUSING, /*[C]*/
1956 U_JT_DUAL_JOINING, /*[D]*/
1957 U_JT_LEFT_JOINING, /*[L]*/
1958 U_JT_RIGHT_JOINING, /*[R]*/
1959 U_JT_TRANSPARENT, /*[T]*/
1960#ifndef U_HIDE_DEPRECATED_API
1967 U_JT_COUNT /* 6 */
1968#endif // U_HIDE_DEPRECATED_API
1970
1977typedef enum UJoiningGroup {
1978 /*
1979 * Note: UJoiningGroup constants are parsed by preparseucd.py.
1980 * It matches lines like
1981 * U_JG_<Unicode Joining_Group value name>
1982 */
1983
1984 U_JG_NO_JOINING_GROUP,
1985 U_JG_AIN,
1986 U_JG_ALAPH,
1987 U_JG_ALEF,
1988 U_JG_BEH,
1989 U_JG_BETH,
1990 U_JG_DAL,
1991 U_JG_DALATH_RISH,
1992 U_JG_E,
1993 U_JG_FEH,
1994 U_JG_FINAL_SEMKATH,
1995 U_JG_GAF,
1996 U_JG_GAMAL,
1997 U_JG_HAH,
1999 U_JG_HAMZA_ON_HEH_GOAL=U_JG_TEH_MARBUTA_GOAL,
2000 U_JG_HE,
2001 U_JG_HEH,
2002 U_JG_HEH_GOAL,
2003 U_JG_HETH,
2004 U_JG_KAF,
2005 U_JG_KAPH,
2006 U_JG_KNOTTED_HEH,
2007 U_JG_LAM,
2008 U_JG_LAMADH,
2009 U_JG_MEEM,
2010 U_JG_MIM,
2011 U_JG_NOON,
2012 U_JG_NUN,
2013 U_JG_PE,
2014 U_JG_QAF,
2015 U_JG_QAPH,
2016 U_JG_REH,
2017 U_JG_REVERSED_PE,
2018 U_JG_SAD,
2019 U_JG_SADHE,
2020 U_JG_SEEN,
2021 U_JG_SEMKATH,
2022 U_JG_SHIN,
2023 U_JG_SWASH_KAF,
2024 U_JG_SYRIAC_WAW,
2025 U_JG_TAH,
2026 U_JG_TAW,
2027 U_JG_TEH_MARBUTA,
2028 U_JG_TETH,
2029 U_JG_WAW,
2030 U_JG_YEH,
2031 U_JG_YEH_BARREE,
2032 U_JG_YEH_WITH_TAIL,
2033 U_JG_YUDH,
2034 U_JG_YUDH_HE,
2035 U_JG_ZAIN,
2090#ifndef U_HIDE_DEPRECATED_API
2098#endif // U_HIDE_DEPRECATED_API
2100
2108 /*
2109 * Note: UGraphemeClusterBreak constants are parsed by preparseucd.py.
2110 * It matches lines like
2111 * U_GCB_<Unicode Grapheme_Cluster_Break value name>
2112 */
2113
2114 U_GCB_OTHER = 0, /*[XX]*/
2115 U_GCB_CONTROL = 1, /*[CN]*/
2116 U_GCB_CR = 2, /*[CR]*/
2117 U_GCB_EXTEND = 3, /*[EX]*/
2118 U_GCB_L = 4, /*[L]*/
2119 U_GCB_LF = 5, /*[LF]*/
2120 U_GCB_LV = 6, /*[LV]*/
2121 U_GCB_LVT = 7, /*[LVT]*/
2122 U_GCB_T = 8, /*[T]*/
2123 U_GCB_V = 9, /*[V]*/
2125 U_GCB_SPACING_MARK = 10, /*[SM]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2127 U_GCB_PREPEND = 11, /*[PP]*/
2129 U_GCB_REGIONAL_INDICATOR = 12, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2131 U_GCB_E_BASE = 13, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2133 U_GCB_E_BASE_GAZ = 14, /*[EBG]*/
2135 U_GCB_E_MODIFIER = 15, /*[EM]*/
2137 U_GCB_GLUE_AFTER_ZWJ = 16, /*[GAZ]*/
2139 U_GCB_ZWJ = 17, /*[ZWJ]*/
2140
2141#ifndef U_HIDE_DEPRECATED_API
2148 U_GCB_COUNT = 18
2149#endif // U_HIDE_DEPRECATED_API
2151
2159typedef enum UWordBreakValues {
2160 /*
2161 * Note: UWordBreakValues constants are parsed by preparseucd.py.
2162 * It matches lines like
2163 * U_WB_<Unicode Word_Break value name>
2164 */
2165
2166 U_WB_OTHER = 0, /*[XX]*/
2167 U_WB_ALETTER = 1, /*[LE]*/
2168 U_WB_FORMAT = 2, /*[FO]*/
2169 U_WB_KATAKANA = 3, /*[KA]*/
2170 U_WB_MIDLETTER = 4, /*[ML]*/
2171 U_WB_MIDNUM = 5, /*[MN]*/
2172 U_WB_NUMERIC = 6, /*[NU]*/
2173 U_WB_EXTENDNUMLET = 7, /*[EX]*/
2175 U_WB_CR = 8, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2177 U_WB_EXTEND = 9, /*[Extend]*/
2179 U_WB_LF = 10, /*[LF]*/
2181 U_WB_MIDNUMLET =11, /*[MB]*/
2183 U_WB_NEWLINE =12, /*[NL]*/
2185 U_WB_REGIONAL_INDICATOR = 13, /*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2187 U_WB_HEBREW_LETTER = 14, /*[HL]*/ /* from here on: new in Unicode 6.3/ICU 52 */
2189 U_WB_SINGLE_QUOTE = 15, /*[SQ]*/
2191 U_WB_DOUBLE_QUOTE = 16, /*[DQ]*/
2193 U_WB_E_BASE = 17, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2195 U_WB_E_BASE_GAZ = 18, /*[EBG]*/
2197 U_WB_E_MODIFIER = 19, /*[EM]*/
2199 U_WB_GLUE_AFTER_ZWJ = 20, /*[GAZ]*/
2201 U_WB_ZWJ = 21, /*[ZWJ]*/
2203 U_WB_WSEGSPACE = 22, /*[WSEGSPACE]*/
2204
2205#ifndef U_HIDE_DEPRECATED_API
2212 U_WB_COUNT = 23
2213#endif // U_HIDE_DEPRECATED_API
2215
2222typedef enum USentenceBreak {
2223 /*
2224 * Note: USentenceBreak constants are parsed by preparseucd.py.
2225 * It matches lines like
2226 * U_SB_<Unicode Sentence_Break value name>
2227 */
2228
2229 U_SB_OTHER = 0, /*[XX]*/
2230 U_SB_ATERM = 1, /*[AT]*/
2231 U_SB_CLOSE = 2, /*[CL]*/
2232 U_SB_FORMAT = 3, /*[FO]*/
2233 U_SB_LOWER = 4, /*[LO]*/
2234 U_SB_NUMERIC = 5, /*[NU]*/
2235 U_SB_OLETTER = 6, /*[LE]*/
2236 U_SB_SEP = 7, /*[SE]*/
2237 U_SB_SP = 8, /*[SP]*/
2238 U_SB_STERM = 9, /*[ST]*/
2239 U_SB_UPPER = 10, /*[UP]*/
2240 U_SB_CR = 11, /*[CR]*/ /* from here on: new in Unicode 5.1/ICU 4.0 */
2241 U_SB_EXTEND = 12, /*[EX]*/
2242 U_SB_LF = 13, /*[LF]*/
2243 U_SB_SCONTINUE = 14, /*[SC]*/
2244#ifndef U_HIDE_DEPRECATED_API
2251 U_SB_COUNT = 15
2252#endif // U_HIDE_DEPRECATED_API
2254
2261typedef enum ULineBreak {
2262 /*
2263 * Note: ULineBreak constants are parsed by preparseucd.py.
2264 * It matches lines like
2265 * U_LB_<Unicode Line_Break value name>
2266 */
2267
2268 U_LB_UNKNOWN = 0, /*[XX]*/
2269 U_LB_AMBIGUOUS = 1, /*[AI]*/
2270 U_LB_ALPHABETIC = 2, /*[AL]*/
2271 U_LB_BREAK_BOTH = 3, /*[B2]*/
2272 U_LB_BREAK_AFTER = 4, /*[BA]*/
2273 U_LB_BREAK_BEFORE = 5, /*[BB]*/
2274 U_LB_MANDATORY_BREAK = 6, /*[BK]*/
2275 U_LB_CONTINGENT_BREAK = 7, /*[CB]*/
2276 U_LB_CLOSE_PUNCTUATION = 8, /*[CL]*/
2277 U_LB_COMBINING_MARK = 9, /*[CM]*/
2278 U_LB_CARRIAGE_RETURN = 10, /*[CR]*/
2279 U_LB_EXCLAMATION = 11, /*[EX]*/
2280 U_LB_GLUE = 12, /*[GL]*/
2281 U_LB_HYPHEN = 13, /*[HY]*/
2282 U_LB_IDEOGRAPHIC = 14, /*[ID]*/
2284 U_LB_INSEPARABLE = 15, /*[IN]*/
2285 U_LB_INSEPERABLE = U_LB_INSEPARABLE,
2286 U_LB_INFIX_NUMERIC = 16, /*[IS]*/
2287 U_LB_LINE_FEED = 17, /*[LF]*/
2288 U_LB_NONSTARTER = 18, /*[NS]*/
2289 U_LB_NUMERIC = 19, /*[NU]*/
2290 U_LB_OPEN_PUNCTUATION = 20, /*[OP]*/
2291 U_LB_POSTFIX_NUMERIC = 21, /*[PO]*/
2292 U_LB_PREFIX_NUMERIC = 22, /*[PR]*/
2293 U_LB_QUOTATION = 23, /*[QU]*/
2294 U_LB_COMPLEX_CONTEXT = 24, /*[SA]*/
2295 U_LB_SURROGATE = 25, /*[SG]*/
2296 U_LB_SPACE = 26, /*[SP]*/
2297 U_LB_BREAK_SYMBOLS = 27, /*[SY]*/
2298 U_LB_ZWSPACE = 28, /*[ZW]*/
2300 U_LB_NEXT_LINE = 29, /*[NL]*/ /* from here on: new in Unicode 4/ICU 2.6 */
2302 U_LB_WORD_JOINER = 30, /*[WJ]*/
2304 U_LB_H2 = 31, /*[H2]*/ /* from here on: new in Unicode 4.1/ICU 3.4 */
2306 U_LB_H3 = 32, /*[H3]*/
2308 U_LB_JL = 33, /*[JL]*/
2310 U_LB_JT = 34, /*[JT]*/
2312 U_LB_JV = 35, /*[JV]*/
2314 U_LB_CLOSE_PARENTHESIS = 36, /*[CP]*/ /* new in Unicode 5.2/ICU 4.4 */
2316 U_LB_CONDITIONAL_JAPANESE_STARTER = 37,/*[CJ]*/ /* new in Unicode 6.1/ICU 49 */
2318 U_LB_HEBREW_LETTER = 38, /*[HL]*/ /* new in Unicode 6.1/ICU 49 */
2320 U_LB_REGIONAL_INDICATOR = 39,/*[RI]*/ /* new in Unicode 6.2/ICU 50 */
2322 U_LB_E_BASE = 40, /*[EB]*/ /* from here on: new in Unicode 9.0/ICU 58 */
2324 U_LB_E_MODIFIER = 41, /*[EM]*/
2326 U_LB_ZWJ = 42, /*[ZWJ]*/
2327#ifndef U_HIDE_DEPRECATED_API
2334 U_LB_COUNT = 43
2335#endif // U_HIDE_DEPRECATED_API
2337
2344typedef enum UNumericType {
2345 /*
2346 * Note: UNumericType constants are parsed by preparseucd.py.
2347 * It matches lines like
2348 * U_NT_<Unicode Numeric_Type value name>
2349 */
2350
2351 U_NT_NONE, /*[None]*/
2352 U_NT_DECIMAL, /*[de]*/
2353 U_NT_DIGIT, /*[di]*/
2354 U_NT_NUMERIC, /*[nu]*/
2355#ifndef U_HIDE_DEPRECATED_API
2363#endif // U_HIDE_DEPRECATED_API
2365
2373 /*
2374 * Note: UHangulSyllableType constants are parsed by preparseucd.py.
2375 * It matches lines like
2376 * U_HST_<Unicode Hangul_Syllable_Type value name>
2377 */
2378
2379 U_HST_NOT_APPLICABLE, /*[NA]*/
2380 U_HST_LEADING_JAMO, /*[L]*/
2381 U_HST_VOWEL_JAMO, /*[V]*/
2382 U_HST_TRAILING_JAMO, /*[T]*/
2383 U_HST_LV_SYLLABLE, /*[LV]*/
2384 U_HST_LVT_SYLLABLE, /*[LVT]*/
2385#ifndef U_HIDE_DEPRECATED_API
2393#endif // U_HIDE_DEPRECATED_API
2395
2440
2527
2535 /*
2536 * Note: UVerticalOrientation constants are parsed by preparseucd.py.
2537 * It matches lines like
2538 * U_VO_<Unicode Vertical_Orientation value name>
2539 */
2540
2550
2578U_STABLE UBool U_EXPORT2
2580
2596U_CAPI const USet * U_EXPORT2
2598
2611U_STABLE UBool U_EXPORT2
2613
2626U_STABLE UBool U_EXPORT2
2628
2641U_STABLE UBool U_EXPORT2
2643
2662U_STABLE UBool U_EXPORT2
2664
2703U_STABLE int32_t U_EXPORT2
2705
2724U_STABLE int32_t U_EXPORT2
2726
2753U_STABLE int32_t U_EXPORT2
2755
2771U_CAPI const UCPMap * U_EXPORT2
2773
2796U_STABLE double U_EXPORT2
2798
2806#define U_NO_NUMERIC_VALUE ((double)-123456789.)
2807
2831U_STABLE UBool U_EXPORT2
2833
2858U_STABLE UBool U_EXPORT2
2860
2875U_STABLE UBool U_EXPORT2
2877
2896U_STABLE UBool U_EXPORT2
2898
2917U_STABLE UBool U_EXPORT2
2919
2938U_STABLE UBool U_EXPORT2
2940
2961U_STABLE UBool U_EXPORT2
2963
2977U_STABLE UBool U_EXPORT2
2979
2996U_STABLE UBool U_EXPORT2
2998
3024U_STABLE UBool U_EXPORT2
3026
3049U_STABLE UBool U_EXPORT2
3051
3070U_STABLE UBool U_EXPORT2
3072
3091U_STABLE UBool U_EXPORT2
3093
3131U_STABLE UBool U_EXPORT2
3133
3155U_STABLE UBool U_EXPORT2
3157
3170U_STABLE UBool U_EXPORT2
3172
3188U_STABLE UBool U_EXPORT2
3190
3208U_STABLE UBool U_EXPORT2
3210
3227U_STABLE UCharDirection U_EXPORT2
3229
3245U_STABLE UBool U_EXPORT2
3247
3267U_STABLE UChar32 U_EXPORT2
3269
3286U_STABLE UChar32 U_EXPORT2
3288
3300U_STABLE int8_t U_EXPORT2
3302
3316#define U_GET_GC_MASK(c) U_MASK(u_charType(c))
3317
3335typedef UBool U_CALLCONV
3336UCharEnumTypeRange(const void *context, UChar32 start, UChar32 limit, UCharCategory type);
3337
3357U_STABLE void U_EXPORT2
3358u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context);
3359
3360#if !UCONFIG_NO_NORMALIZATION
3361
3369U_STABLE uint8_t U_EXPORT2
3371
3372#endif
3373
3397U_STABLE int32_t U_EXPORT2
3399
3409U_STABLE UBlockCode U_EXPORT2
3411
3444U_STABLE int32_t U_EXPORT2
3446 char *buffer, int32_t bufferLength,
3447 UErrorCode *pErrorCode);
3448
3449#ifndef U_HIDE_DEPRECATED_API
3468U_DEPRECATED int32_t U_EXPORT2
3470 char *dest, int32_t destCapacity,
3471 UErrorCode *pErrorCode);
3472#endif /* U_HIDE_DEPRECATED_API */
3473
3494U_STABLE UChar32 U_EXPORT2
3496 const char *name,
3497 UErrorCode *pErrorCode);
3498
3516typedef UBool U_CALLCONV UEnumCharNamesFn(void *context,
3517 UChar32 code,
3518 UCharNameChoice nameChoice,
3519 const char *name,
3520 int32_t length);
3521
3543U_STABLE void U_EXPORT2
3545 UEnumCharNamesFn *fn,
3546 void *context,
3547 UCharNameChoice nameChoice,
3548 UErrorCode *pErrorCode);
3549
3581U_STABLE const char* U_EXPORT2
3583 UPropertyNameChoice nameChoice);
3584
3604U_STABLE UProperty U_EXPORT2
3605u_getPropertyEnum(const char* alias);
3606
3654U_STABLE const char* U_EXPORT2
3656 int32_t value,
3657 UPropertyNameChoice nameChoice);
3658
3690U_STABLE int32_t U_EXPORT2
3692 const char* alias);
3693
3711U_STABLE UBool U_EXPORT2
3713
3735U_STABLE UBool U_EXPORT2
3737
3758U_STABLE UBool U_EXPORT2
3760
3777U_STABLE UBool U_EXPORT2
3779
3798U_STABLE UBool U_EXPORT2
3800
3823U_STABLE UChar32 U_EXPORT2
3825
3848U_STABLE UChar32 U_EXPORT2
3850
3873U_STABLE UChar32 U_EXPORT2
3875
3898U_STABLE UChar32 U_EXPORT2
3899u_foldCase(UChar32 c, uint32_t options);
3900
3939U_STABLE int32_t U_EXPORT2
3940u_digit(UChar32 ch, int8_t radix);
3941
3970U_STABLE UChar32 U_EXPORT2
3971u_forDigit(int32_t digit, int8_t radix);
3972
3987U_STABLE void U_EXPORT2
3989
4001U_STABLE void U_EXPORT2
4003
4004#if !UCONFIG_NO_NORMALIZATION
4026U_STABLE int32_t U_EXPORT2
4027u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode);
4028
4029#endif
4030
4031
4033
4034#endif /*_UCHAR*/
4035/*eof*/
#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
C API: Bit set option bit constants for various string and character processing functions.
UCharDirection u_charDirection(UChar32 c)
Returns the bidirectional category value for the code point, which is used in the Unicode bidirection...
UChar32 u_charFromName(UCharNameChoice nameChoice, const char *name, UErrorCode *pErrorCode)
Find a Unicode character by its name and return its code point value.
uint8_t u_getCombiningClass(UChar32 c)
Returns the combining class of the code point as specified in UnicodeData.txt.
UBool u_iscntrl(UChar32 c)
Determines whether the specified code point is a control character (as defined by this function).
UBlockCode
Constants for Unicode blocks, see the Unicode Data file Blocks.txt.
Definition uchar.h:1020
@ UBLOCK_NANDINAGARI
Definition uchar.h:1777
@ UBLOCK_MRO
Definition uchar.h:1648
@ UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED
Definition uchar.h:1511
@ UBLOCK_COUNT
One more than the highest normal UBlockCode value.
Definition uchar.h:1798
@ UBLOCK_OPTICAL_CHARACTER_RECOGNITION
Definition uchar.h:1184
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B
Definition uchar.h:1330
@ UBLOCK_INSCRIPTIONAL_PAHLAVI
Definition uchar.h:1545
@ UBLOCK_KHMER
Definition uchar.h:1139
@ UBLOCK_BHAIKSUKI
Definition uchar.h:1708
@ UBLOCK_DUPLOYAN
Definition uchar.h:1624
@ UBLOCK_MEDEFAIDRIN
Definition uchar.h:1764
@ UBLOCK_BALINESE
Definition uchar.h:1453
@ UBLOCK_HEBREW
Definition uchar.h:1064
@ UBLOCK_YIJING_HEXAGRAM_SYMBOLS
Definition uchar.h:1385
@ UBLOCK_SUPPLEMENTAL_SYMBOLS_AND_PICTOGRAPHS
Definition uchar.h:1699
@ UBLOCK_CYRILLIC
Definition uchar.h:1058
@ UBLOCK_SPECIALS
Definition uchar.h:1310
@ UBLOCK_HALFWIDTH_AND_FULLWIDTH_FORMS
Definition uchar.h:1313
@ UBLOCK_NEWA
Definition uchar.h:1720
@ UBLOCK_ARROWS
Definition uchar.h:1172
@ UBLOCK_CJK_SYMBOLS_AND_PUNCTUATION
Definition uchar.h:1217
@ UBLOCK_TAKRI
Definition uchar.h:1611
@ UBLOCK_BOX_DRAWING
Definition uchar.h:1190
@ UBLOCK_CURRENCY_SYMBOLS
Definition uchar.h:1157
@ UBLOCK_MISCELLANEOUS_SYMBOLS
Definition uchar.h:1199
@ UBLOCK_ANCIENT_GREEK_NUMBERS
Definition uchar.h:1410
@ UBLOCK_OLD_PERMIC
Definition uchar.h:1656
@ UBLOCK_CJK_STROKES
Definition uchar.h:1416
@ UBLOCK_PALMYRENE
Definition uchar.h:1662
@ UBLOCK_OGHAM
Definition uchar.h:1133
@ UBLOCK_SUTTON_SIGNWRITING
Definition uchar.h:1701
@ UBLOCK_ANATOLIAN_HIEROGLYPHS
Definition uchar.h:1685
@ UBLOCK_SUPPLEMENTAL_PUNCTUATION
Definition uchar.h:1440
@ UBLOCK_MEROITIC_HIEROGLYPHS
Definition uchar.h:1601
@ UBLOCK_EMOTICONS
Definition uchar.h:1580
@ UBLOCK_INDIC_SIYAQ_NUMBERS
Definition uchar.h:1758
@ UBLOCK_REJANG
Definition uchar.h:1488
@ UBLOCK_TAGALOG
Definition uchar.h:1346
@ UBLOCK_LISU
Definition uchar.h:1517
@ UBLOCK_COPTIC_EPACT_NUMBERS
Definition uchar.h:1620
@ UBLOCK_SUPERSCRIPTS_AND_SUBSCRIPTS
Definition uchar.h:1154
@ UBLOCK_TAGS
Definition uchar.h:1334
@ UBLOCK_MIAO
Definition uchar.h:1603
@ UBLOCK_TAMIL_SUPPLEMENT
Definition uchar.h:1787
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C
Definition uchar.h:1559
@ UBLOCK_LIMBU
Definition uchar.h:1375
@ UBLOCK_LYDIAN
Definition uchar.h:1500
@ UBLOCK_NYIAKENG_PUACHUE_HMONG
Definition uchar.h:1779
@ UBLOCK_PHONETIC_EXTENSIONS_SUPPLEMENT
Definition uchar.h:1438
@ UBLOCK_TAI_THAM
Definition uchar.h:1513
@ UBLOCK_MAHJONG_TILES
Definition uchar.h:1502
@ UBLOCK_PLAYING_CARDS
Definition uchar.h:1576
@ UBLOCK_TRANSPORT_AND_MAP_SYMBOLS
Definition uchar.h:1582
@ UBLOCK_IDEOGRAPHIC_DESCRIPTION_CHARACTERS
Definition uchar.h:1214
@ UBLOCK_BASIC_LATIN
Definition uchar.h:1031
@ UBLOCK_GURMUKHI
Definition uchar.h:1082
@ UBLOCK_SYMBOLS_AND_PICTOGRAPHS_EXTENDED_A
Definition uchar.h:1785
@ UBLOCK_HATRAN
Definition uchar.h:1693
@ UBLOCK_PRIVATE_USE_AREA
Same as UBLOCK_PRIVATE_USE.
Definition uchar.h:1276
@ UBLOCK_ARMENIAN
Definition uchar.h:1061
@ UBLOCK_HANGUL_JAMO
Definition uchar.h:1121
@ UBLOCK_VERTICAL_FORMS
Definition uchar.h:1446
@ UBLOCK_SINHALA_ARCHAIC_NUMBERS
Definition uchar.h:1672
@ UBLOCK_SOGDIAN
Definition uchar.h:1768
@ UBLOCK_TIRHUTA
Definition uchar.h:1676
@ UBLOCK_ELBASAN
Definition uchar.h:1626
@ UBLOCK_MATHEMATICAL_ALPHANUMERIC_SYMBOLS
Definition uchar.h:1328
@ UBLOCK_NUMBER_FORMS
Definition uchar.h:1169
@ UBLOCK_CHAM
Definition uchar.h:1490
@ UBLOCK_TANGUT
Definition uchar.h:1724
@ UBLOCK_TAMIL
Definition uchar.h:1091
@ UBLOCK_BLOCK_ELEMENTS
Definition uchar.h:1193
@ UBLOCK_ENCLOSED_ALPHANUMERIC_SUPPLEMENT
Definition uchar.h:1555
@ UBLOCK_CUNEIFORM_NUMBERS_AND_PUNCTUATION
Definition uchar.h:1465
@ UBLOCK_LEPCHA
Definition uchar.h:1474
@ UBLOCK_COMBINING_DIACRITICAL_MARKS_SUPPLEMENT
Definition uchar.h:1418
@ UBLOCK_VARIATION_SELECTORS_SUPPLEMENT
Definition uchar.h:1403
@ UBLOCK_DOMINO_TILES
Definition uchar.h:1504
@ UBLOCK_COMBINING_DIACRITICAL_MARKS_EXTENDED
Definition uchar.h:1622
@ UBLOCK_EGYPTIAN_HIEROGLYPHS
Definition uchar.h:1553
@ UBLOCK_IPA_EXTENSIONS
Definition uchar.h:1043
@ UBLOCK_TANGUT_COMPONENTS
Definition uchar.h:1726
@ UBLOCK_CHEROKEE_SUPPLEMENT
Definition uchar.h:1687
@ UBLOCK_ELYMAIC
Definition uchar.h:1775
@ UBLOCK_LATIN_EXTENDED_A
Definition uchar.h:1037
@ UBLOCK_HANUNOO
Definition uchar.h:1348
@ UBLOCK_LAO
Definition uchar.h:1109
@ UBLOCK_SUNDANESE_SUPPLEMENT
Definition uchar.h:1609
@ UBLOCK_CYRILLIC_EXTENDED_A
Definition uchar.h:1478
@ UBLOCK_HANGUL_JAMO_EXTENDED_B
Definition uchar.h:1535
@ UBLOCK_SAMARITAN
Definition uchar.h:1509
@ UBLOCK_SUPPLEMENTAL_ARROWS_A
Definition uchar.h:1356
@ UBLOCK_GLAGOLITIC
Definition uchar.h:1428
@ UBLOCK_INSCRIPTIONAL_PARTHIAN
Definition uchar.h:1543
@ UBLOCK_WARANG_CITI
Definition uchar.h:1678
@ UBLOCK_MONGOLIAN
Definition uchar.h:1142
@ UBLOCK_NKO
Definition uchar.h:1451
@ UBLOCK_ARABIC_EXTENDED_A
Definition uchar.h:1591
@ UBLOCK_LYCIAN
Definition uchar.h:1496
@ UBLOCK_CYRILLIC_EXTENDED_C
Definition uchar.h:1710
@ UBLOCK_DINGBATS
Definition uchar.h:1202
@ UBLOCK_OTTOMAN_SIYAQ_NUMBERS
Definition uchar.h:1781
@ UBLOCK_LINEAR_B_SYLLABARY
Definition uchar.h:1387
@ UBLOCK_ENCLOSED_CJK_LETTERS_AND_MONTHS
Definition uchar.h:1238
@ UBLOCK_CHAKMA
Definition uchar.h:1595
@ UBLOCK_SPACING_MODIFIER_LETTERS
Definition uchar.h:1046
@ UBLOCK_MEROITIC_CURSIVE
Definition uchar.h:1599
@ UBLOCK_CYRILLIC_EXTENDED_B
Definition uchar.h:1482
@ UBLOCK_GENERAL_PUNCTUATION
Definition uchar.h:1151
@ UBLOCK_MONGOLIAN_SUPPLEMENT
Definition uchar.h:1718
@ UBLOCK_MISCELLANEOUS_TECHNICAL
Definition uchar.h:1178
@ UBLOCK_TAI_XUAN_JING_SYMBOLS
Definition uchar.h:1401
@ UBLOCK_NABATAEAN
Definition uchar.h:1652
@ UBLOCK_TAGBANWA
Definition uchar.h:1352
@ UBLOCK_CUNEIFORM
Definition uchar.h:1463
@ UBLOCK_CONTROL_PICTURES
Definition uchar.h:1181
@ UBLOCK_GREEK
Unicode 3.2 renames this block to "Greek and Coptic".
Definition uchar.h:1055
@ UBLOCK_HIGH_SURROGATES
Definition uchar.h:1259
@ UBLOCK_COUNTING_ROD_NUMERALS
Definition uchar.h:1467
@ UBLOCK_AEGEAN_NUMBERS
Definition uchar.h:1391
@ UBLOCK_BENGALI
Definition uchar.h:1079
@ UBLOCK_LATIN_EXTENDED_E
Definition uchar.h:1636
@ UBLOCK_LINEAR_B_IDEOGRAMS
Definition uchar.h:1389
@ UBLOCK_PAU_CIN_HAU
Definition uchar.h:1664
@ UBLOCK_BUHID
Definition uchar.h:1350
@ UBLOCK_GUNJALA_GONDI
Definition uchar.h:1754
@ UBLOCK_RUMI_NUMERAL_SYMBOLS
Definition uchar.h:1549
@ UBLOCK_HIGH_PRIVATE_USE_SURROGATES
Definition uchar.h:1262
@ UBLOCK_MEETEI_MAYEK_EXTENSIONS
Definition uchar.h:1597
@ UBLOCK_COMMON_INDIC_NUMBER_FORMS
Definition uchar.h:1521
@ UBLOCK_ZANABAZAR_SQUARE
Definition uchar.h:1743
@ UBLOCK_SUNDANESE
Definition uchar.h:1472
@ UBLOCK_CHESS_SYMBOLS
Definition uchar.h:1748
@ UBLOCK_ENCLOSED_ALPHANUMERICS
Definition uchar.h:1187
@ UBLOCK_COMBINING_HALF_MARKS
Definition uchar.h:1298
@ UBLOCK_MANDAIC
Definition uchar.h:1564
@ UBLOCK_GLAGOLITIC_SUPPLEMENT
Definition uchar.h:1712
@ UBLOCK_OSMANYA
Definition uchar.h:1397
@ UBLOCK_DOGRA
Definition uchar.h:1750
@ UBLOCK_IMPERIAL_ARAMAIC
Definition uchar.h:1537
@ UBLOCK_BRAILLE_PATTERNS
Definition uchar.h:1205
@ UBLOCK_OLD_HUNGARIAN
Definition uchar.h:1697
@ UBLOCK_MATHEMATICAL_OPERATORS
Definition uchar.h:1175
@ UBLOCK_KHOJKI
Definition uchar.h:1632
@ UBLOCK_ORIYA
Definition uchar.h:1088
@ UBLOCK_NO_BLOCK
New No_Block value in Unicode 4.
Definition uchar.h:1028
@ UBLOCK_YI_RADICALS
Definition uchar.h:1253
@ UBLOCK_BUGINESE
Definition uchar.h:1414
@ UBLOCK_SMALL_FORM_VARIANTS
Definition uchar.h:1304
@ UBLOCK_OLD_SOGDIAN
Definition uchar.h:1766
@ UBLOCK_GEORGIAN_EXTENDED
Definition uchar.h:1752
@ UBLOCK_TIFINAGH
Definition uchar.h:1444
@ UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B
Definition uchar.h:1360
@ UBLOCK_LATIN_EXTENDED_D
Definition uchar.h:1457
@ UBLOCK_LATIN_EXTENDED_ADDITIONAL
Definition uchar.h:1145
@ UBLOCK_MEETEI_MAYEK
Definition uchar.h:1533
@ UBLOCK_EGYPTIAN_HIEROGLYPH_FORMAT_CONTROLS
Definition uchar.h:1773
@ UBLOCK_MYANMAR
Definition uchar.h:1115
@ UBLOCK_OL_CHIKI
Definition uchar.h:1476
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D
Definition uchar.h:1586
@ UBLOCK_SHORTHAND_FORMAT_CONTROLS
Definition uchar.h:1668
@ UBLOCK_COMBINING_MARKS_FOR_SYMBOLS
Unicode 3.2 renames this block to "Combining Diacritical Marks for Symbols".
Definition uchar.h:1163
@ UBLOCK_MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS
Definition uchar.h:1578
@ UBLOCK_UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS
Definition uchar.h:1130
@ UBLOCK_BOPOMOFO
Definition uchar.h:1226
@ UBLOCK_KATAKANA_PHONETIC_EXTENSIONS
Definition uchar.h:1364
@ UBLOCK_MARCHEN
Definition uchar.h:1716
@ UBLOCK_MODI
Definition uchar.h:1646
@ UBLOCK_SUPPLEMENTAL_MATHEMATICAL_OPERATORS
Definition uchar.h:1362
@ UBLOCK_MULTANI
Definition uchar.h:1695
@ UBLOCK_ETHIOPIC_EXTENDED
Definition uchar.h:1422
@ UBLOCK_AVESTAN
Definition uchar.h:1541
@ UBLOCK_PRIVATE_USE
Same as UBLOCK_PRIVATE_USE_AREA.
Definition uchar.h:1286
@ UBLOCK_GEORGIAN_SUPPLEMENT
Definition uchar.h:1426
@ UBLOCK_COPTIC
Definition uchar.h:1420
@ UBLOCK_HANGUL_COMPATIBILITY_JAMO
Definition uchar.h:1229
@ UBLOCK_LOW_SURROGATES
Definition uchar.h:1265
@ UBLOCK_SOYOMBO
Definition uchar.h:1739
@ UBLOCK_ARABIC_SUPPLEMENT
Definition uchar.h:1412
@ UBLOCK_HANGUL_SYLLABLES
Definition uchar.h:1256
@ UBLOCK_MISCELLANEOUS_SYMBOLS_AND_ARROWS
Definition uchar.h:1383
@ UBLOCK_CJK_COMPATIBILITY
Definition uchar.h:1241
@ UBLOCK_CARIAN
Definition uchar.h:1498
@ UBLOCK_HANIFI_ROHINGYA
Definition uchar.h:1756
@ UBLOCK_TIBETAN
Definition uchar.h:1112
@ UBLOCK_BASSA_VAH
Definition uchar.h:1616
@ UBLOCK_OLD_TURKIC
Definition uchar.h:1547
@ UBLOCK_CJK_RADICALS_SUPPLEMENT
Definition uchar.h:1208
@ UBLOCK_ADLAM
Definition uchar.h:1706
@ UBLOCK_MASARAM_GONDI
Definition uchar.h:1735
@ UBLOCK_ARABIC_PRESENTATION_FORMS_B
Definition uchar.h:1307
@ UBLOCK_RUNIC
Definition uchar.h:1136
@ UBLOCK_ARABIC_PRESENTATION_FORMS_A
Definition uchar.h:1295
@ UBLOCK_ENCLOSED_IDEOGRAPHIC_SUPPLEMENT
Definition uchar.h:1557
@ UBLOCK_BOPOMOFO_EXTENDED
Definition uchar.h:1235
@ UBLOCK_IDEOGRAPHIC_SYMBOLS_AND_PUNCTUATION
Definition uchar.h:1714
@ UBLOCK_KANGXI_RADICALS
Definition uchar.h:1211
@ UBLOCK_SORA_SOMPENG
Definition uchar.h:1607
@ UBLOCK_KANA_SUPPLEMENT
Definition uchar.h:1574
@ UBLOCK_CHEROKEE
Definition uchar.h:1127
@ UBLOCK_BRAHMI
Definition uchar.h:1570
@ UBLOCK_SYLOTI_NAGRI
Definition uchar.h:1442
@ UBLOCK_CYPRIOT_SYLLABARY
Definition uchar.h:1399
@ UBLOCK_ETHIOPIC_SUPPLEMENT
Definition uchar.h:1424
@ UBLOCK_KHAROSHTHI
Definition uchar.h:1430
@ UBLOCK_OLD_ITALIC
Definition uchar.h:1318
@ UBLOCK_OLD_SOUTH_ARABIAN
Definition uchar.h:1539
@ UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_B
Definition uchar.h:1370
@ UBLOCK_MALAYALAM
Definition uchar.h:1100
@ UBLOCK_KHMER_SYMBOLS
Definition uchar.h:1379
@ UBLOCK_JAVANESE
Definition uchar.h:1527
@ UBLOCK_BAMUM_SUPPLEMENT
Definition uchar.h:1572
@ UBLOCK_CYRILLIC_SUPPLEMENT
Definition uchar.h:1339
@ UBLOCK_MYANMAR_EXTENDED_B
Definition uchar.h:1650
@ UBLOCK_ANCIENT_GREEK_MUSICAL_NOTATION
Definition uchar.h:1408
@ UBLOCK_BATAK
Definition uchar.h:1566
@ UBLOCK_SYRIAC_SUPPLEMENT
Definition uchar.h:1741
@ UBLOCK_PAHAWH_HMONG
Definition uchar.h:1660
@ UBLOCK_ORNAMENTAL_DINGBATS
Definition uchar.h:1658
@ UBLOCK_SINHALA
Definition uchar.h:1103
@ UBLOCK_AHOM
Definition uchar.h:1683
@ UBLOCK_NEW_TAI_LUE
Definition uchar.h:1434
@ UBLOCK_SIDDHAM
Definition uchar.h:1670
@ UBLOCK_OLD_NORTH_ARABIAN
Definition uchar.h:1654
@ UBLOCK_YI_SYLLABLES
Definition uchar.h:1250
@ UBLOCK_TELUGU
Definition uchar.h:1094
@ UBLOCK_DEVANAGARI_EXTENDED
Definition uchar.h:1523
@ UBLOCK_SHARADA
Definition uchar.h:1605
@ UBLOCK_SUPPLEMENTARY_PRIVATE_USE_AREA_A
Definition uchar.h:1368
@ UBLOCK_DESERET
Definition uchar.h:1322
@ UBLOCK_ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS
Definition uchar.h:1593
@ UBLOCK_SAURASHTRA
Definition uchar.h:1484
@ UBLOCK_KANBUN
Definition uchar.h:1232
@ UBLOCK_HIRAGANA
Definition uchar.h:1220
@ UBLOCK_KANNADA
Definition uchar.h:1097
@ UBLOCK_SUPPLEMENTAL_ARROWS_C
Definition uchar.h:1674
@ UBLOCK_GEORGIAN
Definition uchar.h:1118
@ UBLOCK_DEVANAGARI
Definition uchar.h:1076
@ UBLOCK_KATAKANA
Definition uchar.h:1223
@ UBLOCK_TAI_VIET
Definition uchar.h:1531
@ UBLOCK_LATIN_EXTENDED_B
Definition uchar.h:1040
@ UBLOCK_THAANA
Definition uchar.h:1073
@ UBLOCK_MODIFIER_TONE_LETTERS
Definition uchar.h:1432
@ UBLOCK_KHUDAWADI
Definition uchar.h:1634
@ UBLOCK_LINEAR_A
Definition uchar.h:1638
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS
Definition uchar.h:1247
@ UBLOCK_GRANTHA
Definition uchar.h:1630
@ UBLOCK_ETHIOPIC
Definition uchar.h:1124
@ UBLOCK_SYRIAC
Definition uchar.h:1070
@ UBLOCK_GREEK_EXTENDED
Definition uchar.h:1148
@ UBLOCK_CJK_COMPATIBILITY_FORMS
Definition uchar.h:1301
@ UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS
Definition uchar.h:1289
@ UBLOCK_MAYAN_NUMERALS
Definition uchar.h:1762
@ UBLOCK_KAYAH_LI
Definition uchar.h:1486
@ UBLOCK_VEDIC_EXTENSIONS
Definition uchar.h:1515
@ UBLOCK_PHAISTOS_DISC
Definition uchar.h:1494
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A
Definition uchar.h:1244
@ UBLOCK_LATIN_1_SUPPLEMENT
Definition uchar.h:1034
@ UBLOCK_PHOENICIAN
Definition uchar.h:1461
@ UBLOCK_MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A
Definition uchar.h:1354
@ UBLOCK_CAUCASIAN_ALBANIAN
Definition uchar.h:1618
@ UBLOCK_ALCHEMICAL_SYMBOLS
Definition uchar.h:1584
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_F
Definition uchar.h:1731
@ UBLOCK_MYANMAR_EXTENDED_A
Definition uchar.h:1529
@ UBLOCK_EARLY_DYNASTIC_CUNEIFORM
Definition uchar.h:1691
@ UBLOCK_MENDE_KIKAKUI
Definition uchar.h:1644
@ UBLOCK_VAI
Definition uchar.h:1480
@ UBLOCK_PHONETIC_EXTENSIONS
Definition uchar.h:1381
@ UBLOCK_PHAGS_PA
Definition uchar.h:1459
@ UBLOCK_OLD_PERSIAN
Definition uchar.h:1436
@ UBLOCK_KAITHI
Definition uchar.h:1551
@ UBLOCK_GEOMETRIC_SHAPES
Definition uchar.h:1196
@ UBLOCK_LATIN_EXTENDED_C
Definition uchar.h:1455
@ UBLOCK_SUPPLEMENTAL_ARROWS_B
Definition uchar.h:1358
@ UBLOCK_MAHAJANI
Definition uchar.h:1640
@ UBLOCK_OSAGE
Definition uchar.h:1722
@ UBLOCK_INVALID_CODE
Definition uchar.h:1802
@ UBLOCK_TAI_LE
Definition uchar.h:1377
@ UBLOCK_MAKASAR
Definition uchar.h:1760
@ UBLOCK_GOTHIC
Definition uchar.h:1320
@ UBLOCK_ETHIOPIC_EXTENDED_A
Definition uchar.h:1568
@ UBLOCK_GEOMETRIC_SHAPES_EXTENDED
Definition uchar.h:1628
@ UBLOCK_MANICHAEAN
Definition uchar.h:1642
@ UBLOCK_ALPHABETIC_PRESENTATION_FORMS
Definition uchar.h:1292
@ UBLOCK_MUSICAL_SYMBOLS
Definition uchar.h:1326
@ UBLOCK_LETTERLIKE_SYMBOLS
Definition uchar.h:1166
@ UBLOCK_ARABIC
Definition uchar.h:1067
@ UBLOCK_NUSHU
Definition uchar.h:1737
@ UBLOCK_CYRILLIC_SUPPLEMENTARY
Unicode 4.0.1 renames the "Cyrillic Supplementary" block to "Cyrillic Supplement".
Definition uchar.h:1344
@ UBLOCK_HANGUL_JAMO_EXTENDED_A
Definition uchar.h:1525
@ UBLOCK_BYZANTINE_MUSICAL_SYMBOLS
Definition uchar.h:1324
@ UBLOCK_ANCIENT_SYMBOLS
Definition uchar.h:1492
@ UBLOCK_UGARITIC
Definition uchar.h:1393
@ UBLOCK_SMALL_KANA_EXTENSION
Definition uchar.h:1783
@ UBLOCK_GUJARATI
Definition uchar.h:1085
@ UBLOCK_CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT
Definition uchar.h:1332
@ UBLOCK_COMBINING_DIACRITICAL_MARKS
Definition uchar.h:1049
@ UBLOCK_PSALTER_PAHLAVI
Definition uchar.h:1666
@ UBLOCK_THAI
Definition uchar.h:1106
@ UBLOCK_CJK_UNIFIED_IDEOGRAPHS_EXTENSION_E
Definition uchar.h:1689
@ UBLOCK_KANA_EXTENDED_A
Definition uchar.h:1733
@ UBLOCK_VARIATION_SELECTORS
Definition uchar.h:1366
@ UBLOCK_BAMUM
Definition uchar.h:1519
@ UBLOCK_WANCHO
Definition uchar.h:1789
@ UBLOCK_SHAVIAN
Definition uchar.h:1395
void u_getUnicodeVersion(UVersionInfo versionArray)
Gets the Unicode version information.
UBool u_isdigit(UChar32 c)
Determines whether the specified code point is a digit character according to Java.
UBool u_isMirrored(UChar32 c)
Determines whether the code point has the Bidi_Mirrored property.
UBool u_hasBinaryProperty(UChar32 c, UProperty which)
Check a binary Unicode property for a code point.
UCharNameChoice
Selector constants for u_charName().
Definition uchar.h:1850
@ U_CHAR_NAME_CHOICE_COUNT
One more than the highest normal UCharNameChoice value.
Definition uchar.h:1870
@ U_UNICODE_10_CHAR_NAME
The Unicode_1_Name property value which is of little practical value.
Definition uchar.h:1859
@ U_CHAR_NAME_ALIAS
Corrected name from NameAliases.txt.
Definition uchar.h:1864
@ U_EXTENDED_CHAR_NAME
Standard or synthetic character name.
Definition uchar.h:1862
@ U_UNICODE_CHAR_NAME
Unicode character name (Name property).
Definition uchar.h:1852
int32_t u_getIntPropertyValue(UChar32 c, UProperty which)
Get the property value for an enumerated or integer Unicode property for a code point.
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.
double u_getNumericValue(UChar32 c)
Get the numeric value for a Unicode code point as defined in the Unicode Character Database.
int32_t u_digit(UChar32 ch, int8_t radix)
Returns the decimal digit value of the code point in the specified radix.
UJoiningType
Joining Type constants.
Definition uchar.h:1947
@ U_JT_COUNT
One more than the highest normal UJoiningType value.
Definition uchar.h:1967
UBool u_isUAlphabetic(UChar32 c)
Check if a code point has the Alphabetic Unicode property.
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.
void u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context)
Enumerate efficiently all code points with their Unicode general categories.
UBool u_ispunct(UChar32 c)
Determines whether the specified code point is a punctuation character.
UIndicPositionalCategory
Indic Positional Category constants.
Definition uchar.h:2402
@ U_INPC_BOTTOM
Definition uchar.h:2412
@ U_INPC_VISUAL_ORDER_LEFT
Definition uchar.h:2438
@ U_INPC_TOP_AND_BOTTOM_AND_RIGHT
Definition uchar.h:2430
@ U_INPC_TOP_AND_RIGHT
Definition uchar.h:2436
@ U_INPC_BOTTOM_AND_LEFT
Definition uchar.h:2414
@ U_INPC_LEFT_AND_RIGHT
Definition uchar.h:2420
@ U_INPC_RIGHT
Definition uchar.h:2424
@ U_INPC_TOP_AND_LEFT_AND_RIGHT
Definition uchar.h:2434
@ U_INPC_LEFT
Definition uchar.h:2418
@ U_INPC_NA
Definition uchar.h:2410
@ U_INPC_TOP
Definition uchar.h:2426
@ U_INPC_TOP_AND_LEFT
Definition uchar.h:2432
@ U_INPC_OVERSTRUCK
Definition uchar.h:2422
@ U_INPC_TOP_AND_BOTTOM
Definition uchar.h:2428
@ U_INPC_BOTTOM_AND_RIGHT
Definition uchar.h:2416
UBlockCode ublock_getCode(UChar32 c)
Returns the Unicode allocation block that contains the character.
UCharDirection
This specifies the language directional property of a character set.
Definition uchar.h:921
@ U_SEGMENT_SEPARATOR
S.
Definition uchar.h:946
@ U_EUROPEAN_NUMBER_TERMINATOR
ET.
Definition uchar.h:938
@ U_BOUNDARY_NEUTRAL
BN.
Definition uchar.h:966
@ U_RIGHT_TO_LEFT_ARABIC
AL.
Definition uchar.h:956
@ U_RIGHT_TO_LEFT
R.
Definition uchar.h:932
@ U_POP_DIRECTIONAL_ISOLATE
PDI.
Definition uchar.h:974
@ U_COMMON_NUMBER_SEPARATOR
CS.
Definition uchar.h:942
@ U_LEFT_TO_RIGHT
L.
Definition uchar.h:930
@ U_DIR_NON_SPACING_MARK
NSM.
Definition uchar.h:964
@ U_FIRST_STRONG_ISOLATE
FSI.
Definition uchar.h:968
@ U_ARABIC_NUMBER
AN.
Definition uchar.h:940
@ U_POP_DIRECTIONAL_FORMAT
PDF.
Definition uchar.h:962
@ U_OTHER_NEUTRAL
ON.
Definition uchar.h:950
@ U_CHAR_DIRECTION_COUNT
One more than the highest UCharDirection value.
Definition uchar.h:982
@ U_WHITE_SPACE_NEUTRAL
WS.
Definition uchar.h:948
@ U_RIGHT_TO_LEFT_OVERRIDE
RLO.
Definition uchar.h:960
@ U_RIGHT_TO_LEFT_EMBEDDING
RLE.
Definition uchar.h:958
@ U_EUROPEAN_NUMBER_SEPARATOR
ES.
Definition uchar.h:936
@ U_LEFT_TO_RIGHT_ISOLATE
LRI.
Definition uchar.h:970
@ U_LEFT_TO_RIGHT_OVERRIDE
LRO.
Definition uchar.h:954
@ U_EUROPEAN_NUMBER
EN.
Definition uchar.h:934
@ U_BLOCK_SEPARATOR
B.
Definition uchar.h:944
@ U_LEFT_TO_RIGHT_EMBEDDING
LRE.
Definition uchar.h:952
@ U_RIGHT_TO_LEFT_ISOLATE
RLI.
Definition uchar.h:972
UPropertyNameChoice
Selector constants for u_getPropertyName() and u_getPropertyValueName().
Definition uchar.h:1887
@ U_PROPERTY_NAME_CHOICE_COUNT
One more than the highest normal UPropertyNameChoice value.
Definition uchar.h:1895
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:3516
UBool u_isgraph(UChar32 c)
Determines whether the specified code point is a "graphic" character (printable, excluding spaces).
UBool u_isbase(UChar32 c)
Non-standard: Determines whether the specified code point is a base character.
UChar32 u_tolower(UChar32 c)
The given character is mapped to its lowercase equivalent according to UnicodeData....
const char * u_getPropertyName(UProperty property, UPropertyNameChoice nameChoice)
Return the Unicode name for a given property, as given in the Unicode database file PropertyAliases....
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,...
UBool u_isUUppercase(UChar32 c)
Check if a code point has the Uppercase Unicode property.
UBool u_isWhitespace(UChar32 c)
Determines if the specified code point is a whitespace character according to Java/ICU.
ULineBreak
Line Break constants.
Definition uchar.h:2261
@ U_LB_E_MODIFIER
Definition uchar.h:2324
@ U_LB_WORD_JOINER
Definition uchar.h:2302
@ U_LB_H3
Definition uchar.h:2306
@ U_LB_H2
Definition uchar.h:2304
@ U_LB_CONDITIONAL_JAPANESE_STARTER
Definition uchar.h:2316
@ U_LB_NEXT_LINE
Definition uchar.h:2300
@ U_LB_INSEPARABLE
Renamed from the misspelled "inseperable" in Unicode 4.0.1/ICU 3.0.
Definition uchar.h:2284
@ U_LB_E_BASE
Definition uchar.h:2322
@ U_LB_JV
Definition uchar.h:2312
@ U_LB_JL
Definition uchar.h:2308
@ U_LB_CLOSE_PARENTHESIS
Definition uchar.h:2314
@ U_LB_ZWJ
Definition uchar.h:2326
@ U_LB_COUNT
One more than the highest normal ULineBreak value.
Definition uchar.h:2334
@ U_LB_REGIONAL_INDICATOR
Definition uchar.h:2320
@ U_LB_HEBREW_LETTER
Definition uchar.h:2318
@ U_LB_JT
Definition uchar.h:2310
UBool u_isalnum(UChar32 c)
Determines whether the specified code point is an alphanumeric character (letter or digit) according ...
int32_t u_getISOComment(UChar32 c, char *dest, int32_t destCapacity, UErrorCode *pErrorCode)
Returns an empty string.
UBool u_isJavaSpaceChar(UChar32 c)
Determine if the specified code point is a space character according to Java.
UBool u_isdefined(UChar32 c)
Determines whether the specified code point is "defined", which usually means that it is assigned a c...
UChar32 u_charMirror(UChar32 c)
Maps the specified character to a "mirror-image" character.
UCharCategory
Data for enumerated Unicode general category types.
Definition uchar.h:725
@ U_FORMAT_CHAR
Cf.
Definition uchar.h:768
@ U_SPACE_SEPARATOR
Zs.
Definition uchar.h:760
@ U_MODIFIER_SYMBOL
Sk.
Definition uchar.h:788
@ U_GENERAL_OTHER_TYPES
Cn "Other, Not Assigned (no characters in [UnicodeData.txt] have this property)" (same as U_UNASSIGNE...
Definition uchar.h:736
@ U_TITLECASE_LETTER
Lt.
Definition uchar.h:742
@ U_PRIVATE_USE_CHAR
Co.
Definition uchar.h:770
@ U_OTHER_SYMBOL
So.
Definition uchar.h:790
@ U_UPPERCASE_LETTER
Lu.
Definition uchar.h:738
@ U_MODIFIER_LETTER
Lm.
Definition uchar.h:744
@ U_PARAGRAPH_SEPARATOR
Zp.
Definition uchar.h:764
@ U_OTHER_PUNCTUATION
Po.
Definition uchar.h:782
@ U_ENCLOSING_MARK
Me.
Definition uchar.h:750
@ U_INITIAL_PUNCTUATION
Pi.
Definition uchar.h:792
@ U_UNASSIGNED
Non-category for unassigned and non-character code points.
Definition uchar.h:734
@ U_CURRENCY_SYMBOL
Sc.
Definition uchar.h:786
@ U_COMBINING_SPACING_MARK
Mc.
Definition uchar.h:752
@ U_CONTROL_CHAR
Cc.
Definition uchar.h:766
@ U_OTHER_LETTER
Lo.
Definition uchar.h:746
@ U_NON_SPACING_MARK
Mn.
Definition uchar.h:748
@ U_START_PUNCTUATION
Ps.
Definition uchar.h:776
@ U_CONNECTOR_PUNCTUATION
Pc.
Definition uchar.h:780
@ U_END_PUNCTUATION
Pe.
Definition uchar.h:778
@ U_OTHER_NUMBER
No.
Definition uchar.h:758
@ U_LINE_SEPARATOR
Zl.
Definition uchar.h:762
@ U_DASH_PUNCTUATION
Pd.
Definition uchar.h:774
@ U_MATH_SYMBOL
Sm.
Definition uchar.h:784
@ U_CHAR_CATEGORY_COUNT
One higher than the last enum UCharCategory constant.
Definition uchar.h:802
@ U_DECIMAL_DIGIT_NUMBER
Nd.
Definition uchar.h:754
@ U_LOWERCASE_LETTER
Ll.
Definition uchar.h:740
@ U_FINAL_PUNCTUATION
Pf.
Definition uchar.h:794
@ U_LETTER_NUMBER
Nl.
Definition uchar.h:756
@ U_SURROGATE
Cs.
Definition uchar.h:772
UVerticalOrientation
Vertical Orientation constants.
Definition uchar.h:2534
@ U_VO_UPRIGHT
Definition uchar.h:2548
@ U_VO_TRANSFORMED_UPRIGHT
Definition uchar.h:2546
@ U_VO_ROTATED
Definition uchar.h:2542
@ U_VO_TRANSFORMED_ROTATED
Definition uchar.h:2544
UBool u_isIDIgnorable(UChar32 c)
Determines if the specified character should be regarded as an ignorable character in an identifier,...
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...
UJoiningGroup
Joining Group constants.
Definition uchar.h:1977
@ U_JG_FE
Definition uchar.h:2036
@ U_JG_BURUSHASKI_YEH_BARREE
Definition uchar.h:2039
@ U_JG_MANICHAEAN_KAPH
Definition uchar.h:2052
@ U_JG_MANICHAEAN_SAMEKH
Definition uchar.h:2061
@ U_JG_MANICHAEAN_ZAYIN
Definition uchar.h:2069
@ U_JG_MALAYALAM_TTA
Definition uchar.h:2085
@ U_JG_MANICHAEAN_AYIN
Definition uchar.h:2044
@ U_JG_MANICHAEAN_ALEPH
Definition uchar.h:2043
@ U_JG_FARSI_YEH
Definition uchar.h:2040
@ U_JG_MANICHAEAN_DALETH
Definition uchar.h:2046
@ U_JG_MANICHAEAN_FIVE
Definition uchar.h:2048
@ U_JG_MANICHAEAN_SADHE
Definition uchar.h:2060
@ U_JG_MANICHAEAN_TAW
Definition uchar.h:2062
@ U_JG_MALAYALAM_LLLA
Definition uchar.h:2078
@ U_JG_MALAYALAM_NNNA
Definition uchar.h:2081
@ U_JG_MANICHAEAN_ONE
Definition uchar.h:2056
@ U_JG_MALAYALAM_LLA
Definition uchar.h:2077
@ U_JG_AFRICAN_FEH
Definition uchar.h:2071
@ U_JG_HANIFI_ROHINGYA_PA
Definition uchar.h:2088
@ U_JG_MALAYALAM_BHA
Definition uchar.h:2075
@ U_JG_MANICHAEAN_TWENTY
Definition uchar.h:2066
@ U_JG_MANICHAEAN_THAMEDH
Definition uchar.h:2065
@ U_JG_MANICHAEAN_PE
Definition uchar.h:2057
@ U_JG_ZHAIN
Definition uchar.h:2038
@ U_JG_MANICHAEAN_TETH
Definition uchar.h:2064
@ U_JG_MANICHAEAN_BETH
Definition uchar.h:2045
@ U_JG_COUNT
One more than the highest normal UJoiningGroup value.
Definition uchar.h:2097
@ U_JG_MANICHAEAN_TEN
Definition uchar.h:2063
@ U_JG_MANICHAEAN_GIMEL
Definition uchar.h:2049
@ U_JG_TEH_MARBUTA_GOAL
Definition uchar.h:1998
@ U_JG_AFRICAN_NOON
Definition uchar.h:2072
@ U_JG_MALAYALAM_NGA
Definition uchar.h:2079
@ U_JG_MANICHAEAN_MEM
Definition uchar.h:2054
@ U_JG_KHAPH
Definition uchar.h:2037
@ U_JG_MALAYALAM_RA
Definition uchar.h:2083
@ U_JG_MALAYALAM_NNA
Definition uchar.h:2080
@ U_JG_MALAYALAM_JA
Definition uchar.h:2076
@ U_JG_MALAYALAM_NYA
Definition uchar.h:2082
@ U_JG_MALAYALAM_SSA
Definition uchar.h:2084
@ U_JG_MANICHAEAN_WAW
Definition uchar.h:2067
@ U_JG_AFRICAN_QAF
Definition uchar.h:2073
@ U_JG_MANICHAEAN_DHAMEDH
Definition uchar.h:2047
@ U_JG_STRAIGHT_WAW
Definition uchar.h:2070
@ U_JG_ROHINGYA_YEH
Definition uchar.h:2042
@ U_JG_MANICHAEAN_HETH
Definition uchar.h:2050
@ U_JG_MANICHAEAN_LAMEDH
Definition uchar.h:2053
@ U_JG_MANICHAEAN_HUNDRED
Definition uchar.h:2051
@ U_JG_HANIFI_ROHINGYA_KINNA_YA
Definition uchar.h:2087
@ U_JG_MANICHAEAN_NUN
Definition uchar.h:2055
@ U_JG_NYA
Definition uchar.h:2041
@ U_JG_MANICHAEAN_YODH
Definition uchar.h:2068
@ U_JG_MANICHAEAN_RESH
Definition uchar.h:2059
@ U_JG_MANICHAEAN_QOPH
Definition uchar.h:2058
UBool u_isIDPart(UChar32 c)
Determines if the specified character is permissible in an identifier according to Java.
UHangulSyllableType
Hangul Syllable Type constants.
Definition uchar.h:2372
@ U_HST_COUNT
One more than the highest normal UHangulSyllableType value.
Definition uchar.h:2392
UBool u_isULowercase(UChar32 c)
Check if a code point has the Lowercase Unicode property.
UChar32 u_toupper(UChar32 c)
The given character is mapped to its uppercase equivalent according to UnicodeData....
UBool u_isalpha(UChar32 c)
Determines whether the specified code point is a letter character.
int32_t u_getIntPropertyMaxValue(UProperty which)
Get the maximum value for an enumerated/integer/binary Unicode property.
UProperty u_getPropertyEnum(const char *alias)
Return the UProperty enum for a given property name, as specified in the Unicode database file Proper...
USentenceBreak
Sentence Break constants.
Definition uchar.h:2222
@ U_SB_COUNT
One more than the highest normal USentenceBreak value.
Definition uchar.h:2251
UBool u_isblank(UChar32 c)
Determines whether the specified code point is a "blank" or "horizontal space", a character that visi...
UBool u_isJavaIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in a Java identifier.
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:3336
UChar32 u_foldCase(UChar32 c, uint32_t options)
The given character is mapped to its case folding equivalent according to UnicodeData....
UBool u_isISOControl(UChar32 c)
Determines whether the specified code point is an ISO control code.
UEastAsianWidth
East Asian Width constants.
Definition uchar.h:1815
@ U_EA_COUNT
One more than the highest normal UEastAsianWidth value.
Definition uchar.h:1835
UBool u_isspace(UChar32 c)
Determines if the specified character is a space character or not.
U_CAPI const UCPMap * u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode)
Returns an immutable UCPMap for an enumerated/catalog/int-valued property.
UBool u_islower(UChar32 c)
Determines whether the specified code point has the general category "Ll" (lowercase letter).
int32_t u_charName(UChar32 code, UCharNameChoice nameChoice, char *buffer, int32_t bufferLength, UErrorCode *pErrorCode)
Retrieve the name of a Unicode character.
UBool u_isIDStart(UChar32 c)
Determines if the specified character is permissible as the first character in an identifier accordin...
UBool u_isJavaIDPart(UChar32 c)
Determines if the specified character is permissible in a Java identifier.
UChar32 u_forDigit(int32_t digit, int8_t radix)
Determines the character representation for a specific digit in the specified radix.
UBool u_isxdigit(UChar32 c)
Determines whether the specified code point is a hexadecimal digit.
int32_t u_getIntPropertyMinValue(UProperty which)
Get the minimum value for an enumerated/integer/binary Unicode property.
int8_t u_charType(UChar32 c)
Returns the general category value for the code point.
UGraphemeClusterBreak
Grapheme Cluster Break constants.
Definition uchar.h:2107
@ U_GCB_SPACING_MARK
Definition uchar.h:2125
@ U_GCB_E_BASE
Definition uchar.h:2131
@ U_GCB_REGIONAL_INDICATOR
Definition uchar.h:2129
@ U_GCB_GLUE_AFTER_ZWJ
Definition uchar.h:2137
@ U_GCB_ZWJ
Definition uchar.h:2139
@ U_GCB_PREPEND
Definition uchar.h:2127
@ U_GCB_COUNT
One more than the highest normal UGraphemeClusterBreak value.
Definition uchar.h:2148
@ U_GCB_E_MODIFIER
Definition uchar.h:2135
@ U_GCB_E_BASE_GAZ
Definition uchar.h:2133
U_CAPI const USet * u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode)
Returns a frozen USet for a binary property.
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...
UBool u_isupper(UChar32 c)
Determines whether the specified code point has the general category "Lu" (uppercase letter).
void u_charAge(UChar32 c, UVersionInfo versionArray)
Get the "age" of the code point.
UChar32 u_totitle(UChar32 c)
The given character is mapped to its titlecase equivalent according to UnicodeData....
UChar32 u_getBidiPairedBracket(UChar32 c)
Maps the specified character to its paired bracket character.
UNumericType
Numeric Type constants.
Definition uchar.h:2344
@ U_NT_COUNT
One more than the highest normal UNumericType value.
Definition uchar.h:2362
UDecompositionType
Decomposition Type constants.
Definition uchar.h:1905
@ U_DT_COUNT
One more than the highest normal UDecompositionType value.
Definition uchar.h:1937
UProperty
Selection constants for Unicode properties.
Definition uchar.h:195
@ UCHAR_GRAPHEME_CLUSTER_BREAK
Enumerated property Grapheme_Cluster_Break (new in Unicode 4.1).
Definition uchar.h:563
@ UCHAR_BIDI_PAIRED_BRACKET
String property Bidi_Paired_Bracket (new in Unicode 6.3).
Definition uchar.h:690
@ UCHAR_ALPHABETIC
Binary property Alphabetic.
Definition uchar.h:209
@ UCHAR_POSIX_GRAPH
Binary property graph (a C/POSIX character class).
Definition uchar.h:407
@ UCHAR_BLOCK
Enumerated property Block.
Definition uchar.h:501
@ UCHAR_INVALID_CODE
Represents a nonexistent or invalid property or property value.
Definition uchar.h:716
@ UCHAR_SEGMENT_STARTER
Binary Property Segment_Starter.
Definition uchar.h:382
@ UCHAR_PREPENDED_CONCATENATION_MARK
Binary property Prepended_Concatenation_Mark.
Definition uchar.h:478
@ UCHAR_REGIONAL_INDICATOR
Binary property Regional_Indicator.
Definition uchar.h:473
@ UCHAR_OTHER_PROPERTY_START
First constant for Unicode properties with unusual value types.
Definition uchar.h:706
@ UCHAR_S_TERM
Binary property STerm (new in Unicode 4.0.1).
Definition uchar.h:337
@ UCHAR_WHITE_SPACE
Binary property White_Space.
Definition uchar.h:321
@ UCHAR_CANONICAL_COMBINING_CLASS
Enumerated property Canonical_Combining_Class.
Definition uchar.h:504
@ UCHAR_SOFT_DOTTED
Binary property Soft_Dotted (new in Unicode 3.2).
Definition uchar.h:306
@ UCHAR_GRAPHEME_LINK
Binary property Grapheme_Link (new in Unicode 3.2).
Definition uchar.h:253
@ UCHAR_PATTERN_SYNTAX
Binary property Pattern_Syntax (new in Unicode 4.1).
Definition uchar.h:387
@ UCHAR_GRAPHEME_EXTEND
Binary property Grapheme_Extend (new in Unicode 3.2).
Definition uchar.h:250
@ UCHAR_CASED
Binary property Cased.
Definition uchar.h:419
@ UCHAR_XID_START
Binary property XID_Start.
Definition uchar.h:328
@ UCHAR_VERTICAL_ORIENTATION
Enumerated property Vertical_Orientation.
Definition uchar.h:600
@ UCHAR_NFKC_INERT
Binary property NFKC_Inert.
Definition uchar.h:371
@ UCHAR_INT_LIMIT
One more than the last constant for enumerated/integer Unicode properties.
Definition uchar.h:606
@ UCHAR_OTHER_PROPERTY_LIMIT
One more than the last constant for Unicode properties with unusual value types.
Definition uchar.h:712
@ UCHAR_PATTERN_WHITE_SPACE
Binary property Pattern_White_Space (new in Unicode 4.1).
Definition uchar.h:392
@ UCHAR_VARIATION_SELECTOR
Binary property Variation_Selector (new in Unicode 4.0.1).
Definition uchar.h:343
@ UCHAR_NUMERIC_VALUE
Double property Numeric_Value.
Definition uchar.h:630
@ UCHAR_DOUBLE_START
First constant for double Unicode properties.
Definition uchar.h:632
@ UCHAR_HEX_DIGIT
Binary property Hex_Digit.
Definition uchar.h:256
@ UCHAR_SIMPLE_TITLECASE_MAPPING
String property Simple_Titlecase_Mapping.
Definition uchar.h:671
@ UCHAR_TRAIL_CANONICAL_COMBINING_CLASS
Enumerated property Trail_Canonical_Combining_Class.
Definition uchar.h:558
@ UCHAR_DIACRITIC
Binary property Diacritic.
Definition uchar.h:234
@ UCHAR_XID_CONTINUE
Binary property XID_Continue.
Definition uchar.h:325
@ UCHAR_NFKC_QUICK_CHECK
Enumerated property NFKC_Quick_Check.
Definition uchar.h:544
@ UCHAR_HYPHEN
Binary property Hyphen.
Definition uchar.h:259
@ UCHAR_RADICAL
Binary property Radical (new in Unicode 3.2).
Definition uchar.h:301
@ UCHAR_BIDI_PAIRED_BRACKET_TYPE
Enumerated property Bidi_Paired_Bracket_Type (new in Unicode 6.3).
Definition uchar.h:578
@ UCHAR_DASH
Binary property Dash.
Definition uchar.h:224
@ UCHAR_ASCII_HEX_DIGIT
Binary property ASCII_Hex_Digit.
Definition uchar.h:213
@ UCHAR_ID_CONTINUE
Binary property ID_Continue.
Definition uchar.h:264
@ UCHAR_NFKD_INERT
Binary property NFKD_Inert.
Definition uchar.h:357
@ UCHAR_BIDI_MIRRORING_GLYPH
String property Bidi_Mirroring_Glyph.
Definition uchar.h:648
@ UCHAR_LEAD_CANONICAL_COMBINING_CLASS
Enumerated property Lead_Canonical_Combining_Class.
Definition uchar.h:551
@ UCHAR_UNICODE_1_NAME
String property Unicode_1_Name.
Definition uchar.h:683
@ UCHAR_BIDI_MIRRORED
Binary property Bidi_Mirrored.
Definition uchar.h:222
@ UCHAR_CASE_IGNORABLE
Binary property Case_Ignorable.
Definition uchar.h:421
@ UCHAR_EAST_ASIAN_WIDTH
Enumerated property East_Asian_Width.
Definition uchar.h:511
@ UCHAR_STRING_LIMIT
One more than the last constant for string Unicode properties.
Definition uchar.h:696
@ UCHAR_EMOJI_PRESENTATION
Binary property Emoji_Presentation.
Definition uchar.h:447
@ UCHAR_GRAPHEME_BASE
Binary property Grapheme_Base (new in Unicode 3.2).
Definition uchar.h:246
@ UCHAR_NFKD_QUICK_CHECK
Enumerated property NFKD_Quick_Check.
Definition uchar.h:538
@ UCHAR_NAME
String property Name.
Definition uchar.h:662
@ UCHAR_UPPERCASE
Binary property Uppercase.
Definition uchar.h:317
@ UCHAR_CASE_FOLDING
String property Case_Folding.
Definition uchar.h:651
@ UCHAR_INDIC_SYLLABIC_CATEGORY
Enumerated property Indic_Syllabic_Category.
Definition uchar.h:593
@ UCHAR_MATH
Binary property Math.
Definition uchar.h:291
@ UCHAR_NUMERIC_TYPE
Enumerated property Numeric_Type.
Definition uchar.h:526
@ UCHAR_ISO_COMMENT
Deprecated string property ISO_Comment.
Definition uchar.h:655
@ UCHAR_SCRIPT
Enumerated property Script.
Definition uchar.h:529
@ UCHAR_CHANGES_WHEN_TITLECASED
Binary property Changes_When_Titlecased.
Definition uchar.h:427
@ UCHAR_EXTENDED_PICTOGRAPHIC
Binary property Extended_Pictographic.
Definition uchar.h:485
@ UCHAR_DEPRECATED
Binary property Deprecated (new in Unicode 3.2).
Definition uchar.h:231
@ UCHAR_CHANGES_WHEN_UPPERCASED
Binary property Changes_When_Uppercased.
Definition uchar.h:425
@ UCHAR_CHANGES_WHEN_CASEFOLDED
Binary property Changes_When_Casefolded.
Definition uchar.h:429
@ UCHAR_INDIC_POSITIONAL_CATEGORY
Enumerated property Indic_Positional_Category.
Definition uchar.h:586
@ UCHAR_DEFAULT_IGNORABLE_CODE_POINT
Binary property Default_Ignorable_Code_Point (new in Unicode 3.2).
Definition uchar.h:228
@ UCHAR_GENERAL_CATEGORY
Enumerated property General_Category.
Definition uchar.h:514
@ UCHAR_INT_START
First constant for enumerated/integer Unicode properties.
Definition uchar.h:498
@ UCHAR_EMOJI_MODIFIER
Binary property Emoji_Modifier.
Definition uchar.h:454
@ UCHAR_NFD_QUICK_CHECK
Enumerated property NFD_Quick_Check.
Definition uchar.h:535
@ UCHAR_IDS_BINARY_OPERATOR
Binary property IDS_Binary_Operator (new in Unicode 3.2).
Definition uchar.h:275
@ UCHAR_BINARY_START
First constant for binary Unicode properties.
Definition uchar.h:211
@ UCHAR_TERMINAL_PUNCTUATION
Binary property Terminal_Punctuation.
Definition uchar.h:310
@ UCHAR_GENERAL_CATEGORY_MASK
Bitmask property General_Category_Mask.
Definition uchar.h:617
@ UCHAR_MASK_START
First constant for bit-mask Unicode properties.
Definition uchar.h:619
@ UCHAR_DECOMPOSITION_TYPE
Enumerated property Decomposition_Type.
Definition uchar.h:507
@ UCHAR_TITLECASE_MAPPING
String property Titlecase_Mapping.
Definition uchar.h:677
@ UCHAR_HANGUL_SYLLABLE_TYPE
Enumerated property Hangul_Syllable_Type, new in Unicode 4.
Definition uchar.h:532
@ UCHAR_LINE_BREAK
Enumerated property Line_Break.
Definition uchar.h:523
@ UCHAR_SIMPLE_UPPERCASE_MAPPING
String property Simple_Uppercase_Mapping.
Definition uchar.h:674
@ UCHAR_POSIX_ALNUM
Binary property alnum (a C/POSIX character class).
Definition uchar.h:397
@ UCHAR_JOINING_TYPE
Enumerated property Joining_Type.
Definition uchar.h:520
@ UCHAR_QUOTATION_MARK
Binary property Quotation_Mark.
Definition uchar.h:297
@ UCHAR_NFC_INERT
Binary property NFC_Inert.
Definition uchar.h:364
@ UCHAR_LOWERCASE_MAPPING
String property Lowercase_Mapping.
Definition uchar.h:659
@ UCHAR_SIMPLE_CASE_FOLDING
String property Simple_Case_Folding.
Definition uchar.h:665
@ UCHAR_JOIN_CONTROL
Binary property Join_Control.
Definition uchar.h:282
@ UCHAR_NONCHARACTER_CODE_POINT
Binary property Noncharacter_Code_Point.
Definition uchar.h:295
@ UCHAR_BIDI_CONTROL
Binary property Bidi_Control.
Definition uchar.h:217
@ UCHAR_CHANGES_WHEN_LOWERCASED
Binary property Changes_When_Lowercased.
Definition uchar.h:423
@ UCHAR_BINARY_LIMIT
One more than the last constant for binary Unicode properties.
Definition uchar.h:491
@ UCHAR_IDS_TRINARY_OPERATOR
Binary property IDS_Trinary_Operator (new in Unicode 3.2).
Definition uchar.h:279
@ UCHAR_ID_START
Binary property ID_Start.
Definition uchar.h:268
@ UCHAR_AGE
String property Age.
Definition uchar.h:643
@ UCHAR_WORD_BREAK
Enumerated property Word_Break (new in Unicode 4.1).
Definition uchar.h:573
@ UCHAR_DOUBLE_LIMIT
One more than the last constant for double Unicode properties.
Definition uchar.h:638
@ UCHAR_EMOJI_MODIFIER_BASE
Binary property Emoji_Modifier_Base.
Definition uchar.h:461
@ UCHAR_EMOJI_COMPONENT
Binary property Emoji_Component.
Definition uchar.h:468
@ UCHAR_POSIX_BLANK
Binary property blank (a C/POSIX character class).
Definition uchar.h:402
@ UCHAR_SIMPLE_LOWERCASE_MAPPING
String property Simple_Lowercase_Mapping.
Definition uchar.h:668
@ UCHAR_NFD_INERT
Binary property NFD_Inert.
Definition uchar.h:350
@ UCHAR_POSIX_PRINT
Binary property print (a C/POSIX character class).
Definition uchar.h:412
@ UCHAR_SENTENCE_BREAK
Enumerated property Sentence_Break (new in Unicode 4.1).
Definition uchar.h:568
@ UCHAR_IDEOGRAPHIC
Binary property Ideographic.
Definition uchar.h:271
@ UCHAR_UNIFIED_IDEOGRAPH
Binary property Unified_Ideograph (new in Unicode 3.2).
Definition uchar.h:314
@ UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED
Binary property Changes_When_NFKC_Casefolded.
Definition uchar.h:433
@ UCHAR_NFC_QUICK_CHECK
Enumerated property NFC_Quick_Check.
Definition uchar.h:541
@ UCHAR_CASE_SENSITIVE
Binary property Case_Sensitive.
Definition uchar.h:332
@ UCHAR_UPPERCASE_MAPPING
String property Uppercase_Mapping.
Definition uchar.h:687
@ UCHAR_BIDI_CLASS
Enumerated property Bidi_Class.
Definition uchar.h:496
@ UCHAR_MASK_LIMIT
One more than the last constant for bit-mask Unicode properties.
Definition uchar.h:625
@ UCHAR_JOINING_GROUP
Enumerated property Joining_Group.
Definition uchar.h:517
@ UCHAR_LOGICAL_ORDER_EXCEPTION
Binary property Logical_Order_Exception (new in Unicode 3.2).
Definition uchar.h:286
@ UCHAR_EXTENDER
Binary property Extender.
Definition uchar.h:238
@ UCHAR_STRING_START
First constant for string Unicode properties.
Definition uchar.h:645
@ UCHAR_SCRIPT_EXTENSIONS
Miscellaneous property Script_Extensions (new in Unicode 6.0).
Definition uchar.h:704
@ UCHAR_FULL_COMPOSITION_EXCLUSION
Binary property Full_Composition_Exclusion.
Definition uchar.h:242
@ UCHAR_EMOJI
Binary property Emoji.
Definition uchar.h:440
@ UCHAR_LOWERCASE
Binary property Lowercase.
Definition uchar.h:289
@ UCHAR_CHANGES_WHEN_CASEMAPPED
Binary property Changes_When_Casemapped.
Definition uchar.h:431
@ UCHAR_POSIX_XDIGIT
Binary property xdigit (a C/POSIX character class).
Definition uchar.h:417
UIndicSyllabicCategory
Indic Syllabic Category constants.
Definition uchar.h:2447
@ U_INSC_VISARGA
Definition uchar.h:2519
@ U_INSC_BINDU
Definition uchar.h:2459
@ U_INSC_INVISIBLE_STACKER
Definition uchar.h:2493
@ U_INSC_SYLLABLE_MODIFIER
Definition uchar.h:2511
@ U_INSC_VIRAMA
Definition uchar.h:2517
@ U_INSC_MODIFYING_LETTER
Definition uchar.h:2497
@ U_INSC_REGISTER_SHIFTER
Definition uchar.h:2509
@ U_INSC_NUKTA
Definition uchar.h:2501
@ U_INSC_NON_JOINER
Definition uchar.h:2499
@ U_INSC_TONE_MARK
Definition uchar.h:2515
@ U_INSC_GEMINATION_MARK
Definition uchar.h:2491
@ U_INSC_CONSONANT_KILLER
Definition uchar.h:2475
@ U_INSC_CONSONANT_HEAD_LETTER
Definition uchar.h:2471
@ U_INSC_BRAHMI_JOINING_NUMBER
Definition uchar.h:2461
@ U_INSC_CONSONANT_DEAD
Definition uchar.h:2467
@ U_INSC_OTHER
Definition uchar.h:2455
@ U_INSC_NUMBER
Definition uchar.h:2503
@ U_INSC_TONE_LETTER
Definition uchar.h:2513
@ U_INSC_JOINER
Definition uchar.h:2495
@ U_INSC_CANTILLATION_MARK
Definition uchar.h:2463
@ U_INSC_PURE_KILLER
Definition uchar.h:2507
@ U_INSC_VOWEL_INDEPENDENT
Definition uchar.h:2525
@ U_INSC_CONSONANT_MEDIAL
Definition uchar.h:2477
@ U_INSC_CONSONANT_WITH_STACKER
Definition uchar.h:2489
@ U_INSC_CONSONANT_PREFIXED
Definition uchar.h:2483
@ U_INSC_CONSONANT
Definition uchar.h:2465
@ U_INSC_CONSONANT_SUCCEEDING_REPHA
Definition uchar.h:2487
@ U_INSC_CONSONANT_PLACEHOLDER
Definition uchar.h:2479
@ U_INSC_VOWEL
Definition uchar.h:2521
@ U_INSC_NUMBER_JOINER
Definition uchar.h:2505
@ U_INSC_CONSONANT_PRECEDING_REPHA
Definition uchar.h:2481
@ U_INSC_CONSONANT_SUBJOINED
Definition uchar.h:2485
@ U_INSC_VOWEL_DEPENDENT
Definition uchar.h:2523
@ U_INSC_CONSONANT_INITIAL_POSTFIXED
Definition uchar.h:2473
@ U_INSC_AVAGRAHA
Definition uchar.h:2457
@ U_INSC_CONSONANT_FINAL
Definition uchar.h:2469
UWordBreakValues
Word Break constants.
Definition uchar.h:2159
@ U_WB_E_MODIFIER
Definition uchar.h:2197
@ U_WB_E_BASE_GAZ
Definition uchar.h:2195
@ U_WB_REGIONAL_INDICATOR
Definition uchar.h:2185
@ U_WB_DOUBLE_QUOTE
Definition uchar.h:2191
@ U_WB_LF
Definition uchar.h:2179
@ U_WB_HEBREW_LETTER
Definition uchar.h:2187
@ U_WB_WSEGSPACE
Definition uchar.h:2203
@ U_WB_COUNT
One more than the highest normal UWordBreakValues value.
Definition uchar.h:2212
@ U_WB_GLUE_AFTER_ZWJ
Definition uchar.h:2199
@ U_WB_NEWLINE
Definition uchar.h:2183
@ U_WB_SINGLE_QUOTE
Definition uchar.h:2189
@ U_WB_CR
Definition uchar.h:2175
@ U_WB_E_BASE
Definition uchar.h:2193
@ U_WB_ZWJ
Definition uchar.h:2201
@ U_WB_EXTEND
Definition uchar.h:2177
@ U_WB_MIDNUMLET
Definition uchar.h:2181
UBidiPairedBracketType
Bidi Paired Bracket Type constants.
Definition uchar.h:992
@ U_BPT_CLOSE
Close paired bracket.
Definition uchar.h:1004
@ U_BPT_COUNT
One more than the highest normal UBidiPairedBracketType value.
Definition uchar.h:1012
@ U_BPT_NONE
Not a paired bracket.
Definition uchar.h:1000
@ U_BPT_OPEN
Open paired bracket.
Definition uchar.h:1002
UBool u_isprint(UChar32 c)
Determines whether the specified code point is a printable character.
int32_t u_charDigitValue(UChar32 c)
Returns the decimal digit value of a decimal digit character.
This file defines an abstract map from Unicode code points to integer values.
struct UCPMap UCPMap
Abstract map from Unicode code points (U+0000..U+10FFFF) to integer values.
Definition ucpmap.h:31
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition umachine.h:425
#define U_DEPRECATED
This is used to declare a function as a deprecated public ICU C API
Definition umachine.h:115
#define U_CDECL_END
This is used to end a declaration of a library private ICU C API.
Definition umachine.h:85
int8_t UBool
The ICU boolean type.
Definition umachine.h:261
#define U_CAPI
This is used to declare a function as a public ICU C API.
Definition umachine.h:109
uint16_t UChar
The base type for UTF-16 code units and pointers.
Definition umachine.h:378
#define U_CDECL_BEGIN
This is used to begin a declaration of a library private ICU C API.
Definition umachine.h:84
#define U_STABLE
This is used to declare a function as a stable public ICU C API.
Definition umachine.h:111
struct USet USet
USet is the C API type corresponding to C++ class UnicodeSet.
Definition uset.h:47
Basic definitions for ICU, for both C and C++ APIs.
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition utypes.h:415
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